Coupling Electrophysiology with Single Cell RNA Sequencing to Uncover Transcriptional Markers of Chemosensitive Serotonergic Neurons

Abstract

Central respiratory chemoreceptors are specialized neurons with intrinsic sensitivity to hypercapnia and/or acidosis that couple breathing and pH/CO2 levels. Previous data support the hypothesis that a sub‐population of brainstem serotonin (5‐HT) neurons serve as central respiratory chemoreceptors, where they become sensitive to pH/CO2 on or after post‐natal day 12 (P12) in rats. However, to date it remains unclear which 5‐HT neurons develop chemosensitivity and what molecular markers may identify this unique sub‐population of 5‐HT neurons. To address these questions, we employed a “patch‐to‐seq” technique to first distinguish 5‐HT neurons as chemosensitive (or insensitive) using electrophysiologic recording, and then to isolate each neurons’ RNA for single cell RNA Sequencing to identify potential cellular sensing mechanisms and/or unique transcriptional markers in each phenotypically‐identified 5‐HT neuron population. Acute brainstem slices (200 μm) were made from young (P18–23) transgenic rats which express eGFP in all 5‐HT neurons (SSeGFP), and action potential firing rates measured via a cell‐attached patch configuration in eGFP+ 5‐HT neurons during superfusion with artificial CSF (aCSF) containing inhibitors for synaptic blockade (10 mM CNQX, 50 mM D‐AP5, 20 mM Gabazine) bubbled with either 5% CO2 (bal. O2; pH= 7.36; 5 min) or 15% CO2 (bal. O2; pH= 7.10). Cellular phenotypes were assessed by calculating a Chemosensitivity Index (C.I.). pH‐sensitive eGFP+ neurons (n=7) had an average C.I. of 177.0±14.6 whereas pH‐insensitive eGFP+ neurons (n=9) had an average C.I. of 98.0 ± 2.9 (P < 0.0001). Cellular contents of each recorded cell were collected and subjected to single cell RNA Sequencing (scRNA‐Seq), from which an average of 19.3M reads were generated with a quality score of ~33.7 and ~83% mapping rate per sample (32,883 known genes quantified). Preliminary analyses indicate very few differentially expressed genes (q < 0.05) between pH‐sensitive and ‐ insensitive 5‐HT neuron populations, and of those differentially expressed no gene obviously contributes to pH regulation or sensing. In addition, unbiased hierarchical clustering and principal component analyses failed to clearly separate samples by phenotype. However, a supervised analysis using Support Vector Machine learning and Recursive Feature Elimination (SVM‐RFE) identified several genes in which expression levels significantly predict the predefined phenotype, but the biological/functional significance of these genes in CO2/pH sensing remains unclear and instead these genes may provide a molecular signature to distinguish pH/CO2 sensitive vs insensitive 5‐HT neurons. We conclude from these preliminary data that pH/CO2 sensitivity in 5‐HT neurons is not a transcriptionally regulated phenotype.Support or Funding InformationParker B. Francis FoundationNHLBI ‐ R01 HL122358