Biology subjects

Roda, S.

Publications and source records attributed to Roda, S..

4 recordsLinked to original sources

bioRxiv2026

Iterative computational bioprospecting of hydroxymethylfurfural oxidases combining molecular simulation and machine learning

The enzymatic oxidation of 5-hydroxymethylfurfural (HMF) to 2,5-furandicarboxylic acid (FDCA) is a key step in the sustainable production of polyethylene furanoate (PEF), yet the final oxidation of 5-formyl-2-furancarboxylic acid (FFCA) to FDCA remains the principal bottleneck for industrial application. Although hydroxymethylfurfural oxidases (HMFOs) can catalyze this transformation, naturally occurring enzymes with high FFCA-oxidizing activity remain largely unexplored. Here, we present an iterative computational bioprospecting campaign to identify natural HMFOs with improved activity toward this rate-limiting reaction. Across four successive rounds, the computational workflow evolved from conservative homology-based screening to large-scale sequence mining assisted by machine learning, with each iteration redesigned according to the principal limitations identified through molecular simulations and experimental characterization. This adaptive strategy led to the discovery of several natural HMFOs with substantially improved FFCA oxidation activity, including candidates whose performance under the tested conditions approached that of the engineered benchmark 8BxHMFO. Beyond the identification of promising biocatalysts, the study demonstrates the value of continuous integration of computational prediction and experimental feedback for refining enzyme discovery campaigns, and provides practical insights into balancing catalytic prioritization, sequence diversity, developability, and computational scalability in future simulation-guided bioprospecting efforts.

bioinformatics↗

bioRxiv2026

Multi-objective Engineering of Trimethylamine Monooxygenase for Improved Thermostability and Cofactor Use

Trimethylamine (TMA) is a major contributor to undesirable odours in protein hydrolysates derived from marine by-products, limiting their industrial use. Flavin-containing monooxygenases (FMOs) catalyse the conversion of TMA to the odourless trimethylamine N-oxide (TMAO); however, industrial applications demand enzymes that are both thermally stable and compatible with cost-effective cofactors. A thermostable variant of the Methylophaga aminisulfidivorans FMO (mFMO_20) can function at elevated temperatures but depends exclusively on the expensive and unstable cofactor NADPH. In this study, we investigated whether it is possible to simultaneously enhance thermostability and NADH compatibility using a multi-objective engineering strategy. We first targeted residues in the cofactor binding site of mFMO_20 to restore NADH activity, which had been completely lost despite the wild type enzyme being naturally active with both cofactors. Variants derived from the thermostable scaffold partially recovered NADH activity but showed reduced NADPH activity. Given the wild types inherent NADH compatibility, we next pursued a stability-improvement approach, introducing highly conserved stabilizing mutations. This preserved cofactor competence but produced only modest improvements in thermostability. Finally, by combining physical, evolutionary, and statistical metrics, we obtained variants that retained higher NADPH activity after heat treatment than any previously reported thermostable mutants, while a subset also retained measurable NADH activity before heat treatment. These findings show that combining complementary scoring strategies helps navigate the trade-off between stability and activity; while, robust NADH function under thermal stress remains elusive, with only one variant retaining detectable NADH activity after heat treatment, the results provide valuable insight into the underlying constraints linking stability and cofactor usage and highlights possible directions for engineering FMOs with both enhanced thermostability and cofactor compatibility. Author summaryIn this work, we aimed to improve an enzyme that could be useful in industrial applications but is limited by two common constraints: poor stability at high temperatures and dependence on an expensive cofactor. To make the enzyme more suitable for large-scale applications, we sought to engineer variants that are both more thermostable and compatible with a cheaper cofactor, NADH. For enzyme engineering, we used a strategy that balances several properties rather than prioritizing a single trait. We combined tools that capture evolutionary patterns, protein physics, and AI-based predictions to explore which mutations might provide the right combination of stability and function. Through this approach, we obtained variants with improved heat resistance and higher cofactor activity retention.

molecular biology↗

bioRxiv2025

Efficient Design of Affilin(R) Protein Binders for HER3

Engineered scaffold-based proteins that bind to concrete targets with high affinity offer significant advantages over traditional antibodies in theranostic applications. Their development often relies on display methods, where large libraries of variants are physically contacted with the desired target protein and pools of binding variants can be selected. Herein, we use a combined artificial intelligence/physics-based computational framework and phage display approach to obtain ubiquitin based Affilin(R) proteins targeting the HER3 extracellular domain, a relevant tumor target. We demonstrate that the developed in silico pipeline can generate de novo Affilin(R) proteins with high experimental success rate using a small training set of sequences (<1000 sequences). The classical phage display yielded primary candidates with low nanomolar affinities. These binders could be further optimized by phage display and computational maturation alike. These combined efforts resulted in four HER3 ligands with high affinity, cell binding, and serum stability that have theranostic potential.

biochemistry↗

bioRxiv2023

Efficient and accurate sequence generation with small-scale protein language models

Large Language Models (LLMs) have demonstrated exceptional capabilities in understanding contextual relationships, outperforming traditional methodologies in downstream tasks such as text generation and sentence classification. This success has been mirrored in the realm of protein language models (pLMs), where proteins are encoded as text via their amino acid sequences. However, the training of pLMs, which involves tens to hundreds of millions of sequences and hundreds of millions to billions of parameters, poses a significant computational challenge. In this study, we introduce a Small-Scale Protein Language Model (SS-pLM), a more accessible approach that requires training on merely millions of representative sequences, reducing the number of trainable parameters to 14.8M. This model significantly reduces the computational load, thereby democratizing the use of foundational models in protein studies. We demonstrate that the performance of our model, when fine-tuned to a specific set of sequences for generation, is comparable to that of larger, more computationally demanding pLM.

bioinformatics↗