Biology subjects

Stephenson, M.

Publications and source records attributed to Stephenson, M..

4 recordsLinked to original sources

bioRxiv2026

ChatGEM: An Agentic Architecture Enabling Interactive Simulation of Genome-Scale Metabolic Models

Genome-scale metabolic models (GEMs) are powerful tools for predicting cellular phenotypes and guiding microbial strain engineering, yet broad adoption remains challenging due to the computational expertise required. To overcome that, we present ChatGEM, an agentic platform that enables interactive GEM simulation through natural language. Built on the multi-agent ADEPT framework, ChatGEM integrates COBRApy within a retrieval-augmented generation (RAG) architecture that coordinates code generation and execution through specialized agents. Benchmarking across three tasks of increasing complexity showed that RAG-enabled code generation improved the mean overall performance score from 2.63 to 4.20 while reducing the execution time significantly starting from routine to complex tasks. Application of ChatGEM using an enzyme-constrained GEM (ecGEM) for four engineered Pseudomonas putida KT2440 strains identified the constitutive strain as the optimal chassis for succinate overproduction using a succinate leakage index - a prediction observed experimentally. Therefore, ChatGEM democratizes metabolic modeling by enabling researchers without computational expertise to perform sophisticated GEM-based analyses through natural language, and, hence, accelerating scientific discovery.

systems biology↗

bioRxiv2026

Systematic prediction and functional analysis of amino acid residues determining product specificity in the plant oxidosqualene cyclase superfamily

Oxidosqualene cyclases (OSCs) catalyse one of natures most intricate enzyme reactions, converting the linear precursor 2,3-oxidosqualene into an array of cyclic triterpene scaffolds through sequential carbocation cascades. Predicting OSC function based on sequence is challenging beyond broad family-level classification. Here, we develop a structure-based computational framework to identify amino acid determinants of OSC product specificity. Using 169 functionally characterised OSCs, we deploy a multifaceted approach combining differential conservation along with structural information, physico-chemical properties of amino acids and binding pocket electrostatics in order to understand the determinants of product specificity. Using Arabidopsis thaliana cycloartenol synthase AtCAS as a model, we then validate our predictions through targeted mutagenesis, achieving stepwise reprogramming towards the protosteryl-type products cucurbitadienol and lanosterol, including complete product switches. Molecular dynamics simulations support a mechanism in which subtle pocket remodelling alters active-site volume, water access and proton-elimination chemistry. These findings provide a blueprint for OSC engineering.

plant biology↗

bioRxiv2025

Reinforced CRISPR interference enables reliable multiplex gene repression in phylogenetically distant bacteria

Genetic screens are essential for uncovering novel molecular mechanisms and identifying the functions of hypothetical proteins. CRISPR interference (CRISPRi) is a powerful, programmable, and sequence-specific gene repression technology that can be used for high-throughput screening and targeted gene repression. Despite its ease of use, the initial development of CRISPRi systems is labor-intensive in many non-model organisms. Our goal is to simplify this by establishing a host-agnostic CRISPRi platform that utilizes the serine recombinase-assisted genome engineering (SAGE) system. This system integrates CRISPRi machinery directly into the bacterial chromosome, overcoming the limitations of plasmid-based systems and enabling wide sharing across diverse bacteria. We demonstrate the design and optimization of multiplexed CRISPRi to repress multiple genes simultaneously in phylogenetically distant bacteria. We use a Francisella novicida-derived Cas12a system that processes multiple distinct CRISPR RNAs, each targeting a unique gene sequence, from a single transcript. This allows easy multi-gene repression. By reinforcing gene repression with multiple guides targeting a single gene, we achieve robust genetic perturbations without the need to pre-screen the efficacy of guide RNAs. Using this toolkit, we perturb multiple combinations of growth and visual phenotypes in Pseudomonas fluorescens and demonstrate simultaneous repression of multiple fluorescent proteins to near background levels in bacteria from various other genera. While the tools are directly portable to all SAGE-compatible microbes, we illustrate the utility of SAGE by optimizing CRISPRi performance in Rhodococcus jostii through a combinatorial screen of Cas protein and CRISPR array expression variants. The efficient integration of CRISPRi machinery via the SAGE system paves the way for versatile genetic screening, enabling profound insights into gene functions both in laboratory conditions and relevant naturalistic scenarios.

microbiology↗

bioRxiv2025

Efficient genetic code expansion tools enable in vivo study of lysine acetylation in non-model bacteria

Recent proteomic advancements have revealed widespread N{varepsilon}-lysine acetylation in pathways governing pathogenicity, metabolism, and antibiotic resistance in bacteria. The spontaneous, non-specific nature of this modification in prokaryotes obscures its biological role, necessitating prokaryotic specific in vivo interrogation systems. Genetic Code Expansion (GCE) offers a powerful method to investigate the roles and regulation dynamics of acetyl-lysine in vivo with the precise incorporation of a suite of non-canonical amino acids, including acetyl-lysine analogs. However, its use has been largely restricted to E. coli strains due to challenges associated with implementation and optimization of the technology in more diverse bacterial strains. Here, we present a bacterial host-agnostic, readily optimizable GCE platform designed to site-specifically incorporate non-canonical amino acids into target proteins within living bacteria. We further demonstrate the versatility of this technology by showcasing, for the first time, the successful incorporation of acetyl-lysine in a non-E. coli bacterium.

synthetic biology↗