Introduction

The development of artificial nucleic acids, or xeno nucleic acids (XNAs), that form stable duplexes with natural nucleic acids has been challenging. A vast number of XNAs have been reported, and some have found utility in biotechnology, medicine and nanotechnology.1, 2, 3 Among these, a major strategy for the synthesis of oligomers that form stable duplexes with natural nucleic acids has been to use neutral backbones. Peptide nucleic acids (PNAs) are one of the most widely used XNAs in the first category (Figure 1).4 PNAs have a neutral peptide backbone instead of a phosphodiester backbone and thus have no charge. PNAs can form stable duplexes with complementary DNA and RNA without requiring cation condensation. Chiral PNAs have been reported that differentially recognize natural nucleic acids depending on their chirality.5, 6, 7 Morpholino nucleic acids, which also have no charge, are also representative of the class of nucleic acids with neutral backbones (Figure 1).8

Figure 1
figure 1

Chemical structures of representative XNAs.

XNAs of another type have a constrained backbone. Locked nucleic acids (LNAs) belong to this category (Figure 1).9, 10 LNAs contain a methylene linkage between the 2′-oxygen and the 4′-carbon of D-ribose, locking the sugar puckers into the 3′-endo conformation adopted by RNA. Consequently, LNA has high affinity for RNA due to a lower decrease in entropy upon duplex formation. Other XNAs with bicyclic or tricyclic backbones, such as tricyclo-DNA, have also been reported (Figure 1).11, 12

Another strategy involves incorporation of acyclic base surrogates onto a DNA backbone.13, 14, 15, 16, 17 Benner’s group first reported flexible nucleic acids, which have nucleobases on flexible propanediol linkers.18, 19, 20 Unlocked nucleic acids, in which there is no bond between 2′ and 3′ carbons of the ribose sugar, were reported by Wengel’s group.21, 22 However, incorporation of these modified nucleotides into DNA or RNA often severely lowered the stability of the duplexes. Meggers et al. reported glycerol nucleic acids (GNAs), which tether nucleobases through an acyclic C2 linker (Figure 1).23, 24, 25, 26 Quite surprisingly, a GNA oligomer, which is composed of only GNA monomers, formed a highly stable duplex with a complementary GNA oligomer even though GNA has a relatively flexible structure and phosphodiester bonds. The melting temperature of the GNA homo-duplex was much higher than those of DNA and RNA homo-duplexes of the same sequence. This high stability of the GNA homo-duplex showed that a cyclic sugar is not necessary for stable duplex formation. GNAs showed strong sequence dependence in the context of hetero-duplexes; they do not form stable duplexes with either DNA or RNA when the GC content is high.26

Inspired by these pioneering studies, several groups, including ours, have synthesized and characterized fully modified acyclic XNAs with phosphodiester linkages.27, 28, 29 There are several advantages of acyclic scaffolds over other XNAs. First, the synthetic costs of acyclic XNA are often low due to their simple structures. Second, chemical modification is usually facile, allowing XNAs with novel structures to be easily prepared. Third, because their chemical structures are very different from natural nucleic acids, they are highly resistant to nucleases. In addition, acyclic XNAs with phosphodiester bonds are usually highly water-soluble, in contrast to neutral XNAs. Furthermore, acyclic XNAs are candidates for genetic material in the ‘pre-RNA world’ because of their simple structures.30, 31 Here we review our recent work on acyclic XNAs. Our group has developed three acyclic XNAs: acyclic D-threoninol nucleic acid (D-aTNA), serinol nucleic acid (SNA) and acyclic L-threoninol nucleic acid (L-aTNA). Though structural differences are marginal, each of these XNAs has unique properties.

D-aTNA is highly orthogonal to natural nucleic acids

In our previous studies, various types of molecules, including fluorophores and photoresponsive moieties, were incorporated into DNA via D-threoninol.32, 33 Although the chemical structures of these base surrogates are very different from those of natural nucleobases, duplexes between natural nucleic acids and D-threoninol are stabilized through stacking interactions. Therefore, we selected D-threoninol as a linker for tethering natural nucleobases and developed D-aTNA.34 We synthesized D-aTNA phosphoramidite monomers, and D-aTNA oligomers were synthesized using a DNA/RNA synthesizer (Figure 2). D-aTNA formed extremely stable duplexes with complementary D-aTNA; the melting temperature (Tm) of homo-duplexes of an 8-mer of D-aTNA was as high as 62.7 °C, much higher than the Tm of either DNA or RNA homo-duplexes of the same sequence (29.0 °C or 38.9 °C, respectively). The Tm of homo-duplexes of other acyclic XNAs, including PNA, GNA and SNA, is also lower.35 Thermodynamic analyses indicated that the D-aTNA homo-duplexes are more stable than those of natural nucleic acids due to enthalpic contributions. The molecular mechanism may involve strong stacking interactions between bases, allowed by the flexible backbone and/or the hydrogen bonds between the amide group and the phosphate backbone.

Figure 2
figure 2

Chemical structure of D-aTNA and melting curves of homo-duplexes of D-aTNA, DNA and RNA of the same sequence: 5′-GCATCAGT-3′/3′-CGTAGTCA-5′. Note that D-aTNA has a 1′ terminus rather than a 5′ terminus. In the RNA sequence, T was replaced with U. Bases of phosphoramidite monomers are protected by acyl groups.

In spite of the high stability of D-aTNA homo-duplexes, D-aTNA did not form stable duplexes with natural DNA or RNA. The Tms of D-aTNA/DNA and D-aTNA/RNA 8-mer duplexes were too low to measure (Table 1). D-aTNA exhibits high orthogonality to natural nucleic acids. There may be applications in which this is an asset: for example, D-aTNA could hybridize with complementary D-aTNA without interference from natural nucleic acids in cells.

Table 1 Melting temperatures of XNAs and natural nucleic acids

Development of SNA with unique chiroptical properties

D-aTNA cannot hybridize with DNA or RNA in spite of the extremely high stability of its homo-duplex. We hypothesized that D-aTNA is not flexible enough to conform to the more rigid DNA or RNA. We then synthesized a more flexible XNA, SNA, which has natural bases linked through symmetric serinols (Figure 1).36 The difference between the D-aTNA and SNA monomer is a single methyl group; however, SNA exhibited several unique properties. Though serinol itself is achiral, we synthesized enantiopure oligomers from chiral phosphoramidite monomers. The enantiopure SNA oligomer exhibits interesting chiroptical properties, as shown in Figure 3a. The SNA dimer T→A is the enantiomer of the A→T sequence. Thus, the chirality of the SNA oligomer can be inverted by reversing the sequence. In contrast, T→T is the enantiomer of itself. Therefore, an SNA oligomer with a symmetric sequence is achiral (the meso form).

Figure 3
figure 3

(a) The chemical structure of the SNA phosphoramidite monomer synthesized from L-serine and schematic illustration of chirality inversion of SNA oligomers. (b) Helicities of SNA homo-duplexes can be controlled by sequence design. (S) and (R) termini are named based on the chirality of terminal residues. Bases of phosphoramidite monomers are protected by acyl groups.

The CD spectrum of an 8-mer SNA homo-duplex Sa/Sb is shown in Figure 3b. Positive and negative cotton effects are observed. In contrast, for the duplex of the reversed sequences of Sa and Sb (Sd/Sc), the complete inverse of the CD signals are observed. Thus, the helicity of SNA homo-duplexes can be inverted by reversing their sequences. On the other hand, the Se/Sf duplex, which has a symmetrical sequence, gives no CD signal, revealing that this duplex is achiral. This unique property is due to the symmetric structure of SNA and cannot be seen in other asymmetric XNAs.

The Tm of an 8-mer SNA homo-duplex is 51.1 °C (Table 1), much higher than that of natural nucleic acid duplexes of the same sequence. However, the SNA homo-duplex is less stable than the D-aTNA duplex because of the larger entropic loss upon hybridization that is caused by SNA’s flexibility. Interestingly, although SNA has a structure similar to that of D-aTNA, SNA forms stable hybrid duplexes with both DNA and RNA. The Tms of 8-mer SNA/DNA and SNA/RNA hetero-duplexes are 23.5 °C and 35.0 °C, respectively. The Tm of a SNA/RNA duplex is even higher than that of a DNA/RNA duplex, clearly demonstrating that SNAs have the potential to be more efficient than DNAs as probes and drugs that target RNA. To the best of our knowledge, SNA is the first acyclic XNA with phosphodiester bonds that can form stable hetero-duplexes with both DNA and RNA. Our analyses of SNA indicate that a rigid structure or a neutral charge is not required for hetero-duplex formation with natural nucleic acids.

L-aTNA has high affinity for natural nucleic acids

The striking differences between D-aTNA and SNA prompted us to synthesize another acyclic XNA, L-aTNA (Figure 1), in which nucleobases are tethered via L-threoninol.37 As expected, because L-aTNA is an enantiomer of D-aTNA, the Tm of the 8-mer L-aTNA homo-duplex is almost the same as that of the D-aTNA homo-duplex (Table 1). In sharp contrast, significant differences are observed in the stabilities of their hetero-duplexes with DNA and RNA. Unlike D-aTNA oligomers, L-aTNA oligomers can recognize complementary DNA and RNA. Interestingly, L-aTNA prefers to form hetero-duplexes not in an antiparallel but in a parallel manner. The Tm of the 8-mer L-aTNA/RNA duplex with a sequence designed to enforce hybridization in the parallel direction is 41.0 °C, whereas the melting profile of the L-aTNA/RNA duplex designed to hybridize in the antiparallel direction did not show a sigmoidal curve (Figure 4 and Table 1). The Tm of the parallel hetero-duplex is even higher than those of the corresponding SNA/RNA and RNA/RNA duplexes. Thus, L-aTNA forms a more stable complex with complementary RNA than SNA does. Similarly, the stability of the L-aTNA/DNA duplex is higher than that of the SNA/DNA duplex.

Figure 4
figure 4

Melting curves of L-aTNA/RNA and L-aTNA/DNA duplexes. The sequences are as follows: 3′-GCATCAGT-1′/3′-CGTAGTCA-5′ (parallel direction) and 1′-GCATCAGT-3′/3′-CGTAGTCA-5′ (antiparallel direction). In RNA, U was substituted for T.

L-aTNA hybridizes with DNA and RNA only in the parallel direction. This is the complete opposite of the preference of SNA, which forms hetero-duplexes only in the antiparallel direction. We attribute this difference to the configuration of the 2′ carbon of these XNAs (Figure 5). As mentioned above, the chirality of the SNA oligomer can be inverted by reversing its sequence. This means that the configuration at the central carbon of SNA can be inverted by reversing its sequence. The strong preference of SNA for the antiparallel direction indicates that the configuration of the 2′ carbon is essential for DNA/RNA recognition. In contrast, because 2′ carbon of the L-aTNA phosphoramidite monomer has the opposite configuration of that of the SNA monomer, the direction of the amide group in the parallel L-aTNA/RNA duplex corresponds to that in the antiparallel SNA/RNA duplex. Accordingly, L-aTNA can form hetero-duplexes only in the parallel orientation. On the other hand, although the antiparallel D-aTNA/RNA duplex has the same configuration, D-aTNA does not form stable duplexes with DNA or RNA. This result clearly demonstrates that the hybridization properties of these XNAs are highly dependent on the position of the methyl group and on the configuration of 2′ carbon.

Figure 5
figure 5

Relationship between sequence direction and configuration of the central carbon of (a) SNA and (b) L-aTNA. SNA forms a duplex with complementary RNA in an antiparallel manner, whereas L-aTNA hybridizes with RNA only in the parallel direction. Directions of amide groups are emphasized with circles. Bases of phosphoramidite monomers are protected by acyl groups.

The drastic difference between D-aTNA and L-aTNA is also likely due to their helical preferences. The CD spectrum of the D-aTNA homo-duplex is shown in Figure 6. This CD signal resembles that of chiral PNA forming a left-handed helix,6, 38, 39 indicating that D-aTNA prefers to adopt a left-handed helical structure. In contrast, L-aTNA shows the opposite CD signal, which is similar to that of chiral PNA forming a right-handed helix. These results strongly suggest that L-aTNA forms a stable duplex with DNA and RNA due to its preference for a right-handed helical structure formation. The helical preference of L-aTNA underlies the higher affinity of L-aTNA for RNA compared with SNA.

Figure 6
figure 6

CD spectra of 8-mer L-aTNA and D-aTNA duplexes. Sequences are as follows; 1′-TGACTACG-3′/3′-ACTGATGC-1′.

Summary and outlook

In summary, we have developed three acyclic XNAs with unique properties (Figure 7). These XNAs have simple structures, making their synthesis straightforward and each can form an extremely stable homo-duplex. Their abilities to recognize natural nucleic acids vary and are highly dependent on chemical structure. D-aTNA and L-aTNA form the most stable homo-duplexes among these acyclic XNAs. In an example of an important contrast, D-aTNA does not form a stable hetero-duplex with DNA or RNA, whereas L-aTNA does. This high orthogonality to natural nucleic acids could make it possible to use D-aTNA as a nanomaterial that functions inside a cell without interfering with cellular nucleic acids. Our group has begun to investigate the operation of nanomachines composed of D-aTNA.

Figure 7
figure 7

Characteristics and possible applications of acyclic XNAs described in this paper.

SNA, which lacks a methyl group that is present in D-aTNA, has a unique chiroptical property: the chirality and helicity of SNA oligomers can be controlled in the design of the sequence. This property stems from the symmetric structure of SNA and makes it unique because most XNAs are asymmetric. In addition, in contrast to D-aTNA, SNA forms stable hybrid duplexes with both DNA and RNA. The SNA/RNA duplex was more stable than the DNA/RNA duplex, and SNA is resistant to nuclease degradation, clearly demonstrating the potential of SNA as a diagnostic probe and therapeutic. We have developed molecular beacons composed of SNA and demonstrated that these probes can be used to visualize mRNA in cells.40 We have also incorporated SNA monomers at the termini of siRNAs; these conjugates activate RNAi-mediated gene silencing and have excellent enzymatic durability.41

As we reported recently, L-aTNA has the highest affinity to natural nucleic acids among the three acyclic XNAs. The L-aTNA/RNA duplex is even more stable than the SNA/RNA duplex. Thus, L-aTNAs have potential for use as probes and drugs that target natural nucleic acids. Because the three acyclic XNAs presented here have unique properties, each should have applications as tools in biology, biotechnology and nanotechnology. The large difference among the properties of these XNAs was a surprise because their structural differences are small. This suggests that additional acyclic XNAs with novel functions can be developed by tuning the chemical structure of the monomer.