Biology subjects

Abedi, M.

Publications and source records attributed to Abedi, M..

16 recordsLinked to original sources

bioRxiv2026

Designed Minibinders Rewire Receptor Signaling to Enable Functional Human Myogenic Reprogramming

Sarcopenia, loss of muscle mass is a considerable health burden that demands immediate societal attention. Direct myogenic somatic cell reprogramming, a potential muscle regeneration method is constrained by an inability to control the signaling logic that governs cell fate. Here, we show that this barrier can be overcome using AI-designed receptor modulators. Screening de novo minibinders, we identify a synthetic protein cocktail, C6-DPC, that drives efficient human fibroblast-to-muscle transdifferentiation with robust structural and metabolic maturation. C6-DPC reprograms extracellular signaling by activating pro-myogenic FGFR1/2c pathways while suppressing anti-myogenic inputs through ALK1 and TGFBR2; targeted depletion of ALK1 is sufficient to lower the reprogramming barrier. Inflammatory signaling via gp130 emerges as a dominant checkpoint, and its inhibition further enhances conversion. Engineered tissues generate high twitch and tetanic forces in both wild-type and dystrophin-deficient human cells. These findings demonstrate that programmable synthetic ligands can rewrite receptor-level signaling to direct cell fate and enable functional tissue regeneration.

biochemistry↗

bioRxiv2026

De novo design of a macrocycle induced dimerization system for cellular control

Investigating and manipulating cellular events requires precise control of protein function. To enable control over cellular processes, we set out to design a chemically induced dimerization (CID) system consisting of a de novo designed ligand and protein pair. Here we describe the design of a C2 symmetric membrane permeable macrocyclic peptide and a cognate protein homodimer which binds the macrocycle through a large interface with both chains. The designed homodimer binds the macrocycle with a KD of 36 nM, and the x-ray crystal structure of the protein homodimer-macrocycle complex is very close to the computational design model, with the C2 axis of the macrocycle aligned with the homodimer C2 axis. Transcriptional and split luciferase assays in mammalian cells demonstrates conditional control over both a reporter gene expression and luciferase reconstitution.

biochemistry↗

bioRxiv2025

Geometric Tuning of Cytokine Receptor Association Modulates Synthetic Agonist Signaling

Cytokines signal by bringing receptor subunits together, but the role of receptor geometry in shaping signaling remains unclear because natural ligands enforce fixed assemblies. Here, we present a de novo protein design platform that rigidly scaffolds receptor-binding domains into defined spatial arrangements. Applying this across IL-7, type I and III interferons, IL-10, gp130, {beta} common, and synthetic receptor pairs, we show that by varying geometry, we can bias pSTAT pathway usage and tune functional outcomes. Geometric control allowed us to decouple pSTAT1 from pSTAT5 in IL-7, separate antigen presentation (MHC-I) from checkpoint induction (PD-L1) in type I interferons, and suppress pro-inflammatory IFN{gamma} secretion while retaining anti-inflammatory activity in IL-10. We further created minimal IL-6 and IL-3 agonists and strengthened synthetic receptor pairings inaccessible with present cytokines. These results establish receptor geometry as a central determinant of cytokine activity and provide a platform for programmable immune modulation.

synthetic biology↗

bioRxiv2025

Direct cell reprogramming by a designed agonist inducing HER2-FGFR proximity

Growth factor induced receptor dimerization and activation of downstream pathways can modulate cell fate decisions. Here, we investigate the potential of de novo designed synthetic ligands, termed Novokines, to reprogram cell identity by inducing proximity of novel pairs of receptor subunits. We find that a design, H2F, that brings together HER2 (which has no known natural ligand) and the FGF receptor has potent signaling activity. H2F induces robust signaling and reprograms fibroblasts into myogenic cells. Unlike native FGF ligands, H2F selectively activates the MAPK pathway without engaging PLC{gamma}-mediated Ca{superscript 2} signaling. FRET assays confirm H2F-mediated HER2-FGFR proximity, and phosphoproteomic analysis reveals activation of MAPK effectors. H2F-induced ERK phosphorylation is abolished in cells expressing a kinase-dead FGFR1 (K514M) mutant, confirming the requirement for FGFR catalytic activity. H2F treatment significantly increases myofiber formation from adult patient-derived primary myoblasts, demonstrating its capacity to promote myogenic regeneration. Our findings demonstrate that synthetic receptor pairings can rewire signaling outputs to drive regeneration, providing a programmable platform for cell fate engineering.

biochemistry↗

bioRxiv2025

High-Throughput De Novo Protein Design Yields Novel Immunomodulatory Agonists

Cytokines regulate cell behavior by bringing together specific receptor subunits to trigger downstream signaling. Designed molecules that bring together non-natural receptor pairs could have novel signaling responses and cell specificities. We present a high-throughput de-novo design approach to create novel cytokines by generating and fusing pairs of computationally designed binders. By combining 33 designed receptor-binding domains, we generated over a thousand potential de novo designed "Novokines", of which 75 activated pSTAT signaling in peripheral blood mononuclear cells. We characterized 18 of these, including new pairings of established common receptors, cross-family pairings such as TrkA-{gamma}common, and a series of pairings with interferon receptor-1 (IFNAR1), revealing that IFNAR1 can function as a versatile common receptor similar to {gamma}common or {beta}common. We identify novokines that drive monocyte proliferation, T cell survival and CD4+ T cell-specific proliferation. Our framework provides a blueprint for expanding the understanding of cytokine signaling and generating novel therapeutic proteins.

synthetic biology↗

bioRxiv2024

Attentive graph neural network models for the prediction of blood brain barrier permeability

The blood brain barriers (BBB) unique endothelial cells and tight junctions selectively regulate passage of molecules to the central nervous system (CNS) to prevent pathogen entry and maintain neural homeostasis. Various neurological conditions and neurodegenerative diseases benefit from small molecules capable of BBB penetration (BBBP) to elicit a therapeutic effect. Predicting BBBP often involves in silico assessment of molecular properties such as lipophilicity (logP) and polar surface area (PSA) using the CNS multiparameter optimization (MPO) method. This study curated an open-source dataset to benchmark rigorously machine learning (ML) and neural network (NN) models with each other and with MPO for predicting BBBP. Our analysis demonstrated that AI models, especially attentive NNs using stereochemical features, significantly outperform MPO in predicting BBBP. An attentive graph neural network (GNN), we refer to as CANDID-CNS, achieved a 0.23-0.26 higher AUROC score than MPO on full test sets, and a 0.17-0.19 higher score on stereoisomers filtered subsets. Regarding stereoisomers that differ in BBBP, which MPO cannot distinguish, attentive GNNs correctly classify these with AUROC and MCC metrics comparable to or better than MPOs AUROC and MCC on less difficult test molecules. These findings suggest that integrating attentive GNN models into pharmaceutical drug discovery processes can substantially improve prediction rates, and thereby reduce the timeline, cost, and increase probability of success of designing brain penetrant therapeutics for the treatment of a wide variety of neurological and neurodegenerative diseases.

pharmacology and toxicology↗

bioRxiv2024

Aberrant TNF signaling in pancreatic lymph nodes of patients with Type 1 Diabetes

The therapeutic landscape for Type 1 Diabetes (T1D) is rapidly changing as ongoing clinical trials aim to delay beta-cell loss by inhibiting proinflammatory cytokines. However, the precise timing and cellular contexts of cytokine dysregulation remains unknown. We generated the largest existing measurement of gene expression and chromatin accessibility in [~]1 million immune cells from the pancreatic lymph nodes and spleens of 34 T1D and non-diabetic organ donors. Our study revealed heightened gene activity of the tumor necrosis factor (TNF) pathway and subsequent chromatin remodeling in central memory CD4+ T cells residing in the pancreatic lymph nodes of T1D and non-diabetic islet-autoantibody positive donors. These findings, validated in mice, offer a mechanism underlying the efficacy of TNF inhibitors, currently undergoing clinical trials to delay T1D onset.

immunology↗

bioRxiv2023

Designed Endocytosis-Triggering Proteins mediate Targeted Degradation and amplify signaling

Endocytosis and lysosomal trafficking of cell surface receptors can be triggered by interaction with endogenous ligands. Therapeutic approaches such as LYTAC1,2 and KineTAC3, have taken advantage of this to target specific proteins for degradation by fusing modified native ligands to target binding proteins. While powerful, these approaches can be limited by possible competition with the endogenous ligand(s), the requirement in some cases for chemical modification that limits genetic encodability and can complicate manufacturing, and more generally, there may not be natural ligands which stimulate endocytosis through a given receptor. Here we describe general protein design approaches for designing endocytosis triggering binding proteins (EndoTags) that overcome these challenges. We present EndoTags for the IGF-2R, ASGPR, Sortillin, and Transferrin receptors, and show that fusing these tags to proteins which bind to soluble or transmembrane protein leads to lysosomal trafficking and target degradation; as these receptors have different tissue distributions, the different EndoTags could enable targeting of degradation to different tissues. The modularity and genetic encodability of EndoTags enables AND gate control for higher specificity targeted degradation, and the localized secretion of degraders from engineered cells. The tunability and modularity of our genetically encodable EndoTags should contribute to deciphering the relationship between receptor engagement and cellular trafficking, and they have considerable therapeutic potential as targeted degradation inducers, signaling activators for endocytosis-dependent pathways, and cellular uptake inducers for targeted antibody drug and RNA conjugates.

synthetic biology↗

bioRxiv2023

Machine Learning Made Easy (MLme): A Comprehensive Toolkit for Machine Learning-Driven Data Analysis

BackgroundMachine learning (ML) has emerged as a vital asset for researchers to analyze and extract valuable information from complex datasets. However, developing an effective and robust ML pipeline can present a real challenge, demanding considerable time and effort, thereby impeding research progress. Existing tools in this landscape require a profound understanding of ML principles and programming skills. Furthermore, users are required to engage in the comprehensive configuration of their ML pipeline to obtain optimal performance. ResultsTo address these challenges, we have developed a novel tool called Machine Learning Made Easy (MLme) that streamlines the use of ML in research, specifically focusing on classification problems at present. By integrating four essential functionalities, namely Data Exploration, AutoML, CustomML, and Visualization, MLme fulfills the diverse requirements of researchers while eliminating the need for extensive coding efforts. To demonstrate the applicability of MLme, we conducted rigorous testing on six distinct datasets, each presenting unique characteristics and challenges. Our results consistently showed promising performance across different datasets, reaffirming the versatility and effectiveness of the tool. Additionally, by utilizing MLmes feature selection functionality, we successfully identified significant markers for CD8+ naive (BACH2), CD16+ (CD16), and CD14+ (VCAN) cell populations. ConclusionMLme serves as a valuable resource for leveraging machine learning (ML) to facilitate insightful data analysis and enhance research outcomes, while alleviating concerns related to complex coding scripts. The source code and a detailed tutorial for MLme are available at https://github.com/FunctionalUrology/MLme. Key PointsO_LIMLme is a novel tool that simplifies machine learning (ML) for researchers by integrating Data Exploration, AutoML, CustomML, and Visualization functionalities. C_LIO_LIMLme improves efficiency and productivity by streamlining the ML workflow and eliminating the need for extensive coding efforts. C_LIO_LIRigorous testing on diverse datasets demonstrates MLmes promising performance in classification problems. C_LIO_LIMLme provides intuitive interfaces for data exploration, automated ML, customizable ML pipelines, and result visualization. C_LIO_LIFuture developments aim to expand MLmes capabilities to include support for unsupervised learning, regression, hyperparameter tuning, and integration of user-defined algorithms. C_LI

bioinformatics↗

bioRxiv2023

SpheroScan: A User-Friendly Deep Learning Tool for Spheroid Image Analysis

BackgroundIn recent years, three-dimensional (3D) spheroid models have become increasingly popular in scientific research as they provide a more physiologically relevant microenvironment that mimics in vivo conditions. The use of 3D spheroid assays has proven to be advantageous as it offers a better understanding of the cellular behavior, drug efficacy, and toxicity as compared to traditional two-dimensional cell culture methods. However, the use of 3D spheroid assays is impeded by the absence of automated and user-friendly tools for spheroid image analysis, which adversely affects the reproducibility and throughput of these assays. ResultsTo address these issues, we have developed a fully automated, web-based tool called SpheroScan, which uses the deep learning framework called Mask Regions with Convolutional Neural Networks (R-CNN) for image detection and segmentation. To develop a deep learning model that could be applied to spheroid images from a range of experimental conditions, we trained the model using spheroid images captured using IncuCyte Live-Cell Analysis System and a conventional microscope. Performance evaluation of the trained model using validation and test datasets shows promising results. ConclusionSpheroScan allows for easy analysis of large numbers of images and provides interactive visualization features for a more in-depth understanding of the data. Our tool represents a significant advancement in the analysis of spheroid images and will facilitate the widespread adoption of 3D spheroid models in scientific research. The source code and a detailed tutorial for SpheroScan are available at https://github.com/FunctionalUrology/SpheroScan. Key PointsO_LIA deep learning model was trained to detect and segment spheroids in images from microscopes and Incucytes. C_LIO_LIThe model performed well on both types of images with the total loss decreasing significantly during the training process. C_LIO_LIA web tool called SpheroScan was developed to facilitate the analysis of spheroid images, which includes prediction and visualization modules. C_LIO_LISpheroScan is efficient and scalable, making it possible to handle large datasets with ease. C_LIO_LISpheroScan is user-friendly and accessible to researchers, making it a valuable resource for the analysis of spheroid image data. C_LI

bioinformatics↗

bioRxiv2023

Computational design of non-porous, pH-responsive antibody nanoparticles

Programming protein nanomaterials to respond to changes in environmental conditions is a current challenge for protein design and important for targeted delivery of biologics. We describe the design of octahedral non-porous nanoparticles with the three symmetry axes (four-fold, three-fold, and two-fold) occupied by three distinct protein homooligomers: a de novo designed tetramer, an antibody of interest, and a designed trimer programmed to disassemble below a tunable pH transition point. The nanoparticles assemble cooperatively from independently purified components, and a cryo-EM density map reveals that the structure is very close to the computational design model. The designed nanoparticles can package a variety of molecular payloads, are endocytosed following antibody-mediated targeting of cell surface receptors, and undergo tunable pH-dependent disassembly at pH values ranging between to 5.9-6.7. To our knowledge, these are the first designed nanoparticles with more than two structural components and with finely tunable environmental sensitivity, and they provide new routes to antibody-directed targeted delivery.

bioengineering↗

bioRxiv2022

MLcps: Machine Learning Cumulative Performance Score for classification problems

MotivationA performance metric is a tool to measure the correctness of a trained Machine Learning (ML) model. Numerous performance metrics have been developed for classification problems making it overwhelming to select the appropriate one since each of them represents a particular aspect of the model. Furthermore, selection of a performance metric becomes harder for problems with imbalanced and/or small datasets. Therefore, in clinical studies where datasets are frequently imbalanced and, in situations when the prevalence of a disease is low or the collection of patient samples is difficult, deciding on a suitable metric for performance evaluation of an ML model becomes quite challenging. The most common approach to address this problem is measuring multiple metrics and compare them to identify the best-performing ML model. However, comparison of multiple metrics is laborious and prone to user preference bias. Furthermore, evaluation metrics are also required by ML model optimization techniques such as hyperparameter tuning, where we train many models, each with different parameters, and compare their performances to identify the best-performing parameters. In such situations, it becomes almost impossible to assess different models by comparing multiple metrics. ResultsHere, we propose a new metric called Machine Learning Cumulative Performance Score (MLcps) as a Python package for classification problems. MLcps combines multiple pre-computed performance metrics into one metric that conserves the essence of all pre-computed metrics for a particular model. We tested MLcps on 4 different publicly available biological datasets and the results reveal that it provides a comprehensive picture of overall model robustness. AvailabilityMLcps is available at https://pypi.org/project/MLcps/ and cases of use are available at https://mybinder.org/v2/gh/FunctionalUrology/MLcps.git/main. Supplementary informationSupplementary data are available at Bioinformatics online.

bioinformatics↗

bioRxiv2022

Peptide binding specificity prediction using fine-tuned protein structure prediction networks

Peptide binding proteins play key roles in biology, and predicting their binding specificity is a long-standing challenge. While considerable protein structural information is available, the most successful current methods use sequence information alone, in part because it has been a challenge to model the subtle structural changes accompanying sequence substitutions. Protein structure prediction networks such as AlphaFold model sequence-structure relationships very accurately, and we reasoned that if it were possible to specifically train such networks on binding data, more generalizable models could be created. We show that placing a classifier on top of the AlphaFold network and fine-tuning the combined network parameters for both classification and structure prediction accuracy leads to a model with strong generalizable performance on a wide range of Class I and Class II peptide-MHC interactions that approaches the overall performance of the state-of-the-art NetMHCpan sequence-based method. The peptide-MHC optimized model shows excellent performance in distinguishing binding and non-binding peptides to SH3 and PDZ domains. This ability to generalize well beyond the training set far exceeds that of sequence only models, and should be particularly powerful for systems where less experimental data is available. Significance statementPeptide binding proteins carry out a variety of biological functions in cells and predicting their binding specificity could significantly improve our understanding of molecular pathways. Deep neural networks have achieved high structure prediction accuracy, but are not trained to predict binding specificity. Here we describe an approach to extending such networks to jointly predict protein structure and binding specificity. We incorporate AlphaFold into this approach, and fine-tune its parameters on peptide-MHC Class I and II structural and binding data. The fine-tuned model approaches state-of-the-art classification accuracy on peptide-MHC specificity prediction and generalizes to other peptide-binding systems such as the PDZ and SH3 domains.

bioinformatics↗

bioRxiv2021

Acoustic Remote Control of Bacterial Immunotherapy

Rapid advances in synthetic biology are driving the development of genetically engineered microbes as therapeutic agents for a multitude of human diseases, including cancer. In particular, the immunosuppressive microenvironment of solid tumors creates a favorable niche for systemically administered bacteria to engraft in the tumor and release therapeutic payloads. However, such payloads can be harmful if released in healthy tissues where the bacteria also engraft in smaller numbers. To address this limitation, we engineer therapeutic bacteria to be controlled by focused ultrasound, a form of energy that can be applied noninvasively to specific anatomical sites such as solid tumors. This control is provided by a temperature-actuated genetic state switch that produces lasting therapeutic output in response to briefly applied focused ultrasound hyperthermia. Using a combination of rational design and high-throughput screening we optimized the switching circuits of engineered cells and connected their activity to the release of immune checkpoint inhibitors. In a clinically relevant cancer model, ultrasound-activated therapeutic microbes successfully turned on in situ and induced a marked suppression of tumor growth. This technology provides a critical tool for the spatiotemporal targeting of potent bacterial therapeutics in a variety of biological and clinical scenarios.

synthetic biology↗

bioRxiv2019

Using a Motion Sensor to Categorize Low Back Pain Patients: A Machine Learning Approach

Low back pain (LBP) remains a critical health issue impacting literally millions of people worldwide. Currently, clinical practitioners rely on subjective measures such as the STarT Back Screening Tool to categorize LBP patients, which then informs specific treatment regimens. This study sought to develop a machine learning model to classify LBP patients into different groups according to kinematic data. Specifically, an inertial measurement unit (IMU) was attached to each patients chest while he performed trunk flexion/extension motions at a self-selected pace. Machine learning algorithms such as support vector machine (SVM) and multi-layer perceptron (MLP) were implemented to evaluate the efficiency of the models. The results showed that the kinematic data we obtained could be used to categorize the patients into two groups: high vs. low-medium risk. We achieved accuracy levels of ~75% and 60% for SVM and MLP, respectively. Additionally, among a range of variables detailed herein, we determined that time-scaled IMU signal resulted in the highest accuracy. Our findings support the use of body-motion measures in developing prognosis tools for healthcare applications. Our results could help overcome the need for objective clinic-based diagnosis approaches, which in turn would lead to assigning better treatment approaches and rehabilitation services for LBP sufferers.

bioengineering↗

bioRxiv2019

Acoustically Detonated Biomolecules for Genetically Encodable Inertial Cavitation

Recent advances in molecular engineering and synthetic biology have made it possible for biomolecular and cell-based therapies to provide highly specific disease treatment. However, both the ability to spatially target the action of such therapies, and their range of effects on the target tissue remain limited. Here we show that biomolecules and cells can be engineered to deliver potent mechanical effects at specific locations inside the body under the direction of focused ultrasound. This capability is based on gas vesicles, a unique class of air-filled protein nanostructures derived from buoyant photosynthetic microbes. We show that low-frequency ultrasound can convert these nanoscale biomolecules into micron-scale cavitating bubbles, as demonstrated with acoustic measurements and ultrafast optical microscopy. This allows gas vesicles targeted to cell-surface receptors to serve as remotely detonated cell-killing agents. In addition, it allows cells genetically engineered to express gas vesicles to be triggered with ultrasound to lyse and release therapeutic payloads. We demonstrate these capabilities in vitro, in cellulo, and in vivo. This technology equips biomolecular and cellular therapeutics with unique capabilities for spatiotemporal control and mechanical action.

bioengineering↗