Search

An improved dynamic programming algorithm is reported for RNA secondary structure prediction by free energy minimization. Thermodynamic parameters for the stabilities of secondary structure motifs are revised to include expanded sequence dependence as revealed by recent experiments. Additional algorithmic improvements include reduced search time and storage for multibranch loop free energies and improved imposition of folding constraints. An extended database of 151,503 nt in 955 structures? determined by comparative sequence analysis was assembled to allow optimization of parameters not based on experiments and to test the accuracy of the algorithm. On average, the predicted lowest free energy structure contains 73 % of known base-pairs when domains of fewer than 700 nt are folded; this compares with 64 % accuracy for previous versions of the algorithm and parameters. For a given sequence, a set of 750 generated structures contains one structure that, on average, has 86 % of known base-pairs. Experimental constraints, derived from enzymatic and flavin mononucleotide cleavage, improve the accuracy of structure predictions., (Copyright 1999 Academic Press.)

We have analyzed both conformational and functional changes caused by two large cis-acting deletions (delta 159 and delta 549) located within the read-through domain, a 850 nucleotide hairpin, in coliphage Q beta genomic RNA. Studies in vivo show that co-translational regulation of the viral coat and replicase genes has been uncoupled in viral genomes carrying deletion delta 159. Translational regulation is restored in deletion delta 549, a naturally evolved pseudorevertant. Structural analysis by computer modeling shows that structural features within the read-through domain of delta 159 RNA are less well determined than they are in the read-through domain of wild-type RNA, whereas predicted structure in the read-through domain of evolved pseudorevertant delta 549 is unusually well determined. Structural analysis by electron microscopy of the genomic RNAs shows that several long range helices at the base of the read-through domain, that suppress translational initiation of the viral replicase gene in the wild-type genome, have been destabilized in delta 159 RNA. In addition, the structure of local hairpins within the read-through region is more variable in delta 159 RNA than in wild-type RNA. Stable RNA secondary structure is restored in the read-through domain of delta 549 RNA. Our analyses suggest that structure throughout the read-through domain affects the regulation of viral replicase expression by altering the likelihood that long-range interactions at the base of the domain will form. We discuss possible kinetic and equilibrium models that can explain this effect, and argue that observed changes in structural plasticity within the read-through domain of the mutant genomes are key in understanding the process. During the course of these studies, we became aware of the importance of the information contained in the energy dot plot produced by the RNA secondary structure prediction program mfold. As a result, we have improved the graphical representation of this information through the use of color annotation in the predicted optimal folding. The method is presented here for the first time.

We propose a number of distance measures between residues in protein structures based on average, minimum and maximum distances of all atom (backbone and side-chain) coordinates or with respect to side-chain atom coordinates only. The d1-distance (D1-distance) refers to the average distance between side-chain (backbone and side-chain) atoms of a residue pair in a given structure. The dm-distance (Dm-distance) refers to the minimum distance between side-chain atoms (non-trivial minimum distance between all atoms of a residue pair). For each distance measure, averaging and normalizing over representative protein structures, association values and closeness orderings for all amino acid types are determined. The expected associations of side-chain interactions between oppositely charged residues, among hydrophobic residues and of cysteine with cysteine are confirmed. Several surprising associations are observed relative to (1) the aromatic residues tyrosine and tryptophan, but not phenylalanine; (2) multiple histidine residues; (3) asymmetries of arginine versus lysine, aspartate versus glutamate, alanine versus glycine, and asparagine versus glutamine; (4) absence of correlations of alpha-carbon distances with side-chain distances. The all atoms D1-distance attractions are dominated by steric relationships, with glycine and alanine significantly close to all amino acids, whereas large residues are under-associated with all residue types. In contrast, for the closeness ordering corresponding to the minimum side-chain dm-distance, glycine and alanine are among the least associated. However, in the d1-distance alanine is significantly close to all hydrophobic residues with the exception of tryptophan. The dm-distance preferences display a pervasive attraction for tyrosine by almost all residue types, the prominence of tyrosine and tryptophan in cation-aromatic interactions, and the versatility of histidine in functionality. The principal findings suggest a new perspective on the early and intermediate stages of protein folding.

We have predicted the secondary structure of the entire 4217 nucleotide sequence of the genomic RNA of coliphage Q beta in one computer run using the computer program MFOLD that computes RNA structures within any prescribed increment of the computed minimum free energy. The results are presented in the form of an "energy dot plot" that shows both an optimal folding as well as the superposition of all base-pairs that can form in slightly suboptimal foldings. The plot reveals five large, well-determined, independent structural domains that cover approximately 50% of the viral genome. The predicted structural domains are consistent with and provide support for five large structural domains identified previously by quantitative electron microscopy in Q beta RNA. The dot plot also contains cluttered regions that indicate large numbers of alternative foldings within or between segments of an RNA molecule. These reflect the impossibility of accurate structure prediction and/or the biological reality of more than one folding. Weaker, long range structures, that are observed by electron microscopy in two alternate competing conformations, are located in the regions of the Q beta sequence that correspond to cluttered regions of the dot plot. The potential biological significance of these secondary structures is discussed.

A molecular sequence alignment algorithm based on dynamic programming has been extended to allow the computation of all pairs of residues that can be part of optimal and suboptimal sequence alignments. The uncertainties inherent in sequence alignment can be displayed using a new form of dot plot. The method allows the qualitative assessment of whether or not two sequences are related, and can reveal what parts of the alignment are better determined than others. It also permits the computation of representative optimal and suboptimal alignments. The relation between alignment reliability and alignment parameters is discussed. Other applications are to cyclical permutations of sequences and the detection of self-similarity. An application to multiple sequence alignment is noted.

A total of 4051 suboptimal secondary structures are predicted by folding the 5' non-coding region of ten polioviruses, five human rhinoviruses and three coxsackieviruses using our new suboptimal folding algorithm for the prediction of both optimal and suboptimal RNA secondary structures. A comparative analysis of these RNA secondary structures reveals the conservation of common secondary structure that can be supported by phylogenetic data. The thermodynamic stability and statistical significance of these predicted, conserved helical elements are assessed and significant structure motifs in the 5' non-coding region are proposed. The possible roles of these structure motifs in the virus life cycle are discussed.

The structure of single-stranded RNA from the bacteriophage MS2 has been examined by electron microscopy in the presence of the polyamine spermidine. The molecules are found in two alternate conformations. The first of these can be characterized as a cruciform structure composed of three large loops approximately 500 to 700 nucleotides in size. The interior of the molecule has extensive base-paired regions which connect distant regions of the molecule; the farthest being 2500 nucleotides apart. In the second conformation, the molecules appear rod-like. Two of the large loops disappear, and these regions form, instead, extensive long-range helices. Computer modeling has been employed to explore the base-pairing potential of the sequence of bacteriophage MS2 RNA. Double-stranded regions identified by electron microscopy are shown to occur in local G + C-rich stretches of the RNA. Detailed models have been calculated for two regions of long-range contact. One of these includes the ribosome-binding site for the viral coat protein gene. The results are discussed in the context of the known role of RNA structure in the regulation of viral gene expression.