Statistical prediction of single-stranded regions in RNA secondary structure and application to predicting effective antisense target sites and beyond

Antisense gapmers selectively suppress individual oncogenic p73 splice isoforms and inhibit tumor growth in vivo

BACKGROUND

Differential mRNA splicing and alternative promoter usage of the TP73 gene results in the expression of multiple NH2-truncated isoforms that act as oncogenes. Abundant levels of these p73 variants in a variety of human cancers correlated with adverse clinical prognosis and response failure to conventional therapies, underscoring their relevance as marker for disease severity and target for cancer intervention. With respect to an equally important role for amino-truncated p73 splice forms (DeltaTAp73) and DeltaNp73 (summarized as DNp73) in the tumorigenic process, we designed locked nucleic acid (LNA) antisense oligonucleotide (ASO) gapmers against individual species that were complementary to DeltaEx2 and DeltaEx2/3 splice junctions and a region in exon 3B unique for DeltaN' and DeltaN.

RESULTS

Treatment of cancer cells with these ASOs resulted in a strong and specific reduction of tumorigenic p73 transcripts and proteins, importantly, without abolishing the wild-type p73 tumor suppressor form as observed with p73-shRNA. The specific antisense oligonucleotides rescued cells from apoptosis inhibition due to overexpression of their corresponding amino-truncated p73 isoform and decreased tumor cell proliferation. Furthermore, ASO-116 against DeltaEx2/3 coupled to magnetic nanobead polyethyleneimine (MNB/PEI) carriers significantly inhibited malignant melanoma growth, which correlated with a shift in the balance between endogenous TAp73 and DeltaEx2/3 towards apoptotic full-length p73.

CONCLUSION

Our study demonstrates the successful development of LNA-ASOs that selectively differentiate between the closely related p73 oncoproteins, and provide new tools to further delineate their biological properties in different human malignancies and for therapeutic cancer targeting.

Antisense applications for biological control

Although Nature's antisense approaches are clearly impressive, this Perspectives article focuses on the experimental uses of antisense reagents (ASRs) for control of biological processes. ASRs comprise antisense oligonucleotides (ASOs), and their catalytically active counterparts ribozymes and DNAzymes, as well as small interfering RNAs (siRNAs). ASOs and ribozymes/DNAzymes target RNA molecules on the basis of Watson-Crick base pairing in sequence-specific manner. ASOs generally result in destruction of the target RNA by RNase-H mediated mechanisms, although they may also sterically block translation, also resulting in loss of protein production. Ribozymes and DNAzymes cleave target RNAs after base pairing via their antisense flanking arms. siRNAs, which contain both sense and antisense regions from a target RNA, can mediate target RNA destruction via RNAi and the RISC, although they can also function at the transcriptional level. A considerable number of ASRs (mostly ASOs) have progressed into clinical trials, although most have relatively long histories in Phase I/II settings. Clinical trial results are surprisingly difficult to find, although few ASRs appear to have yet established efficacy in Phase III levels. Evolution of ASRs has included: (a) Modifications to ASOs to render them nuclease resistant, with analogous modifications to siRNAs being developed; and (b) Development of strategies to select optimal sites for targeting. Perhaps the biggest barrier to effective therapies with ASRs is the "Delivery Problem." Various liposomal vehicles have been used for systemic delivery with some success, and recent modifications appear to enhance systemic delivery, at least to liver. Various nanoparticle formulations are now being developed which may also enhance delivery. Going forward, topical applications of ASRs would seem to have the best chances for success. In summary, modifications to ASRs to enhance stability, improve targeting, and incremental improvements in delivery vehicles continue to make ASRs attractive as molecular therapeutics, but their advance toward the bedside has been agonizingly slow.

A structural interpretation of the effect of GC-content on efficiency of RNA interference

BACKGROUND

RNA interference (RNAi) mediated by small interfering RNAs (siRNAs) or short hairpin RNAs (shRNAs) has become a powerful technique for eukaryotic gene knockdown. siRNA GC-content negatively correlates with RNAi efficiency, and it is of interest to have a convincing mechanistic interpretation of this observation. We here examine this issue by considering the secondary structures for both the target messenger RNA (mRNA) and the siRNA guide strand.

RESULTS

By analyzing a unique homogeneous data set of 101 shRNAs targeted to 100 endogenous human genes, we find that: 1) target site accessibility is more important than GC-content for efficient RNAi; 2) there is an appreciable negative correlation between GC-content and RNAi activity; 3) for the predicted structure of the siRNA guide strand, there is a lack of correlation between RNAi activity and either the stability or the number of free dangling nucleotides at an end of the structure; 4) there is a high correlation between target site accessibility and GC-content. For a set of representative structural RNAs, the GC content of 62.6% for paired bases is significantly higher than the GC content of 38.7% for unpaired bases. Thus, for a structured RNA, a region with higher GC content is likely to have more stable secondary structure. Furthermore, by partial correlation analysis, the correlation for GC-content is almost completely diminished, when the effect of target accessibility is controlled.

CONCLUSION

These findings provide a target-structure-based interpretation and mechanistic insight for the effect of GC-content on RNAi efficiency.

Development of lead hammerhead ribozyme candidates against human rod opsin mRNA for retinal degeneration therapy

To identify lead candidate allele-independent hammerhead ribozymes (hhRz) for the treatment of autosomal dominant mutations in the human rod opsin (RHO) gene, we tested a series of hhRzs for potential to significantly knockdown human RHO gene expression in a human cell expression system. Multiple computational criteria were used to select target mRNA regions likely to be single stranded and accessible to hhRz annealing and cleavage. Target regions are tested for accessibility in a human cell culture expression system where the hhRz RNA and target mRNA and protein are coexpressed. The hhRz RNA is embedded in an adenoviral VAI RNA chimeric RNA of established structure and properties which are critical to the experimental paradigm. The chimeric hhRz-VAI RNA is abundantly transcribed so that the hhRzs are expected to be in great excess over substrate mRNA. HhRz-VAI traffics predominantly to the cytoplasm to colocalize with the RHO mRNA target. Colocalization is essential for second-order annealing reactions. The VAI chimera protects the hhRz RNA from degradation and provides for a long half-life. With cell lines chosen for high transfection efficiency and a molar excess of hhRz plasmid over target plasmid, the conditions of this experimental paradigm are specifically designed to evaluate for regions of accessibility of the target mRNA in cellulo. Western analysis was used to measure the impact of hhRz expression on RHO protein expression. Three lead candidate hhRz designs were identified that significantly knockdown target protein expression relative to control (p<0.05). Successful lead candidates (hhRz CUC [see in text downward arrow] 266, hhRz CUC [see in text downward arrow] 1411, hhRz AUA [see in text downward arrow] 1414) targeted regions of human RHO mRNA that were predicted to be accessible by a bioinformatics approach, whereas regions predicted to be inaccessible supported no knockdown. The maximum opsin protein level knockdown is approximately 30% over a 48h paradigm of testing. These results validate a rigorous computational bioinformatics approach to detect accessible regions of target mRNAs in cellulo. The opsin knockdown effect could prove to be clinically significant when integrated over longer periods in photoreceptors. Further optimization and animal testing are the next step in this stratified RNA drug discovery program. A recently developed novel and efficient screening assay based upon expression of a dicistronic mRNA (RHO-IRES-SEAP) containing both RHO and reporter (SEAP) cDNAs was used to compare the hhRz 266 lead candidate to another agent (Rz525/hhRz485) already known to partially rescue retinal degeneration in a rodent model. Lead hhRz 266 CUC [see in text downward arrow] proved more efficacious than Rz525/hhRz485 which infers viability for rescue of retinal degeneration in appropriate preclinical models of disease.

Ribozyme cleavage of Plasmodium falciparum gyrase A gene transcript affects the parasite growth

Deoxyribonucleic acid (DNA) gyrase is an important enzyme that facilitates the movement of replication and transcription complexes through DNA by creating negative supercoils ahead of the complex. Its presence in Plasmodium falciparum is now established and considered a good drug target since it is absent in the human host. The sequence of P. falciparum gyrase A subunit was analyzed for its messenger ribonucleic acid (mRNA) folding as well as target accessibility for ribozymes. The four GUC triplet sites identified at 334, 491, 1907, and 2642 nucleotide positions of the Gyrase A mRNA were also accessible to oligos by RNase H assay. Site GUC491 was optimally accessible followed by GUC1907, GUC334, and GUC2642 sites. Ribozymes were produced against all these sites and tested for their in vitro transcript cleavage potentials where RZ491 showed the maximum cleavage rate. Therefore, this ribozyme (RZ491) was chemically synthesized albeit with modifications so as to make it resistant against ribonuclease attack. The modified ribozyme retained its cleavage potential and was able to inhibit the P. falciparum parasite growth up to 49.54% and 74.77% at 20 and 30 microM ribozyme concentrations, respectively, as compared to the untreated culture. However, up to 20% and 24.32% parasite growth inhibition was observed at the same ribozyme concentrations of 20 and 30 microM when compared with control ribozyme-treated cultures. This ribozyme as well as other targets identified here can be investigated further to develop the effective chemotherapeutic agents against malaria.

The impact of target site accessibility on the design of effective siRNAs

Small-interfering RNAs (siRNAs) assemble into RISC, the RNA-induced silencing complex, which cleaves complementary mRNAs. Despite their fluctuating efficacy, siRNAs are widely used to assess gene function. Although this limitation could be ascribed, in part, to variations in the assembly and activation of RISC, downstream events in the RNA interference (RNAi) pathway, such as target site accessibility, have so far not been investigated extensively. In this study we present a comprehensive analysis of target RNA structure effects on RNAi by computing the accessibility of the target site for interaction with the siRNA. Based on our observations, we developed a novel siRNA design tool, RNAxs, by combining known siRNA functionality criteria with target site accessibility. We calibrated our method on two data sets comprising 573 siRNAs for 38 genes, and tested it on an independent set of 360 siRNAs targeting four additional genes. Overall, RNAxs proves to be a robust siRNA selection tool that substantially improves the prediction of highly efficient siRNAs.

UNAFold: software for nucleic acid folding and hybridization

The UNAFold software package is an integrated collection of programs that simulate folding, hybridization, and melting pathways for one or two single-stranded nucleic acid sequences. The name is derived from "Unified Nucleic Acid Folding." Folding (secondary structure) prediction for single-stranded RNA or DNA combines free energy minimization, partition function calculations and stochastic sampling. For melting simulations, the package computes entire melting profiles, not just melting temperatures. UV absorbance at 260 nm, heat capacity change (C(p)), and mole fractions of different molecular species are computed as a function of temperature. The package installs and runs on all Unix and Linux platforms that we have looked at, including Mac OS X. Images of secondary structures, hybridizations, and dot plots may be computed using common formats. Similarly, a variety of melting profile plots is created when appropriate. These latter plots include experimental results if they are provided. The package is "command line" driven. Underlying compiled programs may be used individually, or in special combinations through the use of a variety of Perl scripts. Users are encouraged to create their own scripts to supplement what comes with the package. This evolving software is available for download at http://www.bioinfo.rpi.edu/applications/hybrid/download.php .

A structural analysis of in vitro catalytic activities of hammerhead ribozymes

BACKGROUND

Ribozymes are small catalytic RNAs that possess the dual functions of sequence-specific RNA recognition and site-specific cleavage. Trans-cleaving ribozymes can inhibit translation of genes at the messenger RNA (mRNA) level in both eukaryotic and prokaryotic systems and are thus useful tools for studies of gene function. However, identification of target sites for efficient cleavage poses a challenge. Here, we have considered a number of structural and thermodynamic parameters that can affect the efficiency of target cleavage, in an attempt to identify rules for the selection of functional ribozymes.

RESULTS

We employed the Sfold program for RNA secondary structure prediction, to account for the likely population of target structures that co-exist in dynamic equilibrium for a specific mRNA molecule. We designed and prepared 15 hammerhead ribozymes to target GUC cleavage sites in the mRNA of the breast cancer resistance protein (BCRP). These ribozymes were tested, and their catalytic activities were measured in vitro. We found that target disruption energy owing to the alteration of the local target structure necessary for ribozyme binding, and the total energy change of the ribozyme-target hybridization, are two significant parameters for prediction of ribozyme activity. Importantly, target disruption energy is the major contributor to the predictability of ribozyme activity by the total energy change. Furthermore, for a target-site specific ribozyme, incorrect folding of the catalytic core, or interactions involving the two binding arms and the end sequences of the catalytic core, can have detrimental effects on ribozyme activity.

CONCLUSION

The findings from this study suggest rules for structure-based rational design of trans-cleaving hammerhead ribozymes in gene knockdown studies. Tools implementing these rules are available from the Sribo module and the Srna module of the Sfold program available through Web server at http://sfold.wadsworth.org.

Insights into effective RNAi gained from large-scale siRNA validation screening

Transfection of chemically synthesized short interfering RNAs (siRNAs) enables a high level of sequence-specific gene silencing. Although siRNA design algorithms have been improved in recent years, it is still necessary to prove the functionality of a given siRNA experimentally. We have functionally tested several thousand siRNAs for target genes from various gene families including kinases, phosphatases, and cancer-related genes (e.g., genes involved in apoptosis and the cell cycle). Some targets were difficult to silence above a threshold of 70% knockdown. By working with one design algorithm and a standardized validation procedure, we discovered that the level of silencing achieved was not exclusively dependent on the siRNA sequences. Here we present data showing that neither the gene expression level nor the cellular environment has a direct impact on the knockdown which can be achieved for a given target. Modifications of the experimental setting have been investigated with the aim of improving knockdown efficiencies for siRNA-target combinations that show only moderate knockdown. Use of higher siRNA concentrations did not change the overall performance of the siRNA-target combinations analyzed. Optimal knockdown at the mRNA level was usually reached 48-72 hours after transfection. Target gene-specific characteristics such as the accessibility of the corresponding target sequences to the RNAi machinery appear to have a significant influence on the knockdown observed, making certain targets easy or difficult to knock down using siRNA.

Inhibition of Plasmodium falciparum ispH (lytB) gene expression by hammerhead ribozyme

The nonmevalonate pathway of isoprenoid biosynthesis in the apicoplast of the human malaria parasite Plasmodium falciparum is distinct from the mevalonate-dependent pathway of humans and thus a good drug target. We describe here the hammerhead ribozyme based cleavage of the ispH (lytB) gene transcript involved in the last step of this nonmevalonate pathway. Using RNA folding program, three hammerhead ribozymes named as RZ(876), RZ(1260), and RZ(1331) were predicted against ispH (lytB) mRNA. Messenger walk screening (RNaseH) assay confirmed the target accessibility for these ribozymes. All three ribozymes cleaved the target RNA in vitro but RZ(876) exhibited the highest catalytic potential (62.92%). Therefore, RZ(876) was chemically synthesized with appropriate chemical modifications to protect it from nuclease attack while using it for in vitro parasite growth inhibition assay. This ribozyme RZ(876) was able to inhibit 87.36% parasite growth at 30 microM concentration compared to the untreated culture. However, an absolute inhibition of 29.41% was achieved compared to the control ribozyme (RZ(ctrl)). Nonetheless, the growth inhibition effect was found to be sequence-specific as indicated by the decreased level of ispH (lytB) transcript after ribozyme treatment. In conclusion, we have identified the ispH (lytB) as a potential target whose transcript can be cleaved by a ribozyme RZ(876).

Effect of target secondary structure on RNAi efficiency

RNA interference (RNAi) mediated by small interfering RNAs (siRNAs) or short hairpin RNAs (shRNAs) has become a powerful tool for gene knockdown studies. However, the levels of knockdown vary greatly. Here, we examine the effect of target disruption energy, a novel measure of target accessibility, along with other parameters that may affect RNAi efficiency. Based on target secondary structures predicted by the Sfold program, the target disruption energy represents the free energy cost for local alteration of the target structure to allow target binding by the siRNA guide strand. In analyses of 100 siRNAs and 101 shRNAs targeted to 103 endogenous human genes, we find that the disruption energy is an important determinant of RNAi activity and the asymmetry of siRNA duplex asymmetry is important for facilitating the assembly of the RNA-induced silencing complex (RISC). We estimate that target accessibility and duplex asymmetry can improve the target knockdown level significantly by nearly 40% and 26%, respectively. In the RNAi pathway, RISC assembly precedes target binding by the siRNA guide strand. Thus, our findings suggest that duplex asymmetry has significant upstream effect on RISC assembly and target accessibility has strong downstream effect on target recognition. The results of the analyses suggest criteria for improving the design of siRNAs and shRNAs.

Antisense oligonucleotides: from design to therapeutic application

1. An antisense oligonucleotide (ASO) is a short strand of deoxyribonucleotide analogue that hybridizes with the complementary mRNA in a sequence-specific manner via Watson-Crick base pairing. Formation of the ASO-mRNA heteroduplex either triggers RNase H activity, leading to mRNA degradation, induces translational arrest by steric hindrance of ribosomal activity, interferes with mRNA maturation by inhibiting splicing or destabilizes pre-mRNA in the nucleus, resulting in downregulation of target protein expression. 2. The ASO is not only a useful experimental tool in protein target identification and validation, but also a highly selective therapeutic strategy for diseases with dysregulated protein expression. 3. In the present review, we discuss various theoretical approaches to rational design of ASO, chemical modifications of ASO, ASO delivery systems and ASO-related toxicology. Finally, we survey ASO drugs in various current clinical studies.

Inhibition of bacterial translation and growth by peptide nucleic acids targeted to domain II of 23S rRNA

The objective of this work was to study the inhibitory effects of antisense peptide nucleic acids (PNAs) targeted to domain II of 23S rRNA on bacterial translation and growth. In this paper, we report that PNA(G1138) or peptide-PNA(G1138) targeted to domain II of 23S rRNA can inhibit both translation in vitro (in a cell-free translation system) and bacterial growth in vivo. The inhibitory concentration (IC50) and the minimum inhibiting concentration (MIC) are 0.15 and 10 microM, respectively. The inhibition effect of PNA(G1138) in vitro is somewhat lower than that of tetracycline (IC50 = 0.12 microM), but the MIC of peptide-PNA(G1138) against Escherichia coli is significantly higher than that of tetracycline (MIC = 4 microM). Further studies based on similar colony-forming unit (CFU) assays showed that peptide-PNA(G1138) at 10 microM is bactericidal, but the bactericidal effect is less effective than that of tetracycline. Nevertheless, the results demonstrated that the peptide-PNA(G1138) treatment is bactericidal in a dose- and sequence-dependent manner and that the G1138 site of 23S rRNA is a possible sequence target for designing novel PNA-based antibiotics.

siRNA selection criteria--statistical analyses of applicability and significance

RNA interference is a powerful tool for gene silencing, which is mediated by introducing siRNA. In the present study, statistical analyses of published siRNA selection criteria, the interpretation of some criteria and systematic searching for new criteria have been carried out for CGB siRNA and siRecords databases. The results of the analyses are as follows: (i) Our study supports the two-state model of the RNA-induced silencing complex (RISC). (ii) Stable 5'-S ends of a siRNA sequence, higher stability of the whole siRNA, and low breaking energy of siRNA duplex occurs in effective siRNA sequences. Also low internal stability of the 5'-AS terminus is preferred. (iii) Secondary structure can be successfully used as an RNAi selection criterion. (iv) Several published sequence criteria have been confirmed and also new criteria have been developed. (v) Also a Target Patterns criterion, which is comparable or better than the best known criteria, has been created.

Potent effect of target structure on microRNA function

MicroRNAs (miRNAs) are small noncoding RNAs that repress protein synthesis by binding to target messenger RNAs. We investigated the effect of target secondary structure on the efficacy of repression by miRNAs. Using structures predicted by the Sfold program, we model the interaction between an miRNA and a target as a two-step hybridization reaction: nucleation at an accessible target site followed by hybrid elongation to disrupt local target secondary structure and form the complete miRNA-target duplex. This model accurately accounts for the sensitivity to repression by let-7 of various mutant forms of the Caenorhabditis elegans lin-41 3' untranslated region and for other experimentally tested miRNA-target interactions in C. elegans and Drosophila melanogaster. These findings indicate a potent effect of target structure on target recognition by miRNAs and establish a structure-based framework for genome-wide identification of animal miRNA targets.

More complete gene silencing by fewer siRNAs: transparent optimized design and biophysical signature

Highly accurate knockdown functional analyses based on RNA interference (RNAi) require the possible most complete hydrolysis of the targeted mRNA while avoiding the degradation of untargeted genes (off-target effects). This in turn requires significant improvements to target selection for two reasons. First, the average silencing activity of randomly selected siRNAs is as low as 62%. Second, applying more than five different siRNAs may lead to saturation of the RNA-induced silencing complex (RISC) and to the degradation of untargeted genes. Therefore, selecting a small number of highly active siRNAs is critical for maximizing knockdown and minimizing off-target effects. To satisfy these needs, a publicly available and transparent machine learning tool is presented that ranks all possible siRNAs for each targeted gene. Support vector machines (SVMs) with polynomial kernels and constrained optimization models select and utilize the most predictive effective combinations from 572 sequence, thermodynamic, accessibility and self-hairpin features over 2200 published siRNAs. This tool reaches an accuracy of 92.3% in cross-validation experiments. We fully present the underlying biophysical signature that involves free energy, accessibility and dinucleotide characteristics. We show that while complete silencing is possible at certain structured target sites, accessibility information improves the prediction of the 90% active siRNA target sites. Fast siRNA activity predictions can be performed on our web server at http://optirna.unl.edu/.

Rational design and rapid screening of antisense oligonucleotides for prokaryotic gene modulation

Antisense oligodeoxynucleotides (oligos) are widely used for functional studies of both prokaryotic and eukaryotic genes. However, the identification of effective target sites is a major issue in antisense applications. Here, we study a number of thermodynamic and structural parameters that may affect the potency of antisense inhibition. We develop a cell-free assay for rapid oligo screening. This assay is used for measuring the expression of Escherichia coli lacZ, the antisense target for experimental testing and validation. Based on a training set of 18 oligos, we found that structural accessibility predicted by local folding of the target mRNA is the most important predictor for antisense activity. This finding was further confirmed by a direct validation study. In this study, a set of 10 oligos was designed to target accessible sites, and another set of 10 oligos was selected to target inaccessible sites. Seven of the 10 oligos for accessible sites were found to be effective (>50% inhibition), but none of the oligos for inaccessible sites was effective. The difference in the antisense activity between the two sets of oligos was statistically significant. We also found that the predictability of antisense activity by target accessibility was greatly improved for oligos targeted to the regions upstream of the end of the active domain for β-galactosidase, the protein encoded by lacZ. The combination of the structure-based antisense design and extension of the lacZ assay to include gene fusions will be applicable to high-throughput gene functional screening, and to the identification of new drug targets in pathogenic microbes. Design tools are available through the Sfold Web server at .

OligoMatcher: analysis and selection of specific oligonucleotide sequences for gene silencing by antisense or siRNA

OligoMatcher is a web-based tool for analysis and selection of unique oligonucleotide sequences for gene silencing by antisense oligonucleotides (ASOs) or small interfering RNA (siRNA). A specific BLAST server was built for analysing sequences of ASOs that target pre-mRNA in the cell nucleus. Tissue- and cell-specific expression data of potential cross-reactive genes are integrated in the OligoMatcher program, which allows biologists to select unique oligonucleotide sequences for their target genes in specific experimental systems.

AVAILABILITY

The OligoMatcher web server is available at http://shelob.cs.iupui.edu:18081/oligomatch.php. The source code is freely available for non-profit use on request to the authors.

CONTACT

Mathew Palakal (mpalakal@cs.iupui.edu) or Shuyu Li (li_shuyu_dan@lilly.com).

Partition function and base pairing probabilities of RNA heterodimers

BACKGROUND

RNA has been recognized as a key player in cellular regulation in recent years. In many cases, non-coding RNAs exert their function by binding to other nucleic acids, as in the case of microRNAs and snoRNAs. The specificity of these interactions derives from the stability of inter-molecular base pairing. The accurate computational treatment of RNA-RNA binding therefore lies at the heart of target prediction algorithms.

METHODS

The standard dynamic programming algorithms for computing secondary structures of linear single-stranded RNA molecules are extended to the co-folding of two interacting RNAs.

RESULTS

We present a program, RNAcofold, that computes the hybridization energy and base pairing pattern of a pair of interacting RNA molecules. In contrast to earlier approaches, complex internal structures in both RNAs are fully taken into account. RNAcofold supports the calculation of the minimum energy structure and of a complete set of suboptimal structures in an energy band above the ground state. Furthermore, it provides an extension of McCaskill's partition function algorithm to compute base pairing probabilities, realistic interaction energies, and equilibrium concentrations of duplex structures.

Selection of antisense oligonucleotides based on multiple predicted target mRNA structures

Background

Local structures of target mRNAs play a significant role in determining the efficacies of antisense oligonucleotides (ODNs), but some structure-based target site selection methods are limited by uncertainties in RNA secondary structure prediction. If all the predicted structures of a given mRNA within a certain energy limit could be used simultaneously, target site selection would obviously be improved in both reliability and efficiency. In this study, some key problems in ODN target selection on the basis of multiple predicted target mRNA structures are systematically discussed.

Results

Two methods were considered for merging topologically different RNA structures into integrated representations. Several parameters were derived to characterize local target site structures. Statistical analysis on a dataset with 448 ODNs against 28 different mRNAs revealed 9 features quantitatively associated with efficacy. Features of structural consistency seemed to be more highly correlated with efficacy than indices of the proportion of bases in single-stranded or double-stranded regions. The local structures of the target site 5' and 3' termini were also shown to be important in target selection. Neural network efficacy predictors using these features, defined on integrated structures as inputs, performed well in "minus-one-gene" cross-validation experiments.

Conclusion

Topologically different target mRNA structures can be merged into integrated representations and then used in computer-aided ODN design. The results of this paper imply that some features characterizing multiple predicted target site structures can be used to predict ODN efficacy.

Adenovirus-Delivered Antisense RNA and shRNA Exhibit Different Silencing Efficiencies for the Endogenous Transforming Growth Factor-β(TGF-β) Type II Receptor

Gene silencing is an essential tool in gene discovery and gene therapy. Traditionally, viral delivery of antisense RNA and, more recently, small interfering RNA (siRNA) molecules in the form of small hairpin RNAs (shRNA) has been used as a strategy to achieve gene silencing. Nevertheless, the enduring challenge is to identify molecules that specifically and optimally silence a given target gene. In this study, we tested a set of adenovirus-delivered antisense RNA fragments and adenovirus-delivered shRNA molecules for their ability to target human transforming growth factor-beta type II receptor (TGFbetaRII). We used a dicistronic reporter, consisting of the coding sequences for TGFbetaRII and green fluorescent protein (GFP) to screen for optimal silencing agents targeting TGFbetaRII. Our results show, for both antisense RNA and shRNA molecules, that their effectiveness in the GFP screen correlated directly with their ability to reduce exogenously expressed TGFbetaRII. Unexpectedly, the antisense RNAs were unable to silence endogenous TGFbetaRII. In contrast, the shRNAs were able to silence endogenous TGFbetaRII. The shRNA that demonstrated the most pronounced effect on the dicistronic TGFbetaRII/GFP reporter reduced endogenous TGFbetaRII protein expression by 70% in A549 cells and reduced TGFbeta signaling by >80% in HeLa cells.

Statistical and Bayesian approaches to RNA secondary structure prediction

Prediction of RNA secondary structure is a fundamental problem in computational structural biology. For several decades, free energy minimization has been the most popular method for prediction from a single sequence. In recent years, the McCaskill algorithm for computation of partition function and base-pair probabilities has become increasingly appreciated. This paradigm-shifting work has inspired the developments of extended partition function algorithms, statistical sampling and clustering, and application of Bayesian statistical inference. The performance of thermodynamics-based methods is limited by thermodynamic rules and parameters. However, further improvements may come from statistical estimates derived from structural databases for thermodynamics parameters with weak or little experimental data. The Bayesian inference approach appears to be promising in this context.

AOBase: a database for antisense oligonucleotides selection and design

Antisense oligonucleotides (ODNs) technology is one of the important approaches for the sequence-specific knockdown of gene expression. ODNs have been used as research tools in the post-genome era, as well as new types of therapeutic agents. Since finding effective target sites within RNA is a hard work for antisense ODNs design, various experimental methods and computational approaches have been proposed. For better sharing of the experimented and published ODNs, valid and invalid ODNs reported in literatures are screened, collected and stored in AOBase. Till now, ∼700 ODNs against 46 target mRNAs are contained in AOBase. Entries can be explored via TargetSearch and AOSearch web retrieval interfaces. AOBase can not only be useful in ODNs selection for gene function exploration, but also contribute to mining rules and developing algorithms for rational ODNs design. AOBase is freely accessible via .

Targeted gene suppression by RNA interference: an efficient method for production of high-amylose potato lines

Production of high-amylose potato lines can be achieved by inhibition of two genes coding for starch branching enzymes. The use of antisense technology for gene inhibition have yielded a low frequency of high-amylose lines that mostly was correlated with high numbers of integrated T-DNA copies. To investigate whether the production of high-amylose lines could be improved, RNA interference was used for gene inhibition of the genes Sbe1 and Sbe2. Two constructs with 100 bp segments (pHAS2) or 200 bp segments (pHAS3) of both branching enzyme genes were cloned as inverted repeats controlled by a potato granule-bound starch synthase promoter. The construct pHAS3 was shown to be very efficient, yielding high-amylose quality in more than 50% of the transgenic lines. An antisense construct, included in the study as a comparator, resulted in only 3% of the transgenic lines being of high-amylose type. Noticeable was also that pHAS3 yielded low T-DNA copy inserts with an average of 83% of backbone-free transgenic lines being single copy events.

Revolutions in RNA secondary structure prediction

RNA structure formation is hierarchical and, therefore, secondary structure, the sum of canonical base-pairs, can generally be predicted without knowledge of the three-dimensional structure. Secondary structure prediction algorithms evolved from predicting a single, lowest free energy structure to their current state where statistics can be determined from the thermodynamic ensemble. This article reviews the free energy minimization technique and the salient revolutions in the dynamic programming algorithm methods for secondary structure prediction. Emphasis is placed on highlighting the recently developed method, which statistically samples structures from the complete Boltzmann ensemble.

Energy and structural analysis of double nucleic acid triplets

In order to investigate the energy and structural character of RNA-RNA triplets and RNA-DNA duplex base triplets, 64 sets of three-dimensional models of RNA-DNA duplex base triplets and mRNA-tRNA triplex base triplets were constructed and optimized by homologous modeling method using the software InsightII. The comparative statistical method and cluster analysis were adopted to study these features. The result showed: (i) all energy parameters of monomer RNA-DNA hybrid triplets and ternary complexes appeared significantly different; and some parameters related with overall molecules such as overall energy, bond energy and coulomb energy have statistically significant correlations between the structures in vacuum and aquatic solutions while other parameters, including theta energy, phi energy, hydrogen bond energy and non-bond energy, changed significantly, but not continuously. (ii) However, the case of mRNA-tRNA triplets was much more complicated in that only the bond energy's correlation coefficient is -0.8. Typically, the main contribution of GC pairs and G/A/U bases were interesting. The models of RNA-DNA hybrid triplets and mRNA-tRNA triplet should be helpful for the study of base pairing in codons and the biological effectiveness of antisense nucleic acids.

A potent inhibitor of prothrombin gene expression as a result of standardized target site selection and design of antisense oligonucleotides

The development of antisense oligonucleotides (AS-ODN) always had the limitation that because of complex mRNA secondary structures, not every designed AS-ODN inhibited the expression of its target. There have been many investigations to overcome this problem in the last few years. This produced a great deal of theoretical and empirical findings about characteristics of effective AS-ODNs in respect to their target regions but no standardized selection procedure of AS-ODN target regions within a given mRNA or standardized design of AS-ODNs against a specific target region. We present here a standardized method based on secondary structure prediction for target site selection and AS-ODN design, followed by validation of the antisense effect caused by our predicted AS-ODNs in cell culture. The combination of theoretical design and experimental selection procedure led to an AS-ODN that efficiently and specifically reduces prothrombin mRNA and antigen.

A secondary structural common core in the ribosomal ITS2 (internal transcribed spacer) of Culexspecies from diverse geographical locations

In the present study, sequence and structural analysis of ITS2 region (the spacer segment between 5.8S and 28S rRNA of mature rRNA sequences) of 7 Culex species belonging to 5 different geographical locations was carried out. Alignment of the ITS2 sequence from the 7 species revealed 8 homologous domains. Four species namely C. vishnui, C. annulus, C. pipiens, C. quiquefasciatusshowed high sequence (98-100%) and RNA secondary structure similarity. The ITS2 similarity among different species is high despite their varying geographical locations. Several common features of secondary structure are shared among these species, with some of them supported by compensatory changes, suggesting the significant role by ITS2 as an RNA domain during ribosome biogenesis.

Computational identification of antisense oligonucleotides that rapidly hybridize to RNA

The ability of a computational model to determine the relative rate of hybridization between anti-sense oligonucleotides and RNA was tested using HIV-1 tat mRNA. The model, which was based on the assumptions that hybridization is a second-order reaction and that early in the hybridization reaction the concentrations of intermediates are approximately constant (steady-state), allows calculation of a rate factor that is proportional to the reaction constant. Formation of oligodeoxynucleotide (ODN)-RNA hybrid, detected by RNase H-dependent cleavage, increased nearly linearly during an initial incubation period, consistent with the steady-state approximation. The initial hybridization rate increased linearly with substrate RNA concentration and with ODN concentration, indicating a second-order reaction. The logarithm of the second-order reaction constant, determined from the initial rate for hybridization between tat mRNA and 16 ODNs targeted to various sites, was linearly related to the logarithm of the calculated rate factor (r = 0.83, p < 0.001). Thus, the rate factor can be used to identify rapidly hybridizing antisense sequences using target nucleotide sequence information.

RNA secondary structure prediction by centroids in a Boltzmann weighted ensemble

Prediction of RNA secondary structure by free energy minimization has been the standard for over two decades. Here we describe a novel method that forsakes this paradigm for predictions based on Boltzmann-weighted structure ensemble. We introduce the notion of a centroid structure as a representative for a set of structures and describe a procedure for its identification. In comparison with the minimum free energy (MFE) structure using diverse types of structural RNAs, the centroid of the ensemble makes 30.0% fewer prediction errors as measured by the positive predictive value (PPV) with marginally improved sensitivity. The Boltzmann ensemble can be separated into a small number (3.2 on average) of clusters. Among the centroids of these clusters, the "best cluster centroid" as determined by comparison to the known structure simultaneously improves PPV by 46.5% and sensitivity by 21.7%. For 58% of the studied sequences for which the MFE structure is outside the cluster containing the best centroid, the improvements by the best centroid are 62.5% for PPV and 31.4% for sensitivity. These results suggest that the energy well containing the MFE structure under the current incomplete energy model is often different from the one for the unavailable complete model that presumably contains the unique native structure. Centroids are available on the Sfold server at http://sfold.wadsworth.org.

Genome-wide selection of unique and valid oligonucleotides

Functional genomics methods are used to investigate the huge amount of information contained in genomes. Numerous experimental methods rely on the use of oligo- or polynucleotides. Nucleotide strand hybridization forms the underlying principle for these methods. For all these techniques, the probes should be unique for analyzed genes. In addition to being unique for the studied genes, the probes should fulfill a large number of criteria to be usable and valid. The criteria include for example, avoidance of self-annealing, suitable melting temperature and nucleotide composition. We developed a method for searching unique and valid oligonucleotides or probes for genes so that there is not even a similar (approximate) occurrence in any other location of the whole genome. By using probe size 25, we analyzed 17 complete genomes representing a wide range of both prokaryotic and eukaryotic organisms. More than 92% of all the genes in the investigated genomes contained valid oligonucleotides. Extensive statistical tests were performed to characterize the properties of unique and valid oligonucleotides. Unique and valid oligonucleotides were relatively evenly distributed in genes except for the beginning and end, which were somewhat overrepresented. The flanking regions in eukaryotes were clearly underrepresented among suitable oligonucleotides. In addition to distributions within genes, the effects on codon and amino acid usage were also studied.

Advances in the development of therapeutic nucleic acids against cervical cancer

Cervical cancer is the second most common neoplastic disease affecting women worldwide. Basic, clinical and epidemiological analyses indicate that expression of high-risk human papillomaviruses (HPVs) E6/E7 genes is the primary cause of cervical cancer and represent ideal targets for the application of therapeutic nucleic acids (TNAs). Antisense oligodeoxyribonucleotides (AS-ODNs) and ribozymes (RZs) are the most effective TNAs able to inhibit in vivo tumour growth by eliminating HPV-16 and HPV-18 E6/E7 transcripts. Expression of multiple RZs directed against alternative target sites by triplex expression systems may result in the abrogation of highly variable HPVs. More recently, RNA interference (RNAi) gene knockdown phenomenon, induced by small interfering RNA (siRNA), has demonstrated its potential value as an effective TNA for cervical cancer. siRNA and aptamers as TNAs will have a place in the armament for cervical cancer. TNAs against cervical cancer is in a dynamic state, and clinical trials will define the TNAs in preventive and therapeutic roles to control tumour growth, debulk tumour mass, prevent metastasis and facilitate immune interaction.

An efficient algorithm to compute the landscape of locally optimal RNA secondary structures with respect to the Nussinov-Jacobson energy model

We make a novel contribution to the theory of biopolymer folding, by developing an efficient algorithm to compute the number of locally optimal secondary structures of an RNA molecule, with respect to the Nussinov-Jacobson energy model. Additionally, we apply our algorithm to analyze the folding landscape of selenocysteine insertion sequence (SECIS) elements from A. Bock (personal communication), hammerhead ribozymes from Rfam (Griffiths-Jones et al., 2003), and tRNAs from Sprinzl's database (Sprinzl et al., 1998). It had previously been reported that tRNA has lower minimum free energy than random RNA of the same compositional frequency (Clote et al., 2003; Rivas and Eddy, 2000), although the situation is less clear for mRNA (Seffens and Digby, 1999; Workman and Krogh, 1999; Cohen and Skienna, 2002),(1) which plays no structural role. Applications of our algorithm extend knowledge of the energy landscape differences between naturally occurring and random RNA. Given an RNA molecule a(1), ... , a(n) and an integer k > or = 0, a k-locally optimal secondary structure S is a secondary structure on a(1), ... , a(n) which has k fewer base pairs than the maximum possible number, yet for which no basepairs can be added without violation of the definition of secondary structure (e.g., introducing a pseudoknot). Despite the fact that the number numStr(k) of k-locally optimal structures for a given RNA molecule in general is exponential in n, we present an algorithm running in time O(n (4)) and space O(n (3)), which computes numStr(k) for each k. Structurally important RNA, such as SECIS elements, hammerhead ribozymes, and tRNA, all have a markedly smaller number of k-locally optimal structures than that of random RNA of the same dinucleotide frequency, for small and moderate values of k. This suggests a potential future role of our algorithm as a tool to detect noncoding RNA genes.

Effective intracellular delivery of oligonucleotides in order to make sense of antisense

For more than two decades, antisense oligonucleotides (ODNs) have been used to modulate gene expression for the purpose of applications in cell biology and for development of novel sophisticated medical therapeutics. Conceptually, the antisense approach represents an elegant strategy, involving the targeting to and association of an ODN sequence with a specific mRNA via base-pairing, resulting in an impairment of functional and/or harmful protein expression in normal and diseased cells/tissue, respectively. Apart from ODN stability, its efficiency very much depends on intracellular delivery and release/access to the target side, issues that are still relatively poorly understood. Since free ODNs enter cells relatively poorly, appropriate carriers, often composed of polymers and cationic lipids, have been developed. Such carriers allow efficient delivery of ODNs into cells in vitro, and the mechanisms of delivery, both in terms of biophysical requirements for the carrier and cell biological features of uptake, are gradually becoming apparent. To become effective, ODNs require delivery into the nucleus, which necessitates release of internalized ODNs from endosomal compartments, an event that seems to depend on the nature of the delivery vehicle and distinct structural shape changes. Interestingly, evidence is accumulating which suggests that by modulating the surface properties of the carrier, the kinetics of such changes can be controlled, thus providing possibilities for programmable release of the carrier contents. Here, consideration will also be given to antisense design and chemistry, and the challenge of extra- and intracellular barriers to be overcome in the delivery process.

Secondary structure in the target as a confounding factor in synthetic oligomer microarray design

Background

Secondary structure in the target is a property not usually considered in software applications for design of optimal custom oligonucleotide probes. It is frequently assumed that eliminating self-complementarity, or screening for secondary structure in the probe, is sufficient to avoid interference with hybridization by stable secondary structures in the probe binding site. Prediction and thermodynamic analysis of secondary structure formation in a genome-wide set of transcripts from Brucella suis 1330 demonstrates that the properties of the target molecule have the potential to strongly influence the rate and extent of hybridization between transcript and tethered oligonucleotide probe in a microarray experiment.

Results

Despite the relatively high hybridization temperatures and 1M monovalent salt imposed in the modeling process to approximate hybridization conditions used in the laboratory, we find that parts of the target molecules are likely to be inaccessible to intermolecular hybridization due to the formation of stable intramolecular secondary structure. For example, at 65°C, 28 ± 7% of the average cDNA target sequence is predicted to be inaccessible to hybridization. We also analyzed the specific binding sites of a set of 70mer probes previously designed for Brucella using a freely available oligo design software package. 21 ± 13% of the nucleotides in each probe binding site are within a double-stranded structure in over half of the folds predicted for the cDNA target at 65°C. The intramolecular structures formed are more stable and extensive when an RNA target is modeled rather than cDNA. When random shearing of the target is modeled for fragments of 200, 100 and 50 nt, an overall destabilization of secondary structure is predicted, but shearing does not eliminate secondary structure.

Conclusion

Secondary structure in the target is pervasive, and a significant fraction of the target is found in double stranded conformations even at high temperature. Stable structure in the target has the potential to interfere with hybridization and should be a factor in interpretation of microarray results, as well as an explicit criterion in array design. Inclusion of this property in an oligonucleotide design procedure would change the definition of an optimal oligonucleotide significantly.

Mechanistic insights aid computational short interfering RNA design

RNA interference is widely recognized for its utility as a functional genomics tool. In the absence of reliable target site selection tools, however, the impact of RNA interference (RNAi) may be diminished. The primary determinants of silencing are influenced by highly coordinated RNA-protein interactions that occur throughout the RNAi process, including short interfering RNA (siRNA) binding and unwinding followed by target recognition, cleavage, and subsequent product release. Recently developed strategies for identification of functional siRNAs reveal that thermodynamic and siRNA sequence-specific properties are crucial to predict functional duplexes (Khvorova et al., 2003; Reynolds et al., 2004; Schwarz et al., 2003). Additional assessments of siRNA specificity reveal that more sophisticated sequence comparison tools are also required to minimize potential off-target effects (Jackson et al., 2003; Semizarov et al., 2003). This chapter reviews the biological basis for current computational design tools and how best to utilize and assess their predictive capabilities for selecting functional and specific siRNAs.

Profiled support vector machines for antisense oligonucleotide efficacy prediction

BACKGROUND

This paper presents the use of Support Vector Machines (SVMs) for prediction and analysis of antisense oligonucleotide (AO) efficacy. The collected database comprises 315 AO molecules including 68 features each, inducing a problem well-suited to SVMs. The task of feature selection is crucial given the presence of noisy or redundant features, and the well-known problem of the curse of dimensionality. We propose a two-stage strategy to develop an optimal model: (1) feature selection using correlation analysis, mutual information, and SVM-based recursive feature elimination (SVM-RFE), and (2) AO prediction using standard and profiled SVM formulations. A profiled SVM gives different weights to different parts of the training data to focus the training on the most important regions.

RESULTS

In the first stage, the SVM-RFE technique was most efficient and robust in the presence of low number of samples and high input space dimension. This method yielded an optimal subset of 14 representative features, which were all related to energy and sequence motifs. The second stage evaluated the performance of the predictors (overall correlation coefficient between observed and predicted efficacy, r; mean error, ME; and root-mean-square-error, RMSE) using 8-fold and minus-one-RNA cross-validation methods. The profiled SVM produced the best results (r = 0.44, ME = 0.022, and RMSE= 0.278) and predicted high (>75% inhibition of gene expression) and low efficacy (<25%) AOs with a success rate of 83.3% and 82.9%, respectively, which is better than by previous approaches. A web server for AO prediction is available online at http://aosvm.cgb.ki.se/.

CONCLUSIONS

The SVM approach is well suited to the AO prediction problem, and yields a prediction accuracy superior to previous methods. The profiled SVM was found to perform better than the standard SVM, suggesting that it could lead to improvements in other prediction problems as well.

Sfold web server for statistical folding and rational design of nucleic acids

The Sfold web server provides user-friendly access to Sfold, a recently developed nucleic acid folding software package, via the World Wide Web (WWW). The software is based on a new statistical sampling paradigm for the prediction of RNA secondary structure. One of the main objectives of this software is to offer computational tools for the rational design of RNA-targeting nucleic acids, which include small interfering RNAs (siRNAs), antisense oligonucleotides and trans-cleaving ribozymes for gene knock-down studies. The methodology for siRNA design is based on a combination of RNA target accessibility prediction, siRNA duplex thermodynamic properties and empirical design rules. Our approach to target accessibility evaluation is an original extension of the underlying RNA folding algorithm to account for the likely existence of a population of structures for the target mRNA. In addition to the application modules Sirna, Soligo and Sribo for siRNAs, antisense oligos and ribozymes, respectively, the module Srna offers comprehensive features for statistical representation of sampled structures. Detailed output in both graphical and text formats is available for all modules. The Sfold server is available at http://sfold.wadsworth.org and http://www.bioinfo.rpi.edu/applications/sfold.

Tissue-specific translation of murine branched-chain alpha-ketoacid dehydrogenase kinase mRNA is dependent upon an upstream open reading frame in the 5'-untranslated region

The committed step in the pathway for leucine, isoleucine, and valine catabolism is catalyzed by branched-chain alpha-ketoacid dehydrogenase (BCKD). This multienzyme complex is itself regulated through reversible subunit phosphorylation by a specific kinase (BCKD-kinase). Although BCKD is present in the mitochondria of all mammalian cells, BCKD-kinase has a tissue-specific pattern of expression. Various experimental, nutritional, and hormonal conditions have been used to alter the expression of BCKD-kinase, yet little is known regarding the regulation of basal BCKD-kinase expression under normal conditions including the mechanism of its tissue specificity in any organism. Here we use tissue-derived cultured cells to explore the mechanisms used to control BCKD-kinase expression. Whereas the amount of BCKD-kinase protein is significantly higher in mitochondria from C2C12 myotubes than in BNL Cl.2 liver cells, gene transcription and stability of BCKD-kinase mRNA share similar properties in these two cell types. Our results show that the amount of protein synthesized is regulated at the level of translation of BCKD-kinase mRNA and that an upstream open reading frame in the 5'-untranslated region of this transcript controls its translation. The location and putative 19-residue peptide are conserved in the mouse, rat, chimpanzee, and human genes. Likewise, gene structure of mouse, chimpanzee, and human BCKD-kinase is conserved, whereas the rat gene has lost intron 9.

Using an RNA secondary structure partition function to determine confidence in base pairs predicted by free energy minimization

A partition function calculation for RNA secondary structure is presented that uses a current set of nearest neighbor parameters for conformational free energy at 37 degrees C, including coaxial stacking. For a diverse database of RNA sequences, base pairs in the predicted minimum free energy structure that are predicted by the partition function to have high base pairing probability have a significantly higher positive predictive value for known base pairs. For example, the average positive predictive value, 65.8%, is increased to 91.0% when only base pairs with probability of 0.99 or above are considered. The quality of base pair predictions can also be increased by the addition of experimentally determined constraints, including enzymatic cleavage, flavin mono-nucleotide cleavage, and chemical modification. Predicted secondary structures can be color annotated to demonstrate pairs with high probability that are therefore well determined as compared to base pairs with lower probability of pairing.

Evaluation of several lightweight stochastic context-free grammars for RNA secondary structure prediction

Background

RNA secondary structure prediction methods based on probabilistic modeling can be developed using stochastic context-free grammars (SCFGs). Such methods can readily combine different sources of information that can be expressed probabilistically, such as an evolutionary model of comparative RNA sequence analysis and a biophysical model of structure plausibility. However, the number of free parameters in an integrated model for consensus RNA structure prediction can become untenable if the underlying SCFG design is too complex. Thus a key question is, what small, simple SCFG designs perform best for RNA secondary structure prediction?

Results

Nine different small SCFGs were implemented to explore the tradeoffs between model complexity and prediction accuracy. Each model was tested for single sequence structure prediction accuracy on a benchmark set of RNA secondary structures.

Conclusions

Four SCFG designs had prediction accuracies near the performance of current energy minimization programs. One of these designs, introduced by Knudsen and Hein in their PFOLD algorithm, has only 21 free parameters and is significantly simpler than the others.

Rational siRNA design for RNA interference

Short-interfering RNAs suppress gene expression through a highly regulated enzyme-mediated process called RNA interference (RNAi). RNAi involves multiple RNA-protein interactions characterized by four major steps: assembly of siRNA with the RNA-induced silencing complex (RISC), activation of the RISC, target recognition and target cleavage. These interactions may bias strand selection during siRNA-RISC assembly and activation, and contribute to the overall efficiency of RNAi. To identify siRNA-specific features likely to contribute to efficient processing at each step, we performed a systematic analysis of 180 siRNAs targeting the mRNA of two genes. Eight characteristics associated with siRNA functionality were identified: low G/C content, a bias towards low internal stability at the sense strand 3'-terminus, lack of inverted repeats, and sense strand base preferences (positions 3, 10, 13 and 19). Further analyses revealed that application of an algorithm incorporating all eight criteria significantly improves potent siRNA selection. This highlights the utility of rational design for selecting potent siRNAs and facilitating functional gene knockdown studies.

Aptamer database

The aptamer database is designed to contain comprehensive sequence information on aptamers and unnatural ribozymes that have been generated by in vitro selection methods. Such data are not normally collected in 'natural' sequence databases, such as GenBank. Besides serving as a storehouse of sequences that may have diagnostic or therapeutic utility, the database serves as a valuable resource for theoretical biologists who describe and explore fitness landscapes. The database is updated monthly and is publicly available at http://aptamer. icmb.utexas.edu/.

A statistical sampling algorithm for RNA secondary structure prediction

An RNA molecule, particularly a long-chain mRNA, may exist as a population of structures. Further more, multiple structures have been demonstrated to play important functional roles. Thus, a representation of the ensemble of probable structures is of interest. We present a statistical algorithm to sample rigorously and exactly from the Boltzmann ensemble of secondary structures. The forward step of the algorithm computes the equilibrium partition functions of RNA secondary structures with recent thermodynamic parameters. Using conditional probabilities computed with the partition functions in a recursive sampling process, the backward step of the algorithm quickly generates a statistically representative sample of structures. With cubic run time for the forward step, quadratic run time in the worst case for the sampling step, and quadratic storage, the algorithm is efficient for broad applicability. We demonstrate that, by classifying sampled structures, the algorithm enables a statistical delineation and representation of the Boltzmann ensemble. Applications of the algorithm show that alternative biological structures are revealed through sampling. Statistical sampling provides a means to estimate the probability of any structural motif, with or without constraints. For example, the algorithm enables probability profiling of single-stranded regions in RNA secondary structure. Probability profiling for specific loop types is also illustrated. By overlaying probability profiles, a mutual accessibility plot can be displayed for predicting RNA:RNA interactions. Boltzmann probability-weighted density of states and free energy distributions of sampled structures can be readily computed. We show that a sample of moderate size from the ensemble of an enormous number of possible structures is sufficient to guarantee statistical reproducibility in the estimates of typical sampling statistics. Our applications suggest that the sampling algorithm may be well suited to prediction of mRNA structure and target accessibility. The algorithm is applicable to the rational design of small interfering RNAs (siRNAs), antisense oligonucleotides, and trans-cleaving ribozymes in gene knock-down studies.

Thermodynamic criteria for high hit rate antisense oligonucleotide design

Antisense oligonucleotides are used for therapeutic applications and in functional genomic studies. In practice, however, many of the oligonucleotides complementary to an mRNA have little or no antisense activity. Theoretical strategies to improve the 'hit rate' in antisense screens will reduce the cost of discovery and may lead to identification of antisense oligonucleotides with increased potency. Statistical analysis performed on data collected from more than 1000 experiments with phosphorothioate-modified oligonucleotides revealed that the oligo-probes, which form stable duplexes with RNA (DeltaG(o)37 < or = -30 kcal/mol) and have small self-interaction potential, are more frequently efficient than molecules that form less stable oligonucleotide-RNA hybrids or more stable self-structures. To achieve optimal statistical preference, the values for self-interaction should be (DeltaG(o)37) > or = -8 kcal/mol for inter-oligonucleotide pairing and (DeltaG(o)37) > or = -1.1 kcal/mol for intra-molecular pairing. Selection of oligonucleotides with these thermodynamic values in the analyzed experiments would have increased the 'hit rate' by as much as 6-fold.