Mitochondrial transport relies on a motor-adaptor complex containing Miro1, a mitochondrial outer membrane protein with two GTPase domains, as well as TRAK1/2, kinesin-1, and dynein. Using a peroxisome-directed Miro1, we quantified the ability of GTPase mutations to influence peroxisomal recruitment of complex components. Miro1 whose N-GTPase is locked in the GDP-state doesn’t recruit TRAK1/2, kinesin or P135 to peroxisomes whereas the GTP-state does. Miro1 C-GTPase mutations have little influence on complex recruitment. Though Miro2 is thought to support mitochondrial motility, peroxisome-directed Miro2 did not recruit the other complex components regardless of the state of its GTPase domains. Neurons expressing peroxisomal Miro1 with the GTP-state form of the N-GTPase had markedly increased peroxisomal transport to growth cones while the GDP-state caused their retention in the soma. Thus, the N-GTPase of Miro1 is critical for regulating Miro1’s interaction with the other components of the motor-adaptor complex and thereby for regulating mitochondrial motility.SummaryA Miro-containing complex mediates mitochondrial motility. Relocalizing Miro1 and 2 to peroxisomes and systematically manipulating each GTPase domain of Miro revealed the importance of the N-terminal GTPase domain of Miro1 for governing interaction with TRAK proteins, motors, and transport.

Mitochondrial transport relies on a motor–adaptor complex containing Miro1, a mitochondrial outer membrane protein with two GTPase domains, and TRAK1/2, kinesin-1, and dynein. Using a peroxisome-directed Miro1, we quantified the ability of GTPase mutations to influence the peroxisomal recruitment of complex components. Miro1 whose N-GTPase is locked in the GDP state does not recruit TRAK1/2, kinesin, or P135 to peroxisomes, whereas the GTP state does. Similarly, the expression of the MiroGAP VopE dislodges TRAK1 from mitochondria. Miro1 C-GTPase mutations have little influence on complex recruitment. Although Miro2 is thought to support mitochondrial motility, peroxisome-directed Miro2 did not recruit the other complex components regardless of the state of its GTPase domains. Neurons expressing peroxisomal Miro1 with the GTP-state form of the N-GTPase had markedly increased peroxisomal transport to growth cones, whereas the GDP-state caused their retention in the soma. Thus, the N-GTPase domain of Miro1 is critical for regulating Miro1’s interaction with the other components of the motor–adaptor complex and thereby for regulating mitochondrial motility.

This work considers the task of representation learning on the attributed relational graph (ARG). Both the nodes and edges in an ARG are associated with attributes/features allowing ARGs to encode rich structural information widely observed in real applications. Existing graph neural networks offer limited ability to capture complex interactions within local structural contexts, which hinders them from taking advantage of the expression power of ARGs. We propose Motif Convolution Module (MCM), a new motif-based graph representation learning technique to better utilize local structural information. The ability to handle continuous edge and node features is one of MCM’s advantages over existing motif-based models. MCM builds a motif vocabulary in an unsupervised way and deploys a novel motif convolution operation to extract the local structural context of individual nodes, which is then used to learn higher-level node representations via multilayer perceptron and/or message passing in graph neural networks. When compared with other graph learning approaches to classifying synthetic graphs, our approach is substantially better in capturing structural context. We also demonstrate the performance and explainability advantages of our approach by applying it to several molecular benchmarks.

This work considers the task of representation learning on the attributed relational graph (ARG). Both the nodes and edges in an ARG are associated with attributes/features allowing ARGs to encode rich structural information widely observed in real applications. Existing graph neural networks offer limited ability to capture complex interactions within local structural contexts, which hinders them from taking advantage of the expression power of ARGs. We propose motif convolution module (MCM), a new motif-based graph representation learning technique to better utilize local structural information. The ability to handle continuous edge and node features is one of MCM’s advantages over existing motif-based models. MCM builds a motif vocabulary in an unsupervised way and deploys a novel motif convolution operation to extract the local structural context of individual nodes, which is then used to learn higher level node representations via multilayer perceptron and/or message passing in graph neural networks. When compared with other graph learning approaches to classifying synthetic graphs, our approach is substantially better at capturing structural context. We also demonstrate the performance and explainability advantages of our approach by applying it to several molecular benchmarks.