Morphometric Identification of Parvalbumin-Positive Interneurons: A Data-Driven Approach

Abstract

Traditionally, anatomical studies of parvalbumin-positive (PV+) labelled interneurons describe them as a homogeneous population of neurons. In contrast, recent single-cell RNAseq studies have identified multiple transcriptomically distinct categories of PV+ cells. That difference between a single anatomical category of PV+ neurons and multiple transcriptomic categories presents a problem in understanding the role of these neurons in cortical function. One gap that might contribute to this discrepancy is that PV+ morphology is typically addressed using qualitative descriptions and simple quantifications, while single-cell RNAseq studies use big data and high dimensional analyses. PV+ neurons play critical roles in the experience-dependent development of the cortex and are often involved in disease-related changes associated with neurodegenerative and neuropsychiatric disorders. Here, we developed a modern data-driven analysis pipeline to quantify PV+ morphology. We quantified 97 morphometric features from 14274 PV+ neurons and applied unsupervised clustering that identified 13 different PV+ morphologies. We extended the analysis to compare PV+ dendritic arbour patterns and cell body morphologies. Finally, we compared the morphologies of PV+ neurons with the cell body morphologies of neurons expressing various genes associated with PV+ transcriptomic cell types. This approach identified a range of PV+ morphologies similar to the number of transcriptomic categories. It also found that the PV+ morphologies have cortical area, laminar, and transcriptomic biases that might contribute to cortical function.