RNA-seq analysis

For each mOSN sample analyzed, 3 C57BL/6J male mice (8- to 9-weeks old) were pooled. Four replicate samples were analyzed per genotype (Upf3b^+/y; Omp-Cre; R26-eYFP and Upf3b^-/y; Omp-Cre; R26-eYFP). Cell sorting experiments were performed on two separate days, with two samples sorted per day. The OE was dissected as described (Gong, 2012) and dissociated using the Papain Dissociation System (Worthington) at 37°C for 15 min, followed by extensive trituration. Cells were filtered using a 40‐μm strainer (Falcon). After spinning at 200 g for 5 min, cells were resuspended in Hanks’ balanced salt solution (HBSS) containing 3% FBS (Gibco) but without Ca²⁺ and Mg²⁺. The cell suspension was mixed with propidium iodide (final concentration of 1 μg/ml) and the OMP-eYFP⁺ cells were sorted by flow-cytometry. RNA was isolated from the OMP-eYFP⁺ cells using TriZOL (Life Technologies), followed by a secondary purification step using a RNeasy column (Qiagen). Total RNA was assessed for quality using an Agilent Bioanalyzer, and samples determined to have an RNA Integrity Number (RIN) of at least 8 or greater were used to generate RNA libraries using Illumina's TruSeq RNA Sample Prep Kit, following the manufacturer's specifications, with the RNA fragmentation time adjusted to 5 min. RNA-seq was performed at the Institute of Genomic Medicine at UCSD. RNA libraries were multiplexed and sequenced with 100 base pair (bp) pair end reads on an Illumina HiSeq4000. The average number of reads per sample ranged from approximately 15 to 22 million reads. Reads were filtered for quality and aligned with STAR (2.5.2b) against Mus musculus release-90, Ensembl genome (GRCm38). The exon counts were aggregated for each gene to build a read count table using SubRead function featureCounts (Liao et al., 2014). Using the exon start/end positions, we extracted the exon sequences from the mm10 mouse genome, and ligated them together in silico for each transcript. For each entry, the entire transcript sequence was subtracted from the known CDS sequence (obtained as above) to identify 3’UTR length. DEGs were defined using DESeq2 (Love et al., 2014) using a threshold q-val of <0.05. The R package program ‘pheatmap’ was used for clustering and to generate heatmap plots. GO analysis was done using database for annotation, visualization and integrated discovery (DAVID), version v6.8. To infer relative RNA stability, we used the REMBRANDTS program (Alkallas et al., 2017) following the tutorial (https://github.com/csglab/REMBRANDTS).