SEARCH · bioRxiv Science

Search bioRxiv Science

Search indexed bioRxiv preprints in genomics, neuroscience, cell biology and bioinformatics. Read source abstracts and check manuscript versions; preprints are not peer reviewed.

Quote a phrase for an exact phrase match. Source license links do not imply unrestricted reuse.

500 records · Page 2Linked to original sources

bioRxiv2026

KAT3 Shuttling Between Neuronal Identity and Activity-Dependent Plasticity Programs Drives Large-Scale Chromatin Remodeling

Activity-dependent transcription is a central feature of neuronal plasticity. Here, we show that neuronal activation triggers genome-wide redistribution of CBP and p300 in hippocampal neurons. Upon stimulation, KAT3 cofactors relocate from super-enhancers supporting neuronal identity to enhancers associated with activity-regulated genes, accompanied by transient changes in H3K27ac, chromatin accessibility, and three-dimensional genome architecture. Mechanistically, distinct TF families control KAT3 shuttling: proneural bHLH factors such as NeuroD2 maintain cofactor occupancy at identity-associated regulatory elements, whereas AP-1 binds de novo at plasticity-associated loci. This dynamic redistribution reshapes enhancer landscapes and chromatin interactions, enabling robust activation of plasticity genes while transiently attenuating neuronal identity programs. Remarkably, FOS overexpression is sufficient to reproduce the repression of neuronal identity genes observed during stimulation. Together, our findings reveal a reversible competition between transcriptional networks governing neuronal identity and plasticity and identify KAT3 redistribution as a key mechanism coupling neuronal activity to large-scale chromatin remodeling.

neuroscience↗

bioRxiv2026

IBD-Derived Colonic Fibroblasts Exhibit an Osteopontin-Enriched Secretome, and Osteopontin Restrains Human Colonic Organoid Maturation

Background: Intestinal fibroblasts are extensively remodeled in inflammatory bowel disease (IBD), yet the soluble stromal signals that directly influence epithelial maturation remain incompletely understood. We examined whether fibroblasts derived from inflamed IBD colon display an osteopontin (OPN; SPP1)-enriched secretory phenotype and whether extracellular OPN directly modifies non-neoplastic human colonic epithelium. Methods: Conditioned media from 5 noninflamed-associated fibroblast (NAF) and 4 inflammatory-associated fibroblast (IAF) cultures were analyzed in the validated multi-donor cytokine-array matrix, with orthogonal SPP1 RT-qPCR validation in a complementary fibroblast cohort. Recombinant OPN was then tested in human colonic organoids from 3 donors using donor-resolved molecular and functional analyses under standard, fibroblast-conditioned, and WNT-modified culture conditions. Donor identity defined biological replication. Results: OPN showed the strongest positive rank-based separation between IAF and NAF cultures: all 4 IAF values were higher than all 5 NAF values (Cliff's delta=1.00; exact Mann-Whitney P=0.0159; median ratio=3.64; Benjamini-Hochberg q=.19). Fibroblast RT-qPCR showed approximately 10-fold higher mean SPP1 expression in IAF than NAF cultures (P<.05). In organoids, OPN consistently reduced KRT20, FABP1, CA2, and MUC2 from Day 5 to Day 9. SOX9, HES1, and NOTCH1 increased at Day 9, whereas LGR5 and ALDH provided no evidence of canonical stem-cell expansion. Organoid-area and EdU responses were modest and donor dependent. Conclusions: IBD-derived colonic fibroblasts can display an OPN-enriched secretory phenotype. In human colonic organoids, OPN is sufficient to impair epithelial maturation, whereas its effects on growth and proliferation are variable and depend on the surrounding niche.

physiology↗

bioRxiv2026

Cell-type separability predicts annotation accuracy and outweighs algorithm choice: a factorial benchmark across seven scRNA paradigms

Automated cell-type annotation is a prerequisite for most single-cell RNA-sequencing (scRNA-seq) analyses, but the rapid proliferation of methods spanning marker-based, correlation-based, classical machine-learning, deep-learning, semi-supervised, large-language-model (LLM), and transformer foundation-model paradigms has outpaced head-to-head evaluation. Existing benchmarks rely on convenience samples of real datasets in which cell count, class imbalance, cell-type number, and differential-expression strength co-vary uncontrollably, precluding causal attribution of performance to any dataset property. To resolve this, we benchmarked 63 tools across seven paradigms using a Taguchi L9(34) orthogonal array that varies four dataset properties independently, progressively reconfiguring experimental control across five phases: fully controlled simulation, within-platform and cross-platform real-data validation, database-connected and LLM-based annotation under ontology-aware scoring, and fine-tuned foundation models. Using standardized oracle inputs and Cohen's {kappa}, we found that, within the ranges tested, the major paradigms achieved comparable accuracy. Accuracy was predicted near-linearly by the separability of cell types in a shared expression embedding, measured as k-nearest-neighbor (kNN) purity, a relationship that held across sequencing platforms and in fine-tuned foundation models. We attributed the vast majority of {kappa} variance to dataset structure and only a small share to tool identity. Computational cost traded against workflow accessibility rather than accuracy: accessible correlation-based and LLM-based approaches performed competitively, while foundation models matched them only after fine-tuning. Because our oracle design isolates algorithmic capability from upstream noise, these results reframe how methods should be selected: the field's near-term gains lie in strengthening infrastructure--prioritizing tool accessibility, standardized evaluation, and robustness to pipeline variation.

bioinformatics↗

bioRxiv2026

From neuropeptide and receptor annotation to ligand-receptor pairing: a sequence- and structure-based framework for mapping the neuropeptide-receptor interactome in Gryllus bimaculatus

Neuropeptides and their G protein-coupled receptors (GPCRs) control much of insect physiology and behaviour, but in Gryllus bimaculatus, an emerging model and edible insect, receptor sequence similarity hinders the mapping of which peptide each GPCR activates. We re-annotated a chromosome-scale genome (BUSCO 95.3%, from 86.7% on insecta_odb12) with comprehensive curation of 48 neuropeptide precursor families (51 loci, including seven not previously identified) and 134 candidate GPCRs (66 rhodopsin-class, 68 secretin-class), providing a near complete neuropeptide-receptor interactome catalogue. We modelled all 15,946 peptide-receptor pairs with AlphaFold3 and Boltz-2 and scored each interface with pLDDT and ipSAE. Ranking these scores, and cross-checking the top candidate for each family against a receptor phylogeny of known ligand specificity, gave a confident, phylogenetically related receptor for 27 of 35 curated receptor groups. These matches confirm the structural scorings with existing deorphanization data and propose receptors for peptides with no prior functional evidence. The annotation, curated peptide and receptor sets, and ranked complexes are available through CricketBase (https://cricket.annotation.jp), a genome browser with a structure viewer of peptide-receptor complexes, providing a resource for G. bimaculatus endocrinology and a workflow to deorphanize GPCRs in other non-model insects.

bioinformatics↗

bioRxiv2026

Observational satiety: watching yourself eat induces more fullness than watching another

Appetite is shaped not only by physiological need but by the sensory experience of eating and its social context. Can watching food being eaten itself induce satiety, and does it matter who is seen eating? In a functional MRI study (N = 41), participants watched videos of a wanted snack being eaten from their own perspective (self) or, using identical footage shown vertically inverted, as another's (other), holding food identity and visual content constant so that only the attributed agent varied. Observed eating reduced wanting, but not liking, for the eaten foods, whether one's own or another's; crucially, reported fullness increased only when the eating was seen as one's own. In the brain, food-value regions responded to watching eating in both conditions, a shared signal that strengthened over time. Yet only self-attributed eating engaged a self-specific, value-related response in the ventral striatum and medial prefrontal cortex, and orbitofrontal activity during self-eating scaled with each person's reported satiety, whereas watching another eat instead engaged the temporoparietal mentalizing network without a comparable rise in fullness. A large online survey (N = 1,000; ages 15-97) reproduced the behavioral effect across the adult lifespan, and a real-eating experiment reproduced its sensory-specific pattern. Watching eating, and whose eating it is, can thus recalibrate appetite through separate food-value and social-cognitive routes. This "observational satiety" offers a non-invasive route to study food wanting, of potential relevance to social eating and to today's food-media environments.

neuroscience↗

bioRxiv2026

EEG Oscillations in Guided Mindfulness versus Mind-Wandering in Young Adults: Effects of Auditory Task Instruction and Naturalistic Video

Alpha and theta band EEG oscillations have been implicated in states of mindfulness meditation. However, results are inconsistent and the influence of testing environment variables is not well-characterized. We used EEG to measure the amplitude of brain oscillations in a mindful-attention versus mind-wandering condition, in which guided auditory instructions were interleaved with periods of silence, with and without accompanying naturalistic video projections. We generated precise measures of alpha and theta in a sample of 20 young-adult non-expert meditators, by identifying each participant's individual alpha peak frequency (IAF) from posterior EEG electrodes and using it to define four individualized frequency bands: two low alpha bands (in 2 Hz increments below the IAF), one upper alpha band (from IAF to 2 Hz above IAF), and one theta band (4 Hz - 6 Hz below IAF). We found that the mindfulness manipulation significantly increased power in alpha ranging between 2 Hz below to 2 Hz above IAF, including peak alpha amplitude, compared with mind-wandering. Meanwhile, central theta amplitude was larger when auditory instructions were on versus off, and naturalistic video did not reliably modulate the EEG effects of mindfulness. We conclude that the acute effects of mindfulness in non-expert meditators are most consistently observed as increases of posterior alpha-band activity, and that auditory task-instructions should be accounted for in studies of guided meditation. Observed alpha increases may reflect induced states of calm or relaxation induced by mindfulness practice, as proposed in previous studies. These results may apply to mindfulness training or biofeedback therapies.

neuroscience↗

bioRxiv2026

INFORME: coupling information-theoretic experimental design with nonlinear mixed-effects modeling for efficient observation scheduling

Mathematical models of treatment response can inform individualized therapy, but their calibration often requires longitudinal measurements that are costly, burdensome, and collected on fixed schedules. Such schedules may be inefficient, over-sampling patients whose response is already well characterized while delaying informative measurements for those whose model parameters remain uncertain. We present INFORME (INFORmation-theoretic design with Mixed Effects), a framework that combines Bayesian information-theoretic experimental design with nonlinear mixed-effects modeling to adaptively select each patients next measurement time. Population and response-subgroup parameter distributions learned from an existing cohort provide informative priors, allowing candidate measurement times to be ranked by their expected reduction in patient-specific parameter uncertainty. As observations accumulate, priors can be updated to reflect the response subgroup most consistent with the patients data. We evaluate INFORME in two radiotherapy datasets: 150 synthetic tumor volume trajectories from a hybrid cellular automaton model of prostate cancer spheroids (HD1) and longitudinal tumor volumes from 39 patients with head-and-neck cancer (HD2). In HD1, population priors allowed omission of both pretreatment scans, while adaptive scheduling reduced the protocol from nine scans to three or four, with the response group identified from a single post-treatment scan on day 27. In HD2, the adaptive schedule used three scans instead of six and improved prediction by delaying the first on-treatment scan from week 1 to week 2, avoiding transient dynamics that produced false-positive and false-negative response projections. Across both datasets, the adaptive schedules used a mean of 2.7 scans in stead of seven and advanced completion of the patient-specific prediction by a mean of 15.5 days (95% CI, 6.7-24.3) relative to the equidistant protocol, while treatment duration remained unchanged. INFORME therefore reduces measurement burden and accelerates patient-specific prediction by concentrating observations at times that are most informative for model calibration.

systems biology↗

bioRxiv2026

Genetic diversity within and between polyploid sugarcane (Saccharum spp.) families obtained via caryopsis using microsatellite markers and multicategory model

Genetic diversity analyses are essential for sugarcane (Saccharum spp.) breeding programs. Crossbreeding, based on genetic distances between parental plants, is a tool used to increase genetic variability and enhance plant selection; however, quantifying variation in highly polyploid species remains a challenge. The present study aimed to evaluate the diversity within and between 12 families of sugarcane derived from caryopses, analyzing 120 individual seedlings arranged in an augmented block design. Genotyping was performed using primers for 16 microsatellite loci, five simple sequence repeat (SSR) loci, and 11 expressed sequence tag-SSR (EST-SSR) loci. To accurately account for polyploidy, similarity calculations were performed using Bruvos distances among individuals and RST distances among the families. Analysis of molecular variance (AMOVA) indicated that most of the genetic variability was within families (72%), with only 28% found between them. This high level of intra-family variation demonstrates that a significant reservoir of genetic diversity remains available within the crosses. The highest genetic similarity was observed between the families RB986952 x RB986960 and RB036122 x RB03611, whereas the lowest genetic similarity was observed between the families RB97319 x RB966928 and RB106802 x RB855036. Although the evaluated families shared high genetic similarity, the pronounced genetic variation within them demonstrates a robust recombination potential, indicating that the genetic basis of sugarcane can be better explored using the high variability that already exists in the selection of desirable morpho-agronomic characteristics within the families. Furthermore, this study highlights the importance of using appropriate distances for diversity studies with codominant markers, such as microsatellites, in polyploid species.

genetics↗

bioRxiv2026

A multiscale analysis of liver lobule fibrosis and its impact on drug propagation and metabolism - a DLA approach

Employing DLA methods, this paper explores the self-assembly of collagen fibers and resulting fibrosis at three scales up to the scale of regular lobule models. This allows a mechanistic exploration of the effects of collagen on drug transport (flow and diffusion) and metabolism. In addition, this method permits an analysis of fiber growth characteristics. First, variations of the DLA method of Parkinson et al (1994) will be used to generate multiple explicit collagen microfibril self-assembly using DLA particles in one dimension using cubic grid blocks of (4 mm)3 in a 240 x 20 x 20 grid model. The second stage will be to assess the consequences of various densities of these fibers in three dimensions on flow reductions at a higher scale. Here we utilize DLA methods in cubic grid blocks of (80 nm)3 to mimic 3D collagen self-assembly of fibrils. We then apply a pressure gradient or specified flow rates across a spatially gridded version of these models to quantify flow effects. This region represents a local zone of liver tissue affected by fibrosis. Analytic models of fibrotic effects on flow are employed for comparison. A third stage explores the implications of fibrosis in a liver lobule model using multiple grid blocks of size 3200 mm to represent the lobule tissue. Here, a continuum model of fiber density is employed, based on the previous two scales. The model also includes the effects of additional grid blocks representing sinusoidal flow paths found in the lobule. We contrast and quantify drug propagation and metabolism of molecular dissolved versus nanoparticle delivery vehicles in fibrotic media, achieved by upscaling explicit collagen distributions to appropriate average values.

physiology↗

bioRxiv2026

Sex-specific long-term alteration of hippocampal excitation/inhibition balance and behavior by transient caffeine exposure during synaptogenesis

Caffeine is the most widely consumed psychoactive substance worldwide, yet the long-term consequences of exposure during critical periods of brain development remain incompletely understood. Synaptogenesis represents a vulnerable window during which environmental factors can shape the maturation of neuronal circuits and influence lifelong brain function. Here, we investigated the impact of caffeine exposure during hippocampal synaptogenesis on synaptic development, neuronal function, behavior, and seizure susceptibility, with a particular focus on sex-dependent effects. Developmental caffeine exposure induced distinct, sex-specific trajectories of hippocampal synaptic remodeling. In the CA1 region, caffeine produced opposite patterns of glutamatergic synapse regulation, characterized by a delayed reduction in excitatory synapse density in males and an increase in females. In contrast, inhibitory synapse organization was selectively altered in males, with a transient increase in CA3 inhibitory synaptic density during development associated with enhanced inhibitory transmission, whereas females exhibited no significant changes. These findings reveal sex-specific and temporally divergent effects of developmental caffeine exposure on hippocampal synaptic maturation and function. At the behavioral level, developmental caffeine exposure produced distinct sex-dependent phenotypes : males exhibited increased anxiety-like behavior, whereas females developed a delayed impairment in recognition memory that became apparent only in adulthood. Furthermore, caffeine exposure selectively increased PTZ-induced seizure susceptibility in juvenile females, an effect that was no longer detected in adulthood. Together, these findings demonstrate that caffeine exposure during hippocampal synaptogenesis induces sex-specific and temporally dynamic alterations in circuit maturation, resulting in distinct behavioral and neuronal excitability outcomes. These results highlight the importance of considering both sex and developmental timing when assessing the neurodevelopmental consequences of caffeine exposure.

neuroscience↗

bioRxiv2026

De novo design of CR2 binder as vaccine scaffold

Efficient B cell activation during vaccine-induced humoral immunity relies on both B cell receptor (BCR) antigen recognition and synergistic signaling from co-receptors. Complement receptor 2 (CR2), the primary BCR co-receptor on B cells, lowers the activation threshold and amplifies downstream kinase signaling by orders of magnitude when engaged by complement fragment C3d decorated antigens. Targeting CR2 therefore represents a rational vaccine enhancement strategy, yet native C3d suffers from low affinity, poor stability, and manufacturing challenges. Here, we report the de novo design of a highly stable, high-affinity CR2 binder using deep learning driving protein design methods. Biophysical characterization, high-resolution cryoEM structural determination, and functional assays in vitro and in vivo confirm that the designed binder matches computational design models and specifically engages CR2 to boost B cell activation. When fused to antigen as a vaccine scaffold, the trimeric CR2 binder elicits robust humoral immune responses comparable to nanoparticle vaccines, while retaining the simplicity of single-chain protein production. Our work establishes a modular CR2 targeting vaccine scaffold platform with broad translational potential for next-generation protein vaccines.

immunology↗

bioRxiv2026

Brain dynamics of memory encoding for simple versus complex musical sequences

Memory encoding is the foundational process by which the brain transforms sensory input into lasting representations. While the neural mechanisms of auditory memory have been extensively studied, how musical complexity modulates the neural activity during memory encoding remains poorly understood. Here, we used magnetoencephalography (MEG) to investigate the encoding of simple (tonal) versus complex (atonal) musical melodies in 67 participants. Behaviorally, the latter melodies were consistently rated as more complex and associated with lower recognition accuracy across three testing sessions (same day, one day later, and ten days after the encoding task). At the neural level, source-localized analyses revealed distinct spatiotemporal dynamics: simple melodies elicited stronger activity in auditory cortices (left and right Heschl's gyrus) and cingulate regions (medial and anterior cingulate gyrus), while complex melodies recruited the left hippocampus more extensively across multiple tones. These findings demonstrate that musical complexity shapes neural encoding processes from the outset, with tonal sequences benefiting from efficient sensory processing and atonal sequences requiring greater memory-related recruitment. Our study provides novel insights into how the human brain encodes complex auditory information, providing a framework for understanding the neural basis of memory formation for temporally structured stimuli.

neuroscience↗

bioRxiv2026

Macrophage signature-based prediction of cancer treatment response using MIL-attention

Predicting immunotherapy response from single-cell data remains difficult due to patient-level labels, extreme class imbalance, and highly heterogeneous macrophage states. We present a Multiple Instance Learning (MIL) framework that treats each patient as a bag of macrophage embeddings derived from a single-cell RNA foundation model. The architecture incorporates an attention-based pooling mechanism with reduced model complexity, dropout-enhanced regularization and explicit attention penalties to improve stability in small-sample regimes. To address imbalanced clinical datasets, MIL outputs are optimized with a combined focal loss and supervised contrastive objective that simultaneously sharpens class boundaries and improves representation clustering. Across three cancer datasets, this approach outperforms pseudobulk aggregation, embedding baselines and standard MIL variants. Attention-weighted attribution and transcriptional regulatory analysis reveal distinct macrophage programs, interferon and antigen-presentation networks in responders versus hypoxia-linked regulatory modules in non-responders. This shows the potential of MIL to uncover predictive and mechanistically interpretable immune states.

bioinformatics↗

bioRxiv2026

In-cell structural analysis reveals a distinctive chloroplast ribosome in Chlamydomonas reinhardtii

Chloroplast ribosomes synthesize plastid-encoded components of photosynthetic machinery, yet their structure and organization remain poorly understood. We combined cryo-focused ion beam milling, cryo-electron tomography and subtomogram averaging to determine native chloroplast ribosomes in Chlamydomonas reinhardtii. The 4.4-4.9 [A] structure revealed a large arch-like extension on the small subunit (SSU). Comparisons with bacterial and plant chloroplast ribosomes, supported by proteomics, AlphaFold3 predictions and a recent atomic model, indicate that the arch is formed by insertions and extensions in SSU proteins. Classification resolved active, thylakoid-associated ribosomes with density adjacent to the nascent peptide exit and an arch-moved state enriched among thylakoid-associated particles, with coordinated displacement of the arch and beak. Phylogenetic analysis revealed an evolutionary mosaic: the uS3c insertion is broadly distributed across Chlorophyceae, whereas the uS2c insertion, uS5c and PSRP7 are concentrated in Chlamydomonadales, with PSRP7 also in Sphaeropleales. Nuclear-encoded components were recruited stepwise onto a plastid-encoded scaffold, with all four under comparable purifying selection. These findings link a lineage-specific SSU extension to ribosome dynamics, thylakoid association and evolution, highlighting the value of in-cell structural analysis.

plant biology↗

bioRxiv2026

Repeated listening induces exposure-specific cortical tracking of intelligible continuous speech

Neural encoding of acoustic and linguistic features of continuous speech is sensitive to cognitive factors, such as attention and comprehension. We investigated whether neural tracking is also sensitive to the predictability of speech. Participants were repeatedly exposed to intelligible or unintelligible versions of the same audiobook segment while EEG was recorded. First, we fit encoding models to predict EEG responses from acoustic, sublexical, and lexical features of the presented speech. Model comparisons revealed no reliable improvement in model fit when lexical features were included; subsequent analyses were performed on models including only acoustic and sublexical predictors. Second, we compared prediction accuracy for models trained and tested on the same exposures with models trained and tested across different exposures. While we observed no overall change in prediction performance across exposures, we found that models were exposure-specific: prediction performance was highest within the same exposure and decreased with increasing temporal distance between the training and test exposure. This effect was observed for intelligible but not for unintelligible speech, suggesting that the effect depends on properties unique to intelligible speech, such as the ability to form increasingly specific predictions about upcoming linguistic input, rather than general, non-linguistic factors related to repeated exposure. This distance effect was associated with increased model weights from -90 ms to 130 ms, indicating an enhancement of familiar input during an early cortical processing stage. In summary, these findings indicate that cortical tracking of sublexical speech features is modulated by repeated exposure to intelligible speech, consistent with a role for linguistic predictability.

neuroscience↗

bioRxiv2026

MIND the gap: methodological considerations and guidance for structural MRI similarity network analysis with MIND

Structural similarity networks quantify the similarity of structural properties across cortical regions, providing a macroscopic window onto the organisation of cortical architecture. Morphometric inverse divergence (MIND) is a multivariate metric of similarity between cortical areas, based on the Kullback-Leibler (KL) divergence between areal distributions of multiple MRI features or morphometric variables locally measured at voxel or vertex resolution. MIND has demonstrated technical robustness and biological validity and is increasingly widely used as a measure of cortico-cortical similarity in clinical and developmental network neuroscience. Here we provide in-depth methodological background on KL divergence and MIND, highlighting possible sources of bias, critical user decision points in the design of a MIND processing pipeline, and recommendations for technical risk mitigation in using MIND as a metric of cortical similarity. We use simulated data and observational MRI datasets from adults (UK Biobank, N = 500 T1-weighted and diffusion scans) and neonates (Developing Human Connectome Project, N = 752 T2-weighted scans), to show how the estimator of KL divergence implemented in MIND is potentially influenced or biased by five properties of input MRI feature maps: (i) their smoothness; (ii) the proportion of identical values; (iii) analysis in native or common space and the choice of vertex mesh resolution; (iv) parcellation choice; and (v) covariance between input features. We offer principled and practical guidance for investigators wanting to specify and implement the MIND processing pipeline that is best suited to the constraints and opportunities of the MRI data available to them. These recommendations outline which pipeline steps should be used sparingly, such as vertex map smoothing; which should be used with informed caution, such as parcellation choice or vertex mesh resampling; and which could be newly implemented for more robust estimation of MIND, such as the use of principal component analysis to preprocess multivariate MRI features. To support further development of structural MRI similarity network analysis, and wider adoption of robust MIND methods, we also publish the code used to generate the results in this paper as an open resource.

neuroscience↗

bioRxiv2026

SEROTONERGIC ANXIETY IS A STRESS INTENSITY-DEPENDENT STATE MEDIATED BY DOPAMINERGIC SIGNALING

Despite the central role attributed to serotonin in anxiety, its involvement and therapeutic efficacy are inconsistent, raising the possibility that serotonergic recruitment depends on the stress history from which anxiety emerges. Using larval zebrafish, we combined graded glucocorticoid exposure with chemogenetic DRN manipulation, whole-brain activity mapping, and pharmacology to test whether stress intensity determines serotonergic involvement in anxiety.Increasing stress intensity did not simply increase anxiety severity but generated distinct anxiety phenotypes. Lower glucocorticoid exposure produced context-general anxiety that required the serotonergic DRN, whereas higher exposure produced context-dependent, DRN-independent anxiety. Whole-brain mapping identified a posterior tubercular/hypothalamic dopaminergic region associated with DRN-dependent anxiolysis, while D1, but not D2 receptor antagonism abolished the anxiolytic effect of DRN ablation. Finally, environmentally relevant nanomolar concentrations of methylphenidate reduced anxiety-like behavior, further supporting dopaminergic modulation of anxiety. Together, our findings identify stress intensity as a determinant of serotonergic recruitment during anxiety, reveal a downstream contribution of D1-dependent dopaminergic signaling, and provide a framework for understanding the mechanistic heterogeneity of anxiety and its variable response to serotonergic treatments.

neuroscience↗

bioRxiv2026

Beyond benchmark accuracy: machine-learning turnover-number predictors require system-level validation

Enzyme turnover numbers (kcat) are essential for kinetic models and enzyme-constrained genome-scale metabolic models (ecGEMs), but measured values are sparse and therefore increasingly estimated using machine learning (ML). Although these predictors are commonly evaluated by global regression metrics, their practical utility depends on how errors propagate through downstream models. We benchmarked six current kcat predictors on a curated BRENDA-derived dataset and five of them on EnzyExtract. To assess the influence of training-set proximity, we compared each benchmark dataset with the available training data for each predictor. We then used the predicted kcat values to parameterize ecGEMs of Saccharomyces cerevisiae and evaluated growth predictions across 19 conditions. We find that benchmark accuracy is moderate even on the BRENDA-derived dataset and drops sharply on EnzyExtract, where all predictors achieve R2 values of 0.20 or lower. This decline is accompanied by substantially lower overlap between the benchmark and training datasets, with exact sequence matches ranging from 24% to 78% for BRENDA, compared with 9% to 26% for EnzyExtract. However, that overlap alone does not explain differences in generalization across predictors. Moreover, downstream performance is also not explained by benchmark ranking. Across 19 conditions, none of the tool-specific ecGEMs consistently reproduces the experimentally observed variation in growth. In glucose minimal medium, the weakest benchmark performer yields the most accurate growth prediction in the downstream ecGEMs, whereas higher-ranked predictors produce larger deviations in growth. We trace this mismatch to localized errors at high-leverage positions in yeast's metabolic network, where underpredicted mitochondrial ADP/ATP carrier turnover numbers restrict adenine nucleotide exchange and impose an apparent limitation on cytosolic ATP supply. Relaxing this constraint shifts predicted growth toward the experimental reference. Thus, ML-derived kcat values can affect not only quantitative growth predictions but also the phenotype a mechanistic model appears to identify. These results argue for application-driven validation of biological parameter predictors in the downstream systems they are intended to support.

bioinformatics↗