Explore the vast range of titles available.

Dynamic change of mitochondrial morphology and distribution along neuronal branches are essential for neural circuitry formation and synaptic efficacy. However, the underlying mechanism remains elusive. We show here that Pink1 knockout (KO) mice display defective dendritic spine maturation, reduced axonal synaptic vesicles, abnormal synaptic connection, and attenuated long-term synaptic potentiation (LTP). Drp1 activation via S616 phosphorylation rescues deficits of spine maturation in Pink1 KO neurons. Notably, mice harboring a knockin (KI) phosphor-null Drp1 S616A recapitulate spine immaturity and synaptic abnormality identified in Pink1 KO mice. Chemical LTP (cLTP) induces Drp1 S616 phosphorylation in a PINK1-dependent manner. Moreover, phosphor-mimetic Drp1 S616D restores reduced dendritic spine localization of mitochondria in Pink1 KO neurons. Together, this study provides the first in vivo evidence of functional regulation of Drp1 by phosphorylation and suggests that PINK1-Drp1 S616 phosphorylation coupling is essential for convergence between mitochondrial dynamics and neural circuitry formation and refinement.

Background Autosomal dominant optic atrophy (ADOA), a leading common inherited optic neuropathy, arises from progressive retinal ganglion cell degeneration, often linked to OPA1 mutations. OPA1, a mitochondrial GTPase, regulates mitochondrial fusion, crista structure, and apoptosis. While GTPase-related dysfunction is well-studied, the role of other OPA1 domains in ADOA pathology remains unclear. Methods To investigate ADOA-linked OPA1 mutations, we assessed mitochondrial morphology, membrane potential, cytochrome c release, and cell viability in primary cortical neurons and N2a cells expressing OPA1 wild-type or mutant constructs. RNA sequencing and structural predictions (SWISS-MODEL) provided insights into molecular pathways and structural impacts. Results Two ADOA-associated mutations were characterized: V465F (GTPase β-fold) and V560F (BSE α-helix). Both mutations impaired mitochondrial fusion and cell survival under apoptotic stimuli. Notably, the BSE-located V560F mutation caused greater deficits in membrane potential maintenance, earlier apoptosis, and distinct molecular pathway changes compared to V465F. Conclusions This study highlights the domain-specific impacts of OPA1 mutations on mitochondrial function and ADOA pathology, revealing unique roles of the BSE domain in apoptosis regulation and mitochondrial integrity. These findings provide insights into ADOA mechanisms and potential therapeutic targets.

Interfacial magnetism in functional oxide heterostructures not only exhibits intriguing physical phenomena but also implies great potential for device applications. In these systems, interfacial structural and electronic reconstructions are essential for improving the stability and tunability of the magnetic properties. In this work, we constructed ultra-thin La 0.67 Ca 0.33 MnO 3 (LCMO) and SrRuO 3 (SRO) layers into superlattices, which exhibited a robust ferromagnetic phase. The high Curie temperature ( T C ) reaches 291 K, more than 30 K higher than that of bulk LCMO. We found that the LCMO/SRO superlattices consisted of atomically-sharp and asymmetric heterointerfaces. Such a unique interface structure can trigger a sizable charge transfer as well as a ferroelectric-like polar distortion. These two interfacial effects cooperatively stabilized the high- T C ferromagnetic phase. Our results could pave a promising approach towards effective control of interfacial magnetism and new designs of oxide-based spintronic devices.

Background Amyotrophic lateral sclerosis (ALS) is a devastating progressive neurodegenerative disease that affects neurons in the central nervous system and the spinal cord. As in many other neurodegenerative disorders, the genetic risk factors and pathogenesis of ALS involve dysregulation of cytoskeleton and neuronal transport. Notably, sensory and motor neuron diseases such as hereditary sensory and autonomic neuropathy type 2 (HSAN2) and spastic paraplegia 30 (SPG30) share several causative genes with ALS, as well as having common clinical phenotypes. KIF1A encodes a kinesin 3 motor that transports presynaptic vesicle precursors (SVPs) and dense core vesicles and has been reported as a causative gene for HSAN2 and SPG30. Methods Here, we analyzed whole-exome sequencing data from 941 patients with ALS to investigate the genetic association of KIF1A with ALS. Results We identified rare damage variants (RDVs) in the KIF1A gene associated with ALS and delineated the clinical characteristics of ALS patients with KIF1A RDVs. Clinically, these patients tended to exhibit sensory disturbance. Interestingly, the majority of these variants are located at the C-terminal cargo-binding region of the KIF1A protein. Functional examination revealed that the ALS-associated KIF1A variants located in the C-terminal region preferentially enhanced the binding of SVPs containing RAB3A, VAMP2, and synaptophysin. Expression of several disease-related KIF1A mutants in cultured mouse cortical neurons led to enhanced colocalization of RAB3A or VAMP2 with the KIF1A motor. Conclusions Our study highlighted the importance of KIF1A motor-mediated transport in the pathogenesis of ALS, indicating KIF1A as an important player in the oligogenic scenario of ALS.

Deep learning has revolutionized the field of artificial intelligence. Based on the statistical correlations uncovered by deep learning-based methods, computer vision tasks, such as autonomous driving and robotics, are growing rapidly. Despite being the basis of deep learning, such correlation strongly depends on the distribution of the original data and is susceptible to uncontrolled factors. Without the guidance of prior knowledge, statistical correlations alone cannot correctly reflect the essential causal relations and may even introduce spurious correlations. As a result, researchers are now trying to enhance deep learning-based methods with causal theory. Causal theory can model the intrinsic causal structure unaffected by data bias and effectively avoids spurious correlations. This paper aims to comprehensively review the existing causal methods in typical vision and vision-language tasks such as semantic segmentation, object detection, and image captioning. The advantages of causality and the approaches for building causal paradigms will be summarized. Future roadmaps are also proposed, including facilitating the development of causal theory and its application in other complex scenarios and systems.

Proper regulation of synapse formation and elimination is critical for establishing mature neuronal circuits and maintaining brain function. Synaptic abnormalities, such as defects in the density and morphology of postsynaptic dendritic spines, underlie the pathology of various neuropsychiatric disorders. Protocadherin 17 (PCDH17) is associated with major mood disorders, including bipolar disorder and depression. However, the molecular mechanisms by which PCDH17 regulates spine number, morphology, and behavior remain elusive. In this study, we found that PCDH17 functions at postsynaptic sites, restricting the number and size of dendritic spines in excitatory neurons. Selective overexpression of PCDH17 in the ventral hippocampal CA1 results in spine loss and anxiety- and depression-like behaviors in mice. Mechanistically, PCDH17 interacts with actin-relevant proteins and regulates actin filament (F-actin) organization. Specifically, PCDH17 binds to ROCK2, increasing its expression and subsequently enhancing the activity of downstream targets such as LIMK1 and the phosphorylation of cofilin serine-3 (Ser3). Inhibition of ROCK2 activity with belumosudil (KD025) ameliorates the defective F-actin organization and spine structure induced by PCDH17 overexpression, suggesting that ROCK2 mediates the effects of PCDH17 on F-actin content and spine development. Hence, these findings reveal a novel mechanism by which PCDH17 regulates synapse development and behavior, providing pathological insights into the neurobiological basis of mood disorders.

Dynamic change of mitochondrial morphology and distribution along neuronal branches are essential for neural circuitry formation and synaptic efficacy. However, the underlying mechanism remains elusive. We show here that Pink1 knockout (KO) mice display defective dendritic spine maturation, reduced axonal synaptic vesicles, abnormal synaptic connection, and attenuated long-term synaptic potentiation (LTP). Drp1 activation via S616 phosphorylation rescues deficits of spine maturation in Pink1 KO neurons. Notably, mice harboring a knockin (KI) phosphor-null Drp1 recapitulate spine immaturity and synaptic abnormality identified in Pink1 KO mice. Chemical LTP (cLTP) induces Drp1 phosphorylation in a PINK1-dependent manner. Moreover, phosphor-mimetic Drp1 restores reduced dendritic spine localization of mitochondria in Pink1 KO neurons. Together, this study provides the first in vivo evidence of functional regulation of Drp1 by phosphorylation and suggests that PINK1-Drp1 phosphorylation coupling is essential for convergence between mitochondrial dynamics and neural circuitry formation and refinement.

This study investigates small-scale pretraining for Small Language Models (SLMs) to enable efficient use of limited data and compute, improve accessibility in low-resource settings and reduce costs. To enhance long-context extrapolation in compact models, we focus on Infini-attention, which builds a compressed memory from past segments while preserving local attention. In our work, we conduct an empirical study using 300M-parameter LLaMA models pretrained with Infini-attention. The model demonstrates training stability and outperforms the baseline in long-context retrieval. We identify the balance factor as a key part of the model performance, and we found that retrieval accuracy drops with repeated memory compressions over long sequences. Even so, Infini-attention still effectively compensates for the SLM's limited parameters. Particularly, despite performance degradation at a 16,384-token context, the Infini-attention model achieves up to 31% higher accuracy than the baseline. Our findings suggest that achieving robust long-context capability in SLMs benefits from architectural memory like Infini-attention.

Time series analysis is a vital task with broad applications in various domains. However, effectively capturing cross-dimension and cross-time dependencies in non-stationary time series poses significant challenges, particularly in the context of environmental factors. The spurious correlation induced by the environment confounds the causal relationships between cross-dimension and cross-time dependencies. In this paper, we introduce a novel framework called Caformer (\\underline{\\textbf{Ca}}usal Trans\\underline{\\textbf{former}}) for time series analysis from a causal perspective. Specifically, our framework comprises three components: Dynamic Learner, Environment Learner, and Dependency Learner. The Dynamic Learner unveils dynamic interactions among dimensions, the Environment Learner mitigates spurious correlations caused by environment with a back-door adjustment, and the Dependency Learner aims to infer robust interactions across both time and dimensions. Our Caformer demonstrates consistent state-of-the-art performance across five mainstream time series analysis tasks, including long- and short-term forecasting, imputation, classification, and anomaly detection, with proper interpretability.

Accurate predictions of spatio-temporal systems are crucial for tasks such as system management, control, and crisis prevention. However, the inherent time variance of many spatio-temporal systems poses challenges to achieving accurate predictions whenever stationarity is not granted. In order to address non-stationarity, we propose a Distribution and Relation Adaptive Network (DRAN) capable of dynamically adapting to relation and distribution changes over time. While temporal normalization and de-normalization are frequently used techniques to adapt to distribution shifts, this operation is not suitable for the spatio-temporal context as temporal normalization scales the time series of nodes and possibly disrupts the spatial relations among nodes. In order to address this problem, a Spatial Factor Learner (SFL) module is developed that enables the normalization and de-normalization process. To adapt to dynamic changes in spatial relationships among sensors, we propose a Dynamic-Static Fusion Learner (DSFL) module that effectively integrates features learned from both dynamic and static relations through an adaptive fusion ratio mechanism. Furthermore, we introduce a Stochastic Learner to capture the noisy components of spatio-temporal representations. Our approach outperforms state-of-the-art methods on weather prediction and traffic flow forecasting tasks.Experimental results show that our SFL efficiently preserves spatial relationships across various temporal normalization operations. Visualizations of the learned dynamic and static relations demonstrate that DSFL can capture both local and distant relationships between nodes.