Research article

Phosphorylation of Argonaute 2 at serine-387 facilitates its localization to processing bodies

Yan Zeng, Heidi Sankala, Xiaoxiao Zhang, Paul R. Graves


Ago (Argonaute) proteins are essential effectors of RNA-mediated gene silencing. To explore potential regulatory mechanisms for Ago proteins, we examined the phosphorylation of human Ago2. We identified serine-387 as the major Ago2 phosphorylation site in vivo. Phosphorylation of Ago2 at serine-387 was significantly induced by treatment with sodium arsenite or anisomycin, and arsenite-induced phosphorylation was inhibited by a p38 MAPK (mitogen-activated protein kinase) inhibitor, but not by inhibitors of JNK (c-Jun N-terminal kinase) or MEK [MAPK/ERK (extracellular-signal-regulated kinase) kinase]. MAPKAPK2 (MAPK-activated protein kinase-2) phosphorylated bacterially expressed full-length human Ago2 at serine-387 in vitro, but not the S387A mutant. Finally, mutation of serine-387 to an alanine residue or treatment of cells with a p38 MAPK inhibitor reduced the localization of Ago2 to processing bodies. These results suggest a potential regulatory mechanism for RNA silencing acting through Ago2 serine-387 phosphorylation mediated by the p38 MAPK pathway.

  • Argonaute (Ago)
  • mitogen-activated protein kinase (MAPK)
  • phosphorylation
  • processing body
  • RNA interference (RNAi)


Non-coding RNAs of approx. 20–30 nt have been shown to act as regulators of chromatin structure, epigenetic modifications, transcriptional regulation, mRNA translation/degradation and as a defence against transposons [1]. Several different classes of these RNAs have been described, including siRNAs (small interfering RNAs) [2], miRNAs (microRNAs) [3] and piwiRNAs [4]. MiRNAs, which constitute the largest class of endogenously expressed small RNAs, have been shown to regulate the expression of a large number of protein-coding genes involved in development, metabolism, cell growth, cell death and cell-fate determination [3]. Moreover, altered expression of miRNAs has been associated with human diseases, including cancer [5] and heart disease [6].

To perform their function, the small RNA molecules interact with one or more members of a highly conserved protein family known as the Agos (Argonautes). Ago proteins have important regulatory functions in organisms ranging from archaea to mammals [711], and associate directly with siRNAs, miRNAs and piwiRNAs [1220]. For some Ago family members, the Ago–RNA complexes have been shown to silence gene expression, either transcriptionally or post-transcriptionally [2,2123].

The mammalian Ago protein family consists of eight members, divided into the Ago and Piwi subfamilies, the latter of which are expressed only in germ cells. Although all four Ago family members have been implicated in translational inhibition of mRNA [24], only one Ago protein, Ago2, possesses endoribonuclease activity, first identified in RISC (RNA-induced silencing complex) [16,17,2527]. Ago2 is an essential protein, as mice deficient in Ago2 display defective neural tube and heart development, and cells cultured from Ago2-deficient mice are defective for RNAi (RNA interference) [16]. Ago2 is also uniquely required for the maturation and expression of miRNAs [28,29].

Although miRNAs play prominent roles in regulating gene expression, the regulation of RNA silencing itself is less well understood. However, it was shown recently that cellular stress can have an impact on the function of miRNAs and even determine whether some miRNAs inhibit or activate mRNA translation [30,31]. One conceivable mechanism for regulation of RNA silencing is through the modulation of Ago proteins. This could occur through the temporal and spatial regulation of Ago expression or via modulation of Ago activity by associated proteins [10,11]. Another mechanism may be via the regulation of Ago2 localization. It was shown recently that the subcellular location of Ago was altered by cellular stress [32]. Ago proteins are found mostly in the cytoplasm, where they partially localize to distinct RNA-containing granules, termed PBs (processing bodies) [3335]. However, treatment with protein translation inhibitors or arsenite, a cell-stress agent, induced the rapid accumulation of Ago to PBs and stress granules [32], indicating that Ago2 localization is dynamic and regulated by the cellular environment.

To explore mechanisms of Ago protein regulation, we examined Ago2 for evidence of phosphorylation. In the present study, we show that human Ago2 is phosphorylated on serine-387 by a stress-activated mechanism acting through the p38 MAPK (mitogen-activated protein kinase) pathway and that serine-387 phosphorylation contributes to the localization of Ago2 to PBs under normal growth conditions.


Plasmid construction

pIRES-neo-FLAG-HA-Ago2 was a gift from Dr Thomas Tuschl (Howard Hughes Medical Institute, Laboratory of RNA Molecular Biology, Rockefeller University, New York, NY, U.S.A.). To create pGEX-4T-1-Ago2, full-length human Ago2 was amplified by PCR using primers containing BamHI and EcoRI sites at the 5′ and 3′ ends of the primers respectively, digested with BamHI and EcoRI, and ligated into the pGEX-4T-1 vector (GE Healthcare). Mutations were generated with the QuikChange® site-directed mutagenesis kit (Stratagene) and confirmed by DNA sequencing.

Cell culture and transfection

HEK-293T cells (human embryonic kidney cells expressing the large T-antigen of simian virus 40) and H1299 cells were maintained at 37 °C in DMEM (Dulbecco's modified Eagle's medium) with 10% (v/v) fetal bovine serum in an atmosphere of 5% CO2. Transfections were performed using Lipofectamine™ 2000 (Invitrogen). Sodium arsenite and anisomycin were both from Sigma.

In vivo labelling, phospho-amino acid analysis and two-dimensional phosphopeptide mapping

HEK-293T cells transfected with FLAG–HA–Ago2 (where HA is haemagglutinin) were grown for 36 h, incubated in phosphate-free DMEM for an additional 3 h and labelled in DMEM containing 1 mCi/ml [32P]H3PO4 (PerkinElmer) for 12 h. Cells were lysed in buffer A [50 mM Tris/HCl (pH 7.5), 150 mM NaCl, 0.5% Nonidet P40, 2 mM EDTA, 1 μM microcystin-LR and 1× EDTA-free protease inhibitor cocktail (Roche)]. Cell lysates were clarified by centrifugation at 13000 g for 15 min and treated with 12.6 Kunitz units/ml RNase A (Sigma) for 20 min at room temperature (23 °C). The supernatant was mixed with anti-FLAG M2 antibody–agarose (Sigma), and, after washing the beads four times with buffer A, Ago2 was eluted with 0.5 mg/ml FLAG peptide in TBS. Silver-stained Ago2 bands were excised and in-gel digested with trypsin (Promega). Phospho-amino acid analysis of labelled Ago2 was performed as described previously [36]. For two-dimensional phosphopeptide mapping, tryptic digests were spotted on to thin-layer cellulose plates and resolved by electrophoresis (60 min at 1000 V) in 2.5% (v/v) formic acid and 7.8% (v/v) acetic acid (pH 1.9). Peptides were resolved in the second dimension by chromatography in phosphopeptide buffer [62.5% (v/v) isobutyric acid, 1.9% (v/v) n-butanol, 4.8% (v/v) pyridine and 2.9% (v/v) acetic acid] [37] and visualized by autoradiography.


Ago2 tryptic peptides were purified with Poros 20 R2 reverse phase packing (Applied Biosystems) and subjected to direct-infusion nanospray using NanoES spray capillaries (Proxeon, Odense, Denmark) on an Applied Biosystems QSTAR® pulsar XL mass spectrometer. MS spectra were collected both in data-dependent acquisition mode and in manual mode using an ion spray voltage of 800 V, a curtain gas of 20, a declustering potential of 75 V and a focusing potential of 280 V. For data-dependent acquisition, MS data were collected for a mass range of 400 m/z to 2000 m/z with a charge state of 2–5 which exceed one count. MS/MS (tandem MS) data were acquired for ions from a mass range of 60–2000 m/z with a dwell time of 15 s per ion. Identification of phosphorylation sites in Ago2 was achieved by manual interpretation of MS/MS spectra with the aid of Analyst QS 1.1 (MDS Sciex, Concord, ON, Canada).

Antibodies, immunoprecipitation and Western blotting

To generate an antibody against the serine-387-phosphorylated form of Ago2 (P-S387), a phosphopeptide (KLMRSApSFNTDPY) was synthesized and used as an immunogen (Abgent). For immunoprecipitation of endogenous Ago2, an antibody was generated against an N-terminal peptide of human Ago2 (YSGAGPALAPPAPPPPIQGY; Protein-Tech, Chicago, IL, U.S.A.) (denoted N-Ago2). Ago2 was immunoblotted with a mouse monoclonal antibody directed against the C-terminal 377 amino acids of human Ago2 (H00027161-M01; Abnova, Taipei, Taiwan) or a rabbit polyclonal anti-Ago2 antibody (07-590, Upstate Biotechnology). Rabbit polyclonal anti-[phospho-p38 MAPK (threonine-180/tyrosine-182)], anti-(p38 MAPK), anti-[phospho-MAPKAPK2 (MAPK-activated protein kinase-2) (threonine-334)] and anti-MAPKAPK2 antibodies were from Cell Signaling Technology. Goat anti-actin antibody was from Santa Cruz Biotechnology (sc-1615). For immunoprecipitation of endogenous Ago2, clarified HEK-293T cell lysates were mixed with the P-S387 antibody bound to Protein A–agarose. Protein A–agarose was washed four times with buffer A and incubated for 5 min at 100 °C in 2× SDS sample buffer. Samples were resolved by SDS/PAGE (8% or 12% gels) and transferred on to nitrocellulose membranes following methods published previously [36]. The primary antibodies used for immunoblotting were (all at 1:1000 dilution): anti-FLAG antibody, P-S387 antibody, rabbit anti-Ago2 antibody, anti-[phospho-p38 MAPK (threonine-180/tyrosine-182)] antibody, anti-(p38 MAPK) antibody, anti-[phospho-MAPKAPK2 (threonine-334)] antibody, anti-MAPKAPK2 antibody, anti-actin antibody and N-Ago2 antibody. Primary rabbit polyclonal antibodies were detected using the following secondary antibodies (all at 1:7000 dilution): IRDye800-conjugated affinity-purified rabbit anti-IgG antibody (Rockland Immunochemicals), Alexa Fluor® 680-conjugated goat anti-mouse IgG antibody (Molecular Probes) and Alexa Fluor® 680-conjugated rabbit anti-goat IgG antibody (Invitrogen). Proteins were visualized using the Odyssey system (LI-COR).

Expression and purification of GST (glutathione transferase)–Ago2 and kinase assays

Escherichia coli DH5α cells were transformed with GST–Ago2 WT (wild-type) or GST–Ago2 S387A and grown at 37 °C until an attenuance (D600) of 0.6 was reached. Cells were induced with 0.5 mM IPTG (isopropyl β-D-thiogalactoside) and grown for 3 h at 37 °C before harvesting. GST–Ago2 was then purified using the method of Frangioni and Neel [38]. Kinase assays were conducted by incubating 50 ng of GST–MAPKAP2 (MAPK-activated protein 2) (#SE-386, BIOMOL) with ∼100 ng of full-length recombinant GST–Ago2 WT or GST–Ago2 S387A in a total volume of 20 μl in kinase assay buffer [2 mM dithiothreitol, 25 mM MgCl2, 10 μM ATP and 10 μCi of [γ-32P]ATP (PerkinElmer)]. Kinase reactions were performed for 30 min at 37 °C and analysed by SDS/PAGE and autoradiography.

Immunofluorescence and microscopy

H1299 cells were grown on 12-mm-diameter round coverslips (VWR International) in 12-well plates and transfected with Myc–Ago2 WT, Myc–Ago2 S387A, FLAG–Ago2 WT, FLAG–Ago2 S387A or FLAG–Ago2 S385A using Lipofectamine™ 2000 (Invitrogen). Cells were washed twice with PBS, fixed for 15 min at room temperature with 4% (w/v) paraformaldehyde in PBS, washed twice with PBS containing 10 mM glycine (pH 7.4) and incubated for 1 h in blocking/permeabilization buffer [10 mM glycine, 1% BSA and 0.5% Triton X-100 in PBS (pH 7.4)]. Cells were incubated for 1 h with mouse monoclonal anti-FLAG antibody (1:1000 dilution; Sigma) or rabbit polyclonal anti-Myc antibody (1:200 dilution; Santa Cruz Biotechnology) and/or mouse monolonal anti-GW182 antibody (1:200 dilution; Abcam) in blocking/permeabilization buffer, washed three times and then incubated for 1 h with Alexa Fluor® 488-conjugated goat anti-mouse antibody (1:400 dilution; Molecular Probes) and/or Alexa Fluor® 594-conjugated goat anti-rabbit antibody (1:400 dilution; Molecular Probes). Two final washes preceded mounting of coverslips on to glass slides (Fisher) using the ProLong® Antifade Kit (Molecular Probes) and examined with a Zeiss LSM 510 laser-confocal microscope using a ×63 oil-immersion objective lens.


Recombinant and endogenous Ago2 is phosphorylated on serine-387

To determine whether human Ago2 is phosphorylated in intact cells, HEK-293T cells were transfected with FLAG-tagged Ago2 labelled with [32P]H3PO4. As shown in Figure 1(A), recombinant Ago2 was labelled with 32P in vivo. Phospho-amino acid analysis of 32P-labelled Ago2 indicated that it was phosphorylated on serine residues (Figure 1A). To identify Ago2 phosphorylation sites, purified Ago2 was digested with trypsin and the peptides were analysed by MS. A total of 29 Ago2 tryptic peptides were identified, resulting in a sequence coverage of 48% (see Supplementary Figure S1 at One tryptic Ago2 peptide, S385ASFNTDPYVR395, was identified in an unmodified form at [M+2H]2+ =628 m/z and as a phosphopeptide at [M+2H]2+=668 m/z (Figure 1B). MS/MS analysis of the ion at [M+2H]2+=668 m/z indicated that the peptide was phosphorylated on residue serine-387, S385ApSFNTDPYVR395 (Figure 1C). To determine the contribution of serine-387 phosphorylation to total Ago2 phosphorylation, in vivo 32P-labelled Ago2 was subjected to two-dimensional tryptic phosphopeptide mapping. Equal amounts of Ago2 WT, Ago2 S387A (the serine-387 phosphorylation-site mutant) or Ago2 S385A (Ago2 with a mutation in a nearby serine residue) were analysed (Figure 2A). One major spot was consistently observed for Ago2 WT (Figure 2B) and this spot was absent in the two-dimensional map of the S387A mutant, but present in that of Ago2 S385A (Figure 2B). Thus S385 is not phosphorylated, whereas serine-387 is the major site of Ago2 phosphorylation in vivo. Serine-387 is located in the mid-domain of Ago2, between the highly conserved PAZ (Piwi, Argonaut and Zwille) and PIWI domains (Figure 2C). Among the human Ago family, serine-387 is conserved in Agos 1, 2 and 4, but is not present in Ago3 (Figure 2D).

Figure 1 Identification of serine-387 as an Ago2 phosphorylation site

(A) HEK-293T cells were mock-transfected (mock) or transfected with FLAG–HA–Ago2 (Ago2) and labelled in the presence of [32P]H3PO4. Ago2 was immunoprecipitated with FLAG–agarose, resolved by SDS/PAGE, and subjected to silver staining and autoradiography. Phospho-amino acid analysis of Ago2 was also performed. Autoradiography and ninhydrin staining of phospho-amino acid standards (P-Ser, phosphoserine; P-Thr, phosphothreonine; P-Tyr, phosphotyrosine) are shown. Molecular masses are indicated on the left-hand side (in Mr). (B) A zoom scan of a portion of the time-of-flight mass spectrum of Ago2 indicating the presence of a putative phosphopeptide. (C) The MS/MS spectrum of the SApSFNTDPYVR phosphopeptide, [M+2H]2+=668. Prominent y- and b-ions are shown. 2e3 is 2×103 etc.

Figure 2 Identification of serine-387 as the major Ago2 phosphorylation site

(A) HEK-293T cells were transfected with FLAG–HA–Ago2 WT or the indicated mutants (Ago2 S385A or S387A), labelled in the presence of [32P]H3PO4, and Ago2 was immunoprecipitated with FLAG–agarose, resolved by SDS/PAGE and silver-stained. (B) 32P-labelled Ago2 WT, Ago2 S385A and Ago2 S387A were excised and subjected to two-dimensional phosphopeptide mapping. Arrows indicate the two-dimensional map origins. (C) Schematic diagram indicating the location of serine-387 with regard to the conserved PAZ (Piwi, Argonaut and Zwille) and PIWI domains in human Ago2. (D) Alignment of human Agos 1, 2 and 4 with Ago2 (amino acids 380–395).

To monitor serine-387 phosphorylation, an antibody against the serine-387-phosphorylated form of Ago2 (P-S387) was generated. Specificity of the antibody was shown by the loss of reactivity of P-S387 towards Ago2 after treatment with CIP (calf intestinal phosphatase) and an absence of reactivity towards the Ago2 S387A mutant (Figure 3A). Evidence that endogenous Ago2 was phosphorylated on serine-387 was revealved by immunoprecipitation of endogenous Ago2 with the P-S387 antibody in the absence, but not the presence, of CIP (Figure 3B). Immunoprecipitated endogenous Ago2 also contained the same phosphopeptide S385ApSFNTDPYVR395 as determined by MS (see Supplementary Figure S2 at

Figure 3 Endogenous Ago2 is phosphorylated on serine-387 in vivo

(A) HEK-293T cells were transfected with FLAG–HA–Ago2 WT or the Ago2 S387A mutant and immunoprecipitated (IP) with FLAG–agarose. WT immunoprecipitates were treated in the absence or presence of 50 units of CIP for 10 min at room temperature and then all immunoprecipitates were immunoblotted (IB) with the indicated antibodies (anti-FLAG antibody or P-S387 antibody). (B) Endogenous Ago2 was immunoprecipitated (IP) from HEK-293T cells using P-S387 antibody or the pre-immune sera (Pre-I; 1:1000 dilution), incubated in the absence (−) or presence (+) of CIP (50 units) and immunoblotted (IP) with an anti-Ago2 antibody.

Ago2 phosphorylation is mediated by the p38 MAPK pathway

To investigate the regulation of Ago2 phosphorylation, FLAG-tagged Ago2 was expressed in HEK-293T cells and serine-387 phosphorylation was monitored under different conditions. We found that sodium arsenite treatment resulted in an ∼8–10-fold increase in Ago2 serine-387 phosphorylation (Figures 4A and 4B). No immunoreactivity was observed with the P-S387 antibody against the Ago2 S387A mutant either in the absence (Figure 3) or the presence (not shown) of sodium arsenite treatment. Sodium arsenite-induced serine-387 phosphorylation was dose- and time-dependent (see Supplementary Figure S3 at and was also observed after transfection of FLAG–Ago2 into MCF-7 or H1299 cells (results not shown). Since several members of the MAPK family are activated by sodium arsenite, we tested whether inhibitors of MAPKs could prevent sodium arsenite-induced Ago2 phosphorylation. We treated cells with sodium arsenite in the presence of SB203580, SP600125 or PD98059, a p38 MAPK, SAPK/JNK (c-Jun N-terminal kinase) or MEK [MAPK/ERK (extracellular-signal-regulated kinase) kinase] inhibitor respectively. As shown in Figure 4, 5 μM SB203580 inhibited sodium arsenite-induced serine-387 phosphorylation to ∼33% of control phosphorylation (without treatment). In contrast, 30 μM SP600125 or 50 μM PD98059 had no effect (Figures 4A and 4B). Sodium arsenite treatment resulted in the activation of both p38 and MAPKAPK2 (as judged by an increase in their phosphorylation), and this activation was blocked by SB203580 (Figure 4C). Because we identified a role for p38 MAPK, we also tested whether anisomycin, a p38 activator, could also affect Ago2 phosphorylation. Indeed, anisomycin induced serine-387 phosphorylation ∼3-fold (Figures 4D and 4E). In addition to recombinant Ago2, we found that sodium arsenite treatment caused a 6–8-fold increase in serine-387 phosphorylation of endogenous Ago2, as shown by immunoprecipitation of Ago2 with the P-S387 antibody (Figure 5). Furthermore, immunoprecipitation of endogenous Ago2 with the P-S387 antibody was reduced by ∼50% after cells were treated with SB203580 (Figures 5A and 5B). The use of a reciprocal immunoprecipitation approach, i.e. immunoprecipitation of Ago2 followed by immunoblotting with the P-S387 antibody, produced similar results (Figure 5C). Thus Ago2 has a basal level of serine-387 phosphorylation, which can be enhanced by activation of the p38 MAPK pathway.

Figure 4 Serine-387 phosphorylation of recombinant Ago2 is induced by sodium arsenite and inhibited by SB203580

(A) HEK-293T cells were mock-transfected (mock) or transfected with FLAG–HA–Ago2 and treated with 250 μM sodium arsenite (AS) for 90 min at 37 °C in the absence or presence of 5 μM SB203580, 30 μM SP600125 (SP) or 50 μM PD98059 (PD) at the indicated concentrations. Kinase inhibitors were added 15 min prior to sodium arsenite treatment and cell lysates were immunoblotted (IB) with the indicated antibodies (anti-FLAG antibody and P-S387 antibody). (B) Quantification of the results in (A). The ratio of serine-387-phosphorylated Ago2 (P-S387) to total Ago2 is shown, with the serine-387-phosphorylated Ago2/total Ago2 ratio from untreated cells set to 1. Results are means±S.D. (n=3). (C) HEK-293T cell lysates from (A) treated with AS and SB203580 were immunoblotted (IB) using anti-FLAG antibody, P-S387 antibody, anti-[phospho-p38 MAPK (threonine-180/tyrosine-182)] antibody (P-p38), anti-(p38 MAPK) antibody (p38), anti-[phospho-MAPKAPK2 (threonine-334)] antibody (P-MK2) and anti-MAPKAPK2 antibody (MK2). (D) HEK-293T cells transfected with FLAG–HA–Ago2 were treated in the absence (−) or presence (+) of 0.1 μg/ml anisomycin for 15 min at 37 °C before immunoblotting (IB) with anti-FLAG antibody, P-S387 antibody, anti-[phospho-p38 MAPK (threonine-180/tyrosine-182)] antibody and anti-(p38 MAPK) antibody. (E) Quantification of the fold change in the serine-387-phosphorylated Ago2 (P-S387)/total Ago2 ratio of results from (D). Results are means±S.D. (n=4).

Figure 5 Serine-387 phosphorylation of endogenous Ago2

(A) HEK-293T cells were treated with vehicle (DMSO) or in the presence (+) or absence (−) of 250 μM sodium arsenite (AsNaO2), 5 μM SB203580, or sodium arsenite plus SB203580 for 90 min at 37 °C. Immunoprecipitation (IP) with the P-S387 antibody was performed, followed by immunoblotting (IB) with the indicated antibodies (anti-Ago2 and anti-actin antibodies). (B) Quantification of the fold change in serine-387-phosphorylated Ago2 (P-S387)/total Ago2 ratio for results in (A). Results are means±S.D. (n=3) (C) HEK-293T cells were incubated in the presence (+) or absence (−) of 250 μM sodium arsenite for 90 min at 37 °C and endogenous Ago2 was immunoprecipitated (IP) with an N-terminal Ago2 antibody (N-Ago2) and immunoblotted (IB) with the P-S387 antibody. Total-cell lysates treated with or without sodium arsenite are shown as a control.

MAPKAPK2 phosphorylates Ago2 in vitro

Examination of the sequence surrounding serine-387 (Figure 2D) indicates that it is not part of a consensus p38 MAPK site, and p38α failed to phosphorylate Ago2 in vitro (results not shown). However, the sequence does conform to a consensus substrate for MAPKAPK2 [39], a kinase known to be activated by the p38 MAPK pathway [40]. Therefore we tested whether MAPKAPK2 could phosphorylate Ago2 in vitro. Indeed, we found that MAPKAPK2 phosphorylated recombinant bacterially produced GST–Ago2 WT, but not the S387A mutant (Figure 6), providing further evidence that the p38 MAPK pathway may be involved in serine-387 phosphorylation.

Figure 6 MAPKAP2 phosphorylates Ago2 on serine-387 in vitro

Full-length recombinant GST–Ago2 WT or the S387A mutant was expressed and purified from E. coli and incubated with recombinant MAPKAP2 in a kinase reaction. An autoradiograph and a Coomassie Brilliant Blue-stained polyacrylamide gel are shown. GST–Ago2 WT not incubated with MAPKAP2 is shown as a control. The experiment was performed twice with similar results.

Serine-387 phosphorylation mediates Ago2 localization to PBs

To investigate the function of Ago2 phosphorylation, we expressed recombinant Ago2 in H1299 cells and monitored its localization by immunofluorescence. In agreement with previous results, we found that FLAG- or Myc-tagged Ago2 WT partially localized to PBs [34,35], as confirmed by co-immunostaining with an antibody against GW-182, a PB marker (Figure 7B). The Ago2 S385A mutant behaved similarly to Ago2 WT; in contrast, the Ago2 S387A mutant exhibited significantly less accumulation in PBs (Figures 7A–7D). Mutation of serine-387 to an alanine residue greatly reduced the percentage of cells containing high numbers of Ago2-stained PBs (defined as ≥5 PBs) and increased the percentage of cells containing low numbers of PBs (defined as 0–1 PB) (Figure 7C). This difference was also reflected in the average number of Ago2-stained PBs per cell, which was reduced by 68% in cells expressing the Ago2 S387A mutant (Figure 7D). All Ago2 proteins were expressed at similar levels (Figure 7E). Treatment of Ago2 with SB203580 also reduced the percentage of cells with high numbers of Ago2-stained PBs and increased the percentage of cells with low numbers of PBs, similar to the S387A mutant. SB203580 treatment also reduced the average number of Ago2-stained PBs per cell by 53%. In contrast, treatment with SP600125 or PD980059 had no effect (Figure 7A). Finally, because sodium arsenite was shown to induce the localization of Ago1–Ago4 to foci such as PBs [32], we treated cells expressing Ago2 WT or the Ago2 S387A mutant with sodium arsenite. Sodium arsenite treatment led to an increase in the localization of both Ago2 WT and the S387A mutant to PBs (Figure 7B), indicating that sodium arsenite-induced Ago2 PB localization does not require serine-387 phosphorylation.

Figure 7 Serine-387 phosphorylation facilitates Ago2 accumulation in PBs

(A) H1299 cells were transfected with FLAG–Ago2 WT, Ago2 S385A or Ago2 S387A, and Ago2 localization was determined by immunofluorescence using the anti-FLAG antibody. H1299 cells were co-stained with DAPI (4′,6-diamidino-2-phenylindole) to identify nuclei. Where indicated, cells were treated with 5 μM SB203580 (SB), 30 μM SP600125 (SP) or 50 μM PD980059 (PD) for 90 min at 37 °C prior to fixation. (B) H1299 cells were transfected with Myc–Ago2 WT or Myc–Ago2 S387A and left untreated or treated with 250 μM sodium arsenite (arsenic) for 90 min at 37 °C prior to fixation. Following treatment, cells were co-stained with rabbit polyclonal anti-Myc (green) and mouse monoclonal anti-GW-182 (red) antibodies and analysed at ×3 zoom using a ×63 lens. (C) The percentage of cells containing the indicated number of FLAG–Ago2-stained PBs was determined from counting at least 40 cells for each mutant (n=3) (D) The average number of FLAG–Ago2-stained PBs per cell was calculated from counting at least 40 cells for each sample and P-values are shown (n=3). (E) H1299 cells were transfected with FLAG–Ago2 WT, S387A or S385A, and total-cell lysates were immunoblotted (IB) with the anti-FLAG antibody and the anti-actin antibody to control for protein loading.

Our findings are summarized in the model in Figure 8. Ago2 is phosphorylated at serine-387 in cultured cells, and this phosphorylation is necessary for mediating the localization of Ago2 to PBs under normal growth conditions. The p38 MAPK pathway is necessary for PB localization of Ago2, as treatment with SB203580 blocks Ago2 accumulation in PBs. One form of cellular stress, sodium arsenite treatment, induces serine-387 phosphorylation further through the activation of the p38 MAPK pathway; under these conditions, localization of Ago2 to PBs is significantly enhanced, but serine-387 phosphorylation is not required for the localization.

Figure 8 Model of Ago2 regulation by serine-387 phosphorylation

Under normal growth conditions, Ago2 is phosphorylated on serine-387, and this phosphorylation is necessary for the accumulation of Ago2 in PBs. Phosphorylation of serine-387 can be induced by cellular stress mediated by the p38 MAPK pathway. Sodium arsenite treatment induces Ago2 phosphorylation and the localization of Ago2 to PBs, but phosphorylation at serine-387 is not required for sodium arsenite-induced localization. Additional forms of cellular stress may act to regulate serine-387 phosphorylation and the localization of Ago2.


A large body of evidence has demonstrated the importance of RNA silencing in a variety of biological processes [2,41]. However, little is known about how the mechanisms of RNA silencing are regulated. Ago2 plays critical roles in RNA silencing [42] and, despite high sequence identity between mammalian Ago isoforms, Ago2 is uniquely required for RNAi [16] and miRNA expression [28,29]. Therefore Ago2 may represent a nodal point for the regulation of RNA silencing. In the present study, we show for the first time that human Ago2 is a phosphoprotein and identify serine-387 as the major phosphorylation site. Notably, serine-387 phosphorylation is regulated by cellular stress. It was strongly induced by sodium sodium arsenite, a known activator of p38 MAPK. The p38 MAPK inhibitor SB203580 consistently blocked the majority of sodium arsenite-induced serine-387 phosphorylation, but inhibitors of JNK and MEK did not (Figure 4). Moreover, the activation state of p38 MAPK and MAPKAPK2 in vivo directly paralleled the level of serine-387 phosphorylation in Ago2. Anisomycin, a p38 activator, also induced serine-387 phosphorylation, indicating that activation of p38 MAPK by different mechanisms was sufficient to induce Ago2 phosphorylation, and that Ago2 phosphorylation may be regulated by a variety of stress signals acting through the p38 MAPK pathway. Furthermore, we showed that, in vitro, Ago2 can be phosphorylated directly by MAPKAPK2, and that this phosphorylation occurred exclusively on serine-387 of Ago2 (Figure 6). Experiments are underway to determine the role of MAPKAPK2 in mediating Ago2 phosphorylation in vivo.

What might be the function of serine-387 phosphorylation? We found that Ago2 phosphorylation plays a critical role in mediating its localization to PBs under normal cell-culture growth conditions. Mutation of serine-387 to an alanine residue, or treatment of cells expressing Ago2 WT with SB203580 (but not SP600125 or PD980059), greatly reduced the localization of Ago2 to PBs. Mutation of a nearby serine residue, serine-385, to an alanine residue, had no effect, arguing against a general disruption of protein structure. Analysis of PB formation with the PB marker GW-182 indicated that PBs were still present in Ago2 S387A-transfected cells, but that Ago2 was not localized to the PBs.

Although evidence suggests that the presence of visible PBs is not required for RNAi [43,44], the proper localization of Ago2 to PBs requires an intact miRNA-binding domain, suggesting that disruption of Ago2 function may inhibit its localization to PBs [35]. Moreover, the formation of PBs, under certain circumstances, may be related to RNAi activity [45]. Thus a decrease in the localization of Ago2 to PBs, through the lack of serine-387 phosphorylation, may be a reflection of the loss of Ago2 activity and disruption of the RNA-silencing pathway in cells. It was shown recently that sodium arsenite caused rapid re-localization of Ago1–Ago4 from the cytosol to PBs and stress granules [32]. We found that although sodium arsenite did induce serine-387 phosphorylation, both Ago2 WT and Ago2 S387A localized to PBs after sodium arsenite treatment, indicating that other mechanisms are in place after sodium arsenite treatment that mediate Ago2 localization to PBs.

In summary, our results indicate that Ago2, the key protein responsible for mediating RNAi and miRNA expression, is phosphorylated in cells, and its phosphorylation can be induced by cellular stress acting through the p38 MAPK pathway. Under normal growth conditions, phosphorylation of Ago2 is involved in mediating its localization to PBs. Thus the phosphorylation of Ago2 may represent one mechanism by which cellular signals act to regulate RNAi, therefore co-ordinating gene expression with the cellular environment. Further experiments will determine what other signals and pathways may be involved in mediating Ago protein phosphorylation and examine in more detail how phosphorylation regulates Ago2 function and RNAi.


We thank Dr Thomas Tuschl (Howard Hughes Medical Institute, Laboratory of RNA Molecular Biology, Rockefeller University, New York, NY, U.S.A.) for providing the FLAG–HA–Ago2 plasmid. These studies were supported by a NIDA (National Institute on Drug Abuse) Center (DA 011806) and USDOD (U.S. Department of Defense) Army (W81XWH-07-1-0183) grant to Y. Z., and an American Cancer Society Institutional grant (IRG-73-001-34) to P. R. G.

Abbreviations: Ago, Argonaute; CIP, calf intestinal phosphatase; DMEM, Dulbecco's modified Eagle's medium; GST, glutathione transferase; HA, haemagglutinin; HEK-293T cell, human embryonic kidney cell expressing the large T-antigen of simian virus 40; JNK, c-Jun N-terminal kinase; MAPK, mitogen-activated protein kinase; MAPKAPK2, MAPK-activated protein kinase-2; MEK, MAPK/ERK (extracellular-signal-regulated kinase) kinase; miRNA, microRNA; MS/MS, tandem MS; PB, processing body; RNAi, RNA interference; siRNA, small interfering RNA; WT, wild-type


View Abstract