Chemical modification of nucleotides can improve the metabolic stability and target specificity of oligonucleotide therapeutics, and alkylphosphonates have been employed as charge-neutral replacements for naturally-occurring phosphodiester backbones in these compounds. However, at present, the alkyl moieties that can be attached to phosphorus atoms in these compounds are limited to methyl groups or primary/secondary alkyls, and such alkylphosphonate moieties can degrade during oligonucleotide synthesis. The present work demonstrates the tertiary alkylation of the phosphorus atoms of phosphites bearing two 2’-deoxynuclosides. This process utilizes a carbocation generated via a light-driven radical-polar crossover mechanism. This protocol provides tertiary alkylphosphonate structures that are difficult to synthesize using existing methods. The conversion of these species to oligonucleotides having charge-neutral alkylphosphonate linkages through a phosphoramidite-based approach was also confirmed in this study.
Charge-neutral backbone modifications are expected to maintain a high degree of metabolic stability while avoiding this toxicity. Present approaches to obtaining charge-neutral backbones comprise triester-type P–O linkages, several P–N linkages (such as phosphorodiamidate backbones in phosphorodiamidate morpholino oligomers (PMO) and phosphoranylguanidine backbones) and P–C linkages, such as in the case of methylphosphonate (MP) and methoxypropylphosphonate (MOP) moieties. It should be noted that P–C backbones can be associated with challenges related to chemical stability, although high levels of nuclease resistance following the bonding of alkyl groups not found in natural products to phosphorus atoms has been demonstrated. On this basis, we anticipated that sterically-hindered tertiary alkylphosphonates could serve as robust charge-neutral backbones ensuring high chemical stability. The conventional synthesis of oligomers containing P–C backbones involves the preparation of phosphoramidites in which a P–C bond is already introduced. However, to the best of our knowledge, there are presently no techniques for introducing bulky, multi-substituted carbon groups onto a phosphorus atom and then bonding this atom to a phosphoramidite. Moreover, steric hindrance effects might be expected to prohibit the use of tertiary alkylphosphonates in this coupling step. Because alternative approaches to oligonucleotide synthesis have not yet been unexplored, there is a need to develop a fundamentally different approach that is compatible with a high degree of steric hindrance and with a wide range of functional groups.
Carbocations are highly reactive and hence can be employed to attach bulky alkyl substituents to heteroatom centers. However, present-day methods require the use of strong acids to generate the carbocation species, which limits the range of functional groups that can be accepted. Newer electrochemical and photochemical approaches have enabled the generation of carbocations under milder conditions without the use of strong acids as a means of forging C(sp3)–heteroatom bonds with various heteroatom nucleophiles. Our own group previously developed a light-driven radical-polar crossover (RPC) protocol that combines visible light-mediated photoredox catalysis with an RPC mechanism and requires only mild conditions. In this process, a photo-excited benzo[b]phenothiazine donates an electron to a redox-active ester derived from a carboxylic acid in response to visible light irradiation to form a benzo[b]phenothiazine radical cation along with an alkyl radical, with the liberation of carbon dioxide. Subsequently, the alkyl radical is oxidized by the radical cation with the simultaneous combination of these species to form an alkylsulfonium compound. This alkylsulfonium species can subsequently react with a nucleophile, serving as a carbocation analogue. This method permits various tertiary and secondary alkyl substituents to be attached to heteroatom nucleophiles without the use of strong acids, such that a number of different highly functionalized molecules can be obtained. We envisioned that the carbocation equivalent (that is, the alkylsulfonium) generated by the light-driven RPC protocol could be applied to the tertiary alkylation of the phosphorus atoms of oligonucleotides. To assess the viability of this process, the present work examined the Michaelis-Arbuzov-type alkylation reactions of phosphites bearing two deoxynucleosides and a suitable leaving group with aliphatic carboxylic acid-derived redox-active esters via a visible-light-driven RPC mechanism. This reaction is believed to proceed via the nucleophilic attack of the phosphorus atom of the phosphite on the carbocation equivalent, followed by a β-elimination associated with the loss of the phosphonium species. Recently, several alkylphosphonates have been prepared by electrochemical or photochemical approaches involving reactions with radicals or carbocation species and phosphites.
With the optimal reaction conditions in hand, the ranges of phosphites and redox-active esters that could be employed were investigated. Various phosphites were initially examined together with a 2-phenylisobutyric acid derivative (2a). Phosphites bearing two different 2ʹ-deoxynucleoside scaffolds such as deoxycytidine (3ba–3da), deoxyadenosine (3ea) and deoxyguanosine (3fa) were found to participate in this P-alkylation protocol to afford the corresponding alkylphosphonates. The relatively low yields obtained with 3ea and 3fa could possibly have resulted from the low oxidation potentials of the purine bases. The reaction proceeded smoothly even with a tertiary butyldimethylsilyl (TBS) group as the 3’-OH protecting group (3ca).
Automated solid-phase synthesis of oligonucleotides
The synthesis of oligonucleotides having backbone structures comprising advantageous bulky alkylphosphonate dimers was also demonstrated. Initial attempts involved the synthesis of 5ʹ-O-4,4′-dimethoxytrityl (DMTr)-phosphoramidite (3aa−3) from 3aa, during which the isolated diastereomers (RP)−3aa and (SP)-3aa were converted to the corresponding phosphoramidites. The 5ʹ-O-acetyl and 3ʹ-O-TIPS protecting groups of 3aa were removed in series without purification to give the corresponding diols (RP)-3aa-1 and (SP)-3aa-1. These diols were then reacted with DMTr-Cl to protect the 5ʹ-OH group and obtain the free 3ʹ-OH products (RP)−3aa-2 and (SP)−3aa-2. Following this, a reaction with 2-cyanoethyl N,N-diisopropylchlorophosphoramidite gave phosphoramidites with RP or SP phosphorous centers for oligonucleotide synthesis [(RP)-3aa-3 and (SP)-3aa-3]. The bulky alkylphosphonate backbone was found to remain stable in the presence of nucleophiles such as ammonia, methylamine and tetrabutylammonium fluoride (TBAF) and to be unaffected by the basic pH values imparted to the reaction solution by these amines.
The solid-phase synthesis of oligonucleotides having bulky alkylphosphonate backbones was achieved by employing the widely-used phosphoroamidite method together with an automated oligo-synthesizer. In this process, reactive phosphorus (III) compounds were incorporated into the growing oligonucleotide chain through a cycle of coupling, oxidation, and deprotection of the DMTr group steps. The synthesis cycle was repeated until the desired chain length was reached and the target oligonucleotide was then obtained through cleavage and deprotection. Using the phosphoroamidite 3aa-3 and a 5-Oʹ-DMTr-thymidine (dT)-derived phosphoroamidite bearing a 2-cyanoethoxy group (4a), several oligonucleotides (Oligo-1–6) were synthesized using the automated oligo-synthesizer with a solid support. Analyses by reverse phase liquid chromatography indicated that the retention time for the RP form was longer than that for the SP form, suggesting that the former was more globally hydrophobic. The retention times of Oligo-1 and Oligo-2 were also greater than those for Oligo-7 and 8 and the difference between the retention times of the SP and RP configurations of these compounds was expanded. The effect of chirality on global hydrophilicity/hydrophobicity was therefore increased as the molecules became larger. This effect could also modify the protein binding profiles of these compounds and lead to eutomer/distomer differences in the case that they are utilized as pharmaceuticals. The introduction of the alkyl group at the 3′ end of the oligomer had a greater effect on retention time than introduction at the center (Oligo-1–4).