Biology subjects

Katori, Y.

Publications and source records attributed to Katori, Y..

9 recordsLinked to original sources

bioRxiv2026

Information coding strategies in biological reservoir computing for multiclass time-series classification

Physical reservoir computing (RC) using biological neural networks (BNNs) offers a framework for probing how high-dimensional dynamics arising from multicellular systems support complex information processing in the brain. However, how electrical inputs and extracellular readouts should be designed to suit different task demands remains unclear. Here, using primary rat cortical neurons cultured on a high-density microelectrode array, we investigated how input encoding and spike detection methods influence neural representations and the resulting RC performance in BNNs. Comparison of pulse amplitude modulation and latency coding revealed that the latter exhibited superior performance in demanding multiclass conditions, including a 10% improvement in a 10-class spoken digit classification task. Spike sorting further improved classification performance by enhancing class separability, while largely preserving the low-dimensional structure of population activity. These findings highlight the importance of designing information coding strategies according to task difficulty.

neuroscience↗

bioRxiv2026

KDM6A Haploinsufficiency Is Insufficient to Disrupt Cochlear Structure and Auditory Function in Mouse Models of Kabuki Syndrome Type 2

Objective: Kabuki syndrome type 2 (KS2) is a rare X-linked disorder caused by pathogenic variants in KDM6A and is frequently associated with hearing impairment. Because auditory phenotypes include sensorineural, conductive, and mixed hearing loss, we examined whether Kdm6a deficiency directly impairs cochlear morphology or auditory function in mice. Methods: We generated Nestin-Cre; Kdm6a conditional mutants and a germline-derived global Kdm6a deletion model using Sox2-Cre-mediated germline recombination. Auditory function and cochlear morphology were assessed by auditory brainstem responses (ABRs), distortion product otoacoustic emissions (DPOAEs), and histological analyses. Results: Whole-body hemizygous or homozygous Kdm6a deletion resulted in embryonic lethality. Analyses therefore focused on Nestin-Cre; Kdm6aflox/Y males and females carrying a germline-derived heterozygous Kdm6a deletion. In Nestin-Cre mutants, ABR thresholds, wave I amplitudes and latencies, and DPOAE measures at 8, 16, and 32 kHz were comparable with littermate controls at 4 and 8 weeks. In heterozygous deletion females, ABR thresholds, wave I amplitudes and latencies at 4-64 kHz and DPOAE measures at 8, 16, and 32 kHz were comparable with male and female wild-type controls. No overt cochlear pathology was observed in either model. Conclusion: Kdm6a deficiency in the viable mouse models examined did not produce overt cochlear pathology or baseline auditory dysfunction. These findings suggest that partial Kdm6a loss alone is insufficient to produce a robust primary cochlear phenotype in young mice and highlight the potential contributions of non-cochlear, developmental, central, and later-onset mechanisms to hearing impairment in KS2. Clinically, these findings support careful evaluation for potentially treatable middle-ear disease.

developmental biology↗

bioRxiv2026

Predictive E-prop: A biologically inspired approach to train predictive coding-based recurrent spiking neural networks

The predictive coding framework offers a compelling model for temporal signal processing in the cortex. Recent studies explored its implementation in spiking architectures using Hebbian plasticity rules or offline learning; however, a biologically inspired model that enables gradient-based minimization of prediction errors remains an open challenge. In this work, we demonstrate that the predictive coding objective can be optimized using the online and local nature of the e-prop learning algorithm in recurrent spiking neural networks, creating the Predictive E-prop model. We demonstrate that the model is capable of learning complex time-series signals purely from self-supervised learning, using only its own prediction error as input, maintaining self-sustaining activity and reproducing the targets underlying dynamics even in the absence of external stimuli. Furthermore, Predictive E-prop shows robust signal reconstruction abilities, effectively filtering noise and successfully interpolating sparse data. A comparative study against a backpropagation-based approach reveals that the two achieve comparable performance after training, confirming the viability of our model for timeseries generation tasks. These findings are particularly relevant for future developments in neuromorphic hardware, offering a purely self-supervised, gradient-based model that could provide significant advantages in power efficiency and computational ability.

neuroscience↗

bioRxiv2025

Online supervised learning of temporal patterns in biological neural networks under feedback control

In vitro biological neural networks (BNNs) provide a well-defined model system to constructively investigate how living cells interact with their environment to shape high-dimensional dynamics that could be used to generate a coherent temporal output, such as those required for motor control. Here, we developed a real-time closed-loop BNN system capable of generating periodic and chaotic temporal signals by integrating cultured cortical neurons with microfluidic devices and high-density microelectrode arrays. We show that training a simple linear decoder with fixed feedback weights enables the system to learn and autonomously generate diverse temporal patterns. When feedback was switched on, irregular activity in BNNs is transformed into low-dimensional, structured dynamics, producing coherent trajectories characterized by stable transitions between neural states. BNNs trained on different target frequencies--ranging from 4 to 30 s--could be trained to sustain oscillations at distinct frequencies, demonstrating their adaptability. Importantly, a top-down control of self-organized network formation with microfluidic devices is the key to suppress excessive synchronization and increase dynamical complexity in BNNs, facilitating the training and robust output generation. This work offers a biologically inspired platform for understanding the physical basis of cortical computation and for advancing energy-efficient neuromorphic computation. Significance StatementReservoir computing is a machine learning paradigm that exploits the transient dynamics of high-dimensional nonlinear systems. Although it was originally inspired by the mammalian brain and widely explored in physical systems, its implementations in biological neural networks (BNNs) have been limited due to their excessive connectivity and global synchrony in vitro. Here, we use microfluidic devices to construct modular, nonrandomly connected BNNs and integrate them with microelectrode arrays in a closed-loop reservoir computing environment. We show that the system can be trained to autonomously output various temporal signals, with the modular connectivity that is essential for learning. In vitro BNNs provide unique alternatives for physical reservoirs with dynamic adaptability.

neuroscience↗

bioRxiv2025

Efficient learning and intrinsic noise filtering in recurrent spiking neural networks trained with e-prop

ObjectiveBiologically plausible learning rules for neural networks, such as e-prop (eligibility propagation), are essential both for advancing neuromorphic computing and for understanding fundamental mechanisms of learning in animal brains. However, their behavior under different network conditions remains unclear. ApproachHere, we investigate the performance of the e-prop learning algorithm in recurrent spiking neural networks (RSNNs) across different levels of recurrent connectivity and input noise using a complex temporal credit assignment task, a supervised classification problem known to be solvable by rodents. Main resultsWe show that increased sparsity in the recurrent layer significantly enhances learning performance by promoting the generation of more diverse activation patterns. Analysis of the networks evolution further reveals that the e-prop-trained input layer evolves to route distinct inputs to different regions of the recurrent layer while suppressing the contribution of noise. This partially resembles signal routing functions attributed to the thalamus in mammalian sensory systems, providing additional support for the biological plausibility of e-prop. SignificanceThese findings offer promising insights for efficiency and advantages of biologically inspired training in RSNNs.

neuroscience↗

bioRxiv2025

Non-DNA-damaging DNA-PK activation improving hearing and prolonging life due to NAD+ and SIRT upregulation

Emerging evidence strongly supports a close relationship between age-related hearing loss and frailty, highlighting the importance of early detection and intervention. Recently, we invented a mitochondria-homing drug named mitochonic acid 5 (MA-5), that increases the adenosine triphosphate (ATP) levels, rescue mitochondrial function, and protect tissue damages. Currently, the phase I clinical trial has been finished in Japan (jRCT2031210495) and the phase 2 clinical trial has already been approved by PMDA. Here we show that MA-5 improved various types of hearing loss in mouse models. Structural chemical bioanalysis revealed that MA-5 is a mixture of equal amount of S- and R- enantiomer and both S- and R- enantiomer increase ATP by binding mitochondrial protein, mitofilin. However, S-enantiomer significantly increased the NAD+ levels by binding to the NAD+-producing key enzyme nicotinamide phosphoribosyltransferase (NAMPT). Moreover, the S-enantiomer increased the sirtuin 1 protein by suppressing polyubiquitination induced by tripartite motif containing 28 (TRIM28) phosphorylation which was triggered by DNA-dependent protein kinase (DNA-PK) activation in the absence of DNA damage. Transcriptomic signatures showed that the signature of MA-5 shows an inverse correlation with aging and mortality and is oriented in the same direction as the OSKM-related iPSCs, suggesting the modification of aging pathways. Oral administration of MA-5 to mitochondrial disease model mouse showed increased survival. Our findings suggest that, in addition to enhancing ATP levels, the coordinated regulation of NAD+ metabolism, SIRT protein expression, and DNA-PK activity-constituting a novel therapeutic triad may contribute to the amelioration of hearing impairment and mitochondrial dysfunction, thereby improving life prognosis.

cell biology↗

bioRxiv2025

Metabolic cell death of labile iron deficiency as a vulnerability of head and neck squamous cancer cells evaded by BACH1

Head and neck squamous cell carcinoma (HNSCC) remains difficult to treat due to the lack of molecularly targeted drugs with high anti-tumor efficacy and safety. To investigate the involvement of the transcription factor BACH1, which is known to promote the metastasis of various cancers, in the malignant properties of HNSCC cells, we examined the effects of BACH1 depletion with short interfering RNAs in human HNSCC cell lines. We found that knockdown of BACH1 induced cell death due to depletion of intracellular labile iron which is not tightly bound to protein. Mitochodrial electron transfer chain activity was severely reduced but efficiently resuced with supplementation of iron in the medium. By combining chromatin immunoprecipitation-sequence and RNA-sequence analyses, we found that BACH1 represses ferritin genes in HNSCC cells. BACH1 knockdown was found to enhance the effect of tipifarnib, a selective inhibitor of farnesyltransferase required for HRAS activation, resulting in efficient inhibition of proliferation of HNSCC cell lines in vitro. These results indicate that BACH1 is essential for maintaining the amount of labile iron in HNSCC cells to evade metabolic cell death of iron deficiency and to proliferate. BACH1 can be a target for conferring Tipifarnib sensitivity to HNSCC cells.

cancer biology↗

bioRxiv2025

Modular architecture confers robustness to damage and facilitates recovery in spiking neural networks modeling in vitro neurons

Impaired brain function is restored following injury through dynamic processes that involve synaptic plasticity. This restoration is supported by the brains inherent modular organization, which promotes functional separation and redundancy. However, it remains unclear how the modular structure interacts with synaptic plasticity, most notably in the form of spike-timing-dependent plasticity (STDP), to define the damage response and recovery efficiency. In this work, we numerically modeled the response and recovery to damage of a neuronal network in vitro bearing a modular structure. Consistent with the in vitro observations, the in silico numerical model effectively captured the decline and subsequent recovery of spontaneous activity following the injury. We revealed that the modular structure confers robustness to injury, minimizes the decrease in neuronal activity, and promotes recovery via STDP. Finally, using the reservoir computing framework, we show that information representation in the neuronal network improves with the recovery of synchronous activity. Our work provides an experimental-numerical platform for predicting recovery in damaged neuronal networks and may help developing effective models for brain injury.

neuroscience↗

bioRxiv2024

A Mutation in Tmem135 Causes Progressive Sensorineural Hearing Loss

Transmembrane protein 135 (TMEM135) is a 52 kDa protein with five predicted transmembrane domains that is highly conserved across species. Previous studies have shown that TMEM135 is involved in mitochondrial dynamics, thermogenesis, and lipid metabolism in multiple tissues; however, its role in the inner ear or the auditory system is unknown. We investigated the function of TMEM135 in hearing using wild-type (WT) and Tmem135FUN025/FUN025 (FUN025) mutant mice on a CBA/CaJ background, a normal-hearing mouse strain. Although FUN025 mice displayed normal auditory brainstem response (ABR) at 1 month, we observed significantly elevated ABR thresholds at 8, 16, and 64 kHz by 3 months, which progressed to profound hearing loss by 12 months. Consistent with our auditory testing, 13-month-old FUN025 mice exhibited a severe loss of outer hair cells and spiral ganglion neurons in the cochlea. Our results using BaseScope in situ hybridization indicate that TMEM135 is expressed in the inner hair cells, outer hair cells, and supporting cells. Together, these results demonstrate that the FUN025 mutation in Tmem135 causes progressive sensorineural hearing loss, and suggest that TMEM135 is crucial for maintaining key cochlear cell types and normal sensory function in the aging cochlea.

neuroscience↗