In CO2-rich AWIs, nucleic acids are exposed to periodic changes in Mg2+ concentration and pH level, the latter of which originates from dissolved carbonate in dew droplets20,21 (Fig. 1A, Figure S1). While the low pH conditions enable transient melting of otherwise stable nucleic acid duplexes, the co-accumulation of RNA and magnesium ions at water-gas interfaces by capillary flow promotes folding and catalysis22,23. We therefore wondered if AWIs are a suitable environment for repeated RNA-dependent RNA replication. Derivatives of the self-splicing sunY intron from bacteriophage T4 can catalyse template-dependent oligonucleotide ligation using 5'-guanosines as a leaving group4,5. Due to their robust activity and simple activation chemistry, sunY variants are an attractive model system for primitive enzymatic RNA-replication24,25. Initially, we explored the Mg2+ concentration requirements for RNA ligation by a 182 nucleotide (nt) variant of the sunY ribozyme (Figure S2) in CO2-rich AWI conditions (Figure S3-4). To this end, we probed the template-dependent ligation of a 30 nt RNA from three oligonucleotide substrates (Fig. 1B, Table S1): a 5’-Cy5 labelled substrate (M1, 13 nt), which allows direct fluorescence-based PAGE analysis, and two downstream 5’-guanosine activated substrates, M2 (7 nt), and M3 (10 nt). Due to the concentration effect at the evaporation zone of the heated side of the air-water interface, a significantly lower Mg2+ bulk concentration (5 mM) was required to observe near-complete substrate ligation ratios (89%) compared to isothermal conditions (50 mM, 86%) after 4 hours of incubation (Fig. 1C). We observed ligation in the AWI system even at concentrations as low as 1 mM MgCl2 (17%), whereas no detectable ligation occurred in a comparable equilibrated system at constant temperature (Figure S5). This confirmed that the non-equilibrium conditions in the AWI-system efficiently drive sunY-dependent RNA ligation at magnesium concentrations lower than those required in isothermal conditions.
We hypothesized that the periodic local pH decreases, resulting from precipitation of acidified dew droplets (Figure S6), could lead to a transient decrease in RNA melting temperatures thereby promoting the release of ligation products from their template and allowing intramolecular folding into functional RNAs. This encouraged us to attempt to synthesize strands with the same or similar sequence as sunY. The Szostak group previously demonstrated that an active version of sunY can assemble non-covalently from three oligonucleotides (A, B, and C) between 43 and 75 nt in length4. We reasoned that the AWI-system might circumvent template-inhibition during templated RNA ligation and therefore explored if AWI-based non-equilibrium environments could also support replication of sunY-derived RNA strands. We initially explored if AWI-systems enable sunY to catalyse the synthesis of each of the three fragments A, B, and C from three short oligonucleotides substrates (A1-3, B1-3 and C1-3) (Table S1). To this end, we probed the ligation of each fragment from three substrate strands starting from equal concentrations (2.5 µM) of sunY and template and 4-fold excess of each ligation fragment (Figure 2A). We observed good yields of full-length products (38.5±1.5% A123, 19.5±2.5% B123 and 22±7% C123) after 2 h of reaction in the AWI. In contrast, full-length yields were considerably lower under isothermal conditions (~8% A123, ~1% B123 and ~1.3% C123 after 2 h), agreeing with the predicted melting temperatures and suggesting substrate inhibition as the cause (Table S2). In the AWI-system, synthesis of full-length products was favoured, and only low amounts of remaining intermediate species were observed after incubation, while under isothermal conditions, incomplete intermediates were produced in excess over full-length products for all three sunY fragments (Figure S7).
After demonstrating that all three sunY fragments can be synthesized by the full-length ribozyme in situ, we sought to explore if AWIs can also support full cycles of RNA replication, synthesizing both sense and antisense strands of an RNA in a single reaction environment (Figure 2B). To this end, we included substrate oligonucleotides for the sunY fragment (e.g. C1-3) as well as its respective template (e.g. tC1-3) into AWI ligation reactions. We also included a seed amount of full-length template to initiate templated ligation of the full-length sunY fragment. After 3 h of incubation under AWI conditions, we detected formation of sense (C123, A123) as well as antisense strands (tC123, tA123) (Figure 2C). Intriguingly, we detected the formation of both products in the absence of seeding template irrespective of whether the complementary ligation fragments formed a blunt duplex or one with overhangs (Figure 2D, Figure S8), suggesting that the sunY ribozyme itself acted partially as both, catalyst for ligation and template for the synthesis of the complementary fragments.
We speculated that release of the ligation products from their template might lead to folding of the products into functionally active RNAs. To test this hypothesis, we designed a sunY based replication scenario for a minimal version of the hammerhead ribozyme (HH-min) derived from the tobacco ring spot virus satellite RNA (Figure 3A)26,27. In this system, HH-min is synthesized by sunY from three fragments (HH1-3, 11, 11, 16 nt), with ligation junctions interrupting critical catalytic motifs of HH-min, such that catalysis of the fragmented HH-min was prevented (Figure S2, Table S1). While experiments in both the AWI-system and under standard isothermal conditions led to the synthesis of HH-min, the AWI-system promoted the synthesis of HH-min even at a 10-fold lower MgCl2 concentrations (5 mM) than required for isothermal ligation (50 mM, Figure S9), corroborating our previous data regarding magnesium dependence under non-equilibrium conditions. Kinetic measurements for both conditions further revealed that the AWI-system indeed enabled multi-turnover synthesis of HH-min as the resulting yields of the full-length HH-min, which peaked after about 4 h of incubation and exceeded the amount of input-template RNA (Figures S10-12). In contrast, no excess synthesis of HH-min over template could be observed under isothermal conditions, indicating that all product strands remained tightly associated with the template. The observation of multi-turnover ligation in the AWI-system suggested the presence of single-stranded, de novo synthesized HH-min in solution. To probe if this dissociated RNA species was catalytically active, we repeated the experiment in presence of the cognate HH-min substrate (HH-sub) with the aim of observing catalytic substrate cleavage. Satisfyingly, PAGE analysis of the reaction mixture confirmed that HH-min formation was accompanied by cleavage of HH-sub under non-equilibrium conditions (cleavage ratio of 40±13% and 60±7% after 2 h and 4 h, respectively), validating that product dissociation, folding, binding and cleavage occur in addition to sunY-catalysed ribozyme assembly in a one-pot system (Figure 3B-C, Table S3). In contrast, no HH-sub cleavage was observed under isothermal conditions, suggesting that the newly synthesized HH-min was unable to dissociate from the template under these conditions (Figure S13). This interpretation agrees with the predicted melting temperature of the HH-min-template complex at the magnesium concentrations used in the reaction (90 °C, Table S4). As expected, control experiments in the AWI system without sunY resulted in no observable HH-min assembly or HH-sub cleavage.