A Fluorescent G-quadruplex Sensor for Chemical RNA Copying

Abstract

Non-enzymatic RNA replication may have been one of the processes involved in the appearance of life on Earth. Attempts to recreate this process in a laboratory setting have not been successful thus far, highlighting a critical need for finding prebiotic conditions that increase the rate and the yield. Here, we present a highly parallel assay for template directed RNA synthesis that relies on the intrinsic fluorescence of a 2-aminopurine modified G-quadruplex. We demonstrate the application of the assay to examine the combined influence of multiple variables including pH, divalent metal concentrations and ribonucleotide concentrations on the copying of RNA sequences. The assay enables a direct survey of physical and chemical conditions, potentially prebiotic, which could enable the chemical replication of RNA.

COMMUNICATION

RNA lights up: The non-enzymatic replication of RNA might have been crucial to the emergence of life. We developed an assay based on the intrinsic fluorescence of a G-quadruplex that can be used to measure the kinetics of the reaction and screen for improved conditions and catalysts.

The RNA world hypothesis is a central concept in origin of life research. First proposed by Orgel, Crick and Woese^[1–3] and then elegantly enunciated by Gilbert^[4], the hypothesis states that early life forms relied on RNA for both catalysis and the transmission of genetic information. In this context, the primitive RNA genome was hypothesized to be replicated by an RNA enzyme, a replicase. However, the evolution of an effective RNA replicase is thought to have required an initial phase of chemical RNA replication. Chemical RNA replication starts with the binding of activated ribonucleotides to an RNA template through Watson-Crick base pairing. The ribonucleotide monomers react with each other to form a double-stranded RNA duplex; subsequent amplification may then require separation of the strands, so that they can then act as templates for further copying rounds. Multiple rounds of copying would lead to an exponential increase in the population of RNAs, and selective pressures for more efficient replication would eventually give rise to catalytic RNAs. Although significant progress in reconstructing the initial steps of such a process has been made over the last 50 years^[5], several problems persist, including the poor rate and fidelity of non-enzymatic copying, the difficulty of achieving multiple rounds of replication, and incompatibility with prebiotically plausible cellular membranes^[6].

A useful laboratory model for the study of templated RNA synthesis is non-enzymatic primer extension (Scheme 1). The reaction is performed by adding activated ribonucleotide monomers to a pre-formed primer-template duplex, after which the product distribution is analysed by various methods. Conventional methods used to measure the rate of and to characterize the products of templated RNA synthesis include high performance liquid chromatography (HPLC), polyacrylamide gel electrophoresis (PAGE), and mass spectrometry, all of which are cumbersome and low throughput. The current methods of analysis are often denaturing and do not provide information about the folding ability of the RNA strand that is generated in the copying process. The folding ability of the RNA strands is crucial if RNA replication is hypothesized to have given rise to RNAs with structure and function. As such, a quick and facile assay for RNA copying that also informs on the structural integrity of the newly synthesized RNA strand is needed. Such an assay would accelerate the discovery of catalysts and conditions for the template directed chemical synthesis of RNA. Here we present a high-throughput fluorescence assay which enables the screening of multiple reaction conditions in a fraction of the time required by conventional methods. The assay relies on the formation of a correctly folded RNA G-quadruplex, which is, in turn, conditional on the success of the primer extension reaction.

Scheme 1.

Template directed RNA synthesis. Activated ribonucleotide monomers are represented in yellow.

We set out to adapt a previously reported assay for enzymatic primer extension developed by Kankia and colleagues.^[7,8] The assay is based on a DNA primer that contains the fluorescent adenine analog 2-aminopurine.^[9] 2-aminopurine fluoresces when exposed to the solvent, but is quenched when incorporated in a double-stranded duplex. Starting with a primer-template duplex, the addition of two guanosines at the 3’ end of the primer by a polymerase enzyme in the presence of K⁺ removes the primer from the template to form a parallel G-quadruplex structure. In this structure, the 2-aminopurine nucleobase is exposed to the solvent, resulting in a 90-fold increase in fluorescence.^[10] We hypothesized that changing the template and primer from DNA to RNA would not affect the formation of the secondary structure since RNA is known to form stable parallel G-quadruplexes.^[11] Furthermore, instead of using a polymerase and nucleoside triphosphates to extend the primers, we used 5^’-phosphorimidazolide guanosine monomers (Figure 1a). Our group has previously demonstrated that non-enzymatic primer extension can be correlated with an increase in fluorescence of the malachite green aptamer^[12]. Here we opted to use mononucleotides activated with 2-aminoimidazole (2-AI), since the reaction proceeds quickly and in high yield^[13], and a potentially prebiotic synthetic path to 2AI has been reported recently^[14].

Figure 1.

(a) Non-enzymatic RNA primer extension using 2-aminoimidazole activated guanosine phosphorimidazolide (2AImpG) followed by a heating step liberates a fluorescent G-quadruplex. The red squares represent the 2-aminopurine residues (b) The P-GG oligonucleotide forms a G-quadruplex that contains two solvent exposed 2’-deoxyribose 2-aminopurine residues (highlighted in red and labelled dA*). (c) Incubation of P with 2AImpG for 3 hours results in similar levels of fluorescence to identically treated P-GG (positive control). In the second column 2AImpG was replaced by an equal amount of guanosine monophosphate (GMP). In the third column 20 mM EDTA was added to ensure that trace Mg²⁺ is chelated. Data are reported as the mean ± the standard error of the mean from triplicate experiments.

We began by synthesizing the RNA sequence corresponding to the fully extended primer with two 2-aminopurine residues (P-GG), which is expected to form a parallel G-quadruplex (Figure 1b). We reasoned that having two 2-aminopurine residues would improve the signal intensity and thus the sensitivity of the assay. In the presence of K⁺, the oligonucleotide forms a G-quadruplex, as shown by its characteristic circular dichroism spectrum^[10], as well as UV-melting data (Figure S1). In addition, we observed a significant shift in the melting temperature of the duplex formed by P-GG and its complementary strand in the presence of KCl, which is not observed in the case of P (Figure S1), as previously reported for the DNA analogues^[10]. The different behaviour of the two duplexes suggested that only P-GG might form a fluorescent G-quadruplex following denaturation, and that conversion of P to P-GG could therefore be monitored by a fluorescence assay. To test this idea, pre-annealed P duplex was incubated with guanosine 2-aminophosphorimidazolide (2AImpG), in the presence of 50 mM Mg²⁺, in a model primer extension reaction. The reaction was then quenched by adding EDTA, and the mixture was heated in the presence of KCl. The measured fluorescence was identical to that of the identically treated P-GG duplex (Figure 1c). In contrast, when guanosine monophosphate (GMP) was used instead of 2AImpG, we observed a 40-fold lower fluorescence intensity relative to P-GG. Most previous work on non-enzymatic primer extension has been carried out in the presence of > 0.1 M Mg²⁺ concentration^[6]. At lower concentrations, the reaction rate decreases considerably. When no divalent cations were present during the reaction, the fluorescence was similar to the GMP reaction, as expected. Taken together, these results show that the increase in fluorescence is consistent with primer extension.

To validate the assay, we synthesized a modified P primer that contains an orthogonal fluorescent label (6-FAM) at its 5’ end. The modified primer enabled us to simultaneously monitor a copying reaction by 2-aminopurine fluorescence and PAGE (Figure 2). We observed an increase in 2AP fluorescence over the time course of the reaction which was strongly correlated with the extension of the primer by at least 2 nucleotides (Figure S2 a). To show that the +1-extended primer does not contribute to the fluorescence, we synthesized the P-G primer and showed that it fluoresces similarly to the unextended primer P (Figure S2b). We confirmed that the assay is suitable for high-throughput screening given its Z’-factor of 0.90^[15]. The Z’-factor is a statistical parameter developed to measure the reproducibility of an assay for high-throughput screening by determining the separation between positive and negative controls. A Z’ of 1 is ideal, while a value below 0.5 is considered inadequate for high-throughput screening.

Figure 2.

Results from the fluorescence assay for primer extension are consistent with the standard PAGE assay for primer extension (a) A primer extension reaction was carried out with a dually labelled primer, with two internal 2-aminopurine residues and an additional 5’-fluorescein modification (FAM). (b) 2-aminopurine fluorescence was measured at different times for the primer extension reaction. Data points are reported as the mean ± s.e.m. from triplicate experiments. (c) Electrophoretogram showing the primer extension of the FAM labelled primer. The primer extension reaction was carried out at pH 8.0 in the presence of 50 mM Mg²⁺, using 10 mM of 2AImpG.

Non-enzymatic RNA synthesis of homopolymeric G and C stretches has been shown to be relatively quick and high yielding^[16]. In contrast, copying mixed sequences or homopolymeric A and U stretches is slow and often results in a mixture of truncated products and unreacted primer. The poor efficiency of mixed sequence copying precludes sequence general RNA replication and the generation of catalytic RNAs. To examine the copying of mixed sequences we removed the 3’-UG sequence of the primer P. When the resulting oligonucleotide is used as a primer in a copying reaction, extension by three nucleotides is required to generate a fluorescence signal. We then examined primer extension on templates designed to give products containing both purine and pyrimidine nucleotides (CGGG and UGGG). In comparison to the homopolymeric G extension, which is complete in 1 hour, the synthesis of the CGG moiety plateaus after 3 hours, reaching 80% of its maximum fluorescence level (Figure 3a). In contrast, the UGGG extension, although as efficient as the CGGG extension, requires 24 hours to plateau (Figure 3b). When the activated pyrimidine nucleotides are replaced by the respective monophosphates, the reaction rate and efficiency is drastically reduced. However, a significant increase in fluorescence is observed when 2AImpG and UMP are incubated with the primer-template complex for extended periods of time, presumably due to the formation of a mismatched product through G-A mispairing or through a strand slipping process^[17], in which the template forms a mononucleotide loop, excluding the adenosine residue from the templating region.

Figure 3.

Monitoring copying of mixed sequences. (a) Fluorescence time course of the copying reaction on a GCCCC template. Reactions were initiated by adding 2AImpG and 2AImpC (black) or CMP (red) (b) Fluorescence time course of the copying reaction on a ACCCC template. The reactions were initiated by adding 2AImpG and 2AImpU (black) or UMP (red). All data points are reported as the mean ± s.e.m. from triplicate experiments. Fluorescence was normalized with respect to the identically treated P-GG duplex.

Divalent metal cations are known to catalyze the chemical copying of RNA, but the mechanism of catalysis remains incompletely understood. A potential mechanism by which divalent cations increase the rate of primer extension is by presenting a coordinated hydroxide to the diol, or coordinating it, or either of its hydroxyl groups directly, to facilitate diol deprotonation. The resulting alkoxide, a strong nucleophile, would attack the activated phosphate and form a phosphodiester bond. If that were the case, then the reaction rate would depend on the ability of the metal to act as a Lewis acid, provided that nucleophilic attack is the rate-determining step. We therefore looked at how different concentrations of two different cations (Mg²⁺ and Ca²⁺) and 2AImpG affect the rate of primer extension. Using our assay, we surveyed 84 different reaction conditions in a single experiment. At all metal concentrations, Mg²⁺ is a better catalyst than Ca²⁺ for the primer extension reaction, and more P-GG is formed (Figure 4a).The difference in pK_a values of the hexaaquaions, [M(OH₂)₆]²⁺, of the two metals^[18] (pK_a =11.2 for Mg²⁺ and 12.7 for Ca²⁺) show that Mg²⁺ is more efficient in promoting the deprotonation of a coordinated water molecule relative to Ca²⁺. If a similar mechanism is operating in the primer extension reaction, higher pH values should relieve the metal dependence of the reaction. To test this hypothesis, we conducted a two-dimensional screen of pH values and different Mg²⁺ concentrations. The reaction fails at low pH values (≤7.1), or at low Mg²⁺ concentrations (≤4 mM). However, at intermediate metal concentrations the yield of the reaction is rescued by increasing the value of the pH. For example, increasing the pH from 7.4 to 8.9 at a fixed concentration of 4 mM Mg²⁺ generates similar amount of P-GG product as increasing the concentration of Mg²⁺ from 4 mM to 16 mM at a fixed pH of 7.4 (Figure 4b). This result supports the hypothesis that the divalent cations help in the deprotonation of the ribose secondary hydroxyl groups. Incubating the primer-template duplex at the highest pH value and Mg²⁺ concentration, in the absence of 2AImpG, does not lead to an increase in fluorescence (Figure S3). This result confirms that values measured in the two-dimensional screen correspond to successful primer extension and not to any other processes that might increase the fluorescence, such as hydrolysis of the primer-template duplex. Further investigations into the role of divalent metals in the primer extension reactions are currently being carried out in our laboratory.

Figure 4.

(a) Mg²⁺ outperforms Ca²⁺ as a catalyst of non-enzymatic primer extension. Reactions were incubated for 135 minutes at room temperature. Fluorescence values were normalized with respect to a 25 mM 2AImpG, P-GG replicate for each of the different metal concentrations. (b) Non-enzymatic RNA primer extension works best at higher pH values and higher Mg²⁺ concentrations. Values were normalized with respect to the control incubated at each of the Mg²⁺ concentrations. The numerical values of the normalized fluorescence represent the conditions referred to in the main text.

The templated chemical copying of RNA is one of many hypotheses that aims to explain the transition from inanimate, chemical matter to self-replicating entities capable of Darwinian evolution. Although encouraging early results have sustained research in the area for over 50 years, considerable advances are still required to achieve continued cycles of nonenzymatic RNA replication. To address the remaining hurdles, we developed a fluorescence-based assay for non-enzymatic copying of RNA. Under conditions in which RNA copying proceeds quickly and in good yield, a large gain in signal is observed that is strongly correlated with the formation of a fluorescent RNA G-quadruplex. We have shown that the method can be used to measure the kinetics of the copying reaction. In addition, the assay can be used in high-throughput screens of the parameters affecting the copying process.

An ideal fluorescence assay for chemical RNA copying would be able to measure the extent of copying for any given template sequence in real-time. Our assay relies on the formation of a G-quadruplex and is therefore limited to generating guanosine-rich RNA strands. Additionally, after the copying is concluded a heating step is required to liberate the G-quadruplex from its complementary strand. However, a simple heating step is more amenable to highly parallel screening than traditional PAGE assays or mass spectrometry, which typically require lengthy sample preparations. However, mass spectrometry can identify mismatched copying, as well as the identity of the mismatched nucleotide, while PAGE, despite being laborious, is sensitive and versatile, and remains the gold standard for quantification. Our newly developed fluorescence assay complements mass spectrometry and PAGE with its much higher throughput. The parallel nature of the assay enabled us to observe the combinatorial effects of multiple variables such as the concentration of activated monomer, the pH of the reaction, as well as the concentration and the nature of the divalent metal cations on the efficiency of the copying reaction. In addition, the assay is not limited to RNA, and could potentially work for any nucleic acid that can form G-quadruplexes.

Prebiotic synthetic pathways to biomolecules have benefited greatly in recent years from adopting a systems chemistry approach^[19]. The chemical routes to the different classes of biomolecules (nucleic acids, lipids, amino acids) have been evaluated by taking into consideration the possible interactions between the starting materials, the types of chemistries and the nature of the reagents used. As a result, more robust pathways, common to all three classes of biomolecules, have been uncovered^[20]. The same approach has not yet been used to explore superior routes to chemical RNA replication. The nonenzymatic assay developed herein should facilitate a systems approach to the long-standing problem of chemical RNA replication.