Optimizing the delivery systems of chimeric RNA·DNA oligonucleotides

Beyond general oligonucleotide transfer


  • Li Liang,

    1. National Laboratory of Medical Molecular Biology, Institute of Basic Medical Sciences, Chinese Academy of Medical Sciences & Peking Union Medical College, Beijing, People's Republic of China
    Search for more papers by this author
  • De-Pei Liu,

    1. National Laboratory of Medical Molecular Biology, Institute of Basic Medical Sciences, Chinese Academy of Medical Sciences & Peking Union Medical College, Beijing, People's Republic of China
    Search for more papers by this author
  • Chih-Chuan Liang

    1. National Laboratory of Medical Molecular Biology, Institute of Basic Medical Sciences, Chinese Academy of Medical Sciences & Peking Union Medical College, Beijing, People's Republic of China
    Search for more papers by this author

D.-P. Liu, National Laboratory of Medical Molecular Biology, Institute of Basic Medical Sciences, Chinese Academy of Medical Sciences & Peking Union Medical College, 5 Dong Dan San Tiao, Beijing 100005, People's Republic of China.


Special oligonucleotides for targeted gene correction have attracted increasing attention recently, one of which is the chimeric RNA·DNA oligonucleotide (RDO) system. RDOs for targeted gene correction were first designed in 1996, and are typically 68 nucleotides in length including continuous RNA and DNA sequences (RNA is 2′-O-methyl-modified). They have a 25 bp double stranded region homologous to the targeted gene, two hairpin ends of T loop and a 5 bp GC clamp, that give the molecule much greater stability [Fig. 1]. One mismatch site in the middle of the double-stranded region is designed for targeted gene therapy. RDOs have been used recently for targeted gene correction of point mutations both in vitro and in vivo, but many problems must be solved before clinical application. One of the solutions is to optimize the delivery vectors for RDOs. To date, few RDO delivery systems have been used. Therefore, new vectors should be tried for RDO transfer, such as the use of nanoparticles. Additionally, different kinds of modifications should be applied to RDO carrier systems to increase the total correction efficiency in vivo. Only with the development of delivery systems can RDOs be used for gene therapy, and successfully applied to functional genomics.

Figure 1.

Diagrammatic structure of RDO. RDOs are typically 68 nucleotides in length, including continuous 2′-O-methyl-modified RNA and DNA sequences, a 25 bp double-stranded region homologous to the targeted gene, two hairpin ends of T loop and a 5 bp GC clamp, which give the molecule greater stability. One mismatch site in the middle of the double-stranded region is designed for targeted gene therapy.


biodegradable pH-sensitive surfactants


Chinese hamster ovary cells


dimyristyloxypropyl dimethylhydroxyethylammonium bromide






dioleoyloxypropyl trimethylammonium methylsulfate










chimeric RNA·DNA oligonucleotide


single-stranded oligonucleotide


triplex-forming oligonucleotide

Targeted gene therapies are ideal strategies for gene therapy, which include gene replacement in situ and gene correction. Gene replacement substitutes the mutated gene with a normal one, but it is a low efficiency method due to the limitation of homologous recombination. Gene correction is the most putative strategy of targeted gene therapy [1]. Recently, three kinds of oligonucleotides have been used for targeted gene correction: triplex-forming oligonucleotides (TFOs), modified single-stranded oligonucleotides and chimeric RNA·DNA oligonucleotides (RDOs). Triplex-forming oligonucleotides are oligonucleotides with high affinity for polypurine/polypyrimidine sequences, and are capable of forming a triplex after binding to the major groove of DNA. However, the application of TFOs is limited, because they can only repair gene defects near homopurine sequences [2]. Modified single-stranded oligonucleotides (SDOs) are also used for targeted gene correction. The most common SDO is phosphorothioate-modified oligonucleotide (ON), but unfortunately this is toxic to cells when used in vivo. The third kind of ON for gene correction is the chimeric RNA·DNA oligonucleotide.

Chimeric RNAḃDNA oligonucleotide for targeted gene correction

Chimeric RNA·DNA oligonucleotides were first used in antisense gene therapy in late 1980s. Inoue et al. constructed single-stranded chimeric oligonucleotides containing both DNA and RNA sequences [3]. Monia et al. also reported that such chimeric oligonucleotides improved the efficiency of antisense therapy [4]. In 1996, Kmiec's group constructed a novel chimeric oligonucleotide for targeted gene correction [5]. RDO for targeted gene correction is a single-stranded molecule, typically 68 nucleotides or more in length, including continuous RNA and DNA sequences (RNA is 2′-O-methyl-modified). It has a 25 bp double stranded region homologous to the targeted gene, two hairpin ends of T loop and a 5 bp GC clamp that gives the molecule much greater stability. One mismatch site in the middle of the double-stranded region is designed for targeted gene correction [6–8][Fig. 1]. The RDO is different from other oligonucleotides in several respects. First, it is a self-complementary oligonucleotide that folds into a double-hairpin configuration, different from plasmids that are circular, or general oligonucleotides that are linear. Second, it is chimeric, with both RNA and DNA sequences. Third, its length is different from most oligonucleotides used for antisense therapy, which are usually 12–40 bp in length [9], but RDO is 68 nucleotides or more in length.

RDO is used for targeted gene correction both in vitro and in vivo. Its mechanism is not completely known. The gene correction of RDO is based on the mechanism of endogenous DNA repair systems after homologous pairing, rather than homologous recombination. Therefore, RDO is not limited by the frequency of homologous recombination. The two parts of RDO play different roles in gene repair. The DNA region enables gene correction to occur, and the 2′-O-methyl-modified RNA region stabilizes the structure [10]. Transcription may also be involved in the process of gene correction. RDOs designed for sense strands are much more efficient than those for antisense strands [Fig. 2].

Figure 2.

The mechanism of RDO action for targeted gene correction is to repair mismatch after homologous alignment. The DNA strand is responsible for the correction, while the RNA strand stabilizes the structure. The RDO designed for reacting with the sense strand is more efficient than that for the antisense strand.

Although RDOs corrected point mutations in animal models, there is still a long way to go before clinical applicationis possible. The correction efficiency is distinct (from 1–40%) in different cell lines or tissues [11–17]. Optimizing the length and structure of the RDO is crucial, such as lengthening the homologous region or changing the place of the mismatch on the chimeric double-stranded region. Another key step toward clinical application is to optimize RDO delivery vectors. Only with the development of carrier systems can RDOs be applied to functional genomics and used in human gene therapy.

Recent progress in RDO delivery

Several strategies have been attempted to deliver RDOs. Microinjection and microparticle bombardment are efficient methods to be used in vitro or on stem cells, but can not be used in vivo. Two delivery systems have been used for RDO transfer in vivo. One is the use of liposomes. Liposomes were believed to encapsulate nucleic acids within their aqueous core in the past [18–20]. However, some cationic lipids may attract DNA by electrostatic charges [21]. Lipofectin (a commercial liposome) was the first transfection reagent used in RDO transfer. When the lipofectin–RDO complex was transferred into Chinese hamster ovary (CHO) cells containing extra-chromosomal plasmids, 30% correction rate was accomplished at the episomal targets in CHO cells [5]. Because this gene correction was in episomal DNA, not nuclear DNA, it cannot be compared with other experiments. A variety of liposomes have now been used for RDO delivery, especially commercially available liposomes [13]. Dioleoyloxypropyl trimethylammonium methyl-sulfate (DOTAP) was used to transfer RDOs to lymphoblastoid cells, and corrected point mutations of the α-hemoglobin gene. DOTAP also delivered RDOs to MEL-D7 cells and corrected the αE gene, which was introduced into the MEL cells, to the normal α gene. In HeLa cells and CMK cells transfected with two other types of liposomes, DMRIE-C and FuGene 6, detectable corrections can be achieved by RDOs [22]. Anionic liposomes, such as dioleoylphosphatidylserine (DOPS) were also used, and were more effective than the neutral or cationic liposomes for in vitro delivery. But all these delivery vectors were all at a low efficiency level.

Another RDO vector is polyethylenimine (PEI). PEI is a polymer with a backbone of two carbons followed by a nitrogen atom, and can be either linear or branched [23,24]. PEI attracts oligonucleotides by electrostatic charge. It was used to transfer RDOs into HeLa cells, and the point mutation at the −202 residue of the β-globin gene was corrected.

However, all these carrier systems without modifications are not tissue-specific. Nowadays, RDO delivery systems with modifications are used for gene correction in vivo and can concentrate RDO within a specific organ. Some modified liposomes are under study. It was reported that galatocerebroside [25], a type of polysaccharide that can recognize the asialoglycoprotein receptors on hepatocytes, was added to three different liposomes for organ targeting. All such decorated anionic, neutral and cationic liposomes delivered RDOs effectively, and promoted A→C conversion at the Ser365 position in the rat factor IX gene [25]. Furthermore, PEI was decorated by lactose, another liver-specific ligand. PEI (25 kDa) was covalently lactosylated, forming a lactose–PEI complex. The complex carrying RDO was administrated by tail vein injection into rats either once or repeatedly at fixed intervals. The results showed that RDO converted A→C at Ser365 of factor IX gene in rat liver specifically. Factor IX coagulant activities of rats decreased to 40% of normal, showing the effect of RDO conversion [26,27]. B. T. Kren reported that, for in vivo transfection, chimeric oligonucleotides were fluorescently labeled, and then complexed with lactose–PEI at a proportion of 1 : 6 (ON phosphate/PEI amine) in 5% (w/v) dextrose. The lactose–PEI–RDO complex was distributed homogeneously throughout the liver as early as 2 h after tail vein injection, and not in lung, heart and kidney [28]. The results showed that the G residue at nucleotide 1206 was replaced and UGT1A1 genetic defect was corrected (the genetic basis of Crigler–Najjar syndrome type I) in Gunn rat liver. In addition, the RDOs were complexed with anionic liposome AVETM-3 after AVETM-3 was coated with protamine sulfate. The complex delivered RDO to the nucleus more effectively than those without modifications [29].

In short, at present the best RDO delivery systems are modified PEI or liposomes, such as lactosylated PEI or polysaccharide modified liposomes [Fig. 3]. Such decoration not only facilitates targeting, but also improves the relative efficiency of RDOs. However, these modified delivery systems are not ideal, because of instability in serum or toxicity to cells.

Figure 3.

RDO delivery systems and transfer barriers.

The major delivery barriers and problems facing RDO transfer

RDO and its delivery systems must overcome several major hurdles for in vivo gene transfer. This is a multistep process [9,18,19]. First, the delivery systems should have low immunogenicity. Those delivery systems that can trigger immune responses can only be used to transfer RDOs into tumor cells as gene vaccines. Second, vectors should be tissue-specific. Because most genes only function in specific tissues, targeted gene therapy of RDOs should only be at specific organs. Gene conversion in all organs is wasteful, and could even increase undesired side-effects. Third, they should pass the cell membrane, be released from endosomes and be transported easily in the cytoplasm. Fourth, they should enter the nucleus easily. For transfection in vivo, this is one of the most important hurdles. Finally, cytotoxity is another major problem. Only low or nontoxic delivery systems could be used in humans. Therefore, we must test different delivery systems on RDOs, and try to lower their cytotoxity and improve their efficiency.

Future directions: optimizing RDO delivery systems

An ideal RDO delivery system should have little or no toxicity and high efficiency. It should be an all-round delivery system that can pass delivery barriers smoothly [30]. The following are clues on finding such a delivery system.

Applying novel delivery systems to RDO transfer

Recently, several novel delivery systems have been used successfully for plasmid DNA transfer or antisense oligonucleotide transfer. These are putative RDO delivery systems.

Pluronic gel is a substance traditionally used for transdermal injection. One special characteristic of pluronic gel is that it exists as a liquid when cold, and becomes solid at body temperature. Recently, pluronic gel has been used for antisense delivery, especially in blood vessels, and it has the advantage of prolonged delivery [31,32].

There are several substitutes for PEI. Chitosan, or poly(d-glucosamine), is a natural cationic amino-polysaccharide [33–38], and attracts oligonucleotides with electrostatic charge. Chitosan may be a substitute of PEI because it has low toxicity, and is biocompatible and resorbable. Regarding its efficiency, at least at 96 h after transfection of HeLa cells, chitosan was found to be 10 times more efficient than PEI [34]. Dendrimers are a new kind of reproducible substances, with a hydrocarbon core and charged surface of amino groups. They have the advantage of having a defined small size. However, their efficiency and toxicity still need evaluation.

Nanoparticles are new delivery systems. By the method of associating with oligonucleotides [39], they can be classified into encapsulating nanoparticles, complexing nanoparticles and conjugating nanoparticles [Fig. 4]. The first type is represented by nanosponges, such as alginate nanosponge [40], which are sponge-like nanoparticles containing many holes that carry the oligonucleotides. Additionally, nanocapsules such as poly(isobutyl-cyanoacrylate) (IBCA) are also encapsulating nanoparticles. They can entrap oligonucleotides in their aqueous core [41].

Figure 4.

Three types of nanoparticles. (A) Nanosponge, an encapsulating nanoparticle, which encapsulates DNA within its core. (B) Complexing nanoparticle, which attracts DNA by electrostatic charges. (C) Conjugating nanoparticle, which links to DNA through covalent bonds.

The second type is complexing nanoparticles, which are coated with cationic polymers. Such nanoparticles associate with oligonucleotides by electrostatic attraction.

Third, oligonucleotides can also be conjugated to nanospheres by covalent bonds. Different nanospheres have different characteristics. Encapsulating nanoparticles, especially nanosponges, may be the best among all nanoparticles, because they protect oligonucleotides from proteins and enzymes, and do not change the DNA conformation through electrostatic force [39]. For example, in alginate nanosponge, 80% of the oligonucleotides were still undegraded after 1 h of incubation with fetal bovine serum [39]. Considering tissue specificity, PACA [poly(alkylcyanoacrylate)] nanoparticles, which are complexing nanoparticles, can deliver oligonucleotides specifically to the liver. In contrast, alginate nanosponges concentrate oligonucleotides in the lung, liver and spleen. Comparing two encapsulating nanospheres, alginate nanosponges accumulated oligonucleotides in the lung 10-fold more than IBCA when the same amount was administrated intravenously [40]. This difference in tissue distribution may be due partially to the polysaccharide nature of some nanoparticles. Such polysaccharides can be recognized by receptors in specific tissues, such as pulmonary tissue. However, the difference of metabolism passage may also explain the distribution difference. Those nanoparticles that are metabolized in liver certainly accumulate in hepatocytes. Therefore, even though nanosponges are generally the best, other nanoparticles should also be developed to find if they are tissue specific.

RDO delivery systems with more than one modification

Oligonucleotide delivery is a multistep process, and must pass a series of barriers. Basic carrier systems normally have difficulties with targeting. To improve this, delivery systems should be decorated with ligands for a variety of purposes, such as tissue targeting, endosomal release and nuclear targeting. Tissue targeting has been realized in RDO transfer, but other modifications, which have been used in antisense oligonucleotide delivery, still need testing on RDO transfer.

Endosomal release is a rate-limiting step of gene delivery. Oligonucleotides must be released from the endosome before entering the nucleus. However, if oligonucleotides are released too early, they will have difficulties in cytoplasmic transport. The best place for endosomal release is the perinucleus. Many fusogenic or pH-sensitive agents are attached to delivery vectors, to facilitate endosomal release [9,20]. DOPE (dioleoylphosphatidylethanolamine) has the ability to form nonbilayer phases and promote destabilization of the bilayer of the endosome membrane. DOPE is added to cationic liposomes and other delivery systems to control endosome rupture. Biodegradable pH-sensitive surfactant (BPS) typically has a lysosomotropic head (pKa 5–7) and a hydrophobic tail [20]. BPS is stable at alkaline conditions. However, in the acid environment of the endosome, as the pH value decreases, BPS will be ionized and will destabilize the endosome membrane. Therefore, BPS is an ideal strategy of controlling endosomal release.

The nuclear membrane of eukaryotic cells is a barrier for chemicals more than 9 nm (for macromolecules greater than 70 kDa) [42]. Nuclear signal peptides can be irreversibly linked to one end of oligonucleotides, forming oligonucleotide–peptide conjugate [43,44]. The nuclear signal peptide allows effective transfection with minute quantities of DNA. Transfection enhancement (10–1000-fold) as the result of the signal peptide was observed irrespective of the cationic vector or the cell type used. Therefore, nuclear signal peptides are a putative strategy for nuclear location and penetration.

RDO is also reported to correct point mutations in mitochondria isolated from hepatocytes, indicating that mitochondria have the machinery required for the repair of single-point mutations [45]. In order to understand the mechanism of RDO for targeted gene correction and apply them to mitochondrial gene therapy, it is essential to develop mitochondrion-specific delivery systems for RDO. Two methods used in plasmid transfer show prospects of delivering oligonucleotides into mitochondria [46–48]. Mitochondriotropic vesicles are cationic amphiphiles containing a hydrophilic charged center and a hydrophobic core, capable of transferring nucleic acids to mitochondria [46]. This strategy is easy to handle, and may be used for RDO delivery. Another strategy is the conjugation of mitochondrial signal peptides to oligonucleotides, similar to nuclear signal peptides. By imitating mitochondrial entry of polypeptides that are synthesized in the nucleus and function in mitochondria, gene transfer is realized [48]. However, this technology is still in its infancy.

However, one modification usually helps to overcome only one hurdle. In this regard, modifying the delivery systems with two or more ligands is a promising method to construct an all-round delivery system, which can pass all barriers smoothly. One example is PEI modified with both polysaccharides and DOPE. The polysaccharide ligand is for cell targeting, and the DOPE ligand is for controlling endosomal release. To date, the most efficient decoration for endosomal release is BPS, while the ligands for tissue targeting differ in different tissues. Good RDO carrier systems may combine these two advantages. Additionally, nanoparticles and nanosponges should also be modified with specific ligands for tissue distribution and targeting [39].

Designing RDO specific delivery system

It is possible that a novel RDO specific delivery system, which is both safe and efficient, will be invented in the near future. This may be a new polysaccharide-based delivery system or RDO conjugated to a short peptide with a special function. Particluar attention should also be given to new encapsulating nanoparticles. By these methods, not only transient transfection, but also prolonged and controlled-release transfection may be realized. A delivery device especially suitable for RDO transfer is also possible. Because RDO has an RNA sequence, molecules that can bind with both RNA and DNA oligonucleotides may be putative complexing agents for RDO delivery.


The authors would like to thank Dr Xue-Song Wu, and Chang-Mei Liu for suggestions on the manuscript. This work is supported by the Chinese High-Technology (863) Program 2001AA217171 and NSFC/RGC 3991061991.