The brain's high energy demand is largely met by ATP produced through oxidative phosphorylation (OxPhos) in mitochondria. Mitochondria influence various aspects of brain function, from neuronal branching to neurotransmitter release. Energetic constraints suggest that understanding brain mitochondrial biology is crucial for understanding brain function and behavior. Mitochondria are not uniform; their molecular composition, morphology, and function vary depending on cellular demands. Therefore, studying the mitochondria-behavior relationship requires assessing multiple mitochondrial features across multiple brain areas. This study aims to map mitochondria-to-behavior associations across multiple brain regions in mice, leveraging naturally occurring and experimentally induced behavioral and mitochondrial variations. This approach allows for identifying potential brain "mitochondrial networks", similar to the concept of large-scale brain circuitry and metabolism. Previous research has demonstrated links between mitochondrial function and various behavioral domains, including working memory, social dominance, and anxiety. Mitochondrial RC defects are implicated in human neurological and psychiatric disorders, and in vivo brain metabolic imaging studies show that energy metabolism predicts cognitive performance and anxiety. While the importance of mitochondria for brain structure and function is clear, a systematic understanding of area-specific differences in mitochondrial biology and their relationship to behavior is lacking. This research addresses this gap by miniaturizing assays for mitochondrial OxPhos enzyme activities and applying them across multiple brain areas in mice with diverse behavioral phenotypes. Network-based connectivity analysis, combined with brain-wide gene expression data, will be used to explore the distribution of mitochondrial phenotypes and their link to behavior.

A substantial body of research indicates that the brain operates under significant energetic constraints, with oxidative phosphorylation (OxPhos) within mitochondria serving as the primary energy source. Mitochondrial function is linked to various aspects of brain development and cell biology, including neuronal morphology, gene expression, neurotransmitter release, and synaptic plasticity. Studies have also shown a connection between mitochondrial variation and animal behavior, with naturally occurring differences and experience-dependent changes in mitochondrial function. For example, chronic stress paradigms, like chronic social defeat or corticosterone exposure, have been shown to alter mitochondrial function in specific brain areas, potentially linking mitochondrial biology to behavior. Moreover, direct causal experiments highlight that mitochondrial respiratory chain (RC) enzyme activities influence specific behavioral domains, including working memory, social dominance, and anxiety. Defects in mitochondrial function are also implicated in several human neurological and psychiatric disorders. In vivo brain metabolic imaging studies have shown a correlation between energy metabolism in specific brain regions, such as the nucleus accumbens, and cognitive performance as well as anxiety levels. However, a comprehensive, brain-wide investigation of area-specific mitochondrial biology in relation to behavior is lacking. The current study fills this knowledge gap.

This study involved miniaturizing existing spectrophotometric assays for mitochondrial respiratory chain (RC) enzyme activities (Complex I, II, IV, and Citrate Synthase) and mtDNA content quantification. These assays were optimized for sub-milligram tissue samples, using 96-well plates to allow for high-throughput analysis. The study utilized a cohort of inbred male mice exhibiting a wide range of behavioral phenotypes and mitochondrial features, augmented by subgroups exposed to chronic corticosterone (CORT) treatment or chronic social defeat stress (CSDS). Behavioral phenotypes were assessed using four tests: open-field test (OFT), elevated plus maze (EPM), novelty suppressed feeding (NSF), and social interaction (SI) tests. A total of 571 samples across 17 brain areas and 5 peripheral tissues were collected and analyzed. Mitochondrial health index (MHI), an integrative measure of energy transformation capacity, was computed. To identify brain areas with similar mitochondrial phenotypes, a correlation-based similarity matrix was generated and subjected to multi-slice community detection analysis. This analysis, which incorporated six layers of mitochondrial features, identified three large-scale mitochondrial networks: a cortico-striatal network, a salience/spatial navigation network, and a threat response network. The study also integrated gene expression data from the Allen Mouse Brain Atlas to examine the overlap between mitochondrial networks, gene co-expression patterns, and structural connectivity, using metrics like strength fraction and quality of modularity. Finally, the study examined the molecular specificity of the behaviorally relevant cortico-striatal network by analyzing gene expression patterns from the Allen Mouse Brain Atlas, focusing on mitochondrial genes and pathways to identify mitotypes.

The study revealed substantial animal-to-animal variation in mitochondrial phenotypes across brain areas (average coefficient of variation of 36%). Both CORT and CSDS treatments altered mitochondrial phenotypes in an area-specific manner, with CORT tending to increase and CSDS tending to decrease activities. Topological data analysis (TDA) revealed differences in regional mitochondrial recalibrations between CORT and CSDS groups, with CORT producing more area-specific changes and CSDS causing a more integrated response. Analysis of mitochondria-behavior associations showed that MHI in several brain areas correlated significantly with OFT, EPM, and SI scores but not NSF. The strongest correlations were observed for CII in the primary motor cortex and OFT (r = 0.51) and for MHI in the nucleus accumbens and EPM (r = 0.92). Importantly, correlations were stronger for brain mitochondria than for peripheral tissues. A correlation-based connectivity analysis revealed significant positive connectivity among brain areas for mitochondrial features (average r = 0.22), with the cerebellum exhibiting the highest connectivity. Multi-slice community detection analysis revealed three distinct mitochondrial networks: a cortico-striatal network, a salience/spatial navigation network, and a threat response network. The cortico-striatal network showed the strongest correlations with behaviors (OFT, EPM, SI), accounting for up to 50% of variance in EPM anxiety-like behavior. Gene expression analysis from the Allen Mouse Brain Atlas revealed distinct molecular mitochondrial phenotypes (mitotypes) for each network, with network 1 showing enrichment for synaptic signaling, neuronal morphogenesis, and enzyme regulation, while under-expressing metabolic processes and oxygen sensing. These findings converged across multiple modalities, indicating the existence of behaviorally relevant mitochondrial networks in the mouse brain.

This study provides compelling evidence for the existence of behaviorally relevant mitochondrial networks in the mouse brain. The strong correlations between mitochondrial phenotypes, particularly within the cortico-striatal network, and anxiety-like behaviors highlight the importance of mitochondrial function in regulating behavior. The findings extend previous research by demonstrating that these relationships are not limited to isolated brain areas but are distributed across functional networks. This research supports the concept of distributed mitochondrial networks, analogous to the distributed nature of neural networks. Several factors could contribute to the observed diversity of mitochondrial phenotypes across brain areas, including differences in neuronal circuit activity, neuroendocrine receptor densities, and metabolic substrates. The study also underscores the importance of using multiple mitochondrial measures, rather than relying on a single feature like mtDNA copy number, to understand mitochondrial phenotypes. The identified mitochondrial networks show significant overlap with known neural circuits and gene expression patterns, providing convergent evidence for their functional significance. The study's limitations include the use of inbred male mice, potentially limiting the generalizability of findings to females and other strains. Future research should investigate the underlying mechanisms driving mitochondrial network organization and their interaction with other biological systems.

This study demonstrates that brain mitochondrial phenotypes are diverse and organized into behaviorally relevant networks. The cortico-striatal network exhibited the strongest correlation with anxiety-like behavior, highlighting the importance of mitochondrial function in regulating behavior. The findings suggest that future research should investigate the mechanisms underlying the formation and function of these networks and the role of mitochondria in various neurological and psychiatric disorders. The study also calls for further research examining the role of sex and genetic background on mitochondrial phenotype variability and its effect on behavior.

The study used a cohort of inbred male mice, limiting the generalizability of results to female mice and other strains. The use of a limited number of behavioral tests may not fully capture the complexity of behavior. Additionally, the study utilized indirect markers of mitochondrial content, which may not fully reflect the dynamic *in vivo* state of mitochondria. Finally, the study used a specific mouse strain and a certain age group, which might not be representative of the entire mouse population.