Figure 1 - uploaded by Mao Y.H.
Content may be subject to copyright.
An example of HSRs in Escherichia coli and Escherichia fergusonii . A) Z-score is the normalized MFE. The threshold used to define HSR is marked by blue dashes. B) shows the secondary structures of HSRs. Although the HSRs between the two species are conserved (see Materials and Methods for details), the secondary structures are non-conserved. doi:10.1371/journal.pone.0073299.g001
Source publication
Message RNA (mRNA) carries a large number of local secondary structures, with structural stability to participate in the regulations of gene expression. A worthy question is how the local structural stability is maintained under the constraint that multiple selective pressures are imposed on mRNA local regions. Here, we performed the first genome-w...
Context in source publication
Context 1
... molecules tend to adopt a folded conformation through the formation of Watson-Crick base pairing between complemen- tary nucleotides. The resulting so-called RNA secondary structure emerges to be a key player in the regulations of gene expression [1–4]. By surveying secondary structures in various genomes, previous studies have revealed that a large number of genomes are being transcribed to produce non-coding RNAs that generally contain a conserved secondary structure [5–9]. The structural conformation of the molecule is often necessary for its functions. Precursor microRNAs (pre-miRNAs) are among the largest examples that illustrate the functions of secondary structure in non-coding RNAs. The pre-miRNA contains a , 70-bp hairpin, which is recognized by the Dicer protein and then the loop region is removed to leave a dsRNA [10,11]. The secondary structure in pre-miRNA is conserved during evolution [12–14], suggesting an important role of structural conformation in miRNA maturation. Interestingly, in recent years, a considerable number of protein- coding RNAs have been reported to contain local secondary structures [15–18]. Some translational processes, including translation initiation [19–21], co-translational folding of protein [22,23], are sensitive to the variation of local structural stability. Moreover, a strong association between structural stability and protein abundance was observed in yeast [24]. These results suggest an important role of mRNA structural stability, which might be different from the roles of conserved secondary structures reported by previous studies. Besides its functions, numerous studies have focused on the evolution of RNA secondary structure [14,25–29] and revealed several mechanisms to maintain the secondary structure, including lower substitution rate [30] and compensatory mutations [27]. Mutations that occur in the primary sequence might lead to a disruption of the paired regions, thus changing the structural conformation or structural stability of the molecule and impairing its original function. Various studies focused on the selective constraints in folded RNAs that are mostly located in non-coding regions [14,30,31]. They found a lower substitution rate in paired regions in comparison with that in unpaired regions [30]. Moreover, previous studies aimed at attributing the variation of GC content to the selection for high structural stability of RNA [32,33]. The association has been observed in several types of noncoding RNA, such as miRNA. In miRNA, GC content is positively correlated with the organism’s physiological temperature [33], suggesting a possible association between the base-pairing strength of miRNA-targets and the temperature of an organism. Unlike non-coding RNAs, multiple selective constraints, including structural stability [34,35] and translation efficiency [36,37], operate on mRNAs. Both two constraints influence the pattern of synonymous mutation. If there is selection on local protein- coding region for high structural stability, codons in such region might be under conflicting selective pressures: the codons promoting RNA folding with high structural stability might be translationally non-optimal. In this case, the locations of synonymous substitutions might be non-random with respect to the translational efficiency and structural stability. Knowledge of this conflict can further our understanding of constraints imposed on protein-coding RNAs. The initial studies of secondary structure, in mammals as well as yeast [38,39], considered the thermodynamic stability of mRNA mediated by the changes in secondary structure, and revealed that C preference at the four-fold degenerate sites might be partly driven by the selection for RNA stability [38]. However, these studies did not focus on local protein-coding regions that form local secondary structures, exhibiting high structural stability. Numerous lines of evidence indicate that mRNA folding windows are small [40,41]. Therefore, it is reasonable to investigate the substitution patterns of structural regions based on local folding instead of global folding. Moreover, the analysis of selective constraints on local structural stability of protein-coding region has not been performed in the genome wide scale. Therefore, the aim of this study is to analyze the natural selection related to the local structural stability from a genome wide perspective. In recent years, there has been a sharp growth in evidence showing widespread secondary structures in protein-coding region. In our previous study, we found that most of these structures exhibit high structural stability, while their structural conformations are non-conserved across different species [16]. We therefore identified the regions with high structural stability in Escherichia coli (HSR, high structural stability regions) using a loose threshold (Figure 1, Table S1-S2) [16], and revealed that number variation of HSR is correlated with gene functions, probably involving the regulations on the rhythm of translation elongation. However, the evolutionary pattern of HSR is still undetermined in that study. In particular, it remains unclear that how the structural stability of HSR is maintained under the constraint that multiple selective pressures are imposed on mRNA local regions. The selective pressure on HSR might be relaxed compared with that on structural RNA (e.g. 5 s rRNA), because there is a higher probability that a second mutation restores the structural stability disrupted by the first mutation. It is also of interesting to investigate the difference in the patterns of compensatory mutations between HSR and structural RNA. Therefore, in current study, we focus on the natural selections on HSRs, and aim at addressing the following questions that might advance our knowledge of selective pressures on mRNA: 1) Does selection for structural stability of HSR favor synonymous codons with high G/ C? 2) How does HSR influence the local substitution rate of mRNA? 3) Since it is the structural stability rather than the structural conformation that is conserved, is the pattern of compensatory mutations in HSR different from that in structural RNA? Protein coding sequences of Escherichia coli K12 MG1655, Escherichia fergusonii ATCC and Salmonella enterica subsp. enterica serovar Typhi CT18 were downloaded from the National Center for Biotechnology Information FTP server ( genomes/). Sequences with length , 200 nucleotides (nt) were excluded. In total, we obtained 4152, 4126 and 4246 ...
Citations
... The minimum free energy (MFE) and GC content were used to evaluate the stability of the mRNA structure in vitro based on its sequence composition [39,[41][42][43]. MFE was significantly negatively correlated with GC, with a Pearson correlation of −0.45 (Figure 2(b)). ...
The structure of mRNA in vivo is unwound to some extent in response to multiple factors involved in the translation process, resulting in significant differences from the structure of the same mRNA in vitro. In this study, we have proposed a novel application of deep neural networks, named DeepDRU, to predict the degree of mRNA structure unwinding in vivo by fitting five quantifiable features that may affect mRNA folding: ribosome density (RD), minimum folding free energy (MFE), GC content, translation initiation ribosome density (INI) and mRNA structure position (POS). mRNA structures with adjustment of the simulated structural features were designed and then fed into the trained DeepDRU model. We found unique effect regions of these five features on mRNA structure in vivo. Strikingly, INI is the most critical factor affecting the structure of mRNA in vivo, and structural sequence features, including MFE and GC content, have relatively smaller effects. DeepDRU provides a new paradigm for predicting the unwinding capability of mRNA structure in vivo. This improved knowledge about the mechanisms of factors influencing the structural capability of mRNA to unwind will facilitate the design and functional analysis of mRNA structure in vivo.
In vivo mRNA structure was influenced by versatile factors in the translation process resulting in a significant difference of mRNA structure in vitro. In this study, we elaborated a novel deep neural networks (DNN) model for predicting stable or unstable structural states of mRNA in vivo by fitting six quantifiable features of mRNA structure: ribosome density, minimum free energy, GC content, mRNA abundance, ribosomal initial density and structural relative position. By mutations simulation and well-trained, high-precision models prediction, we discovered the unique effects of these six mRNA structural features on the structural stability of mRNA in vivo. It was worthwhile to point out that we have found a double-sided effect of ribosomal density on the structural stability of mRNA, and it may be speculated that, more stable mRNA structure leaded to higher ribosome density in the region of low ribosomal density and higher ribosomal density resulted in more unstable mRNA structure in the region of high ribosome density.Additionally, the trend and extent of these six features affecting mRNA structural stability in vivo could also be accurately predicted by the DNN model.This is the first time to use DNN method to decipher the difference of mRNA structure in vivo and in vitro, and a good prediction results have been acquired, revealing a "personally" impact of different mRNA structural features on its stability quantitatively. Thus, it was speculated that this frontier analysis method, DNN model, will have great potential in solving the problem of complex regularity in the translation process.
