Internal ribosome entry site-mediated translation of Smad5 in vivo: requirement for a nuclear event

Abstract

Smad5 is thought to relay signals of the bone morphogenetic protein pathway. The 5′ untranslated region (5′UTR) of human Smad5 mRNA is long, has the potential to form secondary structures and contains five AUG codons. Here we show that the 5′UTR of Smad5 contains an internal ribosome entry site (IRES) located within 100 nt of the 3′ end of the 5′UTR. The Smad5 IRES was 4–8-fold more active than the poliovirus IRES in C2C12 cells, which have osteoblastic differentiation ability, but was 5–10-fold less active than the poliovirus IRES in 293T cells. When an in vitro transcript of a dicistronic Smad5 IRES construct was transfected into C2C12 cells, the Smad5 IRES was not able to stimulate the translation of the downstream cistron, although the cap-dependent translation of the upstream cistron was efficient. In contrast, the poliovirus IRES in a dicistronic in vitro transcript was able to stimulate the translation of the downstream cistron to a similar extent as in the case of transfection of the corresponding dicistronic DNA construct. These results suggest that Smad5 IRES activity displays cell specificity and that some as yet unidentified nuclear event may be required for efficient Smad5 IRES-driven translation initiation.

INTRODUCTION

Translation of most eukaryotic mRNAs is initiated from the cap structure that is commonly found at their 5′ ends. Accordingly, eukaryotic mRNAs are usually monocistronic. This is in contrast to prokaryotic mRNAs, which are often polycistronic, containing multiple internal ribosome binding sites (Shine–Dalgarno sequences). Eukaryotic RNA sequence elements that can function as internal ribosome entry sites (IRESs) were first discovered in picornavirus and later also in some other viral RNAs (1–5). Recently a number of IRESs have been identified in several eukaryotic mRNAs whose protein products are involved in cell growth or cell death (1,3,6–10). These include mRNAs encoding vascular endothelial growth factor, fibroblast growth factor-2, platelet derived growth factor (PDGF), c-myc, N-myc, apoptotic protease activating factor-1 (Apaf-1) and death-associated protease 5 (11–17). These cellular IRES elements, in their natural capped mRNAs, direct translation initiation in cells under normal physiological conditions in which cap-dependent translation is suppressed. During mitosis, cap-dependent translation is shut down, but translation of some proteins important for cell cycle progression continues. Recently, it was found that mRNAs encoding ornithinine decarboxylase and protein kinase P58^PITSLRE (and c-myc) contain IRESs that are regulated at G₂/M phase (7–9). Similarly, under conditions of cell stress and apoptosis, cap-mediated translation is impaired (10). Nevertheless, certain mRNAs such as X-linked inhibitor of apoptosis and c-myc continue to be translated under these conditions (18,19). During cell development or differentiation, IRES-dependent translation is active. The activity of the IRESs of some Drosophila proteins (Antennapedia and Ultrabithorax) and c-myc is developmentally controlled (20,21), and the PDGF2/c-sis IRES is active during megakaryocytic differentiation of K562 cells (13). Eukaryotic initiation factor 4G (eIF4G), La autoantigen and immunoglobulin heavy chain binding protein (Bip) are thought to be subject to cap-independent translation when cap-dependent translation is suppressed since their mRNAs contain IRESs (22–24).

Smad5 is a member of a highly conserved protein family of intracellular mediators of TGF-β/bone morphogenetic protein (BMP) signaling (25). Smad5 has a key role in the BMP-2-induced osteoblastic differentiation of C2C12 mesenchymal cells (26,27). Smad5 is directly activated by BMP type Ia or Ib receptors upon BMP-2 stimulation. Following serine phosphorylation, Smad5 binds to Smad4 and the complex is translocated to the nucleus (27). Two Smad5 mRNAs contain different 5′ untranslated regions (5′UTRs) linked to the same Smad5 coding region. The longer corresponding cDNA, cDNA-1, contains an UTR of 357 nt with a methionine start codon at the 5′ end and an open reading frame of 465 amino acids. The shorter corresponding cDNA, cDNA-2, is identical to cDNA-1 except that it lacks 75 nt in exon 2 (25). Both messages contain a long 5′UTR, which has the potential to form secondary structures and contains five AUGs upstream of the start codon. These facts led us to examine whether the translation initiation of Smad5 may occur by a cap-independent internal ribosome entry mechanism.

In this study we investigated the full-length 5′UTR of Smad5 and showed that it contains an IRES in its 3′ region. The Smad5 IRES activity was low in human 293T or HeLa cells and high in C2C12 cells. Smad5-IRES-mediated translation may require in vivo transcription in the cell nucleus, in contrast to the poliovirus IRES. This is the first example of the identification of an IRES in a gene whose protein product has a key function in osteoblastic differentiation.

MATERIALS AND METHODS

Cells

The mouse myoblast C2C12 cell line was cultured in DMEM with 15% fetal calf serum (FCS) (26). HeLa S3 cells were grown in DMEM with 5% newborn calf serum. 293T, HepG2, glyoma, Cos7, NIH3T3 and MC3T3E1 cells were cultured in DMEM supplemented with 10% FCS. ATDC5 cells were grown in DMEM and F12 (1:1) supplemented with 5% FCS. All cell lines were grown at 37°C in an atmosphere of 5% CO₂. ATDC5 and MC3T3E1 cells were purchased from the Cell Bank of Riken.

Construction of plasmid DNAs

The Smad5 5′UTR was amplified from the cDNA clone HEPD2282 (corresponding to the region between nucleotide positions 1 and 306 of plasmid pME12S-FL3) by PCR using sense primer 5′-TGCACAAGCTTAATAAAGTTGCAGCGAGGA-3′ and antisense primer 3′-TCTTCCATGGTTGACACAAATCTTCGGAGA-5′. To construct monocistronic Smad5 IRES (mnSm5-WT), the amplified PCR products were digested with HindIII and NcoI, purified by gel electrophoresis and cloned into the HindIII–NcoI site of pGL3 Control Vector (Promega). The nucleotide sequence of the Smad5 IRES was confirmed by DNA sequencing. Deletion mutants of the Smad5 IRES (mnSm5-dl1, -dl2, -dl3, -dl4) (Fig. 1) were constructed by the same method using sense primers 5′-GCACAAGCTTAAAGGAAGCTGTTGAAGTTA-3′ (dl1), 5′-GCACAAGCTTGACTTGACCCAATGAAAGAA-3′ (dl2), 5′-GCACAAGCTTCTTCTGCTTAGGACCTGTGT-3′ (dl3) or 5′-TCTTCCATGGTTCCAATTAAAAAGGGAGGA-3′ (dl4), respectively. For construction of the reverse full-size IRES (mnSm5-R), sense primer 5′-CGTGAAGCTTTTGACACAAATCTTCGGAGA-3′ and antisense primer 5′-TAGACCATGGTGCTAATAAAGTTGCAGCGA-3′ were used. To prepare the dicistronic constructs (diSm5-WT, -dl1, -dl2, -dl3, -dl4, -dlR), the Smad5 IRES isolated from each mnSm5 construct was cleaved with HindIII and ligated with XbaI linker. The ligation products were cleaved with XbaI and inserted into the XbaI site of pRL-CMV Vector (Promega). For construction of diSm5-WT-H, an inverted repeat element was introduced into pRL-CMV vector by inserting the synthetic oligonucleotides 5′-pCTAGCTGAACTGGGAGTGGACACCTG/5′-pCTAGCAGGTGTCCACTCCCAGTTCAG into the NheI site 10 nt upstream of the start codon of the Rluc gene according to the procedure reported for the FMR1 IRES (28). The resulting plasmid was designated pRL-CMV-H. Smad5 IRES was isolated from the mnSm5-WT construct by the same procedure as described above and inserted into the XbaI site of pRL-CMV-H vector, to produce hdiSm5-WT.

The poliovirus IRES (nucleotides 1–743) was amplified from a poliovirus type 1 cDNA clone (pOM) (29) by PCR using sense primer 5′-TGCACAAGCTTTTAAAACAGCTCTGGGGTT-3′ and antisense primer 3′-TCTTCCATGGTATGATACAATTGTCTGATT-5′. mnPV-IRES and diPV-IRES were constructed using the same procedures used for construction of mnSmad5 and diSmad5. Plasmid pEF321-2A (pPV-2A) was constructed by inserting the poliovirus 2A coding region (nucleotides 3386–3832) between the EcoRI and NotI sites of pEF321 vector (30). The 2A-coding region was amplified using the following sense and antisense primers from poliovirus type 1 (pOM) cDNA (29): the sense primer corresponded to nucleotides 3386–3403 (GCATCGGAATTCGCCATGGGATTCGGACACCAAAACAAAG) and the antisense primer to nucleotides 3832–3815 (ACGTGCGGCCGCTCATTATTGTTCCATGGTTCTTCTTCG). pcDNA3-4G^Nt was constructed by ligating the HindIII–XhoI fragment of pSP36T-4G^Nt (31) between the HindIII and XhoI sites of pcDNA3.1 (Invitrogen).

Enzymatic assays of reporter protein

Subconfluent 293T, C2C12 or other cells in 12-well plates were transiently transfected with 1 µg/well monocistronic or dicistronic constructs using SuperFect Reagent (Qiagen) according to the manufacturer’s protocol. After incubation for 27 h, Passive Lysis Buffer (Promega) was added to the cells. The activity of both firefly and Renilla luciferase in the lysates prepared from transfected cells was measured using a Dual-Luciferase Reporter Assay System (Promega) according to the manufacturer’s protocol. Light emission was quantified using a luminometer (Lumicounter 700, Microtech Niti-On). All experiments were performed in duplicate on three or more independent occasions. The Luc⁺ and Rluc activity were expressed as the light units measured using a luminometer. The IRES activity was expressed as the ratio of downstream cistron (Luc⁺) expression to upstream cistron (Rluc) expression (Luc⁺/Rluc). In cases where monocistronic constructs were used, the Rluc expression of co-transfected pRL-CMV vector was used as a transfection control, and IRES activity was expressed as the ratio of Luc⁺ expression to Rluc expression (Luc⁺/Rluc).

RNA isolation and northern analysis

Subconfluent 293T and C2C12 cells in 6-well plates were transfected with 2 µg/well dicistronic IRES constructs using SuperFect Reagent. After 16 h, the total RNAs were extracted using an RNeasy Mini Kit (Qiagen). Aliquots of 10 µg of total RNAs were separated in 1% agarose gels in the presence of formaldehyde and MOPS buffer and blotted onto Hybond-N+ (Amersham Pharmacia Biotech) membranes. The blots were hybridized with a ³²P-labeled firefly luciferase DNA probe (632 bp) which was isolated from pGL3 by cleaving with HindIII and XbaI and labeled using an Exo-free Klenow Type Random Primer DNA Labeling Kit Version 2 (TaKaRa).

In vitro transcription and cell-free translation

Template RNA was transcribed from diSm5-WT, -dl3, -R and diPV-IRES constructs linearized with ClaI in the presence of a cap structure analog using a mMESSAGE mMACHINE^TM T7 KIT (Ambion). The RNAs were purified using a MicroSpin^TM S-400 HR Column (Amersham Pharmacia Biotech). These RNAs were used for transfection of cultured cells, injection into fertilized Xenopus eggs or Xenopus oocytes and in vitro translation in a HeLa S10 extract or a rabbit reticulocyte lysate system (RRL). For addition of the cell extract to in vitro translation systems, C2C12 cell extract was prepared by homogenizing 10⁸ C2C12 cells in hypotonic buffer [10 mM HEPES–KOH, 15 mM KCl, 1.5 mM Mg(CH₃COO)₂, 6 mM SH], followed by centrifugation. The HeLa cell extract used was HeLa S100 prepared from HeLa S10 (32,33).

The preparation of HeLa S10 or PV-HeLa S10 (poliovirus infected HeLa S10) from suspension-cultured HeLa cells or poliovirus-infected HeLa suspension cells and the translation reaction in HeLa S10 and RRL were performed as described previously (32,33). Briefly, each reaction mixture consisted of 0.5 µg of RNA, 170 mM K(CH₃COO), 1.5 mM Mg(CH₃COO)₂, and 200–300 µg of HeLa S10 in a total volume of 12.5 µl. The reaction was carried out at 30°C for 1 h. Flexi^TM Rabbit Reticulocyte Lysate System (Promega) was used for translation in a reaction mixture consisting of 12.5 µl of RRL, 0.5 µg of RNA, 170 mM K(CH₃COO), 1.5 mM Mg(CH₃COO)₂. The reaction was carried out at 30°C for 1 h. Sometimes, 40 µg C2C12 or 50 µg HeLa cell extract was added to the reaction mixture containing RRL. Fertilized eggs were obtained by in vitro fertilization as described elsewhere (34). Each egg was injected with 0.6 ng of RNA and incubated at 21°C for 3 h, after which four pools of six eggs were collected, and cell lysates were prepared by homogenization of the eggs, followed by centrifugation and collection of the supernatants. Oocytes were obtained as described elsewhere (31). Each oocyte was injected with 40 µg C2C12 cell extract or buffer and with 5 ng of RNA and incubated at 21°C for 2 h, after which two sets of five oocytes were collected. The preparation of cell lysates and the luciferase assays were performed as described above. The transfection of mRNA into C2C12 cells was performed by using DMRIE-C Reagent (Invitrogen) and the cells were incubated for 12 h at 37°C. The luciferase activities in the translation reaction mixtures containing HeLa S10 or RRL, and in the lysates of the eggs or of the cells were measured as described above.

RESULTS

The 5′UTR of Smad5 mRNA facilitates internal ribosome entry

A monocistronic construct (mnSm5-WT), the Smad5 5′UTR upstream of the firefly luciferase reporter was transfected into 293T or C2C12 cells together with pRL-CMV vector, and the luciferase activity in the cell extracts was measured (Fig. 2A). Co-expression of Renilla luciferase provided a control for the transfection efficiency. Transfection of a construct with the Smad5 5′UTR upstream of the firefly luciferase reporter resulted in a significant decrease in the amount of luciferase produced compared with that produced in cells transfected with the control vector lacking the Smad5 5′UTR. In cells transfected with a construct containing the poliovirus IRES upstream of the luciferase reporter, the luciferase was efficiently produced (Fig. 2A and B). A high degree of secondary structure in the 5′UTR of Smad5 mRNA is predicted by GENETTYX-MAC analysis (data not shown). This suggests that the presence of a stable secondary structure in the Smad5 5′UTR inhibited scanning. Smad5 translation may be initiated by an alternative mechanism, e.g. internal ribosome entry. An analogous finding was reported for Apaf-1 IRES (16).

Figure 2.

The IRES activities of the Smad5 and poliovirus 5′UTRs. Monocistronic construct and pRL-CMV vector were co-transfected into (A) 293T or (B) C2C12 cells. The luciferase activities were assayed 27 h later using a Dual-Luciferase Reporter Assay System (Promega) as described in Materials and Methods. Luc⁺ or Rluc activity is expressed as light units measured using a luminometer. IRES activity is expressed as the ratio of Luc⁺ expression to Rluc expression (Luc⁺/Rluc). PV, mnPV-IRES; Sm, mnSm5-WT; R, mnSm5-R; G, pGL3. The dicistronic construct was transfected into (C) 293T or (D) C2C12 cells. After 27 h, the luciferase activities were assayed as described above. IRES activity is expressed as the ratio of downstream cistron expression to upstream cistron expression (Luc⁺/Rluc). PV, diPV-IRES; Sm, diSm5-WT; hS, hdiSm5-WT (diSm5-WT with hairpin structure at upstream of the Rluc gene); R, diSm5-R; di33, di33-IRES. All experiments were performed in duplicate on three to five independent occasions. Bars represent the average + SD from three to five independent transfections.

To test whether the Smad5 5′UTR can mediate translation initiation by internal ribosome entry, we constructed a vector that could direct the expression of dicistronic mRNA containing the coding regions for Renilla and firefly luciferases as the first and second cistrons respectively (Fig. 1B). The expression of Renilla luciferase provided an internal control for the steady state concentration of the dicistronic RNA. Transfection of diSm5-WT into 293T cells resulted in a low level of expression of firefly luciferase (Fig. 2C). However, the amount of firefly luciferase produced in C2C12 cells was high (Fig. 2D). In C2C12 cells, the Smad5 5′UTR was 4–8-fold more active than the poliovirus IRES, while in 293T cells, the Smad5 5′UTR was 5–10-fold less active than the poliovirus IRES. These results suggest that the Smad5 5′UTR contains an IRES that can function in C2C12 cells. To rule out the possibility that the IRES-directed translation of the Luc⁺ cistron was due to re-initiation at the Rluc cistron, two new bicistronic constructs, hdiSm5WT and diSm5-R (Fig. 1), were analyzed. Comparison of diSm5-WT and hdiSm5-WT showed that production of Renilla luciferase activity was inhibited by the hairpin structure, whereas production of Smad5 IRES-directed firefly luciferases activity was hardly affected. When Smad5 IRES was inverted in diSm5-R, the production of firefly luciferases activity was strongly inhibited in C2C12 cells (Fig. 2C and D). These results indicate that the two cistrons were independently translated and exclude the possibility of reinitiation at the Rluc region. This was confirmed by the effect of poliovirus 2A protein on the luciferase-producing activity of diSm5-WT (see Fig. 8).

Figure 1.

The constructs containing the Smad5 or poliovirus 5′UTR. A schematic representation of the (A) monocistronic and (B) dicistronic constructs containing the 5′UTR used for this work. The monocistronic construct was cloned in pGL3 Control Vector (Promega). This monocistronic construct was transfected into cells together with pRL-CMV Vector (Promega) as a transfection control. The dicistronic construct was cloned in pRL-CMV Vector. hdiSm5-WT is a hairpin-containing diSm5-WT. diSm5-R contains the Smad5 5′UTR in reverse orientation. (C) List of the constructs containing the 5′UTRs of poliovirus and of wild-type (WT) Smad5 and deletion (dl) mutants of Smad5 used for this work.

Figure 8.

Effect of poliovirus 2A on poliovirus or Smad5 IRES activity. The dicistronic construct with or without pEF321-2A(PV-2A) was transfected into (A) 293T or (B) C2C12 cells. After 27 h, the luciferase activities were assayed as described in the legend for Figure 2. Luc⁺ or Rluc activity is expressed as light units measured using a luminometer. IRES activity is expressed as the ratio of Luc⁺ expression to Rluc expression (Luc⁺/Rluc). PV, diPV-IRES; Sm, diSm5-WT. Bars represent the average + SD from four independent transfections. White bars, Luc⁺ activity; diagonal striped bars, Rluc activity; black bars, Luc⁺/Rluc ratio; grey bars, 2A+/2A– ratio

Mapping of the Smad5 IRES by deletion in the Smad5 mRNA 5′UTR

To map the boundaries of the IRES element located in the 5′UTR of the Smad5 mRNA, the translation of second cistrons preceded by Smad5 5′UTRs with specific deletions (Fig. 3A) was tested in 293T and C2C12 cells transiently transfected with plasmid constructs that expressed monocistronic or dicistronic mRNAs. Figure 3B–E shows that gradual removal of the first 208 nt from the 5′ end of Smad5 (diSm5-dl1, -dl2, -dl3) resulted in a gradual increase of IRES activity in C2C12 cells. Removal of nucleotides 208–307 (diSm5-dl4) markedly reduced the IRES activity. These results show that only the 98 nt located at the 3′ end of the Smad5 5′UTR sequence are required for the IRES activity. This region is very short and similarly some cellular IRES activities located in the short region were reported, La1 (80 nt), La1′ (100 nt) and FMR (21 nt) (21,26). The increased IRES activity of Smad5-dl3 may be dependent on the secondary structure of Smad5-dl3 IRES. The secondary structure gradually becomes simpler as the deletion of the Smad5 5′UTR proceeds. These results also suggest that both the mRNAs corresponding to Smad5 cDNA-1 and cDNA-2 (25) contain a functional IRES.

Figure 3.

Deletion mapping of Smad5 IRES by deletion in the Smad5 mRNA 5′UTR. (A) Schematic representation of Smad5 5′UTR, of wild-type (WT) and deletion (dl) mutants. For each construct, the portion of the 5′UTR included is indicated. Monocistronic constructs together with pRL-CMV control vector (B and C) or dicistronic constructs (D and E) were transfected into 293T (B and D) or C2C12 (C and E) cells. After 27 h, the luciferase activity was measured as described in the legend for Figure 2. All experiments were preformed in duplicate on three independent occasions. Bars represent the average + SD of three independent transfections. PV, mnPV-IRES (B and C) and diPV-IRES (D and E); WT, mnSm5-WT (B and C) and diSm5-WT (D and E); dl1, mnSm5-dl1 (B and C) and diSm5-dl1 (D and E); dl2, mnSm5-dl2 (B and C) and diSm5-dl2 (D and E); dl3, mnSm5-dl3 (B and C) and diSm5-dl3 (D and E); dl4, mnSm5-dl4 (B and C) and diSm5-dl4 (D and E); R, mnSm5-R (B and C) and diSm5-R (D and E).

Smad5 IRES activity is not due to the presence of a cryptic promoter or aberrant splicing

We next examined whether the increase in the translation of firefly luciferase protein from the RNAs shown in Figure 4A and B was the result of small amounts of functionally monocistronic transcripts generated by nuclease activity, cryptic splicing or expression of an unknown promoter in the 5′UTR of Smad5. To examine these possibilities, the lengths of dicistronic mRNAs generated in transfected cells or synthesized in vitro by T7 RNA polymerase were investigated.

Figure 4.

Northern analysis of mRNAs transcribed in vivo and in vitro. (A) A schematic representation of the dicistronic constructs and their transcripts is shown. (B) The sizes of the transcripts of diPV-IRES, diSm5-WT, diSm5-dl3 and diSm-dl4 are shown. (C) diPV-IRES (lanes 1 and 3) or diSm5-WT (lanes 2 and 4) was transfected into 293T cells (lanes 1 and 2) or C2C12 cells (lanes 3 and 4). After 16 h, total RNA was extracted and analyzed by northern blot hybridization using Luc⁺ probe [shown schematically in (A)]. An autoradiograph of the blot is shown on the left, and the arrows indicate the positions of 28S and 18S rRNAs. Staining with ethidium bromide of the same blot showed 28S (human, 5025 bases, lanes 1 and 2; and mouse, 4712 bases, lanes 3 and 4) and 18S (human, 1868 bases; mouse, 1869 bases) rRNAs (right panel). (D) diPV-IRES (lanes 1 and 2); diSm5-WT (lanes 3 and 4); diSm5-dl3 (lanes 5 and 6) or diSm5-dl4 (lanes 7 and 8) was transfected into C2C12 cells. For each construct, RNA samples from two independent experiments were analyzed. The northern blot analysis was performed as described in (C). (E) The transcripts were synthesized in vitro using linearized diPV-IRES, diSm5-WT and diSm-dl3 and a mMESSAGE mMACHINE T7 kit (Ambion). The RNAs (diPV-IRES, lane 1; diSm5-WT, lane 2; diSm5-dl3, lane 3) were separated by electrophoresis in a 1% agarose gel containing formaldehyde. The left-most lane shows RNA markers.

The cytoplasmic RNAs were isolated from the transfected 293T and C2C12 cells and analyzed by northern blot hybridization using ³²P-labeled DNA probes complementary to firefly luciferase sequences. Autoradiography of the blots showed transcripts corresponding to the full-length diSm5-WT and diPV-IRES transcripts. These results show that the dicistronic mRNA expressed in transfected cells remained intact (Fig. 4C and D). Some smaller transcripts were also detected in diPV-IRES-transfected cells (Fig. 4D, lanes 1 and 2).

Transcripts synthesized in vitro using T7 RNA polymerase were also shown to be full length (Fig. 4E). Their translational efficiencies were examined in HeLa cell extracts, RRL, fertilized Xenopus eggs, Xenopus oocytes and C2C12 cells, and the results are shown in Figures 5–7 and Table 1. Short transcripts were sometimes detected in diPV-IRES-transfected cells and samples transcribed in vitro from diPV-IRES. In contrast, no short transcripts were detected in cells transfected with diSm5-WT or diSm5-deletion mutants. These findings suggest that it is unlikely that nuclease activities, splicing or cryptic promoter elements mediated the translation of the second cistrons in the dicistronic mRNAs.

Figure 5.

Translation of the transcripts containing Smad5 or poliovirus IRES in RRL. The transcripts synthesized in vitro from diSm5-WT, diSm5-dl3, diSm5-R and diPV-IRES DNAs were translated in the Flexi^TM Rabbit Reticulocyte Lysate System. Where indicated, 40 µg of C2C12 or 50 µg of HeLa cell extract was added to the reaction mixture. After incubation for 1 h at 30°C, the luciferase activity was measured as described in the legend for Figure 2. Luc⁺ or Rluc activity is expressed as light units measured using a luminometer. IRES activity is expressed as the ratio of Luc⁺ expression to Rluc expression (Luc⁺/Rluc). All experiments were performed in duplicate on four independent occasions. PV, diPV-IRES; SmWT, diSm5-WT; Smdl3, diSm5-dl3; SmR, diSm5-R. Black bars, without cell extract; diagonal striped bars, with HeLa cell extract; gray bars, with C2C12 cell extract.

Figure 7.

Translation directed in C2C12 cells by the dicistronic transcripts containing Smad5 or poliovirus 5′UTR or the corresponding DNA constructs encoding them. C2C12 cells were transfected with the transcripts synthesized in vitro from diSm5-WT, diSm5-dl3, diSm5-R and diPV-IRES DNAs or with the DNAs themselves. The luciferase activity was measured at 18 h after RNA transfection and at 27 h after DNA transfection, as described in the legend for Figure 2. All experiments were performed in duplicate on three independent occasions. The expression of the Renilla luciferase in each cell line was taken as 1 (white bars), and firefly luciferase activity was expressed as a percentage of the Renilla luciferase activity. Bars represent the average + SD from three independent transfections. PV, diPV-IRES; SmWT, diSm5-WT; Smdl3, diSm5-dl3; SmR, diSm5-R. Grey bars, Luc⁺ activity in mRNA-transfected C2C12 cells; diagonal striped bars, Luc⁺ activity in DNA construct-transfected C2C12 cells.

Table 1. Translation of the transcripts containing Smad5 IRES in fertilized Xenopus eggs and oocytes.

	Extracta	Luc+ activity (light units ×103)	Rluc activity (light units ×107)	IRES activity (Luc+/Rluc) ×10–3
Fertilized eggs	–	3.46	2.622	0.132
	–	3.91	2.794	0.140
	–	3.67	2.646	0.138
	–	4.82	3.140	0.154
Oocytes	–	53.43	5.157	1.04
	–	60.69	6.159	0.98
	+	61.64	6.619	0.93
	+	61.51	6.624	0.93

^a+, C2C12 cell extract was added; –, no C2C12 cell extract was added.

Translation of mRNAs transcribed from dicistronic constructs in vitro and in vivo

Dicistronic IRES transcripts containing an m7GpppG cap structure at the 5′ terminus were synthesized from each of the dicistronic IRES constructs (Fig. 4A). When the diSm5-WT transcript was injected into fertilized Xenopus eggs or Xenopus oocytes, the Renilla luciferase cistron was translated efficiently, while little expression of the firefly luciferase was observed. Even when C2C12 cell extract was co-injected, the expression of the firefly luciferase was not stimulated (Table 1).

In vitro translation in RRL using the dicistronic transcript produced high Renilla luciferase activity but low firefly luciferase activity from diPV-IRES, and very low firefly luciferase activity from diSm5-WT. The production of firefly luciferase activity in diPV-IRES was stimulated 10-fold by the addition of HeLa cell extract to the RRL system. The IRES activity of diSm5-WT, diSm5-dl3 and diSm5-R was not stimulated by the addition of HeLa or C2C12 cell extract (Fig. 5). These results concerning the poliovirus IRES activity confirmed the findings of previous studies (1,5,32,33).

In vitro translation in HeLa S10 using the dicistronic Smad5 transcript produced high Renilla luciferase activity but very low firefly luciferase activity (Fig. 6). In the case of the transcript of diPV-IRES, both the firefly and Renilla luciferase activities produced in HeLa S10 were high. Furthermore, when translation was performed using PV-HeLa S10, the poliovirus IRES activity was very high (Fig. 6). This stimulation of translation in PV-HeLa S10 may have been the effect of poliovirus 2A and 3C proteins in the extract.

Figure 6.

Translation of the transcripts containing Smad5 or poliovirus IRES in HeLa S10. The transcripts synthesized in vitro from diSm5-WT, diSm5-dl3, diSm5-R and diPV-IRES DNAs were translated in HeLa S10 or PV-HeLa S10. After incubation for 1 h at 30°C, the luciferase activity was measured as described in the legend for Figure 2. Luc⁺ or Rluc activity is expressed as light units measured using a luminometer. IRES activity is expressed as the ratio of Luc⁺ expression to Rluc expression (Luc⁺/Rluc). All experiments were performed in duplicate on four independent occasions. PV, diPV-IRES; SmWT, diSm5-WT; Smdl3, diSm5-dl3; SmR, diSm5-R. Black bars, HeLa S10; diagonal striped bars, PV-HeLa S10.

To confirm the translational efficiency (Luc⁺/Rluc) of diSm5-WT, diSm5-dl3 and diSm5-R transcripts in C2C12 cells, in vitro-transcribed mRNAs were transfected into C2C12 cells and their expression was examined. Unexpectedly, the firefly luciferase cistron was translated very inefficiently, while the Renilla luciferase cistron was translated efficiently (Fig. 7). The firefly luciferase activity in mRNA-transfected cells was 100-fold lower than that in cells transfected with the corresponding distronic DNA construct. Similar results were obtained in 293T cells transfected with the same transcripts (data not shown). In C2C12 cells transfected with the diPV-IRES transcript, the level of firefly luciferase activity was similar to that in the cells transfected with diPV-IRES DNA. Thus, the Smad5 IRES could not stimulate the translation of the downstream cistron (firefly luciferase) from an in vitro-transcribed dicistronic Smad5 transcript introduced directly into the cytoplasm.

The effects of 4G^Nt and poliovirus 2A on the activity of Smad5 IRES

To examine whether the Smad5 IRES element could promote translational up-regulation in C2C12 or 293T cells in which eIF4G is cleaved, the Smad5 IRES activity was assayed in cells co-transfected with poliovirus 2A cDNA (pPV-2A). In the 2A-co-transfected 293T cells compared with 2A-non-transfected 293T cells, the Smad5 IRES activity was 2–3-fold higher, and the poliovirus IRES activity was 7-fold higher (Fig. 8A, right panel). In 2A-co-transfected C2C12 cells, the Smad5 IRES activity was 2–3-fold higher, while the poliovirus IRES activity was much higher (16–25-fold) (Fig. 8B, right panel). The Smad5 IRES activity in 2A-co-transfected C2C12 and 293T cells was stimulated as the result of a reduction in Renilla luciferase activity. These results in 2A-co-expressing C2C12 cells show that Smad5 IRES activity did not depend on ribosome scanning from the upstream Rluc cistron to the downstream Luc⁺ cistron. In pPV2A-co-transfected cells, the firefly luciferase activity of diPV-IRES was greatly stimulated and the Renilla luciferase activity was slightly reduced. The stimulatory effect of 2A protein on the poliovirus IRES activity was especially strong in C2C12 cells, in which the poliovirus IRES was not efficient. These results also show that 2A protein appears to have a somewhat specific auto-stimulatory effect on the poliovirus IRES. Both of the IRES activities in 2A-co-expressing 293T cells were similar to those in 2A-co-expressing HeLa cells. The level of stimulation of both IRES activities in 2A-co-expressing C2C12 cells were similar to those in 2A-co-expressing MC3T3 cells (data not shown), which have osteoblastic differentiation ability.

Over-expression of pcDNA3-4G^Nt (encoding the N-terminal portion of eIF4GI protein, which contains the binding sites for eIF4E and PABP) was shown previously to have a dominant-negative effect on translation initiation, implying that the cap-dependent translation may be suppressed and cap-independent translation may not be suppressed (31). To confirm this, the cells were co-transfected with diSm5-WT or diPV-IRES plus pcDNA3-4G^Nt. The translation of the upstream cistron (Rluc) in both of these IRES-containing constructs in co-transfected 293T and C2C12 cells was suppressed ∼2-fold as a result, whereas both IRES activities were stimulated ∼2-fold (Fig. 9). Consequently, the 4G+/4G– ratios of these IRES activities were ∼2-fold enhanced (Fig. 9). These results in 2A or 4G^Nt-co-expressing cells show that the Smad5 5′UTR contains an IRES, and that the IRES is active in cells containing cleaved eIF4G.

Figure 9.

Effect of 4G^Nt on poliovirus or Smad5 IRES activity. The dicistronic construct with or without pcDNA3-4G^Nt was transfected into (A) 293T or (B) C2C12 cells. After 27 h, the luciferase activities were assayed as described in the legend for Figure 2. IRES activity is expressed as the ratio of Luc⁺ expression to Rluc expression (Luc⁺/Rluc). PV, diPV-IRES; Sm, diSm5-WT. Bars represent the average + SD from three independent transfections. Black bars, Luc+/Rluc ratio; grey bars, 4G+/4G– ratio.

The Smad5 IRES is active in cells with osteoblastic differentiation ability

To investigate in what types of cell the Smad5 IRES is active, various cell lines were transfected with diSm5-WT, diSm5-R and diPV-IRES. The Smad5 IRES was active in C2C12, ATDC5 and ME3T3E1 cells, all of which have osteoblastic differentiation ability. The PV-IRES was active in 293T, glioma, HeLa and HEPG2 cells (Fig. 10). The Smad5 IRES was more active in the cells in which poliovirus IRES did not function efficiently and, conversely, the Smad5 IRES was not efficient in the cells in which poliovirus IRES functioned efficiently. These results suggest that the expression of Smad5 IRES activity is dependent on the cell type, and that the protein requirements for Smad5 IRES and poliovirus IRES function are different.

Figure 10.

A comparison of the efficiency of the Smad5 or poliovirus IRESs in various cell lines. (A) diPV-IRES, (B) diSm5-WT or (C) diSm5-R was transfected into the cells. After 27 h, the luciferase activities were determined as described in the legend for Figure 2. IRES activity is expressed as the ratio of Luc⁺ expression to Rluc expression (Luc⁺/Rluc). All experiments were performed in duplicate on two to four independent occasions. Bars represent the average + SD from two or four independent transfections.

DISCUSSION

Smad5, Smad1 and Smad8 constitute a conserved family of proteins that are intracellular mediators of BMP signaling. Smad5 has a key function in the BMP-2-induced osteoblastic differentiation of C2C12 mesenchymal cells (26,27). We showed here that the Smad5 5′UTR functioned as an IRES by analyzing a dicistronic (Rluc and Luc⁺) reporter construct in which the enhanced activity of the downstream reporter (Luc⁺) was not due to read-through from the upstream cistron (Rluc) (Fig. 2C and D and Fig. 8) nor to translation of cryptic monocistronic transcripts (Fig. 4C–E). The Smad5 IRES is located within a 100 nt region at the 3′ end of the 5′UTR. Pozner et al. reported that Runx1/AML1 contains an IRES (35). Recently, Lee et al. reported that the osteoblastic differentiation of C2C12 cells induced by BMP-2 requires the cooperation of Smad5 and Runx2/PEBP2A/Cbfa-1, and that this cooperation induces osteoblast-specific gene expression (36). As the messages of Runx2, PEBP2A2 and cbaf1 have long 5′UTRs, Runx2 mRNA may contain an IRES. These facts together suggest that cap-independent translation may occur during the osteoblastic differentiation induced in C2C12 cells by BMP-2. Currently, we are investigating the translation of Smad5 protein in C2C12 cells, in which BMP-2 induces osteoblastic differentiation.

The translation of firefly luciferase from the diSm5-WT and diSm-dl3 transcripts containing an m7GpppG cap structure was very inefficient in fertilized Xenopus eggs and oocytes, and in RRL and HeLa S10 in vitro translation systems (Table 1, Figs 5 and 6). The addition of HeLa or C2C12 cell extract to the translation system did not enhance the production of firefly luciferase activity (Table 1, Fig. 5). The transfection of dicistronic Smad5 transcripts into C2C12 cells resulted in efficient translation of Renilla luciferase, while little expression of the downstream cistron (firefly luciferase) was observed (Fig. 7). However, when the di-Smad5 DNA constructs were transfected into C2C12 cells, the downstream cistron, firefly luciferase, was translated efficiently (Figs 2D, 3E and 7). In contrast, the translation of firefly luciferase from the di-PV-IRES transcripts was efficient in both HeLa S10 in vitro and in transfected C2C12 cells. The IRES activity of the diPV-IRES transcripts in RRL was inefficient, and the activity was stimulated by the addition of HeLa cell extract. The results suggest that the Smad5 IRES cannot stimulate the translation of the downstream cistron in a dicistronic mRNA introduced directly into the cytoplasm. For efficient Smad5 internal initiation, it may be necessary for some nuclear factor(s) to interact with the Smad5 IRES transcribed in the nucleus, and for the resultant RNA–protein complex to be transported to the cytoplasm. The requirement for a nuclear event for c-myc IRES-driven translation was reported (37). It appears that the Smad5 IRES-driven translation may promote internal initiation only on transcripts expressed in the nucleus. The mechanism of this effect is unknown, and must be clarified in the future.

The Smad5 IRES is active in C2C12 and ATDC5 cells, but not in 293T, HeLa or HepG2 cells, whereas the poliovirus IRES is active in 293T, HeLa and HepG2 cells, but is not active in C2C12 cells (Fig. 10). Thus, the protein requirement for the Smad5 IRES must differ from that for the PV-IRES. La autoantigen, polypyrimidine tract binding protein (PTB) and poly(rC) binding protein 2 have been reported to be trans-acting factors necessary for the function of the PV-IRES (38–42). These proteins are RNA-binding proteins that are localized mainly in the nucleus, but La and PTB are thought to move to the cytoplasm in poliovirus-infected cells. They bind to the poliovirus RNA, and the resultant RNP complex is translated in the cytoplasm. As discussed above, the translation-competent Smad5 RNA must be transcribed in the nucleus and the interaction of some nuclear factor(s) with newly synthesized Smad5 transcripts may be necessary for the Smad5 IRES function.

When the diPV-IRES or diSm5-IRES and pPV-2A DNAs were co-transfected into cells, the IRES activity was enhanced (Fig. 8). In cells transfected with diSm5-WT or diSm5-dl3 alone, the Renilla luciferase activity was reduced while firefly luciferase activity was the same as in the cells co-transfected with pPV-2A DNA, and as a result, Smad5 IRES activity was stimulated. Poliovirus 2A protein shuts off cap-dependent translation by cleavage of eIF4G and thereby indirectly stimulates cap-independent translation (43). These results show that Smad5 IRES activity is not dependent on ribosome scanning from the upstream Rluc cistron to the downstream Luc⁺ cistron. Hambidge and Sarnow reported that poliovirus 2A protein is involved in regulating poliovirus mRNA translation (44). The poliovirus IRES activity was greatly stimulated by co-expression of 2A protein in C2C12 cells, in which the poliovirus IRES was not efficient. In this case, firefly luciferase activity was greatly stimulated in contrast with the slight reduction of Renilla luciferase activity. 2A protein thus appears to have specific auto-stimulation activity for the poliovirus IRES, especially in the non-permissive C2C12 cells. Borman et al. reported that the activity of type 1 IRESs (of enteroviruses and rhinoviruses) was dramatically increased upon co-expression of the poliovirus 2A protein in non-permissive neuro-2A cells (45). The mechanism of this phenomenon is unknown and of great interest.

The monocistronic construct of the Smad5 IRES was inactive in both 293T and C2C12 cells; however, the poliovirus IRES was active in both of these cell lines (Fig. 2A and B). The secondary structure of the Smad5 5′UTR is complex, according to GENETTYX-MAC analysis (data not shown). This secondary structure may be stable and may inhibit scanning. As the Smad5 dl-3 5′UTR has a simpler secondary structure, it had somewhat more activity in mono- or diSm-dl3-transfected cells. This is similar to the finding reported about Apaf-1 IRES (16). The inefficient translation of the Smad5 monocistronic construct observed in C2C12 cells suggests that initiation of translation of Smad5 mRNA in these cells may occur only by internal ribosome entry.