Astrocytes support the energy demands of synaptic transmission and plasticity. Enduring changes in synaptic efficacy are highly sensitive to stress, yet whether changes to astrocyte bioenergetic control of synapses contributes to stress-impaired plasticity is unclear. Here we show in mice that stress constrains the shuttling of glucose and lactate through astrocyte networks, creating a barrier for neuronal access to an astrocytic energy reservoir in the hippocampus and neocortex, compromising long-term potentiation. Impairing astrocytic delivery of energy substrates by reducing astrocyte gap junction coupling with dominant negative connexin 43 or by disrupting lactate efflux was sufficient to mimic the effects of stress on long-term potentiation. Furthermore, direct restoration of the astrocyte lactate supply alone rescued stress-impaired synaptic plasticity, which was blocked by inhibiting neural lactate uptake. This gating of synaptic plasticity in stress by astrocytic metabolic networks indicates a broader role of astrocyte bioenergetics in determining how experiencedependent information is controlled.

This is a PDF file of an article that has undergone enhancements after acceptance, such as the addition of a cover page and metadata, and formatting for readability, but it is not yet the definitive version of record. This version will undergo additional copyediting, typesetting and review before it is published in its final form, but we are providing this version to give early visibility of the article. Please note that, during the production process, errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.

Photoplethysmography (PPG) is a noninvasive way to monitor various aspects of the circulatory system, and is becoming more and more widespread in biomedical processing. Recently, deep learning methods for analyzing PPG have also become prevalent, achieving state of the art results on heart rate estimation, atrial fibrillation detection, and motion artifact identification. Consequently, a need for interpretable deep learning has arisen within the field of biomedical signal processing. In this paper, we pioneer novel explanatory metrics which leverage domain-expert knowledge to validate a deep learning model. We visualize model attention over a whole testset using saliency methods and compare it to human expert annotations. Congruence, our first metric, measures the proportion of model attention within expert-annotated regions. Our second metric, Annotation Classification, measures how much of the expert annotations our deep learning model pays attention to. Finally, we apply our metrics to compare between a signal based model and an image based model for PPG signal quality classification. Both models are deep convolutional networks based on the ResNet architectures. We show that our signal-based one dimensional model acts in a more explainable manner than our image based model; on average 50.78% of the one dimensional model's attention are within expert annotations, whereas 36.03% of the two dimensional model's attention are within expert annotations. Similarly, when thresholding the one dimensional model attention, one can more accurately predict if each pixel of the PPG is annotated as artifactual by an expert. Through this testcase, we demonstrate how our metrics can provide a quantitative and dataset-wide analysis of how explainable the model is.

Interpreting rare variants remains a challenge in personal genomics, especially for disorders with several causal genes and for genes that cause multiple disorders. ZNF423 encodes a transcriptional regulatory protein that intersects several developmental pathways. ZNF423 has been implicated in rare neurodevelopmental disorders, consistent with midline brain defects in Zfp423-mutant mice, but pathogenic potential of most patient variants remains uncertain. We engineered~50 patient-derived and small deletion variants into the highlyconserved mouse ortholog and examined neuroanatomical measures for 791 littermate pairs. Three substitutions previously asserted pathogenic appeared benign, while a fourth was effectively null. Heterozygous premature termination codon (PTC) variants showed mild haploabnormality, consistent with loss-of-function intolerance inferred from human population data. In-frame deletions of specific zinc fingers showed mild to moderate abnormalities, as did low-expression variants. These results affirm the need for functional validation of rare variants in biological context and demonstrate cost-effective modeling of neuroanatomical abnormalities in mice.

No abstract