Rational selection and quantitative evaluation of antisense oligonucleotides

Antisense oligonucleotides are an attractive therapeutic option to modulate specific gene expression. However, not all antisense oligonucleotides are effective in inhibiting gene expression, and currently very few methods exist for selecting the few effective ones from all candidate oligonucleotides. The lack of quantitative methods to rapidly assess the efficacy of antisense oligonucleotides also contributes to the difficulty of discovering potent and specific antisense oligonucleotides. We have previously reported the development of a prediction algorithm for identifying high affinity antisense oligonucleotides based on mRNA-oligonucleotide hybridization. In this study, we report the antisense activity of these rationally selected oligonucleotides against three model target mRNAs (human lactate dehydrogenase A and B and rat gp130) in cell culture. The effectiveness of oligonucleotides was evaluated by a kinetic PCR technique, which allows quantitative evaluation of mRNA levels and thus provides a measure of antisense-mediated decreases in target mRNA, as occurs through RNase H recruitment. Antisense oligonucleotides that were predicted to have high affinity for their target proved effective in almost all cases, including tests against three different targets in two cell types with phosphodiester and phosphorothioate oligonucleotide chemistries. This approach may aid the development of antisense oligonucleotides for a variety of applications.     

Prediction of antisense oligonucleotides using structural and thermodynamic motifs

Specific gene expression regulation strategy using antisense oligonucleotides occupy significant space in recent clinical trials. The therapeutical potential of oligos lies in the identification and prediction of accurate oligonucleotides against specific target mRNA. In this work we present a computational method that is built on Artificial Neural Network (ANN) which could recognize and predict oligonucleotides effectively. In this study first we identified 11 major parameters associated with oligo:mRNA duplex linkage. A feed forward multilayer perceptron ANN classifier is trained with a set of experimentally proven feature vectors. The classifier gives an exact prediction of the input sequences under 2 classes – oligo or non-oligo. On validation, our tool showed comparatively significant accuracy of 92.48% with 91.7% sensitivity and 92.09% specificity. This study was also able to reveal the relative impact of individual parameters we considered on antisense oligonucleotide predictions.     

Computational antisense oligo prediction with a neural network model

MOTIVATION: The expression of a gene can be selectively inhibited by antisense oligonucleotides (AOs) targeting the mRNA. However, if the target site in the mRNA is picked randomly, typically 20% or less of the AOs are effective inhibitors in vivo. The sequence properties that make an AO effective are not well understood, thus many AOs need to be tested to find good inhibitors, which is time consuming and costly. So far computational models have been based exclusively on RNA structure prediction or motif searches while ignoring information from other aspects of AO design into the model. RESULTS: We present a computational model for AO prediction based on a neural network approach using a broad range of input parameters. Collecting sequence and efficacy data from AO scanning experiments in the literature generated a database of 490 AO molecules. Using a set of derived parameters based on AO sequence properties we trained a neural network model. The best model, an ensemble of 10 networks, gave an overall correlation coefficient of 0.30 (p=10(-8)). This model can predict effective AOs (>50% inhibition of gene expression) with a success rate of 92%. Using these thresholds the model predicts on average 12 effective AOs per 1000 base pairs, making it a stringent yet practical method for AO prediction.     

Poly-2'-DNP-RNAs with enhanced efficacy for inhibiting cancer cell growth

It is often believed that small interfering RNA (siRNA) is at least 10-fold more effective than the single-stranded antisense oligonucleotide for silencing the same target gene in the same cells. In view of the recent discovery that the RNA-induced silencing complex (RISC) contains only a single-stranded RNA (ssRNA) molecule and can be reconstituted using single-stranded antisense RNA, such a large difference in efficacy seems puzzling. One possible reason is that hybridization protects siRNA from hydrolysis by endogenous RNase activity until it is incorporated in the RISC, whereas ssRNA is rapidly hydrolyzed. Because the single-stranded poly-2'-O-(2,4-dinitrophenyl)-RNA (DNP-ssRNA) is both RNase resistant and membrane permeable, we synthesized homologous native siRNAs, DNP-siRNAs, native ssRNAs, and DNP-ssRNAs and made a comparative study of their efficacies for inhibiting the growth of two cancer cell lines with different overexpressed target genes under equivalent experimental conditions. It was found that the efficacy of antisense DNP-ssRNA is higher than that of the corresponding siRNA and that the efficacy of native siRNA for inhibiting cell growth can also be enhanced from 2-fold to 6-fold by replacing the native strands of RNA in siRNA with homologous DNP-RNA. Thermal denaturation data show that the hybridization affinity of the DNP-RNA/RNA duplex is higher than that of the native RNA/RNA duplex. Western blotting analysis of A549 cells treated with antisense DNP-ssRNAs containing single mismatching bases shows that the gene silencing by antisense DNP-ssRNA is as sequence specific as that by siRNA. The observed large enhancement of inhibition efficacy of native RNAs by DNP derivatization should be advantageous for both gene silencing studies and therapeutic applications.     

Comparison of the suppressive effects of antisense oligonucleotides and siRNAs directed against the same targets in mammalian cells

RNA interference appears to be a potentially powerful tool for studies of genes of unknown function. However, differences in efficacy at different target sites remain problematic when small interfering RNA (siRNA) is used as an effector. Similar problems are associated with attempts at gene inactivation using antisense oligonucleotides (ODNs) and ribozymes. We performed a comparative analysis of the suppressive effects of three knockdown methods, namely, methods based on RNA interference (RNAi), antisense ODNs, and ribozymes, using a luciferase reporter system. Dose-response experiments revealed that the IC50 value for the siRNA was about 100-fold lower than that of the antisense ODN. Our results provide useful information about the positional effects in RNAi, which might help to improve the design of effective siRNAs.     

Use of fully modified 2'-O-methyl antisense oligos for loss-of-function studies in vertebrate embryos

Antisense oligonucleotides are commonly employed to study the roles of genes in development. Although morpholino phosphorodiamidate oligonucleotides (morpholinos) are widely used to block translation or splicing of target gene products' the usefulness of other modifications in mediating RNase-H independent inhibition of gene activity in embryos has not been investigated. In this study, we investigated the extent that fully modified 2'-O-methyl oligonucleotides (2'-OMe oligos) that can function as translation inhibiting reagents in vivo, using Xenopus and zebrafish embryos. We find that oligos against Xenopus β-catenin, wnt11, and bmp4 and against zebrafish chordin (chd), which can efficiently and specifically generate embryonic loss-of-function phenotypes comparable with morpholino injection and other methods. These results show that fully modified 2'-OMe oligos can function as RNase-H independent antisense reagents in vertebrate embryos and can thus serve as an alternative modification to morpholinos in some cases.     

Antisense inhibitory effect: a comparison between 3'-partial and full phosphorothioate antisense oligonucleotides.

Phosphorothioate (PS) antisense oligonucleotides are currently used to inhibit many cell functions both in vivo and in vitro. However, these modified oligos provide reasonable sequence specificity only within a narrow concentration range. To overcome such a limitation we synthesized antisense oligomers, partially phosphorothioated, targeted against the human N-myc mRNA. We utilized such modified oligomers in a human neuroblastoma cell line where the N-myc gene expression was very high, and compared them to full phosphorothioate oligonucleotides. Both full PS and partial PS antisense oligos produced a maximum reduction in target mRNA after 6 h of treatment. They were able to maintain a good level of inhibition for 20 h only at high concentration. While partial PS oligos produced a dose dependent and sequence specific inhibition of N-myc mRNA, full PS molecules suffer from some disadvantages at the highest concentration used. Our results showed that partial PS molecules were capable of reducing gene expression showing a greater sequence specificity over a far broader concentration range. For this reason we conclude that partial PS antisense oligos, with respect to full PS antisense oligos, might be particularly useful for studying gene function.     

Efficient reduction of target RNAs by small interfering RNA and RNase H-dependent antisense agents. A comparative analysis

RNA interference can be considered as an antisense mechanism of action that utilizes a double-stranded RNase to promote hydrolysis of the target RNA. We have performed a comparative study of optimized antisense oligonucleotides designed to work by an RNA interference mechanism to oligonucleotides designed to work by an RNase H-dependent mechanism in human cells. The potency, maximal effectiveness, duration of action, and sequence specificity of optimized RNase H-dependent oligonucleotides and small interfering RNA (siRNA) oligonucleotide duplexes were evaluated and found to be comparable. Effects of base mismatches on activity were determined to be position-dependent for both siRNA oligonucleotides and RNase H-dependent oligonucleotides. In addition, we determined that the activity of both siRNA oligonucleotides and RNase H-dependent oligonucleotides is affected by the secondary structure of the target mRNA. To determine whether positions on target RNA identified as being susceptible for RNase H-mediated degradation would be coincident with siRNA target sites, we evaluated the effectiveness of siRNAs designed to bind the same position on the target mRNA as RNase H-dependent oligonucleotides. Examination of 80 siRNA oligonucleotide duplexes designed to bind to RNA from four distinct human genes revealed that, in general, activity correlated with the activity to RNase H-dependent oligonucleotides designed to the same site, although some exceptions were noted. The one major difference between the two strategies is that RNase H-dependent oligonucleotides were determined to be active when directed against targets in the pre-mRNA, whereas siRNAs were not. These results demonstrate that siRNA oligonucleotide- and RNase H-dependent antisense strategies are both valid strategies for evaluating function of genes in cell-based assays.     

Potent growth inhibition of leukemic cells by novel ribbon-type antisense oligonucleotides to c-myb1

We studied the effects of antisense oligonucleotides (AS oligos) with a novel structure. The AS oligos were covalently closed to avoid exonuclease activities by enzymatic ligation of two identical molecules. The AS oligos of a ribbon type (RiAS oligos) consist of two loops containing multiple antisense sequences and a stem connecting the two loops. Three antisense sequences targeting different binding sites were placed in a loop that was designed to form a minimal secondary structure by itself. RiAS oligos were found to be stable because they largely preserved their structural integrity after 24 h incubation in the presence of either exonuclease III or serums. When a human promyelocytic cell line, HL-60, was treated with RiAS oligos to c-myb, c-myb expression was effectively ablated. Cell growth was inhibited by >90% determined by both the 3-[4,5-dimethythiazol-2-yl]-2,5-diphenyltetrazolium bromide assay and [(3)H]thymidine incorporation. Further, when the leukemic cell line K562 was treated with c-myb RiAS oligos, colony formation on soft agarose was reduced by 92 +/- 2%. These results suggest that RiAS oligos may be employed for developing molecular antisense drugs as well as for the functional study of a gene.     

A simple strategy for generation of gene knockdown constructs with convergent H1 and U6 promoters

RNA interference (RNAi) is a powerful tool for functional genetic studies in model organisms and mammalian cells. To facilitate rapid construction of gene knockdown constructs and RNAi libraries for known genes of mammalian cells, a new and simple strategy to produce small interfering RNA (siRNA) expression vectors with two opposing polymerase III promoters was developed. The design involved a one-step PCR amplification and single cloning procedure to construct a dual promoter siRNA expression vector. The forward primer is identical for all PCR reactions, only a single reverse primer that contains the siRNA targeting sequence has to be synthesized in the construction of each individual vector. This single primer design is cost-effective and it reduces the risk of sequence errors during synthesis of long oligos. Sense and antisense strands of siRNA duplexes were transcribed from the same template and this eliminated the need to synthesize long hairpin-forming oligonucleotides. Our study demonstrated that this vector design could mediate potent inhibition of expression of both exogenous and endogenous genes in mammalian cells.     

Prediction of siRNA knockdown efficiency using artificial neural network models

Selective knockdown of gene expression by short interference RNAs (siRNAs) has allowed rapid validation of gene functions and made possible a high throughput, genome scale approach to interrogate gene function. However, randomly designed siRNAs display different knockdown efficiencies of target genes. Hence, various prediction algorithms based on siRNA functionality have recently been constructed to increase the likelihood of selecting effective siRNAs, thereby reducing the experimental cost. Toward this end, we have trained three Back-propagation and Bayesian neural network models, previously not used in this context, to predict the knockdown efficiencies of 180 experimentally verified siRNAs on their corresponding target genes. Using our input coding based primarily on RNA structure thermodynamic parameters and cross-validation method, we showed that our neural network models outperformed most other methods and are comparable to the best predicting algorithm thus far published. Furthermore, our neural network models correctly classified 74% of all siRNAs into different efficiency categories; with a correlation coefficient of 0.43 and receiver operating characteristic curve score of 0.78, thus highlighting the potential utility of this method to complement other existing siRNA classification and prediction schemes.     

Improved and automated prediction of effective siRNA

Short interfering RNAs are used in functional genomics studies to knockdown a single gene in a reversible manner. The results of siRNA experiments are highly dependent on the choice of siRNA sequence. In order to evaluate siRNA design rules, we collected a database of 398 siRNAs of known efficacy from 92 genes. We used this database to evaluate previously proposed rules from smaller datasets, and to find a new set of rules that are optimal for the entire database. We also trained a regression tree with full cross-validation. It was however difficult to obtain the same precision as methods previously tested on small datasets from one or two genes. We show that those methods are overfitting as they work poorly on independent validation datasets from multiple genes. Our new design rules can predict siRNAs with efficacy >/= 50% in 91% of cases, and with efficacy >/=90% in 52% of cases, which is more than a twofold improvement over random selection. Software for designing siRNAs is available online via a web server at or as a standalone version for high-throughput applications.     

Antisense oligonucleotides as therapeutic agents.

Antisense oligonucleotides can block the expression of specific target genes involved in the development of human diseases. Therapeutic applications of antisense techniques are currently under investigation in many different fields. The use of antisense molecules to modify gene expression is variable in its efficacy and reliability, raising objections about their use as therapeutic agents. However, preliminary results of several clinical studies demonstrated the safety and to some extent the efficacy of antisense oligodeoxynucleotides (ODNs) in patients with malignant diseases. Clinical response was observed in some patients suffering from ovarian cancer who were treated with antisense targeted against the gene encoding for the protein kinase C-alpha. Some hematological diseases treated with antisense oligos targeted against the bcr/abl and the bcl2 mRNAs have shown promising clinical response. Antisense therapy has been useful in the treatment of cardiovascular disorders such as restenosis after angioplasty, vascular bypass graft occlusion, and transplant coronary vasculopathy. Antisense oligonucleotides also have shown promise as antiviral agents. Several investigators are performing trials with oligonucleotides targeted against the human immunodeficiency virus-1 (HIV-1) and hepatitis viruses. Phosphorothioate ODNs now have reached phase I and II in clinical trials for the treatment of cancer and viral infections, so far demonstrating an acceptable safety and pharmacokinetic profile for continuing their development. The new drug Vitravene, based on a phosphorothioate oligonucleotide designed to inhibit the human cytomegalovirus (CMV), promises that some substantial successes can be reached with the antisense technique.     

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.     

Biotinylated antisense methylphosphonate oligodeoxynucleotides. Inhibition of spliceosome assembly and affinity selection of U1 and U2 small nuclear RNPs.

Methylphosphonate (PC) backbone oligodeoxynucleotides complementary to the 5'-terminal nucleotides of U1 and U2 small nuclear (sn) RNAs do not elicit RNase H action under conditions in which natural (phosphodiester) oligodeoxynucleotides yield extensive RNase H cleavage. We show here that antisense PC oligonucleotides can mask sites in U1 and U2 snRNPs that are required for spliceosome formation. We further report that biotinylated derivatives of antisense PC oligos can be used for affinity selection of U1 and U2 snRNPs.     

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 beta-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 http://sfold.wadsworth.org.