The Apc5 Subunit of the Anaphase-Promoting Complex/Cyclosome Interacts with Poly(A) Binding Protein and Represses Internal Ribosome Entry Site-Mediated Translation

Abstract

The anaphase-promoting complex/cyclosome (APC/C) is a multisubunit ubiquitin ligase that mediates the proteolysis of cell cycle proteins in mitosis and G₁. We used a yeast three-hybrid screen to identify proteins that interact with the internal ribosome entry site (IRES) of platelet-derived growth factor 2 mRNA. Surprisingly, this screen identified Apc5, although it does not harbor a classical RNA binding domain. We found that Apc5 binds the poly(A) binding protein (PABP), which directly binds the IRES element. PABP was found to enhance IRES-mediated translation, whereas Apc5 overexpression counteracted this effect. In addition to its association with the APC/C complex, Apc5 binds much heavier complexes and cosediments with the ribosomal fraction. In contrast to Apc3, which is associated only with the APC/C and remains intact during differentiation, Apc5 is degraded upon megakaryocytic differentiation in correlation with IRES activation. Expression of Apc5 in differentiated cells abolished IRES activation. This is the first report implying an additional role for an APC/C subunit, apart from its being part of the APC/C complex.

Unlike viral internal ribosome entry sites (IRESs), which are usually robust and efficient, the cellular IRESs are often weak and their requirements for IRES trans-acting factors (ITAFs) are more stringent. This allows the cellular IRES elements to serve as modulators of translation in response to specific physiological signals. An active IRES musters the translation machinery to secondary/tertiary structures within the 5′ untranslated region (UTR) upstream of the initiator AUG codon. The proteins regulating this recruitment are largely unknown. We sought to identify these ITAFs by using a three-hybrid screen that selects for RNA-protein interactions in the nucleus of a living yeast cell by supporting the formation of a stable complex with a specific RNA bait (28). We resorted to a human HeLa cell cDNA expression library and a portion of the IRES element of human platelet-derived growth factor 2 (PDGF2) mRNA as the RNA bait. One of the cDNA fragments selected by the screen turned out to encode the 361-amino-acid-long carboxy-terminal part of the 755-amino-acid-long human Apc5 protein (37).

The anaphase-promoting complex/cyclosome (APC/C) is the ubiquitin ligase complex that mediates the degradation of at least 15 different proteins during mitosis and G₁. APC/C degradation substrates include mitotic kinases such as cyclins A and B, plk1, aurora and nek kinases, the metaphase inhibitor securin, the Xkid motor protein, cdc6 and geminin (which are involved in the regulation of DNA synthesis), the product of the APC/C regulator gene fzy, and several other proteins (15). The mammalian APC/C comprises 12 subunits (27) and is regulated by interaction with additional proteins and by phosphorylation of several of its subunits. The large size and complex structure of the APC/C are puzzling, especially in view of the fact that the Apc11 subunit is capable of mediating ubiquitination on its own (12, 21). It is possible that several of its subunits act to connect the APC/C with regulatory pathways that control cell cycle progression under normal conditions and in response to signals, albeit the specific function of most of the subunits is unknown. There has been speculation regarding the role of some of the subunits (e.g., Apc1, Apc2, Apc3, Apc6, and Apc11), based on weak homologies to other proteins. More specifically, the Apc2 and Apc11 subunits are related to components of SCF, another ubiquitin ligase complex (27), Apc1 shares a structural motif with the two large subunits of the 19S cap complex of the 26S proteasome (22). Apc5 does not have sequence similarity to any protein of known function.

Although the Apc5 protein does not contain any classical RNA-binding motif, it was tempting to pursue the possible functional connection between this cell cycle-related protein and IRES function. This was particularly intriguing in view of the fact that the G₂/M cell cycle phase was shown to be specifically permissive to the function of several IRES elements, such as those of hepatitis C virus (16), ODC and c-myc (29), and the p58^PITSLRE cyclin-dependent kinase (5). Moreover, a certain time window during the course of differentiation also proved to be specifically more permissive for IRES function (3, 10, 31). Apc5 is essential to viability in budding yeast (38) and in Drosophila (1), but its function is not clear. We studied the human Apc5 in logarithmically growing and differentiated human K562 cells and verified its effect on IRES-mediated translation.

We show that Apc5 is degraded during megakaryocytic differentiation, in correlation with IRES activation. Apc5 overexpression inhibits differentiation-induced IRES activation. Apart from the primary nuclear location of Apc5 and its association with the APC/C, it is also associated with heavier complexes and with the ribosomal salt wash (RSW) fraction. This is in contrast to Apc3, which is tightly bound to the APC/C, does not take part in other complexes, and is maintained at a constant level during differentiation. In addition, we show that Apc5 binds to poly(A) binding protein (PABP), that PABP enhances PDGF2 IRES activity, and that Apc5 interferes with PABP-related IRES activation. This is the first report implying an additional role for an APC/C subunit, apart from forming part of the APC/C E3 complex.

MATERIALS AND METHODS

Plasmids.

pGAD-GH#134 containing residues 395 to 755 of human Apc5 fused to the GAL4 activation domain, was selected from the previously described HeLa cDNA library (14). The SmaI-XhoI Apc5 fragment of pGAD-GH#134 was fused to GAL4 DNA binding domain by its insertion into the SmaI-SalI sites of pGBT9 (Clontech) to create pGBT-hApc5(395-755). The complete coding region of human Apc5 was generated by reverse transcription-PCR (RT-PCR) using oligonucleotides 5′-GGGGATCCGCTAGCGCCATGGCCAGCGTCC-3′ and 5′-CCGCTCGAGGGGATGTCCTCTAGAG-3′ and the Marathon kit (from human testis; Clontech) as a template. This PCR product was digested with BamHI-XhoI and inserted into BamHI-XhoI sites of pcDNA3 (Invitrogen) to generate pcDNA3-Apc5 and into BamHI and XhoI sites of pACTII (Clontech) to generate pACTII-Apc5. The same PCR product was also digested with NheI-XhoI and inserted into the NheI-XhoI sites of pET28b (Novagen) to generate pET28b-Apc5. The PCR product generated with oligonucleotides 5′-CCGGATCCGAATGGCCAGCGTCCAGGAG-3′ and 5′-GGCTCGAGTGCCGTCTTCCCAGCAAAAG-3′ was digested with BamHI-XhoI and cloned into BamHI-XhoI sites of pACTII to generate pACTII-Apc5Δ2 (N-terminal part, residues 1 to 395). The SmaI-XhoI fragment of pGAD-DH#134 was inserted into the corresponding sites of pACTII to construct pACTII-Apc5Δ1 (C-terminal part, residues 395 to 755). pFlag-Apc5, containing the T7 promoter-driven Apc5 coding region fused to Flag tag at its N terminus, was a kind gift from C. Hoog. pACTII-IRP, expressing the GAL4 activation domain fused to the iron-responsive element binding protein (IRP), was kindly provided by M. Wickens (32). pACTII-hPABP, harboring the human PABP1 fused to GAL4 activation domain, was constructed by generating a PCR product using oligonucleotides 5′-GGGGATCCAGATGAACCCCAGTGCC-3′ and 5′-GGGGATCCTTCGGTGAAGCACAAG-3′, with pGEX2T-PABP (7) as template. The PCR product was digested with BamHI and inserted into the BamHI site of pACTII (Clontech). To generate the different fragments of yeast Pab1 fused to the GAL4 activation domain, the following oligonucleotide pairs were used for PCR, with pKB526 (kindly provided by D. Kornitzer) as a template: 5′-GGGAATTCCCGCATATGGCTGATATTACTG-3′ and 5′-GGGGATCCTTGGAATTGTTCGTCAGTAG-3′ for residues 1 to 237, 5′-GGGGAATTCCCGCATATGCTGTTGAACGGTC-3′ and 5′-GGGGATCCCCATATGGAGCAAATTCTTC-3′ for residues 189 to 346, and 5′-GGGGAATTCCCGCATATGGTACTATCACTTCTG-3′ and 5′-GGGGATCCTAAGCTTGCTCAGTTTGTTGTTCTTG-3′ for residues 345 to 577. The PCR products were digested with BamHI-EcoRI and inserted into the BamHI-EcoRI sites of pGAD424 (Clontech) to generate pGAD424-yPab1(1-237), pGAD424-yPap1(189-346), and pGAD424-yPab1(345-577). pcDNA3-hPABP was constructed by insertion of a BamHI fragment containing the coding region of human PABP into the BamHI site of pcDNA3.

Three-hybrid screen and three- and two-hybrid assays.

The system described by Putz et al. (28) was employed for the three-hybrid screen and analyses. For the screen, 1,000 μg of the cDNA library from human HeLa cells fused to GAL4 activation domain (14) was used to cotransfect the CG1945 yeast strain (Clontech) together with 500 μg of pDBRevM10-B (19). A total of 5 × 10⁵ transformants were allowed to grow for 10 days at 30°C on histidine-lacking plates supplemented with 0.5 mM 3-aminotriazole (3AT; Sigma). Plasmids were rescued from each of the 230 surviving colonies and used for retransformation together with either pDBRevM10-A or pDBRevM10 (19) harboring a different or no RNA bait, respectively. Only clones that exhibited RNA-dependent and RNA-specific growth were further characterized. Double transformants forming stable three- or two-hybrid complexes were selected for HIS3 expression by dropping 10 μl of a liquid culture grown to an optical density at 600 nm of 0.1 on plates lacking tryptophan, leucine, and histidine and containing 0.5 mM 3-AT. Cells were allowed to grow for 5 or 7 days at 30°C for the two- or three-hybrid analysis, respectively. LacZ expression with the Y190 yeast strain was determined as described before (19).

In vitro transcription and translation.

pBS-PDGF-LUC and pBS-LUC (3), harboring the firefly luciferase coding region under the T7 promoter with or without the complete 5′ UTR of PDGF2, respectively, were linearized by SacI and transcribed using the MEGAscript kit (Ambion). Capped RNA was prepared using the Ambion mMESSAGE mMACHINE. Poly(A) tail was added using the Ambion poly(A) tailing kit. Krebs-2 cell translation extracts were prepared as described by Svitkin et al. (35). PABP depletion was performed using glutathione S-transferase (GST)-Paip2 prepared from pGEX-6P-2-Paip2 (17) and glutathione-Sepharose 4B beads (Amersham Pharmacia Biotech) according to the method described by Svitkin and Sonenberg (36). The reactions of translation reaction mixtures (14 μl) containing 20 nM each amino acid were programmed with 50 to 100 ng of RNA and incubated at 37°C for 50 to 90 min followed by measurement of firefly luciferase activity with the Promega firefly luciferase assay system and a Turner TD-20e luminometer.

Cells, differentiation, infections and transfections, and luciferase assays.

Human K562 cells weregrown and induced for megakaryocytic differentiation by 48 h of treatment with 5 nM 12-O-tetradecanoylphorbol-13-acetate (TPA) as described in reference 31. HEK293 cells were grown in Dulbecco's modified Eagle's medium supplemented with 50-U/ml penicillin, 50-μg/ml streptomycin, and 10% fetal calf serum (Biological Industries, Israel). Infection with vTF7-3 followed by transfection using Lipofectine (Invitrogen) was performed as described in reference 8. For overexpression, a multiplicity of infection of 5 PFU per cell was used followed by transfection of 6 μg of plasmid DNA per well of a six-well plate. For the effects of Apc5 and/or PABP on cap-dependent and IRES-mediated translation, a multiplicity of infection of 0.05 PFU/cell was used. Renilla and firefly luciferase activities were determined using a Promega dual-luciferase reporter assay system and a Turner TD-20e luminometer.

Protein separation techniques. (i) Gel filtration chromatography.

Logarithmically growing K562 cells (5 × 10⁷) were lysed with 500 μl of 1% TritonX-100-1% deoxycholate (DOC) buffer (1% Triton X-100, 1% DOC, 20 mM Tris [pH 7.5], 50 mM KCl, 10 mM MgCl₂, 10 mM NaF, 50 mM β-glycerophosphate, 1 mM dithiothreitol [DTT], protease inhibitor cocktail) or with 500 μl of 0.2% Triton X-100-0.2% DOC buffer (also containing 20 mM Tris [pH 7.5], 100 mM NaCl, 10 mM NaF, 50 mM β-glycerophosphate, 1 mM DTT, protease inhibitor cocktail, and 10% glycerol) followed by Dounce homogenization. Five milligrams of total soluble protein was fractionated at 4°C through a Superose 6HR column (Amersham Pharmacia Biotech) equilibrated with a mixture of 20 mM Tris (pH 8.0), 150 mM NaCl, 10 mM MgCl₂, 5% glycerol, and a cocktail of protease inhibitors (leupeptin, aprotinin, pepstatin, and phenylmethylsulfonyl fluoride [PMSF]) at a flow rate of 0.3 ml/min. Fractions of 0.5 ml each were collected and concentrated by ethanol precipitation.

(ii) Density sucrose gradients.

For polysome analysis, we followed the protocol described in reference 23, with modifications. K562 cells were incubated in the presence of 90-μg/ml cycloheximide for 10 min prior to harvest. The washed cell pellet was resuspended in 300 μl of LBA buffer (20 mM Tris [pH 7.5], 50 mM KCl, 10 mM MgCl₂, 10 mM NaF, protease inhibitor cocktail without EDTA [Roche], 50 mM β-glycerophosphate, 1 mM DTT, 100 μM PMSF, 0.5-μg/ml aprotinin, 50-μg/ml pepstatin A, 0.5-μg/ml leupeptin, 50-μg/ml cycloheximide). Forty microliters of LBB buffer (LBA buffer containing 10% Triton X-100 and 10% DOC) was added followed by a short spin and addition of heparin to the supernatant to final concentration of 100 μg/ml. Twenty-five A₂₆₀ units was layered on a 5 to 45% sucrose gradient prepared in LBA buffer and centrifuged for 2.5 h at 39,000 rpm in a Beckman SW41 rotor. The positions of ribosomal species were determined by scanning the gradient at A₂₆₀ with Uvicord SII (Amersham Pharmacia Biotech). Fractions of 0.6 ml were collected, mixed with 1 ml of cold ethanol, and incubated overnight at −20°C followed by spinning. The dried pellet was resuspended in 80 μl of 8 M urea, and 25 μl was used for immunoblot analysis.

Preparation of cell extracts and immunodetection.

For immunoblot analysis, proteins were extracted from K562 cells by using lysis buffer containing 25 mM KOH/HEPES (pH 7.5), 1% Triton X-100, 100 mM KCl, 10 mM β-glycerophosphate, 50 mM NaF with freshly added 1 mM DTT, 2 μM okadaic acid, 1 mM vanadate, and Roche Complete protease inhibitor cocktail. For immunoprecipitation of the APC/C complex, proteins from K562 cells were extracted using buffer containing 0.1% NP-40, 250 mM NaCl, 50 mM Tris-HCl (pH 8.0), 20 mM EGTA, 50 mM NaF, and protease inhibitor cocktail (Sigma) followed by incubation with anti-Apc3 (AF3) antibodies covalently coupled to protein A Affiprep beads as described in reference 39. For coimmunoprecipitation of Apc5 and PABP, HEK293 cells expressing pFlag-Apc5 and pcDNA3-PABP were used. Cell pellets were incubated for 15 min on ice in hypotonic buffer containing 10 mM HEPES (pH 7.3), 1.5 mM MgCl₂, 10 mM KCl, 0.5 mM DTT, 5% glycerol, Roche Complete protease inhibitor cocktail, 0.2 mM PMSF, and 2-μg/ml pepstatin. Proteins were extracted by a Dounce homogenizer (20 strokes) followed by a 5-min spin at 10,000 × g. The supernatant was used for overnight incubation with anti-Flag or anti-PABP antibodies at 4°C. Protein A-Sepharose or protein G-agarose beads, respectively, were added for a further incubation of 1 h, followed by three washes with phosphate-buffered saline and analysis by sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS-PAGE) (8% polyacrylamide).

Antibodies.

Human Apc5 cDNA was cloned into pET-28b vector (Novagene) and expressed in Escherichia coli. The recombinant protein, which was mainly nonsoluble, was separated by SDS-PAGE (8% polyacrylamide) and injected as a crushed gel slice into rabbits for the generation of crude antiserum against Apc5. Crude antiserum against eIF3 was a kind gift from J. Hershey, anti-PABP was a kind gift from N. Sonenberg, 4F4 (monoclonal anti-hnRNP-C) was a kind gift from G. Dreyfuss, and anti-Apc3/CDC27 (AF3) used for immunoprecipitation was a kind gift from J. Gannon. Anti-Apc3/CDC27 used for immunoblotting was purchased from Transduction Laboratories, anti-rpS6 was obtained from Cell Signaling Technologies, anti-FLAG was obtained from Sigma, and antihemagglutinin (anti-HA) (12CA5) was made in our lab.

Purification of recombinant proteins and GST pull-down experiment.

E. coli BL21(DE3) cells were transformed with pGEX-2T (Pharmacia), pGEX2T-PABP (7), pGEX-6P-2-Paip2, pGEX-HA-RRM3 (17), or pET28b-APC5 for preparation of GST, GST-PABP, GST-Paip2, GST-RRM3, or HIS-Apc5, respectively. Expression of recombinant protein was induced by 0.5 mM isopropyl-β-d-thiogalactopyranoside (IPTG) treatment for 3 h at 37°C. GST and GST-fused proteins were purified on glutathione-Sepharose 4B beads (Amersham Pharmacia Biotech), and HIS-Apc5 was purified on nickel-agarose beads (Qiagen) as recommended by the vendor. For the pull-down experiment, glutathione-Sepharose 4B beads coupled to GST or GST-PABP were used. Thirty microliters of beads was incubated with the protein of interest in a final volume of 100 μl of pull-down buffer (20 mM Tris-HCl [pH 7.4], 1 mM MgOAc, 100 mM KCl, 0.1% Triton X-100, Roche Complete protease inhibitor cocktail) for 2 h at 4°C followed by three washes with 400 μl of pull-down buffer. Proteins were eluted by boiling for 5 min in Laemmli sample buffer and resolved by SDS-PAGE (8% polyacrylamide).

Filter binding assay and EMSA.

For filter binding, a radiolabeled RNA spanning nucleotides 475 to 685 of the PDGF B 5′ UTR (IRES probe), or A_286-428 (Amersham Pharmacia Biotech catalog no. 27-4110), was added to different concentrations of GST-PABP in a binding buffer containing 10 mM HEPES-KOH (pH 7.6), 3 mM MgCl₂, 30 mM KCl, 1 mM DTT, 50-μg/ml tRNA, and 5% glycerol in a reaction volume of 50 μl. The final concentration of the labeled RNA was 0.04 nM. The binding reaction was incubated at room temperature for 30 min followed by filtration through nitrocellulose membrane and autoradiography of the retained RNA. For electrophoretic mobility shift assay (EMSA), 3 fmol of the above mentioned IRES probe (64,000 cpm) was incubated with recombinant GST-PABP, HIS-Apc5, or GST in a binding buffer containing 10 mM HEPES-KOH (pH 7.6), 3 mM MgCl₂ 30 mM KCl, 1 mM DTT, 1.3 mM ATP, 0.5-mg/ml tRNA, 10 U of SuperaseIn (Ambion), and 5% glycerol in a final volume of 25 μl. After 25 min at 30°C, the complexes were analyzed on 4% nondenaturing acrylamide gel (60:1 acrylamide-bisacrylamide) in a Tris-glycine running buffer containing 1 mM EDTA (pH 8.3). The running time was 2.5 h at 13 V/cm, and the run was followed by drying and autoradiography.

RESULTS

Apc5 was selected by a three-hybrid screen using a specific RNA sequence as bait.

We previously found that a U-rich region spanning nucleotides 475 to 797 of the PDGF2 5′ UTR (region B) is important for internal translation activity, in contrast to the A-rich region (region A) spanning nucleotides 1 to 227 of the 5′ UTR (31). We used a three-hybrid system (28) to identify proteins that bind to the IRES and which may regulate its activity. This system for selecting specific RNA-protein interactions in the nucleus of living yeast cells was successfully applied to the analysis of hnRNP-C interaction with the abovementioned U-rich region of PDGF2 IRES (19). Using region B as the RNA bait, a human cDNA library derived from HeLa cells (14) was screened for three-hybrid complex formation by selecting for HIS3 expression. Only clones classified as RNA-dependent and RNA specific (i.e., that could form a stable three-hybrid complex with RNA bait harboring region B, but not with region A) were analyzed. One of these clones encoded a protein composed of 361 amino acids identical to the carboxy terminus of the 755-amino-acid-long human Apc5 protein (accession no. AF191339) (37). Figure 1A shows transcription activation of both HIS3 and LacZ reporter genes upon specific interaction of the hybrid RNA containing region B with the C-terminal half of human Apc5 fused to the GAL4 activation domain. The complete Apc5 protein sequences from human, mouse, Drosophila, and budding yeast were aligned and checked for the presence of putative functional domains. The alignment revealed that only the mouse and human proteins, which are 90% identical, share two short, putative tetratricopeptide repeats (TPRs) overlapping a coiled-coil domain in the C-terminal half, selected by the three-hybrid screen. A classical RNA binding domain (RBD) was not found in any of the sequences.

FIG. 1.

(A) The C-terminal half of Apc5 promotes the generation of a stable three-hybrid complex with IRES RNA. pACTII-IRP, pACTII-Apc5, pACTII-Apc5Δ1, or pACTII-Apc5Δ2 expressing IRP, full-length Apc5, or the C- or N-terminal part of Apc5, respectively, were cotransformed along with pDBRevM10-B, pDBRevM10-A, or pDBRevM10 expressing the different RNA baits into CG1945 and Y190 yeast strains. Colonies were selected for HIS3 expression and assayed for LacZ expression as described in Materials and Methods. (B) Total extract from K562 cells was immunoprecipitated by anti-Apc3 antibodies (lanes 1) or a large excess of anti-HA antibodies (lanes 2) or was incubated with beads only (lanes 3). HEK293 cells were infected with vTF7-3 and transfected with pFlag-Apc5 (lanes 4) or pCDNA3-Apc5 (lanes 5). The membranes were immunoblotted with antibodies specific for Apc5, Apc3, or Flag tag, as indicated at the bottom of each panel.

We raised rabbit polyclonal antibodies against human Apc5, using recombinant HIS-tagged Apc5 expressed in E. coli. To confirm the specificity of the antibodies, we immunoprecipitated the APC/C complex with antibodies specific for Apc3 to ascertain that this band is indeed a subunit of the APC/C complex. The antibody detected a band of the appropriate size (Fig. 1B). The specificity of the antibody was further confirmed by the detection of a specific band upon overexpression of the Flag-tagged Apc5 open reading frame in HEK293 cells, which contain low levels of endogenous Apc5 (Fig. 1B).

Apc5 interacts with PABP.

Apc5, selected in this study as a potential RNA binding protein, does not harbor any classical RNA binding motif. To check its ability to bind RNA, recombinant HIS-tagged human Apc5 purified from E. coli was used for EMSA and UV cross-linking analysis. As we failed to demonstrate the direct binding of recombinant Apc5 to RNA, we assumed that under the selection pressure used for the screen, formation of a stable hybrid complex in yeast cells was mediated by one of the endogenous yeast proteins. A yeast two-hybrid assay was then used to check several translation initiation factors as possible candidates. Whereas no interaction was detected between Apc5 and eIF4G1, eIF4A, or one of the eIF3 subunits (data not shown), a significant interaction between Apc5 and PABP was observed. The qualitative two-hybrid assay, demonstrating the interaction of the human Apc5 C terminus with human PABP and with fragments of its yeast homologue, Pab1, is shown in Fig. 2A. To confirm Apc5-PABP interaction in vitro, a recombinant GST-PABP fusion protein was used for a GST pull-down assay. GST-PABP, but not GST alone, was able to pull down recombinant HIS-tagged Apc5, as well as native Apc5 from extracts of logarithmically growing human K562 cells (Fig. 2B). Coimmunoprecipitation experiments further confirmed the interaction in vivo. Because the anti-Apc5 antibodies recognize only the denatured protein, a Flag-tagged Apc5 expressed from a transfected plasmid was used in these experiments. Antibodies against Flag tag, but not against HA, pulled down Apc5, Apc3, and PABP. Antibodies against PABP were able to pull down a small amount of Flag-Apc5, as detected by the anti-Flag antibody (Fig. 2C).

FIG. 2.

Apc5 interacts with PABP. (A) Yeast two-hybrid system. pGBT9-Apc5(395-755) expressing the C-terminal half of human Apc5 fused to the GAL4 DNA binding domain was cotransfected into CG1945 yeast strain along with pACTII-hPABP, pGAD424-yPab1(1-237), pGAD424-yPab1(189-346), pGAD424-yPab1(345-577), or pACTII-IRP expressing the human PABP, different regions of yeast Pab1, or IRP, respectively, fused to the GAL4 activation domain. Colonies were selected for HIS3 expression. (B) GST pull down. GST or GST-hPABP protein was immobilized on glutathione-Sepharose for GST pull-down analysis and incubated with 1 μg of recombinant His-Apc5 or 100 μg of total protein extract from logarithmically growing K562 cells. Bound proteins were eluted in Laemmli sample buffer and immunoblotted with Apc5 or PABP antibodies. Recombinant GST-hPABP and His-Apc5 were used as markers. (C) Coimmunoprecipitations. HEK293 cells were infected with vTF7-3 and transfected with pFlag-Apc5 and pcDNA3-hPABP. Twenty-four hours later, cell extracts were subjected to immunoprecipitation (IP) with anti-Flag, anti-PABP, or anti-HA antibodies followed by immunoblotting with anti-Apc5, anti-PABP, anti-Apc3, or anti-Flag antibodies as indicated.

PABP interacts with IRES RNA.

The three-hybrid assay showed that human PABP can form a stable complex with non-poly(A) RNA sequences (Fig. 3A). The ability of human PABP to bind RNA sequence other than poly(A) was confirmed by a UV cross-linking experiment (data not shown) and by a filter binding assay (Fig. 3B). The latter shows that the affinity of human PABP for the non-poly(A) IRES sequence is ∼2 orders of magnitude lower than that for poly(A) RNA, in accordance with previous findings (13). We realized that selection pressure applied by the yeast three-hybrid assay enables the detection of interactions that cannot be readily detected in mammalian cells under normal conditions and thus may reflect transient interactions that occur under certain physiological conditions. The interaction of certain regions of the yeast Pab1 with the non-poly(A) sequences (Fig. 3A) supports the notion that yeast Pab1 served as the bridging protein between human Apc5 and the bait RNA in the original screen. The PABP-mediated binding of Apc5 to the RNA was confirmed by a supershift of an RNAB-PABP complex by Apc5 in an EMSA (Fig. 3C). Residues 189 to 346 of Pab1 were able to bind only to the B RNA bait, raising the possibility that this region in the yeast protein contributed to the discrimination against the A sequence in the original screen, but the involvement of additional factors cannot be excluded. Leaving aside the events that led to Apc5 selection in the yeast assay, we set out to examine the biological significance of the interaction of human PABP with human Apc5 and IRES RNA.

FIG. 3.

PABP interacts with IRES RNA. (A) Yeast three-hybrid system. pDBRevM10-B, pDBRevM10-A, or pDBRevM10 expressing the different RNA baits was cotransfected into yeast strain CG1945 along with pACTII-hPABP or pGAD424-yPab1(1-237), pGAD424-yPab1(189-346), or pGAD424-yPab1(345-577) expressing the human PABP or different regions of the yeast Pab1, respectively. Colonies were selected for HIS3 expression. (B) Filter binding assay. The filter binding assay was performed as described in Materials and Methods with 3,000 cpm (0.04 nM) of labeled A_286-428 or IRES RNA (open or solid circles, respectively). The percentage of radiolabeled RNA retained on the filter is shown as a function of PABP concentration. (C) EMSA was performed as described in Materials and Methods using IRES RNA as a probe and 450 ng of purified HIS-Apc5, 750 ng of purified GST-PABP, or 750 ng of purified GST. Arrowheads indicate (from bottom to top) the positions of the free, shifted, and supershifted probe.

Apc5 represses IRES-mediated translation in vitro.

First, we wished to assess the effect of PABP and Apc5 on translation in vitro. PABP-depleted and nondepleted Krebs-2 cell-free translation reactions (35) were programmed with poly(A)⁺ or poly(A)⁻ luciferase mRNAs containing different 5′ ends. Figure 4A, which presents the effect of PABP presence on translation efficiency, clearly demonstrates that PABP stimulated the translation of capped as well as IRES-containing RNA in a poly(A)-dependent manner. Apc5 acted as an inhibitory protein, but the IRES-containing transcript was more sensitive to Apc5 inhibition than the cap-containing transcripts (Fig. 4B). Preincubation of Apc5 with a truncated protein containing the RRM3 domain of hPABP (17) specifically prevented the inhibitory effect of Apc5 on the IRES-containing RNA (Fig. 4C).

FIG. 4.

Effect of PABP and Apc5 on translation in vitro. (A) Krebs-2 cell-free translation reaction mixtures that had been pretreated with GST-Paip2 for PABP depletion or with GST as a control were programmed with 100 ng of the indicated transcripts at 37°C for 90 min followed by measurements of firefly luciferase activity. Immunoblot analysis of the translation extracts using antibodies specific for PABP is shown at the top: lane 1, untreated; lane 2, treated with GST-Paip2; lane 3, treated with GST. For each transcript, the value obtained in the GST-treated extract was divided by the value obtained in the PABP-depleted extract. The stimulation values represent the average ± standard error of three independent duplicate experiments. (B) Krebs-2 cell-free translation reaction mixtures preincubated for 30 min at room temperature with the indicated amounts of purified HIS-Apc5 were programmed with 50 ng of the indicated transcripts for a further 50-min incubation at 37°C. Firefly luciferase activity in the absence of Apc5 was set as 100%. The values represent the average ± standard error of three independent experiments. (C) Krebs-2 cell-free translation reaction mixtures preincubated for 30 min at room tem-perature with 1 μg of HIS-Apc5 and the indicated amounts of purified GST-RRM3 were programmed with 50 ng of the indicated transcripts for a further 50-min incubation at 37°C. Firefly luciferase activity in the absence of HIS-Apc5 and GST-RRM3 was set as 100%. The values represent the average ± standard error of three independent experiments.

Apc5 represses IRES-mediated translation in a PABP-dependent manner in HEK293 cells.

Next, we wished to assess the effect of PABP and Apc5 on translation in vivo. For this purpose, HEK293 cells were cotransfected with increasing amounts of plasmids expressing PABP or Apc5, along with a bicistronic vector containing the IRES of PDGF2 between the Renilla and firefly luciferase coding regions. The total amount of transfected plasmid in each sample was kept constant by compensation with pcDNA-CAT DNA, which served as a nonspecific competitor for the transcription/translation machineries. As shown in Fig. 5A, PABP enhanced IRES-mediated translation in a dose-dependent manner, as manifested by the firefly luciferase activity. The presence of 1 μg of the Apc5-expressing plasmid in each transfection sample lowered the ability of PABP to stimulate the IRES. In HEK293 cells, Apc5 alone did not inhibit IRES activity, but in the presence of 1 μg of the PABP-expressing plasmid in each transfection sample, Apc5 was able to reduce in a dose-dependent manner the stimulation of the IRES by PABP. PABP alone did not have any significant effect on the first cistron (Fig. 5B), suggesting that in HEK293 cells the endogenous PABP is near saturation level for its role in cap-dependent translation. Although there is an excess of endogenous PABP in the cell (13), the amount seems to be limited for IRES stimulation. The lack of an inhibitory effect on the first cistron by Apc5 suggests that it cannot efficiently compete with the cap-binding complex for its interaction with PABP. The fact that the stimulatory effect on the IRES was observed upon PABP overexpression brings up the interesting possibility that naturally PABP may become available for IRES stimulation under conditions that inhibit cap-dependent translation. Such conditions may be achieved via manipulation of the activity or availability of the cap-binding complex through eIF4G management. Therefore, the data presented in Fig. 5 suggest a regulatory mechanism that depends on the availability of PABP to stimulate the IRES and the availability of Apc5 to hinder this stimulation.

FIG. 5.

Apc5 represses IRES-mediated translation in a PABP-dependent manner. HEK293 cells were infected with vTF7-3 and cotransfected with 0.4 μg of pLPL, expressing PDGF2 IRES between Renilla and firefly luciferases (10), along with increasing amounts of pcDNA-hPABP or pcDNA-Apc5. The total amount of transfected plasmids was kept constant by compensation with pcDNA-CAT. As indicated, to each pcDNA-hPABP sample was added an additional 1 μg of either pcDNA-Apc5 (open squares) or pcDNA-CAT (solid circles) and to each pcDNA-Apc5 sample was added an additional 1 μg of either pcDNA-hPABP (open squares) or pcDNA-CAT (solid circles). Twenty-four hours after transfection, cells were harvested and assayed for the activity of Renilla (first cistron) (B) and firefly (second, IRES-controlled cistron) (A) luciferases. Basal activity of firefly luciferase, in the absence of additional Apc5 or PABP, was set as 1. The values represent the average ± standard error of three independent experiments. (C) Cells transfected with 0, 0.5, 1, 2, or 3 μg (lanes 1 to 5, respectively) of PABP- or Apc5-expressing plasmid were analyzed by immunoblotting with antibodies specific for PABP or Apc5 as indicated.

Apc5 is associated with additional complexes, aside from APC/C.

To assess the possible association of Apc5 with other cellular components apart from its association with the APC/C complex, we first looked at its cellular location. Logarithmically growing and megakaryocytic differentiated K562 cells were fractionated into nuclear, cytoplasmic, and RSW fractions and checked for Apc3 and Apc5 protein levels. Aside from its nuclear location, Apc5 was also associated with the RSW fraction (Fig. 6A). In contrast, Apc3 was present predominantly in the nuclear fraction. To confirm the association of Apc5 with other complexes in addition to the APC/C, the total cell extract was subjected to gel filtration analysis. Figure 6B shows that Apc3 and Apc5 are not similarly distributed. At low detergent concentrations, most of the Apc3 was attached to complexes with a molecular mass of 840 to 1,700 kDa, whereas a major portion of Apc5 was part of much heavier complexes. Higher detergent concentrations partially dissociated Apc5 from the heavier complexes, resulting in a filtration pattern similar to that of Apc3, peaking around 844 kDa, the reported mass of APC/C (11). Additional corroboration that Apc5 is associated with heavier complexes in addition to the APC/C came from sedimentation analysis of a sucrose gradient. Part of Apc5 cosedimented with Apc3 in fractions lighter than 40S (Fig. 6C), probably corresponding to 22S to 23S, the reported sedimentation coefficient value of the human APC/C (11). However, a fraction of Apc5, but not of Apc3, cosedimented with both lighter particles as well as with heavier complexes, such as the 43S/48S translation preinitiation complexes.

FIG. 6.

Apc5 is associated with other complexes, aside from APC/C. (A) Forty micrograms of proteins of the cytoplasmic S100 (C), RSW (R), and nuclear extracts (N) from control (−TPA) or differentiated (+TPA) K562 cells prepared as previously described (31) was immunoblotted with antibodies specific for Apc5, Apc3, or ribosomal protein S6 (rpS6). (B) Gel filtration chromatography. Total soluble proteins from logarithmically growing K562 cells lysed in the presence of a high (top panel) or low (bottom panel) concentration of detergents were separated by Superose-6 gel filtration chromatography. Equal volumes from each fraction were immunoblotted with antibodies specific for Apc5, Apc3, or rpS6. The elution point of the indicated molecular masses was calculated by using Unicorn software based on the elution points of thyroglobulin (669 kDa), apoferritin (443 kDa), catalase (232 kDa), and bovine serum albumin (66 kDa), which were used as molecular mass markers. (C) Density sucrose gradients. Twenty-five A₂₆₀ units of K562 cells was resolved on a 5 to 45% sucrose gradient as described in Materials and Methods. Peaks containing the 40S, 60S, and 80S ribosomal subunits are indicated. Portions of the fractions were analyzed by immunoblotting with antibodies specific for Apc3, Apc5, or eIF3.

Apc5 is degraded upon megakaryocytic differentiation of K562 cells.

To evaluate the relevance of Apc5 to IRES activity under physiological conditions, we searched for a cellular system in which the Apc5 level changes in response to a specific signal. This was important because nonnatural interference with Apc5 expression in logarithmically growing cells is likely to be toxic due to its role as an essential component of the APC/C complex. Working with human K562 cells, we previously showed that upon megakaryocytic differentiation, they become more permissive for IRES function (3, 10, 31). Interestingly, in correlation with the differentiation-induced IRES activation, we observed the disappearance of Apc5 in the course of the differentiation process, in contrast to Apc3, which remained stable (Fig. 7A). Apc5 degradation began 15 to 20 h after TPA treatment and was sensitive to MG132, implying that the degradation is proteasome dependent (Fig. 7B). In parallel, MG132 inhibited the differentiation-dependent induction of PDGF2 IRES activity (Fig. 7C and Table 1), suggesting that degradation of inhibitory proteins, one of which may be Apc5, is beneficial to IRES activation during differentiation.

FIG. 7.

Apc5 is degraded upon megakaryocytic differentiation. (A) Apc5, but not Apc3, is degraded during differentiation. Fifty micrograms of total cell extract from logarithmically growing (log) or megakaryocytic differentiated (diff) K562 cells was separated by SDS-PAGE (10% polyacrylamide) and immunoblotted with antibodies specific for Apc5 or Apc3. (B) Proteasome inhibitor abolishes the differentiation-induced Apc5 degradation. K562 cells were treated with 5 nM TPA for 6 h followed by further incubation in medium containing 5 nM TPA with or without 30 μM MG132. At the indicated time points of TPA treatment (corresponding to 3 or 14 h of MG132 treatment), cells were harvested and 50 μg of total cell protein was immunoblotted with antibodies specific for Apc5, PABP, or hnRNP-C. (C) Proteasome inhibitor abolished the differentiation-induced IRES activation. K562 cells were transfected with pLPL expressing PDGF2 IRES between Renilla and firefly luciferases or the IRES-less vector pLL. Twenty-four hours after transfection, the cells were treated with 5 nM TPA, and 9 h later, MG132 was added to a final concentration of 30 μM. Forty-eight hours after transfection, cells were harvested and assayed for the activity of Renilla (R) (first cistron) and firefly (F) (second, IRES-controlled cistron) luciferases. The absolute values are presented in Table 1. The F/R values represent the firefly/Renilla activity ratio (arbitrary units) and are the average ± standard error of three independent experiments.

TABLE 1.

Absolute values of Renilla and firefly luciferase activities from the experiment shown in Fig. 7C

Plasmid	Luciferase activity (U/106 cells)
	Renilla				Firefly
	Nondifferentiated		Differentiated		Nondifferentiated		Differentiated
	−MG132	+MG132	−MG132	+MG132	−MG132	+MG132	−MG132	+MG132
pLL	94 ± 8.5	154 ± 11	418 ± 27	219 ± 18	0.27 ± 0.01	0.54 ± 0.03	0.71 ± 0.06	0.55 ± 0.04
pLPL	120 ± 8.8	210 ± 13	512 ± 42	232 ± 15	2.0 ± 0.15	2.1 ± 0.02	23 ± 1.4	3.1 ± 0.25

Apc5 overexpression inhibits D-IRES activation.

To determine the effect of Apc5 expression on IRES function in differentiated cells, plasmids expressing bicistronic transcripts with or without different IRES elements were transfected along with an Apc5- or enhanced green fluorescent protein (EGFP)-expressing plasmid, followed by incubation under normal or differentiation conditions. Although the transfection efficiency was 40 to 60%, as judged by the percentage of fluorescent EGFP-expressing cells, only a modest enhancement of Apc5 level was observed, most probably due to its instability in differentiated cells (Fig. 8A). The enhanced Apc5 protein level did not affect overall protein synthesis in logarithmically growing or differentiated K562 cells, as determined by the rate of incorporation of radiolabeled amino acids (not shown). However, under the same conditions, the inhibitory effect on differentiation-induced IRES activation (D-IRES) was observed (Fig. 8B). The D-IRES value of each IRES element in the presence of the Apc5 was compared with the value obtained in the presence of EGFP, which was set as 1. Figure 8B shows that overexpression of APC5 in differentiated cells did not affect the firefly/Renilla luciferase ratio obtained from the IRES-less vector pLL. Similarly, Apc5 overexpression barely affected the D-IRES value of encephalomyocarditis virus (EMCV) IRES. However, Apc5 overexpression in differentiated cells significantly inhibited the D-IRES values of all the cellular IRES elements tested. Northern blot analysis ruled out the possibility that Apc5 overexpression affected the stability of the IRES-containing RNA (data not shown). These data are in agreement with our finding that Apc5 suppresses IRES activity. Interestingly, we observed significant variations in Apc5 mRNA level in different tissues. The levels of Apc5 mRNA were particularly high in brain, heart, skeletal muscle, kidney, and liver and particularly low in the colon, thymus, spleen, small intestine, placenta, and lung (Fig. 8C), suggesting that Apc5 may serve as a regulator of IRES activity in these tissues in response to specific physiological signals.

FIG. 8.

Apc5 overexpression inhibits differentiation-induced IRES activation. K562 cells were transfected with the IRES-less vector pLL or with pLPL, pLVL, pLML, pLXL, and pLEL harboring the IRESs of PDGF2, VEGF, c-myc, Xiap, and EMCV, respectively, between Renilla (R) and firefly (F) luciferase coding regions (10). Each of the bicistronic vectors was cotransfected along with pEGFP or pcDNA3-Apc5, expressing EGFP or Apc5, respectively. Twenty-four hours after transfection, cells were incubated under normal or differentiation conditions (5 nM TPA). (A) At the indicated time points along the differentiation process, the transfected cells were subjected to immunoblot analysis using anti-Apc5 antibodies. (B) Control and 48-h TPA-treated cells were harvested and assayed for the activity of Renilla (R) (first cistron) and firefly (F) (second, IRES-controlled cistron) luciferases. The D-IRES value is the F/R ratio in differentiated cells relative to that in control cells. The D-IRES value of each IRES in the presence EGFP was set as 1 (solid bars). The D-IRES values in the presence of Apc5 (open bars) represent the average ± standard error of three independent experiments. (C) Apc5 mRNA expression in different tissues. A multiple-tissue Northern blot (human 12-lane MTN #7780-1; Clontech) containing equal amounts of poly(A)⁺ from specific tissues (adjusted to β-actin hybridization signal) was hybridized with labeled Apc5 cDNA. RNA size markers (in kilobases) are indicated.

DISCUSSION

APC/C, the ubiquitin ligase (E3) of many mitotic proteins, plays an essential role in the progression of cells from mitosis to G₁. The APC/C is active throughout G₁, and recent work implies that it also ubiquitinates proteins not directly related to the cell cycle (15). The APC/C is composed of a dozen subunits, but the function of the individual subunits is unknown. Apc1, Apc2, Apc10, and Apc11 have Rpn1/2 homology, Cullin homology, the Doc domain and the ring-H2 finger, respectively, whereas Apc3, Apc6, Apc7, and Apc8 contain TPRs (15). No homology has ever been found for Apc5. Using the ClustalW program for multiple sequence alignment of Apc5 from human, mouse, Drosophila, and budding yeast, we identified two short putative TPR domains in the human and mouse proteins that share 90% identity. The TPR motif in the C-terminal half overlaps a putative coiled-coil domain in both proteins. The lack of putative functional motifs in the yeast and Drosophila proteins, in addition to the relatively low evolutionary conservation between them and the mammalian Apc5, suggests that the latter might have gained additional functions in the course of evolution. Such functions may be related to its ability to bind other cellular components through the TPR and coiled-coil motifs, which are known to mediate protein-protein interactions. Indeed, our study provides evidence for the binding of the C-terminal half of human Apc5 to PABP, a protein involved in many aspects of RNA metabolism. The results of our study also show that Apc5 is associated with a complex or complexes heavier than APC/C and that such binding is resistant to mild detergent (Fig. 6B). The absence of Apc3 from the heavy complex or complexes implies that they do not represent APC/C multimers. The amount of Apc5 associated with APC/C may be governed by unknown cellular events that control its availability and/or binding capacity. The degradation of Apc5, but not of Apc3, during K562 cell differentiation supports this possibility. The association of Apc5 with the RSW fraction and its cosedimentation with heavier complexes such as preinitiation complexes (Fig. 6C) suggest a role for Apc5 in the control of gene expression, in addition to its APC/C function. Two observations support this notion. First, we provide evidence for direct binding of Apc5 with PABP. The activity of PABP as a translational stimulator of 5′-capped 3′-polyadenylated transcripts is well documented (9, 30). Due to antibody limitations, it is not clear at this point whether Apc5, when associated with PABP, binds to the rest of the APC/C complex. Second, we show that PABP stimulates PDGF2 IRES activity and that Apc5 inhibits the beneficial effect of PABP on IRES activation. Further experiments will elucidate what the physiological conditions are that lead to Apc5-PABP binding and how Apc5 counteracts the stimulatory effect of PABP on the IRES.

PABP is a multifunctional RNA-binding protein with at least two distinct RBDs: I/II, which most likely binds to the poly(A) tail; and III/IV, which may function by binding to a different part of the same mRNA molecule (4, 33). The PABP-mediated RNA circularization may be facilitated by the multimerization activity of PABP, conferred by its non-RBD C terminus, which multimerizes upon binding to poly(A) (20). Selection/amplification assays revealed that PABP binds to unrelated RNA sequences with an affinity ∼100-fold lower than that for poly(A) (13). Our findings are consistent with these data. It was previously shown that polyadenylation stimulates translation of mRNAs carrying picornaviral IRES elements in cell-free systems (2, 25, 26). PABP interaction with eIF4G was shown to be required for such stimulation (24, 34). To the best of our knowledge, the current study is the first one to show PABP's ability to stimulate the activity of cellular IRES. We demonstrated it in living cells and also in a cell-free translation system that confirmed the poly(A) dependency of the effect. However, at this point, it is not clear if the stimulatory effect by PABP is eIF4G dependent. Moreover, this is the first demonstration of a stimulatory effect of PABP on the IRES that can be thwarted by Apc5. In our experimental system, the beneficial effect of transfected PABP on cap-dependent translation was insignificant. This is probably due to the high accessibility of PABP to the 5′ cap to which it tightly binds through interaction with the eIF4F complex (30), leading to near-saturation levels of endogenous PABP for its role in cap-dependent translation. The fact that the stimulatory effect on the IRES was observed upon PABP overexpression brings up the possibility that naturally PABP may become available for IRES stimulation only under conditions that inhibit cap-dependent translation. Similarly, the inhibitory effect of transfected Apc5 on cap-dependent translation was insignificant compared with that on IRES-mediated translation, obtained only when PABP was available for IRES activation. The inability of Apc5 to inhibit cap-dependent translation might be due to its inability to compete with eIF4G for interaction with PABP. Moreover, it is likely that Apc5 does not elicit its regulatory effect on the IRES by disturbing the binding of PABP to poly(A): otherwise it would negatively affect cap-mediated translation as well. Therefore, Apc5 is probably working specifically on the IRES by perturbing the PABP-IRES-mediated mRNA circularization. Recently, we were able to confirm by GST pull-down assays using different PABP fragments the interaction of Apc5 with RBDs I and III and more strongly with the N-terminal part of the C-terminal domain of human PABP (not shown). Based on this information, Apc5 may interfere with PABP dimerization in such a way that it can not support the interaction between poly(A)-bound and IRES-bound monomers. Future experiments should elucidate the mechanism by which Apc5 exerts its effect.

Our data suggest a regulatory mechanism that depends on the availability of PABP to stimulate the IRES and the availability of Apc5 to counteract it, in response to specific signals. In the cell, PABP availability and activity are regulated by poly(A)-interacting proteins (PAIPs). In contrast to Paip-1, which stimulates translation, Paip-2 inhibits translation by decreasing the affinity of PABP for poly(A) and by competing with Paip-1 for PABP binding (6, 18). Additional work will help to reveal the physiological signals that control cap-dependent translation in contingence with PABP availability and might uncover an additional mechanism for IRES activation. In the present report, we identified Apc5 as an additional PABP-interacting protein with an inhibitory effect on IRES function and we also showed that during megakaryocytic differentiation, the disappearance of Apc5 is correlated with IRES activation. Further experiments will convey the importance of Apc5 in the complex translational regulation associated with the cell cycle, development, and differentiation.