The RNA binding domain of Pumilio antagonizes poly-adenosine binding protein and accelerates deadenylation

Abstract

PUF proteins are potent repressors that serve important roles in stem cell maintenance, neurological processes, and embryonic development. These functions are driven by PUF protein recognition of specific binding sites within the 3' untranslated regions of target mRNAs. In this study, we investigated mechanisms of repression by the founding PUF, Drosophila Pumilio, and its human orthologs. Here, we evaluated a previously proposed model wherein the Pumilio RNA binding domain (RBD) binds Argonaute, which in turn blocks the translational activity of the eukaryotic elongation factor 1A. Surprisingly, we found that Argonautes are not necessary for repression elicited by Drosophila and human PUFs in vivo. A second model proposed that the RBD of Pumilio represses by recruiting deadenylases to shorten the mRNA's polyadenosine tail. Indeed, the RBD binds to the Pop2 deadenylase and accelerates deadenylation; however, this activity is not crucial for regulation. Rather, we determined that the poly(A) is necessary for repression by the RBD. Our results reveal that poly(A)-dependent repression by the RBD requires the poly(A) binding protein, pAbp. Furthermore, we show that repression by the human PUM2 RBD requires the pAbp ortholog, PABPC1. Pumilio associates with pAbp but does not disrupt binding of pAbp to the mRNA. Taken together, our data support a model wherein the Pumilio RBD antagonizes the ability of pAbp to promote translation. Thus, the conserved function of the PUF RBD is to bind specific mRNAs, antagonize pAbp function, and promote deadenylation.

Keywords: PUF; Pumilio; deadenylation; pAbp; poly(A) tail.

FIGURE 1.

Mutations in the Argonaute and eEF1A binding motifs do not alter PUF repression. (A) Alignment of Argonaute and eEF1A binding regions of PUFs from Caenorhabditis elegans (Ce FBF1), Drosophila melanogaster (Dm Pum), Homo sapiens (Hs PUM1 and PUM2), and Saccharomyces cerevisiae (Sc PUF5), reported by Friend et al. (2012). The conserved Argonaute and eEF1A binding amino acids are indicated at the top, and regions of complete and near complete conservation are highlighted in dark and light gray, respectively. RNA recognition motifs of each PUF repeat are also labeled. (B) Schematic of reporters and PUM constructs used in HEK293 cell experiments. Target Renilla luciferase reporter is labeled with UGG* to denote altered Pum response elements (PREs) in the 3′ untranslated region. A Firefly luciferase gene (FFLuc) served as an internal control for normalization. Human PUM1 and PUM2 were expressed as Halotag (HT) fusions and are marked with * in their RBDs to signify their altered specificity in PUF repeat 6 (R6as) which allows binding to UGG PREs, as previously described (Van Etten et al. 2012). Dark gray rectangles indicate regions of high conservation with Drosophila Pumilio, while light gray ovals represent the N-terminal repression domains (Weidmann and Goldstrohm 2012). Halotag alone served as the negative control. (C) Percent repression in human HEK293 cells of the RnLuc 3xPRE UGG reporter by human PUM1 R6as and PUM2 R6as test proteins. Note that the RnLuc 3xPRE UGG reporter is not regulated by endogenous PUM1 or PUM2 (Van Etten et al. 2012). The amino acid position of Argonaute-binding defective mutant (PUM1 T874E or PUM2 T752E) and eEF1A-binding defective mutant (PUM1 F990R and PUM2 F866R) is indicated at the top. Expression of each test protein was verified by fluorescent detection of TMR-labeled Halotag fusions. (D) Representation of constructs used in D.mel-2 cell experiments. The Renilla reporter 3′ UTR contains three wild-type PREs. Pumilio (Pum) constructs include full length (FL) and the RNA binding domain (RBD). Dark gray boxes mark regions of increased conservation, and light gray ovals denote N-terminal repression domains. The X in negative control Pum mutR7 constructs represents mutations in the RNA recognition amino acids of the seventh PUF repeat which block RNA binding and repression (Weidmann and Goldstrohm 2012). (E) Percent repression of RnLuc 3xPRE reporter by wild-type (WT), Argonaute-binding defective (T1137E), and eEF1A binding-defective (F1251R) Pum FL and Pum RBD test constructs in D.mel-2 cells. Western blot detection of test proteins using anti-V5 antibody is shown at the bottom. In all graphs, mean values derived from four replicates are plotted along with the standard error.

FIGURE 2.

Argonaute associates with human and Drosophila PUFs. (A) Human PUM2 coimmunoprecipitates with Argonaute 1 (AGO1) from RNase-treated HEK293 cell extracts. Western blotting detected FLAG-tagged AGO1 in the inputs and eluates from anti-FLAG immunoprecipitations. Halotag (HT) or Halotag fused to wild-type or mutant (T752E) PUM2 or CNOT6L were detected by covalent labeling of Halotag with fluorescent TMR ligand. As a negative control, FLAG immunoprecipitations were performed from extracts that expressed Halotag prey proteins but not FLAG-AGO1. (B) Drosophila Argonaute 2 (Ago2) associates with the RNA binding domain (RBD) of Pumilio. Halotag alone or Halotag-Pum RBD fusions with V5 epitope tags were coexpressed in D.mel-2 cells with FLAG-tagged Ago2 or, as a negative control, myc-tagged Lsm11. Halotag pulldowns were performed from cell extracts, and bound proteins were eluted via TEV protease cleavage. Western blots of input cell extracts and Halotag eluates were probed with V5 antibody to detect wild-type or mutant (T1137E) Pum RBD. FLAG antibody Western blots detected Ago2, and a myc antibody Western blot detected Lsm11.

FIGURE 3.

The repression and Argonaute binding activities of the Pumilio RNA binding domain can be separated. (A) Diagrams of the tethered function reporter and Pum RBD constructs are shown. The RnLuc MS2 reporter contains two binding sites recognized by the MS2 RNA binding protein (2xMS2 BS) in its 3′ untranslated region. The RBD and depicted truncations were fused to the MS2 RNA binding protein. Amino acid numbers where truncations and PUF repeat deletions were made are indicated. (B) Western blots of anti-FLAG immunoprecipitations (eluates) of FLAG-tagged Ago2 from RNase-treated D.mel-2 cell extracts (Inputs). Pum RBD constructs were detected by anti-V5 Western blot. The negative control Lsm11 was coexpressed and detected by anti-myc Western blotting. (C) Tethered function reporter assays measured repression by Pum RBD constructs in D.mel-2 cells. Percent repression, calculated relative to MS2 negative control, is graphed. Mean values derived from four replicates are plotted along with the standard error.

FIGURE 4.

Depletion of Argonaute proteins does not hinder PUF repression in vivo. (A) Efficient depletion of Argonautes using dsRNAs to Ago1 or Ago2 was demonstrated via knockdown of FLAG-tagged proteins in D.mel-2 cells. Western blots were performed using anti-FLAG antibody and, as a loading control, anti-Actin. Nontargeting control (NTC)-treated extracts served as a control. Four times the amount of Ago-depleted extracts were loaded to highlight knockdown efficiency. (B) Percent repression of RnLuc 3xPRE by Dm Pum FL or Dm Pum RBD was measured in D.mel-2 cells. RNAi was performed using dsRNAs to Ago1, Ago2, or both Ago1 and Ago2. Negative control RNAi was performed using the NTC dsRNA. In each condition, percent repression by Pum FL or RBD was calculated relative to the respective RNA binding-defective mutant controls, Pum FL mutR7 or RBD mutR7. RNAi depletion of endogenous Argonaute mRNAs was verified via qRT-PCR, and the percent depletions relative to the NTC are indicated at the bottom of the panel. (C) Percent repression of RnLuc 3xPRE by human PUM1 was measured in D.mel-2 cells depleted of Ago1, Ago2, or both by RNAi. Repression by PUM1 was calculated relative to the negative control, empty expression vector. As in B, RNAi depletion of endogenous Argonaute mRNAs was verified by qRT-PCR. (D) Depletion of Pum, but not Ago1 and Ago2, alleviates repression of an mRNA with the Hunchback (Hb) 3′ untranslated region. Fold change in expression of the RnLuc reporter with the Hb 3′ UTR reporter was measured in D.mel-2 cells treated with NTC dsRNA, dsRNA targeting endogenous Pum, or dsRNAs targeting Ago1 and Ago2. Pum Response Elements 1 and 2 (PRE1 and PRE2) are binding sites for Pumilio and are also known as Nanos Response Elements. Reporter expression was normalized to a Firefly luciferase control and fold change was calculated relative to the NTC sample. In all experiments, mean values derived from four replicates are plotted with the standard error.

FIGURE 5.

The RBD of Pumilio enhances deadenylation dependent on the Pop2 and Ccr4 deadenylases. (A) (Top) Denaturing polyacrylamide Northern blotting of RNA prepared from D.mel-2 cells over a time course following addition of actinomycin D. Poly(A) shortening of the RnLuc 3xPRE mRNA was induced by wild-type Pum RBD or, as a negative control, the RNA binding-defective RBD mutR7. To resolve the poly(A) tail, the RnLuc 3xPRE mRNA was cleaved using a DNA oligo that is complementary to RnLuc and RNase H. A deoxy-thymidine oligo (dT) was also added to one reaction to remove the poly(A) tail (A₀). Northern blot was probed to detect the RnLuc 3xPRE 3′ cleavage product and the internal control 7SL, a stable noncoding RNA. Length of poly(A) tails are marked on the left. (Bottom) Phosphorimager quantification of poly(A) length distribution for each lane from the Northern blot. (B) Western blotting of myc-tagged Pop2 and Ccr4 depletion by RNAi in D.mel-2 cells. Four times the amount of Pop2- and Ccr4-depleted extract was loaded relative to the NTC extracts to highlight knockdown efficiency. Myc antibody was used for Pop2 and Ccr4 detection. Western blot of actin served as a loading standard. A nonspecific band was also detected by the myc antibody in all cell extracts (*). (C) Denaturing polyacrylamide Northern blotting of RnLuc 3xPRE from D.mel-2 cells expressing Pum RBD and treated with either nontargeting control (NTC) dsRNA or both Pop2 and Ccr4 dsRNAs. (D) Western blotting of Halotag pulldown assay demonstrating association of Pop2 with Pumilio RBD. V5-tagged Halotag (HT-V5) alone or Halotag-Pum RBD (HT-Pum RBD-V5) fusions were coexpressed with myc-tagged Pop2. Cell extracts (Inputs) were treated with RNases. HaloLink-bound complexes were eluted with TEV protease (Eluates). Blots were probed with V5 antibody to detect Pum RBD or with myc antibody to detect Pop2. An actin Western blot served as a negative control.

FIGURE 6.

A poly(A) tail is necessary for Pum RBD-mediated repression. (A) Diagrams represent Renilla luciferase (RnLuc) reporters with polyadenylated or nonadenylated 3′ ends. The RnLuc 3xPRE pA reporter contains three PUF response elements (PREs) and a poly(A) tail generated by an efficient cleavage/polyadenylation element. The RnLuc 3xPRE HSL reporter has a nonadenylated 3′ end generated by histone stem–loop (HSL) processing elements. RnLuc 3xPRE A20 HSL has an internal polyadenosine tract of 20 nt (A20) between the PREs and HSL. (B) Percent repression of the RnLuc 3xPRE reporters by full-length Pum FL and Pum RBD in D.mel-2 cells. Percent repression by Pum FL and RBD was calculated relative to Pum FL mutR7 and RBD mutR7, respectively. (C) RnLuc reporters with two MS2 protein binding sites were used for tethered function assays, including polyadenylated RnLuc MS2 pA and nonadenylated RnLuc MS2 HSL. (D) Percent repression of each MS2 reporter by MS2 protein fusions of Pum FL or Pum RBD. Percent repression was calculated relative to the negative control MS2 protein. All graphs show mean values with the standard error from four replicate samples.

FIGURE 7.

Poly(A) binding protein is necessary for repression activity of the PUF RBD. (A) In the top panel, Western blotting demonstrated efficient RNAi depletion of myc-tagged pAbp from D.mel-2 cells. Four times the amount of pAbp-depleted extract was loaded relative to the nontargeting control (NTC)-treated extracts. Myc antibody was used to detect myc-pAbp, while an actin antibody was used to illustrate differences in loading. At the bottom, a graph of Renilla luciferase reporter activity, in relative light units (RLU), is plotted as measured in nontargeting control- or pAbp dsRNA-treated cells. Reporters included polyadenylated RnLuc (pA) and RnLuc HSL, with a nonadenylated histone stem–loop. (B) Percent repression of RnLuc 3xPRE A20 HSL reporter by the Pum RBD in D.mel-2 cells treated with dsRNA to pAbp or NTC dsRNA. Percent repression was calculated relative to the RBD mutR7 negative control. (C) Percent repression of RnLuc 3xPRE pA reporter by Pum FL and Pum RBD in D.mel-2 cells treated with NTC, Pop2 and Ccr4, or pAbp dsRNAs. Percent repression was calculated relative to the negative control Pum FL mutR7 and RBD mutR7. (D) Percent repression of RnLuc MS2 pA reporter by MS2 alone, a MS2-Pum FL fusion, and a MS2-Pum RBD fusion in D.mel-2 cells treated with NTC or pAbp dsRNAs. Percent repression by tethered constructs was calculated relative to MS2 alone. (E) Western blotting, using anti-Halotag antibody, demonstrating efficient RNAi depletion of human PABPC1, fused to Halotag (HT-PABPC1), in HEK293 cells that coexpressed an internal control Halotag protein. (F) Percent repression of RnLuc 3xPRE UGG reporter by human PUM2 RBD R6as, with altered RNA binding specificity, in HEK293 cells treated with either nontargeting control or PABPC1 siRNAs. Percent repression of the RnLuc 3xPRE UGG reporter was calculated relative to the negative control reporter, RnLuc 3xPREmt, wherein the PREs are inactivated by mutations. For each experiment, Western blotting confirmed expression of the PUF test constructs under each RNAi condition. All graphs represent mean values with standard errors calculated from four replicate samples, with the exception of panel F, which used three replicate samples.

FIGURE 8.

The Pumilio RNA binding domain associates with pAbp. V5-tagged Halotag (HT) alone or Halotag-Pum RBD fusions (HT-Pum RBD-V5) were coexpressed in D.mel-2 cells with myc-tagged pAbp or, as a negative control, myc-tagged Lsm11. Cell extracts were treated with RNases, and Halotag fusions were purified over a HaloLink resin. Bound proteins were eluted with TEV protease. Western blots were probed with V5 antibody for detection of Pum RBD and myc antibody for detection of prey proteins, Lsm11 or pAbp.

FIGURE 9.

The Pumilio RNA binding domain does not displace pAbp from a target mRNA. (A) Control experiments demonstrating selective coimmunoprecipitation of polyadenylated mRNA with pAbp from D.mel-2 cells. Western blotting confirmed FLAG-tagged pAbp expression in inputs and in purified eluates. Northern blotting detected Renilla luciferase (RnLuc) reporter mRNAs in input and eluates, including RnLuc with a poly(A) tail (pA), a histone stem–loop (HSL), or an internal 20-adenosine tract preceding HSL (A20 HSL). Cell extracts were prepared from cells transfected with an FLAG-pAbp expression vector or from mock transfected cells. Renilla mRNA measured in each eluate was quantified as a percentage of the total RNA present in the input cell extracts. From these values, fold enrichment of each RNA in pAbp eluates was calculated relative to the corresponding mock transfection. In this way, values were normalized to the amount of reporter RNA present in the input samples. (B) FLAG-pAbp immunoprecipitations were performed from cells expressing either wild-type (WT) or mutR7 Pum RBD with the RnLuc 3xPRE pA reporter, as diagrammed in Figure 6A. Three replicates were performed for each sample. Expression and enrichment of pAbp in eluates was verified with Western blotting using a FLAG antibody. Northern blotting was performed to detect RnLuc 3xPRE pA RNAs in the input cell extracts and eluates. (C) The same experiment as in panel B except that a RnLuc 3xPRE A20 HSL reporter was used. (D) Verification of RBD repression using triplicate samples from panels B and C. In each case, Western blotting confirmed expression of Pum RBD test constructs. The graph represents mean values with standard errors calculated from three separate transfections with four replicates each. (E) Fold enrichment of RnLuc 3xPRE pA or RnLuc 3xPRE A20 HSL by pAbp with coexpressed wild-type or mutR7 Pum RBD as quantified from the experiments in panels B and C. The graph represents mean values with standard errors calculated from three replicate samples.