Search bioRxiv⌕ Search

Biology subjects

Poltorak, V.

Publications and source records attributed to Poltorak, V..

2 recordsLinked to original sources

Enzymatic carbon-fluorine bond cleavage by human gut microbes

Fluorinated compounds are used for agrochemical, pharmaceutical, and numerous industrial applications, resulting in global contamination. In many molecules, fluorine is incorporated to enhance the half-life and improve bioavailability. Fluorinated compounds enter the human body through food, water, and xenobiotics including pharmaceuticals, exposing gut microbes to these substances. The human gut microbiota is known for its xenobiotic biotransformation capabilities, but it was not previously known whether gut microbial enzymes could break carbon-fluorine bonds, potentially altering the toxicity of these compounds. Here, through the development of a rapid, miniaturized fluoride detection assay for whole-cell screening, we discovered active gut microbial defluorinases. We biochemically characterized enzymes from diverse human gut microbial classes including Clostridia, Bacilli, and Coriobacteriia, with the capacity to hydrolyze (di)fluorinated organic acids and a fluorinated amino acid. Whole-protein alanine scanning, molecular dynamics simulations, and chimeric protein design enabled the identification of a disordered C-terminal protein segment involved in defluorination activity. Domain swapping exclusively of the C-terminus conferred defluorination activity to a non-defluorinating dehalogenase. To advance our understanding of the structural and sequence differences between defluorinating and non-defluorinating dehalogenases, we trained machine learning models which identified protein termini as important features. Models trained on 41-amino acid segments from protein C-termini alone predicted defluorination activity with 83% accuracy (compared to 95% accuracy based on full-length protein features). This work is relevant for therapeutic interventions and environmental and human health by uncovering specificity-determining signatures of fluorine biochemistry from the gut microbiome. SignificanceHumans have introduced carbon-fluorine bonds into numerous manufactured compounds, including pharmaceuticals, leading to the formation of toxic fluorinated byproducts. While the human gut microbiota is known for its ability to metabolize drugs, its encoded capacity to break the strong carbon-fluorine chemical bond was previously unknown. Here we discovered that human gut microbial enzymes are capable of cleaving carbon-fluorine bonds. We developed a 96-well colorimetric fluoride assay amenable to bacterial culture-based screening. We additionally conducted whole-protein alanine scanning mutagenesis and identified through machine learning that flexible C-terminal loop residues were predictive of defluorination. Taken in the context of flexible regions of other enzyme families known to perform fluorine chemistry, this work supports using convergent structural features to predict defluorination specificity.

biochemistry↗

Spec2Class: Accurate Prediction of Plant Secondary Metabolite Class using Deep Learning

Mass spectrometry (MS)-based data is commonly used in studying metabolism and natural products, but typically requires domain-specific skill and experience to analyze. Existing computational tools for non-targeted metabolite analysis (i.e., metabolomics) mostly rely on comparison to reference MS spectral libraries for metabolite identification, limiting the annotation of metabolites for which reference spectra do not exist. This is the case in plant secondary metabolites, where most spectral features remain unidentified. Here, we developed Spec2Class, a deep-learning algorithm for the identification and classification of plant secondary metabolites from liquid chromatography (LC)-MS/MS spectra. We used the in-house spectral library of 7973 plant metabolite chemical standards, alongside publicly available data, to train Spec2Class to classify LC-MS/MS spectra to 43 common plant secondary metabolite classes. Tested on held out sets, our algorithm achieved an overall accuracy of 73%, outperforming state-of-the-art classification. We further established a prediction certainty parameter to set a threshold for low-confidence results. Applying this threshold, we reached an accuracy of 93% on an unseen dataset. We show a high robustness of our prediction to noise and to the data acquisition method. Spec2Class is publicly available and is anticipated to facilitate metabolite identification and accelerate natural product discovery. Significance StatementUntargeted mass spectrometry (MS) is essential for natural product discovery but is limited by product identification, which is often manual and requires domain-specific skills. Spec2Class addresses this limitation by accurately classifying plant secondary metabolites from LC-MS/MS spectra without reliance on reference spectral libraries. Trained on a substantial dataset and using a prediction certainty threshold, it outperforms state-of-the-art algorithms with 93% accuracy. This tool demonstrates high robustness against noise and different data acquisition methods, promising to streamline metabolite identification and expedite natural product research. Spec2Class is open-source, publicly available, and easy to integrate into natural product discovery pipelines.

biochemistry↗