Protocol to measure protein-RNA binding using double filter-binding assays followed by phosphorimaging or high-throughput sequencing

Summary Binding affinity quantitatively describes the strength of a molecular interaction and is reported by the equilibrium dissociation constant (KD). Here, we present a protocol to measure KD of mammalian microRNA-loaded Argonaute2 protein by double filter binding. We describe steps for radiolabeling target RNA, measuring concentration of binding-competent protein, setting up binding reactions, separating protein-bound RNA from protein-unbound RNA, preparing library for Illumina sequencing, and performing data analysis. Our protocol is easily applied to other RNA- or DNA-binding proteins. For complete details on the use and execution of this protocol, please refer to Jouravleva et al.1


Highlights
A single-step binding assay that uses a simple experimental setup Double filter-binding assays quantify concentration of active protein Double filter-binding assays measure the affinity of a protein for a target RNA Readily adapted to high-throughput measurements by RNA Bind-n-Seq Note: A double filter-binding assay for a single RNA target typically requires $20 fmol RISC; for a single RNA Bind-n-Seq (RBNS) assay $170 fmol RISC is needed.
2. Prepare cytosolic S100 extracts according to ref. 7 a. Prepare Buffer A, Buffer B, and Buffer C. Buffers without DTT can be stored at room temperature (20 CÀ25 C) up to 3 months. c. Harvest cells.
i. Pour off cell media from a cell dish into waste pot. ii. Add $5 mL ice-cold PBS onto the plate to cover the cell layer.
iii. Using a cell scraper, gently scrape the cells off the bottom of the plate into PBS. iv. Transfer the detached cells into a 50-mL tube. v. Repeat for the all cell dishes, collecting cells in the same tube. d. Centrifuge cell at 500 3 g for 5 min at 4 C. e. Wash cell pellets once with ice-cold PBS.
i. Remove the supernatant from the previous step.
iii. Centrifuge at 500 3 g for 5 min at 4 C. f. Remove all the supernatant and proceed to the next step. Alternatively, dry cell pellets can be snap frozen in liquid nitrogen and stored at -80 C until ready for the next step. g. Resuspend the cell pellet in twice its volume with Buffer A supplemented with Proteinase Inhibitor Cocktail. h. Incubate on ice for 20 min to allow the cells to swell. i. Lyse cells on ice with a Dounce homogenizer using 40 strokes of a tight pestle (B type). j. Centrifuge the homogenate at 2,000 3 g for 10 min at 4 C to remove nuclei and cell membranes. k. Collect the supernatant and add 0.11 volumes (relative to the volume of the clarified supernatant) of Buffer B supplemented with Proteinase Inhibitor Cocktail. Mix by gentle inversion. l. Centrifuge at 100,000 3 g for 20 min at 4 C. Collect the supernatant; the supernatant corresponds to the S100 extract. m. Add 0.32 volumes ice-cold Buffer C supplemented with Proteinase Inhibitor Cocktail to S100 to achieve a 20% (w/v) final glycerol concentration and mix by gentle inversion. n. Aliquot S100, flash-freeze in liquid nitrogen, and store at À80 C for up to 6 months. 3. Assemble RISC ( Figure 1). a. Use 20 mL anti-FLAG M2 paramagnetic beads (Sigma) per mL S100. b. Briefly wash the beads three times with 1 mL ice-cold Buffer D supplemented with Proteinase Inhibitor Cocktail.
c. Add S100 to the beads and capture FLAG-AGO2 protein by rotating the beads for 2 h at 4 C. d. Briefly wash the beads three times with 1 mL room-temperature Buffer D supplemented with Proteinase Inhibitor Cocktail. e. Load immobilized AGO2 by incubating the beads with 1 mM single-stranded, 5 0 -phosphorylated synthetic miR-449a guide (5 0 -UGG CAG UGU AUU GUU AGC UGG U; resuspended in water) in 250 mL Buffer D rotating for 1 h at 37 C. f. Briefly wash the beads three times with 1 mL room-temperature Buffer D to remove unbound synthetic miRNA. g. Incubate the beads rotating for 1 h at room temperature (20 CÀ25 C) with 100 ng$mL À1 3XFLAG peptide in 400 mL Buffer D. h. Collect the eluate and place it on ice. i. Repeat elution. j. Combine both eluates, comprising unloaded protein and AGO2 loaded with exogenous or endogenous small RNA guides. 4. Purify RISCs as described 8 (Figure 1).
Note: RISC is captured using an RNA oligonucleotide containing a high affinity binding site (complementary to miRNA nucleotides (nt) 2À8 and 13À16). RISC is eluted by adding an excess of competitor cDNA oligonucleotide, which is perfectly complementary to the capture oligonucleotide. The strategy takes advantage of the finding that dissociation of RISC from a target RNA is > 1,000 times more rapid then melting of an RNA:DNA duplex. 8,9 a. Use 20 mL streptavidin paramagnetic beads (Dynabeads MyOne Streptavidin T1, Life Technologies) per mL S100. b. Bind a biotinylated, 2 0 -O-methyl capture oligonucleotide (5 0 -GAU CAA CAA UAA CCC ACC ACU GCC UAU AGA; complementary to miRNA nucleotides 2À8 and 13À16) to the beads according to the manufacturer's instructions. c. Incubate the assembled RISC (step 3j) with the beads rotating 1 h at room temperature (20 CÀ25 C) or overnight (12)(13)(14)(15)(16)(17)(18) h) at 4 C. d. Prepare Buffers E and F. Buffers without DTT can be stored at room temperature (20 CÀ25 C) up to 3 months. The procedure below describes preparation of 5 0 -end labeled target RNA in two steps: (1) dephosphorylation of 5 0 phosphorylated synthetic RNA oligonucleotides followed by (2) 5 0 -end labeling with radioactive ATP ([g-32 P]ATP) using T4 Polynucleotide Kinase (PNK).
Note: To achieve high sensitivity, we radioactively 5 0 -end label the target RNA. A typical double filter-binding assay or RBNS experiment requires $30 fmol and $20 pmol of target RNA, respectively. To ensure a high concentration of target RNA, we radioactively label a small amount (e.g., 100 pmol) of target RNA and mix it with cold target RNA to reach the amount required in an assay. If the experimentalist purchases target RNA both with 5 0 phosphate and without, proceed directly with 5 0 -end labeling (step 2). For best results, do not use a radioactive batch of [g-32 P]ATP that is more than 3 weeks old.
CRITICAL: DEPC-treated water should not be used, because it can be acidic and may cause depurination. Unreacted DEPC or by-products from autoclaving can also inhibit enzymes.  b. Incubate at 37 C for 30 min. c. Inactivate the Antarctic Phosphatase by adding 50 mM EDTA (f.c.) and heating at 80 C for 2 min. Cool at room temperature (20 CÀ25 C) for 1 min. d. Add RNase-free water to adjust sample volume to 300 mL. e. Precipitate the RNA by adding 0.1 volumes of 3 M sodium acetate (pH 5.2) and three volumes of ethanol. f. Incubate the solution 2 h at À80 C or overnight (12-18 h) at À20 C. g. Spin at the maximum speed (typically 21,000 3 g) for 30 min at 4 C in a microcentrifuge. h. Carefully remove nearly all supernatant. i. Add 1 mL of ice-cold 75% ethanol to RNA. Mix well by vigorous vortexing. j. Spin at the maximum speed for 15 min at 4 C in a microcentrifuge. k. Carefully remove nearly all supernatant.  l. Repeat steps 1i-k. m. Spin at the maximum speed for 15 s at 4 C in a microcentrifuge to collect the residual ethanol on the wall of the tube and discard the liquid using an extended pipette tip. n. Air-dry the pellet for 5À20 min. o. Resuspend the pellet in 10 mL nuclease-free water. 2. Set up a 5 0 -end labeling reaction.
CRITICAL: The protocol involves the use of radiolabeled nucleotides; follow institutional guidelines for working with radionuclides.
b. Incubate at 37 C for 30 min.
Note: To ensure that all target RNA molecules are 5 0 -phosphorylated, follow the radiolabeling with a second phosphorylation reaction containing 2 mM non-radioactive ATP. To the radiolabeling reaction, add: i. 7 mL nuclease-free water, ii. 1 mL 103 PNK Reaction Buffer, iii. 1 mL 100 mM ATP, iv. 1 mL 10 U$mL À1 T4 PNK.
Incubate at 37 C for 30 min. 3. Remove unincorporated [g-32 P]ATP with a Sephadex G-25 spin column according to the manufacturer's instructions. Save the empty tube of labeling reaction for preparation of radioactive markers (optional step below).
Optional: we highly recommend gel purifying target RNAs to remove traces of unincorporated [g-32 P]ATP and degraded RNA.
a. Resolve 28-nt target RNA and RNA for RBNS on 10% and 8% polyacrylamide urea gels, respectively. For more details on how to prepare, load, and run polyacrylamide gels for RNA analysis, we refer the experimentalist to existing protocols. 12 b. Visualize the radiolabeled target by phosphorimaging: i. Remove one glass plate and cover the gel with plastic wrap.
ii. Prepare radioactive markers: cut three small triangular shapes out of filter paper ( Figure 2).
iii. Place radioactive markers on top of the gel. iv. Add $300 mL of Loading Buffer to the empty tube of labeling reaction saved in step 3. v. Add this radioactive solution dropwise to the triangular shapes; it should be evenly distributed on the triangular pieces of the filter paper. vi. Cover the gel and the radioactive markers with plastic wrap. vii. Expose the gel to a storage phosphor imaging screen for $15À30 s and print an image at actual size. c. Identify the location of the radiolabeled RNA on the gel, using the radioactive markers on the gel to line up the gel with the marks on the printed image. d. Excise the pieces of the gel containing the RNA with a clean razor blade. e. Recover RNA from the gel by the Crush and Soak method. 13 We typically elute RNA in 300 mM NaCl, 1 mM EDTA (pH 8.2) and 1% (w/v) SDS. 4. Purify RNA by phenol/chloroform extraction followed by ethanol precipitation.
a. If the optional step above has been performed, pass eluate through a small 0.45-mm filter into a new tube. Alternatively, load the eluate on a Spin-X filter, centrifuge at 700 3 g for 10 min at room temperature (20 CÀ25 C); collect the flow-through. b. Add an equal volume of phenol:chloroform 5:1 (pH 4.5) to aqueous RNA solution.
CRITICAL: Phenol and chloroform are toxic upon inhalation. Always handle in a chemical fume hood. Avoid contact with eyes and skin. Strictly follow institutional guidelines for waste disposal. Note: Avoid transferring any of the interphase or organic layer into the pipette when removing the aqueous phase. Angle the tube at 45 C to facilitate the process. e. Add an equal volume of chloroform and mix well by vigorous vortexing. f. Spin at the maximum speed for 15 min at room temperature (20 CÀ25 C) in a microcentrifuge. g. Transfer the aqueous (top) phase to a new 2-mL tube. h. Follow instructions from steps 1dÀn to recover RNA by ethanol precipitation.
Note: If RNA was gel-purified and eluted in in 300 mM NaCl, 1 mM EDTA (pH 8.2) and 1% (w/ v) SDS in step 3e, there is no need to add 0.1 volumes 3 M sodium acetate (step 1e).
i. Resuspend the RNA pellet in 40 mL of nuclease-free water. ll OPEN ACCESS j. Measure absorbance at 260 nm in a spectrophotometer. k. Determine RNA concentration of RNA using the Beer-Lambert law: where A is absorbance at 260 nm, l is the optical path length in cm, ε is the molar extinction coefficient of RNA oligonucleotide in L$mol À1 $cm À1 , and c is RNA concentration in M.
CRITICAL: For accurate quantification, absorbance values must be between 0.1 and 0.8. Solutions of higher and lower concentrations have higher relative error in the measurement. If absorbance is < 0.1, add cold RNA to obtain an absorbance measurement comprised between 0.1 and 0.8. If too much cold RNA has been added and measured absorbance is > 0.8, take an aliquot of RNA solution and measure absorbance of a diluted sample.
l. Aliquot target RNA and store at À20 C up to 6 months (or until the specific radioactivity becomes low) by strictly following institutional guidelines for storage of radioactive material.

General guidelines for handling Bio-Dot microfiltration apparatus
Timing: 2À22.5 h Bio-Dot microfiltration apparatus is simple to operate and enables rapid separation of proteinbound RNA from unbound RNA. We describe below the handling of Bio-Dot apparatus that we use in double filter-binding assays and RBNS ( Figure 3).
Note: Additional information about the apparatus can be found in the manufacturer's manual. a. Soak the sample template of Bio-Dot apparatus in 0.5 M sodium hydroxide overnight (12-18 h) at room temperature (20 CÀ25 C) or for 1 h at 65 C to eliminate RNA and protein molecules remaining from previous assays. b. Wash thoroughly with RNase-free water. c. Wash the apparatus with a surfactant; we typically soak the apparatus in 7X-O-Matic Cleaning Solution for 15 min. d. Wash thoroughly with water and rinse with nuclease-free water. e. Air-dry the apparatus. 6. Assembly.

Cleaning.
a. Cut each of the required membranes to fit the Bio-Dot apparatus.
CRITICAL: Both nylon and nitrocellulose membranes should be the same size and should not extend beyond the edge of the gasket. Membranes of a larger size may obstruct vacuum formation after the Bio-Dot apparatus has been assembled. Trim the membranes if needed.
Note: To reduce retention of free single-stranded RNA, pre-condition nitrocellulose membranes prior to use by soaking in 0.4 M potassium hydroxide for 10 min and washing thoroughly with nuclease-free water. 5,14 Pre-condition nylon membranes by incubating in 0.1 M EDTA (pH 8.2) for 10 min, washing three times in 1 M sodium chloride for 10 min, Step-by-step procedure to assemble the Bio-Dot Microfiltration apparatus for double filter-binding assays ll OPEN ACCESS rapidly rinsing ($15 s) in 0.5 M sodium hydroxide, and washing thoroughly with nuclease-free water. 5 We highly recommend pre-conditioning both nitrocellulose and nylon membranes in RBNS assays.
b. Equilibrate the nitrocellulose and nylon membranes by soaking them in Wash Buffer for at least 1 h. Always wear gloves and use forceps when handing membranes.
Note: We assess binding of mammalian RISC by performing binding reactions at 37 C. To ensure constant temperature throughout the procedure, all incubations, assembly of Bio-Dot apparatus and filtering steps are performed in a 37 C constant-temperature room, using supplies that had been pre-equilibrated to 37 C.
c. Deposit the gasket support plate into the vacuum manifold ( Figure 4). d. Pull down the sealing gasket on top of the vacuum manifold. e. Align the 96 holes in the gasket over the 96 holes in the support plate ( Figure 4). f. Remove the nylon membrane from the Wash Buffer and shake it out gently for a few seconds to remove the excess of liquid.
CRITICAL: dehydrated membranes may cause protein denaturation upon binding and may restrain proper drainage of solutions upon filtering. Therefore, remove the membrane from the wetting solution right before (% 5 min) filtering.
g. Lay the nylon membrane on top of the sealing gasket ( Figure 4). h. Remove the nitrocellulose membrane from the Wash Buffer and shake it out gently for a few seconds to remove the excess of liquid. i. Lay the nitrocellulose membrane on top of the nylon membrane ( Figure 4).

OPEN ACCESS
Note: In RBNS experiments, the nitrocellulose membrane is cut into pieces to enable extracting protein-bound RNA from each binding reaction independently. Use a pencil to draw markers around wells where samples will be applied (Figure 4). Do not accidentally drill the membrane: small holes in the membrane may impair proper vacuum formation in downstream steps.
j. Carefully roll out any air bubbles confined between the membranes and the gasket with forceps. k. Place the sample template on top of the membrane (Figure 4). l. Finger-tighten the four screws. Use a diagonal crossing pattern to apply uniform pressure on the membrane surface.
CRITICAL: if some wells are not used, they must be closed off to ensure proper vacuum to the wells in use. We cover the unused portion of the apparatus with clear packing tape to prevent air from moving through those wells (Figures 4 and 5A).
m. Switch the vacuum source on and adjust air aspiration to gentle flow.

Note:
We are using an in-house vacuum line; alternatively, an electronic laboratory vacuum pump can be used.
n. Adjust the flow valve so that the manifold is exposed to air only ( Figure 5B; the flow valve in position 2). o. Connect the apparatus to the vacuum source. p. Place a multi-channel pipet (10À100 mL) and ice-cold Wash Buffer near the Bio-Dot apparatus ( Figure 5A). q. When ready for sample application, remove the excess of liquid from membranes by applying gentle vacuum to the manifold ( Figure 5B; the flow valve in position 3).
Note: Loose tightening of the apparatus results in leaking between wells and cross contamination with samples. If needed, adjust screw tightening using the diagonal crossing pattern while gentle vacuum is on.
CRITICAL: perform the above step rapidly so that not to dehydrate the membranes. Watch the sample wells. If membranes have dried, rehydrate the wells with some Wash Buffer ($100 mL) and remove the excess of liquid by applying gentle vacuum to the manifold (the flow valve in position 3). Step-by-step procedure to disassemble the Bio-Dot Microfiltration apparatus ll OPEN ACCESS r. Watch the sample wells. As soon as the buffer solution drains from all the wells, set up the flow valve so that the manifold is exposed to air ( Figure 5B; the flow valve in position 2). Imminently proceed to the next step. 7. Loading samples.
a. Rapidly load the samples into the wells with a multi-channel pipet.
Note: If binding reactions are performed in a small volume < 50 mL, apply the sample in the center of the well (and not on the well walls). Ensure that there are no air bubbles in the wells, as they will prevent the sample from binding to the membrane. Air bubbles may be removed by pipetting the liquid in the well up and down.
CRITICAL: RBNS is a highly sensitive assay; therefore, skip a well between samples to prevent cross-contamination.
b. Set up the flow valve so that the manifold is exposed to the vacuum ( Figure 5B; the flow valve in position 1). c. Watch the sample wells. After the samples drain from all the wells, leave the flow valve in position 1 and rapidly add 10À100 mL of ice-cold Wash Buffer into the wells. d. Watch the sample wells. When the Wash Buffer drains from all the wells and the membranes are dehydrated, the manifold is ready for disassembly. 8. Disassembly.
a. Place a sheet of filter paper near the Bio-Dot apparatus. b. While the manifold is exposed to the vacuum (the flow valve in position 1), finger-loosen the four screws using a diagonal crossing pattern. c. Lift the sample template and turn the vacuum off by setting up the flow valve to position 2. d. Grip both nitrocellulose and nylon membranes together by a corner with forceps and pull them horizontally towards to the filter paper ( Figure 6). The RNA side should face up. e. Carefully separate the membranes by lifting the nitrocellulose membrane and placing it RNA face up on the filter paper ( Figure 6).
Note: The filter paper drains excess of Wash Buffer from the membranes.
f. Carefully transfer the membranes to a new piece of filter paper ( Figure 6).
CRITICAL: Air-dry the membranes ($30 min). If wet membranes are used in the next step, the sample dots may expand by diffusion yielding halos around the samples and an eventual cross-contamination.
g. Place a plastic wrap over the membranes. h. Expose the membranes to a storage phosphor imaging screen for 10 s to 15 min and for 2À16 h in RBNS and double filter-binding assays, respectively.
Note: The signals must be distinguishable over the background, but not saturated.

Measuring the fraction of active RISC by titration
Timing: 5À6 days We describe below a titration experiment to quantify the concentration of binding-competent protein, which is a key step for determining the protein concentrations to use in double filter-binding and RBNS reactions. Note: The amount of bound ligand in double filter-binding assays and RBNS experiments is determined not by the total protein concentration but by the concentration of total active protein. Different protein preparations can have different percent activities for the same amount of protein due to misfolding, aggregation, degradation, and potentially inactivation upon purification. We estimate total RISC concentration by northern blot and measure the equilibrium binding of active RISC with a high-affinity RNA target. Concentration of RNA target is chosen to be much greater than its known K D , and RISC concentration is varied by an order of magnitude above and below the target concentration. Fitting the titration data to a quadratic equation enables measuring the amount of active RISC. 17. Incubate the binding reactions at 37 C for 1 h. 18. Separate RISC-bound RNA from unbound RNA by using Bio-Dot apparatus (steps 6À8). Upon binding (step 7a), apply 4.5 mL of each reaction to the membranes. Wash with 10 mL ice-cold Wash Buffer (step 7c). 19. Expose the membranes to a storage phosphor imaging screen for 2À16 h (step 8h). 20. Measure the fraction of active RISC (Quantification and statistical analysis: step 1).

Timing: 5À6 days
We describe below the specific steps for performing double-filter binding assays to measure the binding affinity of a target RNA of interest.
Note: To measure K D , we vary protein concentration, while keeping the concentration of target RNA constant. Importantly, the ''titration'' regime-in which the concentration of a binding site is much greater than K D -must be avoided (reviewed in ref. 16 ). K D can be determined under experimental conditions, in which the concentration of target RNA should be well below or of the same order of magnitude as its K D . The latter regime requires fitting  Note: For the binding equilibrium where RISC interacts with an RNA site i , the equilibration rate constant is described by Estimate the equilibration rate constant for the protein and the target RNA of interest. E.g., for the let-7a 6mer-A1 binding site (a target RNA with Watson-Crick pairing to miRNA nucleotides at positions 2À6 and an adenosine opposite miRNA nucleotide at position 1), k on = 2.0 G 0.1 3 10 8 M À1 $s À1 and k off = 0.24 G 0.001 s À1 (ref. 9 ). For the let-7a non-canonical 11mer-m11.21 site (a target RNA with Watson-Crick pairing to miRNA nucleotides at positions 11À21), k on = 3.6 G 0.1 3 10 7 M À1 $s À1 and k off = 0.79 G 0.08 s À1 (ref. 9 ). At target RNA concentration of 100 pM, the binding reaction with the lowest RISC concentration (1 pM) should reach equilibrium in 4 G 1 s and 1.2 G 0.2 s for 6mer-A1 and 11mer-m11.21, respectively.
28. Separate RISC-bound RNA from unbound RNA by using Bio-Dot apparatus (steps 6À8). Upon binding (step 7a), apply 4.5 mL of each reaction to the membranes. Wash with 10 mL ice-cold Wash Buffer (step 7c). 29. Expose the membranes to a storage phosphor imaging screen for 2À16 h (step 8h). 30. Measure the binding affinity (Quantification and statistical analysis: step 2).

Timing: 17À21 days
We describe below the specific steps for performing double-filter binding assays to measure the binding affinities of many binding sites simultaneously using the high-throughput sequencing method RBNS.
Note: As in double filter-binding assays, the RNA pool concentration is the same for all binding reactions, while the protein concentration varies. As discussed in ref., 1 protein and RNA pool concentrations should be carefully chosen to avoid the ''titration'' regime. In a typical RBNS experiment, the random sequence RNA region is flanked by constant primer-binding regions used for sequencing. This design simplifies library preparation, avoids biases that can result from RNA ligation, and ensures that any RNA carried over from protein purification will not contaminate the sequenced library. The length of the random sequence region of RNA ligands is an important aspect of RBNS design and has been discussed previously. 20 We randomized 20 nucleotides, obtaining an RNA pool of 4 20 = 1.1 3 10 12 distinct RNA sequences. Figure 7B illustrates the method. In addition to simultaneously determining absolute K D values of various binding sites, our computational procedure estimates the concentration of active RISC 1 ; therefore, measuring the fraction of active RISC by titration is not strictly required. Note: This amount of reaction mix is sufficient to perform 10 binding reactions. Linearly scale-up accordingly to the number of samples. CRITICAL: Even small amounts of RNA non-specifically retained on the nitrocellulose filter may dilute protein-bound RNA and reduce the sensitivity of RBNS. Therefore, tRNA and BSA are included to minimize non-specific binding. Although they do not interact with RISC or target RNA, BSA does interact with the nitrocellulose filter and can reduce its binding capacity.
b. Pipet reaction mix into six 200-mL sterile PCR tubes on ice, 12 mL per tube. c. Dilute RISC in ice-cold Storage Buffer. d. Add 8 mL diluted RISC to PCR tubes containing 12 mL reaction mix to generate a 20-mL reaction. For the no-RISC control reaction, add 8 mL Storage Buffer to 12 mL reaction mix.
Note: The final concentration of the RNA input pool is 100 nM; the concentrations of individual binding sites in the pool should be measured by high-throughput sequencing.
e. Incubate reactions for 2 h at 37 C. 36. Separate RISC-bound RNA from unbound RNA by using Bio-Dot apparatus (steps 6À8). Upon binding (step 7a), apply 19.5 mL of each reaction to the membranes. Wash with 100 mL icecold Wash Buffer (step 7c). 37. Expose the membranes to a storage phosphor imaging screen for 10 s to 15 min (step 8h). 38. Measure the fraction bound (Quantification and statistical analysis: steps 1a-b).
Note: Imaging both nitrocellulose and nylon membranes at this step enables estimating the fraction of RNA input pool bound by RISC. Fraction bound is not used in downstream computational analyses of RBNS data. Nevertheless, we highly recommend performing this step as it helps in troubleshooting potential technical issues before proceeding with the timeconsuming steps of RNA extraction, library preparation and high-throughput sequencing. Unbound RNA is not sequenced; therefore, the nylon membrane can be discarded once imaged.
a. Excise sample spots from the nitrocellulose membrane with a clean razor blade by following pencil mark drawn in step 6i. b. Fold each piece of membrane using forceps and put it in a new 1.7-mL tube. Do not touch the sample spots with the forceps. c. Prepare a proteinase K master mix by mixing the following components: Note: The table shows the volumes to perform ten reactions. Linearly scale-up accordingly to the number of samples. Because the expected amount of RNA is low, we add 1 mL 20 mg$mL À1 d. To each tube, add 400 mL the proteinase K master mix. e. Immerse pieces of nitrocellulose membrane into the solution by gently pushing with a sterile pipette tip. f. To 2 pmol of 5 0 -phosphorylated, radioactively labeled RNA input pool (i.e., the substrate in the binding reactions), add 400 mL of the proteinase K master mix.
Note: RNA input pool is sequenced along with protein-bound RNA samples and is used in the downstream computation analyses to account for any sequence biases present in the random region of the substrate RNA. The RNA input pool needs to be sequenced only once; it is not necessary to perform this step for every experiment. However, the sequence composition may differ between independently synthesized pools, so each new RNA input pool must be sequenced.
g. Incubate for 1 h at 45 C shaking at 300 rpm. h. Purify RNA by phenol/chloroform extraction as in steps 4bÀg.
Note: The chloroform step (steps 4e-g) is optional.
i. Precipitate the RNA by adding three volumes of ethanol.
Note: The RNA solution contains 150 mM NaCl from the Proteinase K Buffer, so there is no need to add 0.1 volumes 3 M sodium acetate.
j. Incubate the solution 1 h on ice or overnight (12-18 h) at À20 C. k. Recover the RNA as in steps 1gÀn.
Note: Instead of vortexing, mix gently by inverting the tubes.
l. Dissolve pellets in 9.1 mL nuclease-free water and proceed immediately to the next step.
Note: Alternatively, RNA may be stored at À20 C by strictly following institutional guidelines for storage of radioactive material.
40. Reverse transcribe RNA to produce cDNA suitable for library amplification. a. Transfer 9 mL of each sample to a 200-mL PCR tube. b. Incubate 1 min at 90 C and 5 min at 65 C to denature any secondary structure. c. Add 1 mL 50 mM RT primer. d. Anneal the RT primer to RNA by incubating 5 min at 65 C and 10 min at room temperature (20 CÀ25 C). e. Place tubes on ice. f. Perform reverse transcription in a 20 mL final volume by adding to annealed RNA and RT primer from the previous step: i. 4 mL 53 First-Strand Buffer, ii. 1 mL 100 mM DTT, iii. 3 mL dNTP mix (10 nM each), iv. 2 mL RT Superscript III (200 U$mL -1 ). g. Mix well and incubate at 55 C for 1 h then 70 C for 15 min in a thermal cycler. Place tubes on ice.
Note: Alternatively, store cDNA at -20 C until ready to perform the next step.  (12-18 h) at À20 C. f. Spin at the maximum speed for 30 min at 4 C in a microcentrifuge. g. Carefully remove nearly all supernatant. h. Add 0.5 mL ice-cold 70% ethanol to the cDNA pellet and mix gently by inverting the tubes. i. Spin at the maximum speed for 15 min at 4 C in a microcentrifuge. j. Carefully remove nearly all supernatant. k. Repeat the steps 41h-j. l. Spin at the maximum speed for 15 s at 4 C in a microcentrifuge to collect the residual ethanol on the wall of the tube and discard the liquid using an extended pipette tip. m. Air dry the pellet for 5À20 min. n. Resuspend the pellet in 10.1 mL nuclease-free water. o. Store the samples at À80 C until ready to perform the next steps. 42. PCR amplification of libraries.
Note: After reverse transcription, libraries are amplified by PCR using a universal primer and a barcoded primer to allow sample multiplexing. Primer sequences are listed in Table 1. The number of amplification cycles required to amplify the libraries should be minimized to avoid overamplification, which increases PCR artifacts. To determine the appropriate number of PCR cycles to use, the experimentalist may perform a test PCR amplification as described. 22 We generally perform 16 PCR cycles.
a. Setup PCR reactions in a final volume of 25 mL by combining: i. 10 mL water, ii. 2.5 mL 103 AccuPrime Pfx Reaction mix, iii. 1 mL 10 mM PCR Forward primer, iv. 1 mL 10 mM Multiplexing PCR Reverse Primer PCRIdx (x stands for the barcode number), v. 0.5 mL 2.5 U$mL À1 AccuPrime Pfx DNA Polymerase, vi. 10 mL cDNA (from step 41n). b. Place the PCR tubes in a thermocycler with the heated lid set to 105 C and perform PCR amplification using the following cycling parameters: 43. Gel-purify PCR products. a. Resolve PCR products on a 2% (w/v) agarose gel in 13 TAE containing 0.5 mL$mL À1 (f.c.) ethidium bromide and place the gel on a clean sheet protector.
CRITICAL: Skip a well between samples to prevent cross-contamination. b. View the gel on a UV transilluminator and excise gel fragments between $120 and $160 bp with a clean razor blade.
Note: Expected size of PCR amplicons is 138 bp (Figure 8).
44. Extract DNA using MinElute Gel Extraction kit according to the manufacturer's instructions. Elute DNA in 10 mL of nuclease-free water. 45. Run 1 mL purified DNA libraries (0.1-10 ng) on an Agilent 2100 Bioanalyzer High Sensitivity instrument using the High Sensitivity DNA Kit following the manufacturer's instructions. 46. Quantify concentration of libraries using Kapa Library Quantification Kit following the manufacturer's instructions. 47. Pool libraries to be sequenced in an equimolar ratio to yield 1 nM (f.c.).
Note: Library pooling can be performed by using the ''Illumina Pooling Calculator''.
CRITICAL: we use a NextSeq-500 instrument and the NextSeq 500/550 High Output Kit v2.5 (75 Cycles) (Single-read sequencing). This system uses two-channel sequencing chemistry. To correctly identify DNA clusters and perform accurate base calling, all four DNA bases must be represented in every cycle. RBNS libraries show low diversity: RNA oligonucleotides of the input pool used in RBNS reactions are the same length (60 nt) and contain the same 20 nt 5 0 sequence, among which five last nucleotides are sequenced. To provide the required cycle-to-cycle diversity, we add to the library pool 2À3 libraries ($160À180 bp) originating from a different application (e.g., mouse small RNA-Seq). Ideally, these libraries should represent 10À20% of the total number of sequencing reads.

Sequence pooled RBNS libraries using an Illumina Sequencing platform compatible with TruSeq
Small RNA protocol following the manufacturer's instructions. Note: we use a NextSeq-500 instrument and the NextSeq 500/550 High Output Kit v2.5 (75 Cycles) (Single-read sequencing). Sequencing 14À16 pooled libraries in one run typically yields 12-25 million sequenced reads per library or 10À20 million sequenced reads after filtering and provides enough coverage for de novo site discovery and estimation of K D values for sequence motifs % 10 nt within a 20-nt random region.
49. Measure binding affinity for binding sites of interest as described in Quantification and statistical analysis: step 3.

EXPECTED OUTCOMES
Measuring the fraction of active RISC The nitrocellulose filter preferentially retains protein and protein-bound RNA. The positively charged nylon filter beneath the nitrocellulose traps protein-free RNA not retained by the nitrocellulose. Equilibrium binding is measured in a ''titration regime'', i.e., the concentration of RNA target is much greater than its known K D . Essentially all added RISC binds to target RNA (the first linear portion of the curve in Figure 9A), until there is no more free target RNA left to bind (plateau of the curve in Figure 9A). Assuming that the stoichiometry of the bound complex is known, the breakpoint in fraction bound versus the ratio of protein to ligand indicates the amount of active protein.

Measuring RISC binding affinity by double filter-binding assays
To measure K D , the concentration of target RNA should be well below or of the same order of magnitude as its K D . The latter regime requires fitting the binding data to an appropriate binding equation

OPEN ACCESS
that explicitly accounts for bound protein. Therefore, the K D value is obtained by fitting the data to the Morrison quadratic equation. An example of assessing K D values of let-7a$AGO2 by a double filter-binding assay can be found in figure 6C of ref. 19 Measuring binding affinity by RBNS As protein concentration increases, the amount of RNA bound to the nitrocellulose filter increases ( Figure 9B). Assuming (1) the stoichiometry of the bound complex RISC:target RNA is 1:1 and (2) complete recovery of protein from the nitrocellulose filter, the amount of protein-bound RNA should not exceed the amount of RISC in each binding reaction. Nevertheless, the amount of protein-bound RNA is typically > 1 (the dashed line in Figure 9B), especially at low RISC concentrations, because RNA non-specifically retained on the nitrocellulose filter dilutes the small amount of specifically recovered RNA.
Selection of bound RNA using nitrocellulose filter is based on the fact that nucleic acids do not bind the nitrocellulose filter and pass through, while many proteins display a strong affinity for the nitrocellulose filter without losing their affinity for RNA ligands. Nevertheless, some protein-free nucleic acids, mostly G-rich sequences, bind to nitrocellulose in the absence of protein. 23 A no-protein control reaction enables detecting and correcting for this non-specifically recovered RNA. In our experimental setup, stretches of guanine and cytosine nucleotides (e.g., GGGGGGGGGG and ACCCCCCCCC) are among the most enriched motifs.
During library preparation, recovered RNA is reverse transcribed and amplified by PCR. PCR amplicons, visualized on an agarose gel, correspond to three types of molecules of different sizes (Figure 10A): (1) PCR Forward and Multiplexing PCR Reverse primers have the lowest size (< 100 bp), (2) amplicons containing one sequencing adapter are $100-bp long, and (3) amplicons containing both sequencing adapters are 138-bp long. Therefore, PCR products of the correct size (138 bp) should be gel-purified. The size distribution of the purified libraries is then analyzed using an Agilent Bioanalyzer; this step enables controlling for the absence of residual PCR primers before quantifying and sequencing. Figure 10B provides an example of Agilent Bioanalyzer data.
Library preparation starting with protein-bound RNA resulting from 20-mL binding reactions comprising $5À1,000 pM RISC and 100 nM RNA input pool yields $10À200 ng of library as estimated by KAPA library quantification.

QUANTIFICATION AND STATISTICAL ANALYSIS
1. Measuring the fraction of active RISC by titration. a. Use software (e.g., ImageJ or ImageQuant TL) to quantify signal of protein-bound (S bound ) and unbound RNA (S unbound ) from each binding reaction on nitrocellulose and nylon membranes, respectively, as well as background (b) of the membranes.
CRITICAL: Be sure to define regions of interest of the same area for all measurements.
b. For each binding reaction, compute the fraction bound f as: f = S bound À b ðS bound À bÞ+ðS unbound À bÞ c. Subtract the fraction bound of no-RISC binding reaction from each titration point to correct for non-specific RNA retention. d. Plot the fraction bound f as a function of RISC concentration to target RNA concentration (r = ð½RISCÞ=ð½target RNAÞ). e. Fit the titration data to the quadratic form of the Hill equation 24 : f ðrÞ = f max 3 r+K D +n À ffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi ðr+K D +nÞ 2 À 4 3 r 3 n q 2 3 n ; where K D is the apparent dissociation constant, n is the stoichiometric equivalence point, and f max is the maximum fraction bound. 2. Measuring binding affinity by double filter-binding assays. a. Quantify the fraction bound f as described in steps 1aÀc. b. Fit the binding data to the Morrison equation for tight binding: where ½E T is total enzyme concentration, ½S T is total target RNA concentration, and K D is the apparent dissociation constant. Note: We assign a read to a site category if it contains a single binding motif. Reads containing multiple instances of binding sites (from the same or a different site category) and reads containing partially overlapping sites are discarded. Reads that do not have any binding motifs of interest are classified as reads with no-site.
h. To estimate K D values, run the code by following instructions in the README file.

LIMITATIONS
The method presented here has broad utility in quantitatively assessing the specificity of RNA-or DNA-binding proteins for nucleic acids. Purified proteins are a prerequisite for binding assays. Contaminants carried over from protein purification may bind target RNA molecules and affect K D estimation. We assume that RISC binds one target RNA molecule with 1:1 stoichiometry and a Hill coefficient of 1 and that the recovery of bound RNA is complete. For K D estimation, we also assume that the reaction has reached equilibrium. Upon separating protein-bound RNA from unbound RNA using Bio-Dot apparatus and washing the filters with a buffer without reactants, the forward reaction rate is zero and the products start to dissociate to achieve a new equilibrium. Therefore, we wash rapidly once, and we assume that equilibrium is unperturbed. To the extent possible, we recommend confirming binding affinities by an additional, independent assay (e.g., a kinetic ll OPEN ACCESS experiment). To measure K D , the ''titration'' regime-in which the concentration of target RNA is much greater than K D -must be avoided. Therefore, when the binding affinity is completely unknown, a series of pilot experiments is required to determine the correct range of concentrations of protein and ligand. K D values can be fit to motifs of interest if at least 10-100 sequencing reads are assigned to these binding sites.

TROUBLESHOOTING Problem 1
No filtration is observed in some wells of Bio-Dot apparatus (related to steps 13bÀc).

Potential solution
If sample volume exceeds 100 mL, load 8À10 samples at a time and close the other wells with tape to increase the pressure per area. If there is a well with no filtration, close the other wells with tape and remove potential bubbles obstructing the filtration by pipetting the liquid up and down with a tip.

Problem 2
Bio-Dot apparatus does not assemble well and/or leaks between wells (related to steps 12À13).

Potential solution
Nylon and nitrocellulose membranes should not extend beyond the edge of the gasket. Membranes of a larger size may obstruct vacuum formation after the Bio-Dot apparatus has been assembled. Trim the membranes if needed. Leakage between wells may be due to incomplete tightening of the apparatus. Once the Bio-Dot apparatus has been assembled, adjust screw tightening using a diagonal crossing pattern while under gentle vacuum.

Problem 3
No signal of protein-bound RNA is detected on nitrocellulose membranes (related to step 14f).

Potential solutions
Measure concentration of binding-competent protein (steps 15À23) to ensure that the protein has not lost its activity. Perform a positive control using a high-affinity substrate. If the problem persists, a double filterbinding assay may not be suitable for the protein of interest. For example, detecting binding of fly Dicer to pre-miRNAs by double-filter assay yields poor results, whereas mobility shift experiments with the same protein preparation readily detect binding.

Problem 4
Signal of unbound RNA on the nylon filter is saturated, preventing calculation of fraction bound (related to step 14f).

Potential solution
Unbound RNA retained on the nylon filter may be greater than protein-bound RNA retained on the nitrocellulose filter, especially in RBNS experiments. If exposed to a phosphorimaging screen for the same duration, signal of protein-bound RNA is barely detectable, while signal of unbound RNA is already saturated. In this case, nylon and nitrocellulose filters should be imaged twice: with a short and a long exposure times. Prepare five 5À10-fold serial dilutions of radioactively labeled RNA target and spot them onto a separate nylon filter; these samples constitute imaging standards. Imaging nylon and nitrocellulose filters together with the imaging standards at two different time points enable quantifying both signals of protein-bound and unbound RNA and estimating fraction bound.

Problem 5
Library preparation in RBNS experiments yield small amounts of DNA, insufficient for Illumina sequencing (related to steps 45À46).

Potential solutions
Measure concentration of binding-competent protein (steps 15À23) to ensure that protein used in binding reactions has not lost its activity. RISC can bind a variety of sites with affinities ranging from $1 pM to $10 nM. If the protein of interest binds substrates with lower affinities, we recommend performing the binding reaction in a larger volume and using higher protein concentrations. We also recommend ethanol precipitating the RNA and cDNA (steps 39hÀi and 41d, respectively) 1 h on ice or overnight (12-18 h) at À20 C to increase recovery. Finally, the number of PCR cycles in step 42b may be increased. However, it is important to minimize the number of amplification cycles used to avoid overamplification, which increases PCR artifacts.

Problem 6
PCR amplified sequencing libraries contain molecules migrating slower than DNA of the excepted size (related to electrophoresis in step 43).

Potential solutions
In most cases, this phenomenon is caused by over-amplification of the libraries. In later amplification cycles, the desired library-primer hybridization may be outcompeted by library-library hybridization. The resulting annealing products are called ''PCR-bubbles'' and are partly ll OPEN ACCESS double-stranded (hybridized adapters) and partly single-stranded (noncomplementary inserts). Libraries showing bubble products can be sequenced, as bubble products will be denatured into single-stranded DNA prior to flow cell binding. Nevertheless, formation of bubble products may yield in partial loss of PCR amplicons during the size selection step. To avoid PCR-bubbles, reduce the amount of cDNA input or the number of PCR cycles in step 42b.

RESOURCE AVAILABILITY
Lead contact Further information and requests for resources and reagents should be directed to and will be fulfilled by the lead contact, Phillip D. Zamore (phillip.zamore@umassmed.edu).

Materials availability
This study did not generate new unique reagents.

Data and code availability
This study did not generate new datasets or code.