RECETOX Doctoral Conference 2025
Poster session
Deciphering the placenta chemical exposome using non-targeted LC-HRMS
Thomas Angelo Adrien Contini
During pregnancy, women are exposed to mixtures of chemical compounds, that have been associated with potential long-term health concerns for offspring, such as fertility issues and neurodevelopmental disorders. The placenta is a vital organ in fetal development, facilitating the exchange of nutrients, oxygen, and waste between maternal blood and the developing fetus. Combining several advantageous aspects, the placenta is now recognized as a relevant matrix to study prenatal chemical exposures. However, questions remain to be answered before using the organ to study prenatal chemical exposure at cohort-scale. To address these challenges, DECRIPT implements a standardized methodology to apply non-targeted analyses, using high-resolution mass spectrometry (HRMS) to evaluate the intra- and inter-individual variability of the metabolome and chemical exposome in placentas.
A Novel Compositional Dissimilarity Index for Microbiome Data: Application to Human Gastrointestinal Tract Samples
Jan Böhm
We introduce a novel compositional dissimilarity index (CDI) for microbiome data. CDI analysis was performed on 16S rRNA amplicon sequencing data obtained from various gastrointestinal tract (GIT) samples of 158 patients with gastroesophageal reflux disease. Oral as well as duodenal bacteriomes were similar with bacteriomes of biopsies affected by esophageal adenocarcinoma (EAC), while distances from another esophageal pathologies, such as Barrett’s esophagus and reflux esophagitis, were higher. This result suggests that the oral bacteriome may reflect severe esophageal bacteriome alterations and the duodenogastroesophageal reflux plays a role in the EAC development. The CDI analysis enables robust cross-site comparisons of microbiome compositions and may reveal disease-associated signatures.
The Influence of Visual Features in Scenes of Green and Built Environments on Physiological Markers of Stress
Tadeáš Dvořák
Urban environments are linked to increased risk of stress and mental health issues. Merely viewing images of green and urban spaces can elicit distinct brain responses, yet the underlying visual features remain poorly understood. In this study, 67 participants viewed blocks of urban and natural images during simultaneous fMRI and EEG. Urban scenes evoked greater haemodynamic activity in early (calcarine, lingual gyri) and higher-order (middle occipital) visual areas. Image properties like chromaticity and fractal dimension predicted lingual gyrus activation. These findings suggest that both low- and high-level visual features in urban scenes increase visual processing demands, offering insight into how urban visual environments may contribute to neural mechanisms underlying stress.
Analysis of Aromatic Amines and their Transformation Products in Dust from Five Different Indoor Environments Representing Diverse Sources Profiles
Özge Edebali
Aromatic amines (AAs) are found in a wide range of products, including pharmaceuticals, pesticides, rubber, textiles, hair dyes, and plastic products, and are by-products in tobacco smoke and certain cooking activities. Several AAs and their transformation products are classified as substances of very high concern. However, limited data exist regarding the presence and behavior of AAs in indoor environments, where human exposure may be significant given that people spend over 90% of their time indoors. Indoor dust is a critical environmental matrix in this context for human exposure to various chemicals. This study evaluates the occurrence, composition profiles, and potential sources of emerging AAs, nicotine, and related transformation products within dust across indoor environments with specific AA source profiles.
Capturing Protein Dynamics and Their Determinants Using Explainable Artificial Intelligence
Faraneh Haddadi
We use explainable artificial intelligence (XAI) to analyze molecular dynamics (MD) simulations for protein engineering. We developed networks for the next snapshot prediction of luciferases and prediction whether protein ApoE4 binds a medicine. We developed XAI for their automated and unbiased analysis. We used MD trajectories of image.png for supervised learning, focusing on alpha carbons at 5 ns intervals. Utilizing average and maximum aggregation methods, we derived relevance values per residue across the whole trajectory. We revealed discrepancies between computational B-factors and XAI-derived relevancies. We also calculated residue distances and analyzed binary data to discern the impact of a small molecule, 3-SPA on the dynamics of ApoE4. We identified critically important regions of ApoE4 for binding 3-SPA.
Biomarkers of intertissue dialogue between muscle, adipose and bone tissue
Simona Holotová
Research into inter-tissue communication is transforming our understanding of the adipose–muscle–bone axis, revealing critical biochemical and endocrine interactions among these tissues. Our studies aimed to investigate how adipose and muscle-derived signals influence bone structure and density. Using proteomic and metabolic analyses, we identified key secreted factors, including FSTL1 and FSTL5, that respond to exercise and correlate with bone strength. Additionally, adipose-derived proteins were linked to bone mineral density, emphasizing the metabolic role of fat in bone remodelling. These findings suggest, that bone health is modulated not only by mechanical loading but also by tissue-derived signals influenced by lifestyle. Further research is needed to validate these results and explore therapeutic potential.
Prenatal Exposure to Chemicals and Impact on Immunity in Early Childhood
Maja Klimentić
Many environmental chemicals have been linked with immunomodulation, most significantly poly- and perfluoroalkyl substances (PFAS). Exposure to PFAS begins during prenatal development, when the immune system undergoes rapid development and is especially vulnerable to environmental chemicals. Disturbances in immune cell differentiation or maturation during this period may have life-long implications. The project explores the impact of maternal PFAS exposure on newborns’ immunity by examining immune cell subpopulations in the umbilical cord blood using flow cytometry. Maternal exposure and isolation of cord blood mononuclear cells was performed in a subgroup of mother-newborn pairs from CELSPAC: TNG cohort (Brno, Czech Republic, n=600+ pairs). The poster outlines general flow cytometry methodology, steps in the optimization of method, cell detection and data interpretation.
Foldability Guided Sampling From Protein Evolutionary Structured Latent Spaces Variational Autoencoders
Pavel Kohout
Generative machine learning models show promise for enhancing protein design by using natural language processing techniques to decipher patterns from large protein sequence datasets. Despite their usefulness, understanding how to introduce objective-beneficial and solubilizable mutations remains limited. To this end, we optimized the architecture of variational autoencoders to capture evolutionary dependencies from multiple sequence alignments and defined the foldability score to improve the success rate of generated designs. We generated ancestral variants using machine learning techniques to improve foldability scores by employing different sampling strategies that utilize a foldability-informed latent space. By incorporating foldability scores into our approach, we likely enhanced the generation of viable sequences, as supported by our in silico screening of foldability.
Learning to Engineer Protein Flexibility
Kouba Petr
Generative machine learning models are increasingly being used to design novel proteins. However, their major limitation is the inability to account for protein flexibility, a property crucial for protein function. This challenge stems from the scarcity and heterogeneity of relevant data. We address this in three steps. First, we benchmark methods for evaluating protein flexibility and identify useful data sources. Second, we develop flexibility predictors using a pre-trained protein language model, leveraging either sequence alone or in combination with structure. Third, we propose Flexpert-Design, a method for fine-tuning inverse folding models to steer designs toward desired flexibility at specific regions. Flexpert-Design enables control over structural dynamics, offering a new direction for protein engineering beyond static backbone assumptions.
In Silico Prioritization of Water Micropollutants for Quantification of Their Contribution to the Aryl Hydrocarbon Receptor (AhR)-Mediated Activity
Katarína Lörinczová
Risk assessment of environmental mixtures is challenging, as diverse compounds can alter multiple biological pathways. Identifying effect-drivers and their contribution requires experimental data, often limited or unavailable. In silico approaches like quantitative structure-activity relationship (QSAR) models can address data gaps by predicting toxic potential of data-poor compounds. In this study a QSAR model was developed based on random forest algorithm and ToxCast database to predict aryl hydrocarbon receptor (AhR) activity of compounds from Danube River samples. With high predictive performance (weighted overall accuracy: 77-86.9%), the model classified compounds into four categories and prioritized them for CAFLUX bioassay testing, explained part of the total site-specific activity (by up to 11%), and identified novel AhR-active water pollutants.
Safer and More Effective Thrombolytics: Brnoteplase Through Rational Mutagenesis
Jan Mičan
Intravenous thrombolytics used in ischemic stroke treatment achieve only 10-40% recanalization rate (depending on clot size) and have dangerous side effects. In the STROKE Brno consortium, we designed variants of thrombolytic alteplase using structural bioinformatics and combining mutations from literature. We tested their plasminogen activation, fibrinolysis, and inhibition resistance. Selected variants were tested on biophysical arterial occlusion flow and in vivo rat stroke models. The most promising mutant, Brnoteplase, shows 400% increased inhibition resistance compared to alteplase, 80-fold higher fibrin selectivity, and 400% lower fibrin binding. In the rat ischemic stroke model, Brnoteplase has exhibited comparable recanalization to tenecteplase, less significant hemorrhages, and less post-stroke brain assymetry. These properties make Brnoteplase a good candidate for development and clinical use.
Streptococcus to Prevotella Ratio in Inflammatory Changed Esophageal Tissues a Metagenomic Observational Study
Natálie Mlčůchová
The altered Streptococcus to Prevotella (S:P) ratio is considered as an indicator of dysbiosis in the human esophagus. Here we aimed to determine this S:P ratio and the total bacterial DNA content in paired esophageal tissues from patients with reflux esophagitis (RE), Barrett's esophagus (BE), and esophageal adenocarcinoma (EAC). Total bacterial content in pathological tissues was higher than in adjacent esophageal tissues in all study groups (p < 0.01). A significantly lower S:P ratio was found in pathological tissue compared to adjacent tissue only in patients with RE (p < 0.001); however, the medians S:P ratio were similar among inflammatory changed esophageal tissues regardless of the diagnosis. Acknowledgements: SALVAGE (CZ.02.01.01/00/22_008/0004644), AZV GERD (NU20-03-00126), TAČR (FW02020042), RECETOX Research Infrastructure (ID LM2023069).
BISPHENOL EXPOSURE in the Czech Population Before and After Legislative Changes
Daria Sapunova
Bisphenol A (BPA) and its replacements, BPS and BPF, are endocrine-disrupting chemicals widely used in plastics. This study compares Czech cohorts sampled in 2011–2012 and 2019–2020 to assess urinary bisphenol levels. We explored whether BPA decreased and BPS/BPF increased following legislative changes during this period. BPA levels significantly declined (−35.7% in adults, −18.5% in children), while BPS levels rose (+53.8% in adults, +22.5% in children). BPA concentrations were consistently higher in children. All BPA values exceeded the health-based guideline, and a small percentage of BPS values exceeded its limit. These findings suggest that regulation may influence exposure trends. Continued human biomonitoring is crucial for evaluating public health risks and informing regulatory decisions.
Automated and Derivatization Free GC-MS Analysis of SCFA in Human Biospecimen
Hana Seličová
Accurate quantification of short- and branched-chain fatty acids (S&BCFA) is critical for health and disease studies. We present a fully automated GC-MS assay integrating robotic sample preparation and MS data acquisition. Serum/plasma samples are extracted via liquid-liquid extraction using Thermo TriPlus RSH and analyzed by single quadrupole GC-MS. Extracts (4 µL) are injected into a PTV inlet and separated on an RTX-WAX column; electron ionization (70 eV) and FS-SIM mode (40–250 m/z) target ≥2 ions per analyte. Nine S&BCFAs are quantified with an 18-minute sample-to-sample runtime, enabling 80 samples/day. Key features include enhanced analyte loading, native S&BCFA detection, high sensitivity volatile profiling, and in-line automation, reducing runtime, labor, and errors while improving reproducibility for clinical routine use.
Functional Characterisation of the Luminous Apparatus of the Sea Pen Pennatula Phosphorea
Martin Sitte
Bioluminescence is the production of visible light by living organisms via oxidation of luciferin substrates, catalysed by luciferase enzymes. Auxiliary proteins, such as fluorescent proteins and coelenterazine-binding proteins, can modulate emission spectra or stabilise reactive luciferins. Bioluminescent organisms offer diverse light-emitting enzymes and photoproteins with great bioengineering potential, from biosensors to zero-electricity lighting. However, only a few systems are well characterised. Among anthozoans, Renilla-type bioluminescence is best understood, involving a coelenterazine-dependent luciferase, a calcium-regulated coelenterazine-binding protein, and a green fluorescent protein.
Recently, transcriptomic analyses identified putative luciferase and auxiliary protein sequences in the sea pen Pennatula phosphorea. Here, we biochemically reconstitute and functionally characterise the Pennatula-encoded luciferase, illuminating its emission properties and advancing the understanding of cnidarian bioluminescence.
Seasonal Impacts on Concentrations of Flame Retardants in Car Air and Dust
Petra Svobodová
Car interiors are unique microenvironments influenced by strict flammability requirements and environmental variables like temperature, which affect chemical emissions from materials. This study investigates seasonal differences in flame retardant levels in air, dust, and materials inside a vehicle. We hypothesized that concentrations of deca-BDE and organophosphate esters (TDCIPP, TCIPP, TCEP) would be higher in summer due to increased emissions. To test this, repeated sampling was conducted in one car during winter (≤10 °C) and summer (~30 °C). Samples included indoor air, settled dust, and various materials (seat foam, fibres, plastics). Results showed significantly elevated flame retardant levels in summer air, with polyurethane foam containing the highest material concentrations. These findings highlight strong temperature-driven emission patterns and potential exposure pathways in vehicles.
Endocrine Disruptors in Stagnant Waters with Mass Occurence of Water Blooms
Jaroslava Večerková
This study aimed to assess the endocrine-disrupting potential of samples from stagnant waters with the presence of water blooms at selected locations in the Czech Republic and to compare the observed biological activities of water samples with the concentrations of selected substances that may contribute to the endocrine-disrupting potential.
Water samples were tested using in vitro bioassays to evaluate total estrogenic and retinoid-like activity, while concentrations of specific compounds—including estrogens, sterols, flavonoids, retinoids, and microcystins—were determined using LC-MS/MS. The bioassays, together with chemical analyses, indicated the presence of multiple types of endocrine-disrupting substances in stagnant waters with the occurrence of water blooms. In addition to known cyanotoxins, the presence of estrogenic and retinoid substances was also detected.
Exposure to PFAS in Firefighters and its Associations with Biomarkers of Thyroid Function
Petr Gregor
This study investigates the association between PFAS exposure and thyroid function in male firefighters - novices (n=58) and professional (n=52). Serum levels of twelve PFAS compounds and five thyroid hormones (TSH, T3, T4, FT3 and FT4) were measured and Generalized Linear Model (GLM) with a Gamma distribution (log link) was applied what resulted in significant associations between PFUnDA and PFDA with FT3 and T3, particularly in newly recruited firefighters post-training, We also conducted Quantile G-Computation (QGcomp) which supported that PFUnDA and PFDA exhibited positive associations with FT3 and T3 in novices post-training. These results underscore the potential thyroid-disrupting effects of PFAS exposure, highlighting the need for continued monitoring of occupational health risks in high-exposure groups like firefighters.