A long noncoding RNA lincRNA-EPS acts as a transcriptional brake to restrain inflammation

RNA-seq was performed in biological replicates (2 mice per group) of wild-type and lincRNA-EPS-/- BMDMs at 0, 2, and 6 hours after LPS treatment (100 ng/ml). RNA-seq libraries were prepared as described (Heyer et al., 2015). Briefly, ribosomal RNA (rRNA) was depleted using the Ribozero rRNA removal kit (Epicentre). Purified mRNAs were fragmented with RNA fragmentation reagent (Life Technologies) for 4 minutes and 30 seconds at 70C to obtain 100-150 nt long RNA fragments. After ethanol precipitation and washing, RNAs were re-suspended in 5 l of water and the 3 ends dephosphorylated using PNK (New England BioLabs) for 1 hr at 37C. RNA fragments with a 3 OH were then ligated to a preadenylated DNA adaptor using T4 RNA ligase 2, truncated K227Q (NEB). Following this, ligated RNAs were reverse-transcribed with Superscript III (Invitrogen) using a bar-coded reverse-transcription primer that anneals to the preadenylated adaptor. After reverse-transcription, gel purified cDNAs were circularized using CircLigase I (Epicentre) and PCR amplified using paired-end primers PE1.0 and PE2.0 (Illumina) for 14 cycles. PCR amplicons were gel purified and submitted for sequencing on the Illumina Hiseq2000. Tophat version 2.0.12 was used to align single sequence reads to version 73 of the Ensembl mouse genome (mm10) with options: --library-type fr-firststrand -g 1 -x 1 --read-mismatches 2. Cufflinks version 2.1.1 was used to estimate RNA abundances with Ensembl version 73 GTF and the --library-type fr-firststrand option. The Cuffdiff program was used to perform differential expression analysis between wild-type BMDMs and lincRNA-EPS-/- BMDMs at each corresponding time (0,2, and 6 hours after LPS treatment). To compute fold-change values for genes/transcripts with FPKM values of zero, a pseudocount of 1 was added to all FPKM values first.