Abstract The European Commission asked EFSA to update its 2018 risk assessment on polychlorinated dibenzo‐p‐dioxins and dibenzofurans (PCDD/Fs) and dioxin‐like polychlorinated biphenyls (DL‐PCBs) in feed and food, based on the 2022 WHO Toxic Equivalency Factors (WHO2022‐TEFs). A decrease in sperm concentrations as a consequence of early‐life exposure was still considered the critical effect in rodents and humans. This can only be prevented by ensuring sufficiently low exposure to PCDD/Fs and DL‐PCBs of future mothers. Using the WHO2022‐TEFs, the critical human study in the 2018 Opinion still showed a significant inverse association between PCDD/Fs and sperm concentration, but not when including DL‐PCBs. Therefore, this study was not used to derive a Tolerable Weekly intake (TWI). Instead, a developmental study with TCDD in rats was used to derive a TWI of 0.6 pg. WHO2022‐toxic equivalents (TEQ)/kg bw per week, based on decreased sperm production. Human studies were used as supporting evidence, including an evaluation of necessary uncertainty factors. Dietary exposure assessment based on occurrence data in food from the last decade, showed that using the new TEFs resulted in 27%–35% lower total‐TEQ exposure. Mean exposure (lower to upper bound) to PCDD/Fs and DL‐PCBs for European adult populations varied between 1.87 and 6.10 pg. WHO2022‐TEQ/kg bw per week, and at the P95 between 4.15 and 12.0 pg. WHO2022‐TEQ/kg bw per week. The CONTAM Panel concluded that exposure for European women of childbearing age raises a health concern for their sons at the mean (80%–90% certainty) and P95 (95%–99% certainty) exposure. Including uncertainties regarding the TEFs for DL‐PCBs decreased this certainty. Human milk data confirmed the exceedance of the TWI but to a lesser extent than dietary exposure. The change in TEFs does not affect the transfer rates of individual congeners but it could impact the transfer rates when using Total‐TEQ levels.

Abstract Zinc is an essential trace element involved in numerous biological processes, including cellular signalling, development, and reproduction. Zinc homeostasis is regulated by zinc transporters, yet the physiological roles of many transporters remain poorly understood in vivo . Here, we investigated the function of the zinc transporter ZIP9 (SLC39A9) using a zebrafish ( Danio rerio ) knockout model. Elemental imaging using laser ablation inductively coupled plasma mass spectrometry (LA-ICP-MS) revealed altered zinc distribution in zip9 -deficient larvae. Synchrotron-based X-ray fluorescence (XRF) imaging further showed reduced zinc levels in the brain region of mutant zebrafish. Consistent with these observations, loss of zip9 was associated with altered expression of key neuroendocrine genes within the hypothalamic-pituitary-gonadal (HPG) axis. Zip9 mutant females exhibited disrupted ovarian follicle development, reduced spawning rates, and decreased egg production. In addition, embryos derived from zip9 mutant parents displayed reduced size, impaired early development, and decreased survival. Together, these findings identify ZIP9 as a regulator of zinc distribution in vivo and suggest that ZIP9-mediated zinc signalling contributes to reproductive regulation in zebrafish.

The re-evaluation of feed additives in the European Union was outlined in the Regulation (EC) No 1831/2003. The primary objectives were to ensure safety and verify efficacy, as well as to update authorizations and phase out unsafe additives or those in which the industry expressed no interest. The scientific component of the re-evaluation was assigned to the European Food Safety Authority (EFSA). A total of 35 compounds of eight trace elements (iron, zinc, manganese, copper, iodine, selenium, cobalt, and molybdenum) were re-evaluated. Between 2012 and 2019, the EFSA released 33 scientific opinions concerning generic aspects of trace elements or specific compounds. Characterization, safety, and efficacy were assessed. These opinions and the nine authorizing regulations form the basis of the present overview. This article summarizes the procedures and key outcomes of the re-evaluation of trace elements as feed additives in the EU, and highlights the policy decisions adopted by risk managers. © 2026 The Author(s). Journal of the Science of Food and Agriculture published by John Wiley & Sons Ltd on behalf of Society of Chemical Industry.

Abstract The European Commission asked the European Food Safety Authority (EFSA) to assess the risk related to the presence of plant lectins in food. Based on the available evidence, the CONTAM Panel considered only phytohaemagglutinin (PHA), a legume lectin from beans (Phaseolus sp.), for the risk characterisation. Effects of PHA in the small intestine were considered as the critical effect in subacute studies in rats. A lower confidence limit of the benchmark dose (BMDL)10 of 22.9 mg/kg body weight (bw) per day for an increase in small intestine dry weight was selected as the most appropriate reference point for the risk characterisation. The establishment of a health‐based guidance value for PHA was not considered appropriate due to the limitations and uncertainties in the current toxicological evidence, and the margin of exposure (MOE) approach was used for the risk characterisation. The Panel considered that acute exposure resulting in MOEs above 100 is not expected to raise a health concern. As no occurrence data were submitted to EFSA, data for PHA presence in food were identified by a literature search. An arbitrary acute exposure scenario, where only 50% of the lectins are inactivated due to insufficient cooking of food containing lectins (e.g. beans), would result in MOEs below 100. The Panel, accounting for the uncertainties affecting the exposure and hazard assessments, concluded with at least 95% probability that such a dietary exposure would raise health concerns. The Panel also noted that exposure to completely deactivated lectins in food prepared following adequate food processing practices (e.g. soaking and boiling) would not raise health concerns. No risk characterisation could be performed for other lectins due to the lack of relevant toxicological data and/or in some cases lack of occurrence data.

Copper (Cu) and zinc (Zn) are essential trace metal elements for vertebrates, and insufficient intake of trace metal elements can have adverse effects on animal health. Studies suggested that the interaction among trace metal elements played a crucial role in this process. Therefore, the current study was conducted to explore whether low Zn alleviated low dietary Cu-induced decrease in intestinal Cu level and identify potential mechanisms via the in vivo animal experiment and in vitro cell culture. The regulatory relationship between metal response element binding transcription factor 1 (Mtf-1) and ATPase Cu transporting alpha (Atp7a) under low Cu and low Zn was elucidated through electrophoretic mobility shift assays and chromatin immunoprecipitation analysis. The SUMOylation modification of Mtf-1 and its effects on atp7a promoter were determined through immunoprecipitation and dual-luciferase reporter assay. Compared with the control, the low dietary Cu group significantly promoted the transcriptional regulation of atp7a promoter by Mtf-1, leading to intestinal Cu metabolism disorders. However, compared with the low dietary Cu group, the low Cu + low Zn group had significantly inhibited the transcriptional regulation of atp7a promoter by Mtf-1 and alleviated intestinal Cu metabolism disorder caused by low Cu. Mechanistically, Mtf-1 was modified by SUMOylation, and Sumo1-overexpression significantly reduced atp7a promoter activity. Taken together, dietary low Zn alleviated low Cu-induced decrease in intestinal Cu level through the SUMOylated-Mtf-1 and the Mtf-1/Atp7a axis.

Abstract The European Commission asked EFSA to update its 2012 risk assessment on Emerging and Novel brominated flame retardants (BFRs) in food. Since the previous Opinion, further information was collected for a total of 27 BFRs. The CONTAM Panel concluded that TDBPP, DBPNG and TBNPA are considered genotoxic and carcinogenic. Based on positive in vitro mutagenicity in mammalian cells, DBE‐DBCH is considered genotoxic. EBTEBPI, DBDPE, PBB‐Acr and TTBP‐TAZ are not genotoxic. Insufficient or absent data precluded the Panel to conclude on the genotoxic potential of the remaining BFRs. Reproductive/developmental toxicity and/or neurotoxicity/neurodevelopmental effects were reported for DBNPG, BEH‐TEBP, EH‐TBB, DBDPE and TDBP‐TAZTO. Reference Points were identified for TDBPP, DBNPG, BEH‐TEBP, EH‐TBB, DBDPE and TDBP‐TAZTO. No health‐based guidance values could be derived for any of the BFRs. Occurrence data were submitted to EFSA for eight BFRs (BEH‐TEBP, EH‐TBB, BTBPE, DBDPE, HBB, PBEB, PBT, TBX). Dietary exposure could be estimated only for BTBPE and HBB, due to the lack of quantitative data for the other BFRs. However, no risk characterisation could be performed for these BFRs due to the lack of a Reference Point. Using the limited available information on dietary exposure estimates reported in the scientific literature, risk characterisation could be performed for BEH‐TEBP, EH‐TBB and DBDPE, and the CONTAM Panel concluded that dietary exposure does not raise a health concern. The Panel considered that the level of certainty was 90% or higher for most of the reported conclusions. Very limited or no information on hazard or exposure were available for HBCYD, TBCO, DBHCTD, HCTBPH, BDBP‐TAZTO, DBP‐TAZTO, OBTMPI, DBS, HEEHP‐TEBP, 4’‐PeBPO‐BDE208 or TTBNPP. Recommendations were formulated to address data gaps, including the need for access to original data, and for the collection of biomonitoring data, occurrence data in food, genotoxicity information and toxicological studies to support the identification of a Reference Point.

The global demand for sustainable protein sources has accelerated the need for intelligent tools that can rapidly process and synthesise domain-specific scientific knowledge. In this study, we present a proof-of-concept multi-agent Artificial Intelligence (AI) framework designed to support sustainable protein production research, with an initial focus on microbial protein sources. Our Retrieval-Augmented Generation (RAG)-oriented system consists of two GPT-based LLM agents: (1) a literature search agent that retrieves relevant scientific literature on microbial protein production for a specified microbial strain, and (2) an information extraction agent that processes the retrieved content to extract relevant biological and chemical information. Two parallel methodologies, fine-tuning and prompt engineering, were explored for agent optimisation. Both methods demonstrated effectiveness at improving the performance of the information extraction agent in terms of transformer-based cosine similarity scores between obtained and ideal outputs. Mean cosine similarity scores were increased by up to 25

The global demand for sustainable protein sources has accelerated the need for intelligent tools that can rapidly process and synthesise domain-specific scientific knowledge. In this study, we present a multi-agent Artificial Intelligence (AI) framework designed to support sustainable protein production research, with an initial focus on microbial protein fermentation using side-streams which enables waste carbon capture and utilisation (CCU). Our Retrieval-Augmented Generation (RAG)-oriented system consists of two GPT-based LLM agents: (1) a literature search agent that retrieves relevant scientific literature on microbial protein production for a specified microbial strain, and (2) an information extraction agent that processes the retrieved content to extract relevant biological and chemical information. Two parallel methodologies, fine-tuning and prompt engineering, were explored for agent optimisation. Both methods demonstrated effectiveness at improving the performance of the information extraction agent in terms of transformer-based cosine similarity scores between obtained and ideal outputs. Mean cosine similarity scores were increased by up to 25 %, while universally reaching mean scores of ≥0.89 against ideal output text. Fine-tuning overall improved the mean scores to a greater extent (consistently of ≥0.94) compared to prompt engineering, although lower statistical uncertainties were observed with the latter approach. A user interface was developed and published for enabling the use of the multi-agent AI system, alongside preliminary exploration of additional chemical safety-based search capabilities relevant to CCU-integrated sustainable protein production.

Geometric deep learning is an emerging technique in Artificial Intelligence (AI) driven cheminformatics, however the unique implications of different Graph Neural Network (GNN) architectures are poorly explored, for this space. This study compared performances of Graph Convolutional Networks (GCNs), Graph Attention Networks (GATs) and Graph Isomorphism Networks (GINs), applied to 7 different toxicological assay datasets of varying data abundance and endpoint, to perform binary classification of assay activation. Following pre-processing of molecular graphs, enforcement of class-balance and stratification of all datasets across 5 folds, Bayesian optimisations were carried out, for each GNN applied to each assay dataset (resulting in 21 unique Bayesian optimisations). Optimised GNNs performed at Area Under the Curve (AUC) scores ranging from 0.728-0.849 (averaged across all folds), naturally varying between specific assays and GNNs. GINs were found to consistently outperform GCNs and GATs, for the top 5 of 7 most data-abundant toxicological assays. GATs however significantly outperformed over the remaining 2 most data-scarce assays. This indicates that GINs are a more optimal architecture for data-abundant environments, whereas GATs are a more optimal architecture for data-scarce environments. Subsequent analysis of the explored higher-dimensional hyperparameter spaces, as well as optimised hyperparameter states, found that GCNs and GATs reached measurably closer optimised states with each other, compared to GINs, further indicating the unique nature of GINs as a GNN algorithm.

Abstract The European Commission asked EFSA to update its 2014 risk assessment on perchlorate in food. Perchlorate is a contaminant of both natural and anthropogenic sources present in food and drinking water. It is a substrate for the sodium iodide symporter (NIS) and competitively inhibits the uptake of iodide into the thyroid. Experimental animal studies show that perchlorate exposure during pregnancy can result in neurodevelopmental toxicity. The CONTAM Panel established a tolerable daily intake of 1.4 μg/kg body weight per day, based on the inhibition of thyroid iodine uptake in healthy adults. The tolerable daily intake (TDI) takes into account the sensitivity of the fetus to maternal thyroid hormone disturbance and uncertainty around the impact of iodine deficiency on the effects of perchlorate during fetal development. This TDI is applicable for both a short‐term (approximately 2‐week period) and chronic exposures based on the mode of action of perchlorate, its toxicokinetic properties and the key study used to derive the TDI. An acute reference dose (ARfD) was not deemed necessary. EFSA received a total of 40,356 analytical results, between 2016 and 2022, which were considered for the dietary exposure assessments. A chronic dietary exposure assessment for all age groups and a short‐term dietary exposure assessment for pregnant women were calculated. The CONTAM Panel concluded that chronic and short‐term dietary exposure estimates to perchlorate were below the TDI for all age groups including pregnant women, with the exception at the upper bound of the P95 for infants, breastfed infants and formula‐fed infants. Even if the limitations in analytical methods, leading to a large difference between lower bound (LB) and upper bound (UB) dietary exposure, reduces the certainty in this conclusion for infants, breastfed infants and formula‐fed infants, the uncertainty analysis indicates a higher (above 50%) likelihood of ‘no concern’ for all scenarios.

Novel Quantitative Structure-Activity Relationship (QSAR) models were constructed using Graph Convolutional Networks (GCNs), to predict Drug-Induced Liver Injury (DILI), Drug-Induced Renal Injury (DIRI) and Drug-Induced Cardiotoxicity (DICT) of Selective Androgen Receptor Modulators (SARMs) – an emerging class of performance-enhancing drugs. Prior to training on DILI, DIRI and DICT datasets, the GCN QSAR models were first pre-trained on a variety of unrelated biomedical assay datasets, as an attempt to improve model performance via transfer learning. The success of the transfer learning was mixed; model performances were measurably improved via pre-training on certain datasets, by statistically weak increases. The optimal final QSAR models achieved overall accuracy scores of 68% for DILI (no significant improvement via ensemble modelling), 76% for DIRI (improved to 77% via ensemble modelling) and 65% for DICT (improved to 67% via ensemble modelling). Application of the most optimal singular models to a dataset of 25 SARMs predicted that 21 of the 25 SARMs are either DILI-positive, DIRI-positive, or both – which raises concern, given the rising use of SARMs. All SARMs except for one were predicted as DICT-negative. A novel definition of the Applicability Domain (AD) was used, intended for close relevance to the models, via generating three-dimensional graph embeddings, for each model. Convex hulls were fitted around training data embeddings, with a ±10% buffer, defining the AD as the region of embedded chemical space covered by the convex hull, for a given model. Subsequent analysis found that a majority of DILI, DIRI and DICT testing data lay within the AD, alongside a majority of the SARMs – adding consensus to the reliability of the predictions. ### Competing Interest Statement The authors have declared no competing interest. Biotechnology and Biological Sciences Research Council, BB/T008709/1 Food Standards Agency, https://ror.org/05p20a626, FS900120

Abstract The European Food Safety Authority (EFSA) was asked to deliver a scientific opinion on the derivation of a health‐based guidance value (HBGV) for Δ8‐tetrahydrocannabinol (Δ8‐THC) in food with an assessment of the occurrence of Δ8‐THC and the co‐occurrence with Δ9‐THC in hemp and hemp‐derived products. Data from a clinical study were used to inform on the relative potency between Δ9‐THC and Δ8‐THC. The point estimate of the relative potency (ratio Δ9‐THC/Δ8‐THC) was in the range between 1 and 1.4, with 95% confidence between 0.97 and 1.63. Based on this range, the CONTAM Panel, using a conservative approach, set a relative potency factor of one for Δ8‐THC. The CONTAM Panel had previously set an acute reference dose (ARfD) of 1 μg/kg body weight for Δ9‐THC derived from adverse effects on human central nervous system (CNS). Given the similarity in the effects and the mode of action between Δ8‐THC and Δ9‐THC, the Panel considered that the established ARfD can be considered as a group ARfD for the sum of Δ8‐ and Δ9 THC. Regarding the occurrence of Δ8‐THC, the majority of samples were left censored, in particular for hemp infusion leaves, hemp seed oil and hemp seeds (96%–99%). Highest detection rates and levels were found in the categories ‘Sugar and similar, confectionery and water‐based sweet desserts’ and ‘Products for non‐standard diets, food imitates and food supplements’. Of 1145 samples, both substances were detected together in only 96 samples. If the two substances are produced naturally, a Δ8‐THC to Δ9‐THC ratio below 1 is expected; however, many of the samples, positive for Δ8‐THC, were above this ratio indicating either addition of semi‐synthetic Δ8‐THC, formation during processing or enrichment of the natural Δ8‐THC.

Zinc and several physiologically relevant ligands of the aryl hydrocarbon receptor (AHR) are nutrients that promote intestinal barrier function. We have identified that AHR activation upregulates the expression of zinc importers in the intestinal epithelium to increase intracellular zinc concentrations, which leads to improved epithelial barrier function. Here, we investigated if an amino acid chelate of zinc, in cooperation with AHR activation, can improve the barrier function of a differentiated Caco-2 cell epithelium. Functional assays of the Caco-2 cell epithelium demonstrate that both ZnSO4 and a lysine and glutamic acid chelate of Zn, in combination with the physiological AHR agonist 6-formylindolo[3,2-b]carbazole (FICZ), increase expression of tight junction proteins at the mRNA and protein levels. FICZ increases uptake of zinc into the epithelium in the presence of ZnSO4 or the amino acid Zn chelate in the medium to equal extents. We conclude that the lysine and glutamic acid chelate of Zn is as efficacious as ZnSO4 in reducing permeability of the Caco-2 cell epithelium in the presence of FICZ. The results suggest that dietary supplementation with bioavailable forms of zinc together with nutritional AHR agonists may be beneficial in improving gut barrier function and help prevent inflammatory bowel disease (IBD).

The SLC39A (ZIP) family of zinc ion transporters play a pivotal role in maintaining zinc homeostasis, which is essential for numerous physiological processes involving enzyme catalysis, protein structure and regulation in signal transduction. This investigation employed AlphaFold3 to predict and analyze the 3D structures of all 14 human ZIP family members and revealed key structural and functional features, including transmembrane domains with eight alpha-helices, extracellular and cytoplasmic domains, dimerization, and zinc ion transport pathways. Unique zinc-binding motifs—composed of histidine, aspartic acid, and glutamic acid—were identified. They facilitate zinc ion attraction, selection, and transport. The findings highlight significant structural diversity in these proteins, with additional alpha-helices, disulfide bonds, and other conserved motifs that together contribute to functional specialization across the ZIP family members. Compared to predictions, which exist only for ZIP4, the models incorporate dimeric structures, rationalize loop conformations, and achieve a higher resolution. The predicted 3D structures offer enhanced insights into zinc ion transport mechanisms and provide a foundation for future research into the structural biology of these proteins and their interacting partners in physiology and pathology. ### Competing Interest Statement The authors have declared no competing interest. * AF : AlphaFold ECD : extracellular domain pTM : predicted TM-score ipTM : interface pTM-score HB : helix bundle HRD : histidine-rich domain ICD : cytoplasmic domain IL2 : second intracellular loop Men : entrance site Mex : exit site PCD : PAL-containing domain TM : transmembrane TMD : transmembrane domain ZIP : Zrt,Irt-like protein King's-China Scholarship Council PhD Scholarship programme

Introduction: Zinc (Zn) deficiency has been described not only on general human health but also within the sports context −as negatively affecting performance–. Thus, Zn status assessment is of great interest for athletes, especially in order to correct deficiency states of this mineral. Objective : The overall objective of this work was to assess Zn status in professional handball players during the competitive period (through plasma levels, dietary intake and gene expression of the Zn transporters), as well as to determine the effect of Zn supplementation. Methods : A total of twenty-two participants were recruited, −twelve belonged to the Control Group (CG) and ten male handball players comprised the experimental group (ATH-G)−, being monitored over a 2-month period with 2 evaluation moments: baseline (i.e., initial conditions) and follow-up (i.e., after 8 weeks of training and competition). Zn intake, plasma Zn levels, and gene expression of Zn transporters were obtained. Results : Plasma Zn levels were higher in ATH-G than in CG at the end of Zn intervention ( p ≤0.010). Moreover, differences in the gene expression profile of Zn transporters were observed in ATH-G −with the down-regulation of several Zn transporters−, compared to the CG at baseline ( p ≤0.05). Likewise, Zn intervention modified the expression of Zn transporters in ATH-G at 8 weeks (all, p ≤0.001) −ZnT2, ZnT5, ZIP3, ZIP5, ZIP11, ZIP13 and ZIP14 transporters being up-regulated−. Conclusion: Handball players seemed to have different nutritional needs for Zn, with differences in the gene expression of Zn transporters compared to controls. Zn intervention, in our athletes, modified the expression of Zn transporters, so we could deduce that Zn transporters up-regulation may have increased to mobilise Zn at the cellular level at 8 weeks of Zn intervention.

The European Commission asked EFSA to update its 2009 risk assessment on arsenic in food carrying out a hazard assessment of inorganic arsenic (iAs) and using the revised exposure assessment issued by EFSA in 2021. Epidemiological studies show that the chronic intake of iAs via diet and/or drinking water is associated with increased risk of several adverse outcomes including cancers of the skin, bladder and lung. The CONTAM Panel used the benchmark dose lower confidence limit based on a benchmark response (BMR) of 5% (relative increase of the background incidence after adjustment for confounders, BMDL05) of 0.06 μg iAs/kg bw per day obtained from a study on skin cancer as a Reference Point (RP). Inorganic As is a genotoxic carcinogen with additional epigenetic effects and the CONTAM Panel applied a margin of exposure (MOE) approach for the risk characterisation. In adults, the MOEs are low (range between 2 and 0.4 for mean consumers and between 0.9 and 0.2 at the 95th percentile exposure, respectively) and as such raise a health concern despite the uncertainties.

2-ethylhexyl diphenyl phosphate (EHDPP) strongly binds to transthyretin (TTR) and affects the expression of genes involved in the thyroid hormone (TH) pathway in vitro. However, it is still unknown whether EHDPP induces endocrine disruption of THs in vivo. In this study, zebrafish (Danio rerio) embryos (< 2 h post-fertilization (hpf)) were exposed to environmentally relevant concentrations of EHDPP (0, 0.1, 1, 10, and 100 mu gL-1) for 120 h. EHDPP was detected in 120 hpf larvae at concentrations of 0.06, 0.15, 3.71, and 59.77 mu gg(-1) dry weight in the 0.1, 1, 10, and 100 mu gL-1 exposure groups, respectively. Zebrafish development and growth were inhibited by EHDPP, as indicated by the increased malformation rate, decreased survival rate, and shortened body length. Exposure to lower concentrations of EHDPP (0.1 and 1 mu gL-1) significantly decreased the whole-body thyroxine (T4) and triiodothyronine (T3) levels and altered the expressions of genes and proteins involved in the hypothalamic-pituitary-thyroid axis. Downregulation of genes related to TH synthesis (nis and tg) and TH metabolism (dio1 and dio2) may be partially responsible for the decreased T4 and T3 levels, respectively. EHDPP exposure also significantly increased the transcription of genes involved in thyroid development (nkx2.1 and pax8), which may stimulate the growth of thyroid primordium to compensate for hypothyroidism. Moreover, EHDPP exposure significantly decreased the gene and protein expression of the transport protein transthyretin (TTR) in a concentration-dependent manner, suggesting a significant inhibitory effect of EHDPP on TTR. Molecular docking results showed that EHDPP and T4 partly share the same mode of action of binding to the TTR protein, which might result in decreased T4 transport due to the binding of EHDPP to the TTR protein. Taken together, our findings indicate that EHDPP can cause TH disruption in zebrafish and help elucidate the mechanisms underlying EHDPP toxicity.