Cognitive flexibility, the capacity to update behavior when the rules of a task change, is a cornerstone of everyday independence. Unlike memory loss, which has been studied extensively in aging rodents, the mechanisms behind age-related inflexibility have remained comparatively opaque. The research team, based at Kobe University Graduate School of Medicine, tackled the problem using an attentional set-shifting test delivered through touchscreen-based operant chambers. Mice first learned to discriminate between vertical and horizontal visual stimuli for a food reward, then had to abandon that rule and respond instead to a spatial side, left or right, regardless of what appeared on the screen. Success in the second phase demands exactly the kind of executive updating that deteriorates in aging humans.

A crucial decision in the experimental design involved mouse substrains. Aged C57BL/6N mice performed poorly in both the visual discrimination and response-direction phases, indicating broad learning deficits that would have confounded any analysis of flexibility specifically. Aged C57BL/6J mice, by contrast, learned the visual discrimination task normally, yet splintered dramatically in the response-direction phase: some failed almost completely while others matched the performance of young animals. Middle-aged mice showed no such variability, demonstrating that this heterogeneity emerges specifically in advanced age rather than accumulating gradually across adulthood. Females showed a similar pattern to males, though the mechanistic work focused on males. Importantly, individual differences in the flexibility phase did not correlate with any measure of visual discrimination learning, confirming that the variability reflected a selective executive deficit rather than uneven general learning ability.

With a reliable behavioral signature in hand, the researchers turned to volume electron microscopy to examine synaptic ultrastructure in the infralimbic cortex, a component of the medial prefrontal cortex. Using serial block-face scanning electron microscopy, they reconstructed three-dimensional cubes of neuropil and annotated hundreds of individual synapses per mouse, quantifying synapse density, the axon-spine interface as a proxy for synapse size, spine apparatuses, astrocytic coverage, and the presence of presynaptic mitochondria. Aging left visible marks: spine synapse density trended downward, average synapse size increased, and the proportion of synapses containing a spine apparatus rose significantly, a pattern consistent with the preferential loss of smaller synapses. Yet none of these structural changes correlated with how well individual aged mice performed the flexibility task.

The exception was presynaptic mitochondria. Neither their overall frequency nor their morphology changed with age, but in aged mice the proportion of synapses containing a presynaptic mitochondrion was inversely correlated with cognitive performance, a relationship entirely absent in young animals. In other words, aged mice with more mitochondria crowding their synaptic terminals were precisely the mice that struggled to shift behavioral strategies. This counterintuitive association, more synaptic mitochondria predicting worse performance, became the central thread of the study.

To uncover the molecular underpinnings, the team performed quantitative proteomics on both whole tissue and synaptosome fractions, purified preparations of pinched-off nerve endings, from the medial prefrontal cortex. A striking dissociation emerged. The proteins whose abundance changed with chronological age overlapped with the proteins correlated with cognitive flexibility only at chance levels, indicating that the biology of getting older and the biology of losing mental flexibility are largely separable molecular programs. Correlation analyses, verified with exact permutation tests that exhaustively reshuffled behavioral scores, and rank-based gene set enrichment analysis converged on the same conclusion: proteins associated with poorer flexibility in aged mice were overwhelmingly mitochondrial, enriched for oxidative phosphorylation, the tricarboxylic acid cycle, and mitochondrial translation. Hub analysis of protein-protein interaction networks likewise placed mitochondrial oxidative phosphorylation at the center of the synaptosomal signature of inflexibility, while cytoplasmic ribosomal proteins showed the opposite, positive relationship with performance. Of 1,140 mitochondrial proteins catalogued in the MitoCarta3.0 database, 135 detected in synaptosomes were negatively correlated with cognition, against only five positively correlated.

These correlations raised a causal question: is mitochondrial dysfunction at synapses merely a correlate of inflexibility, or does it help drive it? The researchers answered with a pharmacological intervention. MitoQ, an antioxidant molecule consisting of ubiquinone conjugated to a lipophilic triphenylphosphonium cation that actively accumulates inside mitochondria, was administered in drinking water to mice from 55 to 75 weeks of age. A control group received dTPP, a compound identical to MitoQ but lacking the antioxidant quinone moiety. After twenty weeks of treatment, MitoQ significantly improved performance in the response-direction test, with a large effect size, while leaving visual discrimination learning untouched, a double dissociation mirroring the behavioral specificity of the natural deficit. The dTPP control group performed comparably to untreated aged mice, indicating that the benefit derived from antioxidant activity rather than the carrier molecule.

Proteomic profiling of MitoQ-treated brains revealed how the drug works. In synaptosomes, but not whole tissue, MitoQ-induced changes were inversely correlated with age-associated changes, meaning the antioxidant partially reversed the molecular remodeling of the aging synapse. MitoQ selectively lowered synapse-associated mitochondrial proteins while leaving cytoplasmic and mitochondrial ribosomal proteins intact, and among apoptosis-related mitochondrial proteins it specifically reduced pro-apoptotic species, including BNIP3, BAD, and FAM162A, without altering anti-apoptotic counterparts. Gene ontology analysis linked the downregulated proteins to regulation of cytochrome c release and mitochondrial membrane permeabilization, hinting that oxidative stress at synapses may incite a latent apoptotic signaling program that could, in principle, recruit microglial synapse elimination. MitoQ also upregulated actin-reorganization factors such as cofilin and ADF in synaptosomes, suggesting that mitochondrial redox state influences the cytoskeletal dynamics on which synaptic plasticity depends. The authors caution that apoptotic signaling was inferred from protein abundance rather than directly measured, for example through caspase-3 activation, and that the behavioral cohort was modest, with five control and seven treated animals, warranting replication in larger samples.

Conceptually, the study reframes synaptic mitochondria as a latent vulnerability that only reveals its consequences in advanced age. Mice destined for inflexibility may carry a higher complement of synaptic mitochondria, and a richer oxidative phosphorylation machinery, from youth onward, but the relationship with cognition materializes only when age-related factors, such as dysregulated prefrontal hyperactivity and declining peroxisomal support, amplify mitochondrial reactive oxygen species production beyond what synapses can tolerate. This model may also reconcile apparent contradictions with earlier reports in which impaired reversal learning in aged C57BL/6N mice coincided with reduced, rather than increased, mitochondrial gene expression in the hippocampus, a discrepancy attributable to differences in brain region, subcellular compartment, strain, and cognitive domain. Because dysfunctional mitochondria are implicated in dementia and other neurodegenerative diseases, the authors argue that synaptic mitochondrial oxidative stress deserves scrutiny as a shared mechanism of inflexibility across conditions, and that mitochondria-targeted antioxidants represent a promising, testable avenue for preserving mental flexibility into old age.

Subject of Research: Synaptic mitochondrial oxidative stress and individual variability in age-related cognitive inflexibility in mice

Article Title: Synaptic Mitochondrial Oxidative Stress Contributes to Individual Variability in Age‐Related Cognitive Inflexibility in Mice

Article References: Yamada, R., Nagai, H., Numa, C., Zhu, Y., Nagai, M., Ota, K., Kawashima, Y., Ohno, N., & Furuyashiki, T. (2026). Synaptic Mitochondrial Oxidative Stress Contributes to Individual Variability in Age‐Related Cognitive Inflexibility in Mice. Aging Cell, 25(9), Article e70716. https://doi.org/10.1111/acel.70716

Image Credits: AI Generated

DOI: 10.1111/acel.70716

Keywords: cognitive flexibility, aging, mitochondria, oxidative stress, prefrontal cortex, synapses, MitoQ, attentional set shifting, proteomics, electron microscopy, executive function, cognitive aging

The team, led by Yanan Lv, Tong Xin, Jiangduo Liu, Haoran Chen, Hua Han and Xi Chen at the Institute of Automation of the Chinese Academy of Sciences and collaborators, frames the problem as cross-modal ultrastructural inference. Rather than hallucinating structures from scratch, their framework attempts to recover fine morphological information already latent in the optical signal, latent cues that the physics of diffraction has smeared but not entirely destroyed. In other words, the goal is not synthesis de novo but amplification of what the light microscope actually captured, rendered in the visual idiom of electron microscopy.

Technically, the heart of the method is a Schrödinger Bridge, a class of generative model grounded in stochastic optimal transport. Unlike standard diffusion models that gradually denoise toward a target distribution through a fixed reference process, a Schrödinger Bridge learns the most probable stochastic path between two empirically observed distributions—in this case, the distribution of light-microscopy images and that of electron-microscopy images. This makes the translation between modalities more principled, steering each light-microscopy patch toward its most plausible electron-microscopy counterpart while respecting the geometry of the data.

What distinguishes CDSB from a generic image-to-image translator is its content-decoupling strategy. The architecture disentangles modality-invariant physical content—the actual biological structure shared by both imaging modes—from imaging-specific attributes such as contrast, texture and noise introduced by each microscope’s physics. A Content-Decoupling Autoencoder separates these factors, and Feature-wise Linear Modulation injects the modality-specific attributes back in at the right stage of generation. The result is a system that can swap the imaging style while preserving the underlying anatomy, a property the authors argue is essential for scientific credibility: the generated image should change how structures look, not what they are.

To keep the translated images structurally plausible rather than merely plausible-looking, the researchers incorporate a physics-informed cross-modality perceptual loss. This loss draws on knowledge of the imaging process itself, notably the point spread function that governs how a light microscope blurs point sources, and it penalizes generated structures that would be physically inconsistent with the optical evidence. The combined objective encourages the network to sharpen real morphological cues rather than invent convenient detail, addressing one of the most serious objections to generative methods in biology: that they might fabricate structures no microscope ever saw.

The practical payoff shows up in three parts of the connectomics workflow. First, because the generated electron-microscopy-like images are markedly clearer than raw light-microscopy data, human experts selecting regions of interest for targeted electron-microscopy acquisition made fewer misses. In large-scale connectomics, where full-volume electron microscopy is impractical, researchers routinely use fast optical imaging to scout for interesting structures and then acquire high-resolution electron microscopy only at selected spots. Sharper scout images mean fewer important targets overlooked and less wasted time at the electron microscope.

Second, the generated images are inherently aligned with their source light-microscopy data while matching the appearance of the electron-microscopy target, which simplifies multi-modal registration. Registering images from two microscopes with different resolution, contrast and distortion is notoriously difficult; a bridge image that belongs to both worlds gives registration algorithms a much easier anchor. Third, and perhaps most strikingly, the translated images allow pre-trained electron-microscopy segmentation models to be applied directly to light-microscopy data. Segmentation networks trained on the relatively small pool of annotated electron-microscopy volumes are among the most valuable assets in the field, and CDSB effectively extends their reach to the much larger and faster-growing pool of optical data, improving segmentation accuracy without retraining on new modalities.

The broader significance lies in what it suggests about the economics of brain mapping. Projects such as whole-brain connectomes are limited less by algorithms than by acquisition time: electron microscopy throughput is the bottleneck, and every improvement in downstream efficiency multiplies across petabytes of data. By enhancing the analytical value of each light-microscopy image, CDSB shifts some of the burden from slow hardware to fast computation, letting researchers triage tissue, register datasets and run quantitative analysis at optical speeds while retaining electron-microscopy-grade interpretability where it matters.

The work also reflects a wider trend in biomedical imaging: physics-informed generative modeling that respects, rather than ignores, the measurement process. Earlier cross-modal translation approaches built on generative adversarial networks or standard diffusion models often struggled with out-of-distribution tissue and with fidelity guarantees. By combining optimal-transport-based bridges with explicit content decoupling and physics-aware losses, the authors position CDSB as a more trustworthy tool for downstream scientific decisions, an important distinction when generated images guide which regions of a brain get analyzed at all.

As connectomics scales toward完整 brain volumes in model organisms, tools that compress the gap between imaging speed and resolution will increasingly define what is experimentally feasible. CDSB offers a concrete demonstration that the diffraction limit of light need not be the end of the analytical road: with the right generative machinery, fast images can be made to speak the language of slow ones, and the connectomics pipeline—from targeted acquisition to quantitative analysis—can move a decisive step faster.

Subject of Research: A deep learning framework that translates light microscopy images into electron microscopy-like representations to accelerate connectomics workflows.

Article Title: CDSB: accelerating connectomics workflow via Content-Decoupled Schrödinger Bridge

Article References: Lv, Y., Xin, T., Liu, J., Chen, H., Han, H., & Chen, X. (2026). CDSB: accelerating connectomics workflow via Content-Decoupled Schrödinger Bridge. BMC Biology. https://doi.org/10.1186/s12915-026-02715-3

Image Credits: AI Generated

DOI: 10.1186/s12915-026-02715-3

Keywords: connectomics, CDSB, Schrödinger Bridge, light microscopy, electron microscopy, cross-modal image translation, deep learning, generative model, image segmentation, brain mapping, physics-informed loss, ultrastructure