Single-cell RNA-seq

Complementary DNA (cDNA) libraries were prepared from single CRHNs as previously described (12, 29) with some modifications. Briefly, the PVN was isolated from CRH-Cre mice crossed with Ai14 mice, dissociated single cells were plated on coverslips, and single fluorescent cells were transferred to individual tubes using microcapillary pipettes (12). Using the Smart-Seq2 method (29), full-length cDNAs were prepared from mRNAs in each cell. Briefly, single cells were lysed and polyadenylated mRNA was reverse-transcribed by using a template switching method. cDNAs were then amplified using KAPA polymerase (Kapa Biosystems). After purification using AMPure XP beads (Beckman Coulter), cDNA libraries were prepared for sequencing using the Illumina TruSeq DNA Sample Prep Kit, as described previously (12). Briefly, cDNAs were fragmented to ~300 base pairs (bp), ligated to adaptors, and polymerase chain reaction (PCR)–amplified with adaptor primers.

Samples were subjected to multiplexed sequencing using an Illumina HiSeq 2500 instrument and a paired-end, 50-bp read-length sequencing strategy. Image analysis and base calling were conducted with Illumina Real-Time Analysis version 1.18 software, followed by demultiplexing of indexed reads and generation of FASTQ files with Illumina’s bcl2fastq Conversion Software (version 1.8.4). Low-quality reads were discarded before adapter trimming using Trim Galore (version 0.4.4; available at www.bioinformatics.babraham.ac.uk/projects/trim_galore/) with the options “-- adapter AAGCAGTGGTATCAACGCAGAGTAC --stringency 8 --quality 0 -e 0.15 --length 20 --paired --retain_unpaired”. The default options in GSNAP (GSNAP ref) (version 2014-12-29) (30) were used to align reads to the mouse genome (University of California Santa Cruz mm9 assembly and gene models). Gene-level counts were generated from GSNAP alignments using the Python package HTSeq (version 0.6.1) (HTSeq ref), using the “union” overlap mode (31).

We generated a list of receptor genes with Fragments Per Kilobase of transcripts per Million reads (FPKM) ≥1 in at least two cells. Lists of ligand-gated ion channels and GPCRs for neurotransmitters or neuromodulators were obtained from the International Union of Basic and Clinical Pharmacology/British Pharmacological Society website (www.guidetopharmacology.org). For previous sequencing data available in the Gene Expression Omnibus (GSE74672) [from Romanov et al. (15, 16)], we included GPCRs for neurotransmitters/neuromodulators expressed in at least 2 of 86 neurons expressing Crh (≥1 molecule per cell) out of 898 hypothalamic neurons.