Leclerc, G. | bioRxiv Science

bioRxiv2023

A single-cell atlas of ABCA7 loss-of-function reveals lipid disruptions, mitochondrial dysfunction and DNA damage in neurons

Loss-of-function (LoF) variants in the lipid transporter ABCA7 significantly increase Alzheimers disease risk (odds ratio {approx} 2), yet the underlying pathogenic mechanisms and specific neural cell types affected remain unclear. To investigate this, we generated a single-nucleus RNA sequencing atlas of 36 human postmortem prefrontal cortex samples, including 12 carriers of ABCA7 LoF variants and 24 matched non-carriers. ABCA7 LoF variants were associated with transcriptional changes across all major neural cell types. Excitatory neurons, which expressed the highest levels of ABCA7, showed significant alterations in oxidative phosphorylation, lipid metabolism, DNA damage responses, and synaptic signaling pathways. ABCA7 LoF-associated transcriptional changes in neurons were similarly perturbed in carriers of the common AD missense variant ABCA7 p.Ala1527Gly (n = 240 controls, 135 carriers) predicted by molecular dynamic simulations to disrupt ABCA7 structure -, indicating that findings from our study may extend to large portions of the at-risk population. Human induced pluripotent stem cell (iPSC)-derived neurons carrying ABCA7 LoF variants closely recapitulated the transcriptional changes observed in human postmortem neurons. Biochemical experiments further demonstrated that ABCA7 LoF disrupts mitochondrial membrane potential via regulated uncoupling, increases oxidative stress, and alters phospholipid homeostasis in neurons, notably elevating saturated phosphatidylcholine levels. Supplementation with CDP-choline to enhance de novo phosphatidylcholine synthesis effectively reversed these transcriptional changes, restored mitochondrial uncoupling, and reduced oxidative stress. Additionally, CDP-choline normalized amyloid-{beta} secretion and alleviated neuronal hyperexcitability in ABCA7 LoF neurons. This study provides a detailed transcriptomic profile of ABCA7 LoF-induced changes and highlights phosphatidylcholine metabolism as a key driver in ABCA7-induced risk. Our findings suggest a promising therapeutic approach that may benefit a large proportion of individuals at increased risk for Alzheimers disease.

neuroscience↗

bioRxiv2022

Model metamers illuminate divergences between biological and artificial neural networks

Deep neural network models of sensory systems are often proposed to learn representational transformations with invariances like those in the brain. To reveal these invariances we generated "model metamers" - stimuli whose activations within a model stage are matched to those of a natural stimulus. Metamers for state-of-the-art supervised and unsupervised neural network models of vision and audition were often completely unrecognizable to humans when generated from deep model stages, suggesting differences between model and human invariances. Targeted model changes improved human-recognizability of model metamers, but did not eliminate the overall human-model discrepancy. The human-recognizability of a models metamers was well predicted by their recognizability by other models, suggesting that models learn idiosyncratic invariances in addition to those required by the task. Metamer recognition dissociated from both traditional brain-based benchmarks and adversarial vulnerability, revealing a distinct failure mode of existing sensory models and providing a complementary benchmark for model assessment.

neuroscience↗