Get 20M+ Full-Text Papers For Less Than $1.50/day. Start a 14-Day Trial for You or Your Team.

Learn More →

RNA stem–loop enhanced expression of previously non‐expressible genes

RNA stem–loop enhanced expression of previously non‐expressible genes Published online May 26, 2004 Nucleic Acids Research, 2004, Vol. 32, No. 9 e78 DOI: 10.1093/nar/gnh076 RNA stem±loop enhanced expression of previously non-expressible genes Michael Paulus, Martin Haslbeck and Manfred Watzele* Roche Diagnostics, Nonnenwald 2, D-82377 Penzberg, Germany and Institut fu È r Organische Chemie und Biochemie, Fakulta Ètfu È r Chemie, Technische Universita ÈtMu È nchen, D-85747 Garching, Germany Received December 10, 2003; Revised February 21, 2004; Accepted May 4, 2004 ABSTRACT Even today the detailed mechanism of translation is not completely understood. However, much progress has been The key step in bacterial translation is formation of made toward understanding the structure of the complex the pre-initiation complex. This requires initial con- between the ribosome, bound mRNA and the tRNAs (6±8). tacts between mRNA, fMet-tRNA and the 30S sub- Initiation is the key step in bacterial translation (9±13). unit of the ribosome, steps that limit the initiation of Initial contacts between mRNA, fMet-tRNA and the 30S translation. Here we report a method for improving ribosomal subunit play major roles in formation of the pre- translational initiation, which allows expression of initiation complex and the control of translation ef®ciency. several previously non-expressible genes. This Hence, accessibility of the Shine±Dalgarno (SD) sequence method has potential applications in heterologous (ribosome binding site) and the start methionine have been shown to be important pre-requisites for successful initiation protein synthesis and high-throughput expression (9,10,12±16). The so-called SD interaction places the mRNA systems. We introduced a synthetic RNA stem±loop and the fMet-tRNA in the right position on the 30S ribosomal (stem length, 7 bp; DG = ±9.9 kcal/mol) in front of subunit. In this interaction the SD sequence hybridizes with various gene sequences. In each case, the stem± the anti-SD sequence on the 16S rRNA in the 30S ribosomal loop was inserted 15 nt downstream from the start subunit; the ef®ciency of this hybridization correlates with codon. Insertion of the stem±loop allowed in vitro ef®cient translation (17,18). expression of ®ve previously non-expressible genes The next step in initiation is association of the 50S and enhanced the expression of all other genes ribosomal subunit with the pre-initiation complex. Finally, investigated. Analysis of the RNA structure proved interaction of the second codon with its anticodon starts the that the stem±loop was formed in vitro, and demon- elongation process (11,19). strated that stabilization of the ribosome binding More or less stable structures of the mRNA translation site is due to stem±loop introduction. By theoretical initiation region have been investigated (10,14,20). These RNA structure analysis we showed that the inserted studies, which focused on the ribosome binding site and its RNA stem±loop suppresses long-range interactions tendency to hybridize with nearby sequences, reported that such structures, if their free energy was weaker than between the translation initiation domain and gene- ±6 kcal/mol, did not affect the ribosome. However, structures speci®c mRNA sequences. Thus the inserted RNA that were 1.4 kcal/mol more stable reduced expression rates stem±loop supports the formation of a separate 10-fold (10,20). translational initiation domain, which is more Usually, highly expressed Escherichia coli genes have accessible to ribosome binding. AU-rich codons immediately following the start methionine (21); GC-rich codons at these positions decrease expression INTRODUCTION rates (22). Pedersen-Lane and coworkers (22) reported that conversion of purine to thymidine bases at codon positions 3, 4 In recent decades heterologous protein synthesis has become and 5 increased expression up to 25%. They concluded that one of the most important tools in biotechnology. After the expression increased because the structure of the RNA in the genome of an organism has been sequenced, expressing the translation initiation region changed. Because AU-rich genes is a key step in understanding their structure and stretches are less likely to form secondary RNA structures, function. Currently, efforts to improve protein synthesis are the initiation site of the mRNA was more accessible to the 30S focused on high-throughput methods (1) and production of subunit of the ribosome. pharmaceuticals (2). In addition, sequences that enhance expression by allowing Ef®cient synthesis of recombinant proteins depends on, for additional base pairings between the messenger and the 16S example, transcription regulation, initiation of the messenger, rRNA have been described (23,24). These translational leader codon bias, mRNA stability and toxicity of the gene product. sequences improve the interaction of the messenger with the Thus many experimental strategies seek to optimize these ribosome. factors to achieve high expression rates (3±5). *To whom correspondence should be addressed. Tel: +49 8856 603121; Fax: +49 8856 607609; Email: Manfred.Watzele@Roche.com Nucleic Acids Research, Vol. 32 No. 9 ã Oxford University Press 2004; all rights reserved e78 Nucleic Acids Research, 2004, Vol. 32, No. 9 PAGE 2 OF 14 Figure 1. Overview of RNA stem±loop introduction. (A) Initiation region with SD sequence, start codon and gene-speci®c mRNA. (B) Gene-speci®c mRNA sequences fold back and pair with regulatory elements of bacterial protein translation. Thus translation initiation is prevented. (C and D) Introduction of the AU-rich region and the RNA stem±loop prevents regulatory elements from pairing with downstream gene sequences. (E) Diagram of the initiation region of 1049 RNA stem±loop mutant with the RNA stem±loop formed. The diagram shows the oligonucleotides used to probe the formation of the RNA stem±loop and their hybridization targets. However, if the initiation of the messenger is blocked by this leader sequence dramatically enhances the probability secondary RNA structures, no common translational leader ofsuccessful gene expression, and prevents the bacterial sequence to enhance expression has been described yet. initiation site from pairing with heterologous downstream Experiments on the regulation of the A-protein gene in the mRNA sequences. This leader made the translation of a set of RNA phage MS2 demonstrated that long-range interactions previously non-expressible genes possible and improved the expression of already expressible genes (e.g. GFP). In the between the translation initiation region and downstream in vitro system used, linear templates of GFP can achieve mRNA sequences prevented ribosomal initiation (16). expression rates up to 230 mg/ml (25). Similar initiation problems occurred in heterologous in vitro Based on these experiments, we conclude that inserting a protein synthesis, demonstrating the need for a generally 7 bp RNA stem±loop at the appropriate place in the mRNA suitable translational leader sequence. Since the initiation allows formation of an isolated translation initiation domain. region might be hidden by gene-speci®c mRNA structure (Fig. 1B), the leader sequence should alter the structure of wild-type mRNA to make the translation initiation region more accessible to ribosomes. In the present study, we inserted MATERIALS AND METHODS a stable local RNA hairpin loop downstream from the Generation of linear expression constructs initiation region to inhibit long-range interactions between the initiation region and gene-speci®c mRNA sequences Linear expression constructs (containing the T7 promoter, the (Fig. 1C and D). ribosome binding site (SD), the start methionine followed by We designed a translational leader sequence that contained the structural gene, the His tag, the stop codon and the T7 ®ve AU-rich codons, which show little tendency to form terminator sequences) were ampli®ed in a two-step PCR as secondary RNA structures, and a GC-rich RNA stem±loop (7 described (26). For wild-type constructs, the gene-speci®c bp; DG = ±9.9 kcal/mol; positions +19 to +36). Once inserted, sequence (Table 1A) was fused directly after the start codon. 0 PAGE 3 OF 14 Nucleic Acids Research, 2004, Vol. 32, No. 9 e78 Table 1. Generation of stem±loop mutants (A) List of genes used to generate linear expression constructs Gene Source organism DDBJ/EMBL/GenBank Gene-speci®c Gene-speci®c reverse primer Molecular weight accession no. forward primer of protein (kDa) 1049 Human cytomegalovirus M17209 GCTAACACCGCG GCGCCGGGTGCGCGA 9.3 Survivin Human NM001168 GGTGCCCCGACG ATCCATGGCAGCCAGC 17.2 CIITA Human U18259 GAGTTGGGGCCC AGAACCCCC 44 L30 r.p. Human m94314 AAGGTCGAGCTG GCGTTTTCCACCAAC 18 S4X isoform r.p. Human m58458 GCTCGTGGTCCC GCCCACTGCTCTGTTTGG 30 GFP Aequorea victoria U73901 ACTAGCAAAGGA AGAACCCCCCCC 28 Tubulin Human J00314 AGGGAAATCGTG ATGAGAACCCCC 50 (B) List of AT-rich combinations used to generate stem±loop mutants 1 Lys-Tyr-Thr-Tyr-Ser1 6 Lys-Thr-Tyr-Tyr-Ser2 AAATATACATATTCT AAAACATATTATTCA 2 Lys-Thr-Tyr-Tyr-Ser1 7 Lys-Tyr-Ser2-Tyr-Thr AAAACATATTATTCT AAATATTCATATACA 3 Lys-Tyr-Ser1-Tyr-Thr 8 Lys-Tyr-Tyr-Ser2-Thr AAATATTCTTATACA AAATATTATTCAACA 4 Lys-Tyr-Tyr-Ser1-Thr 9 His-His-His-His-His AAATATTATTCTACA CATCATCATCATCAT 5 Lys-Tyr-Thr-Tyr-Ser2 AAATATACATATTCA (C) General design used to construct primers for the ®rst PCR that will lead to an RNA stem±loop-mutant Forward primer: AGGAGATATACCATG-(AT-rich combination)-(RNA stem±loop)-(gene-speci®c forward primer) Reverse primer: ATTCGCCTTTTATTATTA-(His tag)-(gene-speci®c reverse primer) (D) Various stem±loop sequences that were inserted to the 3¢ side of the AT-rich combination RNA stem±loop Sequence Stem length 9 bp CAGACAAATAGATATTTGTCTGTA Stem length 8 bp CGTGCACGTGCATCGTGCACG Stem length 7 bp CTGCACGTGATCGTGCAG Stem length 6 bp CGCACGTGCATCGTGCGA Stem length 5 bp CGCCGTGCATCGGCG Stem length 4 bp GCCGTGATCGGC Stem length 3 bp GCGTGCATCGCA Stem length 2 bp CGTGCATCG Primer pairs listed in the table are the gene-speci®c sequences that hybridize to the template during the ®rst PCR ampli®cation. r.p., ribosomal protein. The overlap region for the second PCR is underlined in each primer. The AT-rich combination (chosen from Table 1B) was placed after the start methionine, followed by the RNA stem±loop and the gene-speci®c forward primer (chosen from Table 1A). The wild-type gene was ampli®ed by forward primers (from Table 1A) that did not contain either the AT-rich combination or the RNA stem±loop sequence. Hybridizing regions are shown in bold. An insert containing an AT-rich amino acid combination TCAAGACCCGTTTAGAGGCCCCAAGGGGTTGGGAG- (Table 1B) followed by the RNA stem±loop sequence TAGAATGTTAAGGATTAGTTTATTA) were used for the (Table 1D) was introduced between the start AUG and the second PCR. The DNA content of the second PCR product was estimated gene-speci®c sequence (Table 1A) to form the RNA stem± loop mutant. using the Lumi-Imager System (Roche, Basel, Switzerland) PCRs were performed in a volume of 50 ml in an Eppendorf and 100 ng of the product was used for in vitro expression. thermocycler (master cycler gradient, Eppendorf, Germany) Cloned linear templates were ampli®ed by one-step PCR, using standard protocols. A portion (2 ml) of the product from using the T7 promoter primer (GAAATTAATACGACTCA- the ®rst PCR was used as template for the second PCR. Primer CTATAGGGAGACCACAACGGTTTC) and the T7 termin- C (GAAATTAATACGACTCACTATAGGGAGACCACA- ator primer (CAAAAAACCCCTCAAGACCCGTTTAGA- ACGGTTTCCCTCTAGAAATAATTTTGTTTAACTTTA- GGCCCCAAGG) as PCR primers. The annealing temperature AGAAGGAGATATACC) and primer D (CAAAAAACCCC- during the ampli®cation was 60°C. e78 Nucleic Acids Research, 2004, Vol. 32, No. 9 PAGE 4 OF 14 Cloning of linear templates Before electrophoresis, samples were denatured in TBE±urea sample buffer (23) (Invitrogen) at 70°C for 5 min. The linear expression constructs were cloned into pBAD Topo After electrophoresis the fragment bands were stained with vectors (Invitrogen, Karlsruhe, Germany) according to the SYBR Green II (Sigma, Munich, Germany) (10 0003 in manufacturer's protocol (pBAD TOPO TA, Expression Kit, DMSO). The mRNA bands were analyzed by densitometry Version L). Recombinant plasmids were ampli®ed by one-step using the Lumi-Imager system (Roche). The molecular weight PCR to generate templates for `run-off' transcription (27) and marker contained a combination of RNA molecular weight for in vivo protein synthesis. marker III (Roche) (50 ng/lane) and a mixture of different oligonucleotides (75 ng/lane). (The sizes of the markers are In vitro protein synthesis given in the ®gures.) For bacterial protein synthesis, the Rapid Translation System Run-off transcription of mRNA RTS 100 E.coli HY Kit (Roche) was used according to the manufacturer's protocol. Each reaction (50 ml ®nal volume) Linear templates were produced from isolated plasmids in a contained 12 ml amino acids, 1.25 ml methionine, 10 ml one-step PCR using the T7 promoter and T7 terminator reaction mix, 4.75 ml reconstitution buffer, 12 ml E.coli lysate, primer. The SP6/T7 Transcription Kit (Roche) was used to 100 ng linear template DNA and 2 mg puri®ed mRNA make run-off transcripts of puri®ed mRNA from these template. The reaction was incubated for 2 h at 30°C without templates. Each transcription reaction contained 0.1±.5 mg stirring. Afterwards samples (0.1±1 ml) of the reaction were PCR product, the T7 RNA Polymerase (20 U) and dNTP analyzed by SDS±PAGE (28). (1.5 mM each). The reaction was incubated for 1 h at 37°C. DNase I (20 U) was then added and the reaction was incubated for an additional 15 min at 37°C. The resulting mRNA was In vivo protein synthesis puri®ed by phenol extraction and ethanol precipitation (29). The orientation of inserts in the pBAD plasmids was veri®ed The concentration and purity of the mRNA was measured by by PCR. Plasmids with the same orientation were transformed the optical density at 280 and 260 nm. into BL21 pLysS (Stratagene, Amsterdam, The Netherlands). RNA structure analysis with RNase H For in vivo protein synthesis, colonies were isolated and grown in 4 ml cultures at 37°C and 200 r.p.m. for 5 h. When cell The structure of the mRNA was analyzed by oligonucleotide- densities reached ~10 cells/ml, T7 transcription was induced mediated cleavage with RNase H (30±32). In the RNase with 1 mM ispropyl-b-D-thiogalactopyranoside (IPTG) and cleavage reaction, each of four oligonucleotides [Fig. 1E, (a)± the cells were incubated for an additional 2 h. Samples (10 (d)] was used at three different concentrations (75, 7.5 and cells) were removed from the culture and centrifuged (3 min at 0.75 mM). Oligonucleotides (e) and (f) (Fig. 1E) were each 14000 r.p.m. on a table-top centrifuge). Pelleted cells were used at only one concentration (36 mM). For titration of the resuspended in 10 ml SDS sample buffer and analyzed by ribosome binding site, an oligonucleotide (TCTCCT) com- SDS±PAGE and western blot. plementary to the site was used. Reactions were performed in a reaction volume of 10 ml SDS±PAGE/western transfer containing 50 ng/ml mRNA. First the mRNA samples were denatured at 70°C for 2 min in incubation buffer (25 mM Tris± Electrophoresis and western transfer were performed in the HCl pH 7.5, 200 mM NaCl, 10 mM MgCl 3 6H O). Then the 2 2 Novex Pre cast gel system (Invitrogen) according to the samples were cooled to 30°C and the mRNA was allowed to manufacturer's protocol. Samples from expression reactions refold for 30 min (32). Next, oligonucleotides were added to were denatured for 10 min at 70°C and separated on 4±12% each sample. To allow hybridization the samples were Tris±glycine polyacrylamide gels (Invitrogen). The Multi- incubated for 10 min at 30°C. RNase H (1 U) was then Tag-Marker (Roche) was used as molecular weight marker. added to each sample and digestion was performed for an After electrophoresis the proteins were transferred to PVDF additional 30 min at 30°C. To stop the reaction, one volume membrane (Roche) by electroblotting, according to the TBE±urea sample buffer (23) (Invitrogen, Karlsruhe, manufacturer's protocol. Germany) was added. Portions of each digest (50±350 ng His-tagged proteins on the membrane were detected with a mRNA) were analyzed by denaturing TBE±PAGE. 1:4000 dilution of Anti-His -POD-MAK (Roche). Antibody was incubated with the membrane in TBST buffer [TBS RNA stability assay (50 mM Tris, 150 mM NaCl pH 7.5) containing 1 ml/l Tween The RTS 100 HY in vitro expression system was used for the 20 (Roche)]. stability assay. Samples containing 1 mg mRNA were mixed Alternatively the His tag was detected with NiNTA-AP- on ice with one of four dilutions of E.coli lysate (13, 0.53, Conjugate (1:1000 dilution in TBST) (Qiagen, Hilden, 0.253, 0.1253). After lysate was added, the samples were Germany) followed by visualization with CDP-star reagent mixed carefully and incubated at 30°C for 20 min. After the (Roche). The chemiluminescent signals were monitored with incubation, two volumes of lysis/binding buffer from the High the Lumi-Imager System (Roche). Pure RNA Isolation Kit (Roche) were added to each sample to inactivate RNases. The mRNA remaining in each sample was TBE±PAGE then isolated with the High Pure Kit. Ribonucleic acids were electrophoretically separated on After isolation the samples were analyzed on polyacryla- denaturing polyacrylamide gels, which contained 6% or mide (6%)±TBE±urea or agarose gels. The mRNA amounts 15% polyacrylamide, TBE and urea (7 M) (Invitrogen). were analyzed by densitometry. PAGE 5 OF 14 Nucleic Acids Research, 2004, Vol. 32, No. 9 e78 Table 2. List of forward primers for the ®rst PCR used to generate linear expression constructs of GFP (A) Forward primers that generate RNA stem±loops of increasing stem length to the 3¢ side of the start AUG Forward primer Sequence Without stem±loop (0 bp) AGGAGATATACCATGACTAGCAAAGGAGAA Stem length 4 bp AGGAGATATACCATGACTAATTTTAGTACTAGCAAAGGAGAA Stem length 5 bp AGGAGATATACCATGACTGTTTATACAGTAACTAGCAAAGGAGAA Stem length 6 bp AGGAGATATACCATGACTGGTCAATTACCAGTAACTAGCAAAGGAGAA Stem length 7 bp AGGAGATATACCATGACTGCTTTACATCAAGCAGTAACTAGCAAAGGAGAA Stem length 8 bp AGGAGATATACCATGACTGCACGTGATCGTGCAGTAACTAGCAAAGGAGAA (B) Different locations chosen for insertion of the 8 bp stem±loop Forward primer Sequence GFP +10 nt AGGAGATATACCATGACTAGCACT...GTAAAAGGAGAAGAACTT GFP +13 nt AGGAGATATACCATGACTAGCAAAACT...GTAGGAGAAGAACTTTTC GFP +16 nt AGGAGATATACCATGACTAGCAAAGGAACT...GTAGAAGAACTTTTCACT GFP +19 nt AGGAGATATACCATGACTAGCAAAGGAGAAACT...GTAGAACTTTTCACTGGA GFP +22 nt AGGAGATATACCATGACTAGCAAAGGAGAAGAAACT...GTACTTTTCACTGGAGTT The start AUG is bold, stem±loop sequences are inverted and the overlap region for the second PCR is underlined. The reverse primer is the same as that listed for GFP in Table 1C. ACT...GTA, 8 bp stem±loop (ACTGCACGTGATCGTGCAGTA). Theoretical RNA structure analysis downstream gene-speci®c sequences) of the introduced stem± loop was then calculated and expressed as a percentage: For theoretical RNA structure analysis, mfold software (33) was used. The messengers were folded at 30°C, and dot plot p = 100% 3 [1 ± p (mutant)/p (WT)] 2 r Gji Gji analysis was used to identify all pairings (j) of regulatory sequences (SD; AUG) with downstream sequences. For The probability (p ) that an RNA stem±loop will form was RST analysis of long-range interactions, only pairings between calculated from the theoretical data (equation 1) by consider- regulatory sequences and downstream gene-speci®c mRNA ing only structures in which the RNA stem±loop actually sequences were considered. occurred (index RST): According to the theory behind the software, if n is the number of structures examined, j is a downstream pairing 1 DG ÿ DG RST max between a regulatory element and a downstream gene-speci®c n DG ÿ DG min max sequence, i is any structure in which pairing j occurs and DG i RST p ˆ 3 RST is the free energy of structure i, which determines the 1 DG ÿ DG n max frequency of occurrence of structure i, the probability p Gji n DG ÿ DG min max that the regulatory mRNA will form secondary RNA struc- tures with downstream gene-speci®c sequences can be calcu- lated as XX 1 DG ÿ DG RESULTS i max n DG ÿ DG min max In vitro expression experiments to de®ne the j i p ˆ 1 ji X translational leader sequence 1 DG ÿ DG n max n DG ÿ DG In the initial set of experiments we studied the in¯uence of the min max stem length and the location of the RNA stem±loop on translation. We used different constructs of a gene (GFP) where the numerator of the equation represents the probability which is usually highly expressed. Various stem±loops were that all downstream pairings of type j will occur, and the introduced by using different forward primers in the ®rst PCR denominator represents the probability that all hypothetically (Table 2). The stem±loops were placed directly after the 3¢ end possible downstream pairings between the regulatory elem- of the start AUG and some of them signi®cantly in¯uenced ents and any downstream sequence will occur. expression rates (Fig. 2A). The probability p for downstream pairing of the wild-type Inserts with a stem length of 5 bp or less did not Gji sequence was compared with that of the stem±loop mutant. signi®cantly affect the expression rates (~100 mg/ml), but The reduced downstream af®nity (potential to release p , i.e. stem lengths from 6 to 8 bp (DG values of ±7.8 to ±11.8 kcal/ potential to decrease pairing of a given RNA sequence with mol) reduced expression signi®cantly (Fig. 2A). Hence, a e78 Nucleic Acids Research, 2004, Vol. 32, No. 9 PAGE 6 OF 14 Figure 2. Expression results of various stem±loop mutants. (A) Expression results of constructs that contain different stem±loops inserted after the 3¢ end of the start AUG. The position of the stem±loops and their free energies are shown. One microliter was taken from each of two independent expression reactions involving each construct. These samples were run on 10% acrylamide±Bis/Tris±SDS gels and then examined by western blot analysis. The His-tagged proteins on the blot were detected with NiNTA-AP conjugate and CDP-star; CDP-star luminescence was measured with the Lumi-Imager. The amounts of luminescence due to His-tagged GFP in each sample are shown in the graph. These amounts were veri®ed by measurement of the GFP ¯uorescence in each sample. (B) Expression results from constructs containing the inhibitory 8 bp stem±loop inserted at different positions in the gene sequence. The diagram shows the various insertion positions of the inhibitory 8 bp stem±loop in the constructs. The graph shows luminescence data representing the relative expression of each construct. stem±loop structure with a stem length of at least 6 bp and a At insertion position +13 the inhibitory effect started to free energy of ±7.8 kcal/mol reduces expression signi®cantly vanish (Fig. 2B); at position +16 expression was no longer by blocking ribosomal initiation. affected. This observation is in good agreement with data on In the subsequent expression experiments (Fig. 2B) we used the structure of the ribosome, since the space required by the GFP constructs with an inhibitory 8 bp stem±loop. The 8 bp messenger in the downstream tunnel of the ribosome ranges element was inserted at different positions downstream from from +11 to +15 nt (6±8). the start codon (Table 2B). This allowed us to investigate the Based on these initial experiments we designed a transla- optimal placement of an 8 bp stem±loop (Fig. 2B). tional leader sequence to enhance expression (Fig. 1). This PAGE 7 OF 14 Nucleic Acids Research, 2004, Vol. 32, No. 9 e78 Table 3. Amounts of protein synthesized in vitro. Gene Relative expression (n 3 GFP WT) CIITA 1 1.4 Survivin 1 2.3 1049 8 1.5 GFP WT 1.0 GFP 6 3.7 For quanti®cation of synthesis, the chemiluminescence signal was measured after immunochemical detection of His-tagged protein. Protein amounts were estimated relative to wild-type GFP. The mutant numbers indicate which AT-rich sequences (listed in Table 1B) were inserted to the 5¢ side of the 7 bp stem±loop. (Fig. 3F and G). When inserted along with the stem±loop, AT- rich combinations 1±8 (Fig. 3, lanes 1±8) produced high expression rates, but combination 9 (Fig. 3, lane 9) produced very little expression. The AT-rich region in this combination (Table 1B, sequence 9, and Fig. 3, lane 9) is the His tag sequence, which seems to be less suitable for expression initiation. Because the amount of synthesized protein depends on the expression system and the type of template (linear DNA or plasmid), in an additional western blot (data not shown) we estimated their expression relative to the GFP wild-type construct (Table 3). According to attenuation mechanisms (34±36), tRNA starvation could have caused the lower expression. Figure 3. In vitro expression results from linear constructs of wild-type and Alternatively, the higher GC content, which increased the RNA stem±loop mutants. (A±E) Western blot analysis of in vitro expression tendency to form secondary RNA structures, might lead to the of ®ve different wild-type genes and their corresponding stem±loop mutants. decrease in expression ef®ciency (21,22). In each blot, lane 10 shows that the wild-type genes were not expressed. However, codon bias could be excluded as the cause of the Lanes 1±9 show the expression of constructs containing identical RNA stem±loop sequences and an AT-rich region. The amino acids of the improved expression because supplementing the reaction with AT-rich region were varied as described in Materials and Methods (Table tRNAs that provide rare codons in E.coli did not lead to in vitro 1B). Lanes 1±9 in the ®gure correspond to constructs 1±9 in Table 1B. expression of the wild-type genes (data not shown). (F and G) Western blot analysis of two expressible wild-type genes and In a subsequent set of experiments, an N-terminal HA tag their corresponding RNA stem±loop mutants. The samples on these gels were loaded in the same order as those on gels (A)±(E). was used as the AT-rich region. It was less suited for initiating expression, but it allowed detection of the N-terminus of the translation product. Thus potential fragmentation during the translation could be detected. HA-tagged constructs of sequence contained 5 AU-rich codons (positions +4 to +18 nt) the S4X ribosomal protein (also shown in Fig. 3D), which had followed by a GC-rich sequence that coded for an RNA stem± coding sequences of increasing length, showed in vitro loop (7 bp, DG = ±9.9 kcal/mol, positions +19 to +36 nt) expression without fragmentation (data not shown). (Table 1). In order to investigate whether the RNA stem±loop acts on translation, puri®ed mRNA was used as a template for in vitro In vitro transcription/translation of linear expression translation. The experimental conditions for the translation constructs were comparable with the coupled transcription/translation reactions (described in Materials and Methods). Variants of To investigate the in¯uence of the arti®cial translational leader CIITA, Survivin and 1049 mRNAs, along with the corres- sequence on gene expression we compared its effect on the ponding wild-type genes, were translated. Expression results expression of seven different genes from different organisms were similar to those obtained in the coupled in vitro (Table 1A). The coding sequences of these genes were transcription/translation. The RNA stem±loop mutants between 246 and 1380 nt long. As described in Materials and showed signi®cant expression, while the wild-type mRNAs Methods, the translational leader sequence was introduced by of all three genes were not detectably translated (data not two-step PCR, using the 7 bp stem±loop (Table 1D) in the shown). forward primer of the ®rst PCR (Table 1C) to generate linear expression constructs. In vitro transcription/translation using different leader Insertion of the RNA stem±loop signi®cantly increased sequences expression of wild-type genes that were initially not expressed or only slightly expressed (Fig. 3A, B, C, D and E). To investigate whether the stem±loop or the AT-rich sequence Additionally, the stem±loop insert signi®cantly increased the was responsible for improved expression, linear expression expression of genes that already had good expression rates constructs containing only one element (AT-rich sequence) e78 Nucleic Acids Research, 2004, Vol. 32, No. 9 PAGE 8 OF 14 presence of the rather unstable 2 bp stem±loop, 1049 was highly expressed while CIITA was less effectively expressed. Stem lengths <5 bp decreased expression of all genes. Stem lengths >6 bp increased expression of all genes; maximum expression was obtained with a 7 bp stem. The results demonstrate that a 7 bp stem±loop improved expression in all cases, while the effects of shorter or longer stem lengths depended on the structural gene. Formation of the introduced RNA stem±loop Next we investigated formation of the RNA stem±loop structure in the mutated mRNA. We used complementary oligonucleotides (Fig. 1E) to target the stem sequence rather than sequences to either side of the stem±loop. Hybridization of these oligonucleotides was detected by RNase H cleavage. Single-stranded RNA is accessible to hybridization and allows cleavage, while double-stranded RNA stretches are not accessible and therefore remain uncleaved by RNase H (30±32). Additionally, hybridization of increasing amounts of oligonucleotides competes against native RNA structure at the targeted site and demonstrates the accessibility, and hence stability, of the secondary RNA structure (32,37). As indicated in Figure 5A, three different oligonucleotide concentrations were used for cleavage reactions. Oligonucleotide (a) mediated cleavage (fragment size, 62 nt); oligonucleotide (b) also resulted in cleavage (fragment size, 98 nt). Different concentrations of oligonucleotides led to comparable amounts of fragments. This indicated that the sites before and after the stem±loop were easily accessible to Figure 4. In vitro expression results using different leader sequences. oligonucleotide hybridization. In contrast, oligonucleotides (A) Expression results obtained from linear templates with and without an (c) and (d) could not hybridize to mediate cleavage at any RNA stem±loop. Lanes 1±5 show expression results from different CIITA concentration investigated (Fig. 5A). Only secondary cleavage constructs. Lanes 6±10 show results from Survivin constructs. Lanes 11±15 products in the lower part of the gel were detected. show results from 1049 constructs. Lanes 1, 6 and 11 show wild-type Because the length of oligonucleotides has previously been expression. Lanes 2, 3, 7, 8, 12 and 13 show expression of constructs con- taining AT-rich combinations (Table 1B, sequences 1 and 2) and the intro- shown to in¯uence hybridization (38), oligonucleotides (e) and duced RNA stem±loop. Lanes 4, 5, 9, 10 14 and 15 show expression of (f), containing sequences complementary to both stem± and constructs containing the same AT-rich combinations (Table 1B, sequences hairpin±loop [Fig. 1E, sequences (e) and (f)] were tested in the 1 and 2) but no RNA stem±loop sequence. (B±E) Expression results ob- RNase H cleavage reaction. Performing this reaction in the tained from linear templates with different stem lengths inserted on the 3¢ side of the AT-rich combination. The stem length and the wild-type con- presence of 36 mM oligonucleotide (e) or (f) generated no structs are indicated above the lanes. Similar amounts of (B) 1049 cleavage products. Thus the RNA stem±loop is formed. (C) CIITA and (D) Survivin were analyzed by western blot and quanti®ed In addition, theoretical structure analysis was performed to with the Lumi-Imager. Each expression result in (B) and (C) was expressed determine the probability of RNA stem±loop formation in as a fraction of the result from the most active sample (the sample contain- different genes (Fig. 5B). This analysis showed signi®cant ing the 7 bp stem±loop) and plotted in (E). formation of the 7 bp stem±loop. Only for Survivin mRNA did theoretical values indicate high af®nity of the gene sequence for the GC-rich sequence of the stem±loop. were generated (Table 1B, sequences 1 and 2). The results of Structural changes of regulatory elements on the mRNA the in vitro expression are shown in Figure 4A. While the by introduction of the RNA stem±loop wild-type genes of CIITA, Survivin and 1049 showed no expression (Fig. 4A, lanes 1, 6 and 11), the constructs To test secondary RNA structure formation in translational containing the RNA stem±loop sequence (Fig. 4A, lanes 2, 3, regulatory elements we analyzed the accessibility of the 7, 8, 12 and 13) were expressed in high yields. The constructs ribosome binding site by oligonucleotide titration and subse- containing the AT-rich region but no RNA stem±loop quent RNase H digestion (Fig. 6). The results obtained with sequence (Fig. 4A, lanes 4, 5, 9, 10, 14 and 15) showed little mutant and wild-type mRNAs were compared (32). or no expression. Thus the RNA stem±loop is absolutely For all genes investigated, comparable oligonucleotide necessary for successful expression of non-expressible wild- concentrations produced more cleavage products with wild- type genes. type mRNAs than with RNA stem±loop mutants. According to To investigate how stem±loop characteristics affected principles of antisense technology the kinetics and thermo- improved translation we varied the length of the stem±loop dynamics of oligonucleotide hybridization are profoundly inserted at position +19 (Table 1D). We tested different stem attenuated when hybridization has to overcome energy lengths between 2 and 8 bp (Fig. 4B, C, D and E). In the barriers caused by secondary RNA structure (37,39). Hence PAGE 9 OF 14 Nucleic Acids Research, 2004, Vol. 32, No. 9 e78 Figure 5. Experimental (RNase H) and theoretical (mfold) analysis of RNA stem±loop formation. (A) TBE±urea gel containing RNase H cleavage products generated in the presence of oligonucleotides that are complementary to parts of the RNA stem±loop. Each sample on the gel contains denatured products from a single cleavage reaction (10 ml volume, containing 250 ng mRNA). Each lane is labeled with the name and concentration of the oligonucleotides used to generate the cleavage pattern. Names of the oligonucleotides correspond to those described in Figure 1E. Oligonucleotide (a) mediated cleavage to the 5¢ side of the RNA stem±loop, while oligonucleotide (b) mediated cleavage to the 3¢ side of the RNA stem±loop. No cleavage of the targeted RNA stem±loop was observed in the presence of oligonucleotides (c), (d), (e) and (f). Lane 0, cleavage reaction produced in the absence of oligonucleotide. (B) Results of theoretical mfold analysis of stem±loop formation in different genes. p , probability of RNA stem±loop formation. RST lower hybridization re¯ects more stable secondary RNA downstream af®nity for both the ribosome binding site (64.8 to structures (32). Thus the ribosome binding sites of the wild- 92.4%) and the start AUG (69.2 to 93.4%). For Survivin, type mRNAs were stabilized by the introduction of the RNA introduction of the stem±loop led to less marked release from stem±loop. downstream pairings for both regulatory elements, 35.1% for the SD and 49.1% for the start AUG. Only CIITA showed more downstream pairings of the start methionine in the RNA Theoretical RNA structure analysis stem±loop mutant than in the wild type. However, stem±loop To predict pairings of the translation initiation site with introduction still released the ribosome binding site (52.1%) of downstream gene-speci®c mRNA sequences, we calculated CIITA from downstream secondary RNA structures. theoretical RNA structures from the primary sequences. The The AUG of CIITA showed more af®nity for downstream probabilities of downstream pairings between the ribosome sequences (increased af®nity, p = ±48.8%) because it was binding site and the start methionine were calculated (Fig. 7A). possible for nucleotides ±2 to +3 to pair with nucleotides +213 The reduction of downstream af®nity after stem±loop inser- to +217 in a 5 bp helix [(a) DG = ±8.10 kcal/mol]; tion is also indicated (Fig. 7B). this could be followed by formation of an interior loop (DG In summary, the data predict that stem±loop introduction = 2.20 kcal/mol) and a 5 bp helix involving nucleotides +7 to decreases downstream pairings of the two regulatory elements +11 and +207 to +211 [(b) DG = ±5.20 kcal/mol]. Because (SD, start methionine) in all mRNAs. this structure occurs with great frequency, the probability for The ribosome binding site of the GFP RNA stem±loop downstream pairings of the AUG increased after stem±loop mutants showed lower downstream af®nity (49.8 to 52.8%), introduction. The free energy of helix (a) from the paired AUG and pairings of the start methionine with sequences down- was compensated by the free energy of the interior loop. stream were also decreased (73.0 to 77.7%). RNA stem±loop Hence the AUG was not bound strongly enough to inhibit mutants of gene 1049 showed the most signi®cant changes in translation (20). In addition, downstream pairings of the e78 Nucleic Acids Research, 2004, Vol. 32, No. 9 PAGE 10 OF 14 After the stability tests the amounts of mRNA remaining were determined by gel electrophoresis and densitometry. For each variant, the stability of mutant and wild-type mRNAs was comparable at different lysate concentrations. At lower lysate concentrations (Fig. 8B, 0.1253 and 0.253) the mRNA of the stem±loop mutant might be slightly more stable. However, at lysate concentrations that mimicked expression conditions (Fig. 8B, 13) wild-type mRNA was more stable than the stem±loop mutant. In vivo protein synthesis of the RNA stem±loop constructs We also tested expression of RNA stem±loop constructs in whole cells. pBAD vectors containing linear templates in the same orientation were transformed into E.coli BL21 pLysS and expressed as described in Materials and Methods. Western analysis of the His-tagged proteins showed that three independent clones of 1049 containing the RNA stem±loop construct (Fig. 9, lanes 1±3) were highly expressed, while three independent clones containing the wild-type 1049 (Fig. 9, lanes 4±6) were poorly expressed. Three independent clones of Survivin containing the RNA stem±loop construct (Fig. 9, lanes 7±9) were highly expressed, while two wild-type clones (Fig. 9, lanes 10 and 11) were not expressed. DISCUSSION Introducing an engineered stem±loop sequence into the gene led to good in vitro expression of previously non-expressible genes. This strategy was effective for several genes, which were obtained from different organisms and varied in length from 246 to 1380 nt. The inserted stem±loops also enhanced expression of these genes in vivo. Thus insertion of stem±loop sequences appears to be a generally applicable strategy for enhancing heterologous gene expression. Whether transcription and translation of mRNA were Figure 6. Determination of structural accessibility of the ribosome binding performed separately or in a coupled system, the results site. TBE±urea gels showing RNase H cleavage products obtained after the ribosome binding site was exposed to increasing concentrations of a were similar: no expression of wild-type mRNAs and good complementary oligonucleotide. The same oligonucleotide was used in all expression of RNA stem±loop mutants. Hence the introduced mRNA titrations. Each lane was loaded with 5 ml from a single cleavage sequence acts at the RNA level, and no other parts of the reaction (representing 125 ng mRNA). The bar graph above each gel shows pathway from DNA to mRNA are affected. It is possible, the cleavage in each sample, expressed as a percentage of the maximum cleavage observed. (A) Titration of 1049 mRNA; lanes 1±5 contained the however that RNA stem±loop introduction might affect steps RNA stem±loop mutant, while lanes 6±10 contained the wild type. As after mRNA production, for example by in¯uencing transla- indicated on the gel, both mRNAs were exposed to increasing amounts of tion initiation or mRNA stability. oligonucleotide (0±1500 ng/reaction). (B) Titration of Survivin mRNA and In addition, the translation results con®rm that the in vitro (C) titration of CIITA mRNA; lanes 1±5 contained the RNA stem±loop expression experiments are a valuable method for investigat- mutant, while lanes 6±10 contained wild type. As indicated, mRNAs were exposed to increasing amounts of oligonucleotide (0±2500 ng/reaction). ing the effects of RNA structures on RNA±protein interactions (10,16,20,41). Constructs that contain only the AT-rich combination have very low expression. This indicates that the RNA stem±loop is interior loop and helix (b) are formed in the presence of the important for suf®cient expression of non-expressible genes. introduced AU-rich sequence; in agreement with data from the Furthermore, expression experiments on constructs containing literature (21,22), the AU-rich sequence enhanced translation different stem±loop sequences showed that expression de- initiation. pends on mRNA structure. Translational pausing is involved in the attenuation mech- mRNA stability anism and is caused by stable RNA secondary structures. It Because differences in mRNA stability could have led to affects transcription and translation ef®ciency (34±36). improved translation results (40), we compared the stability of However, translational pausing and attenuation mechanisms mRNA from the mutant version of gene 1049 with that of can be excluded as causes of improved protein synthesis for mRNAs from wild-type 1049 in an in vitro expression system. two reasons. PAGE 11 OF 14 Nucleic Acids Research, 2004, Vol. 32, No. 9 e78 Figure 7. Downstream af®nity of the translation initiation domain. (A) Probability p that the ribosome binding site (RBS) and the start AUG will pair with Gji downstream sequences. Results are shown for wild-type mRNAs (WT) and stem±loop mutants (numbered) of the genes 1049, Survivin and CIITA. The numbers of the mutants correspond to the amino acid sequences shown in Table 1B. (B) Reduction in downstream af®nity p of the mutant regulatory sequences compared with the wild type. The higher the p value, the less likely it is that the regulatory sequence will pair with downstream sequences. First, the 7 bp stem±loop is positioned at +19 to +36, while downstream gene-speci®c mRNA sequences. Therefore chan- the initiating ribosome spans positions ±15 to +15 (6,8). ging the structure of the ribosome binding site resulted in more During elongation the movement of the ribosome disrupts de®ned structures within the sequences of the 5¢ UTR. This base pairings of nucleotides +19 to +25 in the RNA stem± conclusion is consistent with the hierarchical pathway of RNA loop. After it opens the stem, that ribosome can no longer be structure formation, in which short hairpin structures are more affected by the stem±loop. Also, while one ribosome holds the stable and are formed ®rst, before structures involving long- stem open, there is not enough space on the mRNA for range interactions (43±46). Hence the same oligonucleotide initiation by another ribosome. used for structural titration (targeted to the RBS) hybridizes Second, T7 RNA polymerase is used in the coupled in vitro more ef®ciently to its target if the target is involved in several expression system. Because the T7 RNA polymerase elong- alternative long-range interactions rather than `one' hairpin ates the RNA chain rapidly, transcription is independent of structure. Together with the improved expression observed in translation (42) and RNA structures are formed before the stem±loop mutants, these observations support the ribosomal initiation. conclusion that stem±loop introduction led to formation of a Our experiments on the stability of the RNA structure of the structurally isolated initiation domain, consisting primarily of ribosome binding site (RBS) indicated that it was stabilized short local hairpins with DG < ±6 kcal/mol (10,15,20). During after stem±loop introduction in all three genes investigated. translation initiation such structures can easily be opened by RNA folding analysis with mfold software (33) concentrated the ribosomes. on detecting downstream pairings of the RBS and/or the start Stability of the mRNA plays a fundamental role in methionine. Stem±loop introduction reduced the af®nity of the regulating protein amounts in the cells (47±49). The stability regulatory elements for downstream gene-speci®c mRNA of the messenger controls the number of translation starts that sequences. Compared with the wild-type genes, the RNA occur during the lifetime of the mRNA. Degradation of stem±loop mutants reduced downstream af®nity of the mRNA is not a random process of endonucleolytic cleavage ribosome binding site by 35.1 to 92.4% (49.1 to 93.4% for (50,51), but is mediated by a combination of RNA helicases the start methionine). Hence, stem±loop insertion inhibited the and endo- and exoribonucleases (47), which are combined in formation of alternative structures between the RBS and the E.coli degradasome. The rate-limiting step of degradation e78 Nucleic Acids Research, 2004, Vol. 32, No. 9 PAGE 12 OF 14 Figure 9. Comparison of in vivo expression of RNA stem±loop constructs and wild-type genes. Western blot analysis of protein synthesis from three independent clones of RNA stem±loop mutant 1049 1 (lanes 1±3); wild-type 1049 (lanes 4±6), three independent clones of a Survivin RNA stem±loop mutant (lanes 7±9) and two independent clones of wild-type Survivin (lanes 10 and 11). Similar amounts of protein were analyzed in each sample. Table 4 summarizes expression results obtained by insertion of various stem±loops and theoretical analysis of the RNA structure. In some genes a rather unstable 2 bp stem±loop was suf®cient to initiate expression (Table 4, 2 bp element). Although the constructs containing the AU-rich combination without a stem±loop allowed low expression of 1049, this combination did not allow expression of the other genes investigated (Table 4, AT+/RST±). Theoretical analysis of the Figure 8. mRNA amounts remaining after 20 min incubation with different RNA structures explained these results. The decrease in the concentrations of E.coli lysate. (A) TBE±urea gels show the mRNA remaining in wild-type and RNA stem±loop mutants after 20 min digestion. value of predicted RNA stem±loop formation (Table 4, p ) RST (B) The amounts of mRNA detected after digestion for 20 min, as correlates with the expression results. The Survivin mRNA determined by densitometric analysis of the gel. The error bars show the shows high af®nity for the GC-rich stem±loop sequence, standard deviations obtained for duplicate samples. indicating a general tendency to form secondary RNA structures. The relatively greater effect of the 2 bp element is endonucleolytic cleavage at the 5¢ end, followed by on CIITA expression agrees with the theoretical prediction directional cleavage from the 3¢ to the 5¢ end (47,48). Thus that stem±loop insertion increases stem±loop formation in 5¢ and 3¢ RNA secondary structures in the untranslated region CIITA more than in Survivin, but decreases the downstream act as stabilizers against degradation and extend the lifetime of af®nity of the regulatory elements by about the same amount a messenger (40,52,53). in both genes. Finally, gene 1049 shows the most signi®cant We did not measure mRNA decay precisely, but experi- enhancement of expression by the 2 bp stem±loop, as well as ments on the stability of the mRNA in the in vitro expression producing the highest values in the theoretical analysis. Taken system demonstrate that introduction of the RNA stem±loop together, the results indicate that the degree to which the did not stabilize the wild-type messenger. These observations mRNA structure enables successful expression depends on the are in agreement with data indicating that occurrence of at gene involved. least ®ve unpaired nucleotides upstream from stable second- In summary, the experiments demonstrate the importance of ary RNA structures at the 5¢ end of the messenger reverse the the RNA stem±loop for mediating translation initiation. Once stabilizing effect of these secondary structures (52,54±56). introduced, the RNA stem±loop forms and stabilizes the Additionally, signi®cant amounts (33%) of the wild-type ribosome binding site in a translatable RNA structure. Long- mRNA remained in the undiluted lysate after 20 min. Other range interactions of the regulatory sequences are critically experiments showed that after 20 min most of the protein in important, and the introduced RNA stem±loop acts as a barrier the coupled in vitro system is already synthesized (data not against hybridization of regulatory elements to downstream shown). Thus, if ribosomes initiated synthesis from the wild- gene-speci®c mRNA sequences. Thus introduction of a type messenger, protein would have been detected after stem±loop enhances formation of a distinct bacterial transla- 20 min. tion initiation domain. Table 4. Comparison of experimental results with theoretical RNA structure analysis Gene Expression (2 bp element) Expression (AT+/RST±) p p (RBS) (%) p (AUG) (%) RST r r 1049 0.97 Low 1 64.8±92.4 69.2±93.4 CIITA 0.48 No 1 52.1 ±48.8 Survivin 0 No 0.62 35.1 49.1 Each value for the 2 bp stem±loop element is expressed as a fraction of the value obtained with the 7 bp element (as shown in Fig. 4B±E). AT+/RST± refers to expression of constructs containing the AT-rich codon combination but no RNA stem±loop. PAGE 13 OF 14 Nucleic Acids Research, 2004, Vol. 32, No. 9 e78 18. Jacob,W.F., Santer,M. and Dahlberg,A.E. (1987) A single base change in Furthermore, the formation of a stable local hairpin the Shine±Dalgarno region of 16S rRNA of Escherichia coli affects structure of ~7 bp (DG = ±9.9 kcal/mol), 15 nt downstream translation of many proteins. Proc. Natl Acad. Sci. USA, 84, 4757±4761. from the start methionine, enhances heterologous gene 19. Ringquist,S., Shinedling,S., Barrick,D., Green,L., Binkley,J., expression. For maximum effect, these 15 intervening Stormo,G.D. and Gold,L. (1992) Translation initiation in Escherichia nucleotides should be AU rich. coli: sequences within the ribosome-binding site. Mol. Microbiol., 6, 1219±1229. Although the RNA stem±loop structure is a valuable tool for 20. deSmit,M.H. and van Duin,J. (1994) Control of translation by mRNA enhancing expression yield, the cost of that enhanced expres- secondary structure in Escherichia coli. A quantitative analysis of sion is additional sequences in the ®nal gene product. In some literature data. J. Mol. Biol., 244, 144±150. cases this might impair the function of the protein. To 21. Stenstrom,C.M., Jin,H., Major,L.L., Tate,W.P. and Isaksson,L.A. (2001) overcome this problem, conservative codon replacement Codon bias at the 3¢-side of the initiation codon is correlated with translation initiation ef®ciency in Escherichia coli. Gene, 263, 273±284. might be used to generate the stem±loop structure in the 22. Pedersen-Lane,J., Maley,G.F., Chu,E. and Maley,F. (1997) High-level leader sequence. Alternatively, post-translational chemical or expression of human thymidylate synthase. Protein Expr. Purif., 10, enzymatic processing might be used to remove the additional 256±262. sequences. 23. Studier,F.W., Rosenberg,A.H., Dunn,J.J. and Dubendorff,J.W. (1990) Use of T7 RNA polymerase to direct expression of cloned genes. Methods Enzymol., 185, 60±89. ACKNOWLEDGEMENTS 24. Sprengart,M.L., Fuchs,E. and Porter,A.G. (1996) The downstream box: an ef®cient and independent translation initiation signal in Escherichia We thank Marcus Schmidt, Anette Seidl, Claudia Kirr, coli. EMBO J., 15, 665±674. Thomas Nikolaus and Bjo È rn Voss for experimental support. 25. Watzele,M., Nemetz,C., Obermeier,W., Seidel,A. and Buchberger,B. This work was supported by Roche Diagnostics Germany. (2003) High-throughput expression PCR used to systematically investigate regulation of translation initiation in an Escherichia coli cell- free expression system. In Swartz,J.R. (ed.), Cell-free Protein REFERENCES Expression. Springer, Berlin, pp. 25±30. 26. Watzele,M., Nemetz,C., Metzler,T., Fernholz,E., Hoffmann,T. and 1. Buchanan,M.V., Larimer,F.W., Wiley,H.S., Kennel,S.J., Squier,T.J., Mutter,W. (2001) Rapid Translation System RTS 100Ða high Ramsey,J.M., Rodland,K.D., Hurst,G.B., Smith,R.D., Xu,Y. et al. (2002) throughput cell-free protein expression. Biochemica, 3, 27±28. Genomes to Life `Center for Molecular and Cellular Systems': a research 27. Resto,E., Iida,A., Van Cleve,M.D. and Hecht,S.M. (1992) Ampli®cation program for identi®cation and characterization of protein complexes. of protein expression in a cell free system. Nucleic Acids Res., 20, 5979± Omics, 6, 287±303. 2. Swartz,J.R. (2001) Advances in Escherichia coli production of 28. Laemmli,U.K. (1970) Cleavage of structural proteins during the therapeutic proteins. Curr. Opin. Biotechnol., 12, 195±201. assembly of the head of bacteriophage T4. Nature, 227, 680±685. 3. Jonasson,P., Liljeqvist,S., Nygren,P.A. and Stahl,S. (2002) Genetic 29. Sambrook,J., Russel,D.W. and Sambrook,J. (2001) Molecular Cloning: A design for facilitated production and recovery of recombinant proteins in Laboratory Manual (3rd edn). Cold Spring Harbor Laboratory, Cold Escherichia coli. Biotechnol. Appl. Biochem., 35, 91±105. Spring Harbor, NY. 4. Balbas,P. (2001) Understanding the art of producing protein and 30. Donis-Keller,H. (1979) Site speci®c enzymatic cleavage of RNA. nonprotein molecules in Escherichia coli. Mol. Biotechnol., 19, 251±267. Nucleic Acids Res., 7, 179±192. 5. Makrides,S.C. (1996) Strategies for achieving high-level expression of 31. Gunzl,A., Pal®,Z. and Bindereif,A. (2002) Analysis of RNA-protein genes in Escherichia coli. Microbiol. Rev., 60, 512±538. complexes by oligonucleotide-targeted RNase H digestion. Methods, 26, 6. Yusupova,G.Z., Yusupov,M.M., Cate,J.H. and Noller,H.F. (2001) The 162±169. path of messenger RNA through the ribosome. Cell, 106, 233±241. 32. Phillips,C., Kyriakopoulou,C.B. and Virtanen,A. (1999) Identi®cation of 7. Ramakrishnan,V. (2002) Ribosome structure and the mechanism of a stem±loop structure important for polyadenylation at the murine IgM translation. Cell, 108, 557±572. secretory poly(A) site. Nucleic Acids Res., 27, 429±438. 8. Culver,G.M. (2001) Meanderings of the mRNA through the ribosome. 33. Zuker,M. (2003) Mfold web server for nucleic acid folding and Structure, 9, 751±758. hybridization prediction. Nucleic Acids Res., 31, 3406±3415. 9. Hartz,D., McPheeters,D.S. and Gold,L. (1991) In¯uence of mRNA 34. Landick,R. and Yanofsky,C. (1987) Isolation and structural analysis of determinants on translation initiation in Escherichia coli. J. Mol. Biol., the Escherichia coli trp leader paused transcription complex. J. Mol. 218, 83±97. Biol., 196, 363±377. 10. Gold,L. (1988) Posttranscriptional regulatory mechanisms in Escherichia 35. Landick,R., Yanofsky,C., Choo,K. and Phung,L. (1990) Replacement of coli. Annu. Rev. Biochem., 57, 199±233. the Escherichia coli trp operon attenuation control codons alters operon 11. Gualerzi,C., Risuleo,G. and Pon,C.L. (1977) Initial rate kinetic analysis expression. J. Mol. Biol., 216, 25±37. of the mechanism of initiation complex formation and the role of 36. Gutman,G.A. and Hat®eld,G.W. (1989) Nonrandom utilization of codon initiation factor IF-3. Biochemistry, 16, 1684±1689. pairs in Escherichia coli. Proc. Natl Acad. Sci. USA, 86, 3699±3703. 12. Jacques,N. and Dreyfus,M. (1990) Translation initiation in Escherichia 37. Lima,W.F., Monia,B.P., Ecker,D.J. and Freier,S.M. (1992) Implication of coli: old and new questions. Mol. Microbiol., 4, 1063±1067. RNA structure on antisense oligonucleotide hybridization kinetics. 13. McCarthy,J.E. and Brimacombe,R. (1994) Prokaryotic translation: the Biochemistry, 31, 12055±12061. interactive pathway leading to initiation. Trends Genet., 10, 402±407. 38. Young,S. and Wagner,R.W. (1991) Hybridization and dissociation rates 14. Tessier,L.H., Sondermeyer,P., Faure,T., Dreyer,D., Benavente,A., of phosphodiester or modi®ed oligodeoxynucleotides with RNA at Villeval,D., Courtney,M. and Lecocq,J.P. (1984) The in¯uence of mRNA near-physiological conditions. Nucleic Acids Res., 19, 2463±2470. primary and secondary structure on human IFN-gamma gene expression 39. Nellen,W. and Lichtenstein,C. (1993) What makes an mRNA anti-sense- in E.coli. Nucleic Acids Res., 12, 7663±7675. itive? Trends Biochem. Sci., 18, 419±423. 15. deSmit,M.H. and van Duin,J. (1994) Translational initiation on 40. Carrier,T.A. and Keasling,J.D. (1999) Library of synthetic 5¢ secondary structured messengers. Another role for the Shine±Dalgarno interaction. structures to manipulate mRNA stability in Escherichia coli. Biotechnol. J. Mol. Biol., 235, 173±184. Prog., 15, 58±64. 16. vanMeerten,D., Girard,G. and van Duin,J. (2001) Translational control 41. Ivanovski,G., Gubensek,F. and Pungercar,J. (2002) mRNA secondary by delayed RNA folding: identi®cation of the kinetic trap. RNA, 7, structure can greatly affect production of recombinant phospholipase 483±494. A(2) toxins in bacteria. Toxicon, 40, 543±549. 17. Hui,A. and de Boer,H.A. (1987) Specialized ribosome system: 42. Iost,I., Guillerez,J. and Dreyfus,M. (1992) Bacteriophage T7 RNA preferential translation of a single mRNA species by a subpopulation of mutated ribosomes in Escherichia coli. Proc. Natl Acad. Sci. USA, 84, polymerase travels far ahead of ribosomes in vivo. J. Bacteriol., 174, 4762±4766. 619±622. e78 Nucleic Acids Research, 2004, Vol. 32, No. 9 PAGE 14 OF 14 43. Draper,D.E. (1996) Strategies for RNA folding. Trends Biochem. Sci., 51. McCormick,J.R., Zengel,J.M. and Lindahl,L. (1994) Correlation of 21, 145±149. translation ef®ciency with the decay of lacZ mRNA in Escherichia coli. 44. Misra,V.K. and Draper,D.E. (2002) The linkage between magnesium J. Mol. Biol., 239, 608±622. binding and RNA folding. J. Mol. Biol., 317, 507±521. 52. Emory,S.A., Bouvet,P. and Belasco,J.G. (1992) A 5¢-terminal stem±loop 45. Tinoco,I.,Jr and Bustamante,C. (1999) How RNA folds. J. Mol. Biol., structure can stabilize mRNA in Escherichia coli. Genes Dev., 6, 135± 293, 271±281. 46. Brion,P. and Westhof,E. (1997) Hierarchy and dynamics of RNA folding. 53. Wong,H.C. and Chang,S. (1986) Identi®cation of a positive Annu. Rev. Biophys Biomol. Struct., 26, 113±137. retroregulator that stabilizes mRNAs in bacteria. Proc. Natl Acad. Sci. 47. Carpousis,A.J. (2002) The Escherichia coli RNA degradosome: USA, 83, 3233±3237. structure, function and relationship in other ribonucleolytic multienzyme 54. Bouvet,P. and Belasco,J.G. (1992) Control of RNase E-mediated RNA complexes. Biochem. Soc. Trans., 30, 150±155. degradation by 5¢-terminal base pairing in E. coli. Nature, 360, 488±491. 48. Grunberg-Manago,M. (1999) Messenger RNA stability and its role in 55. Chen,L.H., Emory,S.A., Bricker,A.L., Bouvet,P. and Belasco,J.G. (1991) control of gene expression in bacteria and phages. Annu. Rev. Genet., 33, Structure and function of a bacterial mRNA stabilizer: analysis of the 5¢ 193±227. untranslated region of ompA mRNA. J. Bacteriol., 173, 4578±4586. 49. Ross,J. (1995) mRNA stability in mammalian cells. Microbiol. Rev., 59, 56. Arnold,T.E., Yu,J. and Belasco,J.G. (1998) mRNA stabilization by the 423±450. ompA 5¢ untranslated region: two protective elements hinder distinct 50. Chen,C.Y. and Belasco,J.G. (1990) Degradation of pufLMX mRNA in pathways for mRNA degradation. RNA, 4, 319±330. Rhodobacter capsulatus is initiated by nonrandom endonucleolytic cleavage. J. Bacteriol., 172, 4578±4586. http://www.deepdyve.com/assets/images/DeepDyve-Logo-lg.png Nucleic Acids Research Oxford University Press

RNA stem–loop enhanced expression of previously non‐expressible genes

Download PDF
14 pages

Loading next page...
 
/lp/oxford-university-press/rna-stem-loop-enhanced-expression-of-previously-non-expressible-genes-D1Mq3XYm0f

References (56)

Publisher
Oxford University Press
Copyright
© Published by Oxford University Press.
ISSN
0305-1048
eISSN
1362-4962
DOI
10.1093/nar/gnh076
pmid
15163763
Publisher site
See Article on Publisher Site

Abstract

Published online May 26, 2004 Nucleic Acids Research, 2004, Vol. 32, No. 9 e78 DOI: 10.1093/nar/gnh076 RNA stem±loop enhanced expression of previously non-expressible genes Michael Paulus, Martin Haslbeck and Manfred Watzele* Roche Diagnostics, Nonnenwald 2, D-82377 Penzberg, Germany and Institut fu È r Organische Chemie und Biochemie, Fakulta Ètfu È r Chemie, Technische Universita ÈtMu È nchen, D-85747 Garching, Germany Received December 10, 2003; Revised February 21, 2004; Accepted May 4, 2004 ABSTRACT Even today the detailed mechanism of translation is not completely understood. However, much progress has been The key step in bacterial translation is formation of made toward understanding the structure of the complex the pre-initiation complex. This requires initial con- between the ribosome, bound mRNA and the tRNAs (6±8). tacts between mRNA, fMet-tRNA and the 30S sub- Initiation is the key step in bacterial translation (9±13). unit of the ribosome, steps that limit the initiation of Initial contacts between mRNA, fMet-tRNA and the 30S translation. Here we report a method for improving ribosomal subunit play major roles in formation of the pre- translational initiation, which allows expression of initiation complex and the control of translation ef®ciency. several previously non-expressible genes. This Hence, accessibility of the Shine±Dalgarno (SD) sequence method has potential applications in heterologous (ribosome binding site) and the start methionine have been shown to be important pre-requisites for successful initiation protein synthesis and high-throughput expression (9,10,12±16). The so-called SD interaction places the mRNA systems. We introduced a synthetic RNA stem±loop and the fMet-tRNA in the right position on the 30S ribosomal (stem length, 7 bp; DG = ±9.9 kcal/mol) in front of subunit. In this interaction the SD sequence hybridizes with various gene sequences. In each case, the stem± the anti-SD sequence on the 16S rRNA in the 30S ribosomal loop was inserted 15 nt downstream from the start subunit; the ef®ciency of this hybridization correlates with codon. Insertion of the stem±loop allowed in vitro ef®cient translation (17,18). expression of ®ve previously non-expressible genes The next step in initiation is association of the 50S and enhanced the expression of all other genes ribosomal subunit with the pre-initiation complex. Finally, investigated. Analysis of the RNA structure proved interaction of the second codon with its anticodon starts the that the stem±loop was formed in vitro, and demon- elongation process (11,19). strated that stabilization of the ribosome binding More or less stable structures of the mRNA translation site is due to stem±loop introduction. By theoretical initiation region have been investigated (10,14,20). These RNA structure analysis we showed that the inserted studies, which focused on the ribosome binding site and its RNA stem±loop suppresses long-range interactions tendency to hybridize with nearby sequences, reported that such structures, if their free energy was weaker than between the translation initiation domain and gene- ±6 kcal/mol, did not affect the ribosome. However, structures speci®c mRNA sequences. Thus the inserted RNA that were 1.4 kcal/mol more stable reduced expression rates stem±loop supports the formation of a separate 10-fold (10,20). translational initiation domain, which is more Usually, highly expressed Escherichia coli genes have accessible to ribosome binding. AU-rich codons immediately following the start methionine (21); GC-rich codons at these positions decrease expression INTRODUCTION rates (22). Pedersen-Lane and coworkers (22) reported that conversion of purine to thymidine bases at codon positions 3, 4 In recent decades heterologous protein synthesis has become and 5 increased expression up to 25%. They concluded that one of the most important tools in biotechnology. After the expression increased because the structure of the RNA in the genome of an organism has been sequenced, expressing the translation initiation region changed. Because AU-rich genes is a key step in understanding their structure and stretches are less likely to form secondary RNA structures, function. Currently, efforts to improve protein synthesis are the initiation site of the mRNA was more accessible to the 30S focused on high-throughput methods (1) and production of subunit of the ribosome. pharmaceuticals (2). In addition, sequences that enhance expression by allowing Ef®cient synthesis of recombinant proteins depends on, for additional base pairings between the messenger and the 16S example, transcription regulation, initiation of the messenger, rRNA have been described (23,24). These translational leader codon bias, mRNA stability and toxicity of the gene product. sequences improve the interaction of the messenger with the Thus many experimental strategies seek to optimize these ribosome. factors to achieve high expression rates (3±5). *To whom correspondence should be addressed. Tel: +49 8856 603121; Fax: +49 8856 607609; Email: Manfred.Watzele@Roche.com Nucleic Acids Research, Vol. 32 No. 9 ã Oxford University Press 2004; all rights reserved e78 Nucleic Acids Research, 2004, Vol. 32, No. 9 PAGE 2 OF 14 Figure 1. Overview of RNA stem±loop introduction. (A) Initiation region with SD sequence, start codon and gene-speci®c mRNA. (B) Gene-speci®c mRNA sequences fold back and pair with regulatory elements of bacterial protein translation. Thus translation initiation is prevented. (C and D) Introduction of the AU-rich region and the RNA stem±loop prevents regulatory elements from pairing with downstream gene sequences. (E) Diagram of the initiation region of 1049 RNA stem±loop mutant with the RNA stem±loop formed. The diagram shows the oligonucleotides used to probe the formation of the RNA stem±loop and their hybridization targets. However, if the initiation of the messenger is blocked by this leader sequence dramatically enhances the probability secondary RNA structures, no common translational leader ofsuccessful gene expression, and prevents the bacterial sequence to enhance expression has been described yet. initiation site from pairing with heterologous downstream Experiments on the regulation of the A-protein gene in the mRNA sequences. This leader made the translation of a set of RNA phage MS2 demonstrated that long-range interactions previously non-expressible genes possible and improved the expression of already expressible genes (e.g. GFP). In the between the translation initiation region and downstream in vitro system used, linear templates of GFP can achieve mRNA sequences prevented ribosomal initiation (16). expression rates up to 230 mg/ml (25). Similar initiation problems occurred in heterologous in vitro Based on these experiments, we conclude that inserting a protein synthesis, demonstrating the need for a generally 7 bp RNA stem±loop at the appropriate place in the mRNA suitable translational leader sequence. Since the initiation allows formation of an isolated translation initiation domain. region might be hidden by gene-speci®c mRNA structure (Fig. 1B), the leader sequence should alter the structure of wild-type mRNA to make the translation initiation region more accessible to ribosomes. In the present study, we inserted MATERIALS AND METHODS a stable local RNA hairpin loop downstream from the Generation of linear expression constructs initiation region to inhibit long-range interactions between the initiation region and gene-speci®c mRNA sequences Linear expression constructs (containing the T7 promoter, the (Fig. 1C and D). ribosome binding site (SD), the start methionine followed by We designed a translational leader sequence that contained the structural gene, the His tag, the stop codon and the T7 ®ve AU-rich codons, which show little tendency to form terminator sequences) were ampli®ed in a two-step PCR as secondary RNA structures, and a GC-rich RNA stem±loop (7 described (26). For wild-type constructs, the gene-speci®c bp; DG = ±9.9 kcal/mol; positions +19 to +36). Once inserted, sequence (Table 1A) was fused directly after the start codon. 0 PAGE 3 OF 14 Nucleic Acids Research, 2004, Vol. 32, No. 9 e78 Table 1. Generation of stem±loop mutants (A) List of genes used to generate linear expression constructs Gene Source organism DDBJ/EMBL/GenBank Gene-speci®c Gene-speci®c reverse primer Molecular weight accession no. forward primer of protein (kDa) 1049 Human cytomegalovirus M17209 GCTAACACCGCG GCGCCGGGTGCGCGA 9.3 Survivin Human NM001168 GGTGCCCCGACG ATCCATGGCAGCCAGC 17.2 CIITA Human U18259 GAGTTGGGGCCC AGAACCCCC 44 L30 r.p. Human m94314 AAGGTCGAGCTG GCGTTTTCCACCAAC 18 S4X isoform r.p. Human m58458 GCTCGTGGTCCC GCCCACTGCTCTGTTTGG 30 GFP Aequorea victoria U73901 ACTAGCAAAGGA AGAACCCCCCCC 28 Tubulin Human J00314 AGGGAAATCGTG ATGAGAACCCCC 50 (B) List of AT-rich combinations used to generate stem±loop mutants 1 Lys-Tyr-Thr-Tyr-Ser1 6 Lys-Thr-Tyr-Tyr-Ser2 AAATATACATATTCT AAAACATATTATTCA 2 Lys-Thr-Tyr-Tyr-Ser1 7 Lys-Tyr-Ser2-Tyr-Thr AAAACATATTATTCT AAATATTCATATACA 3 Lys-Tyr-Ser1-Tyr-Thr 8 Lys-Tyr-Tyr-Ser2-Thr AAATATTCTTATACA AAATATTATTCAACA 4 Lys-Tyr-Tyr-Ser1-Thr 9 His-His-His-His-His AAATATTATTCTACA CATCATCATCATCAT 5 Lys-Tyr-Thr-Tyr-Ser2 AAATATACATATTCA (C) General design used to construct primers for the ®rst PCR that will lead to an RNA stem±loop-mutant Forward primer: AGGAGATATACCATG-(AT-rich combination)-(RNA stem±loop)-(gene-speci®c forward primer) Reverse primer: ATTCGCCTTTTATTATTA-(His tag)-(gene-speci®c reverse primer) (D) Various stem±loop sequences that were inserted to the 3¢ side of the AT-rich combination RNA stem±loop Sequence Stem length 9 bp CAGACAAATAGATATTTGTCTGTA Stem length 8 bp CGTGCACGTGCATCGTGCACG Stem length 7 bp CTGCACGTGATCGTGCAG Stem length 6 bp CGCACGTGCATCGTGCGA Stem length 5 bp CGCCGTGCATCGGCG Stem length 4 bp GCCGTGATCGGC Stem length 3 bp GCGTGCATCGCA Stem length 2 bp CGTGCATCG Primer pairs listed in the table are the gene-speci®c sequences that hybridize to the template during the ®rst PCR ampli®cation. r.p., ribosomal protein. The overlap region for the second PCR is underlined in each primer. The AT-rich combination (chosen from Table 1B) was placed after the start methionine, followed by the RNA stem±loop and the gene-speci®c forward primer (chosen from Table 1A). The wild-type gene was ampli®ed by forward primers (from Table 1A) that did not contain either the AT-rich combination or the RNA stem±loop sequence. Hybridizing regions are shown in bold. An insert containing an AT-rich amino acid combination TCAAGACCCGTTTAGAGGCCCCAAGGGGTTGGGAG- (Table 1B) followed by the RNA stem±loop sequence TAGAATGTTAAGGATTAGTTTATTA) were used for the (Table 1D) was introduced between the start AUG and the second PCR. The DNA content of the second PCR product was estimated gene-speci®c sequence (Table 1A) to form the RNA stem± loop mutant. using the Lumi-Imager System (Roche, Basel, Switzerland) PCRs were performed in a volume of 50 ml in an Eppendorf and 100 ng of the product was used for in vitro expression. thermocycler (master cycler gradient, Eppendorf, Germany) Cloned linear templates were ampli®ed by one-step PCR, using standard protocols. A portion (2 ml) of the product from using the T7 promoter primer (GAAATTAATACGACTCA- the ®rst PCR was used as template for the second PCR. Primer CTATAGGGAGACCACAACGGTTTC) and the T7 termin- C (GAAATTAATACGACTCACTATAGGGAGACCACA- ator primer (CAAAAAACCCCTCAAGACCCGTTTAGA- ACGGTTTCCCTCTAGAAATAATTTTGTTTAACTTTA- GGCCCCAAGG) as PCR primers. The annealing temperature AGAAGGAGATATACC) and primer D (CAAAAAACCCC- during the ampli®cation was 60°C. e78 Nucleic Acids Research, 2004, Vol. 32, No. 9 PAGE 4 OF 14 Cloning of linear templates Before electrophoresis, samples were denatured in TBE±urea sample buffer (23) (Invitrogen) at 70°C for 5 min. The linear expression constructs were cloned into pBAD Topo After electrophoresis the fragment bands were stained with vectors (Invitrogen, Karlsruhe, Germany) according to the SYBR Green II (Sigma, Munich, Germany) (10 0003 in manufacturer's protocol (pBAD TOPO TA, Expression Kit, DMSO). The mRNA bands were analyzed by densitometry Version L). Recombinant plasmids were ampli®ed by one-step using the Lumi-Imager system (Roche). The molecular weight PCR to generate templates for `run-off' transcription (27) and marker contained a combination of RNA molecular weight for in vivo protein synthesis. marker III (Roche) (50 ng/lane) and a mixture of different oligonucleotides (75 ng/lane). (The sizes of the markers are In vitro protein synthesis given in the ®gures.) For bacterial protein synthesis, the Rapid Translation System Run-off transcription of mRNA RTS 100 E.coli HY Kit (Roche) was used according to the manufacturer's protocol. Each reaction (50 ml ®nal volume) Linear templates were produced from isolated plasmids in a contained 12 ml amino acids, 1.25 ml methionine, 10 ml one-step PCR using the T7 promoter and T7 terminator reaction mix, 4.75 ml reconstitution buffer, 12 ml E.coli lysate, primer. The SP6/T7 Transcription Kit (Roche) was used to 100 ng linear template DNA and 2 mg puri®ed mRNA make run-off transcripts of puri®ed mRNA from these template. The reaction was incubated for 2 h at 30°C without templates. Each transcription reaction contained 0.1±.5 mg stirring. Afterwards samples (0.1±1 ml) of the reaction were PCR product, the T7 RNA Polymerase (20 U) and dNTP analyzed by SDS±PAGE (28). (1.5 mM each). The reaction was incubated for 1 h at 37°C. DNase I (20 U) was then added and the reaction was incubated for an additional 15 min at 37°C. The resulting mRNA was In vivo protein synthesis puri®ed by phenol extraction and ethanol precipitation (29). The orientation of inserts in the pBAD plasmids was veri®ed The concentration and purity of the mRNA was measured by by PCR. Plasmids with the same orientation were transformed the optical density at 280 and 260 nm. into BL21 pLysS (Stratagene, Amsterdam, The Netherlands). RNA structure analysis with RNase H For in vivo protein synthesis, colonies were isolated and grown in 4 ml cultures at 37°C and 200 r.p.m. for 5 h. When cell The structure of the mRNA was analyzed by oligonucleotide- densities reached ~10 cells/ml, T7 transcription was induced mediated cleavage with RNase H (30±32). In the RNase with 1 mM ispropyl-b-D-thiogalactopyranoside (IPTG) and cleavage reaction, each of four oligonucleotides [Fig. 1E, (a)± the cells were incubated for an additional 2 h. Samples (10 (d)] was used at three different concentrations (75, 7.5 and cells) were removed from the culture and centrifuged (3 min at 0.75 mM). Oligonucleotides (e) and (f) (Fig. 1E) were each 14000 r.p.m. on a table-top centrifuge). Pelleted cells were used at only one concentration (36 mM). For titration of the resuspended in 10 ml SDS sample buffer and analyzed by ribosome binding site, an oligonucleotide (TCTCCT) com- SDS±PAGE and western blot. plementary to the site was used. Reactions were performed in a reaction volume of 10 ml SDS±PAGE/western transfer containing 50 ng/ml mRNA. First the mRNA samples were denatured at 70°C for 2 min in incubation buffer (25 mM Tris± Electrophoresis and western transfer were performed in the HCl pH 7.5, 200 mM NaCl, 10 mM MgCl 3 6H O). Then the 2 2 Novex Pre cast gel system (Invitrogen) according to the samples were cooled to 30°C and the mRNA was allowed to manufacturer's protocol. Samples from expression reactions refold for 30 min (32). Next, oligonucleotides were added to were denatured for 10 min at 70°C and separated on 4±12% each sample. To allow hybridization the samples were Tris±glycine polyacrylamide gels (Invitrogen). The Multi- incubated for 10 min at 30°C. RNase H (1 U) was then Tag-Marker (Roche) was used as molecular weight marker. added to each sample and digestion was performed for an After electrophoresis the proteins were transferred to PVDF additional 30 min at 30°C. To stop the reaction, one volume membrane (Roche) by electroblotting, according to the TBE±urea sample buffer (23) (Invitrogen, Karlsruhe, manufacturer's protocol. Germany) was added. Portions of each digest (50±350 ng His-tagged proteins on the membrane were detected with a mRNA) were analyzed by denaturing TBE±PAGE. 1:4000 dilution of Anti-His -POD-MAK (Roche). Antibody was incubated with the membrane in TBST buffer [TBS RNA stability assay (50 mM Tris, 150 mM NaCl pH 7.5) containing 1 ml/l Tween The RTS 100 HY in vitro expression system was used for the 20 (Roche)]. stability assay. Samples containing 1 mg mRNA were mixed Alternatively the His tag was detected with NiNTA-AP- on ice with one of four dilutions of E.coli lysate (13, 0.53, Conjugate (1:1000 dilution in TBST) (Qiagen, Hilden, 0.253, 0.1253). After lysate was added, the samples were Germany) followed by visualization with CDP-star reagent mixed carefully and incubated at 30°C for 20 min. After the (Roche). The chemiluminescent signals were monitored with incubation, two volumes of lysis/binding buffer from the High the Lumi-Imager System (Roche). Pure RNA Isolation Kit (Roche) were added to each sample to inactivate RNases. The mRNA remaining in each sample was TBE±PAGE then isolated with the High Pure Kit. Ribonucleic acids were electrophoretically separated on After isolation the samples were analyzed on polyacryla- denaturing polyacrylamide gels, which contained 6% or mide (6%)±TBE±urea or agarose gels. The mRNA amounts 15% polyacrylamide, TBE and urea (7 M) (Invitrogen). were analyzed by densitometry. PAGE 5 OF 14 Nucleic Acids Research, 2004, Vol. 32, No. 9 e78 Table 2. List of forward primers for the ®rst PCR used to generate linear expression constructs of GFP (A) Forward primers that generate RNA stem±loops of increasing stem length to the 3¢ side of the start AUG Forward primer Sequence Without stem±loop (0 bp) AGGAGATATACCATGACTAGCAAAGGAGAA Stem length 4 bp AGGAGATATACCATGACTAATTTTAGTACTAGCAAAGGAGAA Stem length 5 bp AGGAGATATACCATGACTGTTTATACAGTAACTAGCAAAGGAGAA Stem length 6 bp AGGAGATATACCATGACTGGTCAATTACCAGTAACTAGCAAAGGAGAA Stem length 7 bp AGGAGATATACCATGACTGCTTTACATCAAGCAGTAACTAGCAAAGGAGAA Stem length 8 bp AGGAGATATACCATGACTGCACGTGATCGTGCAGTAACTAGCAAAGGAGAA (B) Different locations chosen for insertion of the 8 bp stem±loop Forward primer Sequence GFP +10 nt AGGAGATATACCATGACTAGCACT...GTAAAAGGAGAAGAACTT GFP +13 nt AGGAGATATACCATGACTAGCAAAACT...GTAGGAGAAGAACTTTTC GFP +16 nt AGGAGATATACCATGACTAGCAAAGGAACT...GTAGAAGAACTTTTCACT GFP +19 nt AGGAGATATACCATGACTAGCAAAGGAGAAACT...GTAGAACTTTTCACTGGA GFP +22 nt AGGAGATATACCATGACTAGCAAAGGAGAAGAAACT...GTACTTTTCACTGGAGTT The start AUG is bold, stem±loop sequences are inverted and the overlap region for the second PCR is underlined. The reverse primer is the same as that listed for GFP in Table 1C. ACT...GTA, 8 bp stem±loop (ACTGCACGTGATCGTGCAGTA). Theoretical RNA structure analysis downstream gene-speci®c sequences) of the introduced stem± loop was then calculated and expressed as a percentage: For theoretical RNA structure analysis, mfold software (33) was used. The messengers were folded at 30°C, and dot plot p = 100% 3 [1 ± p (mutant)/p (WT)] 2 r Gji Gji analysis was used to identify all pairings (j) of regulatory sequences (SD; AUG) with downstream sequences. For The probability (p ) that an RNA stem±loop will form was RST analysis of long-range interactions, only pairings between calculated from the theoretical data (equation 1) by consider- regulatory sequences and downstream gene-speci®c mRNA ing only structures in which the RNA stem±loop actually sequences were considered. occurred (index RST): According to the theory behind the software, if n is the number of structures examined, j is a downstream pairing 1 DG ÿ DG RST max between a regulatory element and a downstream gene-speci®c n DG ÿ DG min max sequence, i is any structure in which pairing j occurs and DG i RST p ˆ 3 RST is the free energy of structure i, which determines the 1 DG ÿ DG n max frequency of occurrence of structure i, the probability p Gji n DG ÿ DG min max that the regulatory mRNA will form secondary RNA struc- tures with downstream gene-speci®c sequences can be calcu- lated as XX 1 DG ÿ DG RESULTS i max n DG ÿ DG min max In vitro expression experiments to de®ne the j i p ˆ 1 ji X translational leader sequence 1 DG ÿ DG n max n DG ÿ DG In the initial set of experiments we studied the in¯uence of the min max stem length and the location of the RNA stem±loop on translation. We used different constructs of a gene (GFP) where the numerator of the equation represents the probability which is usually highly expressed. Various stem±loops were that all downstream pairings of type j will occur, and the introduced by using different forward primers in the ®rst PCR denominator represents the probability that all hypothetically (Table 2). The stem±loops were placed directly after the 3¢ end possible downstream pairings between the regulatory elem- of the start AUG and some of them signi®cantly in¯uenced ents and any downstream sequence will occur. expression rates (Fig. 2A). The probability p for downstream pairing of the wild-type Inserts with a stem length of 5 bp or less did not Gji sequence was compared with that of the stem±loop mutant. signi®cantly affect the expression rates (~100 mg/ml), but The reduced downstream af®nity (potential to release p , i.e. stem lengths from 6 to 8 bp (DG values of ±7.8 to ±11.8 kcal/ potential to decrease pairing of a given RNA sequence with mol) reduced expression signi®cantly (Fig. 2A). Hence, a e78 Nucleic Acids Research, 2004, Vol. 32, No. 9 PAGE 6 OF 14 Figure 2. Expression results of various stem±loop mutants. (A) Expression results of constructs that contain different stem±loops inserted after the 3¢ end of the start AUG. The position of the stem±loops and their free energies are shown. One microliter was taken from each of two independent expression reactions involving each construct. These samples were run on 10% acrylamide±Bis/Tris±SDS gels and then examined by western blot analysis. The His-tagged proteins on the blot were detected with NiNTA-AP conjugate and CDP-star; CDP-star luminescence was measured with the Lumi-Imager. The amounts of luminescence due to His-tagged GFP in each sample are shown in the graph. These amounts were veri®ed by measurement of the GFP ¯uorescence in each sample. (B) Expression results from constructs containing the inhibitory 8 bp stem±loop inserted at different positions in the gene sequence. The diagram shows the various insertion positions of the inhibitory 8 bp stem±loop in the constructs. The graph shows luminescence data representing the relative expression of each construct. stem±loop structure with a stem length of at least 6 bp and a At insertion position +13 the inhibitory effect started to free energy of ±7.8 kcal/mol reduces expression signi®cantly vanish (Fig. 2B); at position +16 expression was no longer by blocking ribosomal initiation. affected. This observation is in good agreement with data on In the subsequent expression experiments (Fig. 2B) we used the structure of the ribosome, since the space required by the GFP constructs with an inhibitory 8 bp stem±loop. The 8 bp messenger in the downstream tunnel of the ribosome ranges element was inserted at different positions downstream from from +11 to +15 nt (6±8). the start codon (Table 2B). This allowed us to investigate the Based on these initial experiments we designed a transla- optimal placement of an 8 bp stem±loop (Fig. 2B). tional leader sequence to enhance expression (Fig. 1). This PAGE 7 OF 14 Nucleic Acids Research, 2004, Vol. 32, No. 9 e78 Table 3. Amounts of protein synthesized in vitro. Gene Relative expression (n 3 GFP WT) CIITA 1 1.4 Survivin 1 2.3 1049 8 1.5 GFP WT 1.0 GFP 6 3.7 For quanti®cation of synthesis, the chemiluminescence signal was measured after immunochemical detection of His-tagged protein. Protein amounts were estimated relative to wild-type GFP. The mutant numbers indicate which AT-rich sequences (listed in Table 1B) were inserted to the 5¢ side of the 7 bp stem±loop. (Fig. 3F and G). When inserted along with the stem±loop, AT- rich combinations 1±8 (Fig. 3, lanes 1±8) produced high expression rates, but combination 9 (Fig. 3, lane 9) produced very little expression. The AT-rich region in this combination (Table 1B, sequence 9, and Fig. 3, lane 9) is the His tag sequence, which seems to be less suitable for expression initiation. Because the amount of synthesized protein depends on the expression system and the type of template (linear DNA or plasmid), in an additional western blot (data not shown) we estimated their expression relative to the GFP wild-type construct (Table 3). According to attenuation mechanisms (34±36), tRNA starvation could have caused the lower expression. Figure 3. In vitro expression results from linear constructs of wild-type and Alternatively, the higher GC content, which increased the RNA stem±loop mutants. (A±E) Western blot analysis of in vitro expression tendency to form secondary RNA structures, might lead to the of ®ve different wild-type genes and their corresponding stem±loop mutants. decrease in expression ef®ciency (21,22). In each blot, lane 10 shows that the wild-type genes were not expressed. However, codon bias could be excluded as the cause of the Lanes 1±9 show the expression of constructs containing identical RNA stem±loop sequences and an AT-rich region. The amino acids of the improved expression because supplementing the reaction with AT-rich region were varied as described in Materials and Methods (Table tRNAs that provide rare codons in E.coli did not lead to in vitro 1B). Lanes 1±9 in the ®gure correspond to constructs 1±9 in Table 1B. expression of the wild-type genes (data not shown). (F and G) Western blot analysis of two expressible wild-type genes and In a subsequent set of experiments, an N-terminal HA tag their corresponding RNA stem±loop mutants. The samples on these gels were loaded in the same order as those on gels (A)±(E). was used as the AT-rich region. It was less suited for initiating expression, but it allowed detection of the N-terminus of the translation product. Thus potential fragmentation during the translation could be detected. HA-tagged constructs of sequence contained 5 AU-rich codons (positions +4 to +18 nt) the S4X ribosomal protein (also shown in Fig. 3D), which had followed by a GC-rich sequence that coded for an RNA stem± coding sequences of increasing length, showed in vitro loop (7 bp, DG = ±9.9 kcal/mol, positions +19 to +36 nt) expression without fragmentation (data not shown). (Table 1). In order to investigate whether the RNA stem±loop acts on translation, puri®ed mRNA was used as a template for in vitro In vitro transcription/translation of linear expression translation. The experimental conditions for the translation constructs were comparable with the coupled transcription/translation reactions (described in Materials and Methods). Variants of To investigate the in¯uence of the arti®cial translational leader CIITA, Survivin and 1049 mRNAs, along with the corres- sequence on gene expression we compared its effect on the ponding wild-type genes, were translated. Expression results expression of seven different genes from different organisms were similar to those obtained in the coupled in vitro (Table 1A). The coding sequences of these genes were transcription/translation. The RNA stem±loop mutants between 246 and 1380 nt long. As described in Materials and showed signi®cant expression, while the wild-type mRNAs Methods, the translational leader sequence was introduced by of all three genes were not detectably translated (data not two-step PCR, using the 7 bp stem±loop (Table 1D) in the shown). forward primer of the ®rst PCR (Table 1C) to generate linear expression constructs. In vitro transcription/translation using different leader Insertion of the RNA stem±loop signi®cantly increased sequences expression of wild-type genes that were initially not expressed or only slightly expressed (Fig. 3A, B, C, D and E). To investigate whether the stem±loop or the AT-rich sequence Additionally, the stem±loop insert signi®cantly increased the was responsible for improved expression, linear expression expression of genes that already had good expression rates constructs containing only one element (AT-rich sequence) e78 Nucleic Acids Research, 2004, Vol. 32, No. 9 PAGE 8 OF 14 presence of the rather unstable 2 bp stem±loop, 1049 was highly expressed while CIITA was less effectively expressed. Stem lengths <5 bp decreased expression of all genes. Stem lengths >6 bp increased expression of all genes; maximum expression was obtained with a 7 bp stem. The results demonstrate that a 7 bp stem±loop improved expression in all cases, while the effects of shorter or longer stem lengths depended on the structural gene. Formation of the introduced RNA stem±loop Next we investigated formation of the RNA stem±loop structure in the mutated mRNA. We used complementary oligonucleotides (Fig. 1E) to target the stem sequence rather than sequences to either side of the stem±loop. Hybridization of these oligonucleotides was detected by RNase H cleavage. Single-stranded RNA is accessible to hybridization and allows cleavage, while double-stranded RNA stretches are not accessible and therefore remain uncleaved by RNase H (30±32). Additionally, hybridization of increasing amounts of oligonucleotides competes against native RNA structure at the targeted site and demonstrates the accessibility, and hence stability, of the secondary RNA structure (32,37). As indicated in Figure 5A, three different oligonucleotide concentrations were used for cleavage reactions. Oligonucleotide (a) mediated cleavage (fragment size, 62 nt); oligonucleotide (b) also resulted in cleavage (fragment size, 98 nt). Different concentrations of oligonucleotides led to comparable amounts of fragments. This indicated that the sites before and after the stem±loop were easily accessible to Figure 4. In vitro expression results using different leader sequences. oligonucleotide hybridization. In contrast, oligonucleotides (A) Expression results obtained from linear templates with and without an (c) and (d) could not hybridize to mediate cleavage at any RNA stem±loop. Lanes 1±5 show expression results from different CIITA concentration investigated (Fig. 5A). Only secondary cleavage constructs. Lanes 6±10 show results from Survivin constructs. Lanes 11±15 products in the lower part of the gel were detected. show results from 1049 constructs. Lanes 1, 6 and 11 show wild-type Because the length of oligonucleotides has previously been expression. Lanes 2, 3, 7, 8, 12 and 13 show expression of constructs con- taining AT-rich combinations (Table 1B, sequences 1 and 2) and the intro- shown to in¯uence hybridization (38), oligonucleotides (e) and duced RNA stem±loop. Lanes 4, 5, 9, 10 14 and 15 show expression of (f), containing sequences complementary to both stem± and constructs containing the same AT-rich combinations (Table 1B, sequences hairpin±loop [Fig. 1E, sequences (e) and (f)] were tested in the 1 and 2) but no RNA stem±loop sequence. (B±E) Expression results ob- RNase H cleavage reaction. Performing this reaction in the tained from linear templates with different stem lengths inserted on the 3¢ side of the AT-rich combination. The stem length and the wild-type con- presence of 36 mM oligonucleotide (e) or (f) generated no structs are indicated above the lanes. Similar amounts of (B) 1049 cleavage products. Thus the RNA stem±loop is formed. (C) CIITA and (D) Survivin were analyzed by western blot and quanti®ed In addition, theoretical structure analysis was performed to with the Lumi-Imager. Each expression result in (B) and (C) was expressed determine the probability of RNA stem±loop formation in as a fraction of the result from the most active sample (the sample contain- different genes (Fig. 5B). This analysis showed signi®cant ing the 7 bp stem±loop) and plotted in (E). formation of the 7 bp stem±loop. Only for Survivin mRNA did theoretical values indicate high af®nity of the gene sequence for the GC-rich sequence of the stem±loop. were generated (Table 1B, sequences 1 and 2). The results of Structural changes of regulatory elements on the mRNA the in vitro expression are shown in Figure 4A. While the by introduction of the RNA stem±loop wild-type genes of CIITA, Survivin and 1049 showed no expression (Fig. 4A, lanes 1, 6 and 11), the constructs To test secondary RNA structure formation in translational containing the RNA stem±loop sequence (Fig. 4A, lanes 2, 3, regulatory elements we analyzed the accessibility of the 7, 8, 12 and 13) were expressed in high yields. The constructs ribosome binding site by oligonucleotide titration and subse- containing the AT-rich region but no RNA stem±loop quent RNase H digestion (Fig. 6). The results obtained with sequence (Fig. 4A, lanes 4, 5, 9, 10, 14 and 15) showed little mutant and wild-type mRNAs were compared (32). or no expression. Thus the RNA stem±loop is absolutely For all genes investigated, comparable oligonucleotide necessary for successful expression of non-expressible wild- concentrations produced more cleavage products with wild- type genes. type mRNAs than with RNA stem±loop mutants. According to To investigate how stem±loop characteristics affected principles of antisense technology the kinetics and thermo- improved translation we varied the length of the stem±loop dynamics of oligonucleotide hybridization are profoundly inserted at position +19 (Table 1D). We tested different stem attenuated when hybridization has to overcome energy lengths between 2 and 8 bp (Fig. 4B, C, D and E). In the barriers caused by secondary RNA structure (37,39). Hence PAGE 9 OF 14 Nucleic Acids Research, 2004, Vol. 32, No. 9 e78 Figure 5. Experimental (RNase H) and theoretical (mfold) analysis of RNA stem±loop formation. (A) TBE±urea gel containing RNase H cleavage products generated in the presence of oligonucleotides that are complementary to parts of the RNA stem±loop. Each sample on the gel contains denatured products from a single cleavage reaction (10 ml volume, containing 250 ng mRNA). Each lane is labeled with the name and concentration of the oligonucleotides used to generate the cleavage pattern. Names of the oligonucleotides correspond to those described in Figure 1E. Oligonucleotide (a) mediated cleavage to the 5¢ side of the RNA stem±loop, while oligonucleotide (b) mediated cleavage to the 3¢ side of the RNA stem±loop. No cleavage of the targeted RNA stem±loop was observed in the presence of oligonucleotides (c), (d), (e) and (f). Lane 0, cleavage reaction produced in the absence of oligonucleotide. (B) Results of theoretical mfold analysis of stem±loop formation in different genes. p , probability of RNA stem±loop formation. RST lower hybridization re¯ects more stable secondary RNA downstream af®nity for both the ribosome binding site (64.8 to structures (32). Thus the ribosome binding sites of the wild- 92.4%) and the start AUG (69.2 to 93.4%). For Survivin, type mRNAs were stabilized by the introduction of the RNA introduction of the stem±loop led to less marked release from stem±loop. downstream pairings for both regulatory elements, 35.1% for the SD and 49.1% for the start AUG. Only CIITA showed more downstream pairings of the start methionine in the RNA Theoretical RNA structure analysis stem±loop mutant than in the wild type. However, stem±loop To predict pairings of the translation initiation site with introduction still released the ribosome binding site (52.1%) of downstream gene-speci®c mRNA sequences, we calculated CIITA from downstream secondary RNA structures. theoretical RNA structures from the primary sequences. The The AUG of CIITA showed more af®nity for downstream probabilities of downstream pairings between the ribosome sequences (increased af®nity, p = ±48.8%) because it was binding site and the start methionine were calculated (Fig. 7A). possible for nucleotides ±2 to +3 to pair with nucleotides +213 The reduction of downstream af®nity after stem±loop inser- to +217 in a 5 bp helix [(a) DG = ±8.10 kcal/mol]; tion is also indicated (Fig. 7B). this could be followed by formation of an interior loop (DG In summary, the data predict that stem±loop introduction = 2.20 kcal/mol) and a 5 bp helix involving nucleotides +7 to decreases downstream pairings of the two regulatory elements +11 and +207 to +211 [(b) DG = ±5.20 kcal/mol]. Because (SD, start methionine) in all mRNAs. this structure occurs with great frequency, the probability for The ribosome binding site of the GFP RNA stem±loop downstream pairings of the AUG increased after stem±loop mutants showed lower downstream af®nity (49.8 to 52.8%), introduction. The free energy of helix (a) from the paired AUG and pairings of the start methionine with sequences down- was compensated by the free energy of the interior loop. stream were also decreased (73.0 to 77.7%). RNA stem±loop Hence the AUG was not bound strongly enough to inhibit mutants of gene 1049 showed the most signi®cant changes in translation (20). In addition, downstream pairings of the e78 Nucleic Acids Research, 2004, Vol. 32, No. 9 PAGE 10 OF 14 After the stability tests the amounts of mRNA remaining were determined by gel electrophoresis and densitometry. For each variant, the stability of mutant and wild-type mRNAs was comparable at different lysate concentrations. At lower lysate concentrations (Fig. 8B, 0.1253 and 0.253) the mRNA of the stem±loop mutant might be slightly more stable. However, at lysate concentrations that mimicked expression conditions (Fig. 8B, 13) wild-type mRNA was more stable than the stem±loop mutant. In vivo protein synthesis of the RNA stem±loop constructs We also tested expression of RNA stem±loop constructs in whole cells. pBAD vectors containing linear templates in the same orientation were transformed into E.coli BL21 pLysS and expressed as described in Materials and Methods. Western analysis of the His-tagged proteins showed that three independent clones of 1049 containing the RNA stem±loop construct (Fig. 9, lanes 1±3) were highly expressed, while three independent clones containing the wild-type 1049 (Fig. 9, lanes 4±6) were poorly expressed. Three independent clones of Survivin containing the RNA stem±loop construct (Fig. 9, lanes 7±9) were highly expressed, while two wild-type clones (Fig. 9, lanes 10 and 11) were not expressed. DISCUSSION Introducing an engineered stem±loop sequence into the gene led to good in vitro expression of previously non-expressible genes. This strategy was effective for several genes, which were obtained from different organisms and varied in length from 246 to 1380 nt. The inserted stem±loops also enhanced expression of these genes in vivo. Thus insertion of stem±loop sequences appears to be a generally applicable strategy for enhancing heterologous gene expression. Whether transcription and translation of mRNA were Figure 6. Determination of structural accessibility of the ribosome binding performed separately or in a coupled system, the results site. TBE±urea gels showing RNase H cleavage products obtained after the ribosome binding site was exposed to increasing concentrations of a were similar: no expression of wild-type mRNAs and good complementary oligonucleotide. The same oligonucleotide was used in all expression of RNA stem±loop mutants. Hence the introduced mRNA titrations. Each lane was loaded with 5 ml from a single cleavage sequence acts at the RNA level, and no other parts of the reaction (representing 125 ng mRNA). The bar graph above each gel shows pathway from DNA to mRNA are affected. It is possible, the cleavage in each sample, expressed as a percentage of the maximum cleavage observed. (A) Titration of 1049 mRNA; lanes 1±5 contained the however that RNA stem±loop introduction might affect steps RNA stem±loop mutant, while lanes 6±10 contained the wild type. As after mRNA production, for example by in¯uencing transla- indicated on the gel, both mRNAs were exposed to increasing amounts of tion initiation or mRNA stability. oligonucleotide (0±1500 ng/reaction). (B) Titration of Survivin mRNA and In addition, the translation results con®rm that the in vitro (C) titration of CIITA mRNA; lanes 1±5 contained the RNA stem±loop expression experiments are a valuable method for investigat- mutant, while lanes 6±10 contained wild type. As indicated, mRNAs were exposed to increasing amounts of oligonucleotide (0±2500 ng/reaction). ing the effects of RNA structures on RNA±protein interactions (10,16,20,41). Constructs that contain only the AT-rich combination have very low expression. This indicates that the RNA stem±loop is interior loop and helix (b) are formed in the presence of the important for suf®cient expression of non-expressible genes. introduced AU-rich sequence; in agreement with data from the Furthermore, expression experiments on constructs containing literature (21,22), the AU-rich sequence enhanced translation different stem±loop sequences showed that expression de- initiation. pends on mRNA structure. Translational pausing is involved in the attenuation mech- mRNA stability anism and is caused by stable RNA secondary structures. It Because differences in mRNA stability could have led to affects transcription and translation ef®ciency (34±36). improved translation results (40), we compared the stability of However, translational pausing and attenuation mechanisms mRNA from the mutant version of gene 1049 with that of can be excluded as causes of improved protein synthesis for mRNAs from wild-type 1049 in an in vitro expression system. two reasons. PAGE 11 OF 14 Nucleic Acids Research, 2004, Vol. 32, No. 9 e78 Figure 7. Downstream af®nity of the translation initiation domain. (A) Probability p that the ribosome binding site (RBS) and the start AUG will pair with Gji downstream sequences. Results are shown for wild-type mRNAs (WT) and stem±loop mutants (numbered) of the genes 1049, Survivin and CIITA. The numbers of the mutants correspond to the amino acid sequences shown in Table 1B. (B) Reduction in downstream af®nity p of the mutant regulatory sequences compared with the wild type. The higher the p value, the less likely it is that the regulatory sequence will pair with downstream sequences. First, the 7 bp stem±loop is positioned at +19 to +36, while downstream gene-speci®c mRNA sequences. Therefore chan- the initiating ribosome spans positions ±15 to +15 (6,8). ging the structure of the ribosome binding site resulted in more During elongation the movement of the ribosome disrupts de®ned structures within the sequences of the 5¢ UTR. This base pairings of nucleotides +19 to +25 in the RNA stem± conclusion is consistent with the hierarchical pathway of RNA loop. After it opens the stem, that ribosome can no longer be structure formation, in which short hairpin structures are more affected by the stem±loop. Also, while one ribosome holds the stable and are formed ®rst, before structures involving long- stem open, there is not enough space on the mRNA for range interactions (43±46). Hence the same oligonucleotide initiation by another ribosome. used for structural titration (targeted to the RBS) hybridizes Second, T7 RNA polymerase is used in the coupled in vitro more ef®ciently to its target if the target is involved in several expression system. Because the T7 RNA polymerase elong- alternative long-range interactions rather than `one' hairpin ates the RNA chain rapidly, transcription is independent of structure. Together with the improved expression observed in translation (42) and RNA structures are formed before the stem±loop mutants, these observations support the ribosomal initiation. conclusion that stem±loop introduction led to formation of a Our experiments on the stability of the RNA structure of the structurally isolated initiation domain, consisting primarily of ribosome binding site (RBS) indicated that it was stabilized short local hairpins with DG < ±6 kcal/mol (10,15,20). During after stem±loop introduction in all three genes investigated. translation initiation such structures can easily be opened by RNA folding analysis with mfold software (33) concentrated the ribosomes. on detecting downstream pairings of the RBS and/or the start Stability of the mRNA plays a fundamental role in methionine. Stem±loop introduction reduced the af®nity of the regulating protein amounts in the cells (47±49). The stability regulatory elements for downstream gene-speci®c mRNA of the messenger controls the number of translation starts that sequences. Compared with the wild-type genes, the RNA occur during the lifetime of the mRNA. Degradation of stem±loop mutants reduced downstream af®nity of the mRNA is not a random process of endonucleolytic cleavage ribosome binding site by 35.1 to 92.4% (49.1 to 93.4% for (50,51), but is mediated by a combination of RNA helicases the start methionine). Hence, stem±loop insertion inhibited the and endo- and exoribonucleases (47), which are combined in formation of alternative structures between the RBS and the E.coli degradasome. The rate-limiting step of degradation e78 Nucleic Acids Research, 2004, Vol. 32, No. 9 PAGE 12 OF 14 Figure 9. Comparison of in vivo expression of RNA stem±loop constructs and wild-type genes. Western blot analysis of protein synthesis from three independent clones of RNA stem±loop mutant 1049 1 (lanes 1±3); wild-type 1049 (lanes 4±6), three independent clones of a Survivin RNA stem±loop mutant (lanes 7±9) and two independent clones of wild-type Survivin (lanes 10 and 11). Similar amounts of protein were analyzed in each sample. Table 4 summarizes expression results obtained by insertion of various stem±loops and theoretical analysis of the RNA structure. In some genes a rather unstable 2 bp stem±loop was suf®cient to initiate expression (Table 4, 2 bp element). Although the constructs containing the AU-rich combination without a stem±loop allowed low expression of 1049, this combination did not allow expression of the other genes investigated (Table 4, AT+/RST±). Theoretical analysis of the Figure 8. mRNA amounts remaining after 20 min incubation with different RNA structures explained these results. The decrease in the concentrations of E.coli lysate. (A) TBE±urea gels show the mRNA remaining in wild-type and RNA stem±loop mutants after 20 min digestion. value of predicted RNA stem±loop formation (Table 4, p ) RST (B) The amounts of mRNA detected after digestion for 20 min, as correlates with the expression results. The Survivin mRNA determined by densitometric analysis of the gel. The error bars show the shows high af®nity for the GC-rich stem±loop sequence, standard deviations obtained for duplicate samples. indicating a general tendency to form secondary RNA structures. The relatively greater effect of the 2 bp element is endonucleolytic cleavage at the 5¢ end, followed by on CIITA expression agrees with the theoretical prediction directional cleavage from the 3¢ to the 5¢ end (47,48). Thus that stem±loop insertion increases stem±loop formation in 5¢ and 3¢ RNA secondary structures in the untranslated region CIITA more than in Survivin, but decreases the downstream act as stabilizers against degradation and extend the lifetime of af®nity of the regulatory elements by about the same amount a messenger (40,52,53). in both genes. Finally, gene 1049 shows the most signi®cant We did not measure mRNA decay precisely, but experi- enhancement of expression by the 2 bp stem±loop, as well as ments on the stability of the mRNA in the in vitro expression producing the highest values in the theoretical analysis. Taken system demonstrate that introduction of the RNA stem±loop together, the results indicate that the degree to which the did not stabilize the wild-type messenger. These observations mRNA structure enables successful expression depends on the are in agreement with data indicating that occurrence of at gene involved. least ®ve unpaired nucleotides upstream from stable second- In summary, the experiments demonstrate the importance of ary RNA structures at the 5¢ end of the messenger reverse the the RNA stem±loop for mediating translation initiation. Once stabilizing effect of these secondary structures (52,54±56). introduced, the RNA stem±loop forms and stabilizes the Additionally, signi®cant amounts (33%) of the wild-type ribosome binding site in a translatable RNA structure. Long- mRNA remained in the undiluted lysate after 20 min. Other range interactions of the regulatory sequences are critically experiments showed that after 20 min most of the protein in important, and the introduced RNA stem±loop acts as a barrier the coupled in vitro system is already synthesized (data not against hybridization of regulatory elements to downstream shown). Thus, if ribosomes initiated synthesis from the wild- gene-speci®c mRNA sequences. Thus introduction of a type messenger, protein would have been detected after stem±loop enhances formation of a distinct bacterial transla- 20 min. tion initiation domain. Table 4. Comparison of experimental results with theoretical RNA structure analysis Gene Expression (2 bp element) Expression (AT+/RST±) p p (RBS) (%) p (AUG) (%) RST r r 1049 0.97 Low 1 64.8±92.4 69.2±93.4 CIITA 0.48 No 1 52.1 ±48.8 Survivin 0 No 0.62 35.1 49.1 Each value for the 2 bp stem±loop element is expressed as a fraction of the value obtained with the 7 bp element (as shown in Fig. 4B±E). AT+/RST± refers to expression of constructs containing the AT-rich codon combination but no RNA stem±loop. PAGE 13 OF 14 Nucleic Acids Research, 2004, Vol. 32, No. 9 e78 18. Jacob,W.F., Santer,M. and Dahlberg,A.E. (1987) A single base change in Furthermore, the formation of a stable local hairpin the Shine±Dalgarno region of 16S rRNA of Escherichia coli affects structure of ~7 bp (DG = ±9.9 kcal/mol), 15 nt downstream translation of many proteins. Proc. Natl Acad. Sci. USA, 84, 4757±4761. from the start methionine, enhances heterologous gene 19. Ringquist,S., Shinedling,S., Barrick,D., Green,L., Binkley,J., expression. For maximum effect, these 15 intervening Stormo,G.D. and Gold,L. (1992) Translation initiation in Escherichia nucleotides should be AU rich. coli: sequences within the ribosome-binding site. Mol. Microbiol., 6, 1219±1229. Although the RNA stem±loop structure is a valuable tool for 20. deSmit,M.H. and van Duin,J. (1994) Control of translation by mRNA enhancing expression yield, the cost of that enhanced expres- secondary structure in Escherichia coli. A quantitative analysis of sion is additional sequences in the ®nal gene product. In some literature data. J. Mol. Biol., 244, 144±150. cases this might impair the function of the protein. To 21. Stenstrom,C.M., Jin,H., Major,L.L., Tate,W.P. and Isaksson,L.A. (2001) overcome this problem, conservative codon replacement Codon bias at the 3¢-side of the initiation codon is correlated with translation initiation ef®ciency in Escherichia coli. Gene, 263, 273±284. might be used to generate the stem±loop structure in the 22. Pedersen-Lane,J., Maley,G.F., Chu,E. and Maley,F. (1997) High-level leader sequence. Alternatively, post-translational chemical or expression of human thymidylate synthase. Protein Expr. Purif., 10, enzymatic processing might be used to remove the additional 256±262. sequences. 23. Studier,F.W., Rosenberg,A.H., Dunn,J.J. and Dubendorff,J.W. (1990) Use of T7 RNA polymerase to direct expression of cloned genes. Methods Enzymol., 185, 60±89. ACKNOWLEDGEMENTS 24. Sprengart,M.L., Fuchs,E. and Porter,A.G. (1996) The downstream box: an ef®cient and independent translation initiation signal in Escherichia We thank Marcus Schmidt, Anette Seidl, Claudia Kirr, coli. EMBO J., 15, 665±674. Thomas Nikolaus and Bjo È rn Voss for experimental support. 25. Watzele,M., Nemetz,C., Obermeier,W., Seidel,A. and Buchberger,B. This work was supported by Roche Diagnostics Germany. (2003) High-throughput expression PCR used to systematically investigate regulation of translation initiation in an Escherichia coli cell- free expression system. In Swartz,J.R. (ed.), Cell-free Protein REFERENCES Expression. Springer, Berlin, pp. 25±30. 26. Watzele,M., Nemetz,C., Metzler,T., Fernholz,E., Hoffmann,T. and 1. Buchanan,M.V., Larimer,F.W., Wiley,H.S., Kennel,S.J., Squier,T.J., Mutter,W. (2001) Rapid Translation System RTS 100Ða high Ramsey,J.M., Rodland,K.D., Hurst,G.B., Smith,R.D., Xu,Y. et al. (2002) throughput cell-free protein expression. Biochemica, 3, 27±28. Genomes to Life `Center for Molecular and Cellular Systems': a research 27. Resto,E., Iida,A., Van Cleve,M.D. and Hecht,S.M. (1992) Ampli®cation program for identi®cation and characterization of protein complexes. of protein expression in a cell free system. Nucleic Acids Res., 20, 5979± Omics, 6, 287±303. 2. Swartz,J.R. (2001) Advances in Escherichia coli production of 28. Laemmli,U.K. (1970) Cleavage of structural proteins during the therapeutic proteins. Curr. Opin. Biotechnol., 12, 195±201. assembly of the head of bacteriophage T4. Nature, 227, 680±685. 3. Jonasson,P., Liljeqvist,S., Nygren,P.A. and Stahl,S. (2002) Genetic 29. Sambrook,J., Russel,D.W. and Sambrook,J. (2001) Molecular Cloning: A design for facilitated production and recovery of recombinant proteins in Laboratory Manual (3rd edn). Cold Spring Harbor Laboratory, Cold Escherichia coli. Biotechnol. Appl. Biochem., 35, 91±105. Spring Harbor, NY. 4. Balbas,P. (2001) Understanding the art of producing protein and 30. Donis-Keller,H. (1979) Site speci®c enzymatic cleavage of RNA. nonprotein molecules in Escherichia coli. Mol. Biotechnol., 19, 251±267. Nucleic Acids Res., 7, 179±192. 5. Makrides,S.C. (1996) Strategies for achieving high-level expression of 31. Gunzl,A., Pal®,Z. and Bindereif,A. (2002) Analysis of RNA-protein genes in Escherichia coli. Microbiol. Rev., 60, 512±538. complexes by oligonucleotide-targeted RNase H digestion. Methods, 26, 6. Yusupova,G.Z., Yusupov,M.M., Cate,J.H. and Noller,H.F. (2001) The 162±169. path of messenger RNA through the ribosome. Cell, 106, 233±241. 32. Phillips,C., Kyriakopoulou,C.B. and Virtanen,A. (1999) Identi®cation of 7. Ramakrishnan,V. (2002) Ribosome structure and the mechanism of a stem±loop structure important for polyadenylation at the murine IgM translation. Cell, 108, 557±572. secretory poly(A) site. Nucleic Acids Res., 27, 429±438. 8. Culver,G.M. (2001) Meanderings of the mRNA through the ribosome. 33. Zuker,M. (2003) Mfold web server for nucleic acid folding and Structure, 9, 751±758. hybridization prediction. Nucleic Acids Res., 31, 3406±3415. 9. Hartz,D., McPheeters,D.S. and Gold,L. (1991) In¯uence of mRNA 34. Landick,R. and Yanofsky,C. (1987) Isolation and structural analysis of determinants on translation initiation in Escherichia coli. J. Mol. Biol., the Escherichia coli trp leader paused transcription complex. J. Mol. 218, 83±97. Biol., 196, 363±377. 10. Gold,L. (1988) Posttranscriptional regulatory mechanisms in Escherichia 35. Landick,R., Yanofsky,C., Choo,K. and Phung,L. (1990) Replacement of coli. Annu. Rev. Biochem., 57, 199±233. the Escherichia coli trp operon attenuation control codons alters operon 11. Gualerzi,C., Risuleo,G. and Pon,C.L. (1977) Initial rate kinetic analysis expression. J. Mol. Biol., 216, 25±37. of the mechanism of initiation complex formation and the role of 36. Gutman,G.A. and Hat®eld,G.W. (1989) Nonrandom utilization of codon initiation factor IF-3. Biochemistry, 16, 1684±1689. pairs in Escherichia coli. Proc. Natl Acad. Sci. USA, 86, 3699±3703. 12. Jacques,N. and Dreyfus,M. (1990) Translation initiation in Escherichia 37. Lima,W.F., Monia,B.P., Ecker,D.J. and Freier,S.M. (1992) Implication of coli: old and new questions. Mol. Microbiol., 4, 1063±1067. RNA structure on antisense oligonucleotide hybridization kinetics. 13. McCarthy,J.E. and Brimacombe,R. (1994) Prokaryotic translation: the Biochemistry, 31, 12055±12061. interactive pathway leading to initiation. Trends Genet., 10, 402±407. 38. Young,S. and Wagner,R.W. (1991) Hybridization and dissociation rates 14. Tessier,L.H., Sondermeyer,P., Faure,T., Dreyer,D., Benavente,A., of phosphodiester or modi®ed oligodeoxynucleotides with RNA at Villeval,D., Courtney,M. and Lecocq,J.P. (1984) The in¯uence of mRNA near-physiological conditions. Nucleic Acids Res., 19, 2463±2470. primary and secondary structure on human IFN-gamma gene expression 39. Nellen,W. and Lichtenstein,C. (1993) What makes an mRNA anti-sense- in E.coli. Nucleic Acids Res., 12, 7663±7675. itive? Trends Biochem. Sci., 18, 419±423. 15. deSmit,M.H. and van Duin,J. (1994) Translational initiation on 40. Carrier,T.A. and Keasling,J.D. (1999) Library of synthetic 5¢ secondary structured messengers. Another role for the Shine±Dalgarno interaction. structures to manipulate mRNA stability in Escherichia coli. Biotechnol. J. Mol. Biol., 235, 173±184. Prog., 15, 58±64. 16. vanMeerten,D., Girard,G. and van Duin,J. (2001) Translational control 41. Ivanovski,G., Gubensek,F. and Pungercar,J. (2002) mRNA secondary by delayed RNA folding: identi®cation of the kinetic trap. RNA, 7, structure can greatly affect production of recombinant phospholipase 483±494. A(2) toxins in bacteria. Toxicon, 40, 543±549. 17. Hui,A. and de Boer,H.A. (1987) Specialized ribosome system: 42. Iost,I., Guillerez,J. and Dreyfus,M. (1992) Bacteriophage T7 RNA preferential translation of a single mRNA species by a subpopulation of mutated ribosomes in Escherichia coli. Proc. Natl Acad. Sci. USA, 84, polymerase travels far ahead of ribosomes in vivo. J. Bacteriol., 174, 4762±4766. 619±622. e78 Nucleic Acids Research, 2004, Vol. 32, No. 9 PAGE 14 OF 14 43. Draper,D.E. (1996) Strategies for RNA folding. Trends Biochem. Sci., 51. McCormick,J.R., Zengel,J.M. and Lindahl,L. (1994) Correlation of 21, 145±149. translation ef®ciency with the decay of lacZ mRNA in Escherichia coli. 44. Misra,V.K. and Draper,D.E. (2002) The linkage between magnesium J. Mol. Biol., 239, 608±622. binding and RNA folding. J. Mol. Biol., 317, 507±521. 52. Emory,S.A., Bouvet,P. and Belasco,J.G. (1992) A 5¢-terminal stem±loop 45. Tinoco,I.,Jr and Bustamante,C. (1999) How RNA folds. J. Mol. Biol., structure can stabilize mRNA in Escherichia coli. Genes Dev., 6, 135± 293, 271±281. 46. Brion,P. and Westhof,E. (1997) Hierarchy and dynamics of RNA folding. 53. Wong,H.C. and Chang,S. (1986) Identi®cation of a positive Annu. Rev. Biophys Biomol. Struct., 26, 113±137. retroregulator that stabilizes mRNAs in bacteria. Proc. Natl Acad. Sci. 47. Carpousis,A.J. (2002) The Escherichia coli RNA degradosome: USA, 83, 3233±3237. structure, function and relationship in other ribonucleolytic multienzyme 54. Bouvet,P. and Belasco,J.G. (1992) Control of RNase E-mediated RNA complexes. Biochem. Soc. Trans., 30, 150±155. degradation by 5¢-terminal base pairing in E. coli. Nature, 360, 488±491. 48. Grunberg-Manago,M. (1999) Messenger RNA stability and its role in 55. Chen,L.H., Emory,S.A., Bricker,A.L., Bouvet,P. and Belasco,J.G. (1991) control of gene expression in bacteria and phages. Annu. Rev. Genet., 33, Structure and function of a bacterial mRNA stabilizer: analysis of the 5¢ 193±227. untranslated region of ompA mRNA. J. Bacteriol., 173, 4578±4586. 49. Ross,J. (1995) mRNA stability in mammalian cells. Microbiol. Rev., 59, 56. Arnold,T.E., Yu,J. and Belasco,J.G. (1998) mRNA stabilization by the 423±450. ompA 5¢ untranslated region: two protective elements hinder distinct 50. Chen,C.Y. and Belasco,J.G. (1990) Degradation of pufLMX mRNA in pathways for mRNA degradation. RNA, 4, 319±330. Rhodobacter capsulatus is initiated by nonrandom endonucleolytic cleavage. J. Bacteriol., 172, 4578±4586.

Journal

Nucleic Acids Research – Oxford University Press

Published: Jun 1, 2004

There are no references for this article.