Divergent GW182 functional domains in the regulation of translational silencing

Abstract

MicroRNA (miRNA)-mediated gene regulation has become a major focus in many biological processes. GW182 and its long isoform TNGW1 are marker proteins of GW/P bodies and bind to Argonaute proteins of the RNA induced silencing complex. The goal of this study is to further define and distinguish the repression domain(s) in human GW182/TNGW1. Two non-overlapping regions, Δ12 (amino acids 896-1219) containing the Ago hook and Δ5 (amino acids 1670-1962) containing the RRM, both induced comparable silencing in a tethering assay. Mapping data showed that the RRM and its flanking sequences in Δ5, but not the Ago hook in Δ12, were important for silencing. Repression mediated by Δ5 or Δ12 was not differentially affected when known endogenous repressors RCK/p54, GW182/TNGW1, TNRC6B were depleted. Transfected Δ5, but not Δ12, enhanced Ago2-mediated repression in a tethering assay. Transfected Δ12, but not Δ5, released endogenous miRNA reporter silencing without affecting siRNA function. Alanine substitution showed that GW/WG motifs in Δ12 (Δ12a, amino acids 896-1045) were important for silencing activity. Although Δ12 appeared to bind PABPC1 more efficiently than Δ5, neither Δ5 nor Δ12 significantly enhanced reporter mRNA degradation. These different functional characteristics of Δ5 and Δ12 suggest that their roles are distinct, and possibly dynamic, in human GW182-mediated silencing.

Figure 1.

Δ12 and Δ5 are two non-overlapping GW182 domains harboring repression in tethering function assay. The left panel shows GW182/TNGW1 and their series of truncation constructs. Amino acid residues of GW182 constructs are referenced to TNGW1, the longer isoform of GW182 (GenBank Accession NM_014494.2). TNR Q-repeat (green), glutamine repeat at the N-terminal domain of TNGW1; N-GW, M-GW and C-GW (yellow), three glycine/tryptophan-rich regions; Ago hook (red), a region reported to bind Ago protein; Q/N-rich (purple), glutamine/asparagine-rich region; DUF, DUF (orange); RRM (blue), RNA recognition motif. The right panel shows relative repression effects on reporter (either FL-5BoxB or RL-5BoxB; data combined as no difference was observed between reporters) by tethering the corresponding construct to the 3′-UTR of lucifearse mRNA. Their repression effects were normalized to NHA control, which was assigned as 1. Bar graphs show averages with standard errors (error bars); n, numbers of repeated experiments; Asterisk represents significant difference from NHA in t-test, P < 0.01; NS, no statistical significance.

Figure 2.

Roles of Ago hook and RRM domains for repression in tethering function assay and binding to Ago2. (A) The Ago hook in Δ12 was not critical for its repression effect. Compared to Δ12a which still retained 60% repression compared to NHA, Δ12b had only 27% repression. Results are expressed as mean ± standard error from three independent experiments. Asterisk represents significant difference between Δ12a to Δ12 and Δ12a to Δ12b using t-test, P < 0.01. (B) RRM domain and its flanking sequences in Δ5 were required for maximal repression effect. The repression activity of Δ5a was significantly reduced to only 55% of Δ5 (Asterisk represents significantly different compared to Δ5 in t-test, P < 0.01). The repression effect of Δ5b was not significantly different from that of Δ5 (P = 0.12). Results are expressed as mean ± standard error from three independent experiments. (C) Semi-quantitative western blot analysis showed differential binding of NHA-GW182 domains to GST-PIWI in GST pull-down assay. Bands (arrows) in blotting data were quantified and normalized to the total input (left panel) to obtain percent pull-down as shown for each lane at the bottom of the anti-HA panel. NHA-Δ12a, NHA-Δ12b and NHA-Δ5 showed only 0.2, 3.6 and 4.2% pull-down, respectively, whereas NHA-Δ12 showed 18%. Thus NHA-Δ12a, NHA-Δ12b and NHA-Δ5 showed only 1.1, 20 and 23% binding, respectively, to GST-PIWI compared to NHA-Δ12 as 100%. NHA-TNR and NHA-Δ10 served as negative and positive controls respectively. Asterisk, non-specific bands.

Figure 3.

Knockdown of endogenous repressors did not affect repression activity of the two defined domains Δ12 and Δ5 in the tethering assay. (A) Repression by Δ12 and Δ5 were not significantly altered when GW182, TNRC6B, GW182/TNRC6B, or RCK/p54 was knocked down by respective siRNAs. Results are expressed as mean ± standard error from three independent experiments. There is no statistical significance difference comparing each knockdown to siGFP within NHA-Δ12 or -Δ5 group (NS, t-test). Efficiency of siRNA knockdown was monitored in each individual experiment using qRT–PCR for GW182 (B), TNRC6B (C–D) compared to the untreated HeLa cell control, or western blot analysis for RCK/p54 compared to siGFP transfected controls (E). The extract from cells transfected with NHA and siGFP was loaded at three concentrations (100, 50 and 25%) to demonstrate the semi-quantitative detection of RCK/p54.

Figure 4.

Δ5, but not Δ12, enhanced Ago2-mediated repression in the tethering function assay. (A) Co-expressing GST-Δ5 with NHA-Ago2 that was tethered to the RL-5BoxB reporter enhanced the NHA-Ago2-mediating repression from 60 to 80% (33% enhanced repression). In contrast, co-expression of GST-Δ12, -QN, or -Ago2 did not show significant change. Results are expressed as mean ± standard error from three independent experiments. Asterisk represents t-test compared NHA-Ago2+GST-Δ5 with NHA-Ago2 alone, P < 0.01. (B) Identical experiments as in (A) except GFP-tagged proteins were used in place of GST fusion proteins. Results are expressed as mean ± standard error from three independent experiments. Representative cell lysates were analyzed by western blot to demonstrate expression of NHA-Ago2 and GFP fusion proteins with tubulin expression shown as loading controls. Asterisk represents significant difference in t-test compared NHA-Ago2+GFP-Δ5 with NHA-Ago2 alone, P < 0.01.

Figure 5.

Δ12 significantly interfered with endogenous miRNA repression. (A) GST-Δ12, but not GST-Δ5, interfered with miR-20-mediated repression. Three different amounts (0.6, 0.3 and 0.15 µg) of GST-TNR, -Δ1, -Δ12 and -Δ5 plasmids were co-transfected with the RL-20 bulge reporter into HEK293 cells. The RL-20 bulge expression was significantly increased when cells co-expressed Δ12. Overexpression of Δ1 also mildly interfered with miRNA function. Asterisk represents significant difference in t-test compared with +GST-TNR, P < 0.01, n = 3; Double asterisks represent highly significant difference in t-test compared with +GST-TNR, P < 0.0001, n = 3. (B) Δ12 and its deletion constructs only interfered with miRNA but not siRNA mediated repression. NHA tag, NHA-TNGW1, -Δ12, -Δ12a, -Δ12b and -Δ5 were co-transfected with reporters RL-20 bulge/FL or RL-20 perfect/FL in HEK293 cells. Compared to the NHA control, the relative activity of RL-20 bulge was significantly increased in cells expressing NHA-Δ12 and -Δ12a/b but not in cells expressing -TNGW1 or -Δ5. NHA-Δ12 and its deletion constructs did not interfere with RL-20 perfect reporter that repressed by siRNA pathway. Results are expressed as mean ± standard error from three independent experiments. Asterisk represents significant difference in t-test compared with +NHA, P < 0.01.

Figure 6.

Substitution of glycine (G) and tryptophan (W) residues with alanine (A) in GW1Δ12a interfered with its repression on reporter and interference in miRNA repression activity. (A) Amino acid sequence of Δ12a shown with GW and WG residues are underlined; these residues are substituted with AA to generate a mutant Δ12am. (B) Δ12am had no repression activity in RL-5BoxB tethering assay. NHA-Δ5 and its deletion constructs served as positive controls. Asterisk represents significant different in t-test compared with NHA-Δ12, P < 0.01, n = 3. (C) GW/GW mutated Δ12a (Δ12am) no longer affected miRNA mediated repression. Experiment was performed as described in Figure 5B. Asterisk represents significant difference in t-test compared NHA-Δ12a with NHA, P < 0.01.

Figure 7.

Binding of Δ12 and Δ5 to PABPC1 did not significantly affect reporter mRNA degradation. (A) Differential binding of GW182 fragments Δ12, Δ7 and Δ5 to PABPC1. GST-PABPC1 was co-transfected with different NHA-tagged constructs into HeLa cells as shown above the panels for the designed GST pull-down assay. After 24 h, cell lysates were harvested and analyzed by GST pull-down followed by western blot analysis. GST-PABPC1 strongly pulled down NHA-Δ7 compared to NHA-Δ12 and -Δ5 (arrow). (B) Both tethered Δ12 and Δ5 induced primarily translational repression with only moderate reporter mRNA degradation. To determine the mRNA level of the reporter in tethering assay, a pair of each RL and FL primers was utilized in SYBR-Green qRT–PCR. The RL mRNA level was normalized to FL mRNA. All results are expressed as mean ± SD from three independent experiments. Asterisk represents significant difference in t-test compared with NHA, P < 0.01. No significant difference observed in mRNA degradation for any of the constructs compared with NHA.

Figure 8.

Summary of GW182 domain functional characteristics. DUF, sequence identified to be important for PABPC1 binding. A reference schematic map of TNGW1 in one recent review (15) is included for comparison. UBA, Ubiquitin-associated domain; PAM2, PABP-interacting motif 2. Regions M1 and M2 together with PAM2 formed the Mid region. Middle region and C-terminal region but not RRM defined the bipartite silencing domain.