How do eucaryotic ribosomes select initiation regions in messenger RNA?

Primary sequence of ovomucoid messenger RNA as determined from cloned complementary DNA

Ovomucoid messenger RNA (mRNAom) comprises approximately 8% of the total mRNA in the estrogen-stimulated oviduct. The recombinant plasmid pOM100 contained DNA complementary to the 3' end of mRNAom. DNA complementary to the 5' end of mRNAom was obtained from a partially purified preparation of mRNAom by polymerization by reverse transcriptase in the presence of a restriction fragment primer from pOM100. The complementary DNA mixture was amplified by molecular cloning using poly dG/dC tailing to form recombinant bacterial plasmids. Recombinant plasmids containing ovomucoid DNA sequences were selected by in situ hybridization to 32P-labeled pOM100 fragments. The longest plasmid containing ovomucoid DNA sequences was designated pOM502. The complete DNA sequence of both pOM100 and pOM502 was determined. The two plasmids appear to contain sequences complementary to the entire length of mRNAom. The nucleic acid sequence agrees with the known amino acid sequences for both ovomucoid and its N-terminal signal peptide. Highly homologous sequences occur in two regions that coincide with structural domains of the protein. Comparison of the sequence of mRNAom with that for other eucaryotic mRNAs allowed identification of possible functional regions in the mRNA molecule.

Role of ATP in binding and migration of 40S ribosomal subunits

Two assays have been devised to demonstrate ATP-dependent migration of 40S ribosomal subunits on messenger RNA. The first is a two-step runoff assay. Reovirus mRNA was initially loaded with 40S subunits by incubation with wheat germ ribosomes in the presence of the antibiotic edeine. During the second phase of the incubation, in which further attachment of ribosomes was inhibited, the preformed complexes were shown to dissociate (presumably by runoff) only if ATP was included in the reaction. A more direct demonstration of ATP-dependent migration of 40S subunits was carried out using 3' end-labeled brome mosaic virus mRNA. In the presence of edeine and ATP, 40S ribosomal subunits were shown to advance all the way to the 3' end of the message, as shown by protection of the labeled 3'-proximal segment against nuclease digestion. Depletion of ATP by the addition of hexokinase prevented this migration. A variety of observations has raised the possibility that attachment of eucaryotic ribosomes to messenger RNA proceeds via a "scanning mechanism." The hypothesis is that a 40S subunit binds initially at or near the 5' terminus of the message and subsequently migrates toward the interior, stopping when it encounters the first AUG triplet. If migration of 40S subunits requires ATP, as the present studies suggest, the scanning mechanism predicts that in a system depleted of ATP a single 40S ribosome should be trapped near the 5' terminus of the message--upstream of the AUG initiator codon. This prediction was confirmed by analyzing binding of wheat germ ribosomes to a synthetic ribopolymer in which the 5'-proximal region (lacking AUG codons) and the AUG-containing segment near the 3' end of the molecule were differentially labeled.

A technique for expressing eukaryotic genes in bacteria

Methods are described that allow efficient expression in Escherichia coli of cloned eukaryotic genes. The methods require that the coding sequence of the gene in question be available in a form uninterrupted by intervening sequences (for example, as a complementary DNA clone). The gene products are synthesized unfused to other amino acid sequences. The genetic manipulations are simple, and require the plasmids described and commercially available enzymes.

AUG is the only initiation codon in eukaryotes

An analysis of mutants of the yeast Saccharomyces cerevisiae indicates that AUG is the sole codon capable of initiating translation of iso-1-cytochrome c. This result with yeast and the sequence results of numerous eukaryotic genes indicate that AUG is the only initiation codon in eukaryotes; in contrast, results with Escherichia coli and bacteriophages indicate that both AUG and GUG are initiation codons in prokaryotes. The difference can be explained by the lack of the t6 A hypermodified nucleoside (N-[9-(beta-D-ribofuranosyl)purin-6-ylcarbamoyl]threonine) in prokaryotic initiator tRNA and its presence in eukaryotic initiator tRNA.

The putative mRNA for subunit II of human cytochrome c oxidase starts directly at the translation initiator codon

The gene for subunit II of cytochrome c oxidase (COII) in human mitochondrial DNA (mtDNA) is immediately contiguous, on its 5'-end side, to a tRNAAsp gene. Since all eukaryotic mRNAs so far analysed have been shown to have a noncoding stretch, which is presumably used for ribosome attachment on the 5'-side of the coding sequence, it was resonable to ask whether, in the case of the human COII mRNA, the above function is performed by the tRNAAsp sequence or a portion of it, or whether this mRNA lacks a 5' noncoding stretch. We have sequenced a 5'-end proximal segment of the putative COII mRNA from HeLa cells of about 30 nucleotides and have aligned it with the COII coding sequence in human mtDNA. Our results show that this RNA starts directly at the AUG initiator codon.

Sequence of interrupted and uninterrupted mRNAs and cloned DNA coding for the two overlapping nonstructural proteins of influenza virus

We have obtained the complete sequence of cloned full-length DNA (NS DNA) derived from influenza virus gene 8, which codes for two unique polypeptides, NS1 and NS2, and the sequence of the NS2 mRNA. Previously we showed that the mRNA for NS1 (approximately 860 nucleotides) is colinear with the viral RNA and maps from 0.05--0.95 units of the cloned NS DNA, and the body of the NS2 mRNA (approximately 340 nucleotides) maps from 0.59--0.95 units, suggesting that the two mRNAs are 3' co-terminal and share the same poly(A) addition site. Sequencing studies have shown that the NS2 mRNA contains an interrupted sequence of 473 nucleotides. The nucleotide sequences at the junctions of the interrupted segments are similar to those of the consensus sequences at the splicing sites of intervening regions in eucaryotic mRNAs. The first approximately 56 virus-specific nucleotides at the 5' end of the NS2 mRNA are the same nucleotides as are found at the 5' end of the NS1 mRNA, and this leader sequence of the NS2 mRNA contains the initiation codon for protein synthesis and coding information for nine amino acids which would be common to NS1 and NS2. In addition, both mRNAs contain 10--20 heterogeneous nonviral nucleotides at their 5' ends. The approximately 340 nucleotide body region of the NS2 mRNA can be translated in the +1 reading frame, and the sequence indicates that NS1 and NS2 overlap by 70 amino acids that are translated from different reading frames.

The nucleotide sequence of a rabbit beta-globin pseudogene

We report the nucleotide sequence of a rabbit beta-globin pseudogene, psi beta 2. A comparison the psi beta 2 sequence with that of the rabbit adult beta-globin gene, beta 1, reveals the presence of frameshift mutations and premature termination codons in the protein coding sequence which render pse beta 2 unable to encode a functional beta-globin polypeptide. psi beta 2 contains two intervening sequences at the same locations in the globin protein coding sequence as beta 1 and all other sequenced beta-globin genes. An examination of the DNA sequences at the intron/exon junctions suggests that a putative pse beta 2 precursor mRNA could not be spliced normally. We compare the flanking and noncoding sequences of pse beta 2 and beta 1 and discuss the evolutionary relationship between these two genes.

Utilization of polyadenylate mRNA during growth and starvation in Physarum polycephalum

The effect of growth on the efficiency of utilization of poly(A)-containing mRNA for translation has been investigated in microplasmodia of Physarum polycephalum. Measurement of the relative proportions of poly(A)-rich mRNA in polysomal and post-polysomal fractions isolated by sucrose density gradient centrifugation reveals that newly synthesized poly(A)-rich mRNA is present in increasing proportions in the polysomal region during exponential growth. However, the proportion of long-lived poly(A)-rich mRNA observed in actively-translating polysomes declines as starvation approaches. The ribonuclease content and morphology of the microplasmodia were monitored during growth and starvation in an effort to related this phenomenon to the onset of spherulation.

Nucleotide sequence of influenza virus RNA segment 8 indicates that coding regions for NS1 and NS2 proteins overlap

The smallest RNA segment of influenza A viruses (vRNA segment 8) has recently been shown to code for two unrelated nonstructural proteins (NS1 and NS2) translated from separate mRNAs. Molecular weight considerations indicated that there might not be enough space on vRNA segment 8 for the two coding regions unless they overlap. We have recently cloned in bacterial plasmids several genes of an avian influenza A virus, fowl plague virus (EPV), and now present the complete nucleotide sequence of FPV RNA segment 8 largely determined from the cloned DNA. The DNA sequence predicts two open protein synthesis reading frames that can be translated into polypeptides of sizes similar to those of NS1 and NS2. The coding regions for these polypeptides overlap by the equivalent of 43-60 amino acids, the exact amount depending on which of several possible methionines initiates the synthesis of NS2.

The cDNA for the beta-subunit of human chorionic gonadotropin suggests evolution of a gene by readthrough into the 3'-untranslated region

A 579-base pair approximately full-length cDNA which codes for the 145-amino acid long beta-subunit of human chorionic gonadotropin (HCG) has been cloned in the plasmid vector pBR322 and its complete nucleotide sequence determined. A hydrophobic presequence of 20 amino acids can be identified from the nucleotide sequence. The amino acid sequence of the beta-subunit is known to be related to those of the beta-subunits of the other glycoprotein hormones LH, FSH and TSH, but the beta-subunit of HCG is unique in that it contains a C-terminal extension of about 30 amino acids which has no homologous counterpart in the other three hormones. Analysis of the beta HCG cDNA nucleotide sequence suggests that this extension may have arisen by the loss of the termination codon of an ancestral beta-like gene so that most of what was previously the 3'-untranslated region now codes for protein. The beta-subunit of HCG terminates with the codon UAA located 16 bases before the poly(A) in the sequence AAUAAA. This sequence is believed to be a recognition signal involved in either polyadenylation or processing and therefore has a dual role in this gene, serving both a coding and a regulatory function.

Multiple mRNA species for adenovirus type 2 polypeptides III and pVII

Fractionation of messenger activities isolated from the cytoplasm of HeLa cells late in infection with adenovirus type 2 reveals that viral polypeptides III and pVII are each synthesized from two different-sized mRNA's. the major messenger activity for each protein has the same sedimentation rate as that previously reported by Anderson et al. (Proc. Natl. Acad. Sci. U.S.A. 71:2756-2760, 1974). The minor messenger activities for III and pVII sediment more rapidly and are not aggregates of the major mRNA's for these proteins. The two minor messenger activities cosediment with two polyadenylated RNA species which are labeled late in infection with 32P and whose molecular weights are estimated to be 2.9 x 10(6) and 2.4 x 10(6). Both of these species hybridize to adenovirus type 2 DNA specific for the mRNA family that is 3' coterminal at adenovirus type 2 map position 49.5 and the mRNA family that is 3' coterminal at 62.0. This is consistent with the possibility that these RNAs have 5'-terminal sequences identical to those of the normal mRNA's for III and pVII but are 3' coterminal at map position 62, the normal 3' terminus of the mRNA's for polypeptides II and pVI. These species are not found in polyadenylated RNA isolated from the nucleus, suggesting that the minor mRNA species are cytoplasmic RNAs.

Tissue-specific expression of mouse-alpha-amylase genes: nucleotide sequence of isoenzyme mRNAs from pancreas and salivary gland

We have determined the nucleotide sequence of two different mouse alpha-amylase mRNAs, one found in the pancreas and the other in the salivary gland. The 1577 and 1659 nucleotide mRNAs from pancreas and salivary gland, respectively are the major alpha-amylase species which accumulate in each tissue. Differences in mRNA length are primarily in the 5' noncoding regions. Comparable portions of the mRNAs are 89% homologous. The mRNA sequences predict alpha-amylase precursor proteins of 508 and 511 amino acid residues, accounting for nearly the entire coding capacity of the mRNAs; differences in protein length occur as a result of a nine nucleotide segment present within the translated portion of salivary gland, but not pancreas, mRNA. The lengths and amino acid compositions of the predicted proteins concur with those determined empirically by others. These proteins differ 12% in amino acid sequence, explaining previously observed differences in net charge and antigenic properties. Finally, translation of the salivary gland alpha-amylase mRNA is not initiated at the AUG codon nearest the 5' terminus since that codon is almost immediately followed by the termination triplet UAA. This observation may have implications for the mechanism of translation initiation in eucaroytes.

Distinctive sequence of human mitochondrial ribosomal RNA genes

The nucleotide sequence spanning the ribosomal RNA (rRNA) genes of cloned human mitochondrial DNA reveals an extremely compact genome organization wherein the putative tRNA genes are probably 'butt-jointed' around the two rRNA genes. The sequences of the rRNA genes are significantly homologous in some regions to eukaryotic and prokaryotic sequences, but distinctive; the tRNA genes also have unusual nucleotide sequences. It seems that human mitochondria did not originate from recognizable relatives of present day organisms.

Nucleotide sequences of heat shock activated genes in Drosophila melanogaster. I. Sequences in the regions of the 5' and 3' ends of the hsp 70 gene in the hybrid plasmid 56H8

We present the sequences at the 5' and 3' ends of one hsp 70 gene variant which is derived from the chromosomal locus 87A7. The 5' end of the hsp 70 mRNA has also been determined. 550 bp upstream from the 5' end of the hsp 70 mRNA, there is a very A+T rich region shown by heteroduplex analysis to be also present at the same position in other hsp 70 genes9. The 5' end of the hsp 70 mRNA was found 26 bp after a characteristic "Hogness box". The first ATG codon was found 250 bp downstream from the 5' end of the hsp 70 mRNA. We also determined the termination codon at the 3' end of the hsp 70 gene. Comparisons with other genes are discussed.

Insertion of the eukaryotic transposable element Ty1 creates a 5-base pair duplication

The his4-912 mutation results from insertion of a transposable element into the 5'-non-coding region of the his4 gene of yeast. A duplication of 5 base pairs of wild-type his4 DNA flanks the inserted element. The major class of His+ revertants result from excision of most of the transposable element, leaving an inserted segment of 334 base pairs flanked by the 5-base pair repeats.

Proximity of mRNA5'-region and 18S rRNA in eukaryotic initiation complexes

Messenger RNA was covalently linked to 18S ribosomal RNA in eukaryotic 40S and 80S initiation complexes by photoreaction with an RNA cross-linking agent, 4'-substituted psoralen. The sites of interaction of the mRNA capped 5'-region included some 3'-ends of 18S rRNA.

The nucleotide sequence of a cloned human leukocyte interferon cDNA

We have determined the nucleotide sequence of the human leukocyte interferon cDNA carried in hybrid plasmid Z-pBR322(Pst)/HcIF-2h, which has been shown to direct the formation of a polypeptide with human leukocyte interferon activity (Nagata et al., 1980). The 910 base pair insert contains a 567 (or 543) base pair coding sequence, which determines a putative preinterferon polypeptide consisting of a signal peptide of 23 (or less likely 15) amino acids, followed by an interferon polypeptide of 166 amino acids (calculated molecular weight, 19 390). The coding sequence is preceded by a (most likely incomplete) 56 bp leader and followed by a 242 bp trailer and seven A residues from the poly(A) tail: A comparison of the sequence of 35 amino terminal amino acids of lymphoblastoid interferon (Zoon et al., 1980; M. Hunkapiller and L. Hood, personal communication) and the corresponding sequence deducted for leukocyte interferon revealed 9 differences. This suggests that these two interferons are encoded by two non-allelic genes.

The amino-terminal sequence of human fibroblast interferon as deduced from reverse transcripts obtained using synthetic oligonucleotide primers

From recently published data on the amino-terminal structures of human and mouse interferons, we have predicted and synthesised an oligonucleotide capable of priming specifically the reverse transcription of human fibroblast interferon mRNA present within a total mRNA population. From these transcripts we determined the sequence of the 5'-terminus of the mRNA and identified a putative pre-peptide signal sequence. This enabled us to predict the sequence of another primer capable of directing the synthesis of interferon double-stranded cDNA corresponding to the entire coding region of the mRNA. Further sequencing studies also enabled us to establish the identity of 47 consecutive amino acids beginning with the methionine residue at the amino-terminus of the mature protein.

Mutants of yeast initiating translation of iso-1-cytochrome c within a region spanning 37 nucleotides

We used a specially constructed strain, cyc1-345, of the yeast Saccharomyces cerevisiae to isolate revertants that initiated translation of iso-1-cytochrome c at various sites along an extended region of the mRNA. Normal amounts of iso-1-cytochrome c occurred when translation initiated at the abnormal sites corresponding to amino acid positions -3, -2, 3 and 5, as well as the normal position -1; 20% of the normal amounts occurred when translation initiated at the abnormal position 9. These results with cyc1-345 revertants indicate that translation of iso-1-cytochrome c can initiate with the normal efficiency at any site within the region spanning 25 nucleotides. Furthermore, because the lower amount of the short iso-1-cytochrome c in the mutant initiating at position 9 may not necessarily reflect an inefficiency of translation, we believe that translation can initiate with normal or near-normal efficiencies at any site within a 37 nucleotide region, and presumably at any site preceding and following that of the normal initiation codon. These results establish that there is no absolute requirement for a particular sequence 5' to the initiation codon, and are consistent with our previous suggestion that translation starts at the AUG codon closest to the 5' end of the mRNA.

Structure of genes for virus-associated RNAI and RNAII of adenovirus type 2

A DNA sequence, 552 base pairs in length, encoding the two "virus-associated" (VA) RNAs of adenovirus type 2 is presented. Comparison of the oligonucleotide maps of VA RNAI and VA RNII with the established sequence permits identification of the genes for these RNAs. VA RNAI is 157-160 nucleotides long and VA RNAII 158-163 nucleotides long, depending on the exact length of their heterogeneous 3' end. The genes are separated by a spacer of about 98 nucleotides. The RNAs exhibit scattered regions of primary sequence homology and can adopt secondary structures which resemble each other closely in their configuration and stability. VA RNAII is also capable of assuming a different configuration that is energetically more favorable. The data suggest that the two RNA genes may have arisen by duplication of an ancestral gene and that the folding of the RNA chain may be of importance for the function of VA RNAs. Hypothetical RNA polymerase III recognition sequences and the coding potential of the region are discussed.

Structure of a split yeast gene: complete nucleotide sequence of the actin gene in Saccharomyces cerevisiae

The complete nucleotide sequence of the actin gene from Saccharomyces cerevisiae has been determined. The coding region is interrupted by a 304-base-pair intervening sequence that is located within the triplet coding for amino acid 4. DNA sequences of the intron-exon junctions are similar to those found in higher eukaryotes and can be aligned such that the intron starts with the dinucleotide 5'-G-T-3' and ends with 5'-A-G-3'. Regions fo homology within the sequences upstream from the initiation codon and those following the termination codon have been detected between the yeast iso-1-cytochrome c gene and the actin gene. As deduced from the nucleotide sequence, yeast actin has 374 amino acid residues. Its primary structure, especially the NH2-terminal third of the protein, is highly conserved during evolution.

Yeast viral double-stranded RNAs have heterogeneous 3' termini

The yeast virus, ScV, is communicated only by mating. It has two separately encapsidated dsRNAs. One of these, L, codes for the major capsid polypeptide. The other, M, codes for a polypeptide toxic to yeasts without ScV-M particles. Defective interfering particles containing fragments of M (S) displace ScV-M when they arise. We have shown that five independently isolated S dsRNAs are all derived by internal deletion of M. The 3' ends of all the ScV dsRNAs are markedly heterogeneous. For instance, half of the first 35 nucleotides at one 3' end of M and S are variable. Conserved sequences at the 3' ends of M and S are AAACACCCAUCAOH and AUUUCUUUAUUUUUCAOH. Conserved sequences at the 3' ends of L are UAAAAAUUUUUCAOH and AAAAAUXCAOH, where X is variable. We propose that the sequence AUUUUUCAOH is a recognition sequence for the capsid-associated single-stranded RNA polymerase activity. Since all the viral RNAs have pppGp 5' termini, their 3' termini probably extended one nucleotide beyond the terminal pppGp.

Cell-free translation of avian erythroblastosis virus RNA

Avian erythroblastosis virus (AEV) RNA rescued from nonproducer cells by superinfection with a helper virus is translated into three polypeptides in the messenger-dependent rabbit reticulocyte lysate. A 75,000 molecular weight polypeptide (P75AEV) is synthesized from 28S RNA and is encoded by the 5' section of the AEV RNA, including gag-related and AEV-specific sequences. The P75AEV synthesized in infected cells and the P75AEV synthesized in the cell-free system are electrophoretically identical. A 44,000 molecular weight polypeptide (P44AEV) is synthesized from 20-24S RNA, apparently from the 3' section of the AEV-specific RNA sequence. A minor 37,000 molecular weight polypeptide (P37AEV) is synthesized from 20S AEV RNA. A comparison is drawn between the cell-free products of MC29 and AEV RNAs.

Activation of an internal initiation site for protein synthesis during in vitro translation

The major mRNA for adenovirus 2 polypeptide pVIII sediments at 18S as assayed by in vitro translation in the messenger-dependent rabbit reticulocyte lysate system. However, a small amount of messenger activity for pVIII sediments at abut 27S, coincident with the mRNA for 100 K. Isolation and fractionation of poly(A) containing RNA following in vitro translation of 27S 100 K mRNA demonstrated the appearance of an 18S messenger activity for pVIII, which is approximately the size of the authentic mRNA for this protein. Partial degradation of 27S 100 K mRNA with alkali or ribonuclease T1 also results in activation of an 18S messenger activity for pVIII suggesting that in vitro messenger activity for pVIII associated with 27S RNA is due to degradation of 100 K mRNA during translation in the cell-free system.

Leader sequences of Strongylocentrotus purpuratus histone mRNAs start at a unique heptanucleotide common to all five histone genes

We have determined the sequence of the untranslated leader nucleotides of all five histone mRNAs from Strongylocentrotus purpuratus by the dideoxy chain termination method. Total polysomal RNA from sea urchin embryos was used as a substrate for cDNA synthesis primed by specific DNA restriction fragments. Each of the primers was derived from the 5'-terminal part of the coding region for a different histone protein. The five histone mRNA leader sequences are different in length and primary structure. The 5' termini of all five histone mRNAs coincide with the unique heptanucleotide Py-Py-A-T-T-C-Pu in genomic DNA. This sequence, which defines the start of the individual histone mRNAs, is preceded by the A+T-rich octanucleotide identified in front of all eukaryotic structural genes where sequences have been determined to date.

Structure and genomic organization of the mouse dihydrofolate reductase gene

The genomic organization of the mouse dihydrofolate reductase gene has been determined by hybridization of specific cDNA sequences to restriction endonuclease-generated fragments of DNA from methotrexate-resistant S-180 cells. The dihydrofolate reductase gene contains a minimum of five intervening sequences (one in the 5' untranslated region and four in the protein-coding region) and spans a minimum of 42 kilobase pairs on the genome. Genomic sequences at the junction of the intervening sequence and mRNA-coding sequence and at the polyadenylation site have been determined. A similar organization is found in independently isolated methotrexate-resistant cell lines, in the parental sensitive cell line and in several inbred mouse strains, indicating that this organization represents that of the natural gene.

Assignment of reovirus mRNA ribosome binding sites to virion genome segments by nucleotide sequence analyses

All ten reovirus genome RNA segments were radiolabeled at their 3'-termini by incubation with RNA ligase and 32pCp. The extent of radiolabeling was similar for each of the double-stranded RNAs in the genome segment mixture. Radioactivity was equally distributed between the separated plus and minus strands indicating that the 5'-cap in plus strands did not block 3'-end-labeling of minus strands. The 3'-termini of the four S and three M segments included the common sequences: ...U-A-G-C in minus strands and ...U-C-A-U-C in plus strands. By comparing the minus strand 3'-sequences with 5'-sequences of reovirus mRNAs, small-size genome segments S2, S3 and S4 were correlated with the previously sequenced initiation fragments s46, s45 and s54 derived from small class mRNAs. Medium-size genome segments M1, M2 and M3 similarly were correlated with fragments m30, m52 and m44, respectively. The N-terminal amino acid sequences deduced from the mRNA nucleotide sequences can now be assigned to the nascent chains of particular reovirus proteins.

Influence of mRNA secondary structure on binding and migration of 40S ribosomal subunits

Reovirus messenger RNA was modified by reaction with bisulfite (in denaturing conditions) or by incorporation of IMP in place of GMP, thereby irreversibly unfolding the mRNA. Messenger RNA in which the secondary structure was weakened or abolished retained the ability to bind to wheat germ ribosomes, suggesting that conformational features around the AUG codon are not required for ribosome recognition of mRNA. Ribosomes were not able to attach (directly) to spurious internal sites, even in extensively unfolded RNA, indicating that the monocistronic character of eucaryotic messages (in which initiation is limited to a single 5' proximal site) is not simply due to conformational masking of all the internal AUG codons. The secondary structure in eucaryotic messages does contribute to the fidelity of the translation process, however, because when 40S ribosomal subunits were incubated with denatured mRNA they failed to stop at the 5' proximal AUG codon. Extensive migration beyond the 5' region occurred when 40S ribosomes (in the absence of 60S subunits) attached to unfolded mRNA, implying that the secondary structure in native mRNA facilitates correct translation by impeding migration of 40S subunits beyond the 5' proximal initiation region. Secondary structure in mRNA may also modulate the efficiency of translation. Studies with BrUMP-substituted mRNA, in which the secondary structure is enhanced, suggested that the efficiency of mRNA binding to ribosomes decreases as the stability of the secondary structure increases.