Transcript-Specific Decapping and Regulated Stability by the Human Dcp2 Decapping Protein

Abstract

mRNA decapping is a critical step in the control of mRNA stability and gene expression and is carried out by the Dcp2 decapping enzyme. Dcp2 is an RNA binding protein that must bind RNA in order to recognize the cap for hydrolysis. We demonstrate that human Dcp2 (hDcp2) preferentially binds to a subset of mRNAs and identify sequences at the 5′ terminus of the mRNA encoding Rrp41, a core subunit component of the RNA exosome, as a specific hDcp2 substrate. A 60-nucleotide element at the 5′ end of Rrp41 mRNA was identified and shown to confer more efficient decapping on a heterologous RNA both in vitro and upon transfection into cells. Moreover, reduction of hDcp2 protein levels in cells resulted in a selective stabilization of the Rrp41 mRNA, confirming it as a downstream target of hDcp2 regulation. These findings demonstrate that hDcp2 can specifically bind to and regulate the stability of a subset of mRNAs, and its intriguing regulation of the 3′-to-5′ exonuclease exosome subunit suggests a potential interplay between 5′-end mRNA decapping and 3′-end mRNA decay.

mRNA decay plays an important role in the control of gene expression and response to regulatory events. In both yeast and mammalian cells, bulk mRNA decay typically initiates with the removal of the 3′ poly(A) tail, followed by degradation of the mRNA in a 5′-to-3′ direction or a 3′-to-5′ direction (42). Degradation from the 3′ end is carried out by the cytoplasmic RNA exosome, which is a multisubunit 3′-to-5′ exoribonuclease complex (21), and the resulting cap structure is hydrolyzed by the scavenger decapping enzyme DcpS (20). In the 5′-to-3′ decay pathway, the monomethyl guanosine (m⁷G) mRNA cap is cleaved first by the Dcp2 decapping enzyme (9, 22, 35, 39), and the monophosphate RNA is degraded progressively by the 5′-to-3′ exoribonuclease Xrn1 (7, 15).

Decapping is a highly regulated process influenced by both positive and negative regulators. In yeast, Dcp1p forms a complex with Dcp2p and is required for optimum decapping activity (31, 32). The Edc1p, Edc2p, and Edc3p proteins, as well as the Dhh1p and Lsm1-7 protein complex, are all reported to stimulate Dcp2p decapping (6). In mammals, an additional protein, Edc4 (also known as Hedls and Ge-1), is a positive effector of Dcp2 decapping by either directly facilitating human Dcp2 (hDcp2) activity or promoting the association of Dcp1a with hDcp2 to possibly further enhance decapping activity in cells (10). In addition, AU-rich elements (ARE), which confer rapid mRNA decay on an mRNA, have also been shown to stimulate decapping in yeast (37) and mammals (10, 11, 23). In addition to these decapping activators, Dcp2 decapping can also be negatively regulated. In yeast, both the eukaryotic initiation factor 4E (eIF4E) cap-binding protein and the poly(A) tail negatively impact decapping (3, 28-30, 41). In mammals, eIF4E can inhibit Dcp2 decapping in vitro (19), and RNAs with synthetic cap structures that bind eIF4E with higher affinity are more stable in vivo (14). The poly(A) tail can also negatively influence decapping, and the poly(A)-binding protein can also directly inhibit decapping in vitro (19). In a recent report, the testis-specific VCX-A protein was identified as a cap-binding protein and an inhibitor of Dcp2 decapping (18).

Dcp2 is an RNA binding protein and can cleave only cap structure that is linked to an RNA moiety (26). Uncapped RNA, but not cap analog, can efficiently inhibit Dcp2 decapping (26, 35, 39). These facts suggest that Dcp2 detects its cap substrate by RNA binding. A typical RNA binding protein usually has a basal level of nonspecific binding to all RNAs but much higher affinity for a subset of RNAs. If this is the case with Dcp2, then Dcp2 could bind RNAs differentially and preferentially regulate a subset of mRNA. In fact, the X29 protein, which is the nuclear decapping enzyme and a NUDIX fold protein like Dcp2, contains substrate specificity. X29 specifically binds U8 snoRNA (12, 33), and in the presence of Mg²⁺, cap hydrolysis is highly specific for U8 snoRNA (12, 25); in contrast, in the presence of Mn²⁺, all RNAs tested were decapped at high efficiency (25). These data suggest that decapping proteins can have a preference for their RNA substrates and raise the intriguing possibility that Dcp2 can differentially associate with and decap specific mRNAs.

In this study, we demonstrate that, similar to other RNA binding proteins, hDcp2 can preferentially bind specific mRNAs, and we identify the 5′ terminus of the mRNA encoding a core subunit of the exosome, Rrp41, as a specific hDcp2 substrate. Moreover, differential binding of hDcp2 to the Rrp41 mRNA can impact the stability of this mRNA, suggesting an interesting interplay between the nucleases involved in targeting the two termini of an mRNA for decay.

MATERIALS AND METHODS

Plasmid constructs.

The plasmid pGEM-Rrp41 was constructed by inserting the Rrp41 cDNA containing a T7 promoter sequence at the 5′ end into the pGEM-T vector (Promega, Madison, WI). The cDNA was obtained by reverse transcription of 293T cell RNA and PCR amplified with primers 1 and 2. pGEM-Stx7 was similarly constructed, but primers 3 and 4 were used to amplify the syntaxin7 (Stx7) cDNA. pGEM-R5′UTR-S, which contains the chimeric sequence of the Rrp41 5′ untranslated region (UTR) and the Stx7 coding region (Stx7^Δ5′UTR), was constructed in two steps. Initially, the Stx7 coding region was amplified with a primer to insert a 5′ BamHI site (primers 12 and 4) and inserted into the pGEM-TEasy vector (Promega, Madison, WI). The Rrp41 5′ UTR was generated with primers 13 and 14 and inserted into the NcoI and BamHI sites of the pGEM- S^Δ5′UTR plasmid. The plasmid containing the Rrp41 5′ UTR terminal 60 nucleotides, 1 to 60 (pGEM-R5′1-60-S), was generated by digesting pGEM-R5′UTR-S with SacII and BamHI and self-ligating it. The plasmids for glutathione S-transferase (GST)-hDcp2 (pGEX6p1-hDcp2) (39) and GST-murine autosomal Deleted in Azoospermia-like protein (mDazl) (pGEX6p1-mDazl) (17) and pcDNA3-Flag-hDcp2 (10) were described previously. pcDNA3-Flag-mDazl was constructed by removing hDcp2 cDNA from the pcDNA3-Flag-hDcp2 plasmid and replacing it with the mDazl coding region. All primers are listed in Table S2 in the supplemental material.

Coimmunopurification of RNAs associated with hDcp2.

pcDNA3-Flag-hDcp2 (16 μg) or pcDNA3-Flag-mDazl (7 μg) plasmid was transfected into 293T cells in 10-cm dishes by using Lipofectamine 2000 (Invitrogen, Carlsbad, CA) following the recommendations of the manufacturer. These concentrations were chosen to yield equivalent levels of expressed Flag-tagged protein in the cells. Twenty-four hours after transfection, the cells were harvested in 1 ml lysis buffer (phosphate-buffered saline [PBS] plus 0.5% Triton X-100) and disrupted by sonication. After a brief (10-min) spin at 10,000 × g, the amount of supernatant used in the subsequent binding steps was adjusted so that an equivalent amount of Flag-tagged protein was used in the binding studies. The supernatant was precleared with 100 μl agarose-immunoglobulin G (mouse) beads for 30 min to remove nonspecific binding to the beads. The precleared supernatant was incubated with 150 μl anti-Flag M2 beads at 4°C for 2 h, with addition of MgOAc₂ and MnCl₂ to a final concentration of 2 mM and 0.5 mM, respectively. The beads were then washed four times with washing buffer (PBS-150 mM NaCl-300 mM urea-0.5% Triton X-100). The RNAs were isolated from the beads by boiling them for 3 min in 300 μl Tris-EDTA/1% sodium dodecyl sulfate and purified with an RNeasy Mini kit (Qiagen, Valencia, CA).

Microarray analysis.

RNA quality was assessed on an RNA chip using an Agilent 2100 Bioanalyzer (Agilent Technologies, Palo Alto, CA), and microarray experiments were carried out at the Transcriptional Profiling Facility, Cancer Institute of New Jersey. RNA samples were labeled using a BioArray HighYield RNA Transcript Labeling Kit (T7) (Enzo Life Sciences, Inc., Farmingdale, NY) and hybridized to a Human Genome U133A 2.0 GeneChip (Affymetrix, Inc., Santa Clara, CA), following the manufacturers’ recommendations. The arrays were then washed and stained with streptavidin-phycoerythrin conjugate (Invitrogen Corp., Carlsbad, CA) using an Affymetrix Fluidics Station 450 and scanned on a GeneChip scanner (Affymetrix, Inc., Santa Clara, CA). The data were processed using the Microarray Suite 5.0 software (Affymetrix, Inc.), and gene annotations were referenced to the Gene Ontology database.

Reverse transcription (RT)-PCR.

RNAs were reverse transcribed with Moloney murine leukemia virus reverse transcriptase (Invitrogen, Carlsbad, CA) with random primers according to the manufacturer's instructions. Gene-specific PCR amplifications (see Fig. 1A) were carried out with the following primers: Spen, primers 24 and 25; Rrp41, primers 22 and 23; Ndufb7, primers 26 and 27; Ndufs8, primers 28 and 29; Psmc3, primers 30 and 31; Ruvb2, primers 32 and 33; Tceb2, primers 34 and 35; Ada, primers 36 and 37; Ap2s1, primers 38 and 39; Gltscr, primers 40 and 41; Edf1, primers 42 and 43; Narf, primers 44 and 45; Hmox2, primers 46 and 47; Ppib, primers 48 and 49; and Stx7, primers 50 and 51. All primers are listed in Table S2 in the supplemental material.

FIG. 1.

Identification of Rrp41 mRNA as a substrate directly bound by hDcp2. (A) Identification of mRNAs specifically associated with hDcp2. 293T cell mRNAs bound by Flag-hDcp2 (lane 3) or Flag-mDazl (lane 2) were isolated, reverse transcribed with random primer, and amplified by gene-specific primers for the indicated mRNAs. An aliquot of RT-PCR products using 293T total RNA to designate the size of the correct band is shown in lane 1, and RT-PCR products from a mock immunoprecipitation (IP) are indicated in lane 4 as a negative control. RT-PCR results that were reproducibly detected to be preferentially bound by Flag-hDcp2 in three independent sets are indicated by “+.” Those that were not reproducible in all three experiments are indicated by “−.” The Stx7 negative control mRNA was associated with mDazl. (B) Rrp41 mRNA is bound by hDcp2 directly. GST pull-down assays were carried out with 4 μg of the indicated fusion protein and ³²P-labeled Rrp41 or Stx7 RNA. RNAs bound to GST-hDcp2 or GST-mDazl were isolated and resolved on a denaturing PAGE. Five percent of each RNA used was included in the input lanes.

GST fusion protein copurification.

The GST-hDcp2 and GST-mDazl fusion proteins were expressed in Escherichia coli BL21 as previously described (17, 39). Four micrograms of GST fusion protein was bound to 50 μl GST beads in a total volume of 400 μl in PBS with protease inhibitor cocktail (Roche, Mannheim, Germany) at 4°C for 1 h, followed by extensive washes in 300 mM NaCl-500 mM urea-0.5% TritonX-100. The washed beads were resuspended in 400 μl of RBB (10 mM Tris-HCl, pH 7.5, 1.5 mM MgCl₂, 150 mM KCl, 0.5 mM dithiothreitol) with RNase inhibitor (Promega, Madison, WI) and incubated with in vitro-transcribed [α-³²P]UTP-labeled RNA at 4°C for 1 h. After washes in RBB-150 mM NaCl-300 mM urea-0.25% Triton X-100, the RNAs were isolated from the beads as previously described (34). Copurified RNA was resolved on a 5% polyacrylamide gel with 7 M urea.

Generation of RNA in vitro.

RNAs were in vitro transcribed from PCR-generated templates that contained T7 or Sp6 promoter sequence at the 5′ end. Cap labeling was carried out as previously described (40). The 5′ 900 nucleotides of Stx7 mRNA (see Fig. 2 to 5) was transcribed with T7 polymerase from a template generated with primers 3 and 5. Rrp41 mRNA truncations (see Fig. 3A) were generated with T7 polymerase from templates generated by the following primer sets: fragment A was generated with primers 1 and 6, fragment B with primers 7 and 2, C with primers 8 and 9, D with primers 1 and 10, and E with primers 11 and 6. Templates for RNAs (see Fig. 5A) were transcribed with T7 polymerase using the same 3′ primer (primer 5) and the following different 5′ primers: 1-110 and 50-110 were made from pGEM-R5′UTR-S with 5′ primers 1 and 15, respectively; 1-60 and 26-60 were made from pGEM-R5′1-60-S with 5′ primers 1 and 16, respectively; and 1-25 was made from pGEM-S^Δ5′UTR with 5′ primer 17. Control RNA (see Fig. S2 in the supplemental material) was generated with primers 18 and 19 from pGEM-S^Δ5′UTR, and RNA+Rrp41 5′UTR60 was generated using primers 18 and 19 from pGEM-R5′1-60-S. The RNA probe used in the electrophoretic mobility shift assays was in vitro transcribed in the presence of [α-³²P]GTP from a template generated by primers 20 and 21.

FIG. 2.

(A) Rrp41 mRNA is preferentially decapped by hDcp2. Decapping assays were carried out using the indicated amounts of bacterially expressed His-hDcp2 with cap-labeled Rrp41 or Stx7 RNA, and the reaction products were resolved by PEI-TLC. The positions of the capped RNA substrate and the m⁷GDP decapping product are indicated on the left. The decapping efficiency of Stx7 RNA was arbitrarily set to 1. The observed increases in decapping of the Rrp41 RNA relative to the Stx7 RNA obtained from three independent experiments are presented in the graph to the right with standard deviations denoted by the error bars. Quantitations obtained with the decapping observed with 80 ng of His-hDcp2 are shown but are similar to those observed with 40 ng protein. (B) Rrp41 mRNA is preferentially decapped by immunoprecipitated Flag-hDcp2. Decapping assays were carried out as for panel A, except the decapping reactions were carried out with the indicated volumes of Flag-bead slurry containing immunopurified Flag-hDcp2. Labeling and quantitations were as described for panel A. (C) Rrp41 mRNA is preferentially decapped by K562 cell extract. Decapping assays were carried out with the indicated concentrations of K562 cell cytoplasmic P50 extract. His-Dcp2 was used in lane 1 as a positive control. Labeling and quantitations were as described for panel A.

FIG. 5.

Identification of minimal hDE. (A) A schematic representation of the various chimeric RNAs that contain different fragments of the Rrp41 5′ UTR upstream of the Stx7 coding region is shown. The numbers represent the corresponding nucleotides of the Rrp41 5′ UTR included in the RNA. (B) The first 60 nucleotides of the Rrp41 5′ UTR are responsible for the stimulation of decapping. Decapping assays were carried out with the indicated chimeric RNAs and with Stx7 as a negative control. Quantification of the decapping efficiency of each RNA relative to that of Stx7 is shown on the right. Labeling was as described in the legend to Fig. 1. (C) An alignment of the two halves of the first 60 nucleotides of the Rrp41 5′ UTR constituting the hDE are shown, with the corresponding nucleotide numbers on the left. (D) Decapping of the individual hDE fragments. Decapping assays were carried out with the Rrp41 mRNA 5′ sequences consisting of 2×-hDE (nucleotides 1 to 60) or the respective hDE halves of this RNA consisting of nucleotides 1 to 25 or 26 to 60 linked to the Stx7 RNA. The assays were carried out with the indicated concentrations of His-hDcp2, and the labeling was as described in the legend to Fig. 1. (E) Specific binding of hDcp2 to the Rrp41 5′ UTR. An electrophoretic mobility shift assay was carried out with 1 μg of His-hDcp2 and uniformly ³²P-labeled uncapped 2×-hDE RNA corresponding to the first 60 nucleotides of the Rrp41 5′ UTR and competed with increasing amounts of the indicated unlabeled RNAs. The control RNA was the 110-nucleotide α-globin mRNA 3′ UTR. The Rrp41 mRNA sequences 1 to 25 and 26 to 60 were transcribed as chimeric RNAs with 55 nucleotides of the Stx7 RNA to maintain comparable size with the other competitors.

FIG. 3.

The hDcp2 target sequence on Rrp41 mRNA is located in the 5′ UTR. (A) A schematic representation of the various Rrp41 mRNA truncations is shown with the 5′ UTR, coding region, and 3′ UTR indicated. (B) The 5′ UTR of Rrp41 mRNA is the stimulatory region for hDcp2 decapping. The RNAs shown in panel A were used as substrates in decapping assays with the indicated amounts of His-hDcp2 protein. Stx7 RNA substrate was used as a negative control to designate a basal level of hDcp2 decapping. (C) Quantitations of the percent decapping observed in panel B are presented. The decapping efficiency of Stx7 RNA was arbitrarily set to 1, and the observed increases in decapping of the Rrp41 RNA fragments relative to the Stx7 RNA are shown. Quantitations from three independent experiments are presented in the graph, with standard deviations denoted by the error bars. Quantitations obtained with the decapping observed with 50 ng of His-hDcp2 are shown.

In vitro RNA-decapping assay.

In vitro RNA-decapping assays were carried out essentially as described previously (38). The indicated amount of extract or bacterially expressed recombinant hDcp2 protein was incubated with cap-labeled RNA (2,000 cpm) at 37°C for 30 min in IVDA-2 buffer (10 mM Tris-HCl, pH 7.5, 100 mM potassium acetate, 2 mM magnesium acetate, 0.5 mM MnCl₂, 2 mM dithiothreitol). When extract was used, a phenol-chloroform (1:1) extraction was carried out following the reaction. An aliquot of each sample was resolved by polyethyleneimine-cellulose thin-layer chromatography (PEI-TLC), developed in 0.45 M (NH₄)₂SO₄, and exposed to a phosphorimager. Quantifications were carried out using a Molecular Dynamics PhosphorImager using ImageQuant-5 software. Percent decapping was determined as the level of m7GDP relative to total RNA used in the reaction.

Electrophoretic mobility shift assays.

Electrophoretic mobility shift assays were carried out with uniformly in vitro-transcribed [α-³²P]GTP-labeled RNA substrate (∼4,000 cpm per reaction). Binding reactions were carried out in RBB (50 mM KCl) with 1 μg of His-hDcp2 protein in a 20-μl total volume containing 1.5 μg yeast RNA and 0.25 μg heparin. Unlabeled competitor RNAs were added at the beginning of the reaction as indicated. Following a 25-min binding reaction at room temperature, the complexes were resolved on a 5% polyacrylamide gel (60:1 acrylamide-bisacrylamide) in 0.5× Tris-borate-EDTA buffer at 8 V/cm and exposed to a phosphorimager.

siRNA transfection and cell-based mRNA decay assays.

293T cells (30% confluent) were cultured in six-well plates, and transfections were carried out with 80 nM hDcp2 small interfering RNA (siRNA) (SMARTpool M00842500; Dharmacon, Chicago, IL) or control siRNA (1027281; Qiagen, Valencia, CA) by using Lipofectamine 2000 reagent (Invitrogen Corp., Carlsbad, CA) following the recommendations of the manufacturer. A second transfection was carried out 24 h after the first.

293T cells were treated with actinomycin D (5 μg/ml) to stop transcription 72 h after the first transfection. Cells were harvested at the indicated times post-actinomycin D addition, and total RNA was isolated with Trizol reagent (Invitrogen, Carlsbad, CA) following the manufacturer's instructions. mRNA levels were quantified from reverse-transcribed cDNA by real-time PCR using SYBR green PCR core reagent (Applied Biosystems, Foster City, CA), and the abundances of Rrp41 and Stx7 mRNAs were quantified using the standard-curve method according to the recommendations of the manufacturer. Values were normalized to the stable U6 RNA. Each mRNA was amplified using the appropriate specific primers, primers 52 and 53 for Rrp41, primers 54 and 55 for Stx7, and primers 56 and 57 for U6. Real-time PCR was carried out with an ABI Prism 7900HT sequence detection system, and the specificities of the amplified PCR products were assessed by a melting-curve analysis after the last cycle by the manufacturer's suggested program.

Electroporation of RNA.

Cap-labeled RNA (2 × 10⁵ cpm) was electroporated into 293T cells (2 × 10⁶) in a total volume of 200 μl Opti-Mem medium in a 4-mm-gap cuvette using a Bio-Rad Genepulser charged at 300 mV and 250 μF. All reagents were kept on ice prior to electroporation. Following discharge, the cells were resuspended in culture medium containing 200 U/ml micrococcal nuclease (USB Corporation, Cleveland, OH) and 1 mM CaCl₂ for 10 min at 37°C to degrade residual RNA remaining outside the cell. Following the incubation, the cells were transferred to culture medium, and an aliquot was removed for time zero; subsequent aliquots were isolated at the indicated time points. RNA was isolated using Trizol reagent (Invitrogen, Carlsbad, CA) and resolved by denaturing polyacrylamide gel electrophoresis.

RESULTS

Identification of mRNAs specifically bound by hDcp2.

The hDcp2 human decapping enzyme is an RNA binding protein responsible for mRNA decapping. Although it has the capacity to decap any mRNA tested, we reasoned that, similar to other RNA binding proteins, it should have both a basal level of nonspecific binding to RNA and high-affinity binding to a subset of mRNA. To begin the identification of RNAs preferentially bound by hDcp2, we isolated and identified mRNAs specifically associated with hDcp2. Flag epitope-tagged hDcp2 (Flag-hDcp2) or an epitope-tagged control RNA binding protein, mDazl (Flag-mDazl), was overexpressed in 293T cells and immunoprecipitated with anti-Flag antibody. Copurifying RNAs were extracted and subjected to microarray analysis using an Affymetrix Human Genome U133A 2.0 GeneChip. Since we were interested in RNAs specifically bound by hDcp2, RNAs that were bound by both proteins were designated nonspecific and not further considered. A stringent cutoff of sixfold-greater binding to hDcp2 than to the mDazl control was used to designate potential mRNAs specifically bound to hDcp2.

Of the 22,277 probes on the chip, 10,133 RNAs with a discernible signal were detected to be bound by either hDcp2, mDazl, or both. Of those, 1,461 were detected to be preferentially bound by hDcp2 relative to mDazl: 710 were twofold, 351 were threefold, 199 were fourfold, 103 were fivefold, and 98 were sixfold or greater. Genes that were immunopurified with hDcp2 at a level sixfold or greater than that detected with mDazl are listed in Table S1 in the supplemental material. Fourteen of the potential specifically hDcp2-bound mRNAs identified by the microarray were further tested to confirm their binding by RT-PCR. The Stx7 mRNA that was specifically bound by mDazl in the microarray screen was used as a negative control. Figure S1 in the supplemental material demonstrates that equivalent amounts of Flag-tagged hDcp2 or mDazl protein were used in the coimmunopurification assays. Approximately 65% of the candidate mRNAs tested (9 out of 14) were confirmed to specifically associate with hDcp2 relative to mDazl by their selective coimmunopurification with Flag-hDcp2 (Fig. 1A). Surprisingly, the top candidate from the microarray data, with a 16.6-fold-greater association with hDcp2, the Spen mRNA, was not reproducibly detected to specifically bind hDcp2. However, the second-highest candidate in the microarray analysis, the Rrp41 mRNA, was reproducibly bound by hDcp2. In addition to Rrp41, mRNAs encoding Ndufb7, Ndufs8, Psmc3, Ruvb2, Gltscr, Edf1, Hmox2, and Ppib were selectively immunopurified by Flag-hDcp2 expressed in 293T cells, while the Tceb2, Ada, Ap2s1, and Narf mRNAs were not reproducibly coimmunnopurified (Fig. 1A). As expected, the Stx7 mRNA was detected only in the mDazl-bound fraction (Fig. 1A). These data show that despite the ubiquitous requirement for hDcp2 in mRNA decapping, hDcp2 can specifically associate with a set of mRNAs and may exhibit selective regulation of mRNA decay.

The above analysis was carried out with mRNA coimmunopurified with Flag-hDcp2 expressed in cells, and therefore, the isolated mRNA could be a consequence of either direct hDcp2 binding to the mRNA or indirect association through a protein-protein interaction network. To test whether the binding to hDcp2 was direct or indirect, bacterially expressed GST-hDcp2 was used to determine which mRNA could copurify with hDcp2 in the absence of additional proteins. Three of the Flag-hDcp2-copurifying RNAs were randomly chosen and tested for the ability to be bound by recombinant hDcp2 protein: a subunit of the exonuclease exosome complex, Rrp41 mRNA (21); the endothelial differentiation-related factor 1 (Edf1) mRNA, which functions as a transcriptional coactivator in endothelial cell differentiation; and the mRNA that encodes a subunit of the mitochondrial NADH-ubiquinone oxidoreductase (complex I), Ndufb7. mRNA encoding the Stx7 protein, which is involved in post-Golgi vesicle-mediated transport, was included as a negative control mRNA. RNAs corresponding to the full-length Rrp41, Edf1, Ndufb7, and Stx7 mRNAs were transcribed in vitro in the presence of [³²P]UTP. The labeled RNAs were incubated with GST-hDcp2 or a control GST-mDazl protein. After extensive washes, copurified RNAs were isolated and resolved on a polyacrylamide gel. We were unable to detect consistent binding of Edf1 and Ndufb7 to GST-Dcp2, indicating that the above-mentioned detected binding was most likely due to indirect association with hDcp2 (data not shown). However, the Rrp41 RNA was reproducibly and specifically bound to GST-hDcp2, while the Stx7 RNA bound only to GST-mDazl (Fig. 1B). These data demonstrate that the binding of hDcp2 to the Rrp41 RNA is direct and does not require additional proteins.

Collectively, the above data demonstrate that hDcp2 has the capacity to specifically bind a subset of mRNA either directly or indirectly. Furthermore, the specificity of hDcp2 binding to the Rrp41 mRNA was intriguing, considering that Rrp41 is an essential component of the exosome complex. As hDcp2 is the critical component of the 5′-to-3′ mRNA decay pathway and the exosome is the critical 3′-to-5′ decay complex, the binding of hDcp2 to Rrp41 mRNA was of particular interest, and the binding of hDcp2 to this substrate RNA was further characterized.

Rrp41 mRNA is preferentially decapped by hDcp2 in vitro.

We next addressed whether the preferential binding of hDcp2 to the Rrp41 mRNA correlated with increased decapping. The decapping efficiency of the hDcp2-bound Rrp41 RNA substrate was compared to that of the Stx7 RNA, which does not appear to be an hDcp2 binding substrate. The RNAs were in vitro transcribed, cap labeled with [α-³²P]GTP, and used in decapping assays with bacterially expressed His-hDcp2. The Rrp41 RNA consisted of the 900-nucleotide complete Rrp41 mRNA containing a 110-nucleotide 5′ UTR, a 740-nucleotide coding region, and a 50-nucleotide 3′ UTR. To maintain consistency of the RNA substrate size, the 5′ 900 nucleotides of the Stx7 RNA were used, which included a 60-nucleotide 5′ UTR, a 780-nucleotide coding region, and a 60-nucleotide 3′ UTR. The cap-labeled RNAs were incubated with a titration of His-hDcp2, and the decapping products were resolved by PEI-TLC developed in 0.45 M (NH₄)₂SO₄. As expected, the decapping efficiencies for both RNAs increased with increasing His-hDcp2 protein used (Fig. 2A). The increase was also an indication that the reactions were carried out within a linear range for His-hDcp2 protein. Interestingly, decapping of the Rrp41 RNA was considerably more efficient than that of the control Stx7 RNA, with an average decapping value 6.5-fold-greater than the decapping observed with Stx7 RNA. This suggests that specific binding of hDcp2 to an RNA leads to enhanced decapping of the RNA.

The preferential decapping of an hDcp2-bound RNA was not restricted to bacterially expressed hDcp2 and was also observed using both Flag-hDcp2 expressed in cells and endogenous hDcp2. Flag-hDcp2 expressed in 293T cells was immunoprecipitated, and the bead-associated protein was used in a decapping assay with Rrp41 or Stx7 RNA. With this assay system, a consistent average twofold increase in the efficiency of decapping Rrp41 RNA was observed (Fig. 2B). Preferential decapping of the Rrp41 RNA was also observed with endogenous hDcp2. Although hDcp2 activity is not readily detected in cell extract due to decapping inhibitors, low levels of decapping could be detected in the P50 cytoplasmic fraction (18). With the level of K562 cell cytoplasmic P50 fraction used, hDcp2-mediated decapping that generated m⁷GDP could be detected with the Rrp41 RNA and at significantly lower levels with Stx7 RNA (Fig. 2C). Decapping of the Rrp41 RNA was approximately 3.5-fold greater than that observed with the Stx7 RNA. Although the increase in decapping observed with the hDcp2 expressed in eukaryotic cells (Fig. 2B and C) was less than that observed with the bacterially expressed hDcp2 (Fig. 2A), this is consistent with the presence of hDcp2 decapping inhibitors in cells (18). Taken together, these data reveal a novel role of hDcp2. hDcp2 itself contains an intrinsic property of distinguishing different RNA substrates, can preferentially associate with and decap specific mRNAs, and does not indiscriminately hydrolyze all mRNAs equally.

Identification of the hDcp2 target sequence in the Rrp41 mRNA.

In order to identify the sequence element within the Rrp41 mRNA that was responsible for the recruitment of hDcp2, different truncations spanning the Rrp41 mRNA were generated (Fig. 3A) and tested for the ability to be decapped by hDcp2 in an in vitro decapping assay. Three fragments corresponding to the first half, second half, and middle third of the Rrp41 mRNA spanning the junction of the two halves were tested. As shown in Fig. 3B, the 5′ half of the RNA was decapped threefold more efficiently than the other two RNAs, which were comparable to the level detected with the Stx7 negative control. Therefore, the decapping-stimulatory region is contained within the 5′ half of the Rrp41 mRNA.

The hDcp2 recruitment region within the 5′ half of Rrp41 mRNA was further narrowed by separately testing decapping of the 110-nucleotide 5′ UTR and the remaining 400-nucleotide coding region. As shown in Fig. 3B, decapping activity comparable to that of the full-length 5′ half of the Rrp41 RNA was exclusively associated with the 5′ UTR (lanes 13 to 15 and 16 to 18), but not the 5′ portion of the coding region (lanes 19 to 21). From these data (summarized in Fig. 3C), we conclude that the 5′ UTR of Rrp41 mRNA is essential for stimulating hDcp2 decapping and most likely contains the hDcp2 binding region. This suggests a model in which hDcp2 directly binds to the 5′ end of an RNA and preferentially decaps it.

The Rrp41 5′ UTR can function on a heterologous RNA to promote decapping.

The above data indicate that the Rrp41 5′ UTR is involved in recruitment of hDcp2. We next asked whether it can also function when placed onto a heterologous RNA. To test this, the 110-nucleotide Rrp41 5′ UTR was used to replace the 5′ UTR of the Stx7 RNA, since this RNA is normally inefficiently decapped (Fig. 2A). The Rrp41-Stx7 chimeric RNA was used as a substrate for in vitro decapping assays with bacterially expressed recombinant His-hDcp2. Figure 4 shows that while the Stx7 RNA with its own 5′ UTR was poorly decapped by hDcp2 (lanes 7 to 9), the decapping efficiency of the chimeric RNA that bore the Rrp41 5′ UTR was enhanced by approximately 10-fold (lanes 4 to 6). Therefore, the Rrp41 5′ UTR can recruit hDcp2 onto a chimeric RNA and appears to be an autonomous element that can confer higher decapping activity on a heterologous RNA.

FIG. 4.

The Rrp41 5′ UTR is an autonomous element conferring higher decapping activity. Chimeric RNA that harbors the Rrp41 5′ UTR at the 5′ end, followed by Stx7 sequences, was used in decapping assays with bacterially expressed His-hDcp2 (lanes 4 to 6). Stx7 RNA with its own 5′ UTR was used as a negative control (lanes 7 to 9), and the Rrp41 RNA was used as a positive control (lanes 1 to 3). Quantitations of the increase in percent decapping relative the Stx7 RNA are shown in the graph to the right. The decapping efficiency of the chimeric RNA was comparable to that of Rrp41 RNA. The labeling and quantitation of three independent experiments was as described in the legend to Fig. 1.

Identification of a minimal hDcp2 binding element.

The fact that the Rrp41-Stx7 chimeric RNA was efficiently decapped enabled us to use this RNA to further narrow down the minimal sequence capable of stimulating decapping. Initially, two chimeric RNAs were generated containing either the first or second half of the Rrp41 5′ UTR fused to the Stx7 RNA (Fig. 5A). The capacity of hDcp2 to decap each RNA was tested and compared to that for the chimeric RNA containing the full-length Rrp41 5′ UTR. As seen in Fig. 5B, the RNA bearing nucleotides 1 to 60 of Rrp41 was decapped as efficiently as the RNA bearing the full-length Rrp41 5′ UTR (compare lanes 8 and 9 to lanes 2 and 3). In contrast, the decapping efficiency of the RNA bearing nucleotides 50 to 110 was significantly reduced relative to that of the Rrp41 5′UTR-containing RNA.

Examination of the 60-nucleotide sequence revealed it to be rich in guanosine and cytosine with an interesting duplication of a 25-nucleotide motif (Fig. 5C). In an attempt to further subdivide the decapping-stimulatory element and to determine whether either of the duplicated sequences could support increased decapping, two additional chimeras were generated and tested for decapping. Nucleotides 1 to 25 or 26 to 60 of the Rrp41 mRNA were linked to Stx7 (Fig. 5A). Interestingly, neither RNA was as efficiently decapped as the chimeric RNA that contained the entire 60-nucleotide Rrp41 mRNA sequence, and both were 50% as active (Fig. 5D), but nevertheless, they contained greater decapping than the Stx7 RNA lacking Rrp41 sequences (lanes 11 and 12). Therefore, it appears that both elements can support recruitment of, and decapping by, hDcp2, with the two sites adjacent to one another conferring greater decapping stimulation.

The above analysis demonstrates that the 60-nucleotide 5′-terminal sequence of the Rrp41 mRNA can be specifically and efficiently decapped by hDcp2. However, whether this was due to the structure of the RNA being more accessible for decapping or to an active selective binding process remained to be determined. To address this question, we tested whether hDcp2 could specifically bind the RNA by electrophoretic mobility shift assays. Uniformly ³²P-labeled uncapped RNA consisting of the cap-proximal 60 nucleotides of the Rrp41 mRNA was incubated with His-hDcp2, and the bound complex was resolved by native gel electrophoresis. As shown in Fig. 5E, an hDcp2-RNA complex could be detected upon addition of hDcp2 (lane 2). The binding was specific, as it was competed by unlabeled self-competitor but not an unrelated control RNA (compare lanes 3 to 6 to lanes 7 to 10). Moreover, RNA containing either GC-rich repeat within the 60-nucleotide fragment was also capable of competing for hDcp2 binding at molar ratios equal to those of the 60-nucleotide fragment (lanes 11 to 14 and 15 to 18). These data demonstrate that a single GC-rich motif is sufficient to specifically bind hDcp2; however, both are required for the maximal stimulation of decapping observed with the Rrp41 mRNA. Considering that each individual GC-rich motif was competent to augment decapping (Fig. 5D) and compete for hDcp2 binding (Fig. 5E), we will refer to each element as the hDcp2 binding and decapping element (hDE). However, it should be noted that the maximal motif capable of conferring optimal enhanced decapping consisted of the larger 60-nucleotide element of the Rrp41 mRNA containing both hDEs (2×-hDE).

Having shown that the 60-nucleotide 2×-hDE fragment promotes decapping within a heterologous RNA when positioned at the RNA 5′ terminus (Fig. 5B), we next asked whether it would work when placed distal of the 5′ cap. As shown in Fig. S2 in the supplemental material, the 2×-hDE failed to stimulate decapping when placed at a position 650 nucleotides downstream of the cap, indicating that proximity to the 5′ end is important.

hDcp2 can specifically regulate Rrp41 mRNA stability in cells.

To address the functional significance of hDcp2 binding to the Rrp41 mRNA, we determined whether the stability of the Rrp41 mRNA is specifically influenced by hDcp2. Our rationale was that since hDcp2 is a component of the mRNA decay machinery, the Rrp41 mRNA should be selectively stabilized in cells containing a reduction of hDcp2 protein relative to the Stx7 mRNA. siRNA directed against the hDcp2 mRNA was used to reduce hDcp2 expression in 293T cells. hDcp2 protein levels were efficiently reduced by 90% following two successive transfections of siRNA at 72 h posttransfection (Fig. 6A). Therefore, this time interval was used in the subsequent experiments to assess the influence of the hDcp2 protein. 293T cells containing a reduction in hDcp2 levels were subjected to actinomycin D to block transcription at the 72-h post-siRNA transfection time point. RNA was isolated from cells transfected with either hDcp2-specific siRNA or control siRNA at increasing time intervals following actinomycin D up to 8 hours. The Rrp41 mRNA level was determined by quantitative real-time RT-PCR, and the values were normalized to the U6 snRNA. The U6 snRNA is uncapped and should not be affected by hDcp2. As expected, the half-life of the Rrp41 mRNA was increased twofold, from 3.5 h in the control siRNA knockdown cells to 7 h in the hDcp2 knockdown cells (Fig. 6B). In contrast, the decay rate for the Stx7 mRNA, which is a poor substrate for hDcp2 in vitro, did not change when hDcp2 was knocked down and was similar to that observed in the control knockdown cells. These data indicate that the Rrp41 mRNA is a true hDcp2 target in cells and validates the above in vitro analysis demonstrating selective binding and decapping of mRNA by hDcp2.

FIG. 6.

hDcp2 regulates Rrp41 mRNA stability in cells. (A) siRNA-directed knockdown of hDcp2 expression in 293T cells. 293T cells were transfected twice during a period of 72 h with a control siRNA or siRNA specific for hDcp2. Endogenous hDcp2 was detected by Western blot analysis using affinity-purified hDcp2 antibody. The level of an unrelated decapping enzyme, DcpS, was monitored as an internal control. Bacterially expressed His-hDcp2 was used in the last lane as a positive control. (B) Stabilization of the Rrp41 mRNA upon hDcp2 knockdown. 293T cells were treated with actinomycin D to block transcription 72 h after the second of two series of siRNA transfections. The levels of Rrp41 or Stx7 mRNA remaining following 0, 2, 4, and 8 h of actinomycin D treatment were determined by quantitative real-time RT-PCR. The values of quadruplicate quantitative real-time RT-PCRs from two sets of independent RNA preparations normalized to the U6 snRNA are shown, with standard deviations represented by the error bars. (C) The Rrp41 5′ UTR 60-nucleotide 2×-hDE can enhance decapping in cells. Cap-labeled and G-tract-tailed Stx7 RNAs (510 nucleotides) containing either their own 5′ UTRs (60 nucleotides) or the 2×-hDE fragment were electroporated into 293T cells. Cells were harvested at the indicated time points, and RNAs were isolated and resolved on a 5% polyacrylamide gel. A schematic representation of the RNA is on the bottom, with the asterisk representing the ³²P and the G representing the G16 tract. Quantitations of three independent experiments normalized relative to the internal control, which was included in the harvest buffer, are presented in the graph, with standard deviations represented by the error bars.

To further clarify the role of the 2×-hDE in the regulation of Rrp41 mRNA stability in cells, we tested whether this element could enhance decapping in cells. Cap-labeled Stx7 RNA (510 nucleotides) was generated with either its own 60-nucleotide 5′ UTR or the 60-nucleotide Rrp41 5′ UTR 2×-hDE and tailed with a poly(G) tract consisting of 16 guanosines to protect the 3′ terminus from 3′-to-5′ exonucleolytic degradation (40). The two cap-labeled RNAs were electroporated into 293T cells, and RNAs remaining at distinct time intervals up to 4 h were resolved by denaturing polyacrylamide gels. Since the RNAs were labeled exclusively at the cap and were protected from 3′ exonucleolytic decay by the G tract, loss of detectable RNA was directly correlated with decapping (40). As shown in Fig. 6C, the Stx7 RNA containing the 2×-hDE at the 5′ terminus degraded significantly faster, with a half-life of 1 h compared to approximately 2.2 h for the Stx7 RNA lacking the 2×-hDE. These data demonstrate that the 60-nucleotide Dcp2 binding and decapping element can enhance mRNA decay by promoting mRNA decapping and can function on a heterologous RNA in cells.

DISCUSSION

In this study, we demonstrated that the hDcp2 decapping enzyme can selectively bind to a subset of mRNA and identified the Rrp41 mRNA as a substrate that is specifically bound by hDcp2. This binding is direct, as bacterially expressed recombinant hDcp2 can copurify the RNA without additional proteins (Fig. 1B). Furthermore, this RNA is preferentially decapped by hDcp2, most likely because of its specific binding by hDcp2. The 5′ UTR of Rrp41 was identified as an autonomous element capable of stimulating hDcp2 decapping. The maximal stimulatory sequence is contained within the first 60 nucleotides of the 5′ UTR and potentially consists of two minimal hDEs. Lastly, we demonstrate that the Rrp41 mRNA is a true target of hDcp2 in cells and that its stability is regulated by hDcp2.

Microarray analysis of hDcp2-bound mRNA revealed RNA profiles that were associated with hDcp2. A small subset of 98 mRNAs was identified as being bound by hDcp2 at a level sixfold or more greater than that bound to a control unrelated testis-specific RNA binding protein, mDazl (see Table S1 in the supplemental material). These RNAs derived from a wide range of functional classes, including cell metabolism, electron transport, transcription factors, and mRNA encoding translation factors. In Dcp2 knockout plants, the levels of a set of mRNAs were significantly elevated; among them, the best represented were the heat shock proteins (13). In a recent report, degradation of HSP70 mRNA in Drosophila required Dcp2 decapping (2). In our microarray, mRNAs of several heat shock proteins, including HSP70, HSP90B1, and DNAJB12 (an Hsp40 homolog), were found to be associated with hDcp2 approximately fivefold more than the control mDazl protein (unpublished observations). Collectively, these data may indicate a conserved function of Dcp2 in regulating mRNA stability.

RNAs can be associated with hDcp2 either directly or indirectly. Given that hDcp2 can form different complexes with various protein factors, it is conceivable that many of the RNAs that specifically associated with hDcp2 did so through indirect binding via another protein(s) that could recruit hDcp2. For example, TTP can bind an ARE-containing mRNA and recruit a decapping complex including hDcp2 (10, 23). Similarly, mRNAs harboring premature termination codons can be recognized by Upf1 and recruit an hDcp2 decapping complex (22). In yeast, the Rps28 protein can bind its own mRNA and engage the Dcp1p/Dcp2p decapping complex via Edc3p (1). Alternatively, the RNA binding property of hDcp2 suggests that it can also bind its substrates directly without additional factors. In this study, we demonstrated that hDcp2 can directly bind the Rrp41 mRNA and specifically decap it, indicating hDcp2 alone has a preference for RNA substrates. This is similar to the observation with the X29 nuclear decapping enzyme, which can bind and decap the U8 snoRNA (12). It is interesting that the specificity of X29 is dependent on cationic ions. In the presence of Mg²⁺, it preferentially decapped the U8 snoRNA, while in the presence of Mn²⁺, it decapped all RNAs tested (25). However, this does not appear to be true for hDcp2. Similar specific decapping of Rrp41 mRNA by hDcp2 was observed regardless of whether Mg²⁺ or Mn²⁺ was used as the divalent cation (Y. Li and M. Kiledjian, unpublished observations).

We previously demonstrated that hDcp2 is an RNA binding protein (26) and proposed a model in which the RNA binding property anchors hDcp2 to a capped RNA substrate adjacent to the cap structure (26). The presence of a decapping-stimulatory element at the 5′ terminus of Rrp41 mRNA (Fig. 3B) is consistent with this model. Furthermore, the fact that the 2×-hDE motif did not function when positioned several hundred nucleotides downstream of the cap (see Fig. S2 in the supplemental material) suggests that its positioning at the 5′ end of an RNA is important for hDcp2 recognition and decapping. It also suggests that hDcp2 does not bind the body of the RNA and simultaneously recognize the cap. The significance of the cap-proximal 5′ region is also important for the yeast Dcp2p protein, which can be inhibited by annealing antisense oligonucleotides to the 5′ end of its RNA substrate (32). Further significance of the 5′-terminal sequence is apparent in Caenorhabditis elegans, in which the presence of a spliced leader sequence at the 5′ end of an RNA dramatically inhibits Dcp2 decapping activity (5). Collectively, these data support direct binding of Dcp2 to the 5′ end of its RNA substrate to hydrolyze the cap structure, and an RNA containing a sequence preferentially bound by hDcp2 at its 5′ end is decapped more efficiently.

Regulatory roles for mRNA 3′ UTRs have been extensively investigated (8, 16). The ARE instability elements that confer rapid mRNA decay are located in the 3′ UTR (4); the target sites of microRNAs are also predominantly contained within the 3′ UTR. Many RNA binding proteins recognize sequences in the 3′ UTRs of their target RNAs (24). In contrast to the 3′ UTR, which often impacts RNA stability, the 5′ UTR is mainly implicated in translational regulation. The internal ribosome entry site elements that direct cap-independent translation usually feature a long and highly structured 5′ UTR, and binding of the iron response element-binding protein to the 5′ UTR iron response element of ferritin is the prototypic mechanism of translational control by trans-acting factors in the 5′ UTR (reviewed in reference 27). Here, we demonstrated an example of the 5′ UTR influencing mRNA stability mediated by hDcp2. The 5′ UTR of Rrp41 mRNA can bind hDcp2 specifically, and the recruitment of hDcp2 facilitates the removal of the mRNA cap structure and thus regulates the mRNA stability. Our studies provide a potential link between the 5′ UTR and the 5′-to-3′ mRNA decay machinery.

In an attempt to determine the minimal sequence in the Rrp41 5′ UTR required for promoting decapping, we showed that the first 60 nucleotides of the 5′ UTR are critical for the stimulation (Fig. 5B). Upon further dissection of this element into two halves containing similar sequences, each half contained 50% of the stimulatory activity of the full-length sequence (Fig. 5D). Considering that the two halves harbor highly similar sequences, it appears each sequence constitutes an hDE. Consistent with this hypothesis, each hDE was competent to efficiently compete for hDcp2 binding to the 60-nucleotide 5′-UTR fragment. Surprisingly, each hDE could compete for hDcp2 binding to the 60-nucleotide 5′-UTR fragment as efficiently as the full-length 2×-hDE 60-nucleotide 5′-UTR fragment (Fig. 5E), despite the fact that each supported only 50% decapping. Why the binding efficiency of a single hDE was similar to that of the 2×-hDE within the 60-nucleotide Rrp41 5′-UTR fragment and yet the latter was decapped more efficiently is not clear. Perhaps the second site contains a high dissociation constant whereby it provides a mechanism to more efficiently load the adjacent site. This model is currently speculative and is under investigation.

Consistent with the data showing that hDcp2 preferentially decapped Rrp41 mRNA in vitro, the stability of Rrp41 mRNA is regulated by hDcp2 in cells. The fact that Rrp41 is an essential component of the exosome complex (21) is intriguing. It suggests that hDcp2, which is a key component of the 5′-to-3′ RNA decay machinery, can also impact the 3′-to-5′ RNA decay machinery. Even though the exosome is a multisubunit complex and whether the change of a single subunit would affect the activity of the whole complex is undetermined, there is evidence that Rrp41 is an important structural subunit and that its level would affect the whole complex, since the knockdown of hRrp41p led to reduced levels of other core exosome subunits (36). Such a regulatory mechanism could provide an interesting regulation feedback between 5′-to-3′ and 3′-to-5′ RNA decay machineries: when the hDcp2 level is low, 5′-to-3′ decay is compromised, but by increasing the Rrp41 mRNA level, 3′-to-5′ decay by the exosome is possibly enhanced. Conversely, when hDcp2 levels are high, 5′-to-3′ decay is robust and 3′-to-5′ decay by the exosome might be reduced. In fact, the mRNAs of Rrp4 and Rrp46 (two other core components of the exosome) were specifically associated with hDcp2 in the microarray assay at a fourfold-higher level than the control, which suggests that hDcp2 could target multiple subunits of the exosome to strengthen the feedback. Our studies now provide a foundation to begin addressing hDcp2-specific regulation of gene expression and potential regulatory pathways that hDcp2 levels could control directly or indirectly.