Molecular basis of box C/D RNA-protein interactions; cocrystal structure of archaeal L7Ae and a box C/D RNA

We have determined and refined a crystal structure of the initial assembly complex of archaeal box C/D sRNPs comprising the Archaeoglobus fulgidus (AF) L7Ae protein and a box C/D RNA. The box C/D RNA forms a classical kink-turn (K-turn) structure and the resulting protein-RNA complex serves as a distinct platform for recruitment of the fibrillarin-Nop5p complex. The cocrystal structure confirms previously proposed secondary structure of the box C/D RNA that includes a protruded U, a UU mismatch, and two sheared tandem GA base pairs. Detailed structural comparisons of the AF L7Ae-box C/D RNA complex with previously determined crystal structures of L7Ae homologs in complex with functionally distinct K-turn RNAs revealed a set of remarkably conserved principles in protein-RNA interactions. These analyses provide a structural basis for interpreting the functional roles of the box C/D sequences in directing specific assembly of box C/D sRNPs.     

Computational methods for prediction of protein-RNA interactions

Understanding the molecular mechanism of protein-RNA recognition and complex formation is a major challenge in structural biology. Unfortunately, the experimental determination of protein-RNA complexes by X-ray crystallography and nuclear magnetic resonance spectroscopy (NMR) is tedious and difficult. Alternatively, protein-RNA interactions can be predicted by computational methods. Although less accurate than experimental observations, computational predictions can be sufficiently accurate to prompt functional hypotheses and guide experiments, e.g. to identify individual amino acid or nucleotide residues. In this article we review 10 methods for predicting protein-RNA interactions, seven of which predict RNA-binding sites from protein sequences and three from structures. We also developed a meta-predictor that uses the output of top three sequence-based primary predictors to calculate a consensus prediction, which outperforms all the primary predictors. In order to fully cover the software for predicting protein-RNA interactions, we also describe five methods for protein-RNA docking. The article highlights the strengths and shortcomings of existing methods for the prediction of protein-RNA interactions and provides suggestions for their further development.     

Structural insight into the mechanism of double-stranded RNA processing by ribonuclease III

Members of the ribonuclease III (RNase III) family are double-stranded RNA (dsRNA) specific endoribonucleases characterized by a signature motif in their active centers and a two-base 3' overhang in their products. While Dicer, which produces small interfering RNAs, is currently the focus of intense interest, the structurally simpler bacterial RNase III serves as a paradigm for the entire family. Here, we present the crystal structure of an RNase III-product complex, the first catalytic complex observed for the family. A 7 residue linker within the protein facilitates induced fit in protein-RNA recognition. A pattern of protein-RNA interactions, defined by four RNA binding motifs in RNase III and three protein-interacting boxes in dsRNA, is responsible for substrate specificity, while conserved amino acid residues and divalent cations are responsible for scissile-bond cleavage. The structure reveals a wealth of information about the mechanism of RNA hydrolysis that can be extrapolated to other RNase III family members.     

A new residue-nucleotide propensity potential with structural information considered for discriminating protein-RNA docking decoys

Understanding the key factors that influence the preferences of residue-nucleotide interactions in specific protein-RNA interactions has remained a research focus. We propose an effective approach to derive residue-nucleotide propensity potentials through considering both the types of residues and nucleotides, and secondary structure information of proteins and RNAs from the currently largest nonredundant and nonribosomal protein-RNA interaction database. To test the validity of the potentials, we used them to select near-native structures from protein-RNA docking poses. The results show that considering secondary structure information, especially for RNAs, greatly improves the predictive power of pair potentials. The success rate is raised from 50.7 to 65.5% for the top 2000 structures, and the number of cases in which a near-native structure is ranked in top 50 is increased from 7 to 13 out of 17 cases. Furthermore, the exclusion of ribosomes from the database contributes 8.3% to the success rate. In addition, some very interesting findings follow: (i) the protein secondary structure element π-helix is strongly associated with RNA-binding sites; (ii) the nucleotide uracil occurs frequently in the most preferred pairs in which the unpaired and non-Watson-Crick paired uracils are predominant, which is probably significant in evolution. The new residue-nucleotide potentials can be helpful for the progress of protein-RNA docking methods, and for understanding the mechanisms of protein-RNA interactions.     

Binding of U1A protein changes RNA dynamics as observed by 13C NMR relaxation studies

Recognition of RNA by proteins and small molecules often involves large changes in RNA structure and dynamics, yet very few studies have so far characterized these motional changes. Here we extend to the protein-bound RNA recent 13C relaxation studies of motions in the RNA recognized by human U1A protein, a well-known model for protein-RNA recognition. Changes in relaxation observed upon complex formation demonstrate that the protein-binding site becomes rigid in the complex, but the upper stem-loop that defines the secondary structure of this RNA experiences unexpected motional freedom. By using a helix elongation strategy, we observe that the upper stem-loop moves independently of the remainder of the structure also in the absence of U1A. Surprisingly, RNA residues making important intermolecular contacts in the structure of the complex exhibit increased flexibility in the presence of the protein. Both of these results support the hypothesis that RNA-binding proteins select a structure that optimizes intermolecular contacts in the manifold of conformations sampled by the free RNA and that protein binding quenches these motions. Together with previous studies of the RNA-bound protein, they also demonstrate that protein-RNA interfaces experience complex motions that modulate the strength of individual interactions.     

PRI-Modeler: extracting RNA structural elements from PDB files of protein-RNA complexes

A complete understanding of protein and RNA structures and their interactions is important for determining the binding sites in protein-RNA complexes. Computational approaches exist for identifying secondary structural elements in proteins from atomic coordinates. However, similar methods have not been developed for RNA, due in part to the very limited structural data so far available. We have developed a set of algorithms for extracting and visualizing secondary and tertiary structures of RNA and for analyzing protein-RNA complexes. These algorithms have been implemented in a web-based program called PRI-Modeler (protein-RNA interaction modeler). Given one or more protein data bank files of protein-RNA complexes, PRI-Modeler analyzes the conformation of the RNA, calculates the hydrogen bond (H bond) and van der Waals interactions between amino acids and nucleotides, extracts secondary and tertiary RNA structure elements, and identifies the patterns of interactions between the proteins and RNAs. This paper presents PRI-Modeler and its application to the hydrogen bond and van der Waals interactions in the most representative set of protein-RNA complexes. The analysis reveals several interesting interaction patterns at various levels. The information provided by PRI-Modeler should prove useful for determining the binding sites in protein-RNA complexes. PRI-Modeler is accessible at http://wilab.inha.ac.kr/primodeler/, and supplementary materials are available in the analysis results section at http://wilab.inha.ac.kr/primodeler/.     

A protein-RNA docking benchmark (II): extended set from experimental and homology modeling data

We present here an extended protein-RNA docking benchmark composed of 71 test cases in which the coordinates of the interacting protein and RNA molecules are available from experimental structures, plus an additional set of 35 cases in which at least one of the interacting subunits is modeled by homology. All cases in the experimental set have available unbound protein structure, and include five cases with available unbound RNA structure, four cases with a pseudo-unbound RNA structure, and 62 cases with the bound RNA form. The additional set of modeling cases comprises five unbound-model, eight model-unbound, 19 model-bound, and three model-model protein-RNA cases. The benchmark covers all major functional categories and contains cases with different degrees of difficulty for docking, as far as protein and RNA flexibility is concerned. The main objective of this benchmark is to foster the development of protein-RNA docking algorithms and to contribute to the better understanding and prediction of protein-RNA interactions. The benchmark is freely available at http://life.bsc.es/pid/protein-rna-benchmark.     

Discovering the interaction propensities of amino acids and nucleotides from protein-RNA complexes

With the availability of many genome sequences, the mining of biological data is attracting much attention, most of it limited to the sequences of macromolecules. Sequence data are easy to analyze as they can be treated as strings of characters, whereas the structure of a macromolecule is much more complex. We developed a set of algorithms to analyze the structures of protein-RNA complexes at the atomic level and used them to analyze protein-RNA interactions using structural data on 51 protein-RNA complexes. The analysis revealed, among other things, that: (1) polar and charged amino acids have a strong tendency to interact with nucleotides, (2) arginine and asparagine tend to hydrogen bond with uracil, and (3) histidine favors uracil in water-mediated bonding with RNA. We analyzed a large set of structural data of protein-RNA complexes involving water-mediated hydrogen bonds as well as direct hydrogen bonds. The interaction patterns discovered from the analysis provide useful information for predicting the structure of RNA that binds proteins, and of proteins that bind RNA.     

Computational analysis of hydrogen bonds in protein-RNA complexes for interaction patterns

Structural analysis of protein-RNA complexes is labor-intensive, yet provides insight into the interaction patterns between a protein and RNA. As the number of protein-RNA complex structures reported has increased substantially in the last few years, a systematic method is required for automatically identifying interaction patterns. This paper presents a computational analysis of the hydrogen bonds in the most representative set of protein-RNA complexes. The analysis revealed several interesting interaction patterns. (1) While residues in the beta-sheets favored unpaired nucleotides, residues in the helices showed no preference and residues in turns favored paired nucleotides. (2) The backbone hydrogen bonds were more dominant than the base hydrogen bonds in the paired nucleotides, but the reverse was observed in the unpaired nucleotides. (3) The protein-RNA complexes contained more paired nucleotides than unpaired nucleotides, but the unpaired nucleotides were observed more frequently interacting with the proteins. And (4) Arg-U, Thr-A, Lys-A, and Asn-U were the most frequently observed pairs. The interaction patterns discovered from the analysis will provide us with useful information in predicting the structure of the RNA binding protein and the structure of the protein binding RNA.     

Functional protein microarray: an ideal platform for investigating protein binding property

Functional protein microarray is an important tool for high-throughput and large-scale systems biology studies.Besides the progresses that have been made for protein microarray fabrication,significant ...     

Recognition of GU-rich polyadenylation regulatory elements by human CstF-64 protein

Vertebrate polyadenylation sites are identified by the AAUAAA signal and by GU-rich sequences downstream of the cleavage site. These are recognized by a heterotrimeric protein complex (CstF) through its 64 kDa subunit (CstF-64); the strength of this interaction affects the efficiency of poly(A) site utilization. We present the structure of the RNA-binding domain of CstF-64 containing an RNA recognition motif (RRM) augmented by N- and C-terminal helices. The C-terminal helix unfolds upon RNA binding and extends into the hinge domain where interactions with factors responsible for assembly of the polyadenylation complex occur. We propose that this conformational change initiates assembly. Consecutive Us are required for a strong CstF-GU interaction and we show how UU dinucleotides are recognized. Contacts outside the UU pocket fine tune the protein-RNA interaction and provide different affinities for distinct GU-rich elements. The protein-RNA interface remains mobile, most likely a requirement to bind many GU-rich sequences and yet discriminate against other RNAs. The structural distinction between sequences that form stable and unstable complexes provides an operational distinction between weakly and strongly processed poly(A) sites.     

Dissecting the protein-RNA interface: the role of protein surface shapes and RNA secondary structures in protein-RNA recognition

Protein-RNA interactions are essential for many biological processes. However, the structural mechanisms underlying these interactions are not fully understood. Here, we analyzed the protein surface shape (dented, intermediate or protruded) and the RNA base pairing properties (paired or unpaired nucleotides) at the interfaces of 91 protein-RNA complexes derived from the Protein Data Bank. Dented protein surfaces prefer unpaired nucleotides to paired ones at the interface, and hydrogen bonds frequently occur between the protein backbone and RNA bases. In contrast, protruded protein surfaces do not show such a preference, rather, electrostatic interactions initiate the formation of hydrogen bonds between positively charged amino acids and RNA phosphate groups. Interestingly, in many protein-RNA complexes that interact via an RNA loop, an aspartic acid is favored at the interface. Moreover, in most of these complexes, nucleotide bases in the RNA loop are flipped out and form hydrogen bonds with the protein, which suggests that aspartic acid is important for RNA loop recognition through a base-flipping process. This study provides fundamental insights into the role of the shape of the protein surface and RNA secondary structures in mediating protein-RNA interactions.     

Protein-protein interactions required during translation

Protein synthesis requires the involvement of numerous accessory factors that assist the ribosome in translation initiation, elongation, and termination. Extensive protein-protein and protein-RNA interactions are required to bring together the accessory factors, tRNAs, ribosomes, and mRNA into a productive complex and these interactions undergo dynamic alterations during each step of the translation initiation process. Initiation represents the most complex aspect of translation, requiring more accessory proteins, called initiation factors, than either elongation or termination. Not surprisingly, initiation is most often the rate-limiting step of translation and, as such, most (but not all) examples of translational regulation involve the regulation of protein-protein or protein-RNA interactions of the initiation complex. In this review, we focus on those interactions required for efficient translation initiation and how such interactions are regulated by developmental or environmental signals.     

Protein-RNA interface residue prediction using machine learning: an assessment of the state of the art

ABSTRACT: BACKGROUND: RNA molecules play diverse functional and structural roles in cells. They function as messengers for transferring genetic information from DNA to proteins, as the primary genetic material in many viruses, as catalysts (ribozymes) important for protein synthesis and RNA processing, and as essential and ubiquitous regulators of gene expression in living organisms. Many of these functions depend on precisely orchestrated interactions between RNA molecules and specific proteins in cells. Understanding the molecular mechanisms by which proteins recognize and bind RNA is essential for comprehending the functional implications of these interactions, but the recognition 'code' that mediates interactions between proteins and RNA is not yet understood. Success in deciphering this code would dramatically impact the development of new therapeutic strategies for intervening in devastating diseases such as AIDS and cancer. Because of the high cost of experimental determination of protein-RNA interfaces, there is an increasing reliance on statistical machine learning methods for training predictors of RNA-binding residues in proteins. However, because of differences in the choice of datasets, performance measures, and data representations used, it has been difficult to obtain an accurate assessment of the current state of the art in protein-RNA interface prediction. RESULTS: We provide a review of published approaches for predicting RNA-binding residues in proteins and a systematic comparison and critical assessment of protein-RNA interface residue predictors trained using these approaches on three carefully curated non-redundant datasets. We directly compare two widely used machine learning algorithms (Naive Bayes (NB) and Support Vector Machine (SVM)) using three different data representations in which features are encoded using either sequence- or structure-based windows. Our results show that (i) Sequence-based classifiers that use a position-specific scoring matrix (PSSM)-based representation (PSSMSeq) outperform those that use an amino acid identity based representation (IDSeq) or a smoothed PSSM (SmoPSSMSeq); (ii) Structure-based classifiers that use smoothed PSSM representation (SmoPSSMStr) outperform those that use PSSM (PSSMStr) as well as sequence identity based representation (IDStr). PSSMSeq classifiers, when tested on an independent test set of 44 proteins, achieve performance that is comparable to that of three state-of-the-art structure-based predictors (including those that exploit geometric features) in terms of Matthews Correlation Coefficient (MCC), although the structure-based methods achieve substantially higher Specificity (albeit at the expense of Sensitivity) compared to sequence-based methods. We also find that the expected performance of the classifiers on a residue level can be markedly different from that on a protein level. Our experiments show that the classifiers trained on three different non-redundant protein-RNA interface datasets achieve comparable cross-validation performance. However, we find that the results are significantly affected by differences in the distance threshold used to define interface residues. CONCLUSIONS: Our results demonstrate that protein-RNA interface residue predictors that use a PSSM-based encoding of sequence windows outperform classifiers that use other encodings of sequence windows. While structure-based methods that exploit geometric features can yield significant increases in the Specificity of protein-RNA interface residue predictions, such increases are offset by decreases in Sensitivity. These results underscore the importance of comparing alternative methods using rigorous statistical procedures, multiple performance measures, and datasets that are constructed based on several alternative definitions of interface residues and redundancy cutoffs as well as including evaluations on independent test sets into the comparisons.     

The N-terminus of the fragile X mental retardation protein contains a novel domain involved in dimerization and RNA binding

Fragile X syndrome, the most common cause of inherited mental retardation, is caused by the absence of the fragile X mental retardation protein (FMRP). The emerging picture is that FMRP is involved in repression of translation through a complex network of protein-protein and protein-RNA interactions. Very little structural information is, however, available for FMRP that could help to understand its function. In particular, no structural studies are available about the N-terminus of the protein, a highly conserved region which is involved in several molecular interactions. Here, we explore systematically the ability of the FMRP N-terminus to form independently folded units (domains). We produced deletion mutants and tested their fold and functional properties by mutually complementary biophysical and biochemical techniques. On the basis of our data, we conclude that the N-terminus contains a domain, that we named NDF, comprising the first 134 amino acids. Most interestingly, NDF comprises two copies of a newly identified Agenet motif. NDF is thermally stable and has a high content of beta structure. In addition to being able to bind to RNA and to recognize some of the FMRP interacting proteins, NDF forms stable dimers and is able to interact, although weakly, with the full-length protein. Our data provide conclusive evidence that NDF is a novel motif for protein-protein and protein-RNA interactions and contains a previously unidentified dimerization site.     

A novel loop-loop recognition motif in the yeast ribosomal protein L30 autoregulatory RNA complex

The yeast Saccharomyces cerevisiae ribosomal protein L30 negatively autoregulates its production by binding to a helix-loop-helix structure formed in its pre-mRNA and its mRNA. A three-dimensional solution structure of the L30 protein in complex with its regulatory RNA has been solved using NMR spectroscopy. In the complex, the helix-loop-helix RNA adopts a sharply bent conformation at the internal loop region. Unusual RNA features include a purine stack, a reverse Hoogsteen base pair (G11anti-G56syn) and highly distorted backbones. The L30 protein is folded in a three-layer alpha/beta/alpha sandwich topology, and three loops at one end of the sandwich make base-specific contacts with the RNA internal loop. The protein-RNA binding interface is divided into two clusters, including hydrophobic and aromatic stacking interactions centering around G56, and base-specific hydrogen-bonding contacts to A57, G58 and G10-U60 wobble base pair. Both the protein and the RNA exhibit a partially induced fit for binding, where loops in the protein and the internal loop in the RNA become more ordered upon complex formation. The specific interactions formed between loops on L30 and the internal loop on the mRNA constitute a novel loop-loop recognition motif where an intimate RNA-protein interface is formed between regions on both molecules that lack regular secondary structure.     

Protein-RNA contacts at crystal packing surfaces

The crystal packing surfaces comprising protein-RNA interactions were analyzed for 50 RNA-protein crystal structures in the Protein Data Bank database. Protein-RNA crystal contacts, which represent nonspecific protein-RNA interfaces, were investigated for their amino acid propensities, hydrogen bond patterns, and backbone and side chain interactions. When compared to biologically relevant interactions, the protein-RNA crystal contacts exhibit similarities as well as differences with respect to the principles of protein-RNA interactions. Similar to what was observed at cognate protein-RNA interfaces, positively charged amino acids have high propensities at noncognate protein-RNA interfaces and preferentially form hydrogen bonds with RNA phosphate groups. In contrast, nonpolar residues are less frequently associated with noncognate interactions. These results highlight the important roles of both electrostatic and hydrogen bonding interactions, facilitated by positively charged amino acids, in mediating both specific and nonspecific protein-RNA interactions.