NEUROeSTIMator: Using Deep Learning to Quantify Neuronal Activation from Single-Cell and Spatial Transcriptomic Data

Abstract

Neuronal activity-dependent transcription directs molecular processes that regulate synaptic plasticity, brain circuit development, behavioral adaptation, and long-term memory. Single cell RNA-sequencing technologies (scRNAseq) are rapidly developing and allow for the interrogation of activity-dependent transcription at cellular resolution. Here, we present NEUROeSTIMator, a deep learning model that integrates signals of activation distributed throughout the broader transcriptome to estimate neuronal activation in a way that is robust against differences in species, cell type, and brain region. We demonstrate this method's ability to accurately detect neuronal activity in previously published single cell and time course studies of activity-induced gene expression. Further, using spatial transcriptomic techniques, we demonstrate the model's ability to identify patterns of learning-induced activation. In conclusion, NEUROeSTIMator is a powerful and broadly applicable tool for measuring neuronal activation, whether as a critical covariate or a primary readout of interest.

Mapping the spatial transcriptomic signature of the hippocampus during memory consolidation

“…We and others have previously demonstrated that the learning-induced early wave of gene expression peaks at this timepoint after learning [55][56][57] . We further examined the expression profiles by integrating our previous spatial transcriptomics dataset following SOR training 54 (n=3/group). We first obtained cumulative transcriptomic profiles (pseudobulk analysis, total n=7/group) by combining the hippocampal subregions CA1 pyramidal layer, CA1 stratum radiatum, CA1 stratum oriens, CA2 and CA3 pyramidal layers and DG granular layers (Fig.…”

“…Visium spatial transcriptomics ( 10X Genomics ) combines both histology and spatial profiling of RNA expression to provide high-resolution transcriptomic characterization of distinct transcriptional profiles within individual brain subregions 53 . We have recently used this Visium spatial transcriptomic approach to demonstrated neuronal activation patterns within brain regions following spatial exploration using a deep-learning computational tool 54 . In this work, we have extended this novel approach to examine activity-driven spatial transcriptomic diversity within the hippocampal network.…”

Mapping the spatial transcriptomic signature of the hippocampus during memory consolidation

“…We and others have previously demonstrated that the learning-induced early wave of gene expression peaks at this timepoint after learning 39,55,56 . We analyzed additional transcriptomic profiles by integrating our previous spatial transcriptomics dataset following SOR training 54 (GEO GSE201610, n = 3/group). Increasing the number of biological replicates can improve statistical power and improve the robustness of the results as shown previously 57,58 .…”

“…Spatial transcriptomics combines both histology and spatial profiling of RNA expression to provide high-resolution transcriptomic characterization of distinct transcriptional profiles within individual brain subregions 53 . We have recently used the spatial transcriptomic approach to demonstrate neuronal activation patterns within brain regions using a deeplearning computational tool 54 . In this work, we have applied this state-of-the-art approach to examine activity-driven spatial transcriptomic diversity within the hippocampal network.…”