Immunoprecipitation

(See Supplementary Protocol 1 for detailed Standard Operating Procedures for ENCODE-style eCLIP experiments, including oligonucleotide sequences, catalog numbers for all reagents, and specific details for eCLIP experiments). RNA binding protein (RBP)-RNA interactions were stabilized with UV crosslinking (254 nm, 400 mJ/cm²), followed by lysis in iCLIP lysis buffer, limited digestion with RNase I (Ambion), immunoprecipitation of RBP-RNA complexes with a specific primary antibody of interest using magnetic beads with pre-coupled secondary antibody (typically M-280 Sheep Anti-Rabbit IgG Dynabeads, ThermoFisher Scientific 11204D), and stringent washes. After dephosphorylation with FastAP (ThermoFisher) and T4 PNK (NEB), a barcoded RNA adapter was ligated to the 3′ end (T4 RNA Ligase, NEB) (at this step, multiple replicates of the same RBP, or potentially RBPs of similar size and bound RNA amount, can be uniquely barcoded and pooled after ligation to simplify downstream steps - see Supplementary Fig. 2a). Ligations were performed on-bead (to allow washing away unincorporated adapter) in high concentration of PEG8000, which improves ligation efficiency to > 90%. Samples were then run on standard protein gels and transferred to nitrocellulose membranes, and a region 75 kDa (~150 nt of RNA) above the protein size was isolated and proteinase K (NEB) treated to isolate RNA. RNA was reverse transcribed with AffinityScript (Agilent), and treated with ExoSAP-IT (Affymetrix) to remove excess oligonucleotides. A second DNA adapter (containing a random-mer of 5 (N₅) or 10 (N₁₀) random bases at the 5′ end) was then ligated to the cDNA fragment 3′ end (T4 RNA Ligase, NEB), performed with high concentration of PEG8000 (to improve ligation efficiency) and DMSO (to decrease inhibition of ligation due to secondary structure). After cleanup (Dynabeads MyOne Silane, ThermoFisher), an aliquot of each sample was first subjected to qPCR (to identify the proper number of PCR cycles), and then the remainder was PCR amplified (Q5, NEB) and size selected via agarose gel electrophoresis. Samples were sequenced on the Illumina HiSeq 2500 or 4000 platform as two Paired End 50bp (for N₅) or 55bp (for N₁₀) reads. All analyses were performed using identical antibody lots for RBFOX2 (A300-864A lot 002, Bethyl), SLBP (RN045P lot 001, MBL International), and IgG Isotype Control (02-6102 lot 32013, Thermo Fisher Scientific). SLBP experiments were performed with 20×10⁶ cells and 10 ug of primary antibody; RBFOX2 experiments were performed with 20×10⁶ cells and 10 ug (eCLIP Rep1 and Rep2) or 10×10⁶ cells and 5 ug (RNase I variation experiments). All experiments in K562 and HepG2 cells were performed with 20×10⁶ cells and 10 ug of indicated primary antibody (Supplementary Table 2). Antibody validation documentation (including Western images of immunoprecipitation and shRNA knockdown^{19 (link)}) are available at http://www.encodeproject.org/. Additional experiments performed in K562 and HepG2 cells in which the antibody failed to successfully immunoprecipitate the targeted RBP were excluded from analysis. 293T cells were obtained from Clontech (Lenti-X 293T cell line). K562 and HepG2 cells were purchased from ATCC, and were not independently verified. Cells were routinely tested for mycoplasma using MycoAlert PLUS (Lonza).