Higher Order Structures of the 5”Proximal Region Decrease the Efficiency of Translation of the Porcine Pro-opiomelanocortin mRNA*

The SP6 polymerase/promoter system was used to synthesize porcine pro-opiomelanocortin mRNAs with nucleotide sequence deletions in the 5’- as well as 3’- untranslated and coding regions. The translational efficiency of the mutant mRNAs was evaluated by cell- free translation or by monitoring the rate and extent of ribosome binding in the presence of sparsomycin. The results of these experiments indicate that specific nucleotide sequences in the 5’-untranslated and coding regions of the pro-opiomelanocortin mRNA decrease its rate of translation. Structure mapping of the mRNA with double-strand and single-strand specific nucleases suggests that these sequences can form stable secondary structures.

Initiation of protein synthesis in eukaryotic cells is a complex event that involves the molecular interaction of many components (for reviews, see Moldave, 1985;Pain, 1986). In an early step, mRNA molecules in the cellular pool associate with a 43 S preinitiation complex (formed by the 40 S ribosomal subunit, eukaryotic initiation factor-2, -3, and -4C, GTP, and the initiator methionine-tRNA) (Jagus et al., 1981). The binding of mRNAs to the 43 S preinitiation complexes is strongly dependent on the presence of the cap structure, m7GpppN, located at the 5'-end of most eukaryotic mRNAs (Shatkin, 1976). Several polypeptides which appear to specifically interact with the cap structure have been identified (Sonenberg et al., 1978;Tahara et al., 1981;Grifo et al., 1983;Edery et al., 1983). These proteins referred to as cap binding proteins stimulate mRNA binding to ribosomes. The binding of mRNAs to the 43 S complexes requires also the hydrolysis of ATP (Marcus, 1970;Trachsel et al., 1977). The precise role of the cap binding proteins and of ATP hydrolysis is not established, but several lines of evidence suggest that they melt secondary structures in the 5"proximal region of mRNAs (Sonenberg et al., 1981;Lee et al., 1983;Ray et al., 1985). It was postulated that after binding at or near the 5'-* This research was supported in part by grants from the Medical Research Council of Canada (to G. B. and N. S.). The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "aduertisernent" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
Recipient end of mRNAs, the complex moves along the RNA, scanning the sequence for a suitable AUG initiation codon (Kozak, 1983). TWO structural features of mRNAs are believed to modulate the rate at which they are translated. First, Kozak (1984) has shown that the nucleotide sequence around the initiation codon AUG plays an important role in the translational efficiency of the proinsulin mRNA. Second, several studies have associated secondary structures located in the 5'-untranslated region of mRNAs with translational efficiency. By insertion mutagenesis, Pelletier and Sonenberg (1985a) and Kozak (1986) have shown that extensive secondary structures in the 5"untranslated sequence can block the translation of mRNAs. Furthermore, it was suggested by Saito et al. (1983) and Spena et al. (1985) that regions of secondary structure reduce the rate of translation of the c-rnyc and zein mRNA, respectively.
In this paper we have studied the influence of the structure of the 5"untranslated region on the translation of the porcine pro-opiomelanocortin (POMC)' mRNA. Using exonuclease Ba131 and restriction endonucleases we have generated several mutant POMC mRNAs with deletions in the 5'-and 3'untranslated portions of the molecule as well as in the coding region. The translation efficiency of these mutant mRNAs was analyzed in a cell-free rabbit reticulocyte lysate and/or by a ribosome binding assay in the presence of sparsomycin.
The results indicate that in vitro the translational efficiency of the porcine POMC mRNA can be increased when deletions are introduced that either eliminate a hairpin structure (ACP = -15.8 kcal/mol) in the 5"untranslated region or make the 5'-proximal region more accessible to ribosomes.

MATERIALS AND METHODS
Construction of 5"Deletion Mutants-Plasmid pSRT contains a porcine POMC cDNA (Boileau et al., 1983a) cloned between the restriction sites PuuI and PstI of pBR327. The PuuI site is located at the end of the cDNA corresponding to the 5'-end of the POMC mRNA and will be referred to as the 5'-end throughout the paper. A set of mutant porcine POMC cDNAs with random deletions in the sequences corresponding to the mRNA 5'-untranslated region was constructed by digestion of plasmid pSRT with exonuclease Ba131 as follows. The plasmid containing the cDNA was linearized with the restriction endonuclease PuuI and digested with exonuclease Ba131 according to the recommendations of the supplier (Bethesda Research Laboratories); after incubations of 0.5, 1, or 2 min with the nuclease, the plasmid ends were filled in with the Klenow fragment of DNA polymerase I (Pharmacia LKB Biotechnology Inc.) in the presence of 250 ~L M of each dNTP, and XbaI linkers were added with T4 DNA ligase (Pharmacia LKB Biotechnology Inc.). The plasmids were recircularized by ligation with T4 DNA ligase and used to transform Escherichia coli HBlOl competent cells. Tetracycline-resistant colo- The abbreviations used are: POMC, pro-opiomelanocortin; Hepes, 4-(2-hydroxyethyl)-l-piperazineethanesulfonic acid. nies were picked for each incubation time, and the length of the 5'untranslated region of the pSRT mutant cDNAs was precisely determined by analysis of restriction fragments on polyacrylamide gels and by nucleotide sequencing (Maxam and Gilbert, 1980). Five mutants termed pSRA with deletions of 24, 47, 73, 95, and 129 nucleotides were selected, and their truncated cDNAs were cloned downstream of the SP6 promoter in plasmid pSP64 (Melton et al., 1984). This cloning was achieved by digesting the pSRA DNA with endonuclease XBaI and filling in the cohesive ends with the Klenow fragment of DNA polymerase I. The cDNA was then released from the plasmid by digestion with endonuclease PstI. Plasmid pSP64 was restricted with endonuclease HindIII, and the cohesive ends were filled in as described above and digested with restriction endonuclease PstI. The DNAs from both plasmids were mixed, ligated with T4 DNA ligase, and used to transform Escherichia coli HBlOl competent cells. Ampicillin-resistant colonies were screened with restriction endonucleases for the presence of mutant cDNAs, and the positive clones were amplified and characterized by nucleotide sequencing as above. Five mutants corresponding to the deletions mentioned above in the 5'-untranslated region of the POMC cDNA were thus obtained and named pSPP-A24, pSPP-A47, pSPP-A73, pSPP-A95, and pSPP-A129. The porcine POMC cDNA contained in pSRT was also inserted in pSP64 in a manner similar to that described above, except that the Hind111 cohesive ends of pSP64 and PuuI cohesive ends of pSRT were digested with nuclease S1 before treatment with the Klenow fragment of DNA polymerase I. This plasmid was named pSPP-21.
Construction of 3"Deletion Mutants-The 3'-deletion mutants pSPP-21 RE were obtained by linearization of plasmid pSPP-21 with different restriction endonucleases before in uitro transcription. Three endonucleases were used EcoRV (mutants pSPP-21 EcoRV) allowed the synthesis of an RNA lacking the poly(A) tail and the last 46 nucleotides of the 3'-untranslated region; XhoI (mutant pSPP-21 XhoI) allowed the synthesis of a RNA lacking the poly(A) tail, all of the 3"untranslated region, and the last 290 nucleotides of the coding region; finally, BglII (mutant pSPP-21 BglII) allowed the synthesis of an RNA consisting of the 5"untranslated region and of the first 156 nucleotides of the coding region (Boileau et al., 1983a).
In Vitro Transcription-The POMC mRNAs were synthesized by in uitro transcription of the linearized plasmids essentially as described by Melton et al. (1984). Linearization of the plasmids was accomplished by digestion of the DNA with the restriction endonucleases indicated in the legends to the figures. A typical transcription reaction contained 2 pg of linearized plasmid, 40 mM Tris-HCI, pH 7.5, 6 mM MgC12, 10 mM dithiothreitoi, 500 units/ml RNasin (Bethesda Research Laboratories), 500 p M of each ribonucleoside triphosphate (Boehringer Mannheim Canada) and 250 units/ml SP6 polymerase (Du Pont-New England Nuclear) in a final volume of 80 rl.
As a tracer, 1 pCi of [a-32P]GTP (Du Pont-New England Nuclear, specific activity 3000 Ci/mmol) was added, and the reaction was incubated at 37 "C for 1 h. After treatment with 20 pg/ml RNase-free DNase (Bethesda Research Laboratories) in the presence of 10 mM vanadyl ribonucleosides (Bethesda Research Laboratories) for 10 min at 37 "C, the RNA was extracted with phenol/chloroform (l:l, v/v) and purified from unincorporated nucleotides on a 1-ml Sephadex G-100 column.
The reaction mixture was incubated at 37 "C for 30 min, extracted with phenol/chloroform (l:l, v/v), and the 5'-3ZP-labeled RNA purified from unincorporated nucleotides as described above. More than 95% of the radioactivity obtained was associated with the cap structure of the RNA as judged from the amount of counts/min present in the mRNA before and after the capping reaction. The specific activity of the mRNA thus obtained was about 2 X lo5 cpm/pg. RNA for cell-free translation and ribosome binding assays were transcribed from linearized plasmids as described above except that the GTP concentration was decreased to 20 pM, and 500 p~ of cap analog m'GpppG (Pharmacia LKB Biotechnology Inc.) was added in order to obtain capped messages (Pelletier and Sonenberg, 1985a).
Zn Vitro Translation and Polyacrylamide Gel Electrophoresis of the Cell-free Translation Products-Zn vitro translation in micrococcal nuclease-treated rabbit reticulocyte lysates was carried out essentially as described by Pelham and Jackson (1976). Translation of the porcine POMC mRNA was performed for 90 min at 30 "C in a total volume of 14.5 pl containing 10 p1 of lysate (supplemented with creatine kinase and hemin), 2.5 p1 of master mix (0.5 M KCl, 2.5 mM MgCl,, 50 mM creatine phosphate, 0.25 mM each amino acid), and 10 pCi of [35S]methionine (Amersham Corp., specific activity 1000 Ci/ mmol). The reaction was initiated by the addition of 2 pl of a solution containing the amounts of mRNA indicated in the figure legends.
After incubation, 1 volume of twice concentrated gel electrophoresis sample buffer (Laemmli, 1970) was added to the reaction mixture. The samples were then heated at 100 "C for 2 min and analyzed by sodium dodecyl sulfate-polyacrylamide gel electrophoresis as described by Laemmli (1970) using a slab gel system with a polyacrylamide gradient of 12-18%. Gels were stained in 0.2% Coomassie Blue in 45% (v/v) methanol, 10% (v/v) acetic acid, destained in the same solution without Coomassie Blue, dried, and exposed against Fuji XR film at -70 "C for the appropriate period of time. The proteins used as standards to calibrate the gels were: phosphorylase b, 94,000 daltons; bovine serum albumin, 67,000 daltons; ovalbumin, 43,000 daltons; carbonic anhydrase, 30,000 daltons; trypsin inhibitor, 20,100 daltons; lactalbumin, 14,400 daltons.
Secondary Structure Mapping with Nucleuses S I , TI, U2, and V1- The 5'-32P-labeled POMC mRNA was purified by electrophoresis on a polyacrylamide gel containing 7 M urea before nuclease digestion (Pavlakis et al., 1980). Digestions with nuclease S1 (Boehringer Mannheim Canada) were performed in 40 mM sodium acetate, pH 4.5, 200 mM NaC1, and 10 mM ZnSO, for 5 min at 37 "c (Wurst et al., 1978). Digestions with RNases T1 and U2 (Pharmacia LKB Biotechnology Inc.) were performed in 10 mM sodium citrate, pH 5.0, 200 mM NaCl, and 10 mM MgC1, for 4 min at 37 "C (Pavlakis et al., 1980). Digestions with ribonuclease V1 (Pharmacia LKB Biotechnology Inc.) were done in 25 mM Tris-HCI, pH 7.2, 200 mM NaCl, and 10 mM MgCl, for 4 min at 37 "C (Lockard and Kumar, 1981). The enzyme to substrate ratio for each digestion is indicated in the figure legends. In all instances, the RNA was preincubated at 37 "C for 10 min before adding the enzyme, and the reaction mixture contained 1 pg of tRNA and at least 5000 cpm of labeled RNA in a final volume of 10 pl.
Reactions were stopped by the addition of EDTA (pH 7.0) to a final concentration of 25 mM and ethanol precipitation of the RNA. RNA fragments were collected by centrifugation, washed with 70% ethanol at -20 "C, and resuspended in 90% formamide containing 50 mM Tris-borate, pH 8.3, 1 mM EDTA, and 0.2% xylene cyano1 and bromphenol blue.
The nucleotide ladder was obtained either by partial alkaline hydrolysis of the 5'-32P-labeled RNA in 50 mM NaHC03/Na2C03, pH 9.0, 1 mM EDTA at 90 "C for 10 min (Donis-Keller et al., 1977) or by heating the RNA at 100 "C for 30 min in deionized formamide containing 1 mM EDTA. Positions of the guanine residues in the ladder were determined by digesting 5'-3ZP-labeled RNA with T1 in the presence of 7 M urea (Donis-Keller et al., 1977). The reaction was stopped as described above.
The RNA fragments produced by the nuclease digestions or partial hydrolysis were analyzed on an 85-cm-long, 20-cm-wide, and 0.4-mmthick polyacrylamide gel containing 7 M urea (Maxam and Gilbert, 1980). The positions of the fragments were detected by exposing the gel to Fuji XR films.
Ribosome Binding Assay-"P-Labeled POMC mRNA was incubated for various periods of time (specified in the legend to Fig. 4) at 25 "c in 50 pl of an S23 wheat germ extract containing 20 mM Hepes (pH 7.4), 50 p M each of 20 amino acids, 2 mM dithiothreitol, 1 mM ATP, 0.2 mM GTP, 5 mM creatine phosphate, 2.5 pg of creatine phosphokinase, 1.5 mM Mg(OAc)l, and 70 mM KC1. Sparsomycin at a concentration of 0.2 mM was added to inhibit elongation of protein synthesis. Initiation complex formation was determined by analysis on glycerol gradients as described by Sonenberg and Shatkin (1977). Centrifugation was for 3 h at 39,000 rpm and 4 "C in a Beckman SW 41 rotor. Fractions were collected, and Cerenkov counting was used to determine radioactivity.

RESULTS
Transcription of Porcine POMC mRNAs-Porcine POMC mRNAs were synthesized i n uitro from plasmid which contained the SP6 promoter located upstream from the cDNAs. Analysis of the 5"sequence of the RNA transcribed from plasmid pSPP-21 linearized with PstI which contains the full-length cDNA indicated that six nucleotides from the 5'-end of the POMC mRNA were lost and replaced by eight nucleotides from vector pSP64 and from a PuuI linker during the cloning procedure (data not shown). This mRNA is referred to as the full-length POMC mRNA in the present work.
The pSPP-A series of plasmids directed the synthesis of mRNAs with deletions in the 5"untranslated region (see Fig.  1, p a w l A ) . Nucleotide sequence analysis of the plasmids suggested that the mRNAs transcribed from these templates are fusion mRNAs with an extension of 17 nucleotides a t their 5'-end. These nucleotides (5' GAATACAAGCTCTA-GAG 3') originate from vector pSP64 and from the XbaI linker used in the cloning procedure (data not shown). The pSPP-21 RE series of plasmid directed the synthesis of mRNAs with identical 5"untranslated regions but with 3'untranslated and coding regions harboring large deletions (see Fig. 3, p a w l A ) . The 5'-untranslated region of these mutant mRNAs is identical to that of the full-length pSPP-21 mRNA. Translation of the pSPP-A mRNAs-The porcine POMC mRNA has a 5"untranslated region of 129 nucleotides (BO- ileau et al., 1983a). T o determine the effect of this 5"untranslated region on the efficiency of translation, we have selected five different mutants with deletions of 24,47,73,95, and 129 nucleotides in the 5"untranslated sequence, respectively. When translated in a rabbit reticulocyte lysate these mRNAs directed the synthesis of a polypeptide of apparent M , of 37,500 (Fig. 1B) consistent with the previously characterized porcine prepro-opiomelanocortin protein (Boileau et al., 1983b). Scanning of the autoradiogram by soft laser densitometry showed that this polypeptide represents more than 75% of the total radioactivity detected (result not shown). In addition to the major 37,500-dalton species, a minor polypeptide of apparent M , of 34,000 is also synthesized in response to the addition of exogenous POMC mRNA. This polypeptide most probably corresponds to initiation of translation at the second in-frame AUG codon.
The translational efficiency of the different truncated mRNAs was assessed by the amount of protein synthesized by the reticulocyte lysate. As can be seen in Fig. 1B  These results indicate that translational efficiency is not directly correlated with the length of the 5"untranslated region of the POMC mRNA and strongly suggest that specific features confined to discrete regions of the mRNA structure can affect its translational rate.
Structure Mapping of the POMC mRNA-Stable secondary structure is one feature of mRNAs that has been shown to greatly reduce translational efficiency (Pelletier and Sonenberg, 1985a;Kozak, 1986). To examine the possibility that the increased translation of the pSPP-A73 mRNA is a consequence of deletion of secondary structures, we mapped the structure of the 5'-proximal region of the POMC mRNA by partial digestions under nondenaturing conditions of 5'-"'Pend-labeled RNA with T1, S1, U2, and V1 nucleases followed by separation of the fragments by electrophoresis on polyacrylamide gels containing 7 M urea. The radioactive fragments were detected by autoradiography, and the nucleotides susceptible to nuclease attack were located by comparing the mobility of the fragments produced by digestion of the same RNA with T1 nuclease under denaturing conditions or by partial hydrolysis. Fig. 2 shows the structure mapping analysis of the pSPP-21 mRNA on a 15% polyacrylamide gel containing 7 M urea.
An examination of Fig. 2 shows regions of enhanced cleavage by the single strand-specific nucleases S1, T1, and U2 as well as regions of inaccessibility. These results suggest a complex secondary structure for the porcine POMC mRNA. S1 nuclease treatment of the 5'-"P-end-labeled RNA revealed three regions of strong cleavage in the 5'-segment of the molecule. These regions are located between nucleotides 31 and 47, 57 and 60, and around nucleotide 135. These results were substantiated when we digested the 5'-end-labeled RNA with nucleases T1 and U2 which cut 3' to guanine and adenine residues, respectively. The major difference found is between nucleotides 17 and 24 where nucleases T1 and U2 detected a region of accessibility whereas nuclease S1 did not cleave.
The very high content of purine in this sequence is possibly responsible for the failure of nuclease S1 to digest this region. This is consistent with the appearance of a band of medium intensity in the S1 lane corresponding to a cleavage 3' to the cytosine residue in position 21. It is noteworthy that the AUG initiation codon (nucleotides 130-132) of the porcine POMC mRNA is located in a very accessible region of the molecule. Digestion of 5'-end-labeled RNA with the double-strand specific V1 nuclease detected regions of secondary structure around nucleotides 38,53,67,84,118, and 133. These findings are consistent with the results of single-strand specific nuclease digestions presented above except for the region around nucleotide 38 which has been shown to be partially accessible to nucleases S1, T1, and U2. A possible explanation for this observation is the formation of weak alternative secondary interactions in that area.
As noticed in Fig. 1, a deletion of the nucleotide sequence between positions 47 and 73 caused an increase in the translation of the mRNA. Examination of this region in Fig. 2 suggests that it forms an hairpin loop with nucleotides around position 60 very accessible to nuclease S1 (lanes 2 and 3 ) and strong V1 cuts on either side (lanes 11 and 12). Computer analysis (Zuker and Stiegler, 1981) of that sequence shows that a hairpin stem can be formed by pairing nucleotides 50-56 with nucleotides 65-71. Nucleotides 57-64 would then form the loop. The predicted stability of that structure would be -15.8 kcal/mol (Tinoco et al., 1973). Such a structure is perfectly consistent with the structure mapping analysis except that the guanine nucleotides at positions 51 and 71 are slightly accessible to nuclease T1 (lanes 6 and 7). Their position near the beginning of the stem may explain this discrepancy. Since this structure was detected at 37 "C by both the single-and double-strand specific nucleases, it is likely to exist under the cell-free translation conditions. Therefore, the results suggest that the increased rate of translation observed for the pSPP-A73 mutant mRNA is brought about by the deletion of the hairpin structure located betwen nucleotides 50 and 71 of the POMC mRNA sequence.
The hairpin structure located between nucleotides 50 and 71 is not the only secondary structure present in the 5"region of the POMC mRNA. As mentioned above, examination of Fig. 2 reveals other areas protected from nuclease digestion. However, computer analysis (Zuker and Stiegler, 1981) showed that the stability of these structures is lower than the stability of hairpin structure 50-71. The deletion of these other potential secondary structures did not result in a significant increase in the translational rate of the POMC mRNA (Fig. 1). Pelletier and Sonenberg (1985b) and Lawson et al. (1986) have shown that 5'-proximal hybrid structures can reduce the rate of translation of mRNAs by restricting access to the 5'cap structure. To examine the possibility that the nucleotide sequence between positions 47 and 73 of the porcine POMC mRNA reduces translational efficiency by a similar mechanism, we synthesized 5'-32P-end-labeled pSPP-21, pSPP-A47, psPP-A73, and pSPP-A129 mRNAs and assessed the accessibility of the cap structure of each mRNA to tobacco acid pyrophosphatase in the conditions described by Godefroy-Colburn et al. (1985). The results of these experiments (not shown) indicated no significant difference between the mRNAs studied suggesting that nucleotides 47-73 decrease translational efficiency through another mechanism.
Translation of thepSPP-21 RE rnRNAs-The pSPP-21 RE  FIG. 2. Electrophoretic analysis on a 15% polyacrylamide gel containing 7 M urea of partial nuclease digests of 5'-3ZP-end-labeled porcine POMC mRNA. Buffers and reaction conditions are described under "Materials and Methods." Lane I, no enzyme added; lanes 2 and 3, S1 nuclease, 5 X lo-' and 5 X lo-' unitslpg RNA, respectively; lanes 4 and 5, TI ribonuclease, denaturing conditions, 2 X lo-' and 2 X lo-* units/pg RNA, respectively; lanes 6 and 7, T1 ribonuclease, native, 1 X and 1 X lo" units/pg RNA, respectively; lane 8, partial alkaline hydrolysis. The bands in the ladder lane migrate faster than the corresponding bands in the enzymatic digests most probably because of the loss of the 7-methyl group on the cap structure during alkaline hydrolysis which results in the loss of one positive charge on the RNA fragments. Lanes 9 and 10, ribonuclease U2, 1 X lo-' and 1 X unitslpg RNA, respectively. mRNAs are a series of mutant RNAs with large deletions in the 3"untranslated and coding regions created by linearization of plasmid pSPP-21 with different restriction enzymes (Fig. 3A). The translational efficiency of these mRNAs was assessed as described for the pSPP-A mRNAs. to the deletion of approximately 100 amino acids. Fig. 3C shows the densitometric scanning of the autoradiogram presented in B. The value obtained from the scanning of lane 3 (pSPP-21 XhoI mRNA) was corrected for the loss of two of the five pre-POMC methionine residues. For comparison purposes, the pSPP-21 PstI mRNA which is the full-length unmodified mRNA was adjusted to the same level as in Fig.  1C. As can be seen in Fig. 3C no significant variations in the rate of translation of the three mRNAs can be observed. These results suggest that in the rabbit reticulocyte lysate the poly(A) tail, the 3"untranslated region, and the termination codon are not necessary features for maximum translation of the POMC mRNA (the POMC cDNA used in these studies has a poly(A) tail of approximately 75 nucleotides). Ribosome Binding to POMC mRNAs-Since the translation product of the pSPP-21 BglII mRNA could not be detected in the previous experiment, the ribosome binding efficiency of each mRNA was assessed in the presence of sparsomycin. Initiation complex formation was measured by centrifugation through glycerol gradients, and the amount of radioactivity found associated with the initiation complexes was expressed as the percent of total recovered radioactivity. When the ribosome binding efficiency of the pSPP-21 RE mRNAs (Fig.  3, panel A) was assessed, only the BglII mRNA showed a significant increase (factor of 3) over the full-length PstI mRNA (results not presented). The results are in agreement with those presented in the previous section, suggesting that removing the poly(A) tail as well as the entire 3"untranslated region and part of the coding sequences (downstream from the XhoI site) does not affect the ribosome binding efficiency of the POMC mRNA. On the contrary, the removal of sequences located between the BglII and the XhoI restriction sites favors binding of the mRNA to the ribosomes. Ribosome binding of all mRNAs could be totally inhibited by the addition of 100 p~ of the cap analog m7GpppG (results not presented) showing that it is specific. T o further characterize the binding of the POMC mRNAs to the ribosomes, we have determined the kinetics of ribosome binding to the BglII and PstI mRNAs. Fig. 4 shows the formation of 80 S initiation complexes and of a faster moving disome peak when "P-labeled PstI mRNA (panel A ) or BglII mRNA (panel B ) is incubated in a wheat germ extract in the presence of sparsomycin and analyzed on glycerol gradients. It can be seen that the rate of ribosome binding on the shorter BglII mRNA is much faster, since after only 0.25 min of incubation maximum formation of 80 S initiation complexes has been obtained; with the PstI mRNA, binding continues over a period of 10 min. However, the rate of formation of disomes is not increased to the same extent. Formation of a disome peak is observed because the 5"untranslated region of the POMC mRNA is long enough (129 nucleotides) to accommodate more than one ribosome (Filipowicz and Haenni, 1979).

Lanes
The absolute amount of radioactivity found associated with ribosomes is higher for the PstI mRNA than for the BglII mRNA, although the percentage of bound mRNA out of total input is higher for BglII mRNA (maximum binding of 30%)  (panel A , 100,000 cpm) and BglII mRNA (panel B, 25,000 cpm) were incubated at 25 "C in a wheat germ extract for the period of time indicated in each panel in the presence of sparsomysin under the conditions described under "Materials and Methods." The amount of binding was determined by sedimentation of the reaction mixtures through glycerol gradients. The percentage of recovered radioactivity bound to ribosomes after 10 min of incubation was: A , 9.3% and B, 30.0%. The position of 80 S ribosomes and disomes was determined as described by Sonenberg and Shatkin (1977). than for PstI mRNA (maximum binding of 9.3%). This is due to the fact that equimolar amounts of mRNAs were added in the incubation mixture. Since the mRNAs are evenly labeled with ["PIGTP throughout the chain and since the PstI mRNA is roughly four times larger than the BglII mRNA, the amount of radioactivity added in the incubation mixture is four times higher for the PstI mRNA than for the BglII mRNA.

Sensitivity of the BglII and PstI mRNA 5'-Untranslated
Sequences to TI Nuclease Digestion-A possible explanation for the results of the ribosome binding assay would be that removal of the sequences between the BglII and XhoI restriction sites has somehow disrupted the secondary or tertiary structure of the 5"untranslated region of the mRNA making it more available for ribosome binding. To verify this hypothesis, 5'-32P-end-labeled BglII and PstI mRNAs were digested with T1 ribonuclease, and the digestion fragments were separated on a 15% polyacrylamide gel containing 7 M urea. Only T1 ribonuclease was used in this experiment because of the high guanine content of the 5'-proximal region of the porcine POMC mRNA (Boileau et al., 1983a). As can be seen in Fig.  5, several regions of the BglII mRNA are more sensitive to T1 ribonuclease attack than in the PstI mRNA. Specifically it can be noticed that the guanine residues from position 9 to 18 have an increased sensitivity to T1 nuclease digestion. When 5'-3ZP-end-labeled BglII and PstI mRNAs were treated with tobacco acid pyrophosphatase under the conditions described by Godefroy-Colburn et al. (1985), both the BglII and PstI mRNAs showed similar accessibility to the 5'-cap structure (result not shown). These results support the idea that the increase in ribosome binding observed for the BglII mRNA is brought about by an increase in accessibility of the 5'untranslated region of the mRNA and further suggest that in the mutant POMC mRNAs, 5"proximal structures decrease translational efficiency by a mechanism different than restricting access to 5'-cap structure.
It is difficult to assess whether this increased accessibility is caused by disruption of secondary or tertiary structure. However, a computer search for complementary sequences to region 9-18 revealed a sequence located between nucleotides 636 and 646 possessing 10 out of 11 nucleotides complementary to position 8-18. Furthermore, in the XhoI mRNA the guanine residues from position 9-18 do not show the increased sensitivity to ribonuclease T1 observed in the BglII mRNA (data not shown) (XhoI endonuclease cuts plasmid pSPP-21 at nucleotide 643 of the POMC sequence).

DISCUSSION
We have presented evidence that the translational efficiency of in vitro synthesized mutant porcine POMC mRNAs is decreased by the presence of 5"noncoding sequences. This conclusion was reached after analysis of two types of mutant mRNAs. The first mutants (plasmids pSPP-A) were obtained by exonuclease Ba131 treatment and correspond to successive deletions of nucleotide sequences in the 5"noncoding region of the porcine POMC mRNA. Removal of nucleotides 47-73 (mutant psPP-A73) increased the translational rate of mRNA by a factor of about 2. Structure mapping of the in vitro synthesized mRNA showed that this sequence can form a hairpin structure with a Gibbs energy change of -15.8 kcal/ mol. Recently, Pelletier and Sonenberg (1985b) and Lawson et al. (1986) have shown that hybrid structures within the first 15 nucleotides of mRNAs restrict the accessibility of the cap structure to the cap binding proteins, thus reducing translational efficiency. Hybrid structures located downstream from this region had no apparent effect. Since all the pSPP-A mRNAs have the same 17-nucleotide sequence at their 5'end (see "Results") and since the hairpin structure formed by nucleotides 50-71 starts 67, 43, and 20 nucleotides from the 5'-cap structure in the control mRNA, pSPI"A24, and pSPP-A47 mutants, respectively, such a mechanism cannot be invoked to explain their reduced rate of translation. This assertion was supported by cap accessibility studies. Rather it is possible that the hairpin structure decreases the translational efficiency of the porcine POMC mRNA by slowing down the migration of the 43 S initiation complex along the mRNA.

Further deletions
of sequences (mutants pSPP-A95 and -A129) produced a decline in the synthesis of the POMC proteins. The reason for this behavior is not clear, but one possibility is that the close proximity of the AUG initiation endonucleases before in uitro transcription. All these mutant mRNAs have the same 5"untranslated segment but differ by large deletions in the 3"untranslated and coding regions. Deletion of the poly(A) tail (mutant pSPP-21 EcoRV) or of all the 3"untranslated region and part of the coding segment (mutant pSPP-21 XhoI) did not affect the rate of translation nor the affinity for ribosomes of the mutant mRNAs. This is in agreement with in uitro and in vivo experiments where the 3"noncoding sequences from @-globin (Kronenberg et al., 1979) and human y and @* interferon mRNAs (Soreq et al., 1981) were deleted without affecting their translational rate and further suggests that the translational termination codon is not an essential feature of the mRNA for efficient termination of protein synthesis. However, when we removed sequences from the coding region that led to greater accessibility of the 5"untranslated region to T1 nuclease, we noticed an increased affinity of the mutant mRNA (pSPP-21 BglII) for ribosomes.
The results presented here indicate that secondary and possibly tertiary structures located in the 5"untranslated region of the porcine POMC mRNA decrease its rate of translation. Our experimental data suggest that these structures act by decreasing the accessibility of the 5"proximal region of the mRNA to the ribosome and also possibly by interfering with its migration toward the initiation codon.
In conclusion, we have presented evidence that sequences proximal to the 5'-end of the porcine POMC mRNA reduce its in uitro translational efficiency by promoting the formation of secondary structures. Our studies were performed with in uitro transcribed mRNA and thus ignored the role of the RNA-bound proteins in translation. For this reason, we are presently assessing the in uiuo translational efficiency of some of our mutants. end-labeled full-length PstI mRNA and of the shorter form BglII mRNA. Buffers and reaction conditions are described under "Materials and Methods." Lanes 1 and 2, BglII mRNA and PstI mRNA, respectively, without enzyme; lane 3, formamide/magnesium hydrolysis of PstI mRNA; lone 4, T1 digestion of PstI mRNA under denaturing conditions, 2 X lo-* units/pg RNA; lanes 5 and 6, T1 digestion of BglII mRNA, native, 5 X and 5 X lo-' units/pg RNA, respectively; lanes 7 and 8, T1 digestion of PstI mRNA, native; 5 X and 5 X 10" units/pg RNA, respectively. codon may reduce the rate of initiation of protein synthesis (Spena et al., 1985).
The second type of mutants (pSPP-21 RE) was obtained by linearization of plasmid pSPP-21 with different restriction