Stereospecificity of Oligonucleotide Interactions Revisited: No Evidence for Heterochiral Hybridization and Ribozyme/DNAzyme Activity

A major challenge for the application of RNA- or DNA-oligonucleotides in biotechnology and molecular medicine is their susceptibility to abundant nucleases. One intriguing possibility to tackle this problem is the use of mirror-image (l-)oligonucleotides. For aptamers, this concept has successfully been applied to even develop therapeutic agents, so-called Spiegelmers. However, for technologies depending on RNA/RNA or RNA/DNA hybridization, like antisense or RNA interference, it has not been possible to use mirror-image oligonucleotides because Watson-Crick base pairing of complementary strands is (thought to be) stereospecific. Many scientists consider this a general principle if not a dogma. A recent publication proposing heterochiral Watson-Crick base pairing and sequence-specific hydrolysis of natural RNA by mirror-image ribozymes or DNAzymes (and vice versa) prompted us to systematically revisit the stereospecificity of oligonucleotides hybridization and catalytic activity. Using hyperchromicity measurements we demonstrate that hybridization only occurs among homochiral anti-parallel complementary oligonucleotide strands. As expected, achiral PNA hybridizes to RNA and DNA irrespective of their chirality. In functional assays we could not confirm an alleged heterochiral hydrolytic activity of ribozymes or DNAzymes. Our results confirm a strict stereospecificity of oligonucleotide hybridization and clearly argue against the possibility to use mirror-image oligonucleotides for gene silencing or antisense applications.


Introduction
Stereospecificity of chemical and enzymatic reactions is a basic principle of the (bio)chemistry in our homochiral world [1].The translational apparatus for ribosomal protein synthesis is a stereospecific process that only allows for the incorporation of L-amino acids into proteins.D-amino acids which are found in prokaryotic cell walls and also in peptides from higher organisms exclusively derive from post-translational modifications or ribosome-independent peptide synthesis [2].Likewise, nucleic acids occur only in their D-configuration in nature.
The absence of mirror-image L-nucleic acids (L-RNAs and L-DNAs) from living organisms makes them interesting tools for applications in diagnostics, sensoring or therapeutics since they are resistant against nucleolytic degradation and are immunologically passive [3,4].
The current use of oligonucleotides for therapeutic and other purposes is based on either of two general mechanisms: hybridization between nucleic acids through Watson-Crick base pairing or binding of oligonucleotides to other molecules through their three-dimensional shape.Watson-Crick base pairing is exploited in technologies such as antisense, RNA interference, exon skipping or ribozymes in which an oligonucleotide is hybridizing to a messenger RNA either to arrest its translation, to eliminate or to correct it.The ability of oligonucleotides to adopt complex three-dimensional structures and to bind to a target (similar to monoclonal antibodies) is the basis of aptamer technologies [5,6].The question remains if and how mirror-image oligonucleotides can be introduced into these technologies to improve and extend their reach.
To make use of functional macromolecules of non-natural chirality (D-proteins or L-nucleic acids) requires some careful thoughts and also needs appropriate capabilities in synthetic chemistry since current molecular biology or biotechnology methods cannot be applied.A first breakthrough in this regard was published by Kent and co-workers in 1992 [7].By synthesizing the HIV protease in the natural L-as well as the non-natural D-configuration the authors established what is called "reciprocal chiral substrate specificity": The L-enzyme was able to cleave an L-but not a D-substrate whereas the corresponding D-enzyme cleaved only the D-but not the L-substrate.As expected, an achiral inhibitor blocked both the L-and the D-enzyme.The analyzed interactions, however, were kept in their "own homochiral world", i.e. the L-polypeptide acted on the L-substrate whereas the exact mirror-image D-polypeptide acted on the D-substrate.The next logical question was how to implement new, artificial heterochiral interactions.This was first achieved with L-peptide binders that were selected to bind to a D-peptide target (the mirror-image of a naturally occurring L-peptide) using phage display.When the selected L-peptide binders were converted to their D-configuration, the resulting D-peptide binders recognized the natural L-peptide target [8].The same principle was applied to oligonucleotides in order to generate biostable mirror-image (l-)aptamers, later called Spiegelmers (Spiegelmer is a registered trademark of NOXXON Pharma AG) [3].For this purpose, D-nucleic acid libraries are screened against a non-natural mirror-image target (often a d-polypeptide).When the resulting D-aptamers are converted to l-aptamers, they will recognize the intended natural target (L-polypeptide) [9,10].Recently, even an L-RNA aptamer was identified to bind to a structured D-RNA (TAR RNA) by selecting D-aptamers against the mirror-image l-TAR-RNA [11].Importantly, also when looking at these heterochiral interactions, the reciprocal chiral specificity is maintained and there is no binding of aptamers and Spiegelmers to the enantiomer of their target, i.e. the target in its opposite configuration [8,12].However, in case the selection target is achiral (e.g.theophylline), both the D-and L-configuration of an aptamer will bind with equal affinity [13].
In contrast to aptamers, it has not been possible to apply L-oligonucleotides to technologies that rely on the hybridization of complementary RNA or DNA strands.In the early 1990s there have been reports suggesting that L-oligonucleotides could hybridize to natural D-RNA or D-DNA and may therefore potentially be used for antisense applications.In these studies hybridization of l-homo-oligomers (oligo-l-U or -l-dA) to poly-d-T or poly-d-U was investigated and the formation of triple-stranded complexes was reported [14][15][16][17].However, when Garbesi and co-workers later investigated the hybridization of mixed-sequence L-DNAs containing all four bases with complementary D-DNA or D-RNA molecules they did not find any evidence for heterochiral hybridization [18].The authors concluded that "(. ..)L-DNAs have no future as wide scope antimessenger compounds (. ..)".Thereafter, no further progress was reported on the use of L-oligonucleotides for antisense or related strategies.Two decades later in early 2014, a recent publication now re-challenges the principle of stereospecific RNA and DNA hybridization by claiming that ribozymes and DNAzymes can hydrolyze RNA molecules irrespective of their chirality [19].As the conclusions drawn from this study would have groundbreaking consequences not only for the field of oligonucleotide therapeutics but also for the general mechanisms of stereospecificity we sought to thoroughly reevaluate the stereospecificity of oligonucleotide hybridization and ribozyme/DNAzyme activity systematically in independent experiments.

Oligonulceotide hybridization is strictly stereospecific
According to the current understanding, ribozyme/DNAzyme mediated hydrolysis of an RNA target molecule requires the hybridization of the target sequence with anti-parallel, complementary recognition sites of the ribozyme (Fig. 1A) or DNAzyme (Fig. 1B).To explore whether this premise would still be true for heterochiral RNA/ribozyme or RNA/DNAzyme interactions we used enantiomeric oligonucleotides complementary to the 14 nt RNA target sequence and firstly measured temperature-dependent hyperchromicity.In a homochiral setting, anti-parallel, complementary D-RNA/D-RNA (Fig. 1C) and L-RNA/L-RNA (Fig. 1D) molecules showed comparable melting profiles with T m = 70.9± 0.1°C and T m = 70.8± 0.1°C for (mean ± sem of n = 3), respectively.In contrast, no sharp hyperchromicity transition was observed when heterochiral, anti-parallel, complementary RNA strands were tested (Fig. 1C, 1D).Basically, the same results but with reduced melting temperatures were observed for RNA/DNA and DNA/DNA duplexes.Homochiral, anti-parallel, complementary D-RNA/D-DNA (Fig. 1E) and L-RNA/L-DNA (Fig. 1F) strands displayed melting temperatures of T m = 57.8± 0.1°C and T m = 56.9± 0.1°C (mean ± sem of n = 3), respectively, and D-DNA/D-DNA (Fig. 1G) and L-DNA/L-DNA (Fig. 1H) duplexes showed melting temperatures of T m = 60.4 ± 0.1°C and T m = 60.0 ± 0.2°C (mean ± sem of n = 3), respectively.Significant hyperchromicity indicating hybridization of heterochiral strands could not be detected for any of the tested examples (Fig. 1C-1H).Furthermore, we investigated if heterochiral oligonucleotides would interact in parallel, complementary orientation as previously suggested [23].Hyperchromicity transition was neither observed for homochiral nor heterochiral, parallel complementary RNA/RNA (S1A, S1B To finally complete the hybridization experiments, we analyzed the interaction of achiral peptide nucleic acid (PNA) with RNA and DNA oligonucleotides of either chirality.As expected, PNA hybridization was independent of the chirality: the hybridization of D-RNA (Fig. 2A) and L-RNA (Fig. 2B) to anti-parallel complementary PNA yielded almost identical melting temperatures of T m = 84.4± 0.1°C and T m = 84.3± 0.2°C (mean ± sem of n = 3), respectively.A similar picture was observed for D-DNA (Fig. 2C) and L-DNA (Fig. 2D) interacting with anti-parallel complementary PNA.Here the melting temperatures were slightly lower: T m = 80.1 ± 0.3°C and T m = 79.6 ± 0.1°C (mean ± sem of n = 3), respectively.

Ribozyme and DNAzyme activity is strictly stereospecific
For a functional analysis we tested the stereospecificity of RNA hydrolysis using a hammerhead ribozyme (Fig. 1A and Fig. 3A) and a DNAzyme (Fig. 1B and Fig. 3B) following the protocol published by Wyszko et al. [19].Homochirality of RNA substrate and enzymatically active nucleic acid (ribozyme or DNAzyme) resulted in complete cleavage (Fig. 3A and 3B, lanes 2 and 6).In all cases, the homochiral cleavage is MgCl 2 -dependent (Fig. 3A and 3B, lanes 3 and 7).In contrast, no cleavage was observed when RNA substrates were incubated with heterochiral ribozyme or DNAzyme (Fig. 3A and 3B, lanes 4 and 8).The identity of the cleavage products was confirmed by mass spectrometry (S2 Fig. ).Thus, in agreement with the hybridization data, we can show that the activity of ribozyme and DNAzyme is strictly stereospecific.

Experiments focusing on stereospecificity need to be tightly controlled
Enzymatic RNA hydrolysis by ribozymes is a highly efficient process which requires the need for careful planning and tight experimental control, especially in stereospecificity studies.For example, 0.01% of enantiomeric D-ribozyme (200 pM) contaminating an L-ribozyme solution (2 μM) creates a false positive "heterochiral" cleavage product that is indistinguishable from that generated by real homochiral RNA hydrolysis (Fig. 4, lanes 6 and 3, respectively).A first measure to reduce the risk of generating artefacts can be a reduction of the MgCl 2 concentration from 10 mM to a physiological level of 1 mM.Under these conditions, a 0.1% contamination with D-ribozyme (2 nM) is largely tolerated (Fig. 4B, lane 5).
Since the differentiation between L-and D-configured oligonucleotides with standard molecular biology methods and equipment is challenging, careful control strategies have to be established to uncover false-positive results.A rather simple way to control this type of assays is the use of enantiomeric antisense probes.As discussed above, contamination of an L-ribozyme with as little as 0.01% D-ribozyme creates a pretended "heterochiral" hydrolysis product (Fig. 5, lane 5) which is indistinguishable from the D-homochiral product (Fig. 5, lane 8).Addition of 0.1% D-antisense probe completely blocks the alleged "heterochiral" cleavage reaction (Fig. 5, lane 6) while leaving the D-homochiral unaffected (Fig. 5, lane 9).Importantly, this blockade is again stereospecific as addition of 0.1% L-antisense has no impact (Fig. 5, lane 7).

Discussion
We investigated the stereospecificity of oligonucleotide hybridization and RNA hydrolysis by enzymatically active nucleic acids and show that both are strictly limited to homochiral oligonucleotides.In a first paper in 2013, Wyszko et al. described that ribozymes and DNAzymes composed of L-nucleic acids are able to cleave L-RNA substrates [24].This observation is in agreement with the basic concept of chirality that enantiomeric molecules must have identical chemical properties.Furthermore, the outcome could be deduced from previous data generated with either D-proteins or L-nucleic acids [3,7,8,25].As expected, we were able to confirm that homochiral hybridization of anti-parallel complementary strands as well as homochiral RNA cleavage by either a ribozyme or a DNAzyme occurs with the same thermodynamics and efficiency with either L-or D-configured oligonucleotides.
In a second paper published early 2014, Wyszko et al. went one step further claiming that ribozymes and DNAzymes are also able to sequence-specifically hydrolyze RNA in a heterochiral manner [19].In a structural model the authors propose that this phenomenon is based on helices formed by Watson-Crick base pairing between heterochiral strands.Since these results contradict the current understanding, we thoroughly investigated the hybridization of heterochiral oligonucleotides by hyperchromicity measurements.We were not able to detect  any specific hybridization between oligonucleotides of different configuration (Fig. 1).Also in line with the basic principles of stereospecific recognition, PNA oligomers as achiral substances sequence-specifically hybridize to RNA and DNA irrespective of their chirality (Fig. 2) [26].
Consequently, in tightly controlled experiments we were also not able to reproduce any hydrolytic activity of ribozymes or DNAzyms towards RNA substrates of different chirality (Fig. 3).Since ribozyme/DNAzyme-mediated RNA hydrolysis is a very effective catalytic process we experienced in our own hands that (under the given experimental conditions) minimal contaminations of the 'unwanted' ribozyme/DNAzyme may easily lead to false positive results.In this context it needs to be pointed out that the data presented by Wyszko et al. are also not consistent with the principle of identical chemical properties of enantiomers.Cleavage of D-RNA by the L-riboyzme is much less efficient than L-RNA cleavage by the D-riboyzme and furthermore only the L-but not D-DNAzyme is reported to be capable of heterochiral activity [19].If sequence-specific heterochiral cleavage would indeed take place hydrolysis would need to occur at the same efficiency for the D-/L-and L-/D-pairs of target and ribozyme/DNAzyme.Thus, we strongly recommend when investigating the stereospecificity of enzymatic processes to perform this kind of experiments at more physiological conditions (reduced MgCl 2 , no ribozyme/DNAzyme excess) and, what is most important, to include appropriate controls.An example for such a control experiment is the use of antisense probes that may help to discriminate between a de facto cleavage and an artificial cleavage product derived from contaminations (Fig. 5).
We have systematically investigated the hybridization behavior of oligonucleotides displaying the D-and L-configuration.The results confirm that sequence-specific (Watson-Crickbased) hybridization does not occur between oligonucleotides of different chirality.Consequently, heterochiral ribozyme/DNAzyme cleavage of RNA molecules is not possible as long as the mechanisms of the catalytically active nucleic acids rely on hybridization to their respective substrates.However, the powerful techniques of in vitro selection and in vitro evolution are able to deliver mirror-image oligonucleotide candidates [27] that cannot only recognize an RNA substrate by Watson-Crick-independent tertiary interactions [11] but may eventually even be able to catalytically modulate a substrate RNA.In this regard, a cross-chiral RNA polymerase ribozyme has recently been reported [28].

Fig 1 .
Fig 1. Hybridization of anti-parallel complementary oligonucleotides is stereospecific.Schematic representation of (A) hammerhead ribozyme and (B) DNAzyme in complex with the target RNA sequence (green).Hydrolysis sites are indicated by arrows.To analyze (potential) hybridization of enantiomeric, anti-parallel complementary oligonucleotides (3 μM each in 10 mM phosphate buffer, pH 7.4, 100 mM NaCl) temperature-dependent hyperchromicity at 260 nm was measured.(C) D-RNA and (D) L-RNA with anti-parallel complementary D-RNA (red) and L-RNA (blue).(E) D-RNA and (F) L-RNA with anti-parallel complementary D-DNA (red) and L-DNA (blue).(G) D-DNA and (H) L-DNA with anti-parallel complementary D-DNA (red) and L-DNA (blue).Mean of three melting ramps (25°C to 95°C) is given as normalized absorption A / Amax at 260 nm.First derivative is shown as dotted line.Data are representative of three independent experiments.doi:10.1371/journal.pone.0115328.g001

Fig 2 .
Fig 2. Achiral PNA hybridizes to both D-and L-RNA/DNA.Hybridization of achiral PNA and anti-parallel complementary (A) D-RNA, (B) L-RNA, (C) D-DNA, and (D) L-DNA (3 μM each in 10 mM phosphate buffer, pH 7.4, 100 mM NaCl) was analyzed by measuring temperature-dependent hyperchromicity at 260 nm.Mean of three melting ramps (25°C to 95°C) is given as normalized absorption A / A max at 260 nm.First derivative is shown as dotted line.Data are representative of three independent experiments.doi:10.1371/journal.pone.0115328.g002