Translational Control of the Xenopus laevisConnexin-41 5′-Untranslated Region by Three Upstream Open Reading Frames*

The Xenopus laevis Connexin-41 (Cx41) mRNA contains three upstream open reading frames (uORFs) in the 5′-untranslated region (UTR). We analyzed the translation efficiency of constructs containing the Cx41 5′-UTR linked to the green fluorescent protein reporter after injection of transcripts into one-cell stageXenopus embryos. The translational efficiency of the wild-type Cx41 5′-UTR was only 2% compared with that of the β-globin 5′-UTR. Mutation of each of the three uAUGs into AAG codons enhanced translation 82-, 9-, and 4-fold compared with the wild-type Cx41 5′-UTR. Based on these increased translation efficiencies, the percentages of ribosomes that recognized the uAUGs were calculated. Only 0.03% of the ribosomes that entered at the cap structure scanned the entire 5′-UTR and translated the main ORF. The results indicate that all uAUGs are recognized by the majority of the scanning ribosomes and that the three uAUGs strongly modulate translation efficiency inXenopus laevis embryos. Based on these data, a model of ribosomal flow along the mRNA is postulated. We conclude that the three uORFs may play an important role in the regulation of Cx41 expression.

Several lines of evidence indicate that at least some connexins can be regulated at the mRNA level. (i) Cx43 expression is transcriptionally as well as translationally regulated in PC12 cells by Wnt1 (13). (ii) Translational control of rat Cx32 and Cx43 expression was suggested after discovery of an internal ribosome entry site enabling translation of these mRNAs under conditions when cap binding is compromised (14). (iii) The stability of the rat Cx43 messenger is mediated by the binding of neuronal-specific proteins to its 3Ј-UTR (15). (iv) The existence of different mRNAs encoding mouse Cx26 (6) and Cx32 (17), different in the 5Ј-UTR only, suggests translational control.
The ribosomal scanning model can explain translational control of most cellular messengers (18). The 43 S preinitiation complex binds at the 5Ј-terminal cap structure and scans the 5Ј-UTR until the complex recognizes an AUG codon. Subsequently, the 60 S subunit joins and protein synthesis starts. Recognition of the AUG codon is dependent on the surrounding sequences and on the secondary structure of the mRNA (19). Usually, translation starts at the most 5Ј AUG codon. When the 5Ј-UTR contains upstream AUGs (uAUGs), the main ORF can be translated by (i) ribosomes entering on an internal ribosome entry site, (ii) by termination at the upstream ORF and reinitiation at the next AUG, (iii) by leaky scanning, i.e. uAUGs are not recognized by all scanning ribosomal complexes, or (iv) by a combination of these processes (20).
This paper describes the study on the translational control of Xenopus Cx41. The Cx41 5Ј-UTR contains three small open reading frames (uORFs) that might be involved in translational regulation of Cx41 expression. The effect of the three individual upstream AUG codons on translation efficiency was analyzed by mutating each of them into an AAG triplet. The presence of the three uAUGs strongly decreased the flow of ribosomes toward the initiation codon of the main ORF of the Cx41 mRNA. The effects of the single mutations of the uAUGs on reporter gene expression enabled us to calculate the efficiency of uAUG recognition by the ribosome.

EXPERIMENTAL PROCEDURES
Plasmid Construction-The Cx41 5Ј-UTR was amplified with a 5Ј/ 3Ј-RACE Kit (Roche Molecular Biochemicals). Ovaria of adult X. laevis were dissected out of the animals after a 20-min anesthesia on ice, and total RNA was isolated and used for the synthesis of cDNA. The Cx41specific reverse primers (Table I) Cx41-OR2, Cx41-OR1, and Cx41-5R1 were designed based on the published sequence (23). The latter primer mutates the sequence around the ATG into an NcoI restriction site. The PCR cycling conditions were dependent on the primers used. In the first PCR, using Cx41-OR2, after the initial denaturation (5 min, 95°C), 40 cycles were performed for denaturation (1 min, 95°C), annealing (2 min, 55°C), and elongation (2 min, 72°C), followed by an extra elongation period of 10 min (72°C). In the PCR with primer Cx41-OR1 the conditions were the same, except for the annealing temperature (5 cycles at 40°C and 35 cycles at 45°C). The PCR product was cloned into pBluescript SKϩ provided with a NcoI site (kindly provided by A. Buchel, Leiden University, The Netherlands). After confirmation of the sequence (T7 Sequencing Kit, Amersham Pharmacia Biotech) a forward primer (Cx41-5F1) was designed, consisting of a HindIII restriction site and nucleotides 1-18. A PCR was set up with this primer to facilitate the cloning of the PCR product into pT7TS. This vector was a gift of P. A. Krieg (University of Texas) and contains the Xenopus ␤-globin 5Ј-UTR (45 nt), 3Ј-UTR (157 nt), and a track of 30 A and C residues behind the 3Ј-UTR. The GFP ORF was inserted between the 5Ј and 3Ј UTR (see also Fig. 2A).
Based on the newly designed construct (pT7TS-Cx41-GFP-glob-A 30 C 30 ) four different Cx41 constructs were made (see also Fig. 4A). In each construct one or all upstream ATGs were mutated to AAG. Three subsequent PCRs were performed. In the first PCR the WT Cx41 clone was used as template and Cx41-5F3 and Cx41-5R1 as primers to create the two downstream AAG triplets. After the initial denaturation (5 min, 95°C) 5 cycles were performed for denaturation (20 s, 95°C), annealing (40 s, 34°C), and elongation (40 s, 72°C), followed by 35 cycles with an annealing temperature of 62°C and a final elongation period of 10 min (72°C). The resulting PCR product was isolated and used as reverse primer in the second PCR with Cx41-5F2. Annealing was performed at 44 and 72°C respectively. The PCR product was then used as primer in a PCR with primer Cx41-5F1 at the same annealing temperatures, creating a fragment linked by a HindIII and an NcoI restriction site. This fragment was cloned into pT7TS. Clones with the desired mutation combinations were selected after sequencing.
In Vitro Transcription and Translation-Plasmid DNA was linearized with BamHI and purified by phenol/chloroform/isoamyl alcohol extraction and subsequent ethanol precipitation. Capped synthetic mRNA was generated using the mMessage mMachine kit (Ambion). The mRNA was deproteinized and purified on a Sephadex G-50 fine column (Amersham Pharmacia Biotech) followed by ethanol precipitation. Amount and integrity of the mRNA were checked by spectrophotometry and ethidium bromide-agarose gel electrophoresis. Equal amounts of transcript were translated in a reticulocyte lysate system in the presence of [ 35 S]methionine (25). Labeled proteins were separated on a 12.5% acrylamide gel and detected by exposure to Hyperfilm MP (Amersham Pharmacia Biotech). Signals were quantified after overnight exposure to a PhosphorScreen and subsequently analyzed using a PhosphorImager and ImageQuant software (Molecular Dynamics). Small peptides were visualized after incorporation of labeled [ 35 S]methionine/cysteine and separation on a Tris-Tricine acrylamide gel (26).
Embryo Injections-X. laevis frogs were reared as described (27). Female animals were injected with 800 IU human chorionic gonadotropin (Pregnyl, Organon) 16 h before use. The female animals were kept at 14°C after injection. Males were injected twice with 400 IU human chorionic gonadotropin, once in the week prior to the experiment and once 16 h before use. Males were anesthetized on ice for at least 20 min. After decerebration, the testes were dissected, rinsed, and stored at 4°C in 100% MMR (100 mM NaCl, 2 mM KCl, 2 mM CaCl 2 and 5 mM Hepes, 1 mM EDTA, pH 7.4). Immediately after squeezing the female frogs, the eggs were rinsed with tap water and fertilized with sperm squeezed out of a small piece of testis. Twenty minutes after fertilization the embryos were dejellied in 2% cysteine-HCl in tap water, pH 7.9. Embryos were reared in 3% Ficoll in 25% MMR at 18°C and staged according to the normal table for X. laevis (28).
Injection needles (GC150-TF10, Clark Electromedical Instruments) were pulled (micropipette puller, Sutter Instruments) and broken to a tip diameter of ϳ10 m. Needles were calibrated after injection of water into oil (PV830 Pneumatic Pico Pump, World Precision Instruments), measurement of the diameter of the droplet, and adjustment of the pressure. One-cell stage embryos were injected with 0.8 -1 ng of RNA in 8 -10 nl of water. Twenty four hours after fertilization the embryos were pooled, rapidly frozen in liquid nitrogen, and stored at Ϫ80°C for RNA or protein isolation.
Analyses of Injected Embryos-RNA was isolated with RNAzol B (Campro Scientific B.V.) according to the manufacturer's protocol with an extra centrifugation step after homogenization of the embryos. Northern blots were performed with 5 g of total RNA per lane. RNA samples were glyoxylated to melt any secondary structure and analyzed on a sodium phosphate-buffered 1.5% agarose gel (29). RNA was blotted on Hybond-N membrane (Amersham Pharmacia Biotech) by capillary transfer and subsequently UV cross-linked to the blot. After at least 2 h at 65°C in prehybridization solution (5ϫ Denhardt's, 3ϫ SSC, 0.1% SDS, 100 mg of dextran sulfate, and 50 g of salmon sperm DNA per ml), a [ 32 P]dCTP (Amersham Pharmacia Biotech)-labeled GFP probe (T7 Quick Prime kit, Amersham Pharmacia Biotech) was added, and hybridization was performed overnight at 65°C. The probe was purified by Sephadex G-50 fine filtration (Amersham Pharmacia Biotech). Excess probe was removed by washing in 0.1ϫ SSC, 0.1% SDS at room temperature. Signal was detected by exposure of the blot to Hyperfilm MP (Amersham Pharmacia Biotech) and quantified after overnight exposure to a PhosphorScreen, by using a PhosphorImager and Image-Quant software (Molecular Dynamics). Prior to hybridization to a Xenopus Histone 3 (XH3) probe the blot was stripped by pouring boiling 1% SDS directly onto the blot.
Proteins were isolated by resuspending the embryos in 20 mM Tris-HCl, pH 7.6, 100 mM KCl, 10% glycerol, and 0.1 mM EDTA (10 l/ embryo). After repeated freezing and thawing, the mixture was centrifuged at 13,000 rpm at 4°C for 10 min. Fat-free supernatant was transferred to a clean Eppendorf tube. This was repeated once to obtain a clear supernatant. Protein equivalent to one or three embryos was separated on a 12.5% acrylamide gel and blotted onto Hybond-P (Amersham Pharmacia Biotech). After Ponceau Red staining (0.2% in 1% acetic acid) the blot was cut into two parts just below the 40-kDa marker band. Both parts of the blot were blocked in 5% fat-free milk powder in PBS, 0.2% Tween for 1 h. The lower part was incubated with GFP antibody (CLONTECH, 1:100 in blocking solution), and the upper part was incubated with rat antibody raised against eIF4A (1:5000 in blocking solution), for 2 h at room temperature. After another incubation in blocking solution the blots were incubated with 5000-fold diluted peroxidase-conjugated secondary antibody for 1 h at room temperature. After washing in PBS, 0.1% Tween, the blot was developed with chemiluminescence before exposure to film. The amounts of protein were determined by using a densitometer and ImageQuant software (Molecular Dynamics). The quantification was confirmed by loading equal amounts of GFP protein on SDS-polyacrylamide gel electrophoresis and subsequent analysis (results not shown). Translation efficiency was calculated by dividing the amount of GFP protein produced by the amount of injected GFP mRNA. The figures were corrected for the differences in loading, determined using the eIF4A antibody and the XH3 probe. Translation efficiency of the most efficient construct was set at 100%.

RESULTS
Cx41 5Ј-Untranslated Region-The partial sequence of Cx41 cloned earlier (23) contains a 154-nucleotide 5Ј-UTR with three potential uORFs, the complete coding region (1146 nt), and a 3Ј-UTR of 2317 nucleotides. The potential uORFs in the 5Ј-UTR and the extremely long 3Ј-UTR suggest translational control of this messenger. A 5Ј-RACE PCR was performed to obtain the complete Cx41 5Ј-UTR. Eight clones were sequenced, and all clones contained a 5Ј-UTR of 174 nucleotides (Fig. 1A), slightly longer than the sequence reported before (23). Presumably, the cDNA library clone resulted from incomplete reverse transcription of the mRNA (23).
The stability of the 5Ј-UTR and of the strongest hairpin within the leader were calculated for 18°C, the breeding temperature of Xenopus embryos (30). The free energy (⌬G) of the complete 5Ј-UTR was Ϫ62 kcal/mol and that of the strongest hairpin Ϫ38 kcal/mol, corresponding to Ϫ34 and Ϫ21 kcal/mol at 37°C. Structures with a free energy more than Ϫ50 kcal/mol (at 37°C) are not a severe problem for the eukaryotic translational machinery (31,32). Therefore, the stability of the hairpins within this 5Ј-UTR is expected not to cause substantial translational repression. The 5Ј-UTR contains three upstream start codons, all followed by in frame stop codons (Fig. 1B). The upstream open reading frames potentially encode three small peptides of 28, 8, and 6 amino acids (Fig. 1C). These uORFs might have a function in translational control. None of the sequences around the uAUGs conform to the Xenopus consensus sequence (A/C)(A/C)A(A/C)(A/C)AUG(A/G) (33). In contrast, the region around the start codon of the Cx41 open reading frame contains the best conserved nucleotides, i.e. the A at Ϫ3 and the G at ϩ1 relative to the AUG (34).
Translation Efficiency of the Cx41 5Ј-UTR-To investigate the translational capacity of the Cx41 5Ј-UTR, constructs containing the 5Ј-UTR of either Cx41 or Xenopus ␤-globin, the ORF of the GFP reporter, and the 3Ј-UTR of ␤-globin were made ( Fig. 2A). DNA was cut after the A 30 C 30 sequence and used to make transcripts for in vivo and in vitro translation. Addition of the A 30 C 30 tail supports stability and thereby translation of the transcript. 2 The construct with the ␤-globin 5Ј-UTR was used as a positive control since this 5Ј-UTR enables very efficient translation of the downstream ORF (35). The in vitro translation of these transcripts in the reticulocyte lysate system showed that the Cx41 5Ј-UTR was about 3-fold less efficient than the globin 5Ј-UTR construct (Fig. 2, B and C). This indicates that, despite the three uORFs, the Cx41 5Ј-UTR does not cause strong translational repression in vitro. Apparently, the reticulocyte system is relatively insensitive to the presence of three potential upstream initiation codons.
The same constructs were injected into one-cell stage X. laevis embryos. In each embryo 0.8 -1 ng mRNA was injected. This is the smallest amount of transcript resulting in a detectable amount of GFP with the wild-type Cx41 5Ј-UTR (data not shown). Injected embryos were pooled 24 h after fertilization (stage 13-14) for protein and RNA isolation. GFP protein production was visualized by Western blotting. The amount of GFP protein (Fig. 3A) was corrected for the amount of injected GFP mRNA (Fig. 3C), whereas both signals were corrected for the loading controls (eIF4A and Xenopus Histon 3, Fig. 3, B and  D). In vivo translation efficiency of the Cx41 5Ј-UTR was only 2%, compared with the ␤-globin 5Ј-UTR. Apparently, the replacement of the efficient ␤-globin 5Ј-UTR for the Cx41 5Ј-UTR strongly reduces translation. The relative translation efficiencies were also analyzed 8 h after fertilization (stage 8 -9). They 2 A. W. van der Velden, unpublished results. Products were separated on a 12.5% acrylamide gel, and the resulting gel was dried and exposed to an x-ray film. C, the amount of produced protein was quantified as described under "Experimental Procedures." appeared to be similar compared with 24 h after fertilization (data not shown).
Peptide uORF1-Since the uORFs were expected to be the cause of the translational repression, the uAUGs were mutated into AAG. To facilitate the analysis of the contribution of each individual start codon, four different constructs were made with either one or with all uAUGs mutated (Fig. 4A). The translation efficiency of the transcripts with single mutations (⌬1, ⌬2, and ⌬3) was compared with the wild-type Cx41 5Ј-UTR (WT) and the triple mutant (⌬123) in reticulocyte lysate. In this assay, no remarkable differences were found (Fig. 4B).
The in vitro translation assay was also performed with a mixture of [ 35 S]methionine and [ 35 S]cysteine and analyzed on Tris-Tricine gels to enable the visualization of small peptides (Fig. 5). Three transcripts were analyzed as follows: the wildtype (WT), the single mutant ⌬1, and the triple mutant (⌬123). Translation of the wild-type 5Ј-UTR produced a small peptide that migrated slightly slower than a 20-amino acid residue control peptide (2.3 kDa). No peptides were detected with the ⌬1 and the ⌬123 mutant transcripts. Therefore, the peptide produced was due to the presence of uAUG1. Note that the exposure times for GFP (16 h) and the peptide (2 weeks) were very different (Fig. 5). Even taking into account the number of cysteines and methionines in GFP and the 28-amino acid peptide, the amount of peptide formed was very low. The results are in agreement with the moderate translational repression in vitro of the WT Cx41 5Ј-UTR compared with the globin 5Ј-UTR and with the lack of differences in in vitro translational capacities between the Cx41 mutants.
Effect of the Three Upstream AUG Triplets on Translation Efficiency-Wild-type or mutant Cx41 transcripts were injected into one-cell stage Xenopus embryos. The embryos were pooled 24 h after fertilization. Extracts were made and analyzed for the amount of GFP protein and mRNA present at the time of harvesting. The amount of produced GFP protein was corrected for the amount of injected GFP mRNA and for loading differences (Fig. 6).
The translation efficiency of the triple mutant (⌬123) was remarkably high compared with the wild-type transcript. The efficiency of ⌬123 was set at 100%. Only 1% of GFP was formed by the wild-type transcript. The mutation of the three uAUGs enhanced translation efficiency 82-(⌬1), 9-(⌬2), and 4-fold (⌬3) compared with the wild-type Cx41 5Ј-UTR. Calculation of the efficiency of translational repression by each AUG will be discussed in detail later. When the embryos were collected 8 h after fertilization the results were similar (data not shown). The results show that the presence of the uAUGs in the leader of Cx41 strongly represses translation of the downstream GFP ORF in young Xenopus embryos.
One of the differences between the analyses of translation efficiency in reticulocyte lysate and analyses after injection of X. laevis embryos is the incubation temperature. The in vitro translation was performed at 30°C, whereas the injected embryos were cultured at 18°C. The results might reflect a difference in a temperature-dependent mechanism. For example, the secondary structure of an mRNA is dependent on temperature, and alteration in the secondary structure may influence recognition of uAUGs. To test whether the difference between the in vitro and in vivo results was the result of difference in temperature at which the analyses were done, the in vitro translation was performed at 18, 24, and 30°C. The Cx41 WT and the ⌬1 transcripts as well as the globin-GFP mRNA were analyzed (Fig. 7). Translation efficiency in reticulocyte lysate was strongly dependent on the temperature. However, at each temperature tested, the ratio of the translation efficiency of Cx41 WT, ⌬1, and the globin 5Ј-UTR transcripts was similar (see also Fig. 2). The differences between in vitro and in vivo translation efficiencies were apparently not caused by a temperaturedependent recognition of the uAUGs. Therefore, the scanning rate and translation initiation complex formation were not different at the measured temperatures with regard FIG. 3. Translational analysis of the Cx41-GFP transcript after injection into Xenopus embryos. One-cell stage X. laevis embryos were injected with the transcripts shown in Fig. 2A. Embryos were pooled 24 h after fertilization for protein and RNA isolation. A and B, protein expression analysis of injected embryos. Protein derived from an equivalence of three embryos was separated on a 12.5% acrylamide gel and blotted. The blot was cut into two parts; the lower part was incubated with GFP antibody (A) and the upper part was incubated with eIF4A antibody (B) as a control for loading. C and D, Northern blot analysis of injected embryos. Five g of total RNA was glyoxylated and analyzed on a sodium phosphate-buffered 1.5% agarose gel. The blot was subsequently hybridized to a GFP (C) and a Xenopus Histone-3 probe (XH3, D). Translation efficiency was calculated by dividing the amount of GFP protein produced by the amount of injected GFP mRNA. These figures were corrected for the loading controls (eIF4A and XH3, respectively).

FIG. 4. Effect of 5-UTR AUG mutations on translation in vitro.
A, schematic presentation of the upstream AUG mutants. All constructs contain the Cx41 5Ј-UTR, the GFP ORF, the ␤-globin 3Ј-UTR, and an A 30 C 30 tail. The presence (ϩ) or absence (Ϫ) of the respective uAUGs is shown. B, different amounts of transcript were translated in a reticulocyte lysate in the presence of [ 35 S]methionine. Products were separated on a 12.5% acrylamide gel, and the resulting gel was dried and exposed to an x-ray film.
to the relative translation efficiencies of the three RNAs. Probably, regulation of translation by cofactors modulating uAUG recognition is different in young Xenopus embryos and in reticulocyte lysate. The striking difference in upstream AUG recognition between in vitro and in vivo assays was noticed before (36). DISCUSSION After mutation of the uAUGs in the Cx41 5Ј-UTR the translation efficiency increased. Based on this increased translation the percentage of Xenopus ribosomes that recognized the uAUGs after RNA injection was calculated (Fig. 8). Mutation of uAUG3 into an AAG codon caused a 4-fold increase in translation efficiency. According to the leaky scanning model, this implies that for each ribosome that recognized the AUG of the main ORF four times more scanning ribosomal complexes must be present on the 5Ј-UTR upstream of uAUG3. When there are four scanning complexes present upstream of uAUG3, the 9-fold increase due to mutation of uAUG2 means the presence of 36 (9 ϫ 4) scanning complexes upstream of uAUG2. Mutation of uAUG1 enhanced the translation efficiency 82-fold, so for each ribosome that reached the main ORF, about 3000 (82 ϫ 36) ribosomal complexes should enter the mRNA at the cap at the 5Ј end. The number of cap-binding ribosomal complexes was set at 100% which means that only 0.03% of these complexes scan the entire 5Ј-UTR and translate the main ORF.
The model in Fig. 8 accounts for all observations and is fully consistent with the ribosomal scanning model (18). The presented data do not suggest any involvement of the secondary or higher order structure of the mRNA. Furthermore, the minor amino acid changes due to the AUG to AAG mutations are most likely not responsible for the increase in translation efficiency. We have made a mutant construct in which approximately 40% of the amino acid residues was changed. Preliminary data indicate that ablation of uAUG1 had the same effect in this severely mutated 5Ј-UTR as in a wild-type 5Ј-UTR (data not shown).
The number of ribosomes leaving the mRNA is a subtraction of the number of ribosomes upstream and downstream of the  Fig. 4A. The embryos were pooled 24 h after fertilization for protein and RNA isolation. A and B, protein corresponding to one embryo was separated on a 12.5% acrylamide gel and blotted onto Hybond-P. C and D, 5 g of total RNA was glyoxylated and analyzed on a sodium phosphate-buffered 1.5% agarose gel. Analysis was done as in Fig. 3. A, GFP antibody; B, eIF4A antibody; C, GFP probe; D, histone-3 probe (XH3). corresponding uAUG (Fig. 8). By dividing the number of ribosomes leaving the mRNA after recognition of an uAUG by the number of ribosomes scanning the 5Ј-UTR upstream of the same uAUG, the percentage of ribosomes leaving the mRNA before arrival at the GFP initiation codon was calculated. This calculation shows that all uAUGs were recognized by the majority of the scanning complexes (99, 89, and 75%). If reinitiation may have occurred, ribosomes must have recognized the uAUGs even more efficiently. Note that reinitiation can only occur at the main ORF, because of the overlapping uORFs.
Simultaneous mutation of all uAUGs increased translation 100-fold, whereas the proposed model (Fig. 8) predicts a 3000fold (82 ϫ 9 ϫ 4) increase. The replacement of the wild-type Cx41 5Ј-UTR by the 5Ј-UTR of ␤-globin increased translation 50-fold. The ␤-globin is among the most efficiently translated messengers (35). We suggest that the ␤-globin mRNA as well as the ⌬123 transcript are maximally loaded with translating ribosomes and that therefore a more than 100-fold increase in the expression level is not possible. Fig. 8 presents the working model of our data, but certain predictions of this model can be studied further, such as by making in-frame fusions of the uORFs with GFP.
We assumed that all scanning complexes that reached the AUG of the main ORF produced GFP. The sequence flanking this AUG conformed to the consensus sequence of X. laevis (A/C)(A/C)A(A/C)(A/C)AUG(A/G) (33) with respect to the most conserved nucleotides at Ϫ3 and ϩ1 (34). The sequence flanking uAUG1 showed this consensus with respect to the G at ϩ1, whereas uAUG2 and uAUG3 were not flanked by one of the conserved nucleotides. However, the Xenopus consensus sequence has not been experimentally investigated and was deduced from the nucleotides that flank initiation codons of known open reading frames (33). The hypothesis that occurrence of nucleotides at specific sites surrounding the AUG is linked to translational efficiency proved correct in higher vertebrate mRNAs (34) but may not be true for Xenopus mRNAs.
Some examples of translational control by upstream ORFs have been described. The most well studied examples are the 5Ј-UTRs of GCN4, cytomegalovirus, and S-adenosylmethionine decarboxylase mRNA (37)(38)(39)(40)(41). An extensive set of mutants allowed the unraveling of a mechanism in which growth limitations regulate the recognition of uAUGs and thereby translation of GCN4 (37,38). Stalling of ribosomes at uORFs reduced translation of cytomegalovirus and S-adenosylmethionine decarboxylase (39)(40)(41). Regulation of elimination of stalling is not yet understood. A similar case is true for the results shown here as follows: although the uORFs clearly down-regulate expression of the main ORF, reversal of this down-regulation is required to allow more efficient Cx41 synthesis.
In young Xenopus embryos, the Cx41 messenger is present in low amounts (24). From stage 15 onward, messenger is detected by reverse transcriptase-PCR analysis, whereas Northern blotting showed the presence from stage 25 onward. After injection of Xenopus embryos with constructs containing the wild-type Cx41 5Ј-UTR, translation was strongly repressed due to the presence of three uORFs (Fig. 3). Obviously the production of Cx41 protein in young embryos is strongly inhibited by transcriptional and translational control. The amount of Cx41 messenger increases in older embryos, suggesting a requirement for more Cx41 protein. To allow for increased Cx41 protein levels, translational derepression has to occur simultaneously. A similar combined control of gene expression was described for platelet-derived growth factor 2 mRNA during megakaryocytic differentiation (16). Translational control allows for fast adaptation to changing conditions as occurs during development, because the mRNA is already present. Translational derepression may occur at certain stages of development, possibly by proteins masking the uAUGs, thereby allowing more efficient translation initiation.
Of all connexin mRNAs from GenBank TM that have a recognizable, although often incomplete, 5Ј-UTR, more than 40% contain at least one upstream AUG or ORF in contrast to a random analysis of about 700 vertebrate mRNAs, of which only 9% contain an uAUG codon (34). This suggests that the translational control as described here for Xenopus Cx41 may be more general among connexin mRNAs.
We were not able to detect the 28-amino acid peptide produced by translating uORF1 after in vivo translational analyses (data not shown). Currently we are making tagged constructs enabling the visualization of this peptide in vivo. This should facilitate further analyses of the mechanism of translational control of Xenopus Cx41 and provide further evidence for the proposed model. FIG. 8. Ribosome flow on Cx41 5-UTR. Based on the increased translation due to mutation of each one of the uAUGs the percentage of ribosomes leaving the mRNA after recognition of an uAUG was calculated as described in the text. Note that despite the differences in translation increase due to the mutation of the different uAUGs, all uAUGs were recognized by the majority of the scanning complexes. Therefore, only 0.03% of the ribosomes, entering the transcript, scan the entire 5Ј-UTR and translate the main ORF.