Biochemical isolation of Argonaute protein complexes by Ago-APP

Significance

Small RNA-guided gene-silencing pathways regulate fundamental cellular processes. Small RNAs such as microRNAs (miRNAs) directly bind to a member of the Argonaute (Ago) protein family. In animals, Ago proteins interact with a member of the GW protein family (referred to as TNRC6A-C). Based on an Ago-interacting TNRC6 peptide, we have developed a method allowing for the efficient isolation and characterization of Ago protein complexes from any animal organism. We refer to this method as “Ago protein Affinity Purification by Peptides.” Our approach also allows for the identification of Ago-bound small RNAs as well as mRNAs. Expression of this peptide in living cells leads to global miRNA inactivation, thus providing a powerful tool to study miRNA function on various levels.

Abstract

During microRNA (miRNA)-guided gene silencing, Argonaute (Ago) proteins interact with a member of the TNRC6/GW protein family. Here we used a short GW protein-derived peptide fused to GST and demonstrate that it binds to Ago proteins with high affinity. This allows for the simultaneous isolation of all Ago protein complexes expressed in diverse species to identify associated proteins, small RNAs, or target mRNAs. We refer to our method as “Ago protein Affinity Purification by Peptides“ (Ago-APP). Furthermore, expression of this peptide competes for endogenous TNRC6 proteins, leading to global inhibition of miRNA function in mammalian cells.

To repress gene expression, microRNAs (miRNAs) guide Argonaute (Ago) proteins to distinct target sites on mRNAs (1, 2). Ago proteins contain four different domains with distinct functions. The N domain is important for small RNA loading, the PAZ domain binds the 3′ and the MID domain the 5′ end of the small RNA. The PIWI domain is structurally similar to RNase H, and indeed some Ago proteins can function as small RNA-guided endonucleases. In humans, only Ago2 is catalytically active whereas Ago1, Ago3, and Ago4 are inactive (3, 4). Recent structural and biochemical experiments identified small structural elements that affect the endonucleolytic activity of Ago proteins (5–9). For miRNA-guided repression independently of Ago-mediated RNA cleavage, Ago proteins recruit a member of the TNRC6/GW182 protein family (also referred to as GW proteins and TNRC6A-C in human), which coordinate all downstream silencing events including binding to the poly(A)-binding proteins on the poly(A) tail of the mRNA, recruiting deadenylase complexes such as PAN2–PAN3 or the CCR4–NOT complex and translational repression (10–13). GW (glycine-tryptophan) proteins are characterized by an unstructured, tryptophan (Trp)-rich N-terminal half that serves as an Ago-binding domain (Fig. 1A). In particular, two Trp of a GW protein bind into two specific pockets on the Ago surface (10, 14, 15). We have recently identified a short TNRC6B-derived peptide that efficiently interacts with Ago proteins (Fig. 1A, T6B peptide) (15, 16). We hypothesized that such a peptide might be a powerful tool for the isolation of endogenous Ago proteins from cell or tissue lysates. We find that this peptide precipitates all four endogenous human Ago proteins efficiently, and we refer to this method as “Ago protein Affinity Purification by Peptides” (Ago-APP). Furthermore, it also allows for an accurate quantification of Ago protein levels in different primary tissues. Strikingly, Ago-APP binds Ago proteins that are involved in miRNA-guided gene silencing from any animal lysate. Furthermore, in plants, where GW proteins are not conserved, the T6B peptide mimics interactions with components of the RNA-guided DNA methylation pathway, and indeed we have efficiently isolated Ago-associated small RNAs within a length frame indicative for this pathway. Finally, transfection of the T6B peptide leads to a strong repression of endogenous miRNA pathways. Taken together, we have developed and characterized a novel highly efficient tool to study small RNA pathways in many different cell types, tissues, and species.

Fig. 1.

Precipitation of Ago complexes by Ago-APP. (A) Schematic representation of the TNRC6B domain organization and its Ago-interacting regions. The position and amino acid sequence of the peptide used for Ago-APP are shown. (B) Ago-APP from HEK293 cells overexpressing FH-Ago1–4 (lanes 1–6) and FH-tagged human PIWI proteins (lanes 7–9). (C and D) Comparison of Ago2-immunoprecipitation (IP) and Ago-APP in terms of Ago enrichment (C) and coprecipitation of miRNAs (D). A monoclonal antibody against human Ago2 and an unrelated antibody were used for IPs. Recombinant GST-T6B and GST alone were used for Ago-APPs. Asterisks indicate unspecific bands. (E) Schematic representation of an Ago protein and miRNA codepletion experiment using Ago-APP. (F) Codepletion of Ago proteins and associated miRNAs. Supernatants of Ago-APPs were repeatedly incubated with T6B-coupled beads as shown in E. Ago1 and Ago2 protein levels were analyzed by Western blotting. The levels of the highly abundant let-7a and the weakly expressed miRNA-30a were analyzed by Northern blotting. A probe against U6 snRNA was used to control for equal RNA loading. (G) Quantification of Western and Northern blot signals shown in F. Background signals of the blots were deduced to obtain the protein and miRNA signal intensities of input and supernatant samples.

Results and Discussion

Ago-APP Efficiently Precipitates All Four Human Ago Protein Complexes from Cell Lysates.

To test our hypothesis that Ago-APP might be a powerful tool to pull down Ago proteins, we expressed FLAG/HA (FH)-tagged Ago proteins in HEK293 cells and performed glutathione S-transferase (GST)-T6B peptide-mediated pull-down assays (Fig. 1B). Ago-APP isolated all four FH-Ago proteins efficiently but not FH-HIWI, -HIWI2, or -HILI, three Argonaute proteins of the PIWI clade that do not interact with GW proteins (17) (Fig. 1B). Next we analyzed endogenous Ago2 pull-down efficiency and compared it to conventional anti-Ago2 immunoprecipitation (Fig. 1C). Both an anti-Ago2 antibody and the T6B peptide efficiently precipitated endogenous Ago2, whereas a control antibody or GST alone did not (Upper). As expected, the T6B peptide pulled down endogenous Ago1 as well (Lower). Furthermore, we analyzed associated miRNAs by Northern blotting (Fig. 1D). As exemplified by a Northern blot against let-7a, the T6B peptide precipitated miRNA-loaded and presumably functional Ago complexes. It has recently been reported that Ago-free miRNA pools may exist in cells (18, 19). To analyze these findings in more detail, we set out to deplete all Ago proteins from HeLa cell lysate and analyzed codepletion of let-7a and miR-30a (Fig. 1 E and F). After repeated incubation steps, we observed a clear codepletion of let-7a and miR-30a together with Ago1 and Ago2 (Fig. 1G for signal quantification), suggesting that, at least in our cell lysates, mature miRNAs are nearly quantitatively bound by Ago proteins and that Ago-free mature miRNA reservoirs may only be present in trace amounts corresponding to miRNA biogenesis intermediates. However, it should be noted that cell lysates might not fully resemble the conditions within living cells. Furthermore, our biochemical experiments are semiquantitative.

Specific Isolation and Characterization of Ago Protein Complexes Using Ago-APP.

For further specificity examination, human, mouse, Drosophila, and Arabidopsis cell/tissue lysates were subjected to Ago-APP and analyzed by mass spectrometry (Fig. 2A). All four human and mouse Ago proteins were precipitated from the lysates. In Drosophila, we found only Ago1 in our pull-downs, which is the major Ago protein required for miRNA function in this species. In Arabidopsis, GW proteins function only in RNA-directed DNA methylation (RdDM), where an RNA polymerase V subunit binds to Ago4 by Trp interactions resembling those of animal GW and Ago proteins (20). Strikingly, Ago-APP mimics this interaction and Ago4, -6, and -9 are efficiently pulled down (Fig. 2A, lane 5; Fig. 2F). To further validate that Ago-APP enriches Ago proteins equally well, we compared Ago protein levels in total lysates with Ago proteins isolated by Ago-APP (Fig. 2B). For Ago quantification, we used state-of-the-art label-free mass spectrometry [selected reaction monitoring (SRM)] (Materials and Methods). Indeed, the levels of the individual Ago proteins are similar in total lysates and in Ago-APP–enriched samples, suggesting that Ago-APP is not biased toward a specific Ago protein. Of note, Ago4 is in rather low abundance and can therefore be quantified only after Ago-APP enrichment (Fig. 2B).

Fig. 2.

Isolation and characterization of Ago complexes from different species. (A) Ago-APP and mass-spectrometric analysis of different protein extracts. Samples were eluted with PreScission protease (lane 1, Pr) or Laemmli buffer (lanes 2–5), which simultaneously eluted GST-T6B and aggregates of it (asterisks). (B) Total lysates from the indicated cell lines were separated by SDS-PAGE with (Lower) and without (Upper) prior enrichment by Ago-APP. For total lysates without affinity purification, the area around 100 kDa was cut from the gel as indicated. The relative amounts of Ago1–4 were determined by SRM with stable isotope-labeled peptides. The proportion of one Ago protein related to the total Ago pool is shown. (C) Ago-APP was used to purify endogenous Ago1–4 from different murine tissues. The samples were stained with Coomassie (Upper) and analyzed by Western blotting (Lower). BAT, brown adipose tissue. (D) Mass-spectrometric analysis of the Ago-APPs conducted in C. The relative amounts of Ago1–4 were determined as in B. The proportion of one Ago paralogue related to the total Ago pool is shown. Error bars represent the SD of identical samples that were quantified with two different paralogue-specific peptides. (E) Relative amount of Ago4 in different mouse tissues, derived from the same samples/measurements described in C and D. (F) Ago-APP from an Arabidopsis callus with egg cell-like character and mass-spectrometric identification of the purified Ago proteins. A GST-tagged peptide lacking all five tryptophans was included as negative control. (G) Length comparison of small RNA sequencing reads of Arabidopsis input and Ago-APP samples.

To further validate the broad applicability of Ago-APP outside of cell lines, we also performed pull-down experiments from lysates prepared from different mouse tissues. Ago proteins were efficiently isolated as indicated by Western blotting against mouse Ago2 (Fig. 2C). Considering the opportunity to simultaneously isolate all four mammalian Ago proteins, we thought to use our approach to determine the relative amounts of Ago proteins within various mouse tissues (Fig. 2 D and E). Relative Ago protein levels have only been estimated before (21, 22). Such earlier quantifications relied on antibodies with different affinities and thus might not be accurate. Furthermore, Ago protein levels have not been quantified from primary tissues or organs. We find that Ago2 is the most prominent Ago protein in all tissues analyzed, followed by Ago1 and Ago3. Ago4 is found mainly in testes, which is consistent with the testes phenotype of Ago4 knockout mice (23) (Fig. 2E).

The striking observation that Ago-APP selectively isolates RdDM-pathway active Ago proteins prompted us to analyze isolated plant Ago complexes in more detail (Fig. 2F). Lysates from egg cell-like callus material (24) were subjected to Ago-APP, and precipitated proteins were analyzed. Similar to our previous experiment (Fig. 2A), the RdDM Ago proteins Ago4 and -9 were readily identified. In addition, the third RdDM Ago protein in Arabidopsis thaliana, Ago6, was also detected (Fig. 2F). siRNAs involved in RdDM are 24 nt long (25). Using small RNA cloning and deep sequencing from our Ago-APP samples, we find that 24-nt-long siRNAs are selectively enriched by Ago-APP, demonstrating that Ago-APP is a powerful tool to study RdDM mechanisms in various plant species or tissues (Fig. 2G). Furthermore, mass spectrometry analysis could be used to identify novel components involved in RdDM in various plant tissues or species.

Ago-APP Is Compatible with PAR–CLIP and Can Be Used to Identify Ago-Associated miRNA Targets.

In animals, Ago complex isolation is widely used for the identification of associated miRNA target RNAs (26–30). Thus, we asked whether Ago-APP is applicable for miRNA target identification. A common method is based on UV cross-linking of Ago proteins to target mRNAs and the subsequent cloning and sequencing of target RNA fragments (HITS–CLIP and PAR–CLIP) (26, 27). However, the protocol is highly dependent on effective immunoprecipitation of the cross-linked Ago protein complexes. To provide a tool that can be conveniently used in such CLIP protocols, we fused the GST-T6B peptide to FLAG, allowing for anti-FLAG affinity purifications as commonly used in CLIP protocols (Fig. 3A). Both anti-FLAG coupled agarose or control glutathione beads isolated endogenous Ago2 efficiently from HeLa cell lysate (Fig. 3A, lanes 8 and 10). We next performed PAR–CLIP experiments using HEK293 cells (Fig. 3B). UV₃₆₅–cross-linked Ago proteins were isolated, and the associated RNA was radiolabeled and separated by SDS-PAGE. Whereas Ago-APP recovered endogenous Ago proteins with radiolabeled RNA, the control peptide did not, indicating that our approach is highly useful for miRNA target identification using PAR–CLIP.

Fig. 3.

Ago-APP can be used for simultaneous PAR–CLIP of human Ago1–4. (A) Ago-APPs using differently tagged T6B versions (lanes 3, 5, 6, 8, 10, 11) and a mutated control peptide (lanes 2, 4, 7, 9). (B) PAR–CLIP using Ago-APP (lane 2) as well as a mutated control peptide (lane 1) from HEK293 lysates.

T6B Peptide Efficiently Inhibits miRNA Function in Vivo.

The T6B peptide occupies the TNRC6 protein-binding pocket on Ago and thus might be a valuable tool to block global miRNA function in vivo (Fig. 4). For application of the T6B peptide in cells and to protect it from immediate degradation, we fused it to enhanced YFP (eYFP). A peptide with mutated Trp residues was used as control. The expression plasmids were transfected together with a luciferase reporter carrying the 3′ UTR of HMGA2, and luciferase activity was measured 2 d after transfection. Whereas the mutated peptide had no effect on the expression of the reporter, the T6B peptide strongly increased luciferase activity, indicating that miRNA-guided gene silencing was effectively inhibited (Fig. 4A). Finally, we compared the peptide inhibition with common miRNA antisense inhibitors directed against let-7a or let-7e and to knockdowns of the individual TNRC6 proteins using specific siRNA pools (siPools) (31) (Fig. 4B). Both inhibitors against the let-7 miRNAs released repression by two- to threefold. Depletion of individual TNRC6 proteins had even weaker effects, suggesting redundancy as observed before (31). Strikingly, our peptide fused to eYFP led to a seven- and eightfold increase in luciferase activity, suggesting highly efficient inhibition of the miRNA pathway. The fact that the individual miRNA inhibitors show weaker effects suggests that the HMGA2 3′ UTR is under the control of not only the let-7 family but also various other miRNAs (Fig. 4C).

Fig. 4.

T6B peptide expression inhibits endogenous miRNA-guided gene silencing. (A) Luciferase assays conducted on the HMGA2 3′-UTR to monitor T6B-mediated de-repression. Reporter de-repression was monitored in HeLa cells overexpressing the FH- and GFP-tagged T6B peptide (wild type and mutant). (B) Comparison of T6B-mediated de-repression with TNRC6A-C knockdowns and miRNA antisense inhibitors. (C) Predicted miRNA binding sites in the HMGA2 3′-UTR.

In summary, we have developed a novel Ago family protein purification strategy with several unique features and advantages compared with common methods. First, Ago-APP is highly efficient due to high affinity of the peptide to Ago. Second, not only one, but all Ago family proteins involved in mammalian and insect miRNA function can be isolated and even depleted from lysates, which is not possible with available antibodies. This fact is important for the identification of miRNA targets because Ago2 immunoprecipitation is mainly used, and other Ago-protein–specific targets, if they exist, may be recovered by our new approach. Third, Ago proteins and thus miRNAs and associated targets can be isolated and identified from any given animal species. Fourth, our peptide can be used as an inhibitor of global miRNA regulation. Furthermore, it can be induced from stably integrated lentiviral constructs, for example, to shut down the entire miRNA regulation in cell lines that are difficult to transfect or even whole animals at defined time points.

Materials and Methods

Ago-APP.

Per 50 μL of glutathione–sepharose 4 Fast Flow (GE Healthcare) or anti–FLAG-M2 agarose (Sigma-Aldrich), a minimum of 100 μg (FLAG-)GST-tagged peptide (TNRC6B 599–683) was coupled to the beads for 3 h at 4 °C. Excess peptide was removed by washing with PBS three times, and lysate was added to the peptide-coupled beads. After incubation for 3 h at 4 °C, the beads were washed four times with NET buffer (50 mM Tris, pH 7.5, 150 mM NaCl, 5 mM EDTA, 0.5% Nonidet P-40, 10% glycerol, 1 mM NaF; supplemented with 0.5 mM DTT and 1 mM AEBSF before use) and once with PreScission cleavage buffer (50 mM Tris, pH 7.5, 150 mM NaCl, 1 mM EDTA, 1 mM DTT) or PBS. The beads were eluted by PreScission cleavage (∼4 μg PreScission in one bead volume of PreScission cleavage buffer overnight at 4 °C) or by addition of 2× Laemmli sample buffer and incubation at 95 °C for 5 min.

Detailed methods and materials are available in SI Materials and Methods.

SI Materials and Methods

T6B and Argonaute Expression Constructs.

6×His-GST-TNRC6B(599-683) (termed GST-T6B) and 6×His-GST-TNRC6B(599-683) W623A W634A W653A W666A W680A (termed GST-T6B mutant) were expressed from pEC-K-3C constructs (15).

For generation of FLAG-tagged GST-T6B, the TNRC6B fragments (wild type and mutant) were first cloned into pGEX-6p-1 via BamHI/NotI. Therefore, the fragments were amplified from the pEC-K-3C constructs by 5′-CGTAGGATCCGATTGTCAGGCTGTCTTGCAGAC-3′ (forward) and 5′-TACGGCGGCCGCTCAGAGCTCCCCCCATCCAGAC-3′ (wild type, reverse) or 5′-CGTAGCGGCCGCTCAGAGCTCCCCGGCTCCAGACTTC-3′ (mutant, reverse). These constructs were used as template for the amplification of two partially overlapping DNA fragments. Simultaneously, the FLAG sequence and a short linker region containing a BglII restriction site were inserted 5′ of the GST tag. The first part was amplified by 5′-CGTAGATATCTCGGTAGTGGGATACGACG-3′ (forward) and 5′-GGGACATAGATCTGCTCTTGTCATCGTCGTCCTTGTAGTCCATGAATACTGTTTCCTG-TGTGAAATTGTTATCCG-3′ (reverse), the second part by 5′-GATGACAAGAGCAGATCTATGTCCCCTATACTAGGTTATTGGAAAATTAAG-3′ (forward), and the wild-type and mutant T6B reverse primers were listed before. The overlapping parts were annealed, filled up, and inserted into the vector backbone via EcoRV/NotI.

For FLAG/HA-T6B-eYFP, wild-type and mutant T6B versions were amplified by 5′- TAATGCGGCCGCGATTGTCAGGCTGTCTTGCAGAC-3′ and 5′-CGTAGGATCCGGTGGCGACGAGCTCCCCCCATCCAGACTTC-3′ (wild type) or 5′-CGTA GGATCCGGTGGCGACGAGCTCCCCCGCTCCAGACTTC-3′ (mutant) and inserted into a FLAG/HA-eYFP construct based on pIRESneo (Clontech) (3) via NotI/BamHI. Three additional linker amino acids were included before eYFP (the BamHI restriction site after the eYFP-coding region was removed from VP5-eYFP by site-directed mutagenesis).

FLAG/HA-tagged Ago1–4 and FLAG/HA-tagged human Piwi proteins have been described before (3). Dual Luciferase HMGA2 3′-UTR reporter pMIR-HMGA2 and pMIR-HMGA2mut were described previously (32). pMIR-HMGA2mut contains HMGA2 3′-UTR with point mutations in the six let-7–binding sites.

Expression and Purification of Recombinant Protein.

Wild-type and mutant GST-T6B constructs were expressed from pEC-K-3C in BL21 (DE3) Gold pRARE and purified as described before (15). FLAG-GST–tagged T6B constructs were expressed from pGEX-6p1 in Bl21. Therefore, 2–3 L of an isopropyl β-d-1-thiogalactopyranoside–induced culture (OD₆₀₀ = 0.6) were grown overnight at 18 °C. The cells were harvested (4,400 × g; 15 min; 4 °C) and resuspended in buffer GST-A [PBS, 1 mM 4-(2-aminoethyl) benzenesulfonyl fluoride hydrochloride (AEBSF), 1 mM DTT] supplemented with 1 mg/mL lysozyme. After sonication for 3 × 3 min, the lysate was cleared by centrifugation (40,000 × g; 40 min, 4 °C), loaded onto a 15-mL glutathione Sepharose column (GE Healthcare), and washed with 10 column volumes (cv) of GST-A. GST-tagged protein was eluted with 4 cv of buffer GST-B (PBS supplemented with 20 mM Tris, pH 8.0, and 10 mM glutathione). After pooling and concentration, peptide-containing fractions were loaded onto a desalting chromatography column (HiPrep 26/10 Desalting, GE Healthcare) equilibrated in PBS with 5% (vol/vol) glycerol. Desalted fractions were adjusted to ∼2 mg/mL and stored at −80 °C.

Cell Culture and Transfections.

HEK 293T and HeLa cells were cultured in DMEM (Sigma-Aldrich) supplemented with 10% (vol/vol) FBS (Sigma-Aldrich) and penicillin–streptomycin (Sigma-Aldrich) under standard conditions (37 °C, 5% CO₂).

For overexpression of FLAG/HA-tagged Argonaute proteins, one plate (15-cm diameter) per construct was calcium phosphate-transfected with 10 µg DNA.

For dual luciferase assays, HeLa cells were transfected with plasmids, siPools, and 2′-O-methylated oligonucleotides using Lipofectamine 2000 (Life Technologies) as described in the manufacturer’s forward transfection protocol.

Lysate Preparation.

HEK 293T lysates and HeLa lysates were prepared by resuspending a cell pellet of one 15-cm cell culture dish in 0.5–1.5 ml NET buffer (50 mM Tris, pH 7.5, 150 mM NaCl, 5 mM EDTA, 0.5% Nonidet P-40, 10% (vol/vol) glycerol, 1 mM NaF; supplemented with 0.5 mM DTT and 1 mM AEBSF before use). Lysates were incubated on ice for 20 min and cleared by centrifugation (15,000 × g; 20 min, 4 °C).

Mouse tissue lysates were prepared in 800 µL NET buffer using FastPrep-24 with lysing matrix D (45 s at 6.5 M/s; MP Biomedicals). After tissue disruption, ceramic beads were spun down for 1 min at 13,000 × g, and the supernatant was transferred to new reaction tubes and spun down for 30 min at 15,000 × g.

Arabidopsis cell extract was obtained from suspension-cultured PSB-D cells and an RKD2-transformed callus with an egg cell-like transcription profile (24). Protein was extracted in a buffer containing 25 mM Hepes, pH 7.4, 100 mM NaCl, 0.05% Nonidet P-40, 1 mM DTT, 2 mM MgCl₂, 5 mM EGTA, 10% (vol/vol) glycerol, and protease inhibitors (33). For protein extraction from the RKD2-transformed callus, the buffer additionally contained 50 µM MG-132 and, instead of Nonidet P-40, 0.05% IGEPAL CA-630.

Drosophila embryonic extracts were prepared in a buffer containing 10 mM Hepes pH 7.4, 5 mM DTT, and EDTA-free Complete protease inhibitor mixture (Roche) as described in ref. 34.

IPs.

For IPs, an Ago2-specific antibody (35) and a control antibody against RmC, a complement system protein from rat, were coupled to Protein G Sepharose (GE Healthcare). Typically, 50 µL antibody-coupled Protein G Sepharose were used with 0.5–5 mg total protein and incubated for 3 h. The affinity matrix was washed four times with NET buffer and once with PBS. The immunoprecipitates were eluted by addition of Laemmli sample buffer and by incubating the samples at 95 °C for 5 min.

Western Blotting.

Western blot samples were mixed with 2× Laemmli Buffer (eluates) or 5× Laemmli buffer (inputs, supernatants) and incubated at 95 °C for 5 min. Proteins were separated by SDS-PAGE with 10% polyacrylamide gels and semidry-blotted. For immunodetection, the following antibodies were used: mouse–anti-HA (16B12, Covance; 1:1,000), rabbit–anti-FLAG (Sigma, 1:1,000), mouse–anti-α-tubulin (DM 1A, Sigma, 1:10,000), rat–anti-Ago1 1C9 (1:5) (28), rat–anti-Ago2 11A9 (1:5) (35), and rat–anti-Mm_Ago2 6F4 (1:5) (36). Secondary antibodies (IRDye 800CW) against mouse, rat, and rabbit were obtained from LI-COR Inc. The Odyssey Application software (version 3.0.30, LI-COR Inc.) was used to analyze band intensities, which were quantified without background intensities.

Coprecipitation of RNA.

After an Ago2-IP or Ago-APP, 25% of the beads were removed for Western analysis. The remainder was subjected to Proteinase K treatment (200 mM Tris, pH 7.5, 300 mM NaCl, 25 mM EDTA, 2% (wt/vol) SDS, 0.16 mg/mL Proteinase K). The RNA was extracted using phenol/chloroform/isoamylalcohol (25:24:1, Roth) and precipitated overnight with 20 µg of glycogen RNA grade (Thermo Scientific). The resulting pellet was washed once with 70% (vol/vol) ethanol and either solved in RNA sample buffer (for subsequent Northern blot analysis) or resuspended in water (for small RNA cloning).

Northern Blotting.

Northern Blots were carried out as described earlier (37). Briefly, RNA coprecipitating with Ago2-IPs and Ago-APPs was separated on 12% urea gels (UreaGel System, National Diagnostics), semidry-blotted, and crosslinked with 1-ethyl-3-(3-dimethylaminopropyl)-carbodiimide (EDC). Northern blots were hybridized with radiolabeled probes overnight at 50 °C. ³²P-labeled 5′-AACTATACAACCTACTACCTCA-3′ was used for let-7a detection, 5′-CTTCCAGTCGAGGATGTTTACA-3′ for miR-30a detection, and 5′-GAATTTGCGTGTCATCCTTGCGCAGGGGCCATGCTAA-3′ for the detection of U6 snRNA. The blots were washed twice with 5× SSC, 1% SDS, and once with 1× SSC, 1% SDS. Signals were detected by exposure to a screen and scanning with the phospho-imaging system PMI (Bio-Rad). Signal intensities were quantified using Quantity One Software (version 4.6.9, Bio-Rad).

Small RNA Cloning.

To keep compatibility to ILLUMINA miRNA sequencing procedure, all primers for small RNA cloning were designed according to customer sequence information provided by Illumina Inc.

One microgram of isolated total RNA (input samples) or the complete RNA that was coprecipitated by Ago-APP was used for small RNA cloning. Truncated T4 RNA Ligase 2 was used for the first ligation to an adenylated 3′-adapter. In a second ligation step, T4 RNA Ligase 1 catalyzed the addition of an RNA adapter to the 5′-end of the RNAs. The resulting product was reverse-transcribed using the SuperScript III First Strand Synthesis Super Mix (Invitrogen), followed by a PCR amplification, wherein index sequences and other Illumina-specific sequences were added. The samples were run on 6% urea–PAGE, and the bands corresponding to miRNA-containing PCR amplification products were cut out and eluted overnight in 300 mM NaCl, 2 mM EDTA. The supernatants containing the libraries were collected using Spin-X filter tubes (COSTAR), precipitated with ethanol overnight at −20 °C, pelleted, and dissolved in water. After library pooling, the deep sequencing run was performed on an Illumina HiSeq1000 sequencer (50 cycles, single run).

Cluster generation, sequencing, and FastQ file generation were carried out at the local sequencing core facility Center of Excellence for Fluorescent Bioanalytics at Regensburg University. Cluster generation was done with TruSeq SR Cluster Kit v3 on a cBOT (Illumina). Sequencing was done on a HiScan SQ (Illumina) using TruSeq SBS v3 in 50-bp single-end runs with 10 samples per lane. FastQ files were generated by CASAVA 1.8.

The small RNA sequencing data have been deposited in National Center for Biotechnology Center’s (NCBI’s) Gene Expression Omnibus (38) and are accessible through GEO series accession no. GSE70553.

PAR–CLIP.

PAR–CLIP was conducted as described in ref. 26. Per sample, 18 plates of HEK 293 cells were used with 70 and 100 µL magnetic beads, respectively. Beads were washed with PBS twice and coupled to anti-FLAG antibody (Sigma anti-FLAG M2). Beads were washed with PBS to remove excess antibody, and the according amounts of FLAG-GST-tagged T6B peptide (wild type or mutant) were added and coupled for 1 h. Excess peptide was removed by washing with PBS.

Dual Luciferase Assays.

HeLa cells were cotransfected with 125 ng pMIR-HMGA2 reporter and 400 ng pIRES-FLAG/HA-T6B-eYFP in 48-well format. As controls, pMIR-HMGA2 mutant, pIRES-FlAG/HA-eYFP, or pIRES-FLAG/HA-T6B mutant were transfected accordingly. siPools and 2′-O-methylated oligonucleotides were cotransfected with reporters to give a final concentration of 30 and 40 nM, respectively. siPools against TNRC6A, TNRC6B, and TNRC6C and a scrambled siPool (siPool Neg) were obtained from siTool Biotech (31). 2′-O-methylated antisense oligonucleotides for inhibition of let-7a and let-7e and a control inhibitor against miR-122 were generated as described previously (39). Two days after transfection, cells were lyzed in passive lysis buffer (Promega). Luciferase substrates were purchased from PJK Cryosystems. Luciferase activities were read out on a Mithras LB940 luminometer (Berthold Technologies). Firefly/Renilla luminescence ratios were calculated for the individual samples and normalized to the ratio in the pIRES-FLAG/HA-eYFP control sample with the corresponding reporter.

Mass Spectrometry.

Protein bands were cut out from the gel and subjected to in-gel tryptic digest using 2 µg trypsin (Promega) per 100 µL gel volume in 50 mM NH₄HCO₃ overnight at 37 °C. Peptides were eluted by two extractions with 100 mM NH₄HCO₃ and an additional extraction with 50 mM NH₄HCO₃ in 50% (vol/vol) acetonitrile. Before liquid chromatography–mass spectrometry analysis combined eluates were lyophilized and reconstituted in 20 µL 1% formic acid.

For protein identification, peptides were separated on an UltiMate 3000 RSLCnano System (Thermo Scientific) by reversed-phase chromatography using an Acclaim Pepmap 100 C18 nano column (75 µm i.d. × 150 mm, Thermo Fisher) in a linear gradient of 4–40% (vol/vol) acetonitrile in 0.1% formic acid for 60 min at 300 nL/min. The LC system was coupled to a maXis plus UHR-QTOF System (Bruker Daltonics) via a CaptiveSpray nanoBooster Source (Bruker Daltonics). Up to the 20 most abundant precursor ions were selected for fragmentation by collisional dissociation. Data were launched to MASCOT using the ProteinScape software (Bruker Daltonics). Mascot (v2.3.02) was used to search the NCBI nr protein database.

For precise relative quantification within one sample, an SRM-based method was used in combination with synthetic heavy peptides as internal standard. Peptide selection and SRM-assay development were performed using the open source software Skyline and manual evaluation. An MS/MS spectral library was created from information-dependent analysis data of discovery runs acquired on the hybrid triple quadrupole/linear ion trap mass spectrometer QTRAP4500 (SCIEX). Proteotypic peptides with good fragmentation intensities were selected to establish specific SRM transitions. The SRM method included precursors with charge states 2 and 3 with four transitions each. The corresponding transitions of the heavy-labeled peptides were calculated by Skyline. Heavy-labeled synthetic peptides (SpikeTides_TQL, JPT Innovative Peptide Solutions) were spiked into the digests and incubated overnight at 37 °C. For each paralogue, two peptides were chosen: Ago1(a) NIYTVTALPIGNER, Ago1(b) VLPAPILQYGGR, Ago2(a) VLQPPSILYGGR, Ago2(b) DYQPGITFIVVQK, Ago3(a) SFFSAPEGYDHPLGGGR, Ago3(b) SANYETDPFVQEFQFK, Ago4(a) EFGIVVHNEMTELTGR, Ago4(b) QVAWPELIAIR. SRM analyses were performed on the QTRAP4500 operating with the Analyst software (v. 1.6.1). In SRM mode, Q1 and Q3 were set to unit resolution (0.7 amu fwhm). The mass spectrometer was online-coupled with an UltiMate 3000 RSLCnano System (Thermo Fisher) via a NanoSprayIII Ion source (SCIEX). Peptides were trapped on an Acclaim PepMap100 C18 Nano Trap column (300 µm i.d. × 5 mm, Thermo Fisher) in 4% (vol/vol) acetonitrile/0.1% formic acid. Separation of peptides was carried out by reversed-phase chromatography on an analytical Acclaim PepMap C18 nano column (75 µm i.d. × 150 mm, Thermo Fisher) by a linear gradient of 4–40% acetonitrile in 0.1% formic acid in 45 min at a flow rate of 300 nL/min. Raw data (.wiff files) from the SRM runs were loaded into Skyline (40). After analysis of the SRM traces in Skyline, heavy-to-light-ratios of peak areas were exported to Microsoft Excel for further calculations.