Translation initiation region sequence preferences in Escherichia coli

Vladimir Vimberg 1 Age Tats 0 Maido Remm 0 Tanel Tenson 1 0 Department of Bioinformatics, Institute of Molecular and Cell Biology, University of Tartu , Riia 23, Tartu 51010 , Estonia 1 Institute of Technology, University of Tartu , Nooruse 1, Tartu 50411 , Estonia Background: The mRNA translation initiation region (TIR) comprises the initiator codon, ShineDalgarno (SD) sequence and translational enhancers. Probably the most abundant class of enhancers contains A/U-rich sequences. We have tested the influence of SD sequence length and the presence of enhancers on the efficiency of translation initiation. Results: We found that during bacterial growth at 37C, a six-nucleotide SD (AGGAGG) is more efficient than shorter or longer sequences. The A/U-rich enhancer contributes strongly to the efficiency of initiation, having the greatest stimulatory effect in the exponential growth phase of the bacteria. The SD sequences and the A/U-rich enhancer stimulate translation co-operatively: strong SDs are stimulated by the enhancer much more than weak SDs. The bacterial growth rate does not have a major influence on the TIR selection pattern. On the other hand, temperature affects the TIR preference pattern: shorter SD sequences are preferred at lower growth temperatures. We also performed an in silico analysis of the TIRs in all E. coli mRNAs. The base pairing potential of the SD sequences does not correlate with the codon adaptation index, which is used as an estimate of gene expression level. Conclusion: In E. coli the SD selection preferences are influenced by the growth temperature and not influenced by the growth rate. The A/U rich enhancers stimulate translation considerably by acting co-operatively with the SD sequences. - Background The efficiency of initiation is the most important determinant of translation efficiency [1]. In bacteria, the 30S ribosomal subunit, assisted by initiation factors (IF) 1, 2 and 3 and fMet-tRNAfMet, recognizes the translation initiation region (TIR) of the mRNA. This event is followed by binding of the 50S ribosomal subunit and release of the initiation factors [1]. The rate-limiting step in this process is binding of the 30S subunit to the TIR [2]. There are two alternative pathways for mRNA recognition by 30S subunits. In the first, the 30S subunit complexed with IF1 and IF3 binds to the mRNA, followed by IF2 and GTP-dependent binding of fMet-tRNAfMet [2]. In the second, the IF2:GTP:fMet-tRNAfMet complex binds to the 30S subunit followed by mRNA recognition [3]. The relative frequencies with which these pathways are used in bacterial cells are currently not clear. The following sequence elements of the TIR contribute to its efficiency: (a) the initiation codon, which is most commonly AUG but sometimes GUG and very rarely UUG, AUU or CUG [4-7]; (b) the Shine-Dalgarno (SD) sequence [8,9]; (c) regions upstream of the SD sequence and downstream of the initiation codon, which are often described as enhancers of translation [10-15]. In addition, the spacing between these sequence elements is often critical. For example, the distance between the SD sequence and the initiation triplet has a marked effect on the efficiency of translation [16]. The SD sequence base-pairs directly with the anti-ShineDalgarno (aSD) sequence on the 3' end of the 16S rRNA [8]. The maximum known length of the SD:aSD duplex is 12 or 13 nucleotides [17]; in most E. coli genes the SD sequence is shorter. Free energy calculations for all possible duplexes between the 16S rRNA 3' end and a region 21 nucleotides upstream from the start codon in 1159 E. coli genes show that the average number of paired mRNA:rRNA nucleotides is 6.3 [18]. A similar calculation has been made for the ribosomal protein genes and indicates that the average SD length is 4.4 nucleotides [19]. Studies have shown that mRNAs lacking an SD sequence cannot bind the 30S subunit efficiently without the contribution of translational enhancers, additional sequences in the TIR able to increase the efficiency of translation [20]. Also, SD sequences longer than six nucleotides are not very efficient, probably because more time is needed for clearance of the TIR [19,21]. On the other hand, other studies have questioned the importance of the SD for the initiation of translation: Lee et al. [22] report that translation efficiency correlates very poorly with the strength of the SD:aSD interaction. Unfortunately, no systematic study to date has established the correlation between the SD:aSD interaction strength and the efficiency of translation. Recently, it has been shown that before the SD:aSD interaction occurs, the 30S ribosomal subunit can bind to a standby site in the vicinity of the SD [23,24]. Binding to this standby site might increase the local concentration of 30S subunits at the TIR. The ribosome may remain attached to the standby site until the SD sequence is in a conformation appropriate for binding the aSD. Through this mechanism, the standby site could stimulate translation of mRNAs in which the SD can be trapped by secondary structures. One possible way in which a standby site in mRNA could be created is by binding to S1, the largest protein component of the small ribosomal subunit. S1 consists of two major domains with a freely rotatable region between them [25]. One domain is attached to the 30S; the second is exposed on the surface of the small subunit, scanning the space around the ribosome and searching for A/U-rich sequences [14,19,26] that are recognized with the help of four RNA-binding motifs [27]. It has been shown that S1 can destabilize RNA secondary structures [28]. Cross-linking studies have shown that the nucleic acid-binding domain of S1 is aligned with a region of the mRNA upstream of the SD, suggesting that S1 may interact with 5' parts of the TIR [29,30]. Consistent with this observation, A/U-rich sequences in front of the SD or downstream of the initiator codon enhance protein synthesis [15,19]. To date, nine sequences have been shown experimentally to act as translational enhancers. They are all A/U-rich and contain very few Gs [19]. Disruption of the E. coli gene coding for S1 has been reported to be lethal [31]. A decreased level of S1 protein in the cell leads to a rapid decrease in total protein synthesis [32]. Thus it can be speculated that the SD sequence alone cannot mediate efficient initiation of translation but has to be complemented with an enhancer sequence. Unfortunately, information about the effects of combining the enhancers with different SD sequences is very limited [19]. In the current study we have constructed a set of SD sequences, ranging between 1 and 8 nucleotides, and tested their efficiency with a reporter gene. This allowed the most efficient SD sequences in E. coli to be defined. In addition, we have combined all the SD sequences with translational enhancers and determined the effects on reporter gene expression. We have tested all the TIR vari (...truncated)