Phosphorothioates are probably the most popular modifications used in oligonucleotide therapeutics. It involves the substitution of a sulfur atom for a non-bridged oxygen at a phosphate. This modification does not significantly hinder duplex formation, retains the ability of the oligonucleotide duplexes to promote RNase H activity and enhances general nuclease stability. This last property has played key roles in modulating the in vivo half-lives of several approved oligonucleotide therapeutics and many
more candidates in earlier clinical stages.
This modification also has disadvantages, and the most notable of these is that phosphorothioates, introduced via the standard phosphoramidite method, give additional stereocenters. Each phosphorothioate substitution can be either an “Sp” or “Rp” conformation (Figure 1A). As such, a fully phosphorothioate 20 nt antisense oligonucleotide will be a mixture of more than half a million (219) different molecules. To avoid and/or reduce this type of complexity, one workaround is to use achiral phosphorodithioates instead, where both non-bridged oxygen atoms are replaced with sulfurs. Phosphorodithioates share many of the desirable properties of phosphorothioates without the additional undesired stereocenters.1, 2 In addition, phosphorodithiotes are even more stable to nucleases than their phosphorothioate counterparts.
Phosphorodithioates were first developed over thirty years ago. The first oligonucleotide syntheses incorporated dinucleotide phosphoramidites with the phosphorodithioate already in place, and later investigations developed thiophosphoramidites that could be used in place of standard phosphoramidites. One sulfur would come from the thiophosphoramidite while the other would be obtained via the sulfurization reagent (Figure 1A). Several versions of these thiophosphoramidites have been described, and the most popular of these employs pyrrolidinyl thiophosphoramidites where the thiol is protected with a beta-(benzoylmercapto)ethyl group (Figure 1).3 After synthesis is complete, basic deprotection conditions will first remove the benzoyl group and then subsequently eliminate ethylene sulfide to give the desired product (Figure 1B).
Figure 1. Phosphorothioate and phosphorodithioate chemistry.
A: Top, phosphorothioate synthesis; bottom, phosphorodithioate synthesis.
B: Deprotection of phosphorodithioate linkage.
Glen Research introduced DNA thiophosphoramidites in 20084 and subsequently 2’-OMe thiophosphoramidites in 20155 (Figure 2). Since this latter introduction, there have been several studies published on the use of these and other closely related building blocks. Recently, Abeydeera et al. highlighted the utility of phosphorodithioates for optimizing aptamers post-SELEX.6 The researchers synthesized a complete series of singly phosphorodithioate-modified sequences of a VEGF165 RNA aptamer and evaluated them for binding to VEGF165. Although most of these sequences behaved similarly to the native unmodified aptamer, there were two particular sequences that were very different, each with a single G phosphorodithioate substitution. These sequences had binding constants that were approximately one-thousand-fold enhanced relative to that of the control aptamer, a very impressive improvement in performance. This type of investigation was also applied to an α-thrombin RNA aptamer, and similar results were observed. In this case, one U phosphorodithioate substitution also enhanced binding affinity by approximately one-thousand-fold relative to that of the unmodified α-thrombin aptamer control. In addition to elevated binding constants, these sequences exhibited relatively unchanged secondary structure, maintained binding specificity and as expected, showed enhanced serum stability.
Figure 2. DNA and 2’-OMe Thiophosphoramidites
In another aptamer study, Amero et al. employed an approach similar to that of what was used for the VEGF165 and α-thrombin to further improve a previously characterized and partially optimized aptamer.7 In this case, the target was the receptor tyrosine kinase, AXL, which is overexpressed in ovarian cancer. By binding to this receptor, the aptamer would silence the activity of AXL and lead to decreased ovarian cancer tumor growth. In vitro studies showed that two particular phosphorodithioate aptamers, one with a single dA phosphorodithioate and the other with a single dG phosphorodithioate, were quite effective relative to controls. Subsequently, PEG-conjugated versions of these two aptamers were found to significantly inhibit ovarian cancer tumor weight in mice.
In a third publication, the use of phosphorodithioates in LNA/DNA antisense oligonucleotides was evaluated.8 Bleicher et al. synthesized LNA thiophosphoramidites and then used these, along with the commercially available DNA versions, to synthesize several series of RNase H activating gapmers (LNA-DNA-LNA) and steric blocking mixmers (alternating LNA/DNA). These oligonucleotides were fully phosphorothioated with up to seven phosphorodithioate substitutions. In general, phosphorodithioates were well tolerated in the gapmers, particularly in the LNA flanking regions. The phosphorodithioate substitutions did not significantly alter melting temperatures, enhanced serum stability, improved cellular uptake and most importantly, enhanced target reduction in both primary rat hepatocytes as well as mice. For the mixmers, phosphorodithioate substitutions were compatible but sequence dependent. In a set of exon-skipping experiments, DNA phosphorodithioate linkages were generally more beneficial than LNA phosphorodithioate linkages.
As demonstrated in these recent investigations, phosphorodithioates continue to be a compelling alternative or complement to phosphorothioates for therapeutic applications. For those who would like to explore DNA and 2’-OMe phosphorodithioate-containing backbones, the required phosphoramidites and the methods for their use are readily available from Glen Research.
We would like to thank Xianbin Yang for reviewing this document and for his many helpful suggestions.