bims-ainimu 2025-10-26 papers

Issue of 2025–10–26
three papers selected by
Pedro Escoll Guerrero, Institut Pasteur

Microbiol Spectr. 2025 Oct 20. e0078325

Propionic acid toxicity and utilization of α-ketobutyric acid in Neisseria meningitidis via the methylcitrate cycle under specific conditions.

Adelfia Talà, Matteo Calcagnile, Silvia Caterina Resta, Salvatore Maurizio Tredici, Giuseppe Egidio De Benedetto, Cecilia Bucci, Pietro Alifano.

Neisseria meningitidis is a human-specific, transient colonizer of the nasopharynx that occasionally causes invasive disease. It can utilize a limited range of compounds as primary carbon sources, including glucose, maltose, lactate, and pyruvate, which are present in varying concentrations in microenvironments relevant to meningococcal infection. Additionally, intermediates from the tricarboxylic acid cycle, such as succinate, fumarate, and malate, as well as amino acids like glutamate, are utilized as supplementary carbon sources. Notably, N. meningitidis also possesses a functional methylcitrate cycle (MCC), which enables the assimilation of propionic acid and mitigates its toxicity. In this study, we investigated propionate toxicity and MCC functionality in wild-type N. meningitidis strains and prpB-, prpC-, ackA1-, and ackA2-defective mutants under various growth conditions. We observed that propionate toxicity was influenced by the primary carbon source and additional factors, such as bicarbonate. Specifically, prpB- and prpC-defective mutants showed high sensitivity to propionate when cultured with glucose or pyruvate, but were not inhibited even by high concentrations of propionate when grown with lactate. The mechanisms underlying the conditional toxicity of propionate were further explored and discussed. Additionally, in the genome of 41 out of 128 N. meningitidis strains, we identified a gene encoding a transporter from the 4-toluene sulfonate uptake permease family, located between prpC and acnD in the MCC gene cluster. Genetic inactivation of this gene, named kbuT, impaired the ability to take up and oxidize α-ketobutyrate, an α-keto acid abundant in host cells, which can be used as a carbon source through the MCC.
IMPORTANCE: Meningococci are metabolically versatile organisms, switching between intracellular and extracellular lifestyle during colonization and invasive disease. Niche switching impacts on how bacteria communicate with host to find a balance between nutrient assimilation and protection against toxicity of some metabolites. The methylcitrate pathway fulfills this function, providing a compromise between propionate assimilation and propionate detoxification, in relation to the colonized host microenvironments. In this study, we revealed an unexpected difference in the sensitivity of meningococci to propionate when grown with different carbon sources. We also characterized the function of a gene located within the prp operon that encodes a transporter of α-ketobutyrate, an α-ketoacid abundant in host cells. These results contribute to extending our understanding of the metabolic adaptation mechanisms, which are crucial for meningococcal infection and virulence within the host microenvironments.

Keywords: Neisseria meningitidis; methylcitrate cycle; propionic acid toxicity; α-ketobutyric acid metabolism

DOI: https://doi.org/10.1128/spectrum.00783-25

Front Plant Sci. 2025 ;16 1677066

Combining AI and new genomic techniques to 'fine-tune' plants: challenges in risk assessment.

Matthias Juhas, Bernd Rodekohr, Andreas Bauer-Panskus, Christoph Then.

Using new genomic techniques (NGTs) to 'fine-tune' plants typically involves changing just a small number of nucleotides. These small interventions can, nevertheless, lead to effects that go beyond the known plant characteristics, caused by genotypes previously unknown in the breeders' gene pool. The EU is currently discussing a proposal for the future regulation of NGT plants. In essence, the European Commission is proposing that NGT plants with less than 20 deletions, insertions or substitutions should in future no longer undergo mandatory risk assessment. NGT plants up to this threshold would be classified as Category 1 NGT, and therefore treated as equivalent to conventionally-bred plants. Plants in this category would not be subject to mandatory environmental risk assessment. The question thus arises of whether any of these Category1 NGT plants considered, in fact, have novel environmentally hazardous characteristics. Based on our findings from horizon scanning and to exemplify regulatory challenges, we used publicly available generative AI with the aim to design 'fine-tuned' NGT plants that would very likely require environmental risk assessment, but would nevertheless meet the specific the criteria for Category 1 NGT plants. As a proof of principle, we designed a genetic blueprint for an insecticidal maize plant, which could subsequently be developed using NGTs. There are several reasons why these insecticidal NGT plants should be subject to environmental risk assessment prior to being approved for cultivation. For example, they could be toxic to non-target species, cause resistance in pest insects, or show unintended genetic and phenotypic changes. In summary, there is no scientifically justifiable threshold of a certain number of mutations up to which NGT effects could be assumed to be of the same category as conventionally bred plants. Therefore, it is essential that the future regulatory concept is not based on such thresholds. On the contrary, future regulation should be science based and include case-by-case and step-by-step risk assessment, traceability and monitoring requirements to secure the future of food production and to protect biodiversity.

Keywords: GMO regulation; artificial intelligence; cis-regulatory elements; genetic engineering; genetically engineered plants; genome editing; new genomic techniques (NGT); risk assessment

DOI: https://doi.org/10.3389/fpls.2025.1677066

Sci Rep. 2025 Oct 21. 15(1): 36761

Isometric representations in neural networks improve robustness.

Kosio Beshkov, Jonas Verhellen, Mikkel Elle Lepperød.

Artificial and biological agents are unable to learn given completely random and unstructured data. The structure of data is encoded in the distance or similarity relationships between data points. In the context of neural networks, the neuronal activity within a layer forms a representation reflecting the transformation that the layer implements on its inputs. In order to utilize the structure in the data in a truthful manner, such representations should reflect the input distances and thus be continuous and isometric. Supporting this statement, findings in neuroscience propose that generalization and robustness are tied to neural representations being continuously differentiable. Furthermore, representations of objects have the capacity of being hierarchical. Combined together, these two conditions imply that neural networks need to both preserve the distances between inputs as well as have the capacity to apply cuts at different resolutions, corresponding to different levels of a hierarchy. During cross-entropy classification, the metric and structural properties of network representations are usually broken both between and within classes. To achieve and study this behavior, we train neural networks to perform classification while simultaneously maintaining the metric structure within each class at potentially different levels of a hierarchy, leading to continuous and isometric within-class representations. We show that such network representations turn out to be a beneficial component for making accurate and robust inferences about the world. We come up with a network architecture that facilitates hierarchical manipulation of internal neural representations. We verify that our isometric regularization term improves the robustness to adversarial attacks on MNIST and CIFAR10. Finally, we use toy datasets and show that the learned map is isometric everywhere, except around decision boundaries.

DOI: https://doi.org/10.1038/s41598-025-20619-0