Association of 2-oxoacid dehydrogenase complexes with respirasomes in mitochondria.

In the present study, cryo-electron tomography was used to investigate the localization of 2-oxoacid dehydrogenase complexes (OADCs) in cardiac mitochondria and mitochondrial inner membrane samples. Two classes of ordered OADC inner cores with different symmetries were distinguished and their quaternary structures modeled. One class corresponds to pyruvate dehydrogenase complexes and the other to dehydrogenase complexes of α-ketoglutarate and branched-chain α-ketoacids. OADCs were shown to be localized in close proximity to membrane-embedded respirasomes, as observed both in densely packed lamellar cristae of cardiac mitochondria and in ruptured mitochondrial samples where the dense packing is absent. This suggests the specificity of the OADC-respirasome interaction, which allows localized NADH/NAD+ exchange between OADCs and complex I of the respiratory chain. The importance of this local coupling is based on OADCs being the link between respiration, glycolysis and amino acid metabolism. The coupling of these basic metabolic processes can vary in different tissues and conditions and may be involved in the development of various pathologies. The present study shows that this important and previously missing parameter of mitochondrial complex coupling can be successfully assessed using cryo-electron tomography.

Convolutional Neural Network Based on Crossbar Arrays of (Co-Fe-B)x(LiNbO3)100-x Nanocomposite Memristors.

Convolutional neural networks (CNNs) have been widely used in image recognition and processing tasks. Memristor-based CNNs accumulate the advantages of emerging memristive devices, such as nanometer critical dimensions, low power consumption, and functional similarity to biological synapses. Most studies on memristor-based CNNs use either software models of memristors for simulation analysis or full hardware CNN realization. Here, we propose a hybrid CNN, consisting of a hardware fixed pre-trained and explainable feature extractor and a trainable software classifier. The hardware part was realized on passive crossbar arrays of memristors based on nanocomposite (Co-Fe-B)x(LiNbO3)100-x structures. The constructed 2-kernel CNN was able to classify the binarized Fashion-MNIST dataset with ~ 84% accuracy. The performance of the hybrid CNN is comparable to the other reported memristor-based systems, while the number of trainable parameters for the hybrid CNN is substantially lower. Moreover, the hybrid CNN is robust to the variations in the memristive characteristics: dispersion of 20% leads to only a 3% accuracy decrease. The obtained results pave the way for the efficient and reliable realization of neural networks based on partially unreliable analog elements.