A probabilistic model for secondary structure prediction from protein chemical shifts

Protein chemical shifts encode detailed structural information that is difficult and computationally costly to describe at a fundamental level. Statistical and machine learning approaches have been used to infer correlations between chemical shifts and secondary structure from experimental chemical shifts. These methods range from simple statistics such as the chemical shift index to complex methods using neural networks. Notwithstanding their higher accuracy, more complex approaches tend to obscure the relationship between secondary structure and chemical shift and often involve many parameters that need to be trained. We present hidden Markov models (HMMs) with Gaussian emission probabilities to model the dependence between protein chemical shifts and secondary structure. The continuous emission probabilities are modeled as conditional probabilities for a given amino acid and secondary structure type. Using these distributions as outputs of first‐ and second‐order HMMs, we achieve a prediction accuracy of 82.3%, which is competitive with existing methods for predicting secondary structure from protein chemical shifts. Incorporation of sequence‐based secondary structure prediction into our HMM improves the prediction accuracy to 84.0%. Our findings suggest that an HMM with correlated Gaussian distributions conditioned on the secondary structure provides an adequate generative model of chemical shifts. Proteins 2013; © 2012 Wiley Periodicals, Inc.     

Interpreting oligonucleotide microarray data to determine RNA secondary structure: application to the 3' end of Bombyx mori R2 RNA

A method to deduce RNA secondary structure on the basis of data from microarrays of 2'-O-methyl RNA 9-mers immobilized in agarose film on glass slides is tested with a 249 nucleotide RNA from the 3' end of the R2 retrotransposon from Bombyx mori. Various algorithms incorporating binding data and free-energy minimization calculations were compared for interpreting the data to provide possible secondary structures. Two different methods give structures with 100 and 87% of the base pairs determined by sequence comparison. In contrast, structures predicted by free-energy minimization alone by Mfold and RNAstructure contain 52 and 72% of the known base pairs, respectively. This combination of high throughput microarray techniques with algorithms using free-energy calculations has potential to allow for fast determination of RNA secondary structure. It should also facilitate the design of antisense and siRNA oligonucleotides.     

Unnatural amino acid mutagenesis: a precise tool for probing protein structure and function

The first general method for the biosynthetic incorporation of unnatural amino acids into proteins was reported in 1989. The ensuing years have seen the solid development and subsequent implementation of "unnatural amino acid mutagenesis" in a number of groundbreaking studies. Over 100 different amino acids have been incorporated into dozens of soluble and transmembrane proteins, using both cell-extract and cell-intact translation systems. The approach has provided insights into ligand-binding sites, conformational changes, and protein-protein interactions with a level of precision simply unparalleled by conventional mutagenesis. Here, the methodology is outlined, significant applications of the approach are summarized, and recent major improvements in the method are discussed. The future will likely see many more investigators utilizing this approach to manipulate proteins as it realizes its promise of becoming a tool with enormous potential.     

Assessment of a model for intron RNA secondary structure relevant to RNA self-splicing--a review

A widespread class of introns is characterized by a particular RNA secondary structure, based upon four conserved nucleotide sequences. Among such "class I" introns are found the majority of introns in fungal mitochondrial genes and the self-splicing intron of the large ribosomal RNA of several species of Tetrahymena. A model of the RNA secondary structure, which must underlie the self-splicing activity, is here evaluated in the light of data on 16 further introns. The main body or "core structure" of the intron always consists of the base-paired regions P3 to P9 with the associated single-stranded loops, with P2 present also in most cases. Two minority sub-classes of core structure occur, one of which is typical of introns in fungal ribosomal RNA. Introns in which the core structure is close to the 5' splice site all have an internal guide sequence (IGS) which can pair with exon sequences adjacent to the 5' and 3' splice sites to align them precisely, as proposed by Davies et al. [Nature 300 (1982) 719-724]. In these cases, the internal guide model allows us to predict correctly the exact location of splice sites. All other introns probably use other mechanisms of alignment. This analysis provides strong support for the RNA splicing model which we have developed.     

Human immunodeficiency virus rev protein recognizes a target sequence in rev-responsive element RNA within the context of RNA secondary structure.

Human immunodeficiency virus type 1 Rev protein modulates the distribution of viral mRNAs from the nucleus to the cytoplasm by interaction with a highly structured viral RNA sequence, the Rev-responsive element (RRE). To identify the minimal functional elements of RRE, we evaluated mutant RREs for Rev binding in vitro and Rev response in vivo in the context of a Gag expression plasmid. The critical functional elements fold into a structure composed of a stem-loop A, formed by the ends of the RRE, joined to a branched stem-loop B/B1/B2, between bases 49 and 113. The 5' 132 nucleotides of RRE, RREDDE, which possessed a similar structure, bound Rev efficiently but were nonfunctional in vivo, implying separate binding and functional domains within the RRE. Excision of stem-loop A reduced Rev binding significantly and abolished the in vivo Rev response. The B2 branch could be removed without severe impairment of binding, but deletions in the B1 branch significantly reduced binding and function. However, deletion of 12 nucleotides, including the 5' strand of stem B, abolished both binding and function, while excision of the 3' strand of stem B only reduced them. Maintenance of the native RRE secondary structure alone was not sufficient for Rev recognition. Many mutations that altered the primary structure of the critical region while preserving the original RNA conformation were Rev responsive. However, mutations that changed a 5'..CACUAUGGG..3' sequence in the B stem, without affecting the overall structure abolished both in vitro Rev binding and the in vivo Rev response.     

Differential chemical probing of a group II self-splicing intron identifies bases involved in tertiary interactions and supports an alternative secondary structure model of domain V.

Dimethyl sulfate modification was used to probe for tertiary structural elements in the group II intron PI.LSU/2 from the mitochondrial pre-ribosomal RNA of the brown alga Pylaiella littoralis. Modification of the lariat form of the intron under conditions that allow both native folding and conformational homogeneity is found to be generally consistent with secondary and tertiary structural features identified previously for group II ribozymes. A comparison of chemical probing at temperatures just below and above the first melting transition illustrates the cooperative unfolding of tertiary structure and identifies novel candidates for tertiary interactions in addition to defining elements of secondary structure. Substitution of the GAAA terminal loop of domain V is shown to be compatible with retention of conformational homogeneity (despite the loss of an important tertiary interaction), but produces a concise methylation footprint in domain I at the site previously shown to harbor the receptor for that loop. The analysis also identified two nucleotide positions in domain V with novel secondary and potential tertiary structural roles. The proposed refinement of domain V secondary structure is supported by an expanded comparative analysis of group II sequences and bears increased resemblance to U2:U6 snRNA pairing in the spliceosome.     

NMR structure of a 4 x 4 nucleotide RNA internal loop from an R2 retrotransposon: identification of a three purine-purine sheared pair motif and comparison to MC-SYM predictions

The NMR solution structure is reported of a duplex, 5'GUGAAGCCCGU/3'UCACAGGAGGC, containing a 4 × 4 nucleotide internal loop from an R2 retrotransposon RNA. The loop contains three sheared purine-purine pairs and reveals a structural element found in other RNAs, which we refer to as the 3RRs motif. Optical melting measurements of the thermodynamics of the duplex indicate that the internal loop is 1.6 kcal/mol more stable at 37°C than predicted. The results identify the 3RRs motif as a common structural element that can facilitate prediction of 3D structure. Known examples include internal loops having the pairings: 5'GAA/3'AGG, 5'GAG/3'AGG, 5'GAA/3'AAG, and 5'AAG/3'AGG. The structural information is compared with predictions made with the MC-Sym program.     

Phylogenetic and biochemical evidence for a secondary structure model of a small cytoplasmic RNA from Bacilli.

Small cytoplasmic RNA (scRNA; 271 nucleotides) is an abundant, stable RNA identified in the Gram-positive eubacterium Bacillus subtilis. Several findings suggest an important role of scRNA in protein biosynthesis: it shares structural and biochemical features with the Escherichia coli 4.5S RNA (114 nucleotides), a molecule known to be involved in this process, and it can complement the essential function of 4.5S RNA in vivo. The common apical hairpin motif of scRNA and 4.5S RNA also exists in eukaryotic 7SL RNA, the RNA component of the signal recognition particle. To elucidate the higher-order structure of scRNA, we have combined a phylogenetic approach with a biochemical one. The sequence of scRNA from a thermophilic relative of B. subtilis, Bacillus stearothermophilus, was determined and compared with the B. subtilis scRNA. In addition, the solution structure of B. stearothermophilus scRNA was probed with single- and double-strand-specific nucleases. Both types of analysis support a secondary structure model for scRNA that strongly resembles 4.5S RNA and respective parts of 7SL RNA. The results provide further evidence for the suggestion of a functional relationship between these RNAs.     

High‐resolution mapping of function and protein binding in an RNA nuclear enrichment sequence

The functions of long RNAs, including mRNAs and long noncoding RNAs (lncRNAs), critically depend on their subcellular localization. The identity of the sequences that dictate subcellular localization and their high‐resolution anatomy remain largely unknown. We used a suite of massively parallel RNA assays and libraries containing thousands of sequence variants to pinpoint the functional features within the SIRLOIN element, which dictates nuclear enrichment through hnRNPK recruitment. In addition, we profiled the endogenous SIRLOIN RNA‐nucleoprotein complex and identified the nuclear RNA‐binding proteins SLTM and SNRNP70 as novel SIRLOIN binders. Taken together, using massively parallel assays, we identified the features that dictate binding of hnRNPK, SLTM, and SNRNP70 to SIRLOIN and found that these factors are jointly required for SIRLOIN activity. Our study thus provides a roadmap for high‐throughput dissection of functional sequence elements in long RNAs.     

The equilibrium partition function and base pair binding probabilities for RNA secondary structure.

A novel application of dynamic programming to the folding problem for RNA enables one to calculate the full equilibrium partition function for secondary structure and the probabilities of various substructures. In particular, both the partition function and the probabilities of all base pairs are computed by a recursive scheme of polynomial order N3 in the sequence length N. The temperature dependence of the partition function gives information about melting behavior for the secondary structure. The pair binding probabilities, the computation of which depends on the partition function, are visually summarized in a "box matrix" display and this provides a useful tool for examining the full ensemble of probable alternative equilibrium structures. The calculation of this ensemble representation allows a proper application and assessment of the predictive power of the secondary structure method, and yields important information on alternatives and intermediates in addition to local information about base pair opening and slippage. The results are illustrated for representative tRNA, 5S RNA, and self-replicating and self-splicing RNA molecules, and allow a direct comparison with enzymatic structure probes. The effect of changes in the thermodynamic parameters on the equilibrium ensemble provides a further sensitivity check to the predictions.     

Genetic evidence for an interaction between a picornaviral cis-acting RNA replication element and 3CD protein

Internally located, cis-acting RNA replication elements, termed cres, are essential for replication of the genomes of picornaviruses such as human rhinovirus 14 (HRV-14) and poliovirus because they template uridylylation of the protein primer, VPg, by the polymerase 3D(pol). These cres form stem-loop structures sharing a common loop motif, and the HRV-14 cre can substitute functionally for the poliovirus cre in both uridylylation in vitro and RNA replication in vivo. We show, however, that the poliovirus cre is unable to support HRV-14 RNA replication. This lack of complementation maps to the stem of the poliovirus cre and was reversed by single nucleotide substitutions in the stem as well as the base of the loop. Replication-competent, revertant viruses rescued from dicistronic HRV-14 RNAs containing the poliovirus cre, or a chimeric cre containing the poliovirus stem, contained adaptive amino acid substitutions. These mapped to the surface of both the polymerase 3D(pol), at the tip of the "thumb" domain, and the protease 3C(pro), on the side opposing the active site and near the end of an extended strand segment implicated previously in RNA binding. These mutations substantially enhanced replication competence when introduced into HRV-14 RNAs containing the poliovirus cre, and they were additive in their effects. The data support a model in which 3CD or its derivatives 3C(pro) and 3D(pol) interact directly with the stem of the cre during uridylylation of VPg.     

R-CHIE: a web server and R package for visualizing RNA secondary structures

Visually examining RNA structures can greatly aid in understanding their potential functional roles and in evaluating the performance of structure prediction algorithms. As many functional roles of RNA structures can already be studied given the secondary structure of the RNA, various methods have been devised for visualizing RNA secondary structures. Most of these methods depict a given RNA secondary structure as a planar graph consisting of base-paired stems interconnected by roundish loops. In this article, we present an alternative method of depicting RNA secondary structure as arc diagrams. This is well suited for structures that are difficult or impossible to represent as planar stem-loop diagrams. Arc diagrams can intuitively display pseudo-knotted structures, as well as transient and alternative structural features. In addition, they facilitate the comparison of known and predicted RNA secondary structures. An added benefit is that structure information can be displayed in conjunction with a corresponding multiple sequence alignments, thereby highlighting structure and primary sequence conservation and variation. We have implemented the visualization algorithm as a web server R-chie as well as a corresponding R package called R4RNA, which allows users to run the software locally and across a range of common operating systems.     

The iStem, a long-range RNA secondary structure element required for efficient exon inclusion in the Drosophila Dscam pre-mRNA

The Drosophila Dscam gene encodes 38,016 different proteins, due to alternative splicing of 95 of its 115 exons, that function in axon guidance and innate immunity. The alternative exons are organized into four clusters, and the exons within each cluster are spliced in a mutually exclusive manner. Here we describe an evolutionarily conserved RNA secondary structure we call the Inclusion Stem (iStem) that is required for efficient inclusion of all 12 variable exons in the exon 4 cluster. Although the iStem governs inclusion or exclusion of the entire exon 4 cluster, it does not play a significant role in determining which variable exon is selected. Thus, the iStem is a novel type of regulatory element that simultaneously controls the splicing of multiple alternative exons.     

A method for probabilistic mapping between protein structure and function taxonomies through cross training

Background

Prediction of function of proteins on the basis of structure and vice versa is a partially solved problem, largely in the domain of biophysics and biochemistry. This underlies the need of computational and bioinformatics approach to solve the problem. Large and organized latent knowledge on protein classification exists in the form of independently created protein classification databases. By creating probabilistic maps between classes of structural classification databases (e.g. SCOP [1]) and classes of functional classification databases (e.g. PROSITE [2]), structure and function of proteins could be probabilistically related.

Results

We demonstrate that PROSITE and SCOP have significant semantic overlap, in spite of independent classification schemes. By training classifiers of SCOP using classes of PROSITE as attributes and vice versa, accuracy of Support Vector Machine classifiers for both SCOP and PROSITE was improved. Novel attributes, 2-D elastic profiles and Blocks were used to improve time complexity and accuracy. Many relationships were extracted between classes of SCOP and PROSITE using decision trees.

Conclusion

We demonstrate that presented approach can discover new probabilistic relationships between classes of different taxonomies and render a more accurate classification. Extensive mappings between existing protein classification databases can be created to link the large amount of organized data. Probabilistic maps were created between classes of SCOP and PROSITE allowing predictions of structure using function, and vice versa. In our experiments, we also found that functions are indeed more strongly related to structure than are structure to functions.     

Study of the secondary and tertiary structure of phage MS2 RNA using nucleases and fluorescent dyes specific for secondary structure

The binding of ethidium bromide (EtBr) and acridine orange (AO) to RNA in native state or after hydrolysis by S1 and SV nucleases that specifically split single-stranded and double-stranded segments was studied. Nuclease S1 hydrolysis of RNA does not increase the number of EtBr strong binding sites, Tm and hyperchromic effect being also unchanged. Hydrolysis by double-stranded segments accessible to EtBr is followed by the diminishing of Tm and hyperchromism. A supposition is put forward that the main role in stabilization of the RNA tertiary structure is played by double-stranded segments arranged so that some of them are hidden and do not interact with dyes. One of the possible models may be parallel oriented intramolecular "hair-pins" forming compact "rod-like" structures.     

ESSA: an integrated and interactive computer tool for analysing RNA secondary structure.

With ESSA, we propose an approach of RNA secondary structure analysis based on extensive viewing within a friendly graphical interface. This computer program is organized around the display of folding models produced by two complementary methods suitable to draw long RNA molecules. Any feature of interest can be managed directly on the display and highlighted by a rich combination of colours and symbols with emphasis given to structural probe accessibilities. ESSA also includes a word searching procedure allowing easy visual identification of structural features even complex and degenerated. Analysis functions make it possible to calculate the thermodynamic stability of any part of a folding using several models and compare homologous aligned RNA both in primary and secondary structure. The predictive capacities of ESSA which brings together the experimental, thermodynamic and comparative methods, are increased by coupling it with a program dedicated to RNA folding prediction based on constraints management and propagation. The potentialities of ESSA are illustrated by the identification of a possible tertiary motif in the LSU rRNA and the visualization of a pseudoknot in S15 mRNA.