Ribosome profiling

Ribosome profiling libraries were prepared with the TruSeq Ribo Profile (Mammalian) Kit (Illumina). RNA-seq libraries for normalizing ribosomal footprints were prepared in parallel according to kit instructions. Biological duplicates of cells stably integrated with miR-1, miR-122 or the control were induced with 1 μg/ml doxycycline for seven days. To stall ribosomes, cells in 10 cm plates were incubated in media supplemented with 100 μg/ml cycloheximide for two minutes at 37°C. Cells were washed in ice-cold PBS supplemented with 100 μg/ml cycloheximide, lifted from the plates, pelleted and lysed in 750 μl mammalian lysis buffer on ice for 10 min. Lysate was clarified and split into two tubes: (i) 100 μl for preparing total RNA libraries and (ii) 200 μl for preparing ribosome footprint libraries. Library preparation was performed according to the protocol. Ribosomes were treated with 60 U TruSeq Ribo Profile Nuclease to generate ribosome-protected fragments and isolated via size-exclusion with an Illustra MicroSpin S-400 HR column (GE healthcare). Ribosomal RNA was depleted from both ribosome-protected fragments and RNA-seq libraries using the Illumina Ribo-Zero Gold Kit (Human/Mouse/Rat) according to the protocol. Libraries were sequenced on the Illumina NextSeq 500 with the 75 bp kit.

The raw reads were trimmed to remove adapter sequences using cutadapt v1.8.3 (-a AGATCGGAAGAGCACACGTC -m 18). Trimmed reads originating from rRNA were removed using Bowtie2 v2.3.5.1 with default parameters for RNA-seq datasets and with ‘-L 20’ and other default parameters for Ribo-seq datasets. Remaining reads were mapped to the hg19 genome and gencode v19 annotated genes using Tophat v.2.1.1 (–no-novel-juncs –transcriptome-index <indexFile> -p 3 –library-type fr-firststrand). Reads mapping to coding region excluding the ends (initial 45nt and ending 15nt) were counted using featureCounts (-F SAF -s 1 -Q 50 -T 10 –fracOverlap 1). The change in translational efficiency was calculated using edgeR in the same manner as the calculation for change in post-transcriptional regulation. Briefly, the first factor represented experimental conditions and the second factor represented the type of assay (ribosome profiling verses RNA-seq). The models with or without the interaction term between the two factors were compared using likelihood ratio test in edgeR framework. The q-values were computed using qvalue R package (45).