Carbohydrate Modifications in Antisense Oligonucleotide Therapy: New Kids on the Block

Abstract

Chemical modifications to improve the efficacy of an antisense oligonucleotide are designed to increase the binding affinity to target RNA, to enhance the nuclease resistance, and to improve cellular delivery. Among the different sites available for chemical modification in a nucleoside building block, the 2′-position of the carbohydrate moiety¹ has proven to be the most valuable for various reasons: (1) 2′-modification can confer an RNA-like 3′-endo conformation to the antisense oligonucleotide. Such a preorganization for an RNA like conformation^2,3,4,5 greatly improves the binding affinity to the target RNA; (2) 2′-modification provides nuclease resistance to oligonucleotides; (3) 2′-modification provides chemical stability against potential depurination conditions pharmacology evaluations and correlation with pharmacokinetic changes are emerging from these novel chemical modifications. Analytical chemistry of modified oligonucleotides before and after biological administration of antisense oligonucleotides with techniques such as capillary gel electrophoresis (CGE) and mass spectrometry help to determine the purity as well as the in vivo fate of these complex molecules. Large-scale synthesis is becoming a tangible reality for antisense oligonucleotides. Nucleic acid chemists and biologists alike are beginning to understand the structure-biological activity in terms of basic physical-organic parameters such as the gauche effect, the charge effect and conformational constraints. Synthesis of chimeric designer oligonucleotides bringing the attractive features of different modifications to a given antisense oligonucleotide sequence to generate synergistic interactions is forthcoming³⁰. These advances along with the potential availability of complete human genome sequence information promise a bright future for the widespread use of nucleic acid based therapeutics.     

Development of efficient transient transfection systems for introducing antisense oligonucleotides into human epithelial skin cells

Systemic treatment with antisense oligonucleotides is confounded by the dual problems of potential cytotoxicity of antisense oligonucleotides and carrier molecules such as cationic lipids. Treatment of pathologic conditions affecting the skin may avoid these problems to a large degree due to local application. The success of antisense strategies has been limited by the poor uptake of the transfection reagent and inadequate intracellular compartmentalization. Human skin epithelial cells, therefore, are attractive experimental tools for testing both in vitro and in vivo antisense therapies. In the present study, we determined commercially available liposomes which reproducibly induced a nontoxic increase of oligonucleotide uptake in cultured SZ95 sebocytes and keratinocytes. The final protocol for SZ95 sebocytes was a 4-hour incubation with DOTAP in a 2:1 (w/w) lipid/oligonucleotide ratio in serum-free medium. The fluorescein-labeled (ATCG)(5) random oligonucleotide molecules were detected within the nucleus. The optimum transfection system for primary keratinocytes was poly-L-ornithine (12 microg/ml) in a medium without bovine pituitary extract over 4 hours. The uptake of the oligonucleotide increased in the presence of the polycation and oligonucleotide molecules were localized in the cytoplasm of keratinocytes. Oligonucleotide transfection with the help of cationic lipids did not affect the expression of androgen receptor and of the house-keeping gene beta-actin. Thus, cationic lipids are useful for delivery of antisense oligonucleotides into skin cells in vitro and may be used for topical application on animal and human skin.     

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.     

Building oligonucleotide therapeutics using non-natural chemistries

Modified nucleotides are increasingly being utilized in all categories of therapeutic oligonucleotides to increase nuclease-resistance, target affinity and specificity. The extent to which these substitutions are tolerated varies with the different modes of action exploited by various modalities, but fully modified oligonucleotides have now been discovered for most types of therapeutic oligonucleotide. Fully phosphorothioate-substituted antisense oligonucleotides have been used for several years. The first fully modified siRNA was reported in 2006 with a 2'-O-methyl sense strand and a phosphorothioate antisense strand. The first fully modified aptamer (2'-O-methyl) was reported in 2005. It is expected that future candidate therapeutic oligonucleotides will have even more drug-like characteristics as a result of the inclusion of modified nucleotides.     

Synthesis and properties of 2'-O,4'-C-ethylene-bridged nucleic acids (ENA) as effective antisense oligonucleotides

Novel bicyclo nucleosides, 2'-O,4'-C-ethylene nucleosides and 2'-O,4'-C-propylene nucleosides, were synthesized as building blocks for antisense oligonucleotides to further optimize the 2'-O,4'-C-methylene-linkage of bridged nucleic acids (2',4'-BNA) or locked nucleic acids (LNA). Both the 2'-O,4'-C-ethylene- and propylene-linkage within these nucleosides restrict the sugar puckering to the N-conformation of RNA as do 2',4'-BNA/LNA. Furthermore, ethylene-bridged nucleic acids (ENA) having 2'-O,4'-C-ethylene nucleosides had considerably increased the affinity to complementary RNA, and were as high as that of 2',4'-BNA/LNA (DeltaT(m)=+3 approximately 5 degrees C per modification). On the other hand, addition of 2'-O,4'-C-propylene modifications in oligonucleotides led to a decrease in the affinity to complementary RNA. As for the stability against nucleases, incorporation of one 2'-O,4'-C-ethylene or one 2'-O,4'-C-propylene nucleoside into oligonucleotides considerably increased their resistance against exonucleases to an extent greater than 2',4'-BNA/LNA. These results indicate that ENA is more suitable as an antisense oligonucleotide and is expected to have better antisense activity than 2',4'-BNA/LNA.     

Use of Cationic Lipids to Enhance the Biological Activity of Antisense Oligonucleotides

Abstract

Antisense oligonucleotides bind to specific mRNA or pre-mRNA sequences through Watson-Crick base pairing, resulting in decreased expression of the targeted protein. The use of cationic lipids to enhance cellular uptake of antisense oligonucleotides is reviewed herein. Cationic lipids such as N[1-(2,3-dioleyloxy)propyl]-N, N, N-trimethylammonium chloride (DOTMA) were found to enhance the biological activity of phosphorothioate oligonucleotides by at least 1000-fold in cell culture. Cationic lipid preparations enhanced both the rate and amount of oligonucleotide which associated with cells. In addition, DOTMA markedly changed the subcellular distribution of the oligonucleotide. In the absence of lipid, fluorescein labelled phosphorothioate oligonucleotides accumulated in discrete cytoplasmic structures. In the presence of cationic lipids, the oligonucleotides concentrated within the nucleus, were excluded from nucleoli, and localized in punctate cytoplasmic structures. The accumulation of the oligonucleotide in the nucleus was inhibited by incubation of the cells at 4°C and by monensin, but not by chloroquine, ammonium chloride, or nocodazole. Cell lines, both primary and transformed, differ markedly in their sensitivity to inhibition of gene expression with antisense oligonucleotides in the presence of cationic lipids. The differential sensitivity of the cells correlates with the amount of ³⁵S-labelled oligonucleotide associated with the cells and the number of cells in the population which take up the oligonucleotide. Our studies have demonstrated that several types of cationic lipids markedly enhance the activity of phosphorothioate oligonucleotides in cell culture models. We are currently investigating the ability of cationic lipids to enhance activity of antisense oligonucleotides in more complex systems such as organ cultures and in animals.     

Characterization of modified antisense oligonucleotides in Xenopus laevis embryos

A wide variety of modified oligonucleotides have been tested as antisense agents. Each chemical modification produces a distinct profile of potency, toxicity, and specificity. Novel cationic phosphoramidate-modified antisense oligonucleotides have been developed recently that have unique and interesting properties. We compared the relative potency and specificity of a variety of established antisense oligonucleotides, including phosphorothioates (PS), 2'-O-methyl (2'OMe) RNAs, locked nucleic acids (LNAs), and neutral methoxyethyl (MEA) phosphoramidates with new cationic N,N-dimethylethylenediamine (DMED) phosphoramidate-modified antisense oligonucleotides. A series of oligonucleotides was synthesized that targeted two sites in the Xenopus laevis survivin gene and were introduced into Xenopus embryos by microinjection. Effects on survivin gene expression were examined using quantitative real-time PCR. Of the various modified oligonucleotide designs tested, LNA/PS chimeras (which showed the highest melting temperature) and DMED/phosphodiester chimeras (which showed protection of neighboring phosphate bonds) were potent in reducing gene expression. At 40 nM, overall specificity was superior for the LNA/PS-modified compounds compared with the DMED-modified oligonucleotides. However, at 400 nM, both of these compounds led to significant degradation of survivin mRNA, even when up to three mismatches were present in the heteroduplex.     

Neurons are protected from excitotoxic death by p53 antisense oligonucleotides delivered in anionic liposomes.

The potential of anionic liposomes for oligonucleotide delivery was explored because the requirement for a net-positive charge on transfection-competent cationic liposome-DNA complexes is ambiguous. Liposomes composed of phosphatidylglycerol and phosphatidylcholine were monodisperse and encapsulated oligonucleotides with 40-60% efficiency. Ionic strength, bilayer charge density, and oligonucleotide chemistry influenced encapsulation. To demonstrate the biological efficacy of this vector, antisense oligonucleotides to p53 delivered in anionic liposomes were tested in an in vitro model of excitotoxicity. Exposure of hippocampal neurons to glutamate increased p53 protein expression 4-fold and decreased neuronal survival to approximately 35%. Treatment with 1 microm p53 antisense oligonucleotides in anionic liposomes prevented glutamate-induced up-regulation of p53 and increased neuronal survival to approximately 75%. Encapsulated phosphorothioate p53 antisense oligonucleotides were neuroprotective at 5-10-fold lower concentrations than when unencapsulated. Replacing the anionic lipid with phosphatidylserine significantly decreased neuroprotection. p53 antisense oligonucleotides complexed with cationic liposomes were ineffective. Neuroprotection by p53 antisense oligonucleotides in anionic liposomes was comparable with that by glutamate receptor antagonists and a chemical inhibitor of p53. Anionic liposomes were also capable of delivering plasmids and inducing transgene expression in neurons. Anionic liposome-mediated internalization of Cy3-labeled oligonucleotides by neurons and several other cell lines demonstrated the universal applicability of this vector.     

Enhanced delivery of antisense oligonucleotides with fluorophore-conjugated PAMAM dendrimers

PAMAM dendrimers are cationic polymers that have been used for the delivery of genes and oligonucleotides to cells. However, little is known about the behavior of dendrimer–nucleic acid complexes once they reach the cell interior. To pursue this issue, we prepared dendrimers conjugated with the fluorescent dye Oregon green 488. These were used in conjunction with oligonucleotides labeled with a red (TAMRA) fluorophore in order to visualize the sub-cellular distribution of the dendrimer–oligonucleotide complex and of its components by two-color digital fluorescence microscopy. The 2′-O-methyl antisense oligonucleotide sequence used in these studies was designed to correct splicing at an aberrant intron inserted into a luciferase reporter gene; thus effective delivery of the antisense agent results in the expression of the reporter gene product. The dendrimer–oligonucleotide complex remained associated during the process of uptake into vesicular compartments and eventual entry into the nucleus. Since the pharmacological activity of the antisense compound was manifest under these conditions, it suggests that the dendrimer–oligonucleotide complex is functionally active. A surprising result of these studies was that the Oregon green 488-conjugated dendrimer was a much better delivery agent for antisense compounds than unmodified dendrimer. This suggests that coupling of relatively hydrophobic small molecules to PAMAM dendrimers may provide a useful means of enhancing their capabilities as delivery agents for nucleic acids.     

Identification and quantification of protecting groups remaining in commercial oligonucleotide products using monoclonal antibodies

Quality control is paramount to reproducibly achieving oligonucleotide therapeutics and diagnostics of superior value. However, incomplete deprotection of nucleoside reactive groups after the automated chemical synthesis of oligonucleotides would result in diminished antisense activity and in erroneous array analysis of gene expression. Mass spectrometry and capillary electrophoresis are used to detect aborted sequences of oligonucleotides, but not to identify and quantify incompletely deprotected oligonucleotides. To address this problem, monoclonal antibodies (MAbs), ELISA, and dot-blot assays were developed for the specific identification and quantification of the commonly used nucleic acid base- and sugar-protecting groups: benzoyl, isobutyryl, isopropylphenoxyacetyl, and dimethoxytrityl. Each MAb was capable of reproducibly detecting 8-32 pmol of the respectively protected nucleoside in an intact DNA or RNA sample composed of 320-640 nmol of the deprotected nucleoside. In a direct comparison, HPLC nucleoside composition analysis of enzyme-hydrolyzed DNA was limited to the detection of 2-5 nmol of protected nucleoside. Using the MAb dot-blot assay, 5 of 16 commercial DNA products obtained from eight different companies were found to have 1.0-5.2% of the benzoyl and isopropylphenoxyacetyl protecting groups remaining. Thus, MAbs selectively identify and quantify picomoles of remaining protecting groups on antisense therapeutics and oligonucleotide diagnostics.     

4′-C-Aminomethyl-2′-deoxy-2′-fluoroarabinonucleoside increases the nuclease resistance of DNA without inhibiting the ability of a DNA/RNA duplex to activate RNase H

An antisense oligonucleotide is expected as an innovative drug for cancer and hereditary diseases. In this paper, we designed and synthesized DNAs containing a novel nucleoside analog, 1-(4-C-aminomethyl-2-deoxy-2-fluoro-β-d-arabinofuranosyl)thymine, and evaluated their properties. It was revealed that the analog slightly decreases the thermal stability of the DNA/RNA duplex but significantly increases the stability of DNA in a buffer containing bovine serum. Furthermore, it turned out that the DNA/RNA duplex containing the analog is a good substrate for Escherichia coli RNase H. Thus, DNAs containing the nucleoside analog would be good candidates for the development of therapeutic antisense oligonucleotides.     

Uptake of antisense oligonucleotides and functional block of acetylcholine receptor subunit gene expression in primary embryonic neurons

Several recent studies have used antisense oligonucleotides in the nervous system to probe the functional role of particular gene products. Since antisense oligonucleotide-mediated block of gene expression typically involves uptake of the oligonucleotides, we have characterized the mechanism of this uptake into developing neurons from embryonic chickens. Antisense oligonucleotides (15 mers) added to the bathing media are taken up into the embryonic chicken sympathetic neurons maintained in vitro. A portion of the oligonucleotide uptake is temperature dependent and saturates at extracellular oligonucleotide concentrations ≥ 20 μM. This temperature sensitive, saturable component is effectively competed by single nucleotides of ATP and AMP and is reminiscent of receptor-mediated endocytosis of oligonucleotides described in non-neuronal cells. The efficiency of the oligonucleotide uptake system is dependent on the developmental stage of the animal but independent of the number of days that the neurons are maintained in vitro. Following the uptake of antisense oligonucleotides directed against ion channel subunit genes expressed by these neurons (nicotinic acetylcholine receptor subunit α3; nAChR α3), biophysical assays reveal that the functional expression of the target gene is largely blocked. Thus the number of wild type nAChR channels expressed is decreased by =80%–90%. Furthermore, following antisense deletion of α3, “mutant” nAChRs with distinct functional characteristics are expressed. In sum, these studies characterize the uptake of antisense oligonucleotide and demonstrate the functional block of specific gene expression in primary developing neurons. In addition, the functional studies emphasize the need for sensitive and specific assay following antisense deletion, since other homologous gene products may substitute for the targeted gene resulting in new phenotypes that are subtly different from wild type. © 1993Wiley-Liss, Inc.     

Hybridization specificity, enzymatic activity and biological (Ha-ras) activity of oligonucleotides containing 2,4-dideoxy-beta-D-erythro-hexopyranosyl nucleosides.

Antisense oligonucleotides with a 2,4-dideoxyhexopyranosyl nucleoside incorporated at the 3'-end and at a mutation site of the Ha-ras oncogene mRNA were synthesized. Melting temperature studies revealed that an A*-G mismatch is more stable than an A*-T mismatch with these hexopyranosyl nucleosides incorporated at the mutation site. The oligonucleotides are stable against enzymatic degradation. RNase H mediated cleavage studies revealed selective cleavage of mutated Ha-ras mRNA. The oligonucleotide containing two pyranose nucleosides at the penultimate position activates RNase H more strongly than natural oligonucleotides. No correlation, however, was found between DNA - DNA or RNA - DNA melting temperatures and RNase H mediated cleavage capacity. Although the A*-G mismatch gives more stable hybridization than the A*-T base pairing, only the oligonucleotides containing an A*-T base pair are recognized by RNase H. This modification is situated 3 base pairs upstream to the cleavage site. Finally, the double pyranose modified oligonucleotide was able to reduce the growth of T24 cells (bladder carcinoma) while the unmodified antisense oligonucleotide was not.     

Inhibition of the androgen receptor by antisense oligonucleotides regulates the biological activity of androgens in SZ95 sebocytes.

One novel strategy for the blockade of the androgen receptor could be the selective inhibition of androgen receptor by antisense oligonucleotides or small interfering RNA molecules. Here we describe the down regulation of the androgen receptor in cultured human SZ95 sebocytes with antisense oligonucleotides modified with phosphorothioates and 2'- O-methylribosyl residues. The ability of antisense oligonucleotides to cross the cellular membrane was enhanced by establishing a transient transfection system based on cationic lipid vesicles. Both antisense oligonucleotide types administered caused assumedly translational arrest. Dose-dependent inhibition of androgen receptor protein expression was observed after SZ95 sebocyte transfection with modified phosphorothioate oligonucleotides and modified 2'- O-methylribonucleotides which were directed against the translational start of the androgen receptor mRNA. The strongest transient inhibition of androgen receptor expression was detected after 14 hours with 1.0 muM antisense 2'- O-methylribonucleotides (88+/-1.3%, p<0.001). With longer recovery times than 24 hours, androgen receptor protein expression returned to the native control levels. Inhibition of the expression of androgen receptor by antisense oligonucleotides, reduced the enhanced proliferation of SZ95 sebocytes challenged by testosterone and 5alpha-dihydrotestosterone. This administration opens new therapeutic possibilities in androgen-associated skin diseases, since we could also show androgen inhibition with these antisense oligonucleotides in a reconstituted human epidermis model (Horm Metab Res 2007; 39:157-165).     

Inhibition of dengue virus by novel, modified antisense oligonucleotides.

Five different target regions along the length of the dengue virus type 2 genome were compared for inhibition of the virus following intracellular injection of the cognate antisense oligonucleotides and their analogs. Unmodified phosphodiester oligonucleotides as well as the corresponding phosphorothioate oligonucleotides were ineffective in bringing about a significant inhibition of the virus. Novel modified phosphorothioate oligonucleotides in which the C-5 atoms of uridines and cytidines were replaced by propynyl groups caused a significant inhibition of the virus. Antisense oligonucleotide directed against the target region near the translation initiation site of dengue virus RNA was the most effective, followed by antisense oligonucleotide directed against a target in the 3' untranslated region of the virus RNA. It is suggested that the inhibitory effect of these novel modified oligonucleotides is due to their increased affinity for the target sequences and that they probably function via an RNase H cleavage of the oligonucleotide:RNA heteroduplex.     

Selecting optimal oligonucleotide composition for maximal antisense effect following streptolysin O-mediated delivery into human leukaemia cells.

It is widely accepted that most cell types efficiently exclude oligonucleotides in vitro and require specific delivery systems, such as cationic lipids, to enhance uptake and subsequent antisense effects. Oligonucleotides are not readily transfected into leukaemia cell lines using cationic lipid systems and streptolysin O (SLO) is used to effect their delivery. We wished to investigate the optimal oligonucleotide composition for antisense efficacy and specificity following delivery into leukaemia cells using SLO. For this study the well characterised chronic myeloid leukaemia cell line KYO-1 was selected and oligonucleotides (20mers) were targeted to an empirically identified accessible site of c- myc mRNA. The efficiency and specificity of antisense effect was measured 4 and 24 h after SLO-mediated delivery of the oligonucleotides. C5-propyne phosphodiester and phosphorothioate compounds were found to present substantial non-specific effects at 20 microM but were inactive at 0.2 microM. Indeed, no antisense-specific effect was noted at any concentration at either time. All of the other oligonucleotides tested induced some measurable antisense effect, except 7 (chimeric, all-phosphorothioate, 2'-methoxyethoxy termini) which was essentially inactive at 20 microM. The rank efficiency order of the remaining antisense compounds was 4 = 3 >> 9 >> 10 = 8 = 5 = 6 > 11. The efficient antisense effects induced by the chimeric methylphosphonate-phosphodiester compounds were found to be highly specific. Increased phosphorothioate content in the oligonucleotide backbone correlated with reduced antisense activity (efficacy: 2'-methoxyethoxy series 9 >> 8 >> 7, 2'-methoxytriethoxy series 10 > 11). No consistent evidence was obtained for increased activity correlating with increased oligonucleotide-mRNA heteroduplex thermal stability. In conclusion, the chimeric methylphosphonate-phosphodiester oligodeoxynucleotides present the most favourable characteristics of the compounds tested, for efficient and specific antisense suppression of gene expression following SLO-mediated delivery.     

Novel cationic amphiphiles as delivery agents for antisense oligonucleotides.

There has been great interest recently in therapeutic use of nucleic acids including genes, ribozymes and antisense oligonucleotides. Despite recent improvements in delivering antisense oligonucleotides to cells in culture, nucleic acid-based therapy is still often limited by the poor penetration of the nucleic acid into the cytoplasm and nucleus of cells. In this report we describe nucleic acid delivery to cells using a series of novel cationic amphiphiles containing cholic acid moieties linked via alkylamino side chains. We term these agents 'molecular umbrellas' since the cationic alkylamino chains provide a 'handle' for binding of nucleic acids, while the cholic acid moieties are likely to interact with the lipid bilayer allowing the highly charged nucleic acid backbone to traverse across the cell membrane. Optimal gene and oligonucleotide delivery to cells was afforded by a derivative (amphiphile 5) containing four cholic acid moieties. With this amphiphile used as a constituent in cationic liposomes, a 4-5 log increase in reporter gene delivery was measured. This amphiphile used alone provided a 250-fold enhancement of oligo-nucleotide association with cells as observed by flow cytometry. A substantial fraction of cells exposed to complexes of amphiphile 5 and fluorescent oligo-nucleotide showed nuclear accumulation of the fluorophore. Enhanced pharmacological effectiveness of antisense oligonucleotides complexed with amphiphile 5 was observed using an antisense splicing correction assay that activates a Luciferase reporter. Intracellular delivery, nuclear localization and pharmacological effectiveness of oligonucleotides using amphiphile 5 were similar to those afforded by commercial cytofectins. However, in contrast to most commercial cytofectins, the umbrella amphiphile showed substantial delivery activity even in the presence of high concentrations of serum.     

Polyethylenimine but not cationic lipid improves antisense activity of 3'-capped phosphodiester oligonucleotides

Lipofectin, which is a mixture of neutral lipid with a cationic lipid, has been widely used to enhance cellular delivery of phosphorothioate, 2'-sugar-modified, and chimeric antisense oligonucleotides. Phosphodiester oligonucleotides delivered with Lipofectin usually do not elicit antisense activity probably because cationic lipid formulations do not sufficiently protect unmodified oligonucleotides from nuclease degradation. We show that a cationic polymer, polyethylenimine (PEI), improves the uptake and antisense activity of 3'-capped 20-mer and 12-mer antisense phosphodiester oligonucleotides (PO-ODN) targeted to different regions of Ha-ras mRNA and to the 3'-untranslated region (3'-UTR) of C-raf kinase. In contrast, PEI, which forms a very stable complex with the 20-mer phosphorothioate oligonucleotide (PS-ODN), does not enhance its antisense activity. Using fluorescently labeled carriers and ODN, we show that PEI-PS-ODN particles are very efficiently taken up by cells but PS-ODN is not dissociated from the carrier. Our results indicate that carrier-ODN particle size and stability and ODN release kinetics vary with the chemical nature of the ODN and the carrier being transfected into the cells. The very low cost of PEI compared with cytofectins and the increased affinity for target mRNA and decreased affinity for proteins of PO-ODN compared with PS-ODN make the use of PEI-PO-ODN very attractive.     

Predicting the efficacy of short oligonucleotides in antisense and RNAi experiments with boosted genetic programming

MOTIVATION: Both small interfering RNAs (siRNAs) and antisense oligonucleotides can selectively block gene expression. Although the two methods rely on different cellular mechanisms, these methods share the common property that not all oligonucleotides (oligos) are equally effective. That is, if mRNA target sites are picked at random, many of the antisense or siRNA oligos will not be effective. Algorithms that can reliably predict the efficacy of candidate oligos can greatly reduce the cost of knockdown experiments, but previous attempts to predict the efficacy of antisense oligos have had limited success. Machine learning has not previously been used to predict siRNA efficacy. RESULTS: We develop a genetic programming based prediction system that shows promising results on both antisense and siRNA efficacy prediction. We train and evaluate our system on a previously published database of antisense efficacies and our own database of siRNA efficacies collected from the literature. The best models gave an overall correlation between predicted and observed efficacy of 0.46 on both antisense and siRNA data. As a comparison, the best correlations of support vector machine classifiers trained on the same data were 0.40 and 0.30, respectively.