The C-terminal half of human Ago2 binds to multiple GW-rich regions of GW182 and requires GW182 to mediate silencing

Abstract

MicroRNA (miRNA)-mediated silencing is a post-transcriptional mechanism that regulates translation of mRNAs primarily via their 3'-UTR. Ago2 binds miRNA directly and is the core component of miRNA-induced silencing complex. GW182 is another important factor in miRNA-mediated silencing, and its interaction with Ago2 is evolutionarily conserved. However, the GW182-Ago2 interaction in humans has not been characterized thoroughly, and the role of GW182 in the mammalian miRNA pathway remains unclear. In the current study, we generated a set of GST-, green fluorescence protein (GFP)-, or 3xFlag-tagged deletion constructs of GW182 and Ago2 to further analyze GW182-Ago2 interactions. The C-terminal half of Ago2 interacted with four nonoverlapping GW-rich regions of GW182, and this interaction recruited Ago2 to GWB. Furthermore, the interaction with GW182 was observed in all four human Ago proteins. Most interestingly, tethering the C-terminal half of Ago2 to the 3'-UTR of reporter mRNA recapitulated translational repression comparable to that of tethered Ago2, and this repression was greatly impaired upon GW182 knockdown. In comparison, the N-terminal half of Ago2 did not bind GW182 and did not retain the repression function of Ago2. Our data strongly support a model in which Ago2 recruits GW182 to the 3'-UTR of mRNA to mediate silencing, and suggest that GW182 may contribute to enhancement in translational repression by interacting with multiple Ago proteins from multiple miRNA target sites in the same or adjacent 3'UTR.

FIGURE 1.

Identifying the interaction of C-terminal half of Ago2 with GW182 fragments using GST pull-down assays. (A) Schematic of human GW182 and Ago2 deletion constructs used in this study. Amino acid residues of GW182 constructs are referenced to the TNGW1, the longer isoform of GW182 (GenBank Accession NM_014494.2). Q-repeat, glutamine repeat (box in white); Q/N-rich, glutamine/asparagine-rich region (box in magenta); RRM, RNA recognition motif (box in green); GW-rich, glycine/tryptophan-rich region (boxes in yellow); N-GW, N-terminal GW-rich region; M-GW, middle GW-rich region; C-GW, C-terminal GW-rich region. Human Ago2 contains two conserved domains: PAZ domain (box in blue) and PIWI domain (box in red). (B) Endogenous Ago2 coprecipitated with GW182 fragments GW1Δ1 (“aa254-751”) and GW1Δ10 (“aa566-1343”) but not with TNR (“aa1-204”). GST-tagged TNR (lanes 1,4), GW1Δ1 (lanes 2,5), or GW1Δ10 (lanes 3,6) was transfected into HeLa cells. Endogenous Ago2 was detected by using rabbit polyclonal anti-Ago2, which also contained anti-GST activity. In a longer exposure (lanes 7–9), Ago2 was detected more clearly being coprecipitated with GST-GW1Δ10 (lane 9) but still absent in GST-TNR precipitates (lane 7). (C) GW182 fragments GW1Δ1 and GW1Δ10 coprecipitated with Ago2 fragment PIWI (“aa478-860”) but not with PAZ (“aa1-480”). GST-GW1Δ10 (lanes 1–6) or GST-GW1Δ1 (lanes 7–10) was cotransfected with Flag-tagged Ago2, PIWI, or PAZ. Full-length Ago2 (lanes 4,9) and PIWI (lanes 5,10), but not PAZ (lane 6), coprecipitated with GST-GW1Δ10 or GST-GW1Δ1. (D) GW182 fragment TNR did not pull down Ago2. Flag-Ago2 was cotransfected with GST-tagged TNR (“aa1-204,” lanes 1,3) or with GW1Δ1 (lanes 2,4) as a positive control.

FIGURE 2.

GW182 fragment GW1Δ10 (“aa566-1343”) recruited Ago2 to cytoplasmic foci by interacting with the C-terminal half of Ago2. GFP-GW1Δ10 (green, a–c) was cotransfected with Flag-Ago2 (d), PIWI (“aa478-860,” f) or PAZ (“aa1-480,” h) into HeLa cells. As controls, Flag-PIWI (e) or Flag-PAZ (g) was singly transfected. The cells were stained with anti-Flag antibody (red, d–h). Panels in the bottom row are the merged images (i–m). Nuclei were counterstained with 4′,6-diamidino-2-phenylindole (DAPI) (blue). Scale bar, 10 μM.

FIGURE 3.

Ago2 bound to multiple nonoverlapping GW-rich regions of GW182. (A) GW182 fragments coprecipitated with C-terminal half of Ago2. GST-PIWI (“aa478-860”) was cotransfected with GFP-tagged GW1Δ1 (lane 1), GW1Δ1a (lane 2), GW1Δ1b (lane 3), GW1Δ7 (lane 4), TNR (lane 11), or GW1Δ5 (lane 12) into HeLa cells. Similar to positive control GW1Δ1 (lane 6), GW1Δ1a (lane 7), GW1Δ1b (lane 8), GW1Δ7 (lane 9), GFP-GW1Δ5 (lane 14), but no GFP-TNR (lane 13), were detected in GST-PIWI precipitates. GST-tagged N1, N-terminal fragment from an unrelated protein hZW10, was cotransfected with GFP-GW1Δ1 (lane 5) as a negative control and no interaction was detected (lane 10). Asterisks indicate the corresponding GFP-tagged constructs in Western blot. (B) GW182 fragment GW1Δ12 coprecipitated with Ago2. Flag-Ago2 was cotransfected with GST-tagged TNR (lanes 1,3) or with GW1Δ12 (lanes 2,4), which contains the conserved sequence for Ago2 interaction. Flag-Ago2 was coprecipitated with GST-tagged GW1Δ12 but not TNR.

FIGURE 4.

Four subregions of GW1Δ1a were capable of binding Ago2, and the interaction between GW1Δ1a and PIWI was not dependent on the five tryptophan residues. (A) Schematics of the four overlapping fragments of GW1Δ1a each containing two tryptophan residues and the mutant designed with all five tryptophans substituted with alanine. (B) Truncated constructs of GW1Δ1a containing any two continuous tryptophans were able to coprecipitate with PIWI. GST-PIWI was transfected with GFP-tagged TNR, GW1Δ1a, and its truncated constructs: Δ1a_W1-2 (“aa270-346,” contains the first and second tryptophans of GW1Δ1a), W2-3 (“aa318-339,” contains the second and third trytophans), W3-4 (“aa340-439,” contains the third and fourth tryptophan), and W4-5 (“aa409-495,” contains the fourth and fifth tryptophans). Compared with GFP-TNR (lane 2), all truncated constructs of GW1Δ1a (lanes 3–6) coprecipitated with GST-PIWI as full-length GW1Δ1a did (lane 1). (C) GW1Δ1a mutant without tryptophan still coprecipitated with PIWI. A GW1Δ1a mutant with all tryptophan (W) mutated to alanine (A) (GW1Δ1a Mut (W > A), lane 2) was cotransfected with GST-PIWI into HeLa cells. Compared with wild-type GW1Δ1a (lane 1), GW1Δ1a mutant was coprecipitated with PIWI at the comparable level. GFP-TNR (lane 3) served as a negative control in the GST pull-down assay.

FIGURE 5.

Both GW182 fragments GW1Δ1 and GW1Δ10 coprecipitated with other human Ago proteins. (A) Ago 1 and Ago4, but not Ago3 mutant, coprecipitated with GW182 fragments GW1Δ1 and GW1Δ10. GFP-tagged Ago1, Ago3m (Ago3 mutant), or Ago4 was cotransfected with GST-GW1Δ10 (lanes 1–6) or GST-GW1Δ1 (lanes 7–12). Ago3m is missing an exon (“aa757-823”), the C-terminal 66 amino acids of the PIWI domain compared with the reference sequence (NM_024852.2). Both Ago1 (lanes 4,10) and Ago4 (lanes 6,12) were pulled down by GST-GW1Δ10 or GST-GW1Δ1. In comparison, Ago3m was absent from either pull-down (lanes 5,11). (B) Ago3 coprecipitated with GW182 fragment GW1Δ1. Flag-Ago3 (lanes 1,3) or Flag-Ago1 (lanes 2,4) was cotransfected with GST-GW1Δ1 (lanes 1–4). GFP–Ago3m (lanes 5,6) was cotransfected with GST-GW1Δ1 as a negative control.

FIGURE 6.

Gene silencing mediated by tethered C-terminal half of Ago2 required GW182. (A) Tethered PIWI (“aa478-860”) down-regulated protein synthesis to the same extends of other Ago proteins. HeLa cells were transfected with constructs expressing the RL-5BoxB reporter, control FL reporter, and indicated NHA-tagged proteins. Bar graphs represent normalized mean values of RL/FL activities with standard errors. The RL/FL values in cells with tethered NHA-tagged Ago2, PIWI, Ago1, and Ago4 were significantly reduced compared with the value in NHA only group, which was normalized as 1. The NHA tagged PAZ, Ago3m, or HA-Ago2 did not show repression effect on the reporters. The expression of fusion proteins were determined by Western Blot using anti-HA mAb and are indicated below the bar graphs. The assay was repeated a minimum of three times. Asterisks indicate groups have significant difference compared with the NHA only group (unpaired t-test, P < 0.0001). No significant difference was shown between any two groups with asterisk (unpaired t-test, P > 0.05). (B) Translational repression mediated by tethered Ago2 or PIWI (“aa478-860”) was greatly impaired upon GW182 knockdown. HeLa cells were transfected with siRNA for either GW182 (siGW182) or GFP (siGFP). Thirty hours later, cells were transfected again with constructs expressing reporter RL-5BoxB, control FL reporter, and the indicated NHA-tagged proteins. Bar graphs represent the reduction of RL/FL in cells with tethered NHA-Ago2 or NHA-PIWI compared with those in cells with tethered NHA. The reduced values of RL/FL in cell transfected with siGFP were set as 1. Error bars indicate standard errors. The assay was performed in triplicate and was repeated two times. *Significant difference (unpaired t-test, P < 0.01). (C) GW182 knockdown by siRNA was confirmed by quantitative real-time PCR. The bar graphs represent normalized mRNA level of GW182 with standard errors. The mRNA level of GW182 in cells transfected with GFP-siRNA was set as 1. The experiment was performed in triplicate.

FIGURE 7.

At least three nonoverlapping GW-rich regions that are different from the ortholog-conserved GW-rich region can independently bind Ago2. The dot graph on top indicates the distribution of tryptophan in GW182. W, every tryptophan (magenta diamond); WG/GW, a glycine right adjacent to tryptophan (red triangle); AW/WA, an alanine right adjacent to tryptophan (blue diamond); W only, no glycine or alanine right adjacent to tryptophan (green diamond). The majority of the tryptophans are adjacent to either a glycine or alanine. The schematic of GW182 is indicated below the dot graph. GW-rich region, box in yellow; ortholog-conserved GW-rich region (“aa1074-1144”) (Till et al. 2007), box in red. Compared with GW1Δ12, which contains the ortholog-conserved region, GW1Δ1a, GW1Δ1b, and GW1Δ5 are the three nonoverlapping regions identified in the current study that binds Ago2. The amino acid sequence alignment between GW1Δ5/GW1Δ1b and ortholog-conserved GW-rich region was performed using ExPASy website tool. *Represent sequence identity.