Biology subjects

Acevedo-Rocha, C. G.

Publications and source records attributed to Acevedo-Rocha, C. G..

11 recordsLinked to original sources

bioRxiv2026

Translating chemical reactions into enzyme sequences enables formate dehydrogenase engineering

Enzyme engineering has broad applications across chemistry and biotechnology but remains costly and susceptible to low success rates. Generative artificial intelligence can substantially expand the sequence space accessible to enzyme engineering; however, directing generation towards enzymes that catalyse a user-defined chemical reaction remains challenging. Inspired by machine translation, we formulate enzyme generation as translation between two molecular languages: chemical reactions and amino acid sequences. Here, we introduce REXzyme, a translation model that generates enzymes conditioned on user-defined reactions. First, we evaluate its designs using multiple in silico metrics and characterize the strengths and limitations of its generalisation capacity. Then, we steer REXzyme towards the generation of enhanced formate dehydrogenases for their reverse natural reaction, i.e., the reduction of carbon dioxide. All 20 generated enzymes were expressed and purified as soluble proteins and exhibited CO2-reduction activity, with the most active design reaching 2.5-fold the specific activity of the reference enzyme. Notably, all substitutions in the designed variants were performed on distal positions outside the catalytic site. Our results show that reaction-to-sequence translation provides a route for efficiently engineering enhanced biocatalysts.

biochemistry↗

bioRxiv2026

Biocatalytic Production of Galantamine in Yeast through Cytochrome P450 Optimization

Galantamine is a pharmaceutically relevant Amaryllidaceae alkaloid used for the treatment of Alzheimer's disease. Its structural complexity and low abundance in plants motivate the development of alternative manufacturing routes. Here, we established the first engineered yeast platform for the biocatalytic production of galantamine from 4OMe-norbelladine. Heterologous expression of the downstream pathway enzymes NtCYP96T6, NtNMT1, and NtAKR1 from Narcissus cv. Tete-a-Tete in Saccharomyces cerevisiae was complemented by optimization of cultivation temperature, carbon source, and medium pH enabling the first demonstration of galantamine production in yeast. Systematic screening of cytochrome P450 reductase and cytochrome b partners to boost NtCYP96T6 activity led to the identification of a novel reductase mined from the Narcissus pseudonarcissus transcriptome, NpCPR, which supported the highest pathway flux and yielded 7.0 {+/-} 0.6 mg/L galantamine, corresponding to a 7.6 % molar yield from 250 M 4OMe-norbelladine and an approximately 173-fold improvement over the parental strain. We further exploited this yeast cell factory for the precursor-directed biosynthesis of 7F-galantamine, highlighting the potential of pathway enzyme promiscuity to access new-to-nature GAL analogues that may be challenging to produce through conventional chemical synthesis. Together, this work establishes a foundation for microbial galantamine production and biosynthetic diversification of its pharmaceutically relevant scaffold.

synthetic biology↗

bioRxiv2026

Unlocking substrate specificities of human solute carrier proteins using untargeted metabolomics

The limited understanding of transporter substrate spectra constrains our ability to interpret cell and membrane function, highlighting the need for methods that enable transporter deorphanization and characterization of promiscuous transport activities. Here, we present a Xenopus oocyte-based platform for unbiased transporter substrate discovery. Oocytes expressing heterologous solute carrier proteins (SLC) were incubated in human blood serum, a chemically complex metabolite library containing thousands of endogenous metabolites and xenobiotics, followed by paired untargeted LC-MS/MS profiling of intracellular extracts and surrounding medium to capture metabolite exchange events. Across the five human SLC transporters, viz. SLC10A2, SLC10A6, SLC13A2, SLC16A10, and SLC46A1, metabolite exchange signatures were detected, and automated feature annotation was refined by manual chromatographic peak inspection. The workflow recovered known substrates of SLC10A2 and SLC16A10 and identified additional transported metabolites with MS/MS confirmation. This method provides a scalable framework for transporter substrate profiling and prioritization of candidates for targeted validation.

systems biology↗

bioRxiv2026

Hybrid quantum-classical de novo design of MHC-binding peptides

Deep generative models have become a leading approach for designing therapeutic molecules, yet efficiently exploring vast biomolecular sequence spaces remains difficult, particularly for targets with limited training data. The prior distribution that seeds a generative model shapes which regions of sequence space it explores, and recent work suggests that non-classical distributions sampled from quantum processors can serve as a structured alternative to the factorised Gaussian priors used by default. Whether such priors help on complex biological design tasks has been largely untested. Here we present what is, to our knowledge, the first end-to-end hybrid quantum-classical pipeline for de novo design of MHC class I-binding peptides, coupling a generative adversarial network (GAN) to latent vectors sampled from a real photonic quantum processor. Tested in silico across 131 HLA alleles, quantum-derived priors increased the yield of predicted strong binders, with the largest relative gains for understudied alleles where classical baselines perform worst. We selected three understudied alleles for further evaluation, finding that large gains coincided with broader sequence exploration at non-anchor positions while anchor specificity was preserved. On these three alleles, we validated the designs in vitro using peptide-MHC stability ELISAs, confirming that quantum-designed peptides are potent stabilisers of peptide-MHC class I complexes. These results establish structured, hardware-realisable non-classical priors as a useful inductive bias for generative peptide design, with direct relevance to personalised immunotherapies and vaccines.

biochemistry↗

bioRxiv2026

Mycol: A user-friendly app for automating analysis of microscopy images

Microscopy image analysis is central to modern biology, yet many available platforms remain inaccessible to non-specialist users because they require advanced technical expertise, code-based workflows, extensive setup, or paid access. This creates a barrier for researchers who need reliable and fast image quantification but lack dedicated computational support. Here, we introduce Mycol, an open-source, machine-learning-assisted image analysis platform designed to be accessible and run on standard laptops with minimal setup. Mycol supports end-to-end workflows in which users annotate microscopy images, perform human-in-the-loop fine-tuning of machine learning models for automated segmentation and classification, deploy machine learning models, quality control predictions and quantitatively compare morphological and class frequency descriptors through a single intuitive interface. By combining machine-learning analysis with efficient quality control by humans, Mycol makes rapid and high-quality image quantification available to biologists without requiring specialist training. We demonstrate the utility of Mycol in diverse workflows using two economically important organisms, the crop pathogen (Fusarium oxysporum) and the blue mussel (Mytilus edulis). Through Mycol, curated training sets were generated and high quality segmentation and classification models were obtained in each case. Deploying these models through Mycol decreased the time requirements and increased traceability of established cell counting workflows and facilitated a quantitative comparison of morphological parameters that reveals new patterns in early M. edulis larval development.

bioinformatics↗

bioRxiv2026

Engineering Biosensors to Enhance Monoterpene Indole Alkaloid Production in Yeast

Monoterpene Indole Alkaloids (MIAs) are a diverse family of plant natural products with various medicinal applications. Although MIAs, such as vinblastine and reserpine, are clinically validated, sourcing of MIAs for clinical use or drug discovery from natural resources or via chemical synthesis is hampered due to their scarcity and chemical complexity. Refactoring MIA biosynthesis pathways in microbial cell factories could offer an alternative, more stable and potentially sustainable manufacturing route for alkaloid medicines and novel therapies. However, reaching commercially attractive titers, rates and yields remains challenging owing to the length and complexity of these metabolic pathways. One critical bottleneck is the low screening throughput and very high cost of the analytical methods used to quantify MIA for optimizing production. In this study, we evolved RamR, a promiscuous bacterial transcription factor to respond to five different MIAs, resulting in highly sensitive and selective sensor variants (EC50<10 M). X-ray crystallography and computational modeling provided insight into the MIA binding of the evolved biosensor variants. The RamR biosensing platform was functionalized in yeast and subsequently applied in a cost-effective semi-throughput screening campaign of a 188-gene overexpression library to identify high-performing cell factory designs for strictosidine, the common precursor for all MIAs. The fluorescent biosensor signal correlated with HPLC quantification (r2 = 0.932) allowing identification of single metabolic engineering hits which when combined yielded a maximum titer of >220 mg/L strictosidine, 3-fold higher than the parental reference strain. This study demonstrates the development of selective biosensors for MIAs and the cost-effective identification of novel metabolic engineering hits for optimizing MIA production in microbial cell factories.

synthetic biology↗

bioRxiv2024

ProteusAI: An Open-Source and User-Friendly Platform for Machine Learning-Guided Protein Design and Engineering

AO_SCPLOWBSTRACTC_SCPLOWProtein design and engineering are crucial for advancements in biotechnology, medicine, and sustainability. Machine learning (ML) models are used to design or enhance protein properties such as stability, catalytic activity, and selectivity. However, many existing ML tools require specialized expertise or lack open-source availability, limiting broader use and further development. To address this, we developed ProteusAI, a user-friendly and open-source ML platform to streamline protein engineering and design tasks. ProteusAI offers modules to support researchers in various stages of the design-build-test-learn (DBTL) cycle, including protein discovery, structure-based design, zero-shot predictions, and ML-guided directed evolution (MLDE). Our benchmarking results demonstrate ProteusAIs efficiency in improving proteins and enyzmes within a few DBTL-cycle iterations. ProteusAI democratizes access to ML-guided protein engineering and is freely available for academic and commercial use. Future work aims to expand and integrate novel methods in computational protein and enzyme design to further develop ProteusAI.

bioinformatics↗

bioRxiv2024

Metabolic engineering of yeast for de novo production of kratom monoterpene indole alkaloids

Monoterpene indole alkaloids (MIAs) from Mitragyna speciosa ("kratom"), such as mitragynine and speciogynine, are promising novel scaffolds for opioid receptor ligands for treatment of pain, addiction, and depression. While kratom leaves have been used for centuries in South-East Asia as stimulant and pain management substance, the biosynthetic pathway of these psychoactives have only recently been partially elucidated. Here, we demonstrate the de novo production of mitragynine and speciogynine in Saccharomyces cerevisiae through the reconstruction of a five-step synthetic pathway from common MIA precursor strictosidine comprising fungal tryptamine 4-monooxygenase to bypass an unknown kratom hydroxylase. Upon optimizing cultivation conditions, a titer of [~]290 {micro}g/L kratom MIAs from glucose was achieved. Untargeted metabolomics analysis of lead production strains led to the identification of numerous shunt products derived from the activity of strictosidine synthase (STR) and dihydrocorynantheine synthase (DCS), highlighting them as candidates for enzyme engineering to further improve kratom MIAs production in yeast. Finally, by feeding fluorinated tryptamine and expressing a human tailoring enzyme, we further demonstrate production of fluorinated and hydroxylated mitragynine derivatives with potential applications in drug discovery campaigns. Altogether, this study introduces a yeast cell factory platform for the biomanufacturing of complex natural and new-to-nature kratom MIAs derivatives with therapeutic potential.

synthetic biology↗

bioRxiv2024

Harnessing Chemical Space Neural Networks to Systematically Annotate GPCR ligands

Machine learning (ML) has revolutionized drug discovery by enabling the exploration of vast, uncharted chemical spaces essential for discovering novel patentable drugs. Despite the critical role of human G protein-coupled receptors (hGPCRs) in FDA-approved drugs, exhaustive in-distribution drug-target interaction (DTI) testing across all pairs of hGPCRs and known drugs is rare due to significant economic and technical challenges. This often leaves off-target effects unexplored, which poses a considerable risk to drug safety. In contrast to the traditional focus on out-of-distribution (OOD) exploration (drug discovery), we introduce a neighborhood-to-prediction model termed Chemical Space Neural Networks (CSNN) that leverages network homophily and training-free graph neural networks (GNNs) with Labels as Features (LaF). We show that CSNNs ability to make accurate predictions strongly correlates with network homophily. Thus, LaFs strongly increase a ML models capacity to enhance in-distribution prediction accuracy, which we show by integrating labeled data during inference. We validate these advancements in a high-throughput yeast biosensing system (3773 DTIs, 539 compounds, 7 hGPCRs) to discover novel DTIs for FDA-approved drugs and to expand the general understanding of how to build reliable predictors to guide experimental verification.

bioinformatics↗

bioRxiv2020

Pervasive cooperative mutational effects on multiple catalytic enzyme traits emerge via long-range conformational dynamics

Multidimensional fitness landscapes provide insights into the molecular basis of laboratory and natural evolution. Yet such efforts are rare and focus only on limited protein families and a single enzyme trait, with little concern about the relationship between protein epistasis and conformational dynamics. Here, we report the first multiparametric fitness landscape for a cytochrome P450 monooxygenase that was engineered for the regio- and stereoselective hydroxylation of a steroid. We developed a computational program to automatically quantify non-additive effects among all possible mutational pathways, finding pervasive cooperative sign and magnitude epistasis on multiple catalytic traits. By using quantum mechanics and molecular dynamics simulations, we show that these effects are modulated by long-range interactions in loops, helices and beta-strands that gate the substrate access channel allowing for optimal catalysis. Our work highlights the importance of conformational dynamics on epistasis in an enzyme involved in secondary metabolism and offers lessons for engineering P450s.

evolutionary biology↗

bioRxiv2020

In vivo selection for formate dehydrogenases with high efficiency and specificity towards NADP+

Efficient regeneration of cofactors is vital for the establishment of continuous biocatalytic processes. Formate is an ideal electron donor for cofactor regeneration due to its general availability, low reduction potential, and benign byproduct (CO2). However, formate dehydrogenases (FDHs) are usual specific to NAD+, such that NADPH regeneration with formate is challenging. Previous studies reported naturally occurring FDHs or engineered FDHs that accept NADP+, but these enzymes show low kinetic efficiencies and specificities. Here, we harness the power of natural selection to engineer FDH variants to simultaneously optimize three properties: kinetic efficiency with NADP+, specificity towards NADP+, and affinity towards formate. By simultaneously mutating multiple residues of FDH from Pseudomonas sp. 101, which exhibits no initial activity towards NADP+, we generate a library of >106 variants. We introduce this library into an E. coli strain that cannot produce NADPH. By selecting for growth with formate as sole NADPH source, we isolate several enzyme variants that support efficient NADPH regeneration. We find that the kinetically superior enzyme variant, harboring five mutations, has 5-fold higher efficiency and 13-fold higher specificity than the best enzyme previously engineered, while retaining high affinity towards formate. By using molecular dynamics simulations, we reveal the contribution of each mutation to the superior kinetics of this variant. We further determine how non-additive epistatic effects improve multiple parameters simultaneously. Our work demonstrates the capacity of in vivo selection to identify superior enzyme variants carrying multiple mutations which would be almost impossible to find using conventional screening methods.

biochemistry↗