RNA-Templated Concatenation of Triplet Nucleic Acid Probe

Abstract

Template-directed synthesis offers several distinct benefits over conventional laboratory creation, including unsurpassed reaction rate and selectivity. While it is central to many biological processes, such an approach has rarely been applied to the in-situ synthesis and recognition of biomedically relevant target. Towards this goal, we report the development of a three-codon nucleic acid probe containing C-terminal thioester and N-terminal cysteine that is capable of undergoing template-directed oligomerization in the presence of RNA target, and self-deactivation in its absence. The work has implications for the development of millamolecular nucleic acid probes for targeting RNA-repeated expansions associated with myotonic dystrophy type 1 and other related neuromuscular and neurodegenerative disorders.

Graphical Abstract

A trinucleotide gamma peptide nucleic acid containing C-terminal thioester and N-terminal cystine undergoes RNA-templated concatenation upon the reduction of disulfide bond, and self-deactivation (intramolecular cyclization) in the absence of RNA target

The design of molecules for tight and selective binding of biological targets, such as proteins and nucleic acids, is the cornerstone of bioorganic and medicinal chemistries. Despite the heroic efforts made by synthetic chemists, achieving both of these properties at a high degree still remains a challenge due to their discordance. However, nature has devised a creative solution to this dichotomy. Instead of partitioning synthesis from recognition, an approach commonly taken by practitioners in the fields, the two steps are tightly coupled in many biological processes through macromolecule-templated synthesis.^[1,2] Template-directed synthesis, in general, modulates the effective molarity of reactants, enabling highly diluted components to react with astonishing efficiency and selectivity. In the case of polymerization process, involving the conjugation of consecutively-assembled chemical building blocks into larger oligomeric systems, proximity-based reaction modifies the avidity of reactants, allowing the low affinity but highly selective components to coalesce and bind tightly and selectively to the templated target. Such an example is DNA replication, a template-directed polymerization reaction occurring in every living organism on Earth with unparalleled rate and fidelity, and exceptional affinity and selectivity of the copied (daughter) strand for the parent template.

The concept of template-directed synthesis has long been recognized by chemists^[3–5] and has been effectively exploited in the synthesis of organic molecules,^[6] and in the organization and assembly of a variety of macro- and supramolecular systems.^[7,8] Despite the appeal, such an approach has rarely been applied to the in-situ assembly and synthesis of nucleic acid building blocks for recognition of genetic materials in live cells and intact organisms; although, it has been explored in the context of prebiotic chemical synthesis.^[9–11] This is largely due to the fact that natural nucleic acid biopolymers, i.e. DNA and RNA, as well as a majority of the synthetic analogues that have been developed to date, are highly flexible in conformation. For those few rare exceptions that are conformationally more rigid, such as locked nucleic acid (LNA),^[12] they are generally difficult to chemically modify. As with the former, it would be quite challenging to install two highly reactive functional groups on an oligonucleotide molecule, one on each end, and to be able to deter them from undergoing rapid intramolecular reaction. The ability to control the rates of inter- and intra-molecular reactions is essential to a successful implementation of template-directed synthesis in biology and medicine. To enable this process, the mutually reactive nucleic acid components would need to be chemically stable for a defined period, such that in the presence of RNA target they would undergo template-directed concatenation; whereas in its absence they would self-deactivate by undergoing intramolecular (cyclization) reaction. Scheme 1 shows the reaction pathways of the triplet nucleic acid probe reported in this work.

Scheme 1.

(A) Nucleic acid probe (P2) containing C-terminal thioester and N-terminal cystine in the extracellular environments. (B) Probe in the intracellular environments (reduced state) in the presence of RNA target. (C) Probe in the reduced state in the absence of RNA target.

To assess the feasibility of this design concept, we synthesized a series of triplet miniPEG-gamma peptide nucleic acid (MPγPNA)^[13] probes containing C-terminal thioester and N-terminal cystine, along with the various controls (Scheme 2A). The choice of MPγPNA was based on previous studies in which we^[14] and Corradini and coworkers^[15] showed that this particular chiral class of PNA adopts a right-handed helical motif. Such an extended helical conformation, in principle, could impede the mobility of terminal reactive groups and thereby reduce the rate of intramolecular reaction. The N-terminal cystine was chosen to provide a greater control of probe handling so that they remain chemically stable until the disulfide bond is reduced—a chemical state mimicking that in the intracellular environments.^[16] Native chemical ligation (NCL) was selected because of its appealing native backbone product formation (Figure S1, Supplemental Information), bio-orthogonality, fast reaction kinetics, and the fact that such a reaction has been successfully demonstrated both in vitro and in vivo.^[17–19]

Scheme 2.

(A) Chemical compositions of probes. (B) Model RNA targets.

The GAC sequence was chosen so that it is complementary to the CUG-RNA repeats, the expansion of which is the genetic basis of myotonic dystrophy type 1 (DM1).^[20] P1 was included as a negative control because it contained flexible achiral PNA units (G and A), which was expected to undergo rapid intramolecular reaction upon the reduction of disulfide bond. P3 contained the same chemical composition as P2 except that C-nucleobase was replaced by G-clamp (X)—a tricyclic aromatic system capable of forming five H-bonds with G.^[21] This substitution was designed to test the effect of base-stacking on the conformational flexibility of probes, as measured by the half-lives (t_1/2) of the acyclic sulhydryl intermediates. P4 and P5 were the other negative controls: P4, replacement of the C-terminal thioester with amide, and P5, A-to-T single-base mismatch. The sulhydryl-protected C and X monomers were prepared according to the published procedures,^[13] starting with Boc-Cys(Trt)-OH (Figure S2). Probes were made on MBHA-resin using Boc-chemistry, purified by RP-HPLC, and confirmed by MALDI-TOF.

In the synthesis of probes containing C-terminal thioester via conventional SPPS-Boc-chemistry, there was a concern for inverted deletions following the removal of Boc-protecting group and the neutralization of amines (Figure S3). To mitigate this possibility, we performed the neutralization and coupling steps in-situ (Figure S4). Such a condition provided reasonable chemical yields (~ 76%) and purities of P2 through 5 (Figure S5), with only minor deletion products. Figure 1 shows a representative HPLC trace of crude P2. However, under identical conditions, we were not able to prepare P1. No precipitation occurred following the addition of diethyl ether to the cleavage mixture. Injection of the crude sample into HPLC yielded several indiscernible baseline peaks (data not shown). We attribute the failure in the synthesis of P1 to the high conformational flexibility of achiral PNA, which presumably underwent rapid inverted deletions following the deprotection and neutralization steps (Figure S3).

Figure 1.

HPLC trace of crude P2. Inset: reinjection of purified P2.

We next determined the relative conformational flexibility of probes by measuring the half-lives of the reactive intermediates following the reduction of disulfide bond. MALDI-TOF was chosen over HPLC and other analytical techniques in the quantification of reaction products because of the relatively short time-scale of intramolecular NCL, which can be carried out in less than 5 min with the former. DTT (dithiothreitol) was employed as a reducing agent. To assimilate the physiological conditions, all the experiments were performed at a physiologically relevant ionic strength (10 mM NaPi, 150 mM KCl, 2 mM MgCl₂; pH 7.4).^[22] Initial experiments were carried out at an ambient temperature, in which P2 (1 μM) was incubated with DTT (100 μM, within the concentration range of glutathione in cells)¹⁶ for 1 hr. Under this condition, we found that the disulfide bond was completely reduced, as evident by the loss of 120-Dalton mass (Figure 2 and Figure S6). To our surprise, we noticed that the reduced acyclic form of P2 (P2*), with m/z of M-120 Daltons, remained intact without any trace of the cyclic P2 product (cP2) being formed. This is apparent from the absence of a chemical species with 1128-Dalton mass. This result is in line with the earlier finding that MPγPNA adopts an extended helical conformation,^[14] which significantly reduces the rate of intramolecular reaction in comparison to that of a more flexible achiral PNA.

Figure 2.

MALDI-TOF spectrum of P2 at 1 μM concentration after incubation with DTT (100 μM) at an ambient temperature for 1 hr. Inset: MS spectrum of P2 (M = 1466 Daltons). The M-120 peak corresponds to the reduced acyclic P2*, and the 1128-Dalton peak to the cyclic cP2.

To assess the biological relevance of this reaction, we monitored the reaction progress of P2 at a physiological temperature (37 °C) as a function of time, and compared to that of P3. The difference between the two probes is the degree of base-stacking interaction, which is more pronounced for P3 due to the expanded tricyclic-ring of G-clamp. Inspection of Figure 3A reveals that the reduced acyclic P2* has t_1/2 of ~ 1 hr. The conversion from an acyclic to cyclic product was complete within 3 hrs. On the other hand, examination of Figure 3B shows that the reactive intermediate P3* has t_1/2 ~ 4 hrs, four times longer than that of P2*. The kinetic profiles of the two probes are markedly different, with that of P2* and cP2 following exponential decay and growth (Figure S7), respectively, and that of P3* and cP3 following Gaussian profiles (Figures S8). We expected P3* to have a longer half-life than P2* because of the improved base-stacking interaction; however, we did not expect it to be of this magnitude. Interestingly, even after 5 hrs, we did not observe the formation of concatenated intermolecular P2 products in the absence of RNA target (Figure S9). One possibility could be due to the steric hindrance of leucine residue at the C-terminus, and the other due to the relatively low concentration of probe. Overall, the data shows that the lifetime of the reactive intermediate can be extended, if necessary, by increasing the degree of base-stacking interaction.

Figure 3.

MALDI-TOF spectra of (A) P2 and (B) P3 at 1 μM each following their incubation with DTT (100 μM) at 37 °C for the indicated time-points. Square: parent P2 and P3 probes, opened circle: reduced acyclic P2* and P3*, and filled circle: cyclic cP2 and cP3 products. Each chemical species has two masses, M and M+K⁺ (39 Daltons).

With this promising result, we next determined whether RNA can serve as a template for the concatenation of P2. We selected 5’-G(CUG)_nC-3’ as model RNA targets, where n = 1, 2, 4, 8, and 12 (Scheme 2B). This particular sequence is of biomedical interest because its expansion (> n ~ 40-repeats) is responsible for the etiology of DM1, a debilitating neuromuscular disorder for which there is no effective treatment or cure. The pathogenicity of DM1 is largely attributed to RNA toxicity.^[23] Upon transcription, the expanded rCUG-repeats (rCUG^exp) adopt an imperfect hairpin structure (Scheme 2B) which sequesters the muscleblind-like protein 1 (MBNL1), an alternative RNA splicing factor, along with several other key regulators. Formation of the rCUG^exp-MBNL1 complex would not only prevent MBNL1 from carrying out its normal physiological function but would also trap htt transcript in the nucleus, preventing its export to the cytoplasm for production of the essential DMPK protein. Thornton and coworkers^[24] have shown that displacement of MBNL1 protein from rCUG^exp by a 25-nt morpholino oligonucleotide reversed the disease phenotypes in an animal model. Further, it was shown that beyond the repeat-length of five, the thermal stability of r(CUG)_n remains fairly constant, with T_ms in the 70-75 °C range.^[25] This result was corroborated by our finding (Figure S10), indicating that increasing the number of triplet-repeats beyond five does not necessarily make the hairpin structure thermodynamically more stable. This finding suggests that rCUG-repeats might be able to template P2 oligomerization.

UV-melting experiments were performed to determine the effect of P2 on the thermal stability of RNA. In all the experiments, we observed the T_ms of T4, T6, T8, and T12 to be significantly higher with DTT than that without (Figure S11). Figure 4 shows representative UV-melting profiles of T12 and its combination with P2 following their incubation with DTT at an ambient temperature and at 37 °C for 1 hr. At both incubation temperatures, the improvements in the thermal stability of T12 were similar, with ΔT_ms ~ +20 °C—albeit, slightly less efficient at an ambient temperature as compared to that at 37 °C, as inferred from the residual melting of the unbound T12 in the 40-60 °C temperature regimes. A similar experiment was carried out with P3, but the T_ms were too high to be determined by UV-melting measurements (data not shown). Neither DTT nor P2 alone had a notable effect on the T_m of T12 (Figure 4, Inset). Similarly, neither the C-terminal amide P4 probe nor the single-base mismatched P5 had a net positive effect on the thermal stability of T12 following their incubation with DTT (Figure S12). The evidence for RNA-templated concatenation of P2 was further corroborated by MALDI-TOF analysis (Figure S13), which showed new m/z peaks corresponding to that of the concatenated products. However, we were not able to observe a significant amount of oligomeric product larger than a hexamer (6-concatenated probes) of being formed. A likely possibility could be due to the unusually high-binding affinity of this particular class of MPγPNA molecules.^[13] Once formed, they remained bound to RNA template, which cannot be detected by MALDI-TOF, despite concerted efforts to dissociate the complex under extreme thermal and chemical denaturation conditions. This conjecture was based on the finding that the peak intensities of the concatenated P2 products correlate with the concentrations of chemical denaturants up to a certain point (4 mM), beyond which suppression of the ion signals was observed. Nevertheless, the data shows that the (CUG)₁₂-RNA hairpin structure is capable of templating P2 concatenation, with the resulting products bound tightly and selectively to RNA template.

Figure 4.

UV-melting profiles of T12 and its combination with P2, with the latter recorded after incubation with DTT at an ambient temperature and at 37 ° for 1 hr. The concentration of T12 was 1 μM (corresponding to 12 μM ligand binding sites); P2, 12 μM; and DTT, 100 μM. The samples were prepared by incubating the pre-annealed T12 with P2 and DTT at the indicated temperatures for 1 hr before subjecting to UV-melting analyses. Inset: UV-melting profiles of the negative controls carried out under identical conditions.

In conclusion, we have shown that a relatively short MPγPNA probe, 3-nt in length, containing C-terminal thioester and N-terminal cystine, does not undergo immediate intramolecular NCL reaction upon the reduction of disulfide bond. The reactive acyclic P2* intermediate has a half-life of ~ 1 hr at a physiological temperature, well-beyond a typical nucleic acid hybridization time-frame.^[26] The lifetime of the reactive species can be extended, if necessary, by increasing the π-π interactions of nucleobases, as seen with P3. Despite the r(CUG)_n-repeats (n > 5) adopting their own secondary hairpin structures, the triplet nucleic acid probe can still recognize the RNA targets and undergo template-directed concatenation. Nucleic acid-catalyzed oligomerization has been previously demonstrated, particularly, by Lynn,^[27] Liu,^[28] and Seitz;^[29] however, none of these systems contained mutually reactive C-terminal thioester and N-terminal cysteine in the same nucleic acid recognition module, and none were designed to bind to biomedically relevant and structured RNA targets, such as CUG-repeats. Overall the work provides a proof-of-concept for the design of short nucleic acid probes for targeting RNA-repeated expansions through template-directed oligomerization, lending the possibility for the development of therapeutic interventions for DM1 and other related neuromuscular and neurodegenerative disorders. The potential benefits of such relatively small nucleic acid probes for biological and biomedical applications are many, including the ease of chemical synthesis, structural and functional modifications, and scale-up (via solution as opposed to solid-phase chemistry), and the improvements in pharmacokinetic properties in comparison to the conventional antisense molecules.