Here, we present a protocol to enrich endogenous RNA binding sites or "footprints" of RNA:protein (RNP) complexes from mammalian cells. This approach involves two immunoprecipitations of RNP subunits and is therefore dubbed RNA immunoprecipitation in tandem (RIPiT).
RNA immunoprecipitation in tandem (RIPiT) is a method for enriching RNA footprints of a pair of proteins within an RNA:protein (RNP) complex. RIPiT employs two purification steps. First, immunoprecipitation of a tagged RNP subunit is followed by mild RNase digestion and subsequent non-denaturing affinity elution. A second immunoprecipitation of another RNP subunit allows for enrichment of a defined complex. Following a denaturing elution of RNAs and proteins, the RNA footprints are converted into high-throughput DNA sequencing libraries. Unlike the more popular ultraviolet (UV) crosslinking followed by immunoprecipitation (CLIP) approach to enrich RBP binding sites, RIPiT is UV-crosslinking independent. Hence RIPiT can be applied to numerous proteins present in the RNA interactome and beyond that are essential to RNA regulation but do not directly contact the RNA or UV-crosslink poorly to RNA. The two purification steps in RIPiT provide an additional advantage of identifying binding sites where a protein of interest acts in partnership with another cofactor. The double purification strategy also serves to enhance signal by limiting background. Here, we provide a step-wise procedure to perform RIPiT and to generate high-throughput sequencing libraries from isolated RNA footprints. We also outline RIPiT's advantages and applications and discuss some of its limitations.
Within cells, RNA exists in complex with proteins to form RNA:protein complexes (RNPs). RNPs are assembled around RNA binding proteins (RBPs, those that directly bind RNA) but also comprise of non-RBPs (those that bind RBPs but not RNA), and are often dynamic in nature. RBPs and their cofactors function collectively within RNPs to execute regulatory functions. For example, in the nonsense-mediated mRNA decay (NMD) pathway, the UPF proteins (UPF1, UPF2, and UPF3b) recognize the prematurely terminated ribosome. Each of the UPF proteins can bind to RNA, but it is only when they assemble together that an active NMD complex begins to form. Within this complex, UPF1 is further activated by phosphorylation by a non-RBP SMG1, and such UPF1 activation eventually leads to recruitment of mRNA decay inducing factors1,2. In this example, RBPs require non-RBP cofactors for recruitment and activation of the RNP complex that triggers NMD. Yet another property of RNPs is their compositional heterogeneity. Consider the spliceosome, which exists in distinct E, A, B or C complexes. Different spliceosome complexes have overlapping and distinct proteins3. To study RNP functions, it is important to elucidate which RNAs are bound by an RBP and its associated proteins. Many methods exist to accomplish this, with each approach having its distinct advantages and disadvantages4,5,6,7.
The widely popular methods to identify RBP binding sites — crosslinking followed by immunoprecipitation (CLIP) and its various variations - rely on ultraviolet (UV) light to crosslink an RBP to RNA8. However, this is not an effective approach for non-RBPs within RNPs, which do not contact the RNA directly. Here, we describe an alternative approach that is applicable to RBPs and non-RBPs alike, to isolate and identify their RNA binding sites. This approach termed RNA immunoprecipitation in tandem (RIPiT) consists of two immunoprecipitation steps, which help achieve higher specificity as compared to a single purification (Figure 1)9,10. As the individual immunoprecipitation (IP) steps can be carried out at a lower stringency as compared to CLIP, RIPiT does not depend on availability of antibodies that can withstand presence of strong detergents during immunoprecipitation. The most unique advantage of RIPiT is the ability to target two different proteins in two purification steps; this provides a powerful way to enrich a compositionally distinct RNP complex from other similar complexes11.
Small variations to the RIPiT procedure can further enhance RNP enrichment. For instance, some RNA-protein or protein-protein interactions within RNPs are transient and it may be difficult to efficiently purify footprints of such complexes. To stabilize such interactions, RNPs can be crosslinked within cells with formaldehyde prior to cell lysis and RIPiT. For example, we have observed that a weak interaction between the exon junction complex (EJC) core factor, EIF4AIII and the EJC disassembly factor, PYM12 can be stabilized with formaldehyde treatment such that more RNA footprints are enriched (data not shown). Prior to cell harvesting and RIPiT, cells can also be treated with drugs to stabilize or enrich RNPs in a particular state. For example, when studying proteins that are removed from mRNA during translation (e.g., the EJC13, UPF114), treatment with translation inhibitors such as puromycin, cycloheximide or harringtonine can lead to increased occupancy of proteins on RNAs.
The amount of RNA recovered from RIPiT is usually low (0.5-10 pmoles, i.e., 10-250 ng RNA considering an average RNA length of 75 nt). The primary reason for this is that only a small fraction of a given protein is present in complex with other proteins within RNPs (any "free" protein IP'ed in the first step is lost during the second IP). To generate RNA-Seq libraries from this RNA, we also outline here an adaptation of previously published protocol suitable for such low RNA inputs15,16 (Figure 2), which yields high-throughput sequencing ready samples in 3 days.
1. Establishment of Stable HEK293 Cell Lines Expressing Tetracycline-inducible FLAG-tagged Protein of Interest (POI)
2. Culturing Cells for Tetracycline Induction and RIPiT Procedure
3. Cell harvesting, Formaldehyde Treatment, and Cell Lysis
4. FLAG Immunoprecipitation
5. RNase I Digestion
6. Affinity Elution
7. Magnetic Bead-antibody Conjugation
8. Second Immunoprecipitation
9. Denaturing Elution
10. RNA Extraction and End Curing
11. Estimation of RNA Footprint Size and Abundance
12. Adapter Ligation
13. Reverse Transcription
14. Purification of RT Product
15. Circularization of RT Product
16. Test PCR
17. Large-scale PCR
A successful RIPiT will result in the immunoprecipitation of both proteins of interest and other known interacting proteins, and the absence of non-interacting proteins. As seen in Figure 3A, both Magoh and EIF4AIII were detected in the RIPiT elution, but HNRNPA1 was not (lane 6). In parallel, RNA footprints that have co-purified with the RNP complexes was detected via autoradiography (Figure 3B) or bioanalyzer (Figure 3C). Puromycin treatment is expected to increase EJC occupancy on RNA, and a stronger RNA footprint signal was observed in the puromycin treated RIPiT in Figure 3B (compare lanes 2 and 3). Generating samples for deep sequencing requires ligating an adapter to the RNA, and then reverse transcribing the RNA into DNA using a primer that anneals to the adapter sequence. The reverse transcription step incorporates biotinylated nucleotides, for purification of reverse transcription product. Following the reverse transcription, the product must be separated from the unextended adapter by urea-PAGE (Figure 4). The reverse transcription product is then circularized and PCR amplified. The appropriate number of PCR cycles must not overamplify the circularized product. Overamplification will result in primer depletion and aberrant PCR product (see Figure 5, lane 4). The number of cycles with the greatest amplification without evidence of overamplification is most appropriate to use for a large-scale PCR (Figure 5 lane 3 and Figure 6).
Figure 1: Schematic outlining the main steps in RIPiT. Please click here to view a larger version of this figure.
Figure 2: Schematic depicting the workflow for conversion of RIPiT RNA into libraries for high-throughput sequencing. Please click here to view a larger version of this figure.
Figure 3: Estimation of RNA and protein yields from RIPiT. (A) Western blot of the proteins purified from each major step in the RIPiT procedure. (B) Autoradiograph image of RNA footprints from a FLAG-MAGOH:EIF4AIII RIPiT comparing puromycin treated and untreated cells. Red box indicates the RNA footprint size that will ultimately be converted into sequencing libraries. (C) Profile of RNA eluted from RIPiT as in lane 2 in panel B when visualized using a bioanalyzer. Please click here to view a larger version of this figure.
Figure 4: Reverse transcription (RT) product resolved on a 10% urea-PAGE gel and stained with gold nucleic acid gel stain. Red box indicates gel region excised for gel purification of extended RT products. Please click here to view a larger version of this figure.
Figure 5: Test PCR resolved on 8% non-denaturing PAGE. Note the aberrantly large PCR products which appear at 14 cycles (red box) and the parallel depletion of primers (red arrow) indicative of overamplification. For this sample, 11 cycles was chosen for the large-scale PCR (blue box). Please click here to view a larger version of this figure.
Figure 6: Large-scale PCR resolved on 8% non-denaturing PAGE. Red box indicates the gel piece excised for gel purification. Please click here to view a larger version of this figure.
Figure 7: Genome browser screenshot of the MAPK1 gene showing distribution of FLAG-MAGOH:EIF4AIII footprints as representative results from a RIPiT. Red arrows denote the expected canonical EJC binding sites. Please click here to view a larger version of this figure.
PBS | 137 mM NaCl 2.7 mM KCl 10 mM Na2HPO4·7H2O KH2PO4 pH 7.4 |
|
Quenching Buffer | 2.5 M Glycine 2.5 mM Tris-Base |
|
Hypotonic Lysis Buffer (HLB) | 20 mM Tris-HCl pH 7.5 15 mM NaCl 10 mM EDTA 0.5% IGEPAL 0.1% Triton-X-100 1x Aprotinin* 1x Leupeptin* 1x Pepstatin* 1 mM PMSF (phenylmethylsulfonyl fluoride)* |
*must be added fresh every time |
Isotonic Wash Buffer (IsoWB) | 20 mM Tris-HCl pH 7.5 150 mM NaCl 0.1% IGEPAL |
|
Conjugation Buffer | 0.02% Polysorbate-20 1x PBS |
|
Clear Sample Buffer | 100 mM Tris-HCl pH 6.8 4% SDS 10 mM EDTA |
|
4xdNTPmix | 0.25 mM dGTP 0.25 mM dTTP 0.175 mM dATP 0.1625 mM dCTP 0.075 mM biotin-dATP 0.0875 mM biotin-dCTP |
|
5x First-Strand Buffer w/o MgCl2 | 250 mM Tris-HCl pH 8 375 mM KCl |
|
RIPiT dilution buffer (1 mL) | 1 mL IsoWB 5 μL 200x BSA 20 μL 10% Triton-X-100 20 μL 0.5M EDTA 1x Aprotinin* 1x Leupeptin* 1x Pepstatin* 1 mM PMSF* |
*must be added fresh every time |
2x Denaturing Load Buffer | 3 mL 5x TBE 1.8 g Ficoll Type 400 6.3 g urea 3 mg bromophenol blue 3 mg xylene cyanol Adjust volume to 15 mL ddH2O |
To get into solution, place tube in water in a beaker and boil on hot plate for 10–15 min. Add dyes after adjusting the volume to 15 mL |
Urea-PAGE gel | 6 M Urea Acrylamide:bisacrylamide (40% [w/v]) to appropriate percentage 0.5x TBE 0.01% TEMED |
|
DNA Elution Buffer | 300 mM NaCl 1 mM EDTA |
|
Streptavidin Bead Wash Buffer | 0.5 M NaOH 20 mM Tris-HCl pH 7.5 1 mM EDTA |
Table 1: Buffers.
We discuss here some key considerations to successfully perform RIPiT. Foremost, individual IPs must be optimized to achieve highest possible efficiency at each step. The amount of FLAG agarose beads for the input number of cells described here has proven to be robust for a wide range of proteins we have tested. As only a small fraction of partner proteins is co-immunoprecipitated with the FLAG protein, the amount of antibody needed for efficient second IP is usually low (less than 10 µg). Small-scale RIPiT (from one 10 cm plate) followed by western blot verification of proteins in each fraction during the two immunoprecipitation steps prove extremely useful to assess efficiency and specificity of the procedure before scaling up. Both targeted proteins as well as any other expected interacting proteins in the complex should be detected in the elution. It is also beneficial to assay for proteins in the depleted lysates (unbound to the FLAG-agarose or magnetic beads) to have a good estimate of the immunoprecipitation efficiency and the percentage of proteins assembling into a complex. Further, this analysis also informs if the RNase digestion conditions are sufficient to separate RNA-dependent interactions from RNA-independent interactions within an RNP. Therefore, it is just as important to include a negative control, ideally an RBP unrelated to the RNP of interest. For example, in Figure 3A, HNRNPA1 is present in the input but is not detected in the RIPiT elution. HNRNPA1 is an RBP that does not directly interact with the EJC but interacts indirectly with the EJC when the EJC and HNRNPA1 are bound to the same RNA molecule. Detection of the negative control protein in the elution indicates either poor RIPiT specificity or insufficient RNA footprinting. In such a case, the RNA footprints obtained will not completely reflect the footprints of the protein of interest. Footprints of size 50-200 nt are recommended for subsequent RNA-Seq. Duration of RNase I treatment or the amount of enzyme used can be optimized to obtain desired size footprints. Note that the best-case scenario will be to obtain good signal in the desired size range, and it is unavoidable to have longer and shorter RNAs even in the most optimal conditions. RIPiT can also be used to obtain binding sites of a single RBP. In such a case, the same protein can be immunoprecipitated with two different antibodies, first using an antibody against an affinity tag and then with antibodies against the protein itself18. Finally, a negative control RIPiT can be performed in parallel from cells expressing a FLAG-tagged control protein (e.g., green fluorescent protein) in combination with antibody against a protein against an unrelated protein in the second IP.
Despite its many advantages, it is important to consider some limitations of the RIPiT approach, and possible remedies. The requirement of affinity elution after the first purification necessitates the biological source to express a tagged protein. If a site-specific recombination system is not available in the cell line or organism of interest, a short affinity tag such as a FLAG tag (8 amino acids) can be introduced at the endogenous gene locus using CRISPR/Cas-based genome editing approach19. The FLAG tag is an ideal epitope for this approach, because the FLAG antibody is well-suited for affinity elution and can withstand high ionic strengths and mild denaturing conditions that can be used in combination with formaldehyde crosslinking. Another limitation of the RIPiT approach is the requirement for a large input of cellular material. This may remain unavoidable to some extent as only a small percentage of an RBP likely interacts with other proteins in the RNP. Still improved library preparation approaches can help to bring down the large input requirement. Possible ideas to further streamline these steps include carrying out the RNA 3'-end dephosphorylation and adapter ligation on the magnetic beads immediately after second IP washes and prior to the final elution of RNP. Such an approach is successfully implemented in current CLIP-Seq procedures and in a recently described variation of RIPiT20. Such changes will also remove several time-consuming RNA purification steps from the early phases of library preparation procedure. Further, unlike CLIP, which provides a nucleotide level resolution of crosslinking site of an RBP on the RNA, resolution of RIPiT footprints will remain at the level of tens of nucleotides. Finally, as RNPs may include multiple RBPs, the RIPiT enriched RNA sites include a mixture of binding sites of many RBPs. As consensus sequences bound by individual RBPs are being uncovered at a rapidly increasing pace and are now readily available21,22,23, this information can be leveraged to deconvolve the assortment of RBP sites enriched in RIPiT outputs. Notwithstanding these challenges, RIPiT-Seq is an effective procedure for capturing RNA footprints of dynamic, heterogeneous, and even transient RNP complexes, which can provide unique insights into the inner workings of RNA machineries that control cellular function.
The authors have nothing to disclose.
This work was supported by the NIH grant GM120209 (GS). The authors thank the OSUCCC Genomics Shared Resources Core for their services (CCC Support Grant NCI P30 CA16058).
Anti-FLAG Affinity Gel | Sigma | A2220 | |
ATP, [γ-32P]- 3000Ci/mmol 10mCi/ml EasyTide, 250µCi | PerkinElmer | BLU502A250UC | |
BD Disposable Syringes with Luer-Lok Tips (200) | Fisher | 14-823-435 | |
Betaine 5M | Sigma | B0300 | |
biotin-dATP | TriLink | N-5002 | |
biotin-dCTP | Perkin Elmer | NEL540001EA | |
Branson Sonifier, Model SSE-1 | Branson | ||
CircLigase I | VWR | 76081-606 | ssDNA ligase I |
DMEM, High Glucose | ThermoFisher | 11995-065 | |
DNA load buffer NEB | NEB | ||
Dynabeads Protein A | LifeTech | 10002D | |
Flp-In-T-REx 293 Cell Line | ThermoFisher | R78007 | |
GeneRuler Low Range DNA Ladder | ThermoScientific | FERSM1203 | |
Hygromycin B | ThermoFisher | 10687010 | |
Mini-PROTEAN TBE Gel 10 well | Bio-Rad | 4565013 | |
Mini-PROTEAN TBE-Urea Gel | Bio-Rad | 4566033 | |
miRCAT-33 adapter 5′-TGGAATTCTCGGGTGCCAAGGddC-3′ | Any | this protocol is only compatible with the Illumina sequencing platform | |
Mirus transIT-X2 transfection reagent | Mirus | MIR 6004 | |
Mth RNA ligase | NEB | E2610S | |
PE1.0 5′-AATGATACGGCGACCACCGAGATCTACACT CTTTCCCTACACGACGCTCTTCCGATC*T-3′ |
Any | this protocol is only compatible with the Illumina sequencing platform | |
PE2.0 5′-CAAGCAGAAGACGGCATACGAGATCGGTCTC GGCATTCCTGCTGAACCGCTCTTCCGATC*T-3′ |
Any | this protocol is only compatible with the Illumina sequencing platform | |
Phenol/Chloroform/Isoamyl Alcohol (25:24:1, pH 6.7, 100ml) | Fisher | BP1752I-100 | |
Purple Gel Loading Dye (6x) | NEB | NEB #7025 | |
Q5 DNA Polymerase | NEB | M0491S/L | |
RNase I, E. coli, 1000 units | Eppicenter | N6901K | |
SPIN-X column | Corning | CLS8160-24EA | |
Streptavidin beads | ThermoFisher | 60210 | |
Superscript III (SSIII) | ThermoScientific | 18080044 | reverse transcriptase enzyme |
SybrGold | ThermoFisher | S11494 | gold nucleic acid gel stain |
T4 Polynucleotide Kinase-2500U | NEB | M0201L | |
T4RNL2 Tr. K227Q | NEB | M0351S | |
Tetracycline | Sigma | 87128 | |
Thermostable 5´ App DNA/RNA Ligase | NEB | M0319S | |
TruSeq_SE1 5′-pGGCACTANNNNNAGATCGGAAGA GCGTCGTGTAGGGAAAGAGTGT-SPACER 18-CTCGGCATTCCTGCTGAACCGCTC TTCCGATCTCCTTGGCACCCGAGAATTCCA-3′ |
Any | this protocol is only compatible with the Illumina sequencing platform | |
TruSeq_SE10 5′-pGGTGTTCNNNNNAGATCGGAAG AGCGTCGTGTAGGGAAAGAGTGT-SPACER 18-CTCGGCATTCCTGCTGAACCGCT CTTCCGATCTCCTTGGCACCCGAGAATTCCA-3′ |
Any | this protocol is only compatible with the Illumina sequencing platform | |
TruSeq_SE11 5′-pGGTAAGTNNNNNAGATCGGAA GAGCGTCGTGTAGGGAAAGAGTGT-SPACER 18-CTCGGCATTCCTGCTGAACCGCTC TTCCGATCTCCTTGGCACCCGAGAATTCCA-3′ |
Any | this protocol is only compatible with the Illumina sequencing platform | |
TruSeq_SE12 5′-pGGAGATGNNNNNAGATCGGAAGA GCGTCGTGTAGGGAAAGAGTGT-SPACER 18-CTCGGCATTCCTGCTGAACCGCTC TTCCGATCTCCTTGGCACCCGAGAATTCCA-3′ |
Any | this protocol is only compatible with the Illumina sequencing platform | |
TruSeq_SE2 5′-pGGGTAGCNNNNNAGATCGGAAGAG CGTCGTGTAGGGAAAGAGTGT-SPACER 18-CTCGGCATTCCTGCTGAACCGCT CTTCCGATCTCCTTGGCACCCGAGAATTCCA-3′ |
Any | this protocol is only compatible with the Illumina sequencing platform | |
TruSeq_SE35′-pGGTCGATNNNNNAGATCGGAAG AGCGTCGTGTAGGGAAAGAGTGT-SPACER 18-CTCGGCATTCCTGCTGAACCGCT CTTCCGATCTCCTTGGCACCCGAGAATTCCA-3′ |
Any | this protocol is only compatible with the Illumina sequencing platform | |
TruSeq_SE4 5′-pGGCCTCGNNNNNAGATCGGAAGA GCGTCGTGTAGGGAAAGAGTGT-SPACER 18-CTCGGCATTCCTGCTGAACCGCTC TTCCGATCTCCTTGGCACCCGAGAATTCCA-3′ |
Any | this protocol is only compatible with the Illumina sequencing platform | |
TruSeq_SE5 5′-pGGTGACANNNNNAGATCGGAAGA GCGTCGTGTAGGGAAAGAGTGT-SPACER 18-CTCGGCATTCCTGCTGAACCGCTC TTCCGATCTCCTTGGCACCCGAGAATTCCA-3′ |
Any | this protocol is only compatible with the Illumina sequencing platform | |
TruSeq_SE6 5′-pGGTAGACNNNNNAGATCGGAAGAG CGTCGTGTAGGGAAAGAGTGT-SPACER 18-CTCGGCATTCCTGCTGAACCGCTCTTC CGATCTCCTTGGCACCCGAGAATTCCA-3′ |
Any | this protocol is only compatible with the Illumina sequencing platform | |
TruSeq_SE7 5′-pGGGCCCTNNNNNAGATCGGAAG AGCGTCGTGTAGGGAAAGAGTGT-SPACER 18-CTCGGCATTCCTGCTGAACCGCTCT TCCGATCTCCTTGGCACCCGAGAATTCCA-3′ |
Any | this protocol is only compatible with the Illumina sequencing platform | |
TruSeq_SE8 5′-pGGATCGGNNNNNAGATCGGAAGAG CGTCGTGTAGGGAAAGAGTGT-SPACER 18-CTCGGCATTCCTGCTGAACCGCTCTT CCGATCTCCTTGGCACCCGAGAATTCCA-3′ |
Any | this protocol is only compatible with the Illumina sequencing platform | |
TruSeq_SE9 5′-pGGACTGANNNNNAGATCGGAAGAG CGTCGTGTAGGGAAAGAGTGT-SPACER 18-CTCGGCATTCCTGCTGAACCGCTCTTC CGATCTCCTTGGCACCCGAGAATTCCA-3′ |
Any | this protocol is only compatible with the Illumina sequencing platform | |
Typhoon 5 Bimolecular Imager | GE Healthcare Life Science | 29187191 |