Current Topics : Reviews
Chemical Ligation Reactions of Oligonucleotides for Biological and Medicinal Applications
2018 Volume 66 Issue 2 Pages 117-122
Details
2018 Volume 66 Issue 2 Pages 117-122
Chemical ligation of oligonucleotides (ONs) is the key reaction for various ON-based technologies. We have tried to solve the problems of RNA interference (RNAi) technology by applying ON chemical ligation to RNAi. We designed a new RNAi system, called intracellular buildup RNAi (IBR-RNAi), where the RNA fragments are built up into active small-interference RNA (siRNA) in cells through a chemical ligation reaction. Using the phosphorothioate and iodoacetyl groups as reactive functional groups for the ligation, we achieved RNAi effects without inducing immune responses. Additionally, we developed a new chemical ligation for IBR-RNAi, which affords a more native-like structure in the ligated product. The new ligation method should be useful not only for IBR-RNAi but also for the chemical synthesis of biofunctional ONs.
Chemical ligation of oligonucleotides (ONs) connects two ONs by chemical reaction and has been widely used in the field of oligonucleotide chemistry and biology1,2) (Fig. 1). The ligation reaction normally requires a third ON, called a template strand, that has sequence complementary to that of the ligation strands. The dependency on the template strand is utilized for RNA detection probes, where the fluorogenic chemical reaction is coupled with the ligation reaction.3) The chemical ligation reaction is also useful for the synthesis of artificially modified functional ONs, such as molecular probes or the ON-based nanostructure,4,5) for which enzymatic ligation reactions cannot be applied because the non-natural structure hinders recognition of the enzyme.
Among many applications of the ligation reaction, we have been especially intrigued by chemical ligation in the biological context; that is, the construction of biofunctional ONs in cells. Here, we focus on the development of the ligation for a new RNA interference (RNAi) system and its application.
RNAi is a target-specific gene-silencing phenomenon6,7) now widely used for gene knockdown and is regarded as a promising method as therapeutics. Figure 2 illustrates a simplified scheme of the RNAi mechanism. RNAi is mediated by short (20–23-mer) double-stranded RNA, which is referred to as small-interference RNA (siRNA). siRNA is loaded on an RNA-processing enzyme complex, called the RNA-induced silencing complex (RISC), and one of the strands is removed. The RISC containing short RNA fragments accesses the target mRNA, and this target is degraded by the RISC. This results in target-specific gene knockdown.
Although this RNAi is reliable and widely utilized, there are several significant problems that need to be solved. The first significant problem is the low membrane permeability of the double-stranded RNA. Another problem is the induction of immune responses that can lead to side effects of the siRNA.8) In order to solve these problems of siRNA, we designed the new RNAi system based on chemical ligation reactions of fragmented RNA in cells.9) The concept of our new RNAi system is summarized in Fig. 3.
First, the active siRNA is fragmented into shorter strands. By using short fragments, the problems of the permeability and immune responses can be circumvented. The fragments have reactive functional groups at their terminals. After entering the cells, those fragments react with each other through chemical ligation to afford active siRNA or its equivalent, and the expected RNAi effect is induced. In this system, the active siRNA is built up in the cells, and therefore is referred to as intracellular buildup RNAi (IBR-RNAi).
The appropriate choice of reactive functional groups (FGs) is important to achieve IBR. If the reactivity is too high, side reactions with biomolecules in cells will be problematic, and if the reactivity is too low, active siRNA will not be generated. Based on the reactivity reported to date,10,11) we selected the phosphorothioate (PS) group as the nucleophile and iodoacetyl (IAc) group as the electrophile. The PS group can be introduced easily using a sulfur-based oxidant in an automatic ON synthesizer, and the IAc group can be introduced at a terminal oligonucleotide having an amino group at the 5′- or 3′-terminal. The RNAi system designed based on the combination of these reactive FGs is shown in Fig. 4.
The key point is the protection of the PS group masked as a disulfide structure to prevent undesired reactions outside the cells. These RNA fragments can be efficiently uptaken in the cells without immune responses due to their small size. The disulfide structure of PS-ON is deprotected by glutathione (GSH), which exists in high concentrations (1–10 mM) in cells. The four fragments are organized based on their complementary sequences, and the reactive functional groups are positioned in close proximity. Then, the naked nucleophile PS group attacks the electrophilic IAc group of the other fragment, resulting in the formation of the siRNA equivalent structure in cells.
Based on this design, we prepared fragmented strands for the IBR system. The structures of the ONs used in this study are shown in Fig. 5. For the evaluation of siRNA activity, we selected the dual luciferase assay, and the positive control siRNA [25 mer, siRNA-n-1 (A)] is designed to target the luciferase gene. The fragment length is 19+6 mer and 18+7 mer to prevent the recognition by the Toll-like receptor (based on previous reports).12) PS strands (1, 2) were protected as a phenyl-disulfide structure, and the IAc-strands were prepared by introducing 5′-amino-dT and reacting the strands with the iodoacetic acid N-hydroxysuccinimide ester. As the negative control of the IAc strands (nonreactive strands), we also prepared acetyl-terminated strands (5, 6). The structures of the resulting ligated strands and their negative control (non-ligated) are described as siRNA-l-1 (E) and segmented siRNA (F), respectively.
First, we checked whether the target ligation reaction could proceed in vitro. The RNA fragments were incubated at room temperature in the presence/absence of dithiothreitol (DTT). The samples were analyzed by denaturing-polyacrylamide gel electrophoresis (PAGE) (Fig. 6). In the presence of GSH (lane 4), ligated strands were formed, while ligation did not proceed in the absence of GSH (lane 5). The ligation reaction required template strands; the sense strand or antisense strand alone did not afford the ligation product (lanes 6, 7).
After confirming that the ligated strands could be processed by Dicer,9) we evaluated RNAi efficiency in the cell system with HeLa cells stably expressing firefly luciferase genes (Fig. 7). Natural siRNA (siRNA-n-1) showed about 60% knockdown at the concentration of 25 nM (entry 2), while the siRNA-lig-1 (prepared in vitro by chemical ligation) showed about 40% knockdown at the same concentration (entry 3). With administration of the IBR fragments at 25 nM (entry 4), weaker knockdown efficiency was observed, but comparable results were observed when increasing the concentration of the fragments up to 100 nM (entries 5, 6). When using non-reactive fragments (1+2+5+6), no knockdown was observed (data not shown),9) and therefore this RNAi effect is ligation dependent as designed.
Next, we evaluated immune responses using T98G cells. The immune response level was evaluated by quantitative RT-PCR of interferon-β (IFN-β) mRNA (Fig. 8). Native siRNA (100 nM) showed nearly 4-fold stronger immune responses than the positive control polyinosinic : polycytidylic acid (polyI : C) (0.105 ng/µL), while the segmented siRNA (100 nM) or IBR-RNA (100 and 500 nM) showed almost no immune response. These results clearly showed that IBR can induce a nearly comparable RNAi effect as natural siRNA without inducing immune responses.
Although our previous IBR system was useful, several practical problems remained. The most significant drawback of the IBR system was the formation of the unnatural backbone structure by the ligation reaction between the PS and IAc groups. Due to its unnatural structure, it created significant limitations on the segmentation position. If we selected a near-central position of siRNA for segmentation, the resulting ligated product showed very weak RNAi effects. This limitation in the design hampered the further application of IBR to general RNAi targets. In light of this limitation, we planned to develop a new chemical ligation reaction that could be applied to intracellular ligation and afford a more natural linkage structure.
The most established ON ligation reaction that affords a native-like or native structure itself is based on phosphate-activation methods, where the phosphate-terminal of one strand is activated by condensation reagents such as 1-ethyl-3-(dimethylaminopropyl)carbodiimide hydrochloride13–15) and the resulting active ester is reacted with the hydroxyl- or amino-terminal of the other strand. While promising, this ligation reaction is not suitable for applications in cells, because it requires excess amounts of toxic condensation reagents and the intermediate (active ester) is very labile to hydrolysis. For designing a new chemical reaction for ONs, we referred to one of the peptide ligation reactions. In this reaction, a thioacid-terminated peptide is treated with the electrophilic reagent 2,4-dinitro-fluorobenzene (DNFB) to form an active ester, which was reacted with the amino-terminal of the second peptide to afford ligated peptides with amide bonds.16) We envisioned that a similar transformation might be possible for PS-terminated and amino-terminated ONs (Fig. 9) and that the corresponding intermediate (active phosphorothioester) might be stable for hydrolysis and nucleophiles in cells, thus being suitable for the intracellular ligation reaction. The resulting phosphoramidate bond is stable under biological conditions and can function completely as a replacement for the native phosphodiester bond because of their high structural similarity.17–20) Based on this design, we began to develop the new reaction.
First, we checked whether the active ester could be generated from PS-terminated ON and DNFB. PS-ON was treated with an excess amount of DNFB for 30 min, and subsequent HPLC analysis of the reaction mixture showed almost quantitative conversion (data not shown).21) The resulting electrophilic phosphorothioester-ON (EPT-ON) showed a practical level of stability; it could be isolated in HPLC and stored at −30°C without degradation. At room temperature at pH 7.0, the half-life was nearly 8 h, which is suitable for subsequent biological applications like IBR.
Next, we performed the ligation reaction using 10-mer 3′-EPT-DNA, 13-mer 5′-amino DNA, and 27-mer template DNA as substrates at several pH values from 6 to 8 at room temperature. The ligation results were analyzed by denaturing PAGE, as shown Fig. 10. The ligations showed clear dependency on the pH of the reaction system; pH 8 was optimal, where most (75%) of the 5′-amino DNA was converted into the ligated product after 2 h, while only trace amounts (23%) of the product were observed after 2 h at pH 6.21)
Kinetic analysis of the ligation reaction suggested that this pH dependency was due to the protonation state of the terminal amino group of ON; at pH 6 most of the amino group was protonated, while at pH 8 most of it was unprotonated, thus resulting in higher efficiency of the reaction. DNA ligation in the reverse direction (i.e., with 5′-EPT-DNA and 3′-amino DNA) was also successful and afforded very similar reaction profiles.
This EPT-ON-based ligation was also applied to the RNA strands (Fig. 11); there was no significant difference in reactivity between DNA and RNA (75% yield in 2 h). The only difference required for RNA ligation was the use of a modified nucleotide for the ligation of 3′-EPT RNA and 5′-amino RNA; the 2′-OMe derivative should be used for the 3′-terminal of the 3′-EPT RNA strand in order to prevent intramolecular cyclization to afford 2′,3′-cyclic phosphate. Expect for this point, the ligation could be performed without changing the reaction conditions, suggesting that the reaction developed is a fairly universal method for ON ligation.
In addition to short ON strands, the ligation system was successfully applied to the synthesis of long DNA strands (113 mer from 67 mer and 45 mer, Fig. 12). Although the reaction required a longer time for completion, and the yield (41%, 19 h) was not as high as that of the ligation with shorter strands, the long DNA strands, which are difficult to synthesize with an automatic synthesizer, can be prepared successfully.
The possibility of a side reaction occurring with DNFB treatment of ONs was checked by the sequence analysis of the ligated product. The analysis showed that DNFB treatment did not induce severe mutations in DNA. These results suggest that the newly developed reaction is also useful for functional long ON synthesis, which could be useful molecular probes in biology research. After establishing the new ligation reaction, we have been engaged in its applications to buildup RNAi.
The chemical ligation reaction of ON is fundamental to various biological and medicinal applications of ON. However, the reaction types of chemical ligation have been limited so far, and thus the application range was rather narrow. By developing new types of chemical ligation, we can widen the range of application, and IBR RNAi is a typical example. By incorporating the ligation reaction in RNAi, we were able to solve some important problems associated with RNAi technology, that is, the immunogenicity of siRNA. Many attempts have been made for the improvement of RNAi technology, but most involved structural optimization of siRNA molecules. In contrast to this traditional approach focusing on the static profile of siRNA, we have tried to incorporate dynamic processes in RNAi (i.e., chemical ligation), and thus higher-order function could be implemented in RNAi. We believe that this approach focusing on dynamic aspects of ON-based methods should be notably effective in improving ON-based biotechnology and medicine and will continue to advance research in this direction.
We sincerely appreciate all the students and collaborators in the studies presented here, especially Dr. Hideto Maruyama, Mr. Shono Takamori, Prof. Satoshi Shuto, and Prof. Akira Matsuda (Hokkaido University); Dr. Yuko Nakashima and Prof. Yoshihiro Ito (RIKEN); and Dr. Naoko Abe, Dr. Genichiro Tsuji, Mr. Ryota Oikawa, and Ms. Mayu Hayakawa (Nagoya University).
The authors declare no conflict of interest.
