Identification of cellular proteins enhancing activities of internal ribosomal entry sites by competition with oligodeoxynucleotides

Abstract

The translation of numerous eukaryotic mRNAs is mediated by internal ribosomal entry sites (IRESs). IRES-dependent translation requires both canonical translation initiation factors and IRES-specific trans-acting factors (ITAFs). Here we report a strategy to identify and characterize ITAFs required for IRES-dependent translation. This process involves steps for identifying oligodeoxynucleotides affecting IRES-dependent translation, purifying proteins interacting with the inhibitory DNA, identifying the specific proteins with matrix-assisted laser desorption ionization/time-of-flight mass spectrometry, and confirming the roles of these proteins in IRES-dependent translation by depletion and repletion of proteins from an in vitro translation system. Using this strategy, we show that poly(rC)-binding proteins 1 and 2 enhance translation through polioviral and rhinoviral IRES elements.

INTRODUCTION

Translation through the internal ribosomal entry site (IRES) plays an important role in cap-independent translation of numerous viral mRNAs and regulation of cellular gene expression at the translational level (1–5). IRES elements were initially identified in picornaviral mRNAs, such as encephalomyocarditis virus (EMCV) (6) and poliovirus (PV) (7). Subsequently a number of IRES elements have been discovered from different groups of viruses and cellular mRNAs (8–10). Although the number of newly discovered IRES elements has increased dramatically, the molecular basis and regulation mechanisms of function are poorly understood.

Several cellular proteins stimulate translation of IRES-dependent mRNAs. For instance, polypyrimidine tract binding protein (PTB), a cellular RNA-binding protein (57 kDa) that binds to the 5′ region of EMCV IRES (11) is required for translation directed by picornaviral IRES elements, such as PV, human rhinovirus (HRV) and EMCV (12–14). Human La autoantigen, involved in the regulation of initiation and termination of transcription by RNA polymerase III, stimulates IRES-dependent translation of polioviral mRNA, hepatitis C viral (HCV) mRNA, cellular human immunoglobulin heavy chain-binding protein (BiP) mRNA, and X-chromosome-linked inhibitor of apoptosis (XIAP) mRNA (15–18). Poly(rC)-binding protein 2 (PCBP2), also known as α-complex protein 2 (αCP2) and heterogeneous nuclear ribonucleoprotein E2 (hnRNP E2), is required for IRES-dependent translation of PV (19). Other RNA-binding proteins, such as upstream of N-ras (unr) (20,21), heterogeneous nuclear ribonucleoprotein C (hnRNP C) (5,22) and heterogeneous nuclear ribonucleoprotein L (hnRNP L) (23), additionally modulate IRES-dependent translation of mRNAs. These non-canonical translation factors were identified from analysis of IRES-binding proteins by UV cross-linking or investigating proteins enhancing specific IRES functions in an in vitro translation system through fractionation of cellular extracts.

Here we report a new strategy to identify and characterize cellular factors modulating IRES-dependent translation. Initially, the effects of oligodeoxynucleotides (oligo-DNAs) on IRES- and cap-dependent translation are investigated using multiple IRES elements and oligo-DNAs. Next, the relationship between oligo-DNAs that affect IRES activities and IRES elements is analyzed. Oligo-DNAs that inhibit IRES function without evident complementary sequences to IRES elements are selected. Thirdly, cellular proteins interacting with the oligo-DNA are purified with an affinity column and identified by matrix-assisted laser desorption ionization/time-of-flight mass spectrometry (MALDI-TOF-MS). Finally, the roles of proteins in IRES-dependent translation are analyzed via molecular biological and biochemical methods.

Using the strategy described above, we show that an oligonucleotide, CCCCCTT, inhibits translation through PV IRES and HRV IRES. This oligomer interacts with PCBP1 and PCBP2, and inhibits the function of PV and HRV IRES elements through sequestration of these factors from in vitro translation mixtures. We additionally demonstrate that both PCBP1 and PCBP2 augment PV and HRV IRES activities via depletion and repletion of these proteins from in vitro translation mixtures. This procedure is suitable for identifying non-canonical cellular proteins affecting IRES function, since relatively few proteins interact with a specific oligonucleotide affecting IRES activity compared with RNA corresponding to the IRES element (hundreds of nucleotides long), due to the small size of the oligonucleotide (about 10 bases). Moreover, synthesis of oligo-DNAs is easier than that of RNAs of the corresponding sequence.

MATERIALS AND METHODS

Plasmid construction

Plasmid pRPF containing polioviral IRES between Renilla and firefly luciferase was constructed by replacing the c-myc IRES sequence of pRM531F (5) with the 5′ non-translated region (NTR) sequence of PV (24). Plasmid pRRF containing rhinoviral IRES between Renilla and firefly luciferase was constructed by ligating a NcoI–Klenow–BamHI fragment of pHRV2 to a NdeI–Klenow–SalI fragment of pRM531F. The pGL3-H402 plasmid was generated by ligation of a pSKH402CAT fragment (23) treated with Asp718–BamHI–Klenow with pGL3 (Promega) treated with NcoI–Klenow. pGL3-H402 treated with HindIII–XbaI–Klenow and pRL-CMV (Promega) treated with XbaI–Klenow were ligated to generate the dicistronic construct, pRH402F. To produce pSKPSLIV containing stem–loop IV of PV IRES, PCR was performed with pMPS1-ECAT (11) as template and two primers (5′-CGCGGATCCGTTATCCGCTTATGTACTTC-3′ and 5′-ATCGAATTCCATTTCTTATGTAGCTCACTCG-3′), using Amplitaq Gold (Perkin-Elmer). Amplified DNA was ligated to pSK(–) digested with EcoRI and BamHI. Plasmid pSKRSLIV containing the stem–loop IV region of HRV IRES was constructed by ligation of a product yielded by PCR with pHRV2 and two primers (5′-CGCGGATCCGTTAACCGCAAAGCGCCTAC-3′ and 5′-ATCGAATTCCATCTCAAAGCGAGCACACGG-3′) to pSK(–) digested with EcoRI and BamHI. DNA encoding PCBP1 was obtained by amplification from a human fetal cDNA library (Clontech) using two primers (5′-GGAATTCCCATGGATGCCGGTGTGACTG-3′ and 5′-AACTGCAGCTCGAGCTAGCTGCACCCCATGCCCTTC-3′). Amplified PCBP1 was ligated to pSK(–) digested with EcoRI and PstI to generate pSK(–) PCBP1. To confirm the fidelity of the clone, DNA sequencing was performed using an ABI automated sequencer (ABI 3100). For recombinant PCBP1 protein, pGEX-KG/PCBP1 was constructed by ligating an NcoI-XhoI fragment of pSK(–) PCBP1 to pGEX-KG digested with the same restriction enzymes.

Peptide mass fingerprinting by MALDI-TOF-MS

Peptide mass fingerprinting was performed as described previously (25). In brief, CT-oligo-binding proteins were enriched by affinity column chromatography. DNA-bound proteins were resolved by 12% SDS–PAGE. Several bands were excised and digested with trypsin (Roche Molecular Biochemicals) at 37°C for 6 h. The masses of tryptic peptides were determined with a Bruker Reflex III MALDI-TOF mass spectrometer (POSTECH Instrument Center).

Recombinant protein purification

GST-fused proteins (PCBP1 and PCBP2) were expressed in Escherichia coli BL21(DE3)pLysS after transforming cells with plasmids pGEX-KG/PCBP1 and pGEX-KG/PCBP2, respectively (26). IPTG (final concentration of 0.5 mM) was added to the medium at an OD_600nm value of 0.5. After additional cultivation at 27°C for 5 h, cells were harvested and resuspended in lysis buffer [20 mM Na-phosphate (pH 7.6), 300 mM NaCl, 0.5 mM PMSF, 1 mM β-mercaptoethanol, 10% (v/v) glycerol]. GST-fusion proteins were allowed to bind to glutathione Sepharose 4B resin (Amersham-Pharmacia Biotech) in lysis buffer at 4°C for 2 h. After washing protein-bound resin four times with lysis buffer, proteins were treated with thrombin (Roche). Proteins released from the resin by thrombin were loaded onto a phosphocellulose column. After washing the resin with binding buffer containing 0.1 M NaCl [50 mM Tris–HCl (pH 7.6), 100 mM NaCl, 0.5 mM PMSF, 1 mM β-mercaptoethanol, 10% (v/v) glycerol], proteins were eluted with binding buffer containing 1.0 M NaCl. Purified proteins were dialyzed against HT buffer [16.2 mM HEPES–KOH (pH 7.5), 36 mM KCl, 160 mM KOAc, 1.24 mM MgOAc, 1.6 mM DTT and 2.8 mM β-mercaptoethanol].

Preparation of HeLa S10 cell extracts

Cytoplasmic S-10 extracts of HeLa S3 cells were prepared as described by Oh et al. (27). HeLa cell extracts were stored at –80°C.

UV cross-linking

Oligo-DNAs were end-labeled with [γ-³²P]ATP and T4 polynucleotide kinase. ³²P-labeled RNA probes were synthesized by in vitro transcription and isolated by push column chromatography (Stratagene). Probes (3 × 10⁵ c.p.m.) were incubated with 20 µg HeLa cell cytoplasmic extracts or 100 ng purified PCBP1 and PCBP2. RNA–protein interactions were performed in a 30 µl reaction mixture containing 0.5 mM DTT, 5 mM HEPES (pH 7.6), 75 mM KCl, 2 mM MgCl₂, 0.1 mM EDTA, 4% glycerol, 20 U RNasin and 3 µg tRNA. After 20 min incubation at 30°C, samples were irradiated with UV light on ice for 15 min with a UV cross-linker (Vilber Lourmat). Unbound RNA was digested with 5 µl RNase cocktail (2 µl RNase A [10 mg/ml], 2 µl RNase T1 [100 U/ml], 1 µl RNase V1 [700 U/ml]) at 37°C for 45 min and analyzed by 12% SDS–PAGE.

Generation of biotinylated RNA and RNA pull-down assay

For generating biotinylated RNA, 5 µg pSKH402CAT, pSKPSLIV and pSKRSLIV were linearized using BamHI (for pSKH402CAT) and EcoRI (for pSKPSLIV and pSKRSLIV). RNA transcripts were generated with T7 RNA polymerase (Stratagene) for pSKH402CAT and T3 RNA polymerase (Roche) for pSKPSLIV and pSKRSLIV with nucleoside triphosphates in biotinylation buffer [0.5 mM (each) ATP, CTP, GTP and UTP and 0.015 mM Bio-21-UTP (Stratagene)]. After 2 h incubation at 37°C, 5 U RNase-free DNase (Promega) was added to remove template DNA. RNA pull-down experiments were performed using recombinant PCBP1 or PCBP2 and biotinylated RNA corresponding to HCV IRES (nt 18–402), stem–loop IV RNA of PV IRES (nt 226–442) and stem–loop IV RNA of HRV IRES (nt 223–433). Following incubation of biotinylated RNA (final concentration of 80 nM) and recombinant proteins (400 ng) in 1 ml incubation buffer [10 mM HEPES (pH 7.4), 1.5 mM magnesium acetate, 150 mM potassium acetate, 2.5 mM DTT, 0.05% NP-40] at 4°C for 30 min, samples were subjected to streptavidin–agarose resin (Pierce) adsorption and further incubated for 2 h. Yeast tRNA (Roche) (8 µg) was added to binding mixtures as a non-specific competitor. After adsorption, the resin was washed four times with incubation buffer, and resin-bound proteins were resolved by 12% SDS–PAGE. Proteins were transferred to nitrocellulose membranes (Amersham-Pharmacia Biotech). Membranes were incubated overnight at 4°C in blocking solution [20 mM Tris–HCl (pH 7.4), 150 mM NaCl, 0.5% Tween-20, 5% skimmed milk] to minimize non-specific binding. Primary antibodies [polyclonal antibody against PCBP1 and PCBP2 (both from Santa Cruz Biotechnology)] were added to the blocking solution, and incubated for 1 h. Membrane-bound antibodies were detected by enhanced chemiluminescence (Amersham-Pharmacia Biotech) using horseradish peroxidase-linked anti-goat immunoglobulin G as the secondary antibody.

In vitro transcription and translation

Transcription reactions were performed using T7 RNA polymerase (Roche Molecular Biochemicals) at 37°C for 90 min according to the manufacturer’s instructions. The concentrations of the RNA transcripts were determined using a UV spectrophotometer. In vitro translation reactions in micrococcal nuclease-treated HeLa S-10 were performed for 1 h in 12.5 µl reaction mixtures containing 10 nM mRNA at 30°C. The translational efficiency of mRNA was monitored by luciferase activity measured by a Dual Luciferase assay kit (Promega) using the manufacturer’s instructions.

Depletion of endogenous PCBP1/2 from HeLa cell extracts

Endogenous PCBP1/2 in HeLa extracts was depleted by affinity chromatography, using a biotinylated CT-oligomer strongly bound to PCBP1 and PCBP2. Briefly, 100 µl streptavidin agarose beads were incubated with 30 µg biotinylated CT-oligomer and washed four times with HT buffer. HeLa cell extracts (500 µl) were incubated with the CT-oligomer-bead complex at 4°C for 1 h and centrifuged at 3150 g. The supernatant was employed in the translation reaction. A mock-depleted HeLa extract was prepared by incubation of the HeLa extract with streptavidin–agarose beads in the absence of biotinylated CT-oligomer.

RESULTS

Search for oligodeoxynucleotides affecting IRES activity

The effects of various oligo-DNAs complementary to different regions of PV IRES on the translation of PV, HRV and HCV mRNAs were investigated using HeLa cell lysates and artificial dicistronic mRNAs suitable for monitoring IRES activities (Fig. 1A). Dicistronic mRNAs composed of methyl guanosine cap structure–Renilla luciferase gene (Rluc)– various IRES elements–firefly luciferase gene (Fluc) (from the 5′ to 3′ end) were generated by in vitro transcription with T7 RNA polymerase (Fig. 1A). Oligo-DNAs (7–15 nt long) complementary to the loop regions of the RNA secondary structure of PV were selected in this experiment (Table 1). The effects of oligo-DNAs on different IRES activities were determined. Among the oligomers, CCCCCTT oligo-DNA (designated ‘CT-oligomer’) strongly inhibited the translation of PV and HRV mRNAs (Fig. 1B, compare lanes 1 and 4 with lanes 3 and 6). On the other hand, translation of HCV mRNA was not affected by the CT-oligomer (Fig. 1B, compare lanes 7 and 9). The specificity of the primary sequence of the CT-oligomer was monitored by using a CGCGCTT oligomer with a difference of two nucleotides (designated ‘CG-oligomer’). Translation of IRES-dependent mRNAs was not affected by the CG-oligomer (Fig. 1B, lanes 2, 5 and 8). This finding strongly indicates that the inhibitory effect of CT-oligomer is sequence-specific. The inhibition of PV RNA translation is possibly due, at least in part, to the antisense DNA effect that distorts the RNA structure of PV IRES. However, the inhibitory effect of the CT-oligomer on HRV translation is not explained by the antisense effect via direct annealing, since no potential annealing site to the CT-oligomer exists in the 5′ NTR of HRV.

Figure 1.

Effect of the CT-oligomer on IRES-dependent translation. (A) A schematic diagram of the reporter constructs. Capped mRNAs were generated by in vitro transcription with T7 RNA polymerase in the presence of m⁷GpppG. The mRNAs (RPF, RHF and RRF) contain viral IRES sequences (PV, HCV and HRV) between Renilla (RLuc) and firefly luciferases (FLuc). IRES-dependent translation activity was normalized by calculating the ratio of the IRES-dependent translation (measured by FLuc activity) to cap-dependent translation (measured by RLuc activity). (B) Effects of the CT-oligomer and CG-oligomer on IRES-dependent translation in HeLa extracts. Capped mRNAs (final concentration, 10 nM) were used in translation reactions in the presence of 10 µM CG-oligomer (gray bars) and CT-oligomer (open bars) or the absence of oligonucleotide (solid bars).

Table 1. Oligo-DNAs used in measuring inhibitory effects on the activities of PV, HRV and HCV IRES elements.

Sequence	Postion of complementary sequences
GCTGTTTTAA	1–10
GGGGTGGGT	23–31
GTACCGCAATT	61–72
GGGAAGGGAG	91–100
CCCCCTT	153–159
CTGGTTTGT	161–169
CGCAAGCAGTCTATA	191–204
CTCTGGGGTTGAG	292–305
CCGTCACCGG	340–349
ACTGGTTTGT	491–500
CGCGTTACGA	511–520

The numbers indicate the nucleotide positions in the 5′ NTR of PV complementary to the oligomers.

A 40 kDa protein interacts with the CT-oligomer

A possible explanation for the inhibitory effect of the CT-oligomer on HRV translation is the sequestration of cellular factor(s) required for HRV mRNA translation by the CT-oligomer. This theory was investigated by searching for cellular protein(s) specifically interacting with the CT-oligomer. UV cross-linking experiments were performed with HeLa cell extracts and ³²P-labeled CT-oligomer. A protein with an apparent molecular mass of 40 kDa bound strongly to the CT-oligomer (Fig. 2A, lane 1). A few additional bound proteins were detected with the UV cross-linking experiment (Fig. 2A, lane 1). Among the labeled bands, only the 40 kDa band was weakened following the addition of unlabeled oligo CT (Fig. 2A, lanes 2–5). Oligo CG competed poorly for binding to the 40 kDa protein (Fig. 1A, lanes 6–9). Binding of the 40 kDa protein to HRV IRES was investigated by an RNA UV cross-linking experiment in the presence of CT or CG oligomers, using HeLa cell extracts and ³²P-labeled RNA corresponding to HRV IRES. A 40 kDa protein was detected by UV cross-linking with numerous other proteins (Fig. 2B, lane 1). The binding of this protein was inhibited strongly by the CT-oligomer (Fig. 2B, lanes 2–4), but only weakly by the CG-oligomer (Fig. 2B, lanes 5–7). The data indicate that both the CT-oligomer and HRV IRES interact with a 40 kDa protein in HeLa cell extracts.

Figure 2.

A 40 kDa protein interacts with the CT-oligomer. (A) UV cross-linking experiments were performed with HeLa cell extracts (20 µg) and ³²P-labeled CT-oligomer (2 × 10⁵ c.p.m.). Competition experiments were performed by adding 80 nM, 400 nM, 2 µM and 10 µM unlabeled CT-oligomer (lanes 2–5) or CG-oligomer (lanes 6–9) in the UV cross-linking reaction mixtures. After irradiation with UV light to form cross-links, samples were resolved by 12% SDS–PAGE. Arrows depict the 40 kDa protein specifically interacting with the CT-oligomer. (B) UV cross-linking of HRV IRES with HeLa cell extracts. RNA corresponding to full-length HRV IRES was labeled with ³²P and incubated with HeLa cell extracts at increasing concentrations of unlabeled CT- or CG-oligomer. Following irradiation with UV light, samples were treated with an RNase cocktail and analyzed by 12% SDS–PAGE.

PCBP1 interacts with the CT-oligomer

To determine the identity of the 40 kDa protein interacting with the CT-oligomer, purification was performed using affinity column chromatography with biotinylated CT-oligomer. After incubation of HeLa extracts with the biotinylated CT-oligomer, complexes were precipitated with streptavidin–agarose beads, and resin-bound proteins were resolved on an SDS–polyacrylamide gel (Fig. 3A). The 40 kDa protein was observed among the several bands detected in the protein–DNA complex. Several bands that were not detected by UV cross-linking, were additionally observed (compare Figs 2A and 3A). These proteins may exist in a protein–DNA complex through interactions with the 40 kDa protein. The 40 kDa band was excised and analyzed by peptide mass fingerprinting with MALDI-TOF-MS (Fig. 3B). The molecular masses of 16 peaks were matched with tryptic fragments of PCBP1 in the Swiss-Prot database according to the MS-Fit peptide mass search program. The peptide search result was performed with an accuracy of 50 p.p.m. Analyzed peptides comprised 51% of the PCBP1 sequence. PCBP2, a homolog of PCBP1 with over 80% identical amino acids, was not detected by the MALDI-TOF-MS analysis. Among prominent peaks in Figure 3B, 16 peaks, marked by numbers, matched with molecular weights of either tryptic fragments of PCBP1 (12 peaks) or those of modified PCBP1 (four peaks). Among the 12 peaks corresponding to unmodified PCBP1 fragments, four peaks correspond to common fragments of PCBP1 and PCBP2 and eight peaks correspond to PCBP1-specific fragments. PCBP2-specific protein fragments were not identified by the MS-Fit peptide mass search program. This indicates that not enough quantity of PCBP2 protein exists at the p40 band in Figure 3A, which represents the purification by affinity chromatography described in Materials and Methods. Based on these findings, the 40 kDa protein was assigned as PCBP1 (Fig. 3C). The identities of other proteins were determined similarly by MALDI-TOF-MS. Polypeptides p37, p45 and p65 were assigned as heterogeneous nuclear ribonucleoprotein A/B (28), Y box-binding protein I (29,30) and heterogeneous nuclear ribonucleoprotein K (29,30), respectively. Our research was focused on PCBP1, since binding of PCBP1, with molecular mass of 40 kDa, to HRV IRES RNA was specifically inhibited by the CT-oligomer DNA that inhibits HRV IRES activity (Fig. 2B).

Figure 3.

Identification of the 40 kDa protein. (A) Affinity purification of proteins interacting with oligo-CT was performed using biotinylated oligo-CT and HeLa cell extracts. Resin-bound proteins were resolved by 12% SDS–PAGE and the 40 kDa protein band was excised. The isolated protein was trypsinized and analyzed by MALDI-TOF-MS. (B) Peptide mixtures obtained by in-gel digestion of the excised band with trypsin were analyzed by MALDI-TOF-MS. The Swiss-Prot database was used to identify the corresponding protein. Peptide masses marked with arrowheads were matched with the calculated tryptic peptide masses of PCBP1. (C) Polypeptide fragments in PCBP1 were matched with tryptic peptide masses. Matching sequences are underlined.

PCBP1 interacts with the IRES elements of HRV and PV

We performed in vitro RNA binding assays to confirm that PCBP1 interacts with the IRES elements of PV and HRV RNA. PCBP2, a homolog of PCBP1, is reported to interact with stem–loop IV RNA of PV IRES (31). Accordingly, RNAs corresponding to stem–loop IV of PV and HRV and HCV IRESs (nt 18–402 of HCV genome) were employed in RNA–protein interaction assays. For affinity chromatography, biotinylated RNAs were generated by in vitro transcription. Recombinant PCBP1 and PCBP2 proteins were expressed in E.coli and purified. Following incubation of PCBP1 and PCBP2 with biotinylated RNA, RNA–protein complexes were precipitated with streptavidin–agarose beads. The presence of PCBP1 and PCBP2 proteins in these complexes was determined by western blot analysis with anti-PCBP1 and anti-PCBP2 antibodies, respectively. PCBP1 and PCBP2 were detected in precipitated resins associated with stem–loop IV elements of HRV and PV IRES elements (Fig. 4, lanes 4, 5, 9 and 10). PCBP2 precipitated with HCV RNA, albeit weakly compared with PV and HRV (Fig. 4, lane 8). However, PCBP1 did not co-precipitate with HCV RNA (Fig. 4, lane 3). The data imply that both PCBP1 and PCBP2 interact directly with the IRES elements of HRV and PV, and that PCBP2, but not PCBP1, interacts weakly with the HCV IRES element.

Figure 4.

PCBP1 and PCBP2 interact with the IRES elements of HRV and PV. To investigate whether PCBP1/2 interacts with stem–loop IV RNA of HRV and PV IRES elements, RNA pull-down experiments were performed. Recombinant PCBP1 or PCBP2 was incubated in the absence of RNA (lanes 2 and 7) or in the presence of biotinylated stem–loop IV RNA of HRV (lanes 3 and 8) and PV (lanes 4 and 9), and biotinlyated full-length IRES RNA of HCV (lanes 5 and 10). After incubation of biotinylated RNA and PCBP1 or PCBP2, samples were subjected to streptavidin–agarose resin adsorption. After washing the resin, bound proteins were resolved by 12% SDS–PAGE. PCBP1 (lanes 1–5) or PCBP2 (lanes 6–10) was visualized by immunoblot analysis with anti-PCBP1 or anti-PCBP2 antibody.

PCBP1/2 is required for efficient translation through the HRV and PV IRES elements

Depletion and repletion of PCBP1 and PCBP2 was performed to examine whether the inhibitory effects of the CT-oligomer on PV and HRV IRES activities are due to sequestration of PCBP1 and/or PCBP2 through protein–DNA interactions. PCBP1 and PCBP2 proteins are abundant in HeLa cell extracts (Fig. 5A, lane 1; Fig. 5B, lanes 1 and 3). A concentration of 5 µg/mg PCBP proteins was estimated in HeLa cell extracts by western blot analyses (data not shown). Endogenous PCBP1 and PCBP2 proteins (see below) in HeLa extracts were depleted using an oligo CT column. Most oligo-CT-binding proteins were removed by oligo-CT column chromatography, as confirmed by UV cross-linking experiments using ³²P-labeled oligo-CT (Fig. 5A, compare lanes 1 and 2). Depletion of PCBP1 and PCBP2 proteins was confirmed by western blot analysis using specific antibodies. Less than 5% PCBP1 and PCBP2 remained in oligo-CT-treated HeLa cell extracts, compared to mock-treated HeLa extracts (Fig. 5B, compare lanes 2 and 4 with lanes 1 and 3). Both PCBP1 (identified by MALDI-TOF analysis) and PCBP2 were removed by oligo-CT resin treatment. Simultaneous depletion of PCBP1 and PCBP2 is likely due to similar, but not the same, nucleic acid binding preferences of PCBP1 and PCBP2, as depicted in lanes 3–5 and 8–10 in Figure 4. It should be noted that an excess of oligo-CT resin was used in the depletion experiment.

Figure 5.

Depletion of PCBP1/2 in HeLa cell extracts. (A) UV cross-linking experiments were performed with HeLa cell extracts (10 µg proteins in total) and ³²P-labeled CT-oligomer (2 × 10⁵ c.p.m.). HeLa extracts treated with streptavidin agarose were loaded onto (A) lane 1 and (B) lanes 1 and 3. Extracts treated with biotinylated oligo-CT and streptavidin–agarose were loaded onto (A) lane 2 and (B) lanes 2 and 4. After UV irradiation, samples were resolved by 12% SDS–PAGE. (B) Western blot analysis of PCBP1/2- and mock-depleted HeLa extracts using anti-PCBP1 (lanes 1 and 2) and anti-PCBP2 antibody (lanes 3 and 4). HeLa cell extracts were resolved by SDS–PAGE and transferred to a nitrocellulose membrane for immunoblot analysis.

To evaluate the roles of PCBP1 and PCBP2 in IRES-dependent translation, we performed in vitro translation reactions in PCBP1/2-depleted HeLa cell extracts. Translational efficiencies of HRV and PV mRNAs were dramatically reduced by up to 90% (Fig. 6A, lanes 1–4). On the other hand, translation of HCV RNA was affected only marginally, if at all, by PCBP1/2 depletion (Fig. 6A, lanes 5 and 6). Translational efficiencies of PV and HRV mRNAs were restored by the addition of recombinant PCBP1 or PCBP2 proteins in a dose-dependent manner (Fig. 6B and C). On the other hand, depletion and repletion of PCBP1 or PCBP2 protein did not affect the translational efficiency of HCV mRNA (Fig. 6A, lanes 5 and 6; Fig. 6D, lanes 1–11).

Figure 6.

PCBP1/2 enhances IRES-dependent translation of PV and HRV mRNAs. (A) The effect of PCBP1/2 depletion on IRES-dependent translation. In vitro translation reactions were conducted in PCBP1/2-depleted (lanes 2, 4 and 6) and mock-depleted (lanes 1, 3 and 5) HeLa cell extracts using dicistronic mRNA shown in Figure 1A. The mRNAs used in translation reactions are indicated by the IRES elements in the transcripts. IRES-dependent translation (measured by Fluc activity) was normalized by cap-dependent translation (measured by Rluc activity). The bars signify the relative IRES activity of each mRNA in PCBP1/2-depleted HeLa extracts (lanes 2, 4 and 6) compared to that in mock-depleted HeLa extracts (lanes 1, 3 and 5). (B) Effect of PCBP1/2 on PV IRES function. PCBP1/2 depletion and repletion experiments were performed using dicistronic PV mRNA as a reporter. Lane 1 depicts PV IRES activity in mock-depleted HeLa extracts. Lanes 2 and 7 depict the relative PV IRES activity in PCBP1/2-depleted HeLa extracts compared to mock-depleted HeLa extracts. Purified recombinant PCBP1 (25, 50, 100 and 200 ng) (lanes 3, 4, 5 and 6) and PCBP2 (lanes 8, 9, 10 and 11) were supplemented to investigate the effects of PCBP1 and PCBP2 on translation. (C) Effect of PCBP1/2 on HRV IRES function. PCBP1/2 depletion and repletion experiments were performed using dicistronic HRV mRNA as a reporter. The order of lanes is the same as that in (B). (D) The effect of PCBP1/2 on HCV IRES function. PCBP1/2 depletion and repletion experiments were performed using dicistronic HCV mRNA as a reporter. The order of lanes is the same as that in (B). (E) Additive effect of PCBP1 and PCBP2 on PV IRES function. PCBP1/2 depletion and repletion experiments were performed using dicistronic PV mRNA as a reporter. Lane 1 (black bar) depicts PV IRES activity in mock-depleted HeLa extracts. Lane 2 (white bar) depicts the relative PV IRES activity in PCBP1/2-depleted HeLa extracts compared to mock-depleted HeLa extracts. Purified PCBP1 (50 ng for lanes 3 and 5, 100 ng for lane 6) and/or PCBP2 (50 ng for lanes 4 and 5, 100 ng for lane 7) were supplemented to investigate the effects of combined treatment of PCBP1 and PCBP2 on translation. The increased levels of translation by the addition of PCBP1 and/or PCBP2 are shown as gray bars.

Synergistic effect of PCBP1 and PCBP2 on augmenting PV IRES function was monitored using dicistronic PV mRNA RPF (Figs 1B and 6E). PCBP1 and PCBP2 were added in PCBP-depleted translation mixtures separately or together as depicted in the top panel of Figure 6E. Fifty nanograms of PCBP1 and PCBP2 restored translation of PV mRNA by 20 and 38%, respectively (Fig. 6E, bottom panel). And a mixture of PCBP1 and PCBP2, containing 50 ng of each protein, restored translation of PV mRNA by 55%. This is very close to the sum (58%) of the effects of PCBP1 and PCBP2 added separately. This indicates that PCBP1 and PCBP2 augment PV IRES function additively not synergistically and that the mode of action of these two proteins is likely to be the same. We speculate that the discrepancy in activities of these proteins is attributed to binding preference of the proteins for IRES elements.

Removal of other proteins, which may augment IRES-dependent translation through the PCBP depletion process, was monitored by western blot analysis using antibodies against hnRNP C, hnRNP Q, La autoantigen and PTB. The level of hnRNP C, hnRNP Q and La autoantigen remained the same after PCBP depletion, but PTB was partially removed by the depletion process (data not shown). Nevertheless, translational efficiencies of PV and HRV mRNAs were fully restored by addition of either PCBP1 or PCBP2 (Fig. 6B and C). Moreover, addition of purified PTB in the PCBP-depleted translation mixture did not augment translation of PV and HRV mRNAs (data not shown). This indicates that the remaining PTB in the PCBP-depleted translation mixture is sufficient for augmenting IRES-dependent translation of PV and HRV mRNAs.

DISCUSSION

Translation of numerous viral and cellular mRNAs is directed by IRES elements where non-canonical translation factors bind for IRES function. To elucidate the molecular basis of IRES-dependent translation, it is necessary to identify and characterize the proteins interacting with IRES elements. Here we report that both PCBP1 and PCBP2 bind to stem–loop IV of PV and HRV IRES elements and augment translation. In this study, we have devised a strategy to identify proteins activating IRES elements through RNA–protein interactions. Unlike previous methods that identify cellular proteins interacting with IRES elements without information on the functionality of the proteins in IRES-dependent translation, our procedure includes a step that identifies short oligo-DNAs inhibiting activity of a specific IRES element. Next, interacting cellular proteins are identified by affinity chromatography using the oligomer, and purified proteins are subsequently analyzed using MALDI-TOF-MS. The advantages of this strategy are as follows. Firstly, proteins affecting a particular IRES activity are identified with much higher precision, compared with IRES RNA affinity chromatography where RNA spanning a large part of an IRES element (for example, stem–loop IV of PV IRES) is used due to lack of information on the protein-binding site. On the other hand, only a limited number of proteins interact with small oligo-DNA inhibiting a specific IRES function (compare lane 1 in Figs 2A and B). Secondly, the biochemical activity of the oligo-DNA binding protein in IRES-dependent translation is easily investigated by using the specific oligo-DNA that inhibits IRES function. As shown in Figure 5, depletion of proteins specifically interacting with oligo-DNA is accomplished by incubating HeLa cell extracts with biotinylated DNA and subsequent treatment with streptavidin resin. The role of the oligo-DNA-binding protein in translation may be confirmed by adding back the depleted protein into the in vitro translation mixture pretreated with biotinylated DNA (Fig. 6A–D).

The depletion of proteins from an in vitro translation system by oligo-DNA is very specific, since only a few proteins interact with short oligo-DNA compared with protein depletion with RNA spanning a large region of an IRES element. In this respect, it is worth comparing depletion and repletion experiments performed with homopolymeric RNA resin, reported by Walter and colleagues (32), with the oligo-DNA resin experiments described in this report. Walter et al. (32) showed that PCBP2, but not PCBP1, restored PV IRES-dependent translation when added to PCBP-depleted HeLa cell extracts prepared with poly(rC) resin. On the other hand, ITAF activities of both PCBP1 and PCBP2 on PV IRES were suggested by Gamarnik and Andino (31). The authors showed that antibodies against PCBP1 and PCBP2 blocked PV IRES function in Xenopus oocytes and that co-introduction of oligo-peptides specific to the antibodies restored PV IRES activity. However, depletion and repletion of PCBP1 was not shown in the paper. In this report, we show that both PCBP1 and PCBP2 restore PV and HRV IRES-dependent translation in PCBP-depleted HeLa cell extracts prepared with an oligo-DNA resin (Fig. 6). The discrepancy in the depletion and repletion results is possibly due to the difference in the PCBP depletion step. We speculate that a putative cellular factor required for PCBP1, but not PCBP2, activity on PV IRES function is removed by poly(rC) resin, but not by oligo-DNA treatment. This is a plausible explanation, since a significantly higher number of proteins bind homopolymeric RNA, compared with short oligo-DNA. Moreover, several IRES elements need multiple RNA-binding proteins for function (21,33).

Very recently it has been shown that PCBP1, PCBP2 and hnRNP K, all of which contain KH motifs, augment IRES-dependent translation of c-myc mRNA (34). HnRNP K protein, at least in part, must have been removed by the depletion process of PCBP through an interaction with PCBP (see Figs 2A and 3A). Nevertheless, translation of polioviral and rhinoviral mRNAs was fully restored by the addition of PCBP protein. This may suggest that hnRNP K is not essential for translation of these mRNAs. Alternatively, residual hnRNP K in the PCBP-depleted cell extract was sufficient for supporting translation of these mRNAs. The role of hnRNP K in IRES-dependent translation of polioviral and rhinoviral mRNAs remains to be investigated.

An obvious limitation of this strategy is that only the IRES activation factors interacting with short oligo-DNA can be identified. Several RNA-binding proteins interact with DNA. For instance, hnRNP A1 containing two RNA-recognition motifs (RRM) and hnRNP K with three K homology (KH) domains interact with telomere DNA (35) and c-myc promoter sequences, (36) respectively. Since many RNA-binding proteins contain either RRM or KH domains, it is not unusual to identify short DNAs interacting with RNA-binding proteins. A reasonable strategy to identify DNAs interacting with ITAFs is to synthesize oligo-DNAs with sequences corresponding to cis-acting elements in the IRES elements determined by functional analysis, and examine the effects of these sequences on IRES activity.

Investigation of cellular proteins inhibiting IRES function is also possible by using the oligonucleotide competition approach described here. We found an oligonucleotide specifically enhancing XIAP IRES activity (data not shown). It is likely that the oligonucleotide blocks the inhibitory activity of a protein by sequestrating it from translational machinery. Investigation into the putative translational inhibitor is in progress. We speculate that the oligonucleotide competition strategy can be applied to the investigations of various molecular mechanisms involved RNA-binding proteins.