Artificial nucleic acid backbones and their applications in therapeutics, synthetic biology and biotechnology

The preparation of TherONs, which are typically short (∼20mer) chemically modified ONs [14,15], mainly relies on solid support-based oligonucleotide synthesis [16,17]. Artificial backbones are usually introduced by modified on-resin coupling of a monomer to form the unnatural backbone (monomer approach, Figure 2A), or by coupling of a dinucleoside containing the artificial linkage (dimer approach, Figure 2A). The monomer approach presents significant challenges due to the potential chemical incompatibility with standard ON chemistry and synthesis equipment. Nevertheless, on-resin formation of artificial backbones has been reported for boranophosphates (borano) [18], phosphorothioates (PS) [19], phosphorodithioates (PDS) [20], phosphoramidates (PA) [21], methylphosphonates (MP) [22,23], amides (AM) [24–27] and to a limited extend for triazole linkages (TLs) [28]. Indeed, the dimer approach is preferred for backbones that are harder to form on a solid support, such as ureas [29], squaramides (SQAM) [30], or triazoles [31–36]. The artificial linkage can be pre-formed as part of a dinucleoside that is compatible with standard oligonucleotide synthesis. However, 16 modified dinucleosides are required to cover all sequence possibilities and the construction of consecutive artificial backbones is not possible. In general, the limited availability of optimised and easy-to-use protocols for on-resin formation of artificial backbones, and the demands of the dimer approach remain major bottlenecks in research in the TherON area.

From the plethora of artificial backbones, only a limited number of chemical modifications are found in clinically approved TherONs [37]. Synthetic accessibility, efficient target hybridisation, serum stability and the retention of antisense activity are key requirements for a successful artificial backbone (Figure 2B) [1]. As such, the PS modification is fully compatible with two of the predominant antisense approaches: (i) splice-switching to modulate mRNA maturation, or (ii) activation of RNase H to degrade an mRNA [38–40]. Moreover, advanced synthetic protocols [41] and favourable pharmacokinetics [42] of phosphorothioate oligonucleotides all contribute to its prevalence in clinically approved TherONs. However, PS backbones are linked to toxicity [43,44] and exploration of alternative modifications is urgently needed. A recent study showed that toxicity of PS-TherONs can be significantly reduced by a single MP substitution [45]. Other promising modalities of TherONs include peptide [46] and morpholino [47] nucleic acids which combine backbone and sugar modifications: The former combines amide bonds to connect acyclic subunits and the latter combines a phosphorodiamidate backbone and a morpholine ring as a sugar substitute. However, peptide nucleic acids suffer from poor solubility and inefficient cellular uptake [48,49] while morpholino oligonucleotides are associated with concerns for off-target effects [50,51]. Unfortunately, alternative backbones such as triazoles [32,33,52] or carbamates [53,54] reduce RNA target affinity which must be compensated for by additional sugar or base modifications [35,36,55,56]. Lengthy synthetic procedures limit the combination of artificial backbones with other base or sugar modifications. Nevertheless, these studies showcase the vast potential of alternative backbones in TherONs and emphasise the need for easily accessible backbone modifications for therapeutic research beyond specialised synthetic laboratories.

Different from short TherONs, modified long oligonucleotides for synthetic biology can be several hundreds of bases in length. This exceeds the limits of solid-phase oligonucleotide synthesis and requires different strategies. One approach is to assemble long modified oligomers from shorter, chemically modified ONs via ligation reactions (Figure 2C). Such ligation chemistry must be orthogonal to other functional groups within the oligomers and is often facilitated by a splint/template (orthogonal ligation, Figure 2C). A combinatorial approach for the discovery of splint-templated chemical ligations has been reported to identify DNA-compatible reactions to ligate terminally functionalised ONs [57]. Moreover, the generation of artificial backbone mimics has been shown for bridging 5′-S-phosphorothioester linkages (Ps) [58], PA [59–62], AM [61], urea [63], SQAM [63], TL1 [64] and TL3 [61]. Indeed, copper-catalysed azide-alkyne cycloaddition (CuAAC) to form TL3 was reported for the assembly of whole genes [12,65] and long RNA [9,10] from azide and alkyne modified shorter ONs. This approach enables the precise, site-specific introduction of artificial backbones and other modifications, but is limited by the compatibility of the ligation reaction with terminally modified ONs under aqueous conditions. Another approach is the introduction of artificial backbones through enzymatic synthesis using modified nucleotide triphosphates as substrates [4,13,66–68]. This has been demonstrated by the enzymatic synthesis of a PA-modified gene in the presence of an unnatural cytidine triphosphate analogue [13]. However, the controlled introduction of artificial backbones or other modifications at specific sites is not readily achievable with this method, and engineered polymerases are often required. Whilst engineered polymerases and ligases can accept base- and sugar-modified triphosphates as substrates, such incorporations still form PO bonds [69]. The engineering of enzymes to generate unnatural internucleoside linkages is inherently harder but has been recently demonstrated for uncharged ethylphosphonates [4].

Not all backbone-modified ONs have the desired biocompatibility for applications in synthetic biology. For instance, TL1 was recently described for the preparation of next-generation sequencing libraries but suffers from inefficient and inaccurate replication when used with several polymerases (polymerase read-through, Figure 2D) [34,64]. In contrast, TL3 has good read-through compatibility with DNA and RNA polymerases [70,71] and can be replicated with high fidelity [34] enabling expression of click-assembled genes in bacteria [12,59,70] and mammalian cells [11]. Similarly, phosphoramidate backbones can be read by DNA and RNA polymerases [59,61], and translated by ribosomes [60], and introduction of PA-modified genes can lead to the expression of their associated genetic information in bacteria [13]. Apart from gene synthesis, artificial backbones such as TL2, [31] urea [63] and SQAM [63] were reported as components of modified primers in PCR. In the case of SQAM, in situ template assembly by target-templated SQAM formation was utilised for RNA detection [63]. Other examples include the construction of a functional hammerhead ribozyme [9] or bioactive single guide RNAs (sgRNAs) for gene editing using a split-and-click strategy to form TL3 (CRISPR–Cas9 activation, Figure 2D) [10]. The versatility of artificial backbones in synthetic biology and biotechnology emphasises their vast potential. However, not all artificial backbones perform well, and the molecular requirements for biological integrity remain elusive [34]. Hence, easily accessible and structurally diverse artificial backbones are needed to fully exploit the vast potential of artificial nucleic acids in synthetic biology and biotechnology.

Artificial backbones only account for a fraction of ON modifications but hold great potential. Despite many trailblazing discoveries emphasising the beneficial effects of artificial backbones in therapeutics and synthetic biology, their broad application and an in-depth understanding of their molecular requirements are hampered by limited synthetic accessibility. Thus, new chemical approaches are urgently needed for the synthesis of easy-to-access modified ONs to facilitate research on artificial backbones in a broader spectrum of laboratories.

The authors declare that there are no competing interests associated with the manuscript.

Open access for this article was enabled by the participation of University of Oxford in an all-inclusive Read & Publish pilot with Portland Press and the Biochemical Society under a transformative agreement with JISC.

All authors contributed to the writing and editing of the manuscript, reviewing and analysing the literature.

S.E. is grateful to the EPSRC Centre for Doctoral Training in Synthesis for Biology and Medicine (EP/L015838/1) for a studentship, generously supported by AstraZeneca, Diamond Light Source, Defence Science and Technology Laboratory, Evotec, GlaxoSmithKline, Janssen, Novartis, Pfizer, Syngenta, Takeda, UCB and Vertex. A.H.E.-S. is supported by BBSRC grant BB/R008655/1, in partnership with ATDBio Ltd.

 
• AM

  amides

 
• MP

  methylphosphonates

 
• ONs

  Oligonucleotides

 
• PA

  phosphoramidates

 
• PO

  phosphodiester

 
• PS

  phosphorothioates

 
• SQAM

  squaramides

 
• TherONs

  therapeutic oligonucleotides