Data mining of imperfect double-stranded RNA in 3' untranslated regions of eukaryotic mRNAs.

Recent developments in the study of RNA silencing indicate that double-stranded RNA (dsRNA) can be used in eukaryotes to block expression of a corresponding cellular gene. There is also a large class of small non-coding RNAs having potential to form a distinct, stable stem-loop in numbers of eukaryotic genomes. We had reported that a large imperfect dsRNA structure with hundreds of base-pairs (bp) in the 3' untranslated region (3' UTR) of cytotoxic ribonuclease was correlated with the translation suppression. In this study, we search for such dsRNAs in a 3' UTR database. The occurrence rate of large dsRNA in 3' UTRs ranges from 0.01% in plant to 0.30% in vertebrate mRNAs. However, small imperfect dsRNAs of ~ 30 bp are much more prevalent than large ones. The small dsRNAs are statistically very significant and uniquely well-ordered. Most of them have the conserved structural features of pre-miRNAs. Our data mining of the dsRNAs in the 3' UTR database can be used to explore RNA-based regulation of gene expression.

Short-strong hydrogen bonds and a low barrier transition state for the proton transfer reaction in RNase A catalysis: a quantum chemical study.

There is growing evidence that some enzymes catalyze reactions through the formation of short-strong hydrogen bonds as first suggested by Gerlt and Gassman. Support comes from several experimental and quantum chemical studies that include correlation energies on model systems. In the present study, the process of proton transfer between hydroxyl and imidazole groups, a model of the crucial step in the hydrolysis of RNA by the enzymes of the RNase A family, is investigated at the quantum mechanical level of density functional theory and perturbation theory at the MP2 level. The model focuses on the nature of the formation of a complex between the important residues of the protein and the hydroxyl group of the substrate. We have also investigated different configurations of the ground state that are important in the proton transfer reaction. The nature of bonding between the catalytic unit of the enzyme and the substrate in the model is investigated by Bader's atoms in molecule theory. The contributions of solvation and vibrational energies corresponding to the reactant, the transition state and the product configurations are also evaluated. Furthermore, the effect of protein environment is investigated by considering the catalytic unit surrounded by complete proteins--RNase A and Angiogenin. The results, in general, indicate the formation of a short-strong hydrogen bond and the formation of a low barrier transition state for the proton transfer model of the enzyme.

Anatomy of protein structures: visualizing how a one-dimensional protein chain folds into a three-dimensional shape.

Here, we depict the anatomy of protein structures in terms of the protein folding process. Via an iterative, top-down dissecting procedure, tertiary structures are spliced down to reveal their anatomy: first, to produce domains (defined by visual three-dimensional inspection criteria); then, hydrophobic folding units (HFU); and, at the end of a multilevel process, a set of building blocks. The resulting anatomy tree organization not only clearly depicts the organization of a one-dimensional polypeptide chain in three-dimensional space but also straightforwardly describes the most likely folding pathway(s). Comparison of the tree with the formation of the hydrophobic folding units through combinatorial assembly of the building blocks illustrates how the chain folds in a sequential or a complex folding pathway. Further, the tree points to the kinetics of the folding, whether the chain is a fast or a slow folder, and the probability of misfolding. Our ability to successfully dissect the protein into an anatomy tree illustrates that protein folding is a hierarchical process and further validates a building blocks protein folding model.

A gender-specific mRNA encoding a cytotoxic ribonuclease contains a 3' UTR of unusual length and structure.

A cDNA (2855 nt) encoding a putative cytotoxic ribonuclease (rapLR1) related to the antitumor protein onconase was cloned from a library derived from the liver of gravid female amphibian Rana pipiens. The cDNA was mainly comprised (83%) of 3' untranslated region (UTR). Secondary structure analysis predicted two unusual folding regions (UFRs) in the RNA 3' UTR. Two of these regions (711-1442 and 1877-2130 nt) contained remarkable, stalk-like, stem-loop structures greater than 38 and 12 standard deviations more stable than by chance, respectively. Secondary structure modeling demonstrated similar structures in the 3' UTRs of other species at low frequencies (0.01-0.3%). The size of the rapLR1 cDNA corresponded to the major hybridizing RNA cross-reactive with a genomic clone encoding onconase (3.6 kb). The transcript was found only in liver mRNA from female frogs. In contrast, immunoreactive onconase protein was detected only in oocytes. Deletion of the 3' UTR facilitated the in vitro translation of the rapLR1 cDNA. Taken together these results suggest that these unusual UFRs may affect mRNA metabolism and/or translation.

Prediction of common secondary structures of RNAs: a genetic algorithm approach.

In this study we apply a genetic algorithm to a set of RNA sequences to find common RNA secondary structures. Our method is a three-step procedure. At the first stage of the procedure for each sequence, a genetic algorithm is used to optimize the structures in a population to a certain degree of stability. In this step, the free energy of a structure is the fitness criterion for the algorithm. Next, for each structure, we define a measure of structural conservation with respect to those in other sequences. We use this measure in a genetic algorithm to improve the structural similarity among sequences for the structures in the population of a sequence. Finally, we select those structures satisfying certain conditions of structural stability and similarity as predicted common structures for a set of RNA sequences. We have obtained satisfactory results from a set of tRNA, 5S rRNA, rev response elements (RRE) of HIV-1 and RRE of HIV-2/SIV, respectively.

Distinguishing between sequential and nonsequentially folded proteins: implications for folding and misfolding.

We describe here an algorithm for distinguishing sequential from nonsequentially folding proteins. Several experiments have recently suggested that most of the proteins that are synthesized in the eukaryotic cell may fold sequentially. This proposed folding mechanism in vivo is particularly advantageous to the organism. In the absence of chaperones, the probability that a sequentially folding protein will misfold is reduced significantly. The problem we address here is devising a procedure that would differentiate between the two types of folding patterns. Footprints of sequential folding may be found in structures where consecutive fragments of the chain interact with each other. In such cases, the folding complexity may be viewed as being lower. On the other hand, higher folding complexity suggests that at least a portion of the polypeptide backbone folds back upon itself to form three-dimensional (3D) interactions with noncontiguous portion(s) of the chain. Hence, we look at the mechanism of folding of the molecule via analysis of its complexity, that is, through the 3D interactions formed by contiguous segments on the polypeptide chain. To computationally splice the structure into consecutively interacting fragments, we either cut it into compact hydrophobic folding units or into a set of hypothetical, transient, highly populated, contiguous fragments ("building blocks" of the structure). In sequential folding, successive building blocks interact with each other from the amino to the carboxy terminus of the polypeptide chain. Consequently, the results of the parsing differentiate between sequentially vs. nonsequentially folded chains. The automated assessment of the folding complexity provides insight into both the likelihood of misfolding and the kinetic folding rate of the given protein. In terms of the funnel free energy landscape theory, a protein that truly follows the mechanism of sequential folding, in principle, encounters smoother free energy barriers. A simple sequentially folded protein should, therefore, be less error prone and fold faster than a protein with a complex folding pattern.

Ion-RNA interactions in the RNA pseudoknot of a ribosomal frameshifting site: molecular modeling studies.

The three-dimensional (3-D) structure of a RNA pseudoknot that causes the efficient ribosomal frameshifting in the gag-pro region of mouse mammary tumor virus (MMTV) has been determined recently by nuclear magnetic resonance (NMR) studies. But since the structure refinement in the studies did not use metal ions and waters, it is not clear how metal ions participate in the stabilization of the pseudoknot, and what kind of ion-RNA interactions dominate in the tertiary contacts for the RNA pseudoknotting. Based on the reported structure data of the pseudoknot VPK of MMTV, we gradually refined the structure by restrained molecular dynamics (MD) using NMR distance restraints. Restrained MD simulation of the RNA pseudoknot was performed with sodium ions and water molecules. Our results are in good agreement with known NMR data and delineate the importance of the metal ion coordination in the stability of the pseudoknot. In the non-coaxially stacking pseudoknot, stem 1 (S1), stem 2 (S2), and the intervening A14 involves unconventional stacking of base pairs coordinated by Na+ and/or bridging water molecules. A6 and G7 of loop L1 make a perfect base stacking in the major groove and are further stabilized by coordinated Na+ ions and water molecules. The first 4-nucleotide (nt) ACUC of loop L2 form a sharp turn and the following 4-nt AAAA cross the minor groove of S1 and are steadied by interactions with the nucleotides of S , bridging water molecules and coordinated Na+ ions. Our studies suggest that the metal ion plays a crucial role in the RNA pseudoknotting of VPK. In the stacking interior of S1 and S2, the Na+ ion is positioned in the major groove and interacts directly with the carbonyl group O6 of G28 and carbonyl group O4 of U13 in the wobble base pair U13:G28. The ion-RNA interactions in MMTV VPK not only stabilize the RNA pseudoknot but also modify the electrostatic properties of the nucleotides at the critical parts of the pseudoknot VPK.

Evolution of a common structural core in the internal ribosome entry sites of picornavirus.

The translational control involving internal ribosome binding occurs in poliovirus (PV), human rhinoviruses (HRV), encephalomyocarditis virus (EMCV), foot-and-mouth disease virus (FMDV), and hepatitis A virus (HAV). Internal ribosome binding utilizes cis-acting genetic elements of approximately 450 nucleotides (nt) termed the internal ribosome entry sites (IRES) found in these picornaviral 5'-untranslated region (5'UTR). Although these IRES elements are quite different in their primary sequence, a similar folding structure with a conserved 3' structural core exists in the IRES. Phylogenetic analysis and RNA folding of the 5' UTR of picornaviruses, including PV types 1-3, coxsackievirus types A and B, swine vesicular disease virus, echoviruses, enteroviruses (human and bovine), HRV, HAV, EMCV, mengovirus, Theiler's murine encephalomyelitis viruses, FMDV, and equine rhinoviruses, indicates that the predicted conserved structural core is indeed a general structural feature for all members of the picornavirus family. The evolution of a common structural core likely occurred by the gradual addition or deletion of structural domains and elements to preserve a similar tertiary structure that facilitates the utilization of the IRES in specific host-cell environments.

Phylogenetic evidence for the improved RNA higher-order structure in internal ribosome entry sequences of HCV and pestiviruses.

The strong requirement for a small segment of the 5'-proximal coding sequence of hepatitis C virus (HCV) is one of the most remarkable features in the internal initiation of HCV mRNA translation. Phylogenetic analysis and RNA folding indicate a common RNA structure of the 5' untranslated region (UTR) of HCV and the animal pestiviruses, including HCV types 1-11, bovine viral diarrhea (BVDV), border disease virus (BDV) and hog cholera (HoCV). Although the common RNA structure shares similar features to that proposed for the internal ribosome entry sequence (IRES) of picornavirus, phylogenetic evidence suggests four new tertiary interactions between conserved terminal hairpin loops and between the terminal hairpin loop of F2b and the short coding sequence for HCV and pestiviruses. We suggest that the higher-order structures of IRES cis-acting elements for HCV and animal pestivirus are composed of stem-loop structures B-C, domains E-H, stem-loop structure J and four additional tertiary interactions. The common structure of IRES elements for these viruses forms a compact structure by these tertiary interactions and stem stacking. The active structural core is centered in the junction domain of E-H that is also conserved in all members of picornaviruses. Our model suggests that the requirement for a small segment of the 5' coding sequence is to form the distinct tertiary structure that facilitates the cis-acting function of the HCV IRES in the internal initiation of the translational control.

A common RNA structural motif involved in the internal initiation of translation of cellular mRNAs.

The 5'-non-translated regions (5'NTR) of human immunoglobulin heavy chain binding protein (BiP), Antennapedia (Antp) ofDrosophilaand human fibroblast growth factor 2 (FGF-2) mRNAs are reported to mediate translation initiation by an internal ribosome binding mechanism. In this study, we investigate predicted features of the higher order structures folded in these 5'NTR sequences. Statistical analyses of RNA folding detected a 92 nt unusual folding region (UFR) from 129 to 220, close to the initiator AUG in the BiP mRNA. Details of the structural analyses show that the UFR forms a Y-type stem-loop structure with an additional stem-loop in the 3'-end resembling the common structure core found in the internal ribosome entry site (IRES) elements of picornavirus. The Y-type structural motif is also conserved among a number of divergent BiP mRNAs. We also find two RNA elements in the 5'-leader sequence of human FGF-2. The first RNA element (96 nt) is 2 nt upstream of the first CUG start codon located in the reported IRES element of human FGF-2. The second (107 nt) is immediately upstream of the authentic initiator AUG of the main open reading frame. Intriguingly, the folded RNA structural motif in the two RNA elements is conserved in other members of FGF family and shares the same structural features as that found in the 5'NTR of divergent BiP mRNAs. We suggest that the common RNA structural motif conserved in the diverse BiP and FGF-2 mRNAs has a general function in the internal ribosome binding mechanism of cellular mRNAs.

A common structural core in the internal ribosome entry sites of picornavirus, hepatitis C virus, and pestivirus.

Cap-independent translations of viral RNAs of enteroviruses and rhinoviruses, cardioviruses and aphthoviruses, hepatitis A and C viruses (HAV and HCV), and pestivirus are initiated by the direct binding of 40S ribosomal subunits to a cis-acting genetic element termed the internal ribosome entry site (IRES) or ribosome landing pad (RLP) in the 5' noncoding region (5'NCR). RNA higher ordered structure models for these IRES elements were derived by a combined approach using thermodynamic RNA folding, Monte Carlo simulation, and phylogenetic comparative analysis. The structural differences among the three groups of picornaviruses arise not only from point mutations, but also from the addition or deletion of structural domains. However, a common core can be identified in the proposed structural models of these IRES elements from enteroviruses and rhinoviruses, cardioviruses and aphthoviruses, and HAV. The common structural core identified within the picornavirus IRES is also conserved in the 5'NCR of the divergent viruses, HCV, and pestiviruses. Furthermore, the proposed structural motif shares a structural feature similar to that observed in the catalytic core of the group 1 intron. The conserved structural motif from these divergent sequences that looks like the common core region of group 1 introns is probably a crucial element involved in the IRES-dependent translation.

Unusual folding regions and ribosome landing pad within hepatitis C virus and pestivirus RNAs.

A statistically significant folding region is identified in the 5' untranslated region (5'-UTR) of hepatitis C virus (HCV), bovine viral diarrhea virus and hog cholera virus. This unusual folding region (UFR) detected in HCV encompasses 199 nucleotides (nt) and coincides with the reported internal ribosome entry site or ribosome landing pad (RLP), as determined by the 5' and 3' deletions [Tsukiyama-Kohara et al., J. Virol. 66 (1992) 1476-1483]. The RNA structure predicted in the UFR of HCV consists of a large stem-loop and a pseudoknot. The proposed structural model is consistent with RNase sensitivity studies [Brown et al., Nucleic Acids Res. 20 (1992) 5041-5045]. Moreover, the structure is highly conserved among these divergent HCV and pestivirus RNAs. The covariation of paired bases in the helical regions offers support for the proposed structural models. The pseudoknot predicted in these UFR shares a similar structural feature to those proposed in the RLP of cardioviruses, aphthoviruses and hepatitis A virus. Based on the common structural motif, a putative base-pairing model between HCV RNA and 18S rRNA, as well as pestiviral RNAs and 18S rRNA are suggested. Intriguingly, the proposed base-pairing models in this study are comparable to those proposed in picornaviruses in terms of their folded shape and location of the predicted complementary sequences between viral RNAs and 18S rRNA. Taken together, we suggest that the common base-pairing model between the UFR detected in the 5'-UTR of pestivirus and HCV and 18S rRNA have a general function in the internal initiation of cap-independent translation.

A method for predicting common structures of homologous RNAs.

We have developed a procedure, composed of a set of computer programs, for predicting common RNA structures of homologous sequences. Given a set of homologous RNAs, these programs perform a multiple sequence alignment, generate a list of possible helical stems that are thermodynamically favored in RNA folding from a selected individual sequence, establish a conserved stem list by inspecting the equivalent base pairings and/or conserved helical stems from the derived alignment of homologous RNAs, and build common RNA secondary structures with the maximum scores (i.e., compensatory base changes and number of base pairs, etc.). The approach is a combination of phylogenetic and thermodynamic methods and has been applied to the prediction of common folding structures of the 5' untranslated regions in a number of positive RNA viruses.

RNA tertiary structure of the HIV RRE domain II containing non-Watson-Crick base pairs GG and GA: molecular modeling studies.

We have used molecular modeling techniques to model the RNA tertiary structure of the viral RNA element (referred to as domain II of Rev responsive element, RRE) bound by the Rev protein of HIV. In this study, the initial three-dimensional model was built from its established RNA secondary structure, including three non-Watson-Crick G:G, G:A and G:U base pairs. Molecular dynamics (MD) simulations were performed with hydrated or unhydrated sodium ions. Our results indicate that the non-Watson-Crick base pairs in the simulation with unhydrated sodium ions and water are more stable than those with hydrated sodium ions only. The RNA can maintain its compact double helical structure throughout the course of the MD simulations with water and unhydrated sodium ions, although the non-Watson-Crick base pairs and two bulge loops show much more flexibility and conformational distortion than the classical RNA helical region. The distinct distortion of the sugar-phosphate backbone significantly widens the RNA major groove so that the major groove is readily accessible for hydrogen bonding by specific Rev binding. This model emphasizes the importance of specific hydrogen bonding in the stabilization of the three-dimensional structure of the HIV Rev core binding element, not only between the nucleotide bases, but also among the ribose hydroxyls, phosphate anionic oxygens, base oxygens and nitrogens, and bridging water molecules. Moreover, our results suggest that sodium ions play an important role in the formation of base pairs G:G and G:A of the RRE by a manner similar to the arginine of the Rev-RRE complex.

Distinct structural elements and internal entry of ribosomes in mRNA3 encoded by infectious bronchitis virus.

Infectious bronchitis virus (IBV) mRNA3 encodes three small proteins, 3a, 3b, and 3c, at its 5' end. Recently, it was demonstrated that initiation of protein 3c is dependent on the upstream sequence. Monte Carlo simulations of RNA folding in this tricistronic mRNA3 indicate that a highly significant folding region occurs prior to the initiator AUG of 3c. The unusual folding region (UFR) of 265 nucleotides (nt) contains the coding sequences of proteins 3a and 3b. Details of the structural analyses show that five highly significant RNA stem-loops in the UFR can be modeled into a compact superstructure by the interaction of two predicted pseudoknot structures. The folded superstructure comprising nt 44 to 330, with additional 22 nt downstream from this UFR, is suggested to serve as a ribosome landing pad (or an internal ribosomal entry site) in the cap-independent translation of the 3c of IBV. Intriguingly, the proposed structural motif of this coronavirus shares structural features similar to those proposed in a number of picornavirus mRNAs. Based on the common structural features, a plausible base pairing model between mRNA3 and 18 S rRNA is suggested, which is consistent with a general mechanism for regulation of internal initiation described in many picornaviruses.

Conserved tertiary structural elements in the 5' nontranslated region of cardiovirus, aphthovirus and hepatitis A virus RNAs.

Statistical analyses of RNA folding in 5' nontranslated regions (5'NTR) of encephalomyocarditis virus, Theiler's murine encephalomyelitis virus, foot-and-mouth disease virus, and hepatitis A virus indicate that two highly significant folding regions occur in the 5' and 3' portions of the 5'NTR. The conserved tertiary structural elements are predicted in the unusual folding regions (UFR) for these viral RNAs. The theoretical, common structural elements predicted in the 3' parts of the 5'NTR occur in a cis-acting element that is critical for internal ribosome binding. These structural motifs are expected to be highly significant from extensive Monte Carlo simulations. Nucleotides (nt) in the conserved single-stranded polypyrimidine tract for these RNAs are involved in a distinctively tertiary interaction that is located at about 15 nt prior to the initiator AUG. Intriguingly, the proposed common tertiary structure in this study shares a similar structural feature to that proposed in human enteroviruses and rhinoviruses. Based on these common structural features, plausible base pairing models between these viral RNAs and 18 S rRNA are suggested, which are consistent with a general mechanism for regulation of internal initiation of cap-independent translation.

Prediction of alternative RNA secondary structures based on fluctuating thermodynamic parameters.

In this paper we present a new method for predicting a set of RNA secondary structures that are thermodynamically favored in RNA folding simulations. This method uses a large number of 'simulated energy rules' (SER) generated by perturbing the free energy parameters derived experimentally within the range of the experimental errors. The structure with the lowest free energy is computed for each SER. Structural comparisons are used to avoid multiple generation of similar structures. Computed structures are evaluated using the energy distribution of the lowest free energy structures derived in the simulation. Predicted be graphically displayed with their occurring frequencies in the simulation by dot-plot representations. On average, about 90% of phylogenetic helixes in the known models of tRNA, Group I self-splicing intron, and Escherichia coli 16 S rRNA, were predicted using the method.

Identification of unusual RNA folding patterns encoded by bacteriophage T4 gene 60.

A 50-nucleotide (nt) untranslated region (coding gap sequence) that interrupts the amino acid coding sequence in T4 gene 60, plus an additional 5 nt upstream and another 3 nt downstream from the gap sequence, shows unusual folding patterns according to RNA structure prediction. A predicted highly stable and significant hairpin structure in the 5' half of the gap sequence and a plausible tertiary structural element computed in the 3' part of the gap sequence seem significant by statistical tests on the wild-type (wt) sequence. This feature is absent in insertion, deletion and substitution variants of the gap sequence, in which template activities are markedly lower than that of the wt. The proposed feature is consistent with currently available data showing that the translational bypass of the coding gap is correlated with a stop codon involved in a stem-loop structure folded in the gap sequence. We suggest that the role of this segment in 'ribosomal bypass' of a portion of the mRNA sequence is a property of its special folded structure.

Conserved tertiary structure elements in the 5' untranslated region of human enteroviruses and rhinoviruses.

A combination of comparative sequence analysis and thermodynamic methods reveals the conservation of tertiary structure elements in the 5' untranslated region (UTR) of human enteroviruses and rhinoviruses. The predicted common structural elements occur in the 3' end of a segment that is critical for internal ribosome binding, termed "ribosome landing pad" (RLP), of polioviruses. Base pairings between highly conserved 17-nucleotide (nt) and 21-nt sequences in the 5' UTR of human enteroviruses and rhinoviruses constitute a predicted pseudoknot that is significantly more stable than those that can be formed from a large set of randomly shuffled sequences. A conserved single-stranded polypyrimidine tract is located between two conserved tertiary elements. R. Nicholson, J. Pelletier, S.-Y. Le, and N. Sonenberg (1991, J. Virol. 65, 5886-5894) demonstrated that the point mutations of 3-nt UUU out of an essential 4-nt pyrimidine stretch sequence UUUC abolished translation. Structural analysis of the mutant sequence indicates that small point mutations within the short polypyrimidine sequence would destroy the tertiary interaction in the predicted, highly ordered structure. The proposed common tertiary structure can offer experimentalists a model upon which to extend the interpretations for currently available data. Based on these structural features possible base-pairing models between human enteroviruses and 18 S rRNA and between human rhinoviruses and 18 S rRNA are proposed. The proposed common structure implicates a biological function for these sequences in translational initiation.