Protocol for evaluating RNA-protein associations in mammalian cells with RIP-seq and RIP-qPCR

Authors: Jackson B. Trotman, Shuang Li, Quinn E. Eberhard, Zhiyue Zhang, J. Mauro Calabrese

Abstract

RNA-protein interactions drive gene regulation, subcellular organization, and noncoding RNA function. Here, we present a protocol for measuring RNA-protein associations in formaldehyde-crosslinked mammalian cells using RNA immunoprecipitation followed by sequencing (RIP-seq) and quantitative PCR (RIP-qPCR). We include steps and best practices for qualifying reagents, preparing cells, and processing and analyzing data, including a standardized set of steps to quantify signal over noise. This protocol is broadly applicable for the study of RNA-protein interactions in cells.

For complete details on the use and execution of this protocol, please refer to Trotman et al.^1^

Before you begin

We detail a protocol that employs formaldehyde crosslinking followed by immunoprecipitation to capture direct and indirect (protein-bridged) interactions between proteins and RNA transcriptome-wide. We recently used this protocol to demonstrate that the RNA-binding protein HNRNPK is required for recruiting Polycomb repressive complex 1 to C-rich regions of the long noncoding RNAs Xist, Airn, and Kcnq1ot1 in mouse embryonic stem cells (mESCs).^1^ We have successfully used this protocol to investigate RNA-protein associations in mESCs,^2^^,^^3^^,^^4^^,^^5^^,^^6^^,^^7^ mouse trophoblast stem cells,^1^^,^^6^^,^^8^ and human prostate carcinoma epithelial cells,^9^ suggesting broad applicability across mammalian cell lines. In this protocol, cells are crosslinked with formaldehyde to preserve RNA-protein and protein-protein interactions via covalent bonds between residues that situate in close proximity, and RNA is sheared into short fragments via sonication. After immunoprecipitation with a protein-specific antibody, RNA-bound beads are washed using conditions borrowed from chromatin immunoprecipitation (ChIP). Crosslinks are then reversed with heat and proteinase K treatment, and RNA is purified and reverse transcribed. The immunoprecipitated RNA is finally sequenced (RIP-seq) and/or measured by quantitative PCR (RIP-qPCR) against a standard curve prepared from input material processed in parallel.

Innovation

To our knowledge, this protocol provides the most comprehensive end-to-end guide available for designing, conducting, and analyzing RIP-based experiments. This protocol is a modification of the “fRIP” protocol originally described in Hendrickson et al.,^10^ altered to include a modified formaldehyde-crosslinking procedure and more stringent washes, which, in empirical tests, recovered higher signal-to-noise in known regions of RNA-protein interactions. The standard workflow includes the addition of ERCC spike-in RNA controls prior to library preparation and sequencing, enabling the estimation of absolute amounts of non-ribosomal RNA present in RIPs relative to controls. We additionally include a standardized set of bioinformatic steps for processing and analyzing RIP-seq data in a way that carefully considers true signal relative to the noise that is inherent to all antibody-based genomic assays. Finally, we provide guidance and details for the sensitive measurement of differences in RIP signal via input-normalized qPCR.

Planning and preparation

1.Devise an experimental plan.a.Determine the hypothesis that will be tested and carefully design experimental controls that address your hypothesis and monitor signal relative to noise.Note: RIP-seq is best-suited for broadly evaluating the RNA sequences with which a protein associates, while RIP-qPCR can more sensitively (and inexpensively) measure differences in the level of protein association at specific RNA regions upon particular treatments or genetic alterations.b.Identify what protein(s) will be analyzed and determine whether robust IP-qualified antibodies are available for them.Note: If possible, we recommend using antibodies previously demonstrated in the literature to successfully IP the protein(s) of interest. Importantly, if the results of RIPs against a single factor or set of factors will be crucial for a study, we recommend confirming antibody specificity using genetic knockouts and/or parallel RIPs with epitope-tagged proteins or antibodies targeting different epitopes of the same protein or protein complex.^5^ We further recommend supporting results with orthogonal methods. To monitor noise, specificity for the protein(s) of interest should be evaluated by comparison to RIPs with a non-specific IgG antibody and/or with the same epitope-targeting antibody (e.g., anti-FLAG) against a control such as epitope-tagged green fluorescence protein (GFP).c.For RIP-qPCR, determine the region(s) of RNA that will be analyzed.Note: This protocol employs sonication to shear RNA into ∼300-nt fragments, and qPCR signal is determined with ∼75–150-bp amplicons. We recommend evaluating specificity for each region of interest by comparing the qPCR signal to that of a negative control region that is not expected to associate with the protein being tested. RIP-seq can be used as an initial step to assess specificity and determine regions to investigate more quantitatively by RIP-qPCR if they are not previously defined.2.Obtain mESCs or other cell line(s) suitable for addressing the hypothesis. The specific protocol outlined here is for mESCs, but other mammalian cell lines can be used.3.Obtain quantities of antibodies from the same production lots sufficient for all samples and all experimental replicates. We have observed that different lots of the same antibody can IP RNA with variable efficiency.4.For RIP-seq, ensure access to a sequencing facility and a UNIX-compatible computing resource.5.For RIP-qPCR, obtain validated qPCR primers targeting the region(s) of interest (see qPCR primer design and validation (for RIP-qPCR) section below).

qPCR primer design and validation (for RIP-qPCR)

Timing: 1 week 6.Identify and obtain the sequences of RNA to be targeted from the UCSC Genome Browser or other genome/transcriptome source.7.Design and order qPCR primers. Note: We routinely use the PrimerQuest webtool from Integrated DNA Technologies (IDT), using the “2 Primers, Intercalating Dyes” option and default settings to generate suitable primer pairs targeting the sequence of interest. Ideal primers should have melting temperatures of 62°C–64°C, be 17–30 nt in length, and have 35%–65% GC content. Primer pairs should have minimal overlap to prevent primer-dimer formation and should produce an amplicon 75–150 bp in length. Note: Sequence characteristics of the region of interest may necessitate deviations from the ideal. Non-ideal primers can still be used as long as their specificity and ability to amplify the target sequence quantitatively are confirmed through empirical testing. An example of a non-ideal primer pair is JT375 and JT376 in Trotman et al.^3^ Due to experimental and sequence constraints in their design, these primers have only 10%–20% GC content and lower-than-ideal melting temperatures, necessitating an adjustment of qPCR annealing temperature from the standard 60°C to 56°C. Nonetheless, we confirmed that these primers performed robustly. CRITICAL: Primers must only amplify the target of interest. As a first pass to screen potential primer pairs, use a tool such as Primer-BLAST to confirm that, within the transcriptome of your model organism (e.g., “RefSeq RNA” option), the primer pair is predicted to target the region of interest and has no predicted off-target products under a length of 3000 bp. 8.Once primers have been obtained, pulse-spin to collect primers at the bottom of the tubes and resuspend in a new stock of ultrapure nuclease-free water to 100 μM. Also prepare 10-μM working stocks with ultrapure nuclease-free water for use in qPCR. Note: We routinely order custom DNA oligonucleotides from Integrated DNA Technologies (IDT) at 25-nmol scale and purified with standard desalting. 9.Evaluate the performance of primer pairs in qPCR.a.For a detailed qPCR protocol, see the reverse transcription and qPCR of input and RIP samples (RIP-qPCR) section below.b.For qPCR template, use a cDNA sample generated from random-oligo-primed reverse transcription of total RNA prepared from the same cells that will be used for RIP-qPCR.Note: The most experimentally relevant test would be to use cDNA prepared from a representative 5% input RNA from the RIP protocol (below), but this is not necessary here.c.As additional qPCR templates, use serial dilutions of the same cDNA (e.g., 4-fold serial dilutions).Note: qPCR signal (Cq values) from the diluted cDNAs should scale linearly over a wide range of dilutions and will ideally demonstrate a primer efficiency (i.e., the average amount of new amplicons generated per PCR cycle) of 80%–100%.d.If available, prepare parallel qPCR reaction(s) with a cDNA sample that does not contain appreciable levels of the RNA target (e.g., a no-doxycycline-induction control or gene-knockout control).Note: This provides an important specificity control for the primer pair. Also include no-template control reaction(s) containing ultrapure nuclease-free water in place of template to test for potential reagent contamination.e.Run the qPCR products on a 2% (w/v) agarose-TAE-EtBr gel to confirm that only a single band of the expected size is produced.Optional: Sanger sequencing of the qPCR product can provide further confirmation of the primer pair’s target.Note: Cq values above ∼31–32 tend to be noisy and potentially unreliable. As a result, lowly abundant RNA targets may prove difficult to study with RIP-qPCR. Take into consideration whether the Cq values from primer pair validation support the feasibility of RIP-qPCR in evaluating your RNA of interest. RIP efficiencies (i.e., the amounts of RNA recovered relative to input) vary across orders of magnitude and depend on factors such as protein abundance and antibody performance. In our experience with over a dozen antibodies in RIP-qPCR, we have observed RIP efficiencies as high as ∼10% for epitope-tagged, overexpressed proteins and as low as ∼0.01% for IgG controls, with a median of ∼0.2%. As an example, assuming ideal (100%) qPCR primer efficiency, RIPs with an antibody giving 0.2% recovery relative to input would need a maximum input-sample Cq value of 23 [32 - log2(1/0.002) = 23] for RIP-qPCR Cq values to be reliably under 32. If no amplification or weaker-than-expected amplification is observed, see troubleshootingproblem 1.

Cell culture

Timing: Approximately 1 week 10.Thaw mESC lines to be used for RIP-seq and/or RIP-qPCR.a.Pre-warm Quenching Medium and mESC Growth Medium to 37°C in a water bath.b.Coat new plastic cell culture dish(es) with gelatin by adding enough 0.1% (w/v) gelatin in 1X PBS to cover the dish (e.g., 3 mL for a 6-cm dish or 8 mL for a 10-cm dish), let incubate at 20°C–24°C for at least 1 min, and remove by aspiration.c.Remove cryo vials from liquid nitrogen storage and place in a 37°C water bath until the contents have thawed. Do not let these incubate at 37°C longer than is necessary.d.Immediately transfer cells to a 15-mL or 50-mL conical tube containing at least 7 mL pre-warmed Quenching Medium and invert to mix.Note: We prepare cryo vial stocks of mESCs in a “Freezing Medium” containing, by volume, 30% ESC-qualified serum, 7% DMSO, and 63% mESC Growth Medium containing LIF. Thawed cells should not be exposed to high concentrations of DMSO for longer than is necessary. The purpose of adding thawed cells to Quenching Medium here is to quickly reduce the concentration of DMSO.e.Pellet cells by centrifugation (200 x g, 5 min, 20°C–24°C).f.Carefully remove supernatant by aspiration and resuspend cells in pre-warmed mESC Growth Medium using multiple up-down strokes of a serological pipette. Use 4–6 mL for cells to be plated to a 6-cm dish, 10–12 mL for a 10-cm dish, and 20–25 mL for a 15-cm dish.g.Transfer cells to gelatin-coated dish(es) and place in a humidified cell-culture incubator set to 37°C and 5% CO2.CRITICAL: All cell-handling steps must be performed inside a certified biological safety cabinet (BSC) and in compliance with institutional rules and the appropriate biological safety level.CRITICAL: To prevent contamination, only handle cells and objects inside the BSC with gloved hands that have been sterilized with 70% ethanol. Sterilize gloves with 70% ethanol frequently during cell culture work. Keep biological safety cabinets closed when not in use and turn on the air flow for 5 min prior to beginning cell culture work. Spray 70% ethanol over the interior surfaces of the biological safety cabinet, including pipettes, prior to beginning cell culture work.Note: We recommend using cryogenic stocks of mESCs for thawing into 6-cm or 10-cm dishes with a density high enough to reach ∼90% confluence in 48 h. This will limit unnecessary passages and time spent waiting for cells to grow to the scale needed for crosslinking and harvesting. If cells appear dead or differentiated at any point, see troubleshootingproblem 2.11.Give cells fresh mESC Growth Medium every day and passage them every other day. Culture for at least four days before crosslinking and harvesting. As a general guideline, prior to performing end-point RIP-seq or RIP-qPCR assays, culture mESCs in the presence of leukemia inhibitory factor (LIF), off feeder cells, and on gelatin-coated tissue-culture-grade plastic dishes. To passage a.Pre-warm 0.125% trypsin, Quenching Medium, and mESC Growth Medium to 37°C in a water bath.b.Coat new plastic cell culture dish(es) with gelatin by adding enough 0.1% (w/v) gelatin in 1X PBS to cover the dish, let incubate at 20°C–24°C for at least 1 min, and remove by aspiration.c.Remove growth medium from cells by aspiration.d.Wash once with 1X PBS.e.Add just enough 0.125% (w/v) trypsin to coat the cells (e.g., 0.5 mL for a 6-cm dish or 1 mL for a 10-cm dish). Gently tilt the dish to ensure all cells have been coated. Let incubate at 20°C–24°C or 37°C for 3–5 minutes until cells have mostly dissociated.f.Add twice the volume of Quenching Medium (e.g., 1 mL for a 6-cm dish or 2 mL to a 10-cm dish) and gently tilt the dish to mix.g.Using a P1000 pipette set to 1000 μL, mix and dissociate cells with gentle up-down strokes until clusters of cells have been fully dissociated to a suspension of individual cells. This can be observed under a microscope. We typically use ∼15 strokes for a 6-cm dish and ∼30 strokes for a 15-cm dish.h.Transfer a portion of the dissociated cells to a new 15-mL or 50-mL conical tube, as appropriate for the size of the dish to be cultured into.Note: A 5 or 6 split is typical for mESCs that are at ∼90% confluence. Use more cells if cells are at a lower confluence. If “growing up” cells to a larger dish, factor in the ratio between the larger and smaller dishes’ areas. For example, the area of a 10-cm dish is roughly 3 times that of a 6-cm dish, so for an otherwise “1:6” split, 3 x (1/6) = 1/2 of the dissociated cells should be transferred to the new tube.i.Pellet cells by centrifugation (200 x g, 5 min, 20°C–24°C).j.Carefully remove supernatant by aspiration and resuspend cells in pre-warmed mESC Growth Medium using multiple up-down strokes of a serological pipette. Generally, use 4–6 mL for cells in a 6-cm dish, 10–12 mL for cells in a 10-cm dish, and 20–25 mL for cells in a 15-cm dish.k.Transfer cells to gelatin-coated dish(es) and place in a humidified cell-culture incubator set to 37°C and 5% CO2.Note: Trypsin stocks will slowly self-inactivate with time. We recommend preparing 25-mL aliquots of 0.25% (w/v) trypsin to store at −20°C that, after thawing, are brought to 0.125% by the addition of 25 mL 1X PBS and stored at 4°C. Discard 0.125% trypsin stocks ∼1 month after thawing and limit their time spent at 37°C.Optional: After dissociating cells to a single-cell suspension, the concentration of living cells can be determined by mixing a small portion (∼12 μL) of well-mixed cell suspension with an equal volume of trypan blue and counting with an accurate hemacytometer or cell counter.12.At the time of crosslinking and harvesting, cell lines will ideally be ∼90% confluent in a 10-cm or 15-cm dish without feeder cells. Note: Cells will be harvested by scraping after crosslinking in the dish and therefore will not be directly countable. To estimate the number of harvested cells, one of the two options should be followed before the day of Optional: Plate an area-proportional amount of each cell line to an additional 6-cm dish during the passage prior to harvesting. At the time of crosslinking, count trypsin-dissociated cells from the 6-cm dish as a proxy for the larger dish of cells that will be crosslinked and harvested. To estimate the number of cells in the larger dish, multiply the number of counted cells by the ratio of cells plated in the larger dish to those in the 6-cm dish. Optional: Grow a large stock of mESCs (e.g., 5 or 6 90%–100% confluent 15-cm dishes) and pellet 20, 30, 40, 50, 60, 70, and 80 million accurately counted cells in appropriately labeled 15-mL conical tubes. With a fine-point permanent marker, carefully mark the top of each cell pellet on the exterior of the tube. This set of conical tubes can be used as a visually comparative metric for estimating amounts of crosslinked, harvested, and pelleted cells in future RIP experiments.

Key resources table

REAGENT or RESOURCESOURCEIDENTIFIERAntibodiesNonspecific IgGInvitrogen02–6102SRSF1Santa Cruzsc-33652RBM15Proteintech10587-1-APChemicals, peptides, and recombinant proteinsDMEM high glucose plus sodium pyruvateGibco11995–065ESC-qualified fetal bovine serumGibco26140–079Non-essential amino acidsGibco11140–050Penicillin-streptomycinGibco15140–122L-glutamineGibco25030–081β-mercaptoethanolSigma-Aldrich63689Trypsin-EDTA, phenol redGibco252007210X Phosphate-buffered saline (PBS)Corning46-013-CMGelatinSigma-AldrichG9391TrisFisher ScientificAC327360010Hydrochloric acid (HCl)Fisher ScientificA142-212Sodium chloride (NaCl)Fisher ScientificBP358-1Ethylenediaminetetraacetic acid (EDTA)Fisher ScientificBP120-500Sodium hydroxide (NaOH) to pH EDTAFisher ScientificBP359-500EthanolDecon Labs2701Dimethyl sulfoxide (DMSO)Sigma-AldrichD2650Dithiothreitol (DTT)Thermo Fisher Scientific15508013Protease inhibitor cocktailSigma-AldrichP8340SUPERase-In RNase inhibitorInvitrogenAM2696Pierce 16% Formaldehyde (w/v), Methanol-freeThermo Fisher Scientific28906GlycineFisher ScientificBP381-1Tween 20Fisher ScientificBP337Protein A/G PLUS-Agarose beadsSanta Cruz Biotechnologysc-2003Bovine serum albumin (BSA)Fisher ScientificBP1600Potassium chloride (KCl)Fisher ScientificBP366-1Triton X-100Fisher ScientificBP151Sodium deoxycholateSigma-AldrichD6750-100GIGEPAL CA-630Fisher ScientificICN19859650Ethylene glycol-bis(β-aminoethyl ether)-N,N,N′,N′-tetraacetic acid (EGTA)MilliporeSigma324626-25GMLithium chloride (LiCl)Sigma-AldrichL4408-100GN-lauroylsarcosine sodium saltFisher ScientificAC434371000Proteinase KThermo Fisher Scientific25530049RNaseOUT RNase InhibitorThermo Fisher Scientific10777019TRIzolThermo Fisher Scientific15596026ChloroformFisher ScientificBP1145-1ERCC RNA spike-in controlsThermo Fisher Scientific4456740iTaq Universal SYBR Green SupermixBio-Rad17251241 Kb Plus DNA LadderThermo Fisher ScientificMPK10025Apex Quick Dissolve LE AgaroseGenesee Scientific20-102QDGlacial acetic acidSigma-AldrichA6283-1LEthidium bromide (EtBr, 1% solution)Fisher ScientificBP1302-10Critical commercial assaysQubit dsDNA Broad-Range KitThermo Fisher ScientificQ32853KAPA RNA HyperPrep Kit with RiboErase (HMR)Roche08098140702KAPA Unique Dual-Indexed AdaptersRoche08861919702RNA Clean & Concentrator-5 KitZymoR1013Applied Biosystems High-Capacity cDNA Reverse Transcription KitThermo Fisher Scientific4368814Deposited dataGENCODE Release M25 (GRCm38.p6) Mus musculus reference genome GTF filehttps://www.gencodegenes.org/mouse/release_M25.htmlgencode.vM25.basic.annotation.gtf.gzGENCODE Release M25 (GRCm38.p6) Mus musculus reference genome FASTA filehttps://www.gencodegenes.org/mouse/release_M25.htmlGRCm38.p6.genome.fa.gzExperimental Cell linesE14 mouse ESCs (ES-E14TG2a)ATCCCRL-1821OligonucleotidesJT356 Illumina F (AATGATACGGCGACCACCGA)IDTN/AJT357 Illumina R (CAAGCAGAAGACGGCATACGA)IDTN/ASB001 Repeat A F (GCCACGGATACCTGTGTGTC)IDTN/ASB002 Repeat A R (CCCAGATGGGCAAGTTTAGA)IDTN/ASoftware and algorithmsPrimerQuest ToolIDThttps://www.idtdna.com/pages/tools/primerquestPrimer-BLASTNCBIhttps://www.ncbi.nlm.nih.gov/tools/primer-blast/OligoAnalyzer ToolIDThttps://www.idtdna.com/pages/tools/oligoanalyzerSTARDobin et al.^11^v2.7.11bSAMtoolsLi et al.^12^v1.21UCSC Genome BrowserUCSChttps://genome.ucsc.edu/MACS2Zhang et al.^13^v2.2.7.1BEDToolsQuinlan and Hall^14^v2.31.1MEME SuiteBailey et al.^15^v5.5.7Perl script bigsam_to_wig_mm10_wcigar2.plThis studyhttps://github.com/CalabreseLab/Protocol-for-evaluating-RNA-protein-associations-in-mammalian-cells-with-RIP-seq-and-RIP-qPCRPerl script scale_wiggle_rpm.plThis studyhttps://github.com/CalabreseLab/Protocol-for-evaluating-RNA-protein-associations-in-mammalian-cells-with-RIP-seq-and-RIP-qPCRPerl script macs_strand_rand_sam.plThis studyhttps://github.com/CalabreseLab/Protocol-for-evaluating-RNA-protein-associations-in-mammalian-cells-with-RIP-seq-and-RIP-qPCRPython script create_bkg_seq.pyThis studyhttps://github.com/CalabreseLab/Protocol-for-evaluating-RNA-protein-associations-in-mammalian-cells-with-RIP-seq-and-RIP-qPCRCFX ManagerBio-Radv3.1PrismGraphPadv9.5.0OtherForma Series II Water-Jacketed CO2 IncubatorThermo Fisher Scientific3110Vibra-Cell 130-W probe-tip sonicatorSonicsVCX 1302200 TapeStation SystemAgilentG2964AAHigh Sensitivity D1000 ScreenTapeAgilent5067–5584High Sensitivity D1000 ReagentsAgilent5067–5585NextSeq1000 sequencing platformIllumina20038898100-cycle NextSeq 1000/2000 P2 XLEAP-SBS Reagent KitIllumina20100987Hard-Shell 96-Well PCR PlatesBio-RadHSP9601Microseal 'B' PCR Plate Sealing FilmBio-RadMSB1001C1000 Touch Thermal CyclerBio-Rad1851196CFX96 Real-Time SystemBio-Rad1845097RNaseZap RNA Decontamination SolutionThermo Fisher ScientificAM9780

Materials and equipment

0.1% gelatin in 1× PBS

Dissolve 0.5 g glycine in 500 mL 1X PBS and autoclave. Store at 20°C–24°C for up to 12 months.mESC Growth MediumReagentFinal concentrationAmountDMEM (90 mL removed from 500-mL bottle before adding other reagents directly to this bottle)N/A410 mLESC-qualified fetal bovine serum (100%)15%75 mLNon-essential amino acids (10 mM)0.1 mM5 mLPenicillin-streptomycin (10,000 U/mL)100 U/mL5 mLL-glutamine (200 mM)2 mM5 mLβ-mercaptoethanol (100 mM)0.1 mM500 μLLIF-conditioned medium (or recombinant LIF obtained commercially, e.g., Gibco, A35934)1:5001 mLTotalN/A501.5 mLStore at 4°C for up to 3 weeks. Alternatively, mESC Growth Medium can be prepared without LIF and stored at 4°C for up to 2 months. Smaller volumes of medium can be prepared with a 500 dilution of LIF, which should be used within 3 weeks.Quenching MediumReagentFinal concentrationAmountDMEM (60 mL removed from a 500-mL bottle before adding other reagents directly to this bottle)N/A440 mLFetal bovine serum (100%)10%50 mLPenicillin-streptomycin (10,000 U/mL)100 U/mL5 mLL-glutamine (200 mM)2 mM5 mLβ-mercaptoethanol (100 mM)0.1 mM500 μLTotalN/A500.5 mLStore at 4°C for up to 3 months.CRITICAL: To prevent contamination, prepare cell culture buffers inside an approved biological safety cabinet cleaned with 70% ethanol prior to work. Use only RNase-free tubes and pipette tips. Prepare and handle buffers with gloves that have been sterilized with 70% ethanol.0.3% formaldehyde in 1X PBSReagentFinal concentrationAmount1X PBS1X49 mLFormaldehyde (16%, w/v)0.3%1 mLTotalN/A50 mL

Prepare on ice immediately before use. Gently tap the base of a 1-mL 16% formaldehyde ampoule on a lab bench to collect the entire volume in the bottom before breaking off the cap. Use a bulb-fitted glass Pasteur pipette to transfer the entire volume and add to 49 mL ice-cold 1X PBS. If more than 50 mL of 0.3% formaldehyde in 1X PBS will be needed, prepare multiple stocks.

2 M glycine

Dissolve 6.01 g glycine in sterile water to a volume of 40 mL. Store at 4°C for up to 12 months.0.1% Tween 20 in 1X PBSReagentFinal concentrationAmount1X PBS1X25 mLTween 20 (40%, v/v)0.1%62.5 μLTotalN/A25 mLPrepare on ice immediately before use.

0.5% (w/v) bovine serum albumin in 1× PBS

Prepare at 20°C–24°C immediately before use. Dissolve 0.125 g bovine serum albumin in 25 mL 1X PBS. Scale proportionately as needed (0.005 g bovine serum albumin per 1 mL 1X PBS).RIPA BufferReagentFinal concentrationAmountTris-HCl, pH 8.0 (1 M)50 mM2.5 mLTriton X-100 (10%, w/v)1%5 mLSodium deoxycholate (10%, w/v)0.5%2.5 mLSodium dodecylsulfate (10%, w/v)0.1%500 μLEDTA, pH 8.0 (500 mM)5 mM500 μLPotassium chloride (2 M)150 mM3.75 mLRNase-free waterN/A35.25 mLTotalN/A50 mLStore at 4°C for up to 6 months.fRIP BufferReagentFinal concentrationAmountTris-HCl, pH 7.5 (1 M)25 mM1.25 mLEDTA, pH 8.0 (500 mM)5 mM500 μLIGEPAL CA-630 (10%, w/v)0.5%2.5 mLPotassium chloride (2 M)150 mM3.75 mLRNase-free waterN/A42 mLTotalN/A50 mLStore at 4°C for up to 6 months.Pol II ChIP Wash BufferReagentFinal concentrationAmountTris-HCl, pH 7.5 (1 M)50 mM2.5 mLSodium chloride (5 M)140 mM1.4 mLEDTA, pH 8.0 (500 mM)1 mM100 μLEGTA, pH 8.0 (100 mM)1 mM500 μLTriton X-100 (10%, w/v)1%5 mLSodium deoxycholate (10%, w/v)0.1%500 μLSodium dodecylsulfate (10%, w/v)0.1%500 μLRNase-free waterN/A39.5 mLTotalN/A50 mLStore at 4°C for up to 6 months.High-Salt Pol II ChIP Wash BufferReagentFinal concentrationAmountTris-HCl, pH 7.5 (1 M)50 mM2.5 mLSodium chloride (5 M)500 mM5 mLEDTA, pH 8.0 (500 mM)1 mM100 μLEGTA, pH 8.0 (100 mM)1 mM500 μLTriton X-100 (10%, w/v)1%5 mLSodium deoxycholate (10%, w/v)0.1%500 μLSodium dodecylsulfate (10%, w/v)0.1%500 μLRNase-free waterN/A35.9 mLTotalN/A50 mLStore at 4°C for up to 6 months.LiCl Wash BufferReagentFinal concentrationAmountTris-HCl, pH 8.0 (1 M)20 mM1 mLEDTA, pH 8.0 (500 mM)1 mM100 μLLithium chloride (5 M)250 mM2.5 mLIGEPAL CA-630 (10%, w/v)0.5%2.5 mLSodium deoxycholate (10%, w/v)0.5%2.5 mLRNase-free waterN/A41.4 mLTotalN/A50 mLStore at 4°C for up to 6 months.3.5X Reverse Crosslinking BufferReagentFinal concentrationAmountPBS (10X)3X3.33 mLN-lauroylsarcosine (20%, w/v)6%3 mLEDTA, pH 8.0 (500 mM)30 mM600 μLRNase-free waterN/A3.07 mLTotalN/A10 mLStore at 4°C for up to 6 months.Complete RIPA BufferReagentFinal concentrationAmountRIPA BufferN/A560 μLDithiothreitol (1 M)0.5 mM0.56 μLProtease inhibitor cocktail (Sigma P8340)1:1005.6 μLSUPERase-In (Invitrogen AM2696, 20 U/μL)0.1 U/μL2.8 μLTotalN/A568.9 μL

Prepare on ice immediately before use. A volume of 508 μL is needed per RIP lysate. Multiply volumes above by the number of RIP lysates to prepare enough for all lysates (this includes an additional ∼10% to allow for pipetting error).Complete fRIP BufferReagentFinal concentrationAmountfRIP BufferN/A560 μLDithiothreitol (1 M)0.5 mM0.56 μLProtease inhibitor cocktail (Sigma P8340)1:1005.6 μLSUPERase-In (Invitrogen AM2696, 20 U/μL)0.1 U/μL2.8 μLTotalN/A568.9 μL

Prepare on ice immediately before use. A volume of 510 μL is needed per RIP lysate. Multiply volumes above by the number of RIP lysates to prepare enough for all lysates (this includes an additional ∼10% to allow for pipetting error).Complete 1 RIPA:fRIP BufferReagentFinal concentrationAmountRIPA BufferN/A280 μLfRIP BufferN/A280 μLDithiothreitol (1 M)0.5 mM0.56 μLProtease inhibitor cocktail (Sigma P8340)1:1005.6 μLSUPERase-In (Invitrogen AM2696, 20 U/μL)0.1 U/μL2.8 μLTotalN/A568.9 μL

Prepare on ice immediately before use. A volume of 490 μL is needed per RIP sample. (Lysate from one 10-million-cell pellet is sufficient for two 5-million-cell-equivalent RIP samples). Multiply volumes above by the number of RIP samples to prepare enough for all samples (this includes an additional ∼10% to allow for pipetting error).Complete Input-Sample 1X Reverse Crosslinking BufferReagentFinal concentrationAmount3.5X Reverse Crosslinking Buffer1X36.3 μLRNase-free waterN/A45.1 μLDithiothreitol (100 mM)4.3 mM5.5 μLSUPERase-In (Invitrogen AM2696, 20 U/μL)0.17 U/μL1.1 μLProteinase K (Invitrogen 25530049, 20 μg/μL)1.7 μg/μL11 μLTotalN/A99 μL

Prepare on ice immediately before use. A volume of 90 μL is needed per input sample. Multiply volumes above by the number of input samples to prepare enough for all samples (this includes an additional 10% to allow for pipetting error). Final concentrations of components above account for a total volume of 115 μL upon addition to 25-μL input samples.Complete RIP-Sample 1X Reverse Crosslinking BufferReagentFinal concentrationAmount3.5X Reverse Crosslinking Buffer1X36.3 μLRNase-free waterN/A72.6 μLDithiothreitol (100 mM)4.3 mM5.5 μLSUPERase-In (Invitrogen AM2696, 20 U/μL)0.17 U/μL1.1 μLProteinase K (Invitrogen 25530049, 20 μg/μL)1.7 μg/μL11 μLTotalN/A132 μL

Prepare on ice immediately before use. A volume of 115 μL is needed per RIP sample. Multiply volumes above by the number of RIP samples to prepare enough for all samples (this includes an additional 10% to allow for pipetting error).50X Tris-acetic acid-EDTAReagentFinal concentrationAmountTris2 M242 gGlacial acetic acid1 M57.1 mL50 mM EDTA5 mM100 mLNuclease-free waterN/Ato 1 LTotalN/A1 LStore at 20°C–24°C for up to 5 years. From this stock solution, prepare 1X TAE running buffer by bringing 20 mL 50X TAE to 1 L with nuclease-free water.CRITICAL: Prepare buffers in new, certified RNase-free tubes. Prepare buffers at a lab bench that is dust-free and cleaned with 70% ethanol and RNaseZap prior to work. Use only RNase-free tubes and pipette tips. Prepare and handle all buffers with clean gloves. Replace gloves frequently and after touching potential sources of RNase contamination (e.g., skin, door handles, electronic devices).

Step-by-step method details

Crosslinking cells with formaldehyde

Timing: 3 h

This step will generate 10-million-cell pellets of formaldehyde-crosslinked cells that can be stored long-term at −80°C for use in future RIP experiments.CRITICAL: To crosslink multiple dishes in parallel, perform steps in a temporally precise, time-staggered manner to ensure consistent incubation times across samples. Due to the time it takes to move liquids, we recommend processing a maximum of six 10-cm dishes or four 15-cm dishes in parallel. If more dishes than this need to be crosslinked, do so in multiple batches. These steps can be performed on a standard lab bench.CRITICAL: Perform work at a lab bench that is dust-free and cleaned with 70% ethanol and RNaseZap prior to work. Use only RNase-free tubes and pipette tips. Prepare and handle all samples with clean gloves. Replace gloves frequently and after touching potential sources of RNase contamination (e.g., skin, door handles, electronic devices).CRITICAL: Because formaldehyde is a toxic compound, wear proper personal protective equipment, including gloves, laboratory coat, and eye protection. Dispose of formaldehyde-containing solutions in accordance with your institution’s laboratory waste guidelines.1.Observe cells under a microscope and take note of their confluence.Note: A 90% confluent 10-cm dish contains approximately 30 million mESCs, and a 90% confluent 15-cm dish contains approximately 70 million mESCs.Optional: If an area-proportional amount of each cell line was plated to an additional 6-cm dish for counting cells, count these cells before starting the crosslinking and harvesting process. As described in Step 11 of the before you begin section above, dissociate cells from the 6-cm dish(es) with trypsin and count as a proxy for the larger dish(es) of cells that will be crosslinked and harvested. Estimate and record the number of cells in the larger dish by multiplying the total number of cells in the 6-cm dish by the ratio of the amount of cells plated in the larger dish to that in the 6-cm dish.2.Using aspiration, remove growth medium from cells and wash by adding 6 mL or 10 mL ice-cold 1X PBS to a 10-cm or 15-cm dish, respectively.3.Gently tilt the dish and remove wash by aspiration.4.Repeat this 1X PBS wash once more.5.Add 8 mL or 12 mL ice-cold 0.3% formaldehyde in 1X PBS to 10-cm or 15-cm dishes, respectively. Gently tilt dishes to cover all cells with formaldehyde solution. Incubate for 30 min at 4°C.CRITICAL: Handle cells very carefully and keep dishes as level as possible to limit the movement of the 0.3% formaldehyde in 1X PBS, which can cause dissociation of cells from the plate. We recommend incubating the dishes on a level surface in a cold room or refrigerator, whichever is closest to the bench where work is being done.6.Return dishes to the bench, and exactly when 30 min of formaldehyde incubation has elapsed for a dish, add 1 mL or 1.5 mL ice-cold 2 M glycine for a 10-cm or 15-cm dish, respectively. Tilt dishes gently to mix thoroughly. Incubate at 20°C–24°C for 5 min.Note: The purpose of this step is to halt the crosslinking reaction by quenching unreacted formaldehyde with excess glycine.7.After 5 min has elapsed for a dish, remove liquid by aspiration and immediately add 6 mL or 10 mL ice-cold 1X PBS for 10-cm or 15-cm dishes, respectively. After all dishes are in 1X PBS, remove by aspiration.8.Wash dishes again with the same volume of 1X PBS. Keep 1X PBS on cells until harvesting.9.One dish at a time, completely remove 1X PBS by aspiration and harvest cells by scraping.a.Add 2 mL or 4 mL ice-cold 0.1% Tween 20 in 1X PBS to a 10-cm or 15-cm dish, respectively.b.Harvest cells as completely as possible by scraping the entire surface with a cell lifter.c.Transfer the entire volume to an appropriately labeled 15-mL conical tube on ice.d.Add an additional 1 mL or 2 mL 0.1% Tween 20 in 1X PBS to the 10-cm or 15-cm dish, respectively.e.Tilt the dish and use the cell lifter to collect any residual cells into the liquid. Transfer to the same 15-mL conical tube.f.Keep all 15-mL conical tubes of cells on ice until all dishes being processed have been harvested.10.Pellet cells by centrifugation (1200 x g, 5 min, 4°C). Return to ice, doing so carefully to not disturb the pellets.Optional: If a set of marked, known-pellet-size 15-mL conical tubes is available (see Step 12 of the before you begin section above), visually compare these standards to the sizes of the cell pellets currently being processed. Estimate and record the amounts of cells to the nearest 5 million.11.Label 1.7-mL tubes that will receive 10-million-cell aliquots and place these on ice. These can be set up ahead of time during wait steps.12.One 15-mL conical tube of cells at a time, resuspend and prepare 10-million-cell aliquots.a.Very carefully remove supernatant from the cell pellet using a glass Pasteur pipette attached to vacuum aspiration line.b.Using an automatic serological pipettor, gently and thoroughly resuspend each pellet in a volume of ice-cold 1X PBS corresponding to 1 mL per 10 million cells (e.g., 6 mL to a 60-million-cell pellet).c.Transfer 1 mL of evenly suspended cells to appropriate pre-labeled 1.7-mL tubes on ice.13.Pellet cells by centrifugation (1200 x g, 5 min, 4°C).CRITICAL: Visually confirm that the 10-million-cell pellets within and across genotypes or treatment groups are uniform in size.14.Carefully remove supernatants with a P1000 pipette. Keep cell pellets on ice.15.Snap-freeze cell pellets by dipping tubes in a dry ice-methanol bath or liquid nitrogen. Store cell pellets at −80°C.CRITICAL: Take precautions to ensure safety while handling dry ice or liquid nitrogen.**Pause ** Cell pellets can be stored at −80°C for at least 3 years.

RNA immunoprecipitation and purification

Timing: 3–4 days

This step uses bead-bound antibodies to recover sonication-fragmented RNA associated with a protein or proteins of interest. RNA samples obtained from this step can be analyzed by sequencing (RIP-seq) and/or by qPCR (RIP-qPCR).CRITICAL: Perform work at a lab bench that is dust-free and cleaned with 70% ethanol and RNaseZap prior to work. Use only RNase-free tubes and pipette tips. Prepare and handle all samples with clean gloves. Replace gloves frequently and after touching potential sources of RNase contamination (e.g., skin, door handles, electronic devices).16.Prepare antibody-bound beads for RIPs.a.For each immunoprecipitation, transfer 25 μL well-mixed slurry volume of Protein A/G PLUS-Agarose beads (Santa Cruz sc-2003) to a pre-labeled 1.7-mL tube at 20°C–24°C.Note: Prior to removing bead slurry from the stock tube, ensure that the bead suspension is well mixed by inverting the tube ∼10–15 times.b.Wash beads three times with 1 mL 0.5% (w/v) bovine serum albumin in 1X PBS. For each i.Add 1 mL room-temp 0.5% (w/v) bovine serum albumin in 1X PBS to each tube.ii.Invert tubes twice to mix.iii.Pellet beads by centrifugation (2000 x g, 1 min, 20°C–24°C).iv.Carefully remove supernatant with a P1000 pipette.CRITICAL: For all bead-handling steps in this protocol, it is essential to take caution to avoid bead loss, especially after incubating with antibody and lysate. Losing beads will cause a loss in signal and may contribute to experimental variability. We recommend performing wash steps in a well-lit space to enable optimal visualization of beads to limit their loss. It is better to leave a small amount of residual wash on top of the beads (∼10–25 μL) than to risk bead loss by attempting to remove washes completely. Do not mix bead-containing mixtures by pipette as this can also cause beads to be lost to the inside of the tips.c.Add 300 μL room-temp 0.5% (w/v) bovine serum albumin in 1X PBS to each tube.d.Add antibody to each tube. Gently invert to mix.Note: We have found 5 μg of antibody to be generally sufficient when testing new antibodies. For antibodies that perform well in RIP, we have successfully used amounts as low as 1 μg.CRITICAL: As a specificity control, include at least one RIP using a nonspecific IgG antibody at a microgram amount equivalent to the largest amount of antibody in the parallel RIP samples. Alternatively, if using an epitope-recognizing antibody to target epitope-tagged protein(s), prepare control RIPs (e.g., epitope-tagged GFP) with the same epitope-recognizing antibody.e.Incubate overnight at 4°C on an end-over-end rotating platform.**Pause ** The precise timing of this incubation step is not critical. We generally allow 16 to 24 h for this step.17.The next day, prepare cell lysates for RIPs.a.Thaw 10-million-cell pellets of formaldehyde-crosslinked cells on ice.Note: 5 million cells will be used for each RIP, so lysate from one cell pellet can be used for up to two RIPs. Unused lysate can be snap-frozen in a dry ice-methanol bath or liquid nitrogen and stored at −80°C for future RIPs or other experiments. If lysate from the same cell line is needed for more than two RIPs, thaw additional pellets and pool lysates prior to adding them to antibody-bound beads.b.To each pellet, add 508 μL Complete RIPA Buffer and resuspend cells with 10 gentle up-down strokes of a P1000 pipette.c.Keeping cells on ice, sonicate each cell suspension with a Sonics Vibra-Cell 130-watt VCX 130 probe-tip sonicator set to 30% amplitude output.i.Before starting, use Kimwipes to clean the probe tip with RNaseZap and then 70% ethanol. Clean the probe tip with 70% ethanol between samples.ii.Lower the probe tip into the center of the cell suspension such that the probe tip is just above the bottom of the tube and not touching the walls of the 1.7-mL tube.iii.Sonicate twice for 30 s with a 1-min pause in between.d.Pellet cell debris by centrifugation (15,000 x g, 15 min, 4°C).i.Meanwhile, label 1.7-mL tubes that will receive lysates and input samples and place these on ice. If lysates will be pooled from multiple 10-million-cell pellets, instead label larger tube(s) and place on ice.ii.Meanwhile, start bead-washing steps if time allows (Step 19 below). Lysates can incubate on ice until bead-washing steps are complete.e.Transfer 510 μL supernatant lysates to pre-chilled tubes on ice immediately after centrifugation is complete.f.Add 510 μL Complete fRIP Buffer to each lysate and mix thoroughly (∼15 up-down strokes) with a P1000 pipette.18.Transfer a 25-μL portion of each 1020-μL lysate mixture to pre-chilled tubes on ice to be processed the next day as 5% input samples. Store these at −20°C or −80°C.19.Wash antibody-bound beads from the previous day.a.Pellet beads by centrifugation (2000 x g, 1 min, 4°C).b.Place tubes on ice and carefully remove supernatant with a P1000 pipette.Note: After the first centrifugation step to remove 0.5% (w/v) bovine serum albumin in 1X PBS, some beads may cling to the sides of the tube above the surface of the liquid. Take care to avoid touching these beads with the pipette tip to prevent their loss. Beads should not cling to the sides of the tube in subsequent wash steps.c.Add 1 mL ice-cold fRIP Buffer and mix by gently inverting three times.d.Pellet beads by centrifugation (2000 x g, 1 min, 4°C).e.Place tubes on ice and carefully remove supernatant with a P1000 pipette.f.Repeat Steps 19c-e once more for a total of two washes with fRIP Buffer.20.Add 490 μL lysate mixtures to the appropriate tube of washed, antibody-bound beads.21.Add 490 μL Complete 1 RIPA:fRIP Buffer to each tube to bring to 980 μL. Gently invert to mix.22.Incubate overnight at 4°C on an end-over-end rotating platform.**Pause ** The precise timing of this incubation step is not critical. We generally allow 16 to 24 h for this step.23.The next day, pre-chill two appropriately labeled 1.7-mL tubes for each RIP at 4°C (for changing tubes during the washes).24.Pre-warm heat blocks to 42°C, 55°C, and 65°C.25.Wash beads once with 1 mL ice-cold fRIP Buffer.a.Pellet beads by centrifugation (2000 x g, 1 min, 4°C).b.Place tubes on ice and carefully remove supernatant with a P1000 pipette.c.Add 1 mL ice-cold fRIP Buffer and mix by gently inverting three times.d.Pellet beads by centrifugation (2000 x g, 1 min, 4°C).e.Return tubes on ice and carefully remove supernatant with a P1000 pipette.26.Carefully transfer beads to new tubes with 1 mL ice-cold Pol II ChIP Buffer, using a “two-step” transfer technique.a.Place new, pre-chilled tubes on ice.b.One RIP sample at a time, temporarily open the lid of the source tube containing the beads and the lid of the new tube that will receive the beads.c.Fill a P1000 pipette tip with 1000 μL ice-cold Pol II ChIP Buffer.d.Place the pipette tip in the source tube directly above the beads. Carefully depress the P1000 pipette plunger part-way to eject approximately 100 μL of the Pol II ChIP Buffer onto the beads.e.Move the pipette tip to the bottom of the tube, then slowly release the plunger to fill the pipette tip back to 1000 μL, now containing some of the beads at the bottom of the tip. Avoid introducing bubbles into the pipette tip, as this can disperse the beads throughout it.f.Move the pipette to the new tube and carefully depress the plunger to eject approximately 800 μL into the new tube.g.Return the pipette to the source tube and slowly depress the plunger to eject the remaining 200 μL onto the remaining beads.h.Slowly release the plunger to collect the entire volume into the pipette tip, stopping once all is collected to limit introducing bubbles.i.Move the pipette back to the new tube and depress the plunger to eject the remaining volume into the new tube.CRITICAL: Visually inspect the pipette tip and source tube to ensure that no beads were left behind. If any beads were left behind in the source tube, collect them with the pipette tip in residual wash buffer. Alternatively, centrifuge the new tube (2000 x g, 1 min, 4°C) and use ∼20 μL of the supernatant to collect residual beads and transfer to the new tube.Note: Once comfortable with this technique, consider modifying it to have three transfer-ejection steps, which enables an additional opportunity to collect beads with bead-free wash buffer, thus enhancing the ability to collect every bead.Note: In the development of this protocol, we added tube-changing steps as a strategy to reduce carry-over of background signal that could potentially result from lysate-derived unbound protein-RNA complexes adhering to the plastic or being insufficiently flushed from between tube openings and their caps during wash steps. That said, we have not empirically tested whether this protocol outperforms one in which tube-changing steps are omitted.27.Wash beads.a.Incubate tubes end-over-end at 4°C for 5 min.b.Pellet beads by centrifugation (2000 x g, 1 min, 4°C).c.Return tubes to ice and carefully remove supernatant with a P1000 pipette.28.Wash beads as done in Step 27 twice more with 1 mL ice-cold Pol II ChIP Buffer, for a total of three washes.29.Wash beads as done in Step 27 once with 1 mL ice-cold High-Salt Pol II ChIP Buffer.30.As done in Step 26, carefully transfer beads to a new tube with 1 mL ice-cold LiCl Wash Buffer.31.Wash beads as done in Step 27.a.Meanwhile, place 5% input samples from the day before on ice to thaw.32.Carefully remove supernatant from beads, first with a P1000 pipette and then more precisely with a P20 pipette.a.Ensure no loss of beads while leaving behind a supernatant volume of ∼10 μL.33.To all 5% input samples, add 90 μL ice-cold Complete Input-Sample 1X Reverse Crosslinking Buffer and mix thoroughly with a P200 pipette.Note: 1X Reverse Crosslinking Buffer is slightly viscous. When ejecting from the pipette tip, depress the pipette plunger slowly to ensure the entire volume is ejected. Any loss of sample may impact the accuracy of RIP quantification.34.To all RIP samples, eject 115 μL ice-cold Complete Bead-Sample 1X Reverse Crosslinking Buffer directly onto the beads with enough force to resuspend them. Do not pipette the beads into the tip.35.Transfer all 5% input and RIP sample tubes to a heat block pre-set to 42°C. Incubate for 1 h.a.Every ∼30 min during this and the subsequent heat steps, gently resuspend the beads in each RIP sample tube by filling a P200 pipette tip with ∼90 μL of the supernatant and ejecting back down to resuspend them. Ensure complete ejection of sample out of the tip to prevent sample loss.36.Transfer tubes to a heat block pre-set to 55°C. Incubate for 1 h.37.Transfer tubes to a heat block pre-set to 65°C. Incubate for 30 min.38.Remove tubes from the heat block and add 1 mL ice-cold TRIzol Reagent to each sample. Mix thoroughly by inverting several times.**Pause ** TRIzol-containing samples can be immediately processed further for RNA purification or stored at −80°C for up to 3 months. If samples are stored at −80°C, place them at 20°C–24°C until completely thawed before proceeding.39.Add 200 μL chloroform to each TRIzol-containing sample. Vortex for 3–5 s to mix thoroughly.a.Alternatively, if many samples are being processed, firmly hold an empty tube rack on top of the rack holding the tubes to make a “sandwich,” and vigorously shake the sandwich up and down for 15 s.40.Centrifuge tubes (16,100 x g, 15 min, 4°C) and place on ice carefully to not disturb the separated phases.a.Meanwhile, label one new 1.7-mL tube for each sample. Keep at 20°C–24°C.41.From each phase-separated sample, transfer 580 μL upper aqueous phase to a new tube.42.Add 580 μL 100% ethanol to each sample, invert, and vortex for 3–5 s to mix. Pulse down in a microcentrifuge to collect volumes at the bottom of the tubes.43.Purify RNA using the Zymo RNA Clean & Concentrator-5 kit.a.Apply 600 μL of the sample mixtures to appropriately labeled Zymo-Spin IC columns and centrifuge (16,100 x g, 30 s, 20°C–24°C). Remove flowthrough from the collection tubes.b.Apply the remaining sample mixtures to the columns and centrifuge (16,100 x g, 30 s, 20°C–24°C). Remove flowthrough from the collection tubes.c.Add 400 μL RNA Wash Buffer to the columns and centrifuge (16,100 x g, 30 s, 20°C–24°C). Remove flowthrough from the collection tubes.i.Meanwhile, prepare DNA Digestion Master Mix at 20°C–24°C, containing, for each sample, 5 μL 1 U/μL DNase I and 35 μL DNA Digestion Buffer.d.Apply 40 μL DNA Digestion Master Mix to each column and incubate at 20°C–24°C for 20 min.e.Add 400 μL RNA Prep Buffer to the columns and centrifuge (16,100 x g, 30 s, 20°C–24°C). Remove flowthrough from the collection tubes.f.Add 700 μL RNA Wash Buffer to the columns and centrifuge (16,100 x g, 30 s, 20°C–24°C). Remove flowthrough from the collection tubes.g.Add 400 μL RNA Wash Buffer to the columns and centrifuge (16,100 x g, 30 s, 20°C–24°C). Remove flowthrough from the collection tubes.h.Centrifuge (16,100 x g, 2 min, 20°C–24°C) to remove residual wash buffer from the columns.i.Meanwhile, label one new 1.7-mL tube for each sample. Keep at 20°C–24°C.i.Transfer columns to new 1.7-mL tubes and add 15 μL RNase-free water to each. Incubate at 20°C–24°C for 5 min.j.Centrifuge (16,100 x g, 1 min, 20°C–24°C) to elute RNA.k.Immediately place eluted RNA samples on ice.**Pause ** RNA samples can be stored at −80°C indefinitely (>5 years). RNA samples can later be thawed on ice or briefly at 20°C–24°C before placing on ice.

Preparation of RIP-seq libraries and sequencing

Timing: 2–4 days

This step generates RIP-seq libraries that are sequenced using an Illumina short-read sequencer to generate FASTQ sequence files for downstream analysis.Note: Sequencing RIP samples is optional if measurement by RIP-qPCR is the intended endpoint of the experiment. Each ∼14-μL eluted RIP RNA sample has sufficient volume to be analyzed by both RIP-seq and RIP-qPCR.Note: We recommend using the KAPA RNA HyperPrep Kit with RiboErase (HMR) with KAPA Unique Dual-Indexed (UDI) Adapters for RIP-seq library construction. Other RNA-seq-based library preparation kits or protocols may be considered but are untested in our hands with this protocol.CRITICAL: Perform work at a lab bench that is dust-free and cleaned with 70% ethanol and RNaseZap prior to work. Use only RNase-free tubes and pipette tips. Prepare and handle all samples with clean gloves. Replace gloves frequently and after touching potential sources of RNase contamination (e.g., skin, door handles, electronic devices).44.Refer to Steps 1–10 of Chapter 3 of the KAPA RNA HyperPrep Kit with RiboErase (HMR) manufacturer’s protocol, with modifications listed below, for details of library construction through adapter https://elabdoc-prod.roche.com/eLD/api/downloads/9083e64d-6578-f011-3091-005056a772fd?countryIsoCode=XG.a.At Step 1.3, the 10-μL “total RNA in water” sample should be prepared by combining 9 μL column-purified RIP RNA with 1 μL of a 250 dilution of ERCC RNA spike-in controls (Thermo Fisher 4456740).Note: The addition of a 250 dilution of ERCC RNA spike-in controls is optional. Inclusion of spike-in controls enables quantitative comparisons between samples of interest once RIP-seq data have been obtained. For example, when RIP-seq is performed with the same antibody across multiple conditions, inclusion of spike-in controls would enable a quantitative comparison of signal, for example, between wild-type and knockout cells. Likewise, inclusion of spike-in controls would enable a quantitative estimate of how much RIP-seq signal was obtained over a given region using an antibody of interest versus an IgG control. We have used the upper quartile and median of ratios normalization methods to scale RIP-seq RPM values relative to ERCC spike-in controls across a panel of samples. However, in practice, we often do not use spike-in normalization because our RIP-seq experiments often involve generating signal from different commercially available antibodies, each of which has varying and uncontrollable levels of specificity, selectivity, and purity. In our experience, even different lots of the same antibody can have significantly different IP efficiencies. Thus, the absolute signal obtained after RIP-seq with different antibodies can vary dramatically and in ways that are not biologically meaningful. We typically include the spike-in addition at this step, even if spike-ins are not used for normalizations, because they do provide a form of quality control for library preparation and subsequent PCR amplification, and their inclusion does not affect any other downstream analyses.b.At Steps 2.11–2.12, 4.11–4.12, 9a.11-9a.12, and 10.11–10.12, pulse-spin tubes after removal of final 80% ethanol wash, return to magnetic separator, and remove residual wash from the bottom of the tubes with a pipette. Let air-dry for 1 min instead of the 3–5 min listed.c.At Step 5.6, fragment the RNA by incubating at 85°C for 5 min.d.At Step 8.2, use diluted adapter stocks of 1.5 μM.e.At Step 10.14, resuspend beads in 52 μL 10 mM Tris-HCl (pH 8.0–8.5) instead of 22 μL. At Step 10.17, transfer 50 μL of the supernatant to a new tube.Note: Increasing the volume of this final elution enables estimation of the number of needed amplification cycles by qPCR and provides additional pre-amplification library in case the first attempt at library amplification uses too many cycles and needs to be repeated with fewer cycles.Pause ** Pre-amplification libraries can be stored at −20°C for up to 1 month, if not longer. These can later be thawed at 20°C–24°C.45.Use qPCR to estimate the number of PCR cycles needed to sufficiently amplify each library while preventing overamplification.a.Following technical details from the reverse transcription and qPCR of input and RIP samples (RIP-qPCR) section below, set up 10-μL qPCR reactions in duplicate for each pre-amplification library, with each containing the ReagentVolume per reaction10 μM JT3560.5 μL10 μM JT3570.5 μLUltrapure nuclease-free water3 μL2x iTaq Universal SYBR Green Supermix5 μLPre-amplification library1 μLNote:** If available, include parallel reactions with one or more pre-amplification libraries that have been previously amplified and sequenced. Knowledge of the number of PCR cycles that were previously used to amplify these libraries can provide a positive control for the number of cycles that should be used for the present libraries. We also recommend including no-template negative control reactions with ultrapure nuclease-free water in place of pre-amplification library.b.Run reactions on a qPCR machine (e.g., Bio-Rad C1000 Touch Thermal Cycler equipped with a CFX96 Real-Time System) with the following StepsTemperatureTimeCyclesInitial denaturing95°C10 min1Denaturing95°C10 s40 cyclesAnnealing60°C30 sElongation72°C30 sPlate readN/AN/Ac.Load the output plate file in Bio-Rad CFX Manager or other appropriate software.d.For each pre-amplification library, calculate the average of the threshold-determined (default setting) Cq values from the technical duplicate reactions.e.Add 2 to these average Cq values and round up to determine the number of PCR cycles to amplify each library.Note: The number of PCR cycles for library amplification is variable and is highly dependent on each antibody’s RIP efficiency and target protein abundance. In our experience, we have used as few as 13 cycles for highly efficient antibodies (e.g., Santa Cruz sc-33652 anti-SRSF1) and as many as 21 cycles for lowly efficient antibodies (e.g., Invitrogen 02-6102 nonspecific IgG). For low RIP-seq library yield, see troubleshootingproblem 3.46.Refer to Steps 1-3a of Chapter 4 of the KAPA RNA HyperPrep Kit with RiboErase (HMR) manufacturer’s protocol, with modifications listed below, for details of library https://elabdoc-prod.roche.com/eLD/api/downloads/9083e64d-6578-f011-3091-005056a772fd?countryIsoCode=XG.a.At Step 2.1, use the number of PCR amplification cycles determined by qPCR for each library.b.At Step 3.11–3.12, pulse-spin tubes after removal of final 80% ethanol wash, return to magnetic separator, and remove residual wash from the bottom of the tubes with a pipette. Let air-dry for 1 min instead of the 3–5 min listed.47.Measure dsDNA concentration of each amplified library using a Qubit fluorometer and Qubit dsDNA Broad-Range Assay Kit, following the manufacturer’s https://tools.thermofisher.com/content/sfs/manuals/Qubit_dsDNA_BR_Assay_UG.pdf.a.At Steps 1.6 and 1.7, add 2 μL amplified library to 198 μL Qubit working solution for a 100 dilution.Note: Ideally, amplified libraries should have a dsDNA concentration of 5–60 ng/μL. Concentrations lower than 5 ng/μL may be feasible as long as enough amplified library is available when pooling libraries for sequencing. Concentrations greater than 60 ng/μL may be indicative of library overamplification, and repeating Step 46 with fewer PCR amplification cycles is recommended.48.Estimate, to the nearest 50 bp, the average length of DNA molecules comprising each library by running 50–300 ng (5–10 μL) on a 2% (w/v) agarose-TAE-EtBr gel alongside a dsDNA ladder.Note: DNA should appear as a smear with an average size of ∼300–350 bp, as shown in Figure 1.

Optional: We have alternatively used an Agilent TapeStation instead of agarose gel electrophoresis to measure library sizes. Estimates of library sizes by agarose gels are less precise than by TapeStation but are cheaper and sufficient for this purpose. 49.Pool libraries for sequencing such that the total DNA concentration of the pooled sample is 15 nM.Note: We recommend using Microsoft Excel or a similar tool to perform these calculations for all libraries that will be pooled.a.To convert concentrations in ng/μL to nM, assume that each base pair, on average, is 660 g/mol. Use the following i.[DNA] in nM = ([DNA] in ng/μL) ∗ 10^6^/(660 ∗ average length in bp).Note: We routinely achieve ∼550 million reads on a single run of an Illumina NextSeq 1000 sequencer. We recommend a sequencing depth of at least 20 million reads per sample, which equates to pooling a maximum of 27 libraries. Note that other types of libraries (e.g., RNA-seq, ChIP-seq) from other experiments can be pooled with RIP-seq libraries, assuming adapter indexes are all compatible.b.Calculate the desired final concentration of each library in the pooled sample by dividing 15 nM by the number of libraries.Note: This calculation is for the equimolar contribution of each library. Adjust if different read depths are desired among the pooled libraries.c.Calculate the volume of each library to add using the following i.Volume library to add = (Pooled sample volume) ∗ (Library final concentration)/(Current library concentration).Note: We recommend a pooled sample volume of 200 μL to enable accurate pipetting of all libraries that are within the desired concentration range of 5–60 ng/μL. Adjust this volume if needed and/or make dilutions of highly concentrated libraries before pooling to ensure that all volumes can be pipetted accurately with the same pipette (e.g., using a P10 pipette for all samples, adding between 1 and 10 μL of each library).d.Using the volumes calculated in Step 49c, carefully combine each library in a new 1.7-mL tube, using 10 mM Tris-HCl pH 8.0–8.5 (library elution buffer) to bring to the final desired volume.Note: To increase pipetting accuracy, we recommend adding 10 mM Tris-HCl pH 8.0–8.5 to the tube first. When adding each library, place the end of the tip just below the surface of the liquid, eject the library from the tip, and then pipette up and down 2–3 times to rinse any potential residual library volume from the inside of the tip.Note: Steps 50–52 below are routinely performed by our sequencing facility. Defer to your sequencing facility for final sample preparation and sequencing procedures.50.Using RSB with Tween 20 (Illumina), prepare a 2-nM dilution of the pooled sample.51.Prepare a final sample to load onto the sequencer, containing 650 pM pooled sample and 3–4% PhiX DNA spike-in (by volume) as a positive control for sequencing quality.a.For more details, refer to the manufacturer’s https://support-docs.illumina.com/IN/NextSeq1K2K_DnD/Content/NextSeq1K2K/DnD-NS1K2K.htm?protocol=onboard.52.Sequence the pooled sample with an Illumina NextSeq 1000, following manufacturer’s https://support-docs.illumina.com/IN/NextSeq10002000/Content/IN/FrontPages/NextSeq10002000.htm.a.We routinely use the 100-cycle NextSeq 1000/2000 P2 XLEAP-SBS Reagent Kit (Illumina 20100987) for sequencing. Figure 1Estimation of RIP-seq library size by agarose gel electrophoresisRIP-seq libraries should appear as a characteristic smear with an average size of 300–350 bp. Related to Step 48.

RIP-seq data alignment and visualization

Timing: 1–2 days

This step aligns RIP-seq data to a reference genome and generates wiggle files that can be visualized in a genome browser to view read counts at genomic positions for each RIP sample.53.Using UNIX commands, obtain a FASTQ file from each sample that was sequenced and copy these into a directory fastq within your working directory. Unzip these files if they are cd /your/working/directorymkdir fastqcp /source/directory/∗.fasta.gz fastqgunzip fastq/∗54.Use STAR^16^ to map FASTQ files to the desired reference genome (e.g., mouse mm10/vM25).a.Generate a STAR genome index.i.Download a gene annotation GTF file (e.g., gencode.vM25.basic.annotation.gtf.gz from https://www.gencodegenes.org/mouse/release_M25.html) and place this in your working directory. Unzip with gunzip.ii.Download a corresponding whole-genome FASTA file (e.g., GRCm38.p6.genome.fa.gz from https://www.gencodegenes.org/mouse/release_M25.html) and place this in your working directory. Unzip with gunzip.iii.Run the following example mkdir STARv2.7.11b_genome_index_mm10module load star/2.7.11bSTAR --runThreadN 12 --runMode genomeGenerate --genomeDir STARv2.7.11b_genome_index_mm10 --genomeFastaFiles GRCm38.p6.genome.fa --sjdbGTFfile gencode.vM25.basic.annotation.gtfb.Align FASTQ files to the reference genome.mkdir STAR_outputSTAR --genomeDir STARv2.7.11b_genome_index_mm10 --runThreadN 12 --readFilesIn fastq/${sample_name}.fastq --outFilterMultimapNmax 1 --outFileNamePrefix STAR_output /${sample_name} --outSAMtype SAM --outReadsUnmapped FastxNote: The option --outFilterMultimapNmax 1 is used to only consider uniquely mapping reads and thus omits reads mapping to nonunique features such as highly repetitive regions. This option can be omitted or adjusted depending on the need.Note: STAR can be run to align individual FASTQ files if individual wiggle files are desired for each RIP replicate. Alternatively, FASTQ files from multiple replicates can be aligned together for generating a single multi-replicate summary wiggle file. To do this, use option --readFilesIn .55.Parse the output SAM files by strand to make “stranded” SAM mkdir stranded_sammodule load samtools/1.21samtools view -f 0x10 STAR_output/${sample_name}Aligned.out.sam > stranded_sam/${sample_name}.sam_possamtools view -F 0x10 STAR_output/${sample_name}Aligned.out.sam > stranded_sam/${sample_name}.sam_negNote: The option -f specifies retention of reads of a particular attribute, denoted by an integer that follows. The hexadecimal integer 0x10 specifies reads that align to the negative strand of the genome. As an alternative, the decimal integer 16 can be used. Perhaps counterintuitively, the option -f 0x10 (“retain negative-stranded reads”) is used here to identify positive-stranded alignments. This is because the sequencing libraries are prepared with a reverse-stranded kit, which selectively amplifies first-strand, reverse-complemented cDNA sequences due to incorporation of dUTP during second-strand cDNA synthesis. Therefore, reads whose reverse complements align to the positive genomic strand are considered positive-stranded reads. The option -F specifies exclusion of reads and is the opposite of -f, so -F 0x10 specifies negative-stranded alignments. Most stranded RNA-seq library preparation kits are reverse-stranded, but if a forward-stranded kit is used instead, the -f 0x10 and -F 0x10 options here should be switched.56.Generate wiggle files for each stranded SAM file using the custom Perl script bigsam_to_wig_mm10_wcigar2.pl (see key resources table).Note: Here, read depth is summarized in 50-nt bins, which we have found to be a reasonable tradeoff between data resolution and file size. Note that, depending on computing resources available, this Perl script may take a day or longer to run.mkdir wiggles# Make wiggle file for negative strand. Specify desired display color (e.g., red), paired-end data flag (n/y), and log10 normalization flag (n/y)perl bigsam_to_wig_mm10_wcigar2.pl stranded_sam/${sample_name}.sam_neg wiggles/${sample_name}_neg red n n 50gzip {sample_name}_neg.wig# Make wiggle file for positive strand. Specify desired display color (e.g., blue), paired-end data flag (n/y), and log10 normalization flag (n/y)perl bigsam_to_wig_mm10_wcigar2.pl stranded_sam/{sample_name}.sam_pos wiggles/${sample_name}_pos blue n n 50gzip${sample_name}_pos.wigOptional: The step above generates wiggle files in which the y-axis reports the number of reads aligned per 50-nt bin. To scale this signal by the millions of reads sequenced or uniquely aligned (RPM normalization), run the Perl script scale_wiggle_rpm.pl (see key resources table) as follows. Total sequenced or uniquely aligned reads can be obtained from the STAR alignment log.out files.perl scale_wiggle_rpm.pl wiggles/${sample_name}_neg.wig <total-reads> wiggles/${sample_name}_neg_RPM.wiggzip wiggles/${sample_name}_neg_RPM.wigperl scale_wiggle_rpm.pl wiggles/${sample_name}_pos.wig wiggles/${sample_name}_pos_RPM.wiggzip wiggles/${sample_name}_pos_RPM.wigOptional: As an alternate approach to generating wiggle files, see code and instructions at https://github.com/CalabreseLab/seekr2.0_update_manuscript.57.Visualize wiggle tracks.a.Download zipped wiggle files to your local computer.b.Upload zipped wiggle files to the UCSC Genome Browser (which accepts zipped files) in the appropriate genome (e.g., GRCm38/mm10) using the toolbar option My Data > Custom Tracks, then click “Add custom tracks.”c.Upload a wiggle file (one at a time) with “Browse…”, click “Submit,” and then view tracks in the genome browser by clicking “Return to current position.”d.Rescale and reconfigure track displays as desired by right-clicking on each track.

RIP-seq peak calling

Timing: 1 day

The following steps locate peaks of RIP-seq data in the genome, which we typically define as genomic regions that have reads per million total reads (RPM)-normalized RIP-seq signal values of greater than twice that of IgG or another non-specific control (such as RIP-seq performed in knockout cells or RIP-seq performed with epitope-tagged GFP) in at least two biological replicates.Note: The code below is designed to call peaks from a single experimental condition and should be run within a directory that only contains FASTQ files from that single condition. The results of output files and processing steps that involve pooled replicates from the single experimental condition are named in the format of ‘yourRIPname_’. The results of output files from individual replicates within the single experimental condition are named in the format of ‘yourRIPname_rip{j}’, where {j} is the replicate number.58.Into a single directory with a unique name, copy or move over sequence files (.fastq) for all replicates from a single experimental condition, referred to below as ‘yourRIPname’.Note: This directory should only contain FASTQ files for the RIP of interest, not all FASTQ files for all RIPs or experimental conditions in a multi-RIP study.Note: We recommend naming the.fastq file in a way that contains the corresponding date or notebook reference (“B12” in the example below), as well as the name of the cell line used and the protein target of the RIP (“TSCs” and “hnrnpk” in the example below):TSC_B12_hnrnpk_rip_S3_R1_001.fastq59.Separately, in another directory with a unique name, copy or move over sequence files (.fastq) for all replicates of IgG or other negative control, referred to below as ‘igg’.60.Align all replicates of IgG control RIP to the reference genome, merging all replicates into one output SAM file.Note: Run this step in the directory that contains only the IgG RIP FASTQ files. Record this path as ‘/your/path/to/igg_data_folder’ to be used later.module load star/2.7.11bmodule load samtools/1.21# Use UNIX to obtain a comma separated list of the names of all IgG fastq files, "rip_filenames", with all white spaces and newline characters removedrip_filenames=(ls -m ∗.fastq | sed -r 's/∖s+//g' | tr -d '∖n')# Use STAR to align to the reference genome all IgG replicates in the rip_filenames listSTAR --genomeDir /your/STARv2.7.11b_genome_index_folder --runThreadN 12 --outFilterMultimapNmax 1 --outFileNamePrefix igg_ --readFilesIn rip_filenames# Split the filtered SAM file by strandsamtools view -h -F 0x10 igg_Aligned.out.sam > igg_Aligned_neg.out.samsamtools view -h -f 0x10 igg_Aligned.out.sam > igg_Aligned_pos.out.samNote: See Step 55 for an explanation of these SAMtools flags.61.Align RIP-of-interest data to the reference genome.Note: Run this step in a directory that contains only RIP-of-interest FASTQ files. There are two rounds of separate alignments below, one round in which all replicates are merged prior to aligning to the reference, and a second round in which each individual replicate is aligned separately. The results from both alignment steps will be used to call peaks on the RIP data.# Use UNIX to obtain a comma separated list of the names of all RIP FASTQ files, "rip_filenames", with all white spaces and newline characters removedrip_filenames=(ls -m ∗.fastq | sed -r 's/∖s+//g' | tr -d '∖n')# Use STAR to perform a single alignment of all RIP replicates merged together, and assign the prefix of the output file as "yourRIPname_"STAR --genomeDir /your/STARv2.7.11b_genome_index_folder --runThreadN 12 --outFilterMultimapNmax 1 --outFileNamePrefix yourRIPname_ --readFilesIn rip_filenames# The next series of commands will perform STAR alignment on each individual replicate of RIP data in the directory# Count the total number of replicates (assuming each FASTQ file represents one separate biological replicate), and save that number as "rip_file_num"rip_file_num=$(ls ∗.fastq | wc -l)if [$rip_file_num -gt 1 ]# Only perform these subsequent steps if the replicate number is greater than 1, and not when there is only 1 replicatethen # Convert rip_filenames into an array for the loop in the subsequent step IFS=',' read -r -a rip_arr <<< "rip_filenames" # Loop on indices of array for i in {!rip_arr[@]}; do  # For each file name  rip_name="{rip_arr[i]}"  # Save the FASTQ name and its corresponding replicate number  echo The rip${i} file is$rip_name. >> yourRIPname_rip_info.txt  # Index+1 so that it's 1-based for the file name  j=$(($i + 1))  # Align each RIP file to the reference genome, and store each alignment using a filename structure “yourRIPname_rip${j}_”  STAR --genomeDir /your/STARv2.7.11b_genome_index_folder --runThreadN 12 --outFilterMultimapNmax 1 --outFileNamePrefix yourRIPname_rip${j} --readFilesIn rip_name  donefi62.Run the MACS peak-calling algorithm on the positive- and negative-stranded STAR alignments that were generated using all replicates pooled together (e.g., “yourRIPname_” files).***Note:*** The list of coordinates returned by MACS is the preliminary set of RIP peaks, but not the final one. For the Perl script macs_strand_rand_sam.pl, see the key resources table.# Split the filtered SAM by strandsamtools view -h -F 0x10 yourRIPname_Aligned.out.sam > yourRIPname_Aligned_neg.out.samsamtools view -h -f 0x10 yourRIPname_Aligned.out.sam > yourRIPname_Aligned_pos.out.sam# Load randomization Perl script macs_strand_rand_sam.plcp /path/to/macs_strand_rand_sam.pl .# Randomize strand value (+ or -) for each aligned read within each stranded file. This is performed because MACS is designed to read non-strand-specific ChIP-seq data and uses the average distance between positive- and negative-stranded alignments to center its peak calls.perl macs_strand_rand_sam.pl yourRIPname_Aligned_pos.out.sam yourRIPname_Aligned_pos.out.rand.samperl macs_strand_rand_sam.pl yourRIPname_Aligned_neg.out.sam yourRIPname_Aligned_neg.out.rand.sam# Convert SAM to BAM to avoid errors in MACS peak callingsamtools view -S -b yourRIPname_Aligned_pos.out.rand.sam > yourRIPname_Aligned_pos.out.rand.bamsamtools view -S -b yourRIPname_Aligned_neg.out.rand.sam > yourRIPname_Aligned_neg.out.rand.bam# Use MACS to identify potential peak regions from the BAM filesmodule load macs/2.2.7.1macs2 callpeak -t yourRIPname_Aligned_pos.out.rand.bam --keep-dup all --broad --broad-cutoff 0.3 --max-gap 76 --outdir yourRIPname_pos_peaks -n yourRIPname_posmacs2 callpeak -t yourRIPname_Aligned_neg.out.rand.bam --keep-dup all --broad --broad-cutoff 0.3 --max-gap 76 --outdir yourRIPname_neg_peaks -n yourRIPname_neg***Note:*** See **Step 55** for an explanation of these SAMtools flags.63.Use featureCounts to count the number of reads under each potential peak region called by MACS:module load subread/2.0.6# Use the output .broadPeak from MACS peak calling (Step 62)# Convert the broadPeak file to a SAF file in preparation for input into featureCountscat yourRIPname_neg_peaks.broadPeak | awk -v OFS="∖t" '{print 1,$2,$3,"-"}' > yourRIPname_saf.txtcat yourRIPname_pos_peaks.broadPeak | awk -v OFS="∖t" '{print $1,$2,3,"+"}' >> yourRIPname_saf.txt# Add column names to the first line of the file and save it with the suffix .safcat yourRIPname_saf.txt | awk -v OFS="∖t" '{print NR,0}' | sed '1 i∖GeneID∖tChr∖tStart∖tEnd∖tStrand' > yourRIPname_peaks.saf# Count reads under each potential MACS peak using the total merged replicate filefeatureCounts -s 2 -F SAF -a yourRIPname_peaks.saf -o yourRIPname_fc yourRIPname_Aligned.out.sam# Count reads under each potential MACS peak using the merged IgG control file# Using the path in Step /your/path/to/igg_data_folderfeatureCounts -s 2 -F SAF -a yourRIPname_peaks.saf -o yourRIPname_igg_fc /your/path/to/igg_data_folder/igg_Aligned.out.sam# Extract the read counts under each potential MACS peak from the IgG file. These counts will be appended to yourRIPname_fc later, along with other informationcut -f7 yourRIPname_igg_fc > yourRIPname_igg_counts.txtif [ rip_file_num -gt 1 ]# Proceed only if there is more than 1 RIP replicatethen # Loop on indices of array rip_arr for i in {!rip_arr[@]}; do  # Index+1 so that it's 1-based for the file name  j=$(($i + 1))  # Count reads under each MACS peak for individual replicates  featureCounts -s 2 -F SAF -a yourRIPname_peaks.saf -o yourRIPname_rip${j}_fc yourRIPname_rip${j}_Aligned.out.sam  # Extract the read counts under each potential MACS peak  # These counts will be appended to yourRIPname_fc later  cut -f7 yourRIPname_rip${j}_fc > yourRIPname_rip${j}_counts.txt  # Prepare for concatenation of count files  # Build a string called "rip_count_filename" that contains all of the replicates _counts.txt files, separated by spaces  if [ j == 1 ]  then  # Initialize the string properly for the first replicate  rip_count_filename="yourRIPname_rip1_counts.txt"  else  # Append the rest of the replicate file names  rip_count_filename="{rip_count_filename} yourRIPname_rip{j}_counts.txt"   fi done # Make summary file by concatenating in the following counts from all RIP replicates merged together, counts from the merged IgG control, counts from the individual RIP replicates paste yourRIPname_fc yourRIPname_igg_counts.txt rip_count_filename | sed 1d > yourRIPname_fc.txtelse  # For situations in which there is only 1 replicate  # Make a summary file by counts from the single RIP replicate, counts from the merged IgG control  paste yourRIPname_fc yourRIPname_igg_counts.txt | sed 1d > yourRIPname_fc.txtfi64.Convert read counts into RPM (reads per million total reads).Note: We most commonly use total read counts as our denominator in RIP-seq RPM calculations because different RNA-binding proteins have varying tendencies to associate with non-unique sequences within the transcriptome, including splicing-associated snRNAs and other repeat-derived RNAs such as SINEs, LINEs, or satellite repeats. But based on need or preference, it would also be appropriate to use the total number of uniquely aligned reads as the denominator instead of the total read counts.# RPM conversion for IgG # Copy over all IgG FASTQ files to count total read numbercp /your/path/to/igg_data_folder/∗igg∗.fastq .# Count the total line number across all ∗igg∗.fastq filesfastq_lin_num_igg=(wc -l ∗igg∗.fastq | tail -1 | sed 's/ˆ ∗//g' | cut -d ' ' -f1)# Total read count is the total line number divided by 4(( reads_igg = fastq_lin_num_igg / 4 ))# Count the total line number across all ∗yourRIPname∗.fastq filesfastq_lin_num=(wc -l ∗yourRIPname∗.fastq | tail -1 | sed 's/ˆ ∗//g' | cut -d ' ' -f1)# Total read count is the total line number divided by 4(( reads = fastq_lin_num / 4 ))# RPM conversion for all replicates, $7 is the 7th column that contains the raw countcat yourRIPname_fc.txt | sed 1d | awk -v OFS="∖t" -v reads=$reads '{print $0, (($7∗1000000)/reads)}' > yourRIPname_fc_allRpm.txt# RPM conversion for individual if [ rip_file_num -gt 1 ]# Proceed if there is more than 1 replicate for the RIP-of-interestthen # Loop on indices of array rip_arr for i in {!rip_arr[@]}; do  # Index+1 so that it's 1-based for the file name  j=$(($i + 1))  # Count total line number of each individual replicate FASTQ  fastq_lin_num=$(wc -l${rip_arr[i]} | tail -1 | sed 's/ˆ ∗//g' | cut -d ' ' -f1)  # Total read count is the total line divided by 4  (( reads = fastq_lin_num / 4 ))  # RPM conversion for each replicate, 1 is the 1st column that contains the raw count  cat yourRIPname_rip${j}_counts.txt | sed 1,2d | awk -v reads=$reads '{print ((1∗1000000)/reads)}' > yourRIPname_rip{j}_rpm.txt  # Prepare to concatenate rpm files and file header  if [ j == 1 ]  then   # Initiate the table properly for the first replicate   rip_count_header="yourRIPname_rip1_counts"   rip_rpm_header="yourRIPname_rip1_rpm"   rip_rpm_filename="yourRIPname_rip1_rpm.txt"  else rip_count_header="{rip_count_header}∖tyourRIPname_rip${j}_counts" rip_rpm_header="${rip_rpm_header}∖tyourRIPname_rip${j}_rpm" rip_rpm_filename="${rip_rpm_filename} yourRIPname_rip{j}_rpm.txt"  fi done # Make summary file with the following geneID, chr, start, end, strand, length, read count across all RIP replicates, total read count for IgG control, read counts for each RIP replicate, rpm across all RIP replicates, IgG rpm, and rpm for each RIP replicate paste yourRIPname_fc_allRpm.txt yourRIPname_igg_rpm.txt rip_rpm_filename | sed "1 i∖GeneID∖tChr∖tStart∖tEnd∖tStrand∖tLength∖tyourRIPname_counts∖tigg_counts∖t${rip_count_header}∖tyourRIPname_rpm∖tigg_rpm∖t${rip_rpm_header}" > yourRIPname_fc_rpm.txtelse # If there is only 1 replicate of the RIP-of-interest # Make summary file with the following chr, start, end, strand, length, RIP read counts, IgG read counts, RIP rpm, IgG rpm paste yourRIPname_fc_allRpm.txt yourRIPname_igg_rpm.txt | sed "1 i∖GeneID∖tChr∖tStart∖tEnd∖tStrand∖tLength∖tyourRIPname_counts∖tigg_counts∖tyourRIPname_rpm∖tigg_rpm" > yourRIPname_fc_rpm.txtfi65.Identify the set of final peak regions, defined as having RIP-seq RPM values of greater than two times the RPM values for IgG controls in greater than or equal to two separate replicates.Note: As a starting point to define peaks, we recommend using a signal-over-noise threshold (RIP-over-IgG) of >2 in two or more biological replicates. However, this threshold can be adjusted by the end-user. In our experience, requiring that peak regions have signal-over-noise values of >2 in two or more biological replicates strikes a reasonable balance between reducing false positives due to noise and capturing what would be false negatives (true sites of association) due to high background, e.g., in typical exonic regions. To evaluate thresholds for peak calling in your datasets, start by using the threshold we recommend, paste the data from the output file below called “yourRIPname_ripi_2igg.txt” into Microsoft Excel, and rank the peaks by multiplying the [average RIP-seq RPM] value per peak by the ([average RIP-seq RPM]/([average RIP-seq RPM] + [average IgG RPM])) value per peak, which represents a proxy for the fraction of sequencing data under the peak that is comprised of specific RIP-seq signal versus non-specific background. Use this same system to rank potential peak regions under increasing signal-over-noise threshold regimes (e.g. requiring peaks to harbor signal-over-noise values of >3, >4, or >5 in two or more biological replicates). Visually examine the read density under peaks in a UCSC Genome Browser session that has wiggle density tracks loaded for each RIP-seq and IgG replicate (or merged RIP-seq and IgG tracks; see Steps 56–57 above). Determine the peak calling threshold that best captures your perception of likely true signal in the assay versus your perception of likely noise. For an additional approach to arrive at an acceptable signal-to-noise threshold in your assays, identify motifs that are enriched under the top sets of ranked peaks at varying signal-over-noise thresholds (e.g. >2, >3, >4, or >5); see Steps 66–69 below for instructions.if [ {rip_file_num} -gt 1 ]# Proceed if there is more than 1 replicatethen # Append a new column (last column) and store 0 for all lines # This step creates a new column in the file that # will be used to denote for each peak, # the number of replicates where RIP-over-IgG RPM values are >2 cat yourRIPname_fc_rpm.txt | sed 1d | awk -v OFS="∖t" '{print 0, 0}' > yourRIPname_ripi_2igg.txt((igg_col = 10 + rip_file_num)) # Column number of igg_rpm for i in $(seq 1$rip_file_num) # For each replicate do  ((rip_col = 10 + rip_file_num + i)) # Column number of rip_rpm  # (rip_rpm > 2∗igg_rpm) -> if TRUE, increment count in the last column  awk -i inplace -v OFS="∖t" -v c1=$rip_col -v c2=$igg_col '{if($c1 > 2 ∗$c2) {$NF=$NF+1;print} else {print 0}}' yourRIPname_ripi_2igg.txt done # Find peaks whose RIP-over-IgG RPM values are >2 # in >=2 replicates cat yourRIPname_ripi_2igg.txt | awk -v OFS="∖t" '{if(NF > 1) {print 0}}' > yourRIPname_fc_rpm_2igg_filtered.txt # Write BED files based on yourRIPname_fc_rpm_2igg_filtered.txt which identify the peaks whose RIP-over-IgG RPM values are >2 in >=2 replicates cat yourRIPname_fc_rpm_2igg_filtered.txt | awk -v OFS="∖t" '{print 2,$3-1,$4,$1,0,$5}' > yourRIPname_peaks_2igg_2reps.bedfi

Motif identification in RIP-seq peaks

Timing: 1–2 h

The following steps identify the most common N motifs of customizable widths from a set of RIP-seq peaks.Note: Path names for input and output files are demarcated as within blocks of code.66.Extract the sequence for each peak region of the input BED file from the genomic FASTA file.module load bedtools/2.31.1bedtools getfasta -s -fi -bed -fo 67.Calculate the ratio of genomic nucleotides and perform a weighted selection to generate one length-matched control sequence for each peak sequence. To do so, run the script create_bkg_seq.py (see key resources table) with Python:python3 create_bkg_seq.py genome-fasta-file.fa input-peak-seq-fasta.fa output-control-seq-fasta.fa68.Calculate motifs from the RIP datasets using STREME from the MEME Suite.^15^Note: The parameters given here specify a desired output of the top 20 motifs, with a minimum motif length of four nucleotides and a maximum motif length of twelve nucleotides.Note: More information about STREME parameters can be found at https://meme-suite.org. The values given here are recommended but can be altered to fit user needs.module load meme/5.5.7streme --oc -n --maxw 12 --minw 4 --nmotifs 20 --rna --p 69.Download output HTML files to your local computer and open in a web browser to view and download the most common motifs and their enrichment statistics.

Reverse transcription and qPCR of input and RIP samples (RIP-qPCR)

Timing: 1–2 days, longer if analyzing many qPCR plates

This step measures the relative amount of RNA(s) of interest that was recovered by each RIP and generates data that can be presented or analyzed further. We have found RIP-qPCR to be a convenient and robust way to compare signal for an individual RIP(s) over transcript regions of interest between experimental conditions, such as wild-type and knockout cells. RNA sequences in each 5% input and RIP sample are converted to cDNA, which then serves as a template for input-normalized qPCR. Because of its sensitivity, qPCR requires organization, patience, and focus to maintain uniform reaction preparation to achieve a high level of accuracy and precision. To control for variation in PCR amplification efficiencies between primer pairs, we use the “standard curve” method of qPCR quantification, described below.Note: RIP-qPCR is optional if RIP-seq is the intended endpoint of the experiment. Each eluted RIP RNA sample is typically ∼14.5 μL and has sufficient volume to be analyzed by both RIP-seq (9 μL) and RIP-qPCR (5 μL).CRITICAL: Perform work at a lab bench that is dust-free and cleaned with 70% ethanol and RNaseZap prior to work. Use only RNase-free tubes and pipette tips. Prepare and handle all samples with clean gloves. Replace gloves frequently and after touching potential sources of RNase contamination (e.g., skin, door handles, electronic devices).70.Prepare reverse transcription reactions.a.In a pre-chilled tube on ice, prepare a reverse transcription master mix that contains enough volume for all 5% input and RIP RNA samples, plus one or two to allow for pipetting error. Mix thoroughly by pipetting up and down. Each 20-μL reaction will contain the following components (all from the High-Capacity cDNA Reverse Transcription Kit [Thermo Fisher 4368813] except RNaseOUT):ReagentVolume per reaction10X RT Buffer2 μL10X Random Primers2 μL25X dNTP Mix0.8 μLUltrapure nuclease-free water (new stock)8.7 μLRNaseOUT0.5 μLMultiScribe Reverse Transcriptase1 μLTOTAL****15 μLb.To pre-chilled 0.2-mL PCR strip-tubes on ice, add 5 μL 5% input or RIP column-purified RNA sample.c.Add 15 μL well-mixed reverse transcription master mix to each tube and pipette up and down to mix thoroughly.Note: In standard RT-qPCR assays, it is common to use no-reverse-transcriptase (“-RT”) controls to confirm that qPCR signal represents RNA abundance and not the presence of contaminating DNA. However, in this protocol, the 5% input and RIP RNA samples are prepared via DNA-depleting TRIzol/chloroform extraction and on-column DNase I treatment, which together dramatically reduce the presence of DNA that could contaminate qPCR reactions. To illustrate this point, the qPCR results in Figure 2 demonstrate a 7100-fold difference in Xist qPCR signal for a 5% input RNA sample prepared with and without reverse transcriptase. For RIP RNA samples, the difference is even signal for the -RT samples was undetectable after 40 cycles of qPCR, representing at least a ∼1-million-fold difference relative to +RT samples. The absence of detectable DNA in RIP RNA samples is presumably due to these samples undergoing extensive washes during RIP, in addition to TRIzol/chloroform extraction and DNase I treatment. Therefore, as a practical matter, we do not typically include -RT controls for each RNA sample. If including -RT control(s) for RNA samples that will also be sequenced via RIP-seq, we recommend reducing the volume of RNA in each 20-μL reverse transcription reaction from 5 μL to 2.5 μL to have sufficient RNA for all applications. Including -RT controls is especially important for qPCR amplicons residing in lowly abundant transcripts and/or within introns to ensure that signal is not impacted by trace DNA contamination.Figure 2Demonstration of trace or undetectable DNA contamination in RNA samples purified using TRIzol/chloroform extraction and DNase I digestion5% input, SRSF1 RIP, and RBM15 RIP RNA samples were purified as described in Steps 38–43. cDNA samples were generated with or without reverse transcriptase and used as templates in qPCR reactions with primers targeting the Repeat A region of Xist RNA known to associate with the proteins SRSF1 and RBM15.^3^^,^^17^^,^^18^^,^^19^ Repeat A signal is reported as percent of input, derived from a standard curve prepared by serially diluting of the 5% input +RT cDNA. Dots represent technical triplicate qPCR values, and average Cq values are reported. Related to Step 69.71.Run reverse transcription reactions on a thermal cycler with the following StepsTemperatureTimePrimer annealing25°C10 mincDNA synthesis37°C120 minHeat inactivation85°C5 minHold4°Cindefinite

Pause ** Reverse transcription products can be stored at 4°C for up to 1 week or at −20°C for longer-term storage. 72.Prepare dilutions of reverse transcription products to use as templates for qPCR.a.Place the reverse transcription products at 20°C–24°C and add 20 μL ultrapure nuclease-free water to each, pipetting up and down to mix thoroughly.Note: This step reduces the viscosity of the reverse transcription products to promote pipetting accuracy and increases the volume that can serve as a template in more qPCR reactions if necessary. If an RNA target is known to be lowly abundant, this dilution step may be omitted in order to “save” approximately one Cq value.b.Label new 0.2-mL PCR strip-tubes from which the templates will be taken when preparing qPCR reactions.Note: We highly recommend making a “plate map” ahead of time to plan the layout of the 96-well qPCR plate. Prepare the qPCR templates in PCR strip-tubes in the order that they will be used for qPCR. Staying organized reduces the chances of mixing up samples or adding the wrong templates to qPCR plate wells. An example plate map is shown in Figure 3.Figure 3Example of a plate map for planning and guiding qPCR reactionsHere, reactions are prepared in technical triplicate. In this example, two RIPs (1 and 2) were performed, each with its own corresponding input, and the relative recovery of two different RNA targets (A and B) is being analyzed for each. NTC, no-template control. Related to Step 71.c.In labeled PCR-strip tubes at 20°C–24°C, construct a 6-standard standard curve from each 5% input reverse transcription product by preparing 4-fold serial dilutions with a new stock of ultrapure nuclease-free water.i.Transfer the entire 40-μL 5% input reverse transcription product to the first tube in the strip.ii.Transfer 10 μL 5% input reverse transcription product to the second tube in the strip. Add 30 μL nuclease-free water and pipette up and down to mix thoroughly.iii.Transfer 10 μL 4-diluted 5% input reverse transcription product to the third tube in the strip. Add 30 μL nuclease-free water and pipette up and down to mix thoroughly.iv.Repeat this process until the 4^5^ dilution is prepared in the sixth tube of the strip.Optional: If following the example qPCR plate map in Figure 3, consider adding the appropriate RIP reverse transcription product to the seventh tube of the strip and nuclease-free water to the eighth tube of the strip for a no-template control.73.Prepare qPCR reactions.Note: As with all parts of this protocol, wear clean gloves and take care to limit contamination of the qPCR plate with dust, hair, saliva, etc.a.In 1.7-mL tubes at 20°C–24°C, prepare qPCR master mix(es) with each containing a validated primer pair against a target sequence to be measured. Using the qPCR plate map to determine the number of reactions needed for each master mix, prepare master mixes with enough volume for all reactions plus ∼6 to allow for pipetting error and for setting up additional reaction wells if a mistake is caught during plate loading. Mix thoroughly by pipetting up and down. Each 10-μL reaction will contain the following ReagentVolume per reaction10 μM forward primer0.5 μL10 μM reverse primer0.5 μLUltrapure nuclease-free water (new stock)3 μL2X iTaq Universal SYBR Green Supermix5 μLTOTAL****9 μL**b.Obtain a 96-well qPCR plate and two 96-tip boxes of P10 barrier pipette tips, one for adding master mix(es) to the qPCR plate and one for adding template samples to the plate.Note: We strongly recommend using these pipette tips such that the position of each tip in the box matches the same position of the qPCR well being loaded. This can greatly reduce the chance of errors during plate loading. Have an additional, third P10 tip box on hand in case additional tips are needed (e.g., if a loosely added tip falls off the pipette and needs to be replaced).CRITICAL: Use high-quality qPCR plates that can handle thermal stress to limit plate warping that can lead to de-sealing and sample loss. We exclusively use Bio-Rad HSP9601 Hard-Shell qPCR plates for use in our qPCR machine (Bio-Rad C1000 Touch Thermal Cycler equipped with a CFX96 Real-Time System).c.Prepare qPCR plate and pipette tip boxes for loading.i.Use a fine-point marker to draw vertical and horizontal lines on the qPCR plate to help visually organize the wells into technical triplicates for each template sample.ii.Draw the same pattern of lines on one of the two P10 pipette tip boxes (which will be used for loading template samples).Note: These lines help to visually break down the 8-by-12 grid into smaller blocks that can reduce the chance of errors during plate loading. As an example, see the bold lines in Figure 3. Vertical lines serve as reminders to switch to the next template sample. If more than one qPCR master mix will be used on the qPCR plate, draw line(s) on the other P10 pipette tip box to demarcate the tips that will be used for loading each master mix.d.Referring to the plate map, add 9.00 μL of the appropriate qPCR master mix to each well of the qPCR plate.CRITICAL: Firmly press each pipette tip onto the P10 pipet, as loose tips are one potential source of pipetting inaccuracy. Ensure that the pipette tip only goes just below the surface of the master mix source tube to prevent any additional master mix from being carried on the outside of the tip that could cause volume inaccuracies. To better see the surface of the master mix inside the source tube without having to pick it up for loading each well, consider placing the 1.7-mL tube on a rack made for 0.5-mL tubes (often on the underside of 1.7-mL tube racks). After loading each well, visually confirm that the entire volume has been ejected from the tip and add any residual amount back to the well if necessary.CRITICAL: Ensure that the pipette remains set to 9.00 μL before loading each well with master mix, as pipettes with freely rotating volume adjusters can drift with repeated use. We recommend using a pipette with a volume adjuster that has discrete “clicks” that does not drift (e.g., Eppendorf 2231300006).Note: Loading qPCR plates requires sustained focus and involves repetitive movements that can be surprisingly strenuous, especially if multiple plates will be loaded on the same day. Reduce outside distractions that can cause loading errors by wearing headphones or asking colleagues to limit conversation with you. Neatly organize your workspace so it is ergonomic and limits unnecessary movements. We recommend using your non-dominant hand to hold the qPCR plate at a slight angle toward you while your dominant hand holds the pipette. While a full plate may take an experienced user about 45–60 min to load, it is important not to rush and to stretch and take breaks as needed.e.Referring to the plate map and taking the same considerations as in Step 72d, add 1.00 μL of the appropriate template sample to each well of the qPCR plate containing master mix.CRITICAL: To promote accuracy, eject into the master mix at the bottom of the well and pipette up and down an additional 1-2 times to flush any residual template into the reaction mixture. After loading each well, visually confirm that the entire volume has been ejected from the tip and add any residual amount back to the well if necessary.f.Once the entire qPCR plate has been loaded, seal it tightly with an appropriate adhesive sealing film (e.g., Bio-Rad 17010701 Microseal ‘B’ Sealing Film).i.Use a plastic tool (e.g., the non-sharp side of a Bio-Rad 1653320 Gel Releaser) to firmly press the seal onto the plate, first going over the top of all wells, then pressing around the edges of the plate, and finally pressing in the horizontal and vertical lines between each row and column of wells.g.Mix the contents of the reaction wells.i.Invert the qPCR plate and knock it firmly a few times on a soft surface such as a foam block or a stack of C-fold paper towels.ii.Turn the qPCR plate back over, knock again on the soft surface, and repeat a few more times.iii.Mix the plate contents further with a vortex mixer on a low (∼20%–30% maximum intensity) setting.h.Briefly pulse-spin the qPCR plate up to 650 x g in a centrifuge equipped with a plate holder to collect liquid at the bottom of the wells.74.Run qPCR reactions and subsequent (optional) melt-curve analysis on a real-time qPCR machine (e.g., Bio-Rad C1000 Touch Thermal Cycler equipped with a CFX96 Real-Time System) with the following StepsTemperatureTimeCyclesInitial denaturing95°C10 min1Denaturing95°C10 s40 cyclesAnnealing60°C30 sElongation72°C30 sPlate readN/AN/ADenaturing95°C10 s1Reannealing65°C31 s1Melting65 + 0.5x°C5 s61 cyclesPlate readN/AN/AIncrease by 0.5°CN/A1 s

Note: After the run completes, temporarily save the qPCR plate, as reaction products can be run on an agarose gel if necessary. Visually inspect the wells to ensure that no samples evaporated during the run. 75.Export data from the qPCR machine and open the plate file on a computer with the manufacturer’s recommended software. Note: The following steps describe data processing with Bio-Rad CFX Manager software (v3.1); however, the same principles apply to processing data with other software. 76.View, process, and export the qPCR data in Bio-Rad CFX Manager.a.In the menu bar at the top, select Settings > Cq Determination Mode > Regression. Then save the file (under File) to make sure this setting is saved in the file.Note: Regression-based Cq determination tends to be more accurate than threshold-based Cq determination because it considers the shapes of amplification curves at multiple points rather than the single points at which an amplification curves cross an arbitrary threshold. Nonetheless, both Cq determination methods are robust and should yield similar results. Ensure that the same method is always used within the same experiment.b.At the top-right of the menu bar, select Plate Setup > View/Edit Plate to add identifiers to each well, group wells into technical replicates, designate standard curve samples, and assign RIP cDNA samples to the appropriate standard curves to determine relative amounts of target sequence present in the cDNA samples.i.By clicking and dragging (with the use of keyboard shortcuts like Ctrl and Shift on a PC), highlight all wells that contain standards from one input-derived standard curve and the RIP samples (and any no-template controls) that correspond to that input and primer pair. Give this group a unique identifier in the Target Name field at the right, then hit Enter. Repeat for all groups of input-derived standard curves and corresponding RIP samples.ii.Designate standards by highlighting all wells of the first standard curve and selecting Sample Type as “Standard” from the drop-down menu to the right. Designate replicates with Replicate Series, setting Replicate Size to 3 if triplicate wells were used. Set the Starting Replicate # to 1 for the first standard curve; set to a higher unique number for subsequent standard curves (the Replicate # is arbitrary but cannot be re-used among standards). Click Apply. Add standard curve information with the Dilution Series. Set Starting Concentration to a value of 5, as this tells the software that the first (most concentrated) standard corresponds to 5% input and will thus conveniently report values in the units of %RIP/input. Set the Dilution Factor to 4.000 if the standard curve was constructed using 4-fold serial dilutions. Click Apply. Repeat this process for all standard curves.iii.Designate RIP samples as “unknowns” by highlighting all RIP samples and selecting Sample Type as “Unknown.” Assign each well to replicates with the Replicate Series as done for the standards. Give each RIP a sample name (e.g., "HNRNPK RIP") and hit Enter.iv.Designate no-template controls by highlighting them and selecting Sample Type as “NTC.”v.If the same plate layout will be used, even as a template, for future qPCR plates, the entries can be saved as a plate layout file using File > Extract Plate in the menu at the top of the Plate Editor window.vi.Once all information is complete, click OK at the bottom-right. Then save the file (under File) to make sure the plate information is saved.c.View the amplification curves.Note: Hovering the cursor over an amplification curve will highlight the corresponding well below and vice versa. Take note of any curves that deviate from the desired “S” shape, as this can be a sign of reaction failure, possibly from reaction evaporation or contamination. No-template controls should have no amplification (flat curve) or have amplification curves with very high (e.g., >37) Cq values. Lower Cq values in no-template controls can be a sign of contamination (see troubleshootingproblem 4).d.View Cq values for the standard curve(s). If more than one standard curve is present on the plate, first change the drop-down option in the top-right of the menu from “Fluorophore” to “Target” to view each standard curve separately.Note: Circles represent standard values, and Xs represent unknown values in the Cq vs. Log Starting Quantity (SQ) plot. Ideally, all unknowns will have Cq values that fall within the standard curve. Make note of the qPCR efficiency (E) for each standard curve – this ideally will be between 80%–100%. R^∧^2 values should be close to 1; deviations may suggest that serial dilutions were inaccurate or that qPCR technical replicates were variable. For lower-abundance RNA targets, the most dilute standard or two may have very high Cq values that are noisy and may provide unreliable information in the standard curve’s linear regression fit. In this case, consider returning to the Plate Setup > View/Edit Plate editor and changing the “Standard” designation of these dilute standard wells to “Unknown” or “Negative Control.” Ideally, each standard curve will have reliable Cq values from at least 5 standards (see troubleshootingproblems 4 and 5).e.Take note of the precision of qPCR technical replicates for each sample, which should all have similar Cq values.Note: High variability in multiple samples may indicate that efforts to improve the qPCR technique would be beneficial in a repeated qPCR plate (see troubleshootingproblem 5).Optional: If a post-amplification melt curve was run, click on the Melt Curve tab to check that all wells containing the same primer pair have similar derivative melt curves (the graph on the right). Each primer pair should have a single peak at a consistent temperature that corresponds to the temperature at which the most DNA melting occurred. These curves can serve as biophysical markers of amplicon identity and can ensure that samples were not mixed up.f.Export the qPCR data to Microsoft Excel by clicking on the Quantification Data tab, right-clicking on the table, and selecting “Export to Excel.”Note: Starting Quantity (SQ) values are in the units of %RIP/input and reflect fixed values for the standard curve samples and standard-curve-derived, input-normalized values for each RIP.77.Plot %RIP/input values for each RIP sample using software such as GraphPad Prism or R. CRITICAL: Compile data from at least two biological replicate RIP-qPCR experiments before making any biological conclusions. Statistical tests between groups should measure biological variability and use technical-replicate average values as representative for each biological replicate experiment.

Expected outcomes

The RIP-seq portion of this protocol generates wiggle files that can be opened in a genome browser to observe the RNA regions that associate with proteins of interest. Peaks can be called for sequences where the RIP signal is reproducibly above a user-determined background threshold, and these peak sequences can be analyzed further (e.g., to determine transcript features enriched for each RIP or to determine enriched sequence motifs).

The RIP-qPCR portion of this protocol generates input-normalized measurements of the extents of association between a protein(s) and specific RNA target(s) of interest. These values are reported in the unit of percent IP/input and can be plotted with the user’s graphing software of choice. Multiple replicate experiments should be reported together and used for statistical tests comparing values between genotypes or treatment groups, or against control RIPs.

For examples of RIP-seq and RIP-qPCR data generated with this protocol, see Figure 4 of Trotman et al.^1^

Limitations

Formaldehyde crosslinking captures both protein-protein and RNA-protein interactions, so this protocol cannot discern whether a detected association reflects direct RNA-protein binding or indirect interaction mediated via additional proteins. Moreover, like essentially all other methods for detecting RNA-protein interactions, the nonspecific recovery of “sticky,” protein-rich regions of RNA will occur and will contribute to noise; therefore, appropriate controls are essential. In cases where a specific RNA-protein association detected by RIP forms a centerpiece of a biological conclusion, validation with orthogonal methods is highly recommended. In our experience, the regions of highest background association in RIP assays occur surrounding exon-exon junctions of spliced RNAs; these regions associate with many RNA-binding proteins and thus typically exhibit the highest degree of recovery from non-specific IgG preparations, making it challenging to detect true signal over exon-exon junctions. In contrast, intronic regions of RNA typically have low background in RIP and often represent the clearest sites of protein association. Our own analysis of motifs under peaks of RIP-seq data generated using western-blot-validated antibodies raised against canonical RNA-binding proteins usually recovers motifs that match well with those obtained via in-vitro binding assays and UV-based crosslinking and immunoprecipitation (CLIP) assays, supporting the view that many high-confidence peaks obtained by this RIP-seq protocol represent regions of true signal, even within exons.^3^^,^^4^^,^^8^

It is important to note that canonical RNA-binding proteins are often highly abundant in cells and, for this reason, are generally well-suited for RIP. In contrast, RIP-seq datasets generated from proteins such as chromatin-modifying enzymes that are generally lower in abundance than canonical RNA-binding proteins are more likely to generate a signal that hovers close to the threshold of background (e.g., many chromatin-modifying enzymes are >100-fold lower in abundance than the most abundant RNA-binding proteins in HeLa cells^20^). One example is our studies of the Polycomb repressive complexes (PRCs), in which we found that RIP-seq analyses of different protein constituents of the same PRC2 complex rarely recovered the same peaks of RNA in the transcriptome, suggesting that most peaks of PRC2 association detected by RIP-seq represented noise.^1^ Even in our studies of PRC1, where we did observe concordant sites of association using antibodies raised against separate components of the same complex, RIP-seq signal dropped precipitously outside of the lncRNAs Airn, Xist, and Kcnq1ot1, which are among the most potent PRC-controlling RNAs known.^1^ These conclusions echo those from a recent study that utilized the UV-based covalent linkage and affinity purification (CLAP) method to study interactions between proteins and RNA in vivo.^21^ RIP, CLIP, and CLAP can readily detect interactions between abundant RNA-binding proteins and RNA, but the lower the abundance of the protein target of the immunoprecipitation, the more difficult it becomes to detect true interactions above the inherent level of background that arises from more abundant RNA-binding proteins.

Despite the sensitivity of RIP-qPCR, cases of low target RNA abundance will make qPCR detection of immunoprecipitated RNA difficult and potentially infeasible. The RNA target sequence must be uniquely amplifiable by PCR, which precludes study of RNA species smaller than ∼50–60 nucleotides and may preclude study of RNA sequences with low complexity, extreme GC content, or redundancy elsewhere in the transcriptome.

Troubleshooting

Problem 1

No amplification of qPCR product during primer validation (related to Step 9).

Potential solution

Check that the primer sequences were designed correctly and use a tool such as Primer-BLAST to confirm that the primers should theoretically amplify the target. Use a tool such as IDT’s OligoAnalyzer to ensure that the primers are not predicted to form strong secondary structures (hairpins) or homo- or heterodimers that could impede PCR. Repeat the qPCR, ensuring that the reactions are prepared properly, and add positive control reactions with a previously validated primer pair to rule out issues with reagents or the qPCR machine. If no amplification is still observed, consider designing a new pair of primers with 3′ extensions or targeting a slightly different sequence.

Problem 2

Death or differentiation of mESCs (related to Steps 10–11).

Potential solution

If cells appear dead upon thawing, ensure that the freezing medium had been prepared properly and that cells had not been exposed to high concentrations of cryoprotectant DMSO for more than a few minutes (see Step 10c). If mESCs had been living but died unexpectedly while culturing, ensure that β-mercaptoethanol has been added to the culture medium, as this prevents ferroptosis.^22^^,^^23^ If cells lose normal mESC morphology, this is a sign that cells could be differentiating. Ensure that LIF has been added to the culture medium and that cells are not split too thinly by aiming for at least 40% confluence the day following a split.

Problem 3

Low RIP-seq library yield (related to Step 45).

Potential solution

Pre-amplification RIP-seq libraries that require more than ∼21 cycles for library amplification (especially for non-IgG or non-negative-control RIPs) may indicate substandard library preparation. Ensure that beads have not been overdried by following the modifications to the protocol regarding 80% ethanol wash removal in Steps 44b and 46b. Ensure that only RNase-free tubes and pipette tips are used, always work at a clean workspace, and ensure that clean gloves are used during all RNA-handling steps to prevent RNase contamination. If qPCR estimates suggest that more than ∼20 cycles of PCR will be necessary for library amplification, yet sequencing of the library is still desirable, we suggest separating the library amplification into two distinct steps. In the first step, use only 10 cycles of PCR amplification. Then, purify the amplified library using SPRI (e.g., Beckman-Coulter AMPure) beads, and perform a second qPCR quantification to estimate how many additional cycles of PCR will be needed to amplify sufficiently for subsequent sequencing (e.g., repeat Step 45). This second round of PCR should use between 10 and 20 cycles. Separating the PCR amplification into two rounds in this way helps to ensure that PCR reagents do not become limiting at high PCR cycle numbers. The library in question is more likely to be amplified linearly as a result.

Problem 4

Non-linear Cq vs. Log Starting Quantity qPCR standard curves due to higher-than-expected amplification of the most dilute standards and/or qPCR amplification of no-template control samples with Cq < 35 (related to Steps 9, 45, 72).

Potential solution

This is a likely sign of contamination with some foreign material that is causing non-specific amplification. Consider all of the following clean your workspace before working (especially if dust is present), obtain a new stock of ultrapure nuclease-free water, prepare fresh working stocks of primers with new ultrapure nuclease-free water, wear new gloves (and change them if they touch a potential source of contamination), hold the qPCR plate away from your body while loading (i.e., don’t lean over it directly, breathe on it, or talk while loading), and consider wearing a mask or working in a PCR hood if the issue is persistent.

Problem 5

High variability among qPCR technical replicates (related to Steps 75–76).

Potential solution

Technical-replicate variability can be a sign of material issues but is usually caused by human error. Ensure that high-quality qPCR plates and seals are used, as sample evaporation and plate warping can affect PCR amplification and optical measurements. Make adjustments to the technique used to load the qPCR plate, following the recommendations in Step 72. Do not be loading qPCR plates with high technical precision can be difficult and should improve with practice. If the issue persists and is experienced by other users of the same qPCR machine, consider contacting the machine manufacturer’s product support for maintenance and/or recalibration.

Resource availability

Lead contact

Further information and requests for resources and reagents should be directed to and will be fulfilled by the lead contact, J. Mauro Calabrese (jmcalabr@med.unc.edu).

Technical contact

Technical questions on executing this protocol should be directed to and will be answered by the technical contact, Jackson B. Trotman (jbt@unc.edu).

Materials availability

This study did not generate new unique reagents.

Data and code availability

This study did not generate new datasets. All code, including Perl and Python scripts, is available at https://github.com/CalabreseLab/Protocol-for-evaluating-RNA-protein-associations-in-mammalian-cells-with-RIP-seq-and-RIP-qPCR. For more information, please refer to data and code available in Trotman et al.^1^

Acknowledgments

We thank Brian Golitz, Noah Sciaky, and Andrew Snipes of the UNC CRISPR Screening Facility for sequencing. This work was supported by NIH grants R01GM136819, R01GM121806, and R35GM153293; NSF grant DBI-2228805; and the Yang Family Biomedical Scholar Fund (J.M.C.). J.B.T. was supported by NIH fellowship T32CA217824 and a Marzluff Postdoctoral Fellowship from the RNA Discovery Center. Q.E.E. was supported by NIH fellowship F31HG014413.

Author contributions

J.B.T. and J.M.C. conceived the study; J.B.T., S.L., Q.E.E., Z.Z., and J.M.C. performed the experiments and/or data analysis; J.B.T., Q.E.E., and J.M.C. acquired funding; and J.B.T., S.L., Q.E.E., Z.Z., and J.M.C. wrote the paper.

Declaration of interests

The authors declare no competing interests.