Review
Developing Methods for the Synthesis of Ligand–Oligonucleotide Conjugates for Drug Discovery Applications
2025 Volume 73 Issue 9 Pages 762-771
Details
2025 Volume 73 Issue 9 Pages 762-771
Recently, oligonucleotide-based drug discovery has attracted considerable amounts of attention. As oligonucleotide therapeutics have evolved into practical use, research into the development of functional artificial nucleic acids has been vigorously conducted worldwide. However, the synthesis of artificial nucleic acids generally requires long sequences from starting materials; hence, structurally optimizing oligonucleotide therapeutics is extremely difficult. In response to this challenge, we have been developing reactions that use oligonucleotides as starting materials. As a part of this work, we focused on ligand–oligonucleotide conjugates because the conjugates of functional ligands and oligonucleotides have attracted attention as drug-delivery systems that improve the efficacies of oligonucleotide therapeutics; they are also DNA-encoded-library-based drug-discovery tools. In this review, we broadly introduce our research into ligand–oligonucleotide conjugates.
Research and development techniques for oligonucleotide therapeutics, including those associated with antisense oligonucleotides (ASOs) and small interfering RNAs (siRNAs), have advanced rapidly over the past few years.1,2) Oligonucleotide therapeutics are generally chemically modified because natural DNAs and RNAs not only lack in vivo stability, but also the ability to hybridize with the therapeutic target RNA. Consequently, various types of chemically modified nucleic acid have been developed to address these issues.3–5) For example, locked nucleic acids (LNAs)6,7)/2ʹ,4ʹ-bridged nucleic acids (2ʹ,4ʹ-BNAs)8,9) can form stable duplexes with complementary RNAs because they contain structurally restricted N-type sugars, which are preferred when forming stable duplexes with target RNAs (Fig. 1A). Additionally, the 2ʹ,4ʹ-BNA modification enhances the in vivo stability of an oligonucleotide, probably by avoiding nuclease recognition through steric hindrance from the incorporated bridge structure. Hence, 2ʹ,4ʹ-BNA-modified-oligonucleotide therapeutics have already been used in clinical trials owing to these characteristics.10) On the other hand, chemical modification of the phosphate backbone of an oligonucleotide therapeutic is also an effective way of enhancing its in vivo stability. For example, the phosphorothioate moiety, which is typically used to modify a backbone, is essential for enhancing oligonucleotide therapeutic efficacy and is used in most approved ASOs and siRNAs.11) Therefore, we have directed effort toward developing artificial nucleic acids modified with sugar and backbone moieties12–25) (Fig. 1B) with the aim of developing functional artificial nucleic acids that can be used in oligonucleotide therapeutics.
These artificial nucleic-acid-modified oligonucleotide therapeutics may require structural optimization to maximize their clinical efficacies. However, artificial-nucleic-acid syntheses almost always require more than ten steps from the starting material; consequently, the derivative syntheses of artificial nucleic acids required for the structural optimization of oligonucleotide therapeutics is extremely difficult. In response to this challenge, we aimed to develop reactions from oligonucleotides as starting materials. On the other hand, we also believed that subjecting oligonucleotides to artificial-nucleic-acid-selective structural transformations would be difficult. Therefore, we focused on ligand–oligonucleotide conjugates because conjugates of functional ligands and oligonucleotides have attracted attention not only as a drug-delivery system (DDS) that improve the efficacies of oligonucleotide therapeutics, but also as drug-discovery tools using DNA-encoded libraries (DELs). The conformational conversion of a ligand molecule into a ligand conjugate is an essential DEL technique, and many reactions have been reported for DEL syntheses.26,27) Against this background, we developed methods for synthesizing ligand–oligonucleotide conjugates for DEL-based drug discovery and DDSs suitable for oligonucleotide therapeutics. As a part of our efforts, we reported a 3′-aminolinker-modification method for oligonucleotides with base-labile nucleotides.28) This method enables an oligonucleotide therapeutic to be modified with an amido-bridged nucleic acid (AmNA) as a 3′-aminolinker, which is difficult to achieve using conventional methods.29) In addition, as a new synthetic method oriented toward DDS applications for oligonucleotide therapeutics and drug discovery using DELs, we developed a single-step method for constructing oligonucleotides conjugated to bioactive coumarins.30) Furthermore, we developed a simple conjugation method based on the Ugi reaction to prepare structurally diverse dipeptide–oligonucleotide conjugates,31) and explored dipeptide ligands that enhance the activities of nucleic-acid-based drugs in vitro.32) This review presents these three studies into ligand–oligonucleotide conjugates (Fig. 2).
Many chemical reactions are used to synthesize ligand–oligonucleotide conjugates. Amino-linker-modified oligonucleotides are used as starting materials in conjugation reactions that form amide bonds through reactions involving carboxylic acids and appropriate condensing agents. In addition, because a variety of carboxylic acids are readily commercially available, this reaction is suitable for synthesizing structurally diverse ligand–oligonucleotide conjugates. In most cases, while the amino linker is modified at the 3′- or 5′-end of the oligonucleotide using this reaction, ligand conjugates modified at their 3′-ends are particularly useful because such oligonucleotides are highly resistant to nucleases.33) Oligonucleotides modified with amino linkers at their 3′-ends are conventionally synthesized using the phosphoramidite method via a solid-support immobilized with an amino linker protected with a phthaloyl group34,35) (Fig. 3A). The phthaloyl group is removed concurrently with the nucleobase protecting group using 28% ammonia solution. However, this method is not applicable to oligonucleotides, including chemically modified nucleic acids that are unstable under strongly basic conditions because phthalimide ring opening requires prolonged heating (55 °C for 17 h). Therefore, we aimed to develop a new 3′-amino-linker-modification method for oligonucleotides that are sensitive to strong bases. Amines are commonly protected using the 9-fluorenylmethoxycarbonyl (Fmoc) group because it is quickly removed by treatment with a weak base. Therefore, we designed and synthesized phosphoramidite reagents for 3′-amino-linker-modification chemistry based on the Fmoc group (Fig. 3B). The protecting groups can be removed under mildly basic conditions after introducing a phosphoramidite into an oligonucleotide; hence, a base-labile nucleotide can be used to 3′-amino-linker-modify the oligonucleotide.
(A) Conventional synthetic method using a solid-support immobilized with a phthaloyl-protected amino linker. (B) Our method using designed reagents 1a–c.
Model oligonucleotides (dC decamers) were synthesized using phosphoramidite 1a, and their deprotection efficiencies were investigated (Fig. 4, Table 1). In this experiment, the removal of the protecting group of the amino linker was examined under the general conditions for deprotection of oligonucleotides. Initially, the synthesized solid-supported dC decamer was treated with 28% aqueous ammonia, which required 8 h to completely remove the Fmoc-based protecting group at room temperature (Entry 1). Deprotection efficiency was improved using 40% methylamine solution, which only required 2 h at room temperature to completely remove the protecting group (Entry 2), a significantly shorter time than that required under the conventional conditions for removing the phthaloyl group (28% aqueous ammonia at 55 °C for 17 h). The reaction temperature was also successfully lowered from that used under the conventional condition. To explore the applicability of this method for 3′-amino-linker-modifying oligonucleotides with base-labile nucleotides, we compared deprotection conditions using mild basic solutions to those involving aqueous methylamine or ammonia solutions. The protecting group was completely removed over 24 h at room temperature when 50 mM potassium carbonate in methanol was used (Entry 3). Moreover, a 1 : 1 : 2 mixture of tert-BuNH2/MeOH/H2O (as the solvent) was also effective (Entry 4).
| Entry | Base | Temp. | Time |
|---|---|---|---|
| 1 | 28% NH3 aq. | r.t. | 8 h |
| 2 | 40% MeNH2 aq. | r.t. | 2 h |
| 3 | 50 mM K2CO3 in MeOH | r.t. | 24 h |
| 4 | tert-BuNH2/MeOH/H2O (1 : 1 : 2) | 65 °C | 3 h |
We synthesized oligonucleotides modified with TAMRA-T and an amido-bridged nucleic acid (AmNA), which react with aqueous ammonia, to examine the use of the synthesized phosphoramidites 1a–c for 3′-amino-linker-modifying oligonucleotides that cannot be treated with strong bases (Fig. 5A). The TAMRA-T-modified oligonucleotide was prepared without decomposition via deprotection using the tert-BuNH2/MeOH/H2O solvent system (Entry 4, Table 1). The AmNA-modified oligonucleotide was deprotected using 50 mM potassium carbonate in methanol (Entry 3, Table 1), which removed the protecting group without ring-opening the amide bridge in AmNA. We also compared AmNA- and TAMRA-T-modified oligonucleotides synthesized by the conventional method using a solid support with an immobilized phthaloyl-protected amino linker and reverse-phase HPLC. The modified oligonucleotides synthesized by the conventional method were treated with aqueous ammonia for 17 h at 55 °C, and the products resulting from the reactions of the TAMRA-T and AmNA moieties with ammonia were confirmed. Specifically, TAMRA-T was almost completely converted to its lactam form (Fig. 5B), while ammonolysis of the amide bridge in the AmNA moiety was observed (Fig. 5C). In contrast, few peaks derived from such side reactions were observed for the oligonucleotides modified using our method, which highlights the superiority of our synthetic method for 3′-aminolinker-modifying oligonucleotides.
(A) Deprotecting TAMRA-T- and AmNA-T-modified oligonucleotides using our developed phosphoramidite reagents 1a–c. Comparing HPLC traces of the (B) TAMRA-T- and (C) AmNA-T-modified oligonucleotides.
Combinatorial chemistry and high-throughput screening (HTS) methods led to a major drug-discovery paradigm shift in the early 1990s. The DEL concept, one of many HTS technologies, was introduced by Brenner and Lerner in 199236) and used to synthesize oligonucleotide-tagged peptides by Janda and Brenner in 1993.37) Each drug-candidate compound is covalently linked to a barcode DNA using a DEL, where the DNA strand functions as an identification barcode for compound hits. DEL-based drug discovery generally follows the sequence: 1) constructing the DEL, 2) isolating molecules that bind strongly to target molecules, 3) identifying compound hits via DNA barcode sequence analysis, and 4) synthesizing hit compounds and evaluating their binding activities. DELs are powerful platforms for discovering ligands in the early stages of drug discovery because they do not depend on the target protein and can simultaneously evaluate the binding activities of multiple organic compounds.
The coumarin (2H-chromen-2-one) scaffold is found in warfarin, umbelliferone, and todacrine, all of which have pharmacological properties. Because coumarins are attractive compounds, we hypothesized that constructing a coumarin skeleton on DNA is a useful DEL-based synthesis and drug-discovery methodology. Many synthetic methods have been developed for constructing coumarin scaffolds.38–44) As mentioned above, DNA barcodes are indispensable for identifying compound hits during DEL-based drug discovery. As a typical example of chemical damage to DNA, adenine and guanine in DNA are well-known to be removed by cleaving glycosyl bonds under acidic conditions,45) while some transition-metal compounds also cleave DNA.46) In view of these chemical properties of nucleic acids, selecting reactions that do not damage DNA during a DEL-based synthesis is important. Among the many methods for constructing the coumarin skeleton, that based on the Knoevenagel condensation reaction,38,39) which is transition-metal-catalyst-free and constructs coumarin skeletons from malonates and salicylaldehydes under weakly basic conditions, is considered suitable for synthesizing coumarin-containing DELs. Therefore, we developed a method for constructing the coumarin scaffold using malonate-modified DNA as the substrate.
3-1. Synthesizing Coumarin–Oligonucleotide ConjugatesMalonate-modified DNA2, as a starting material for the Knoevenagel condensation, was prepared by condensing N-hydroxysuccinimide (NHS)-ester 1 with DNA1 bearing an aminohexyl linker at its 5ʹ-end, followed by Knoevenagel condensation with salicylaldehyde as a model substrate (Fig. 6). DNA2 reacted completely with 10 µmol of salicylaldehyde, 2 µmol of piperidine, and 20 nmol of DNA2 to produce DNA3a as the desired conjugated coumarin in an isolated yield of 68%. Salicylaldehyde substrate scope and limitations in this reaction were next investigated (Fig. 6) using various salicylaldehydes bearing electron-withdrawing groups, such as nitro and carboxyl, and electron-donating groups, such as hydroxyl and methoxyl. Lower yields were obtained when salicylaldehydes bearing electron-donating groups were used. In particular, desired products DNA3i, 3n, and 3u were not obtained when salicylaldehydes substituted with hydroxyl groups were used. In contrast, Knoevenagel condensation and subsequent intramolecular cyclization proceeded with other substrates to afford the desired coumarin–DNA conjugates in yields of up to 79%. Our method not only facilitated the introduction of most coumarins with substituents at positions 5–8 of the scaffold, but also enabled tricyclic fused coumarins DNA3x and DNA3y to be constructed. Additionally, we were able to synthesize DNA3p and DNA3z bearing 7-aminocoumarin derivatives, which are useful fluorescent materials.
We explored the applicability of our method to DEL-based drug discovery by preparing a model DEL (Chart 1). Here, DNA4, as an amino-linker-modified hairpin duplex, was used as the starting material. Here, construction of the model DEL involved three steps: 1) condensation with 2 followed by deprotection of the Fmoc group, 2) condensing the terminal benzylamine in DNA5 with reagent 1, and 3) Knoevenagel condensation. The DNA5 model coumarin molecule was synthesized and barcode DNAs dsDNA1–3, which correspond to reactions 1)–3), respectively, were connected using T4 ligase. While DELs are generally purified using simple operations, such as ethanol precipitation, to provide structurally diverse ligands, the model DEL required reverse-phase HPLC in each reaction owing to less-than-quantitative reaction efficiencies. Furthermore, nucleobase damage resulting from reactions involving alkylating reagents and electrophiles directly led to the loss of coding information in the DEL.47) The DEL constructed using our method required detailed DNA-sequence analysis by liquid chromatography-tandem mass spectrometry (LC/MS) to determine nucleobase damage because the Knoevenagel condensation reaction involves reactive aldehydes. Although some problems remain to be solved, the synthetic method developed by us is useful for synthesizing DNAs conjugated to a variety of coumarins in one step; consequently, we believe that our method is applicable to DEL-based drug-discovery research.
As mentioned in the Introduction, while oligonucleotide therapeutic efficacy can be enhanced through the use of artificial nucleic acids, some issues that are difficult to resolve by chemically modifying the nucleic acids alone remain. For example, oligonucleotide therapeutics are poorly cell-membrane permeable and rarely reach the cytoplasm or nucleus, where their target RNAs reside.48,49) In addition, most oligonucleotide therapeutics administered systemically accumulate in the liver, which makes it difficult to target organs other than the liver. Consequently, efficiently delivering an oligonucleotide to its therapeutic target and cells requires the urgent development of a DDS suitable for oligonucleotide therapeutics. Ligand conjugation has been actively investigated in recent years in attempts to deliver oligonucleotide therapeutics.50–52) Generally, ligand molecules for the receptors on cell-membrane surfaces are conjugated to oligonucleotide therapeutics, and are often studied as parts of strategies that impart membrane permeability and target-tissue specificity. However, these receptor-targeted ligand conjugates are not always successful; consequently, we aimed to identify DDS ligands that function through novel mechanisms, not found in conventional ligand conjugates, by screening more than 100 ligand conjugates and ten cell lines. To that end, we focused on a multicomponent reaction (MCR) to identify suitable DDS ligands for oligonucleotide therapeutics from a group of ligands with random structures. MCRs are reactions that yield single compounds from three or more reagent components and are useful for synthesizing diverse derivatives. Among the various MCRs developed to date, the Ugi reaction is used to construct structurally diverse dipeptide skeletons from carboxylic acids, aldehydes, amines, and isocyanides. Therefore, we developed a method for conjugating Ugi-reaction products synthesized from carbamate-modified carboxylic acids that are moderately reactive and stable to oligonucleotides. The prepared library, which contained 130 dipeptide–oligonucleotide conjugates was then subjected to in-vitro screening.
4-1. Synthesizing Dipeptide–Oligonucleotide ConjugatesInitially, 20 nmol of amino-linker-bearing DNA1 was reacted separately with purified dipeptides 10–13 bearing various leaving groups, which were prepared via Ugi reactions of carbamates 3–6 with compounds 7–9 in the presence of DIPEA (Chart 2, Table 2). The reactivities of dipeptides 10–13 toward DNA1 were compared (Entries 1–4), which revealed that conjugation proceeded well when DNA1 was reacted with compound 13 bearing the 4-nitrophenyl group, with conjugated product DNA9 obtained in 31% isolated yield (Entry 4). Subsequently, 50 equiv. of compound 4 were required to completely convert DNA1 into DNA9 (Entry 5). Based on the results, we prepared crude dipeptide 13 by the Ugi reaction of N-(nitrophenylcarbamoyl)-6-aminohexanoic acid (6) with DNA1, which was not further purified; the desired conjugate was successfully obtained in 33% yield.
| Entry | Dipeptide reagent | DNA9a) | DNA1a) |
|---|---|---|---|
| 1 | 10 (100 nmol, 5 equiv.) | 0% | 93% |
| 2 | 11 (100 nmol, 5 equiv.) | 0% | 96% |
| 3 | 12 (100 nmol, 5 equiv.) | 10% | 74% |
| 4 | 13 (100 nmol, 5 equiv.) | 31% | 40% |
| 5 | 13 (1000 nmol, 50 equiv.) | 87% | 0% |
a) Isolated yield after HPLC-purification.
We next synthesized a conjugate library under the optimized conditions; here we used two carboxylic acids, eight aldehydes, 16 amines, four isocyanides, and an amino-linker-modified ASO (Chart 3). Many functional groups, including aryl, amino, hydroxyl, and methoxycarbonyl, were introduced under the optimized conjugation conditions to afford 130 different dipeptide–ASO conjugates in yields of up to 83%. However, low isolated yields were obtained in some of the reactions examined. For example, the desired conjugate was not obtained when 1,1-dimethylpropanediamine was used as the substrate in the Ugi reaction, which is possibly ascribable to inhibited iminium-ion formation (which is a key step in the Ugi reaction) by the tertiary amine (Chart 4). Based on this concept, we added one equivalent (equiv.) of hydrochloric acid during the Ugi reaction, which dramatically improved the yield of the conjugate to 32%.
The abovementioned 130 conjugates were subjected to in-vitro screening, in which 10 cell lines transfected with the luciferase gene were treated with ligand–ASO conjugates and their knockdown activities were evaluated based on the degree of attenuation of the luciferase-derived luminescence intensity as an indicator. In addition, this evaluation method maximizes the effect of the ligand by enabling ASO to be taken up by cells without the use of special cell-transfection reagents. As a result, a total of 21 of the 130 conjugates exhibited decreases in luminescence intensity in the 10 cell lines, indicative of increases in ASO activity. Based on these results, we evaluated the dose-dependence of ASO activity and quantified luciferase mRNA levels for these 21 combinations; representative results are shown in Fig. 7, with Fig. 7A showing relative luminescence intensities normalized against that of the ASO-untreated group, while Fig. 7B shows relative mRNA levels normalized in the same way. We compared ligand-unmodified ASO and ligand–ASO conjugates, which revealed that, among the 21 combinations identified earlier, UL1-ASO and UL2-ASO showed dose-dependent antisense effects in DU-145 cells, a human prostate-derived cell line, and significantly enhanced activities compared to ligand-unmodified ASO (Fig. 8). In particular, we found that luciferase mRNA levels were significantly suppressed by UL2-ASO. We quantified and compared the intracellular uptake of UL2-ASO bearing the promising ligand UL2 identified during screening with that of unmodified ASO, which verified the target-RNA-sequence-dependence of the activity-enhancing effect of UL2. Unfortunately, UL2-ASO exhibited little change in intracellular uptake and no effect on ASO targeting in the case of human MALAT1 RNA. Therefore, we intend to clarify the mechanism responsible for the in-vitro ASO activity observed when conjugated with UL2 in future studies.
(A) Dose dependence of luminescence intensity and (B) relative luciferase mRNA levels determined by qPCR (no ASO, treatment without ASO; ASO with LA, treatment with ASO and lipofectamine 3000 (positive control); ASO, ASO without ligand modification (negative control); UL1-ASO and UL2-ASO, ligand–ASO conjugates. Mean values and SDs were determined from triplicate measurements; *: p < 0.05).
We focused on ligand–oligonucleotide conjugates intended for DDSs for use in oligonucleotide therapeutics and DEL-based drug discovery. We developed a method for 3′-amino-linker-modifying oligonucleotides with base-labile nucleotides, which has been difficult to achieve using conventional synthetic methods, as well as methods for synthesizing ligand–oligonucleotide conjugates based on the Knoevenagel condensation and Ugi reactions. A series of studies identified a dipeptide ligand that enhanced the activity of an oligonucleotide therapeutic. We are currently developing reactions for structurally transforming artificial nucleic acids into oligonucleotides that use the knowledge gained through this research. We are also challenging oligonucleotide permeation through cell membranes, which has been a long-standing issue in the field of oligonucleotide therapeutics, by fully exploiting the reactions currently under development. We intend to report the results of our ongoing research in the future.
I would like to thank all my collaborators, especially Prof. Satoshi Obika, for their support and advice. I would like to thank Dr. Yuuya Kasahara of the National Institute of Biomedical Innovation and Nutrition for conducting in-vitro screening. This work was partially supported by the Japan Society for the Promotion of Science (KAKENHI Grant Numbers: JP20K15401, JP23K04930, JP24H00839, and JP24H00840), and the Japan Agency for Medical Research and Development (Grant Numbers: JP19am0401003, JP21ae0121022, JP21ae0121023, JP21ae0121024, and JP24am0521009).
The author declares no conflict of interest.
This review of the author’s work was written by the author upon receiving the 2025 Pharmaceutical Society of Japan Award for Young Scientists.
