Periodic Melting of Oligonucleotides by Oscillating Salt Concentrations Triggered by Microscale Water Cycles Inside Heated Rock Pores

Abstract

To understand the emergence of life, a better understanding of the physical chemistry of primordial non-equilibrium conditions is essential. Significant salt concentrations are required for the catalytic function of RNA. The separation of oligonucleotides into single strands is a difficult problem as the hydrolysis of RNA becomes a limiting factor at high temperatures. Salt concentrations modulate the melting of DNA or RNA, and its periodic modulation would enable melting and annealing cycles at low temperatures. In our experiments, a moderate temperature difference created a miniaturized water cycle, resulting in fluctuations in salt concentration, leading to melting of oligonucleotides at temperatures 20°C below the melting temperature. This would enable the reshuffling of duplex oligonucleotides, necessary for ligation chain replication. The findings suggest an autonomous route to overcome the strand-separation problem of non-enzymatic replication in early evolution.

Introduction

The strand separation of oligonucleotides remains a major challenge for the continuous non-enzymatic genomic replication on the prebiotic Earth.^[1] Without a simple means of DNA/RNA denaturation, the conversion of single strands to double strands by chemical replication leads to a dead-end. Moreover, the fast reannealing of separated strands can prevent template copying, since the timescale of reannealing can be orders of magnitude faster than the timescale of chemical replication.^[2,3]

Salts like NaCl, KCl or MgCl₂ modulate the melting temperature of oligonucleotides over a wide range. By screening the interactions between the phosphate charges, ions have a direct role in stabilizing DNA or RNA double strands. Conversely, low salt concentrations have a significant destabilizing effect.^[4,5]

A natural cause for the modulation of various salt concentrations comes from the terrestrial hydrological cycle powered by solar irradiation.^[6] We found that at smaller scales, for example in a porous rock, a microscale analogue of the water cycle is implemented by a moderate temperature difference (Figure 1a,b). In our experiments, the evaporation on the warm side released salt-free water vapor into the gas void already at moderate temperatures. Instead of generating rain, snow or hail, we found submillimeter-sized water droplets that grew and precipitated at the colder side of the chamber. They fused by surface tension and gravitated stochastically into the original salt solution. This generated spikes of low salt concentration. In this setting, we studied the melting of DNA and RNA by fluorescence.

Figure 1. Water cycle at the microscale.

a) A microscale water cycle between water and gas Is driven by a thermal gradient In a porous rock. b) The process can be compared to the global-scale hydrological cycle on Earth that is driven by solar radiation. c) Experimental geometry with the phase changes of water. A temperature difference between 67°C and 55°C is applied across a 500 μm gap that is filled with water and air. The black arrows indicate the direction of the microfluidic water cycle: water evaporates at the warm side, condenses into droplets at the cold side where they fall back into the water by gravity. d) Photo of the microfluidic setup. The temperature difference was created by a transparent, heated sapphire here shown on top of a cooled silicon back plate. The chamber volume (30 mm × 14 mm) was cut out of a 500 μm thin spacer foil made from Teflon. e) A bright field image through the sapphire revealed the droplets of pure condensed water on the cold wall above the gas-water interface. The bottom remained dark. f) The fluorescence image showed the labeled DNA right after a condensed water droplet fell into the solution. The vortexes are created by diffusion and convection. Also seen is the characteristic accumulation of DNA at the gas-water interface caused by the evaporation dynamics studied recently.^[8] At the top part, the condensed droplets are not seen since DNA did not evaporate in the water cycle

Results and Discussion

We reconstructed a heated rock pore on the millimeter scale (Figure 1c,d). An applied temperature difference was used to drive the microscale water cycle, reconstructing a ubiquitous setting on the early Earth.^[7] The salt-rich solution of NaCl and DNA evaporated pure water predominantly at the warm side. The condensation dynamics of water-vapor droplets on the cold side was imaged with both white light and fluorescence (Figure 1e,f and Supporting Information, Movie 1). In the closed-pore setting, this process ran continuously with a stochastic characteristic.

The highly pure water droplets fell back into the bottom salt solution and were mixed into the bulk solution by surface tension, convection, and diffusion. This led to locally diluted salt and nucleic acid at the gas–liquid interface. Our experiments showed that they are sufficient for the denaturation of RNA or DNA well below their melting temperature.

Interestingly, the dynamics was found to slow down for higher salt concentrations due to the enhanced density difference between the lighter pure water and the denser bulk solution. This led to significantly slower mixing times and kept double-stranded oligonucleotides separated for longer times in a metastable salt gradient. Furthermore, the frequency of the salt oscillations was enhanced by reducing the ambient pressure in the experiments. This simulated possible barometric conditions on higher regions of the Early Earth ^[9,10] and was attributed to increased evaporation rates.^[11]

We investigated the effect of the water cycle on the conformation of duplex DNA or RNA using Förster resonance energy transfer (FRET) under the alternating laser excitation (ALEX)^[12] illumination protocol. Complementary strands were labeled with FRET-compatible probes: FAM (Carboxy-fluorescein) as the donor fluorophore and ROX (Carboxy-X-Rhodamine) as the acceptor fluorophore.^[13] The two fluorescent dyes (FAM and ROX) were positioned on opposite strands (Supporting Information, Figure S2a).

Two excitation alternating light-emitting diodes (LEDs, blue and amber) were associated with an optical image splitter and provided access to the emission channels of FAM and ROX. By exciting the acceptor ROX, we measured the DNA concentration. The FRET intensity measured the separation of strands and was calculated from all three spectral images after correcting for the crosstalk between the channels. A detailed explanation of the FRET calculation and setup is given in the Section 2 of the Supporting Information. A high FRET occurred when oligonucleotides were double strands. When the oligonucleotides denatured and the strands separated, the FRET signal decreased.

FRET imaging was initially used to measure a melting curve in the water phase of the chamber (Figure 2a). The front and back plates of the chamber were heated simultaneously. This also allowed the calibration of the maximal and minimal FRET levels. We started with a 51mer (51 bp) dsDNA that revealed a melting temperature of T_m ≈ 67 °C in 50 mM NaCl solution. The following experiments were performed at an average chamber temperature of 61 °C with a temperature difference between the warm and cold side of 12 °C. The mixing dynamics, which we will describe in the following, were imaged at the cold side where the condensation droplets are localized. There, the temperature was 55 °C, which is 12 °C lower than the melting temperature of the observed DNA.

Figure 2. DNA denaturation at the gas-liquid interface by droplet precipitation.

a) Melting curve of 51mer dsDNA in 50 mM NaCl measured by FRET inside the reaction chamber. The dashed line is a sigmoidal fit. b) Series of images showing the microfluidic precipitation of a pure water droplet at the gas-water interface. Top: DNA fluorescence (acceptor fluorescence), bottom: FRET signal. The white square indicates the region where the FRET signal was averaged. The uncertainty of the FRET signal was estimated to 0.08. c) Averaged FRET signal at the gas-water interface over time. Experimental conditions were: 50 mM NaCl, 10 mM Tris, 1 mM EDTA, pH 7.5, 5 μM DNA, temperature gradient of 12°C (hot and cold temperatures 67°C and 55°C, respectively), average T = 61 °C, P_atm = 0.2±0.1 bar.

As shown in Figure 2b and in Movie 2 in the Supporting Information, the falling of a microfluidic raindrop created a diluted area at the gas–water interface, where the FRET signal dropped from an initial value of 0.81 down to 0.14 and reannealed with a time constant of 16 s (Figure 2c). In the absence of a water cycle, a FRET value of 0.14 could be achieved only for a temperature greater than or equal to 71 °C (Figure 2a), indicating that the salt cycling was decreasing the DNA melting curve by 16 °C. As shown in the time traces of the FRET signal, averaged over the shown white rectangles (Figure 2c), this occurred periodically in our system and with different magnitude, depending on the size of the precipitating droplet.

The size of the condensation droplets could be estimated from low levels of fluorescence (Figure 2b, broken sphere). Especially later in the experiments, droplets revealed a small amount of DNA fluorescence. This could have been picked up from DNA that dried at the cold walls. It was likely deposited there when the water–air interface moved up and down, leaving traces of dried DNA at the moving interface. As seen in Figure 2b, the upward motion correlated with the enhanced water phase volume after dissolution of the droplet.

This remaining fluorescence in the condensation droplet also allowed us to calculate its internal FRET signal. We found a FRET <0.2, reporting a continuously denatured DNA in its low salt condition. To speed up the droplet dynamics, the experiment was performed at reduced pressure (P_atm = 0.2 ± 0.1 bar), as described later.

To understand the impact of salt on the hybridization of DNA, we determined melting curves with FRET for the aforementioned 51mer dsDNA for varying NaCl concentrations. By reducing the salt concentration, the T_m could be tuned widely from 88 °C down to 11 °C. It is therefore easy to imagine how fluctuations in the concentration of solutes could impact the hybridization of oligonucleotides.

To provide an additional control, melting curves were measured in a standard thermocycler (Figure 3a and Supporting Information, Figure S4). Since here the oligonucleotide concentration remained constant, FRET could be monitored simpler by the fluorescence of the donor fluorophore. In the duplex form, the donor fluorescence is quenched by FRET, and increases when the two strands separate. As seen, the melting curves in the thermocycler confirmed the measurements inside the gradient chamber.

Figure 3. DNA melting by NaCl concentration variation.

a) Denaturation curves of the 5lmer DNA at different NaCl concentrations measured by FRET in the thermocycler. Buffer concentrations were: 10 mM Tris and 1 mM EDTA for 500, 250, 100, 50 mM NaCl with DNA at 5 μM. At 10 mM NaCl, we used 1 mM Tris and 0.1 mM EDTA. No salt buffer was used for the pure water condition (Milli-Q water). b) Melting temperature T_m versus NaCl concentration. Dashed lines are fitted curves using a hybridization model. c) An experimental FRET time trace was compared to a simulation of the droplet dissolution. d) Snapshots of the simulation at different times are shown, demonstrating how convection and diffusion leads to fast melting dynamics of the DNA. The full simulation is provided as Movie 3 in the Supporting Information.

We obtained a deeper insight into the strand-separation dynamics by finite element simulations (COMSOL Multiphysics). The melting curves were used to fit the steady state of a salt-dependent kinetic hybridization model of DNA. On and off rates were calculated assuming a salt-independent and temperature-independent annealing rate (k_on) of 0.4 μm⁻¹ S⁻¹ measured by hybridization experiments.^[14] The strand separation rate (k_off) was calculated through the Van’t Hoff relationship:

$$\ln K_{\text{eq}}=-\frac{\Delta H}{RT}+\frac{\Delta S}{R}$$ (1)

where K_eq = k_on/k_off and R is the gas constant. ΔH (the enthalpy change) and ΔS (the entropy change) have been estimated in function of NaCl concentration, based on the Van’t Hoff plots (Supporting Information, Figure S5) obtained from the experimental melting curves of Figure 3a.

This reaction kinetics was inserted into the simulation as a diffusion equation of both complementary strands and subjected to the laminar convection flow of the water due to the temperature difference. For simplicity, details of the gas-water interface were not modeled. A precipitation event was simulated as a 2.5-fold dilution of solutes over a 0.5 mm² area at the top of the chamber, according to our experiments. We have modeled the water properties to account for the NaCl-dependence of the water density.^[15] Details of the simulation are provided in Section 5 of the Supporting Information.

Figure 3c,d show the chamber in its cross-section. It is seen how the incoming droplet diluted the salts and melted the DNA. As the species are subsequently mixed by convection and diffusion, the strands slowly re-annealed. The simulation was in good agreement both in magnitude and recovery timescale with the experiments. This demonstrated how the speed of oligonucleotide hybridization and diffusion is fast enough to be affected by the salt concentration spike.

To probe the generality of the findings, we investigated the effects of the water cycle for double-stranded RNA and DNA of two different lengths (24 or 51 bp, termed 24mer and 51mer, respectively) and at various NaCl concentrations from 50 mM to 500 mM. Typical FRET signals are shown in Figure 4 with parameters collected in Table 1. This shows how a microscale water cycle could periodically melt oligonucleotides under various salt conditions. A peculiar, slower mixing was observed for the case of 500 mM NaCl, which we attributed to the enhanced density difference between the pure water droplet and the high salt concentration in the water phase below. For example, seawater is on average 28 kgm^-3denser than freshwater,^[16] leading to the thermohaline circulation between water masses that occurs in the ocean.^[17] Due to the density difference, the lower FRET signal persisted significantly longer. Indeed,when the finite element model was fed with the salt-dependent density of water, it described this effect very clearly (Figure 4b, dashed line). Additional details on the effects of the salinity gradient in our system can befound in the Section 6 of the Supporting Informatinon.

Figure 4. Investigation of different oligonucleotides and NaCl concentrations.

a) Representative snapshots (DNA or RNA fluorescence and FRET) for various NaCl concentration, nucleic acid type, length, and temperatures. The roman numbers link to Table 1. White squares indicate the region where the FRET signal was averaged. If not otherwise reported, the buffer contained 10 mM Tris, 1 mM EDTA, at pH 7.5. Oligonucleotide concentration was 10 μM for I and II, 5 μM for Ill, and 2 μM for IV. Atmospheric pressure in all experiments was 0.2 bar to enhance the probability of observing spikes in the experiment. b) FRET time traces, simulations, and experiment. The NaCl dependency of water density (ρ) plays a significant role in our microfluidic denaturation system. This is confirmed by the agreement between our experiment (points) and the simulation (solid line). When the NaCl dependency of water density was not accounted (broken line), the reannealing time scale reduced and the agreement between simulation and experiment diminished. The conditions studied here correspond to the 24mer dsDNA, 5 μM in 500 mM NaCl (T_m 43°C) in the following temperature gradient: 15°C (cold side) and 24°C (warm side).

Table 1.

Measurement parameters for experiments in Figure 2 and Figure 4 (roman numerals) such as oligonucleotide type, length, NaCl concentration, melting temperature T_m, the average chamber temperature 〈T〉 and the temperature difference ΔT. For each experiment, the average FRET signal in the water phase is reported as well as after the dissolution event of a droplet. The last column contains the time constant for reannealing.

ID		Oligo type	Length (bp)	[NaCl] [mM]	Tm [°C]	〈T〉 [°C]	Δ〈T〉 [°C]	Initial 〈FRET〉	Minimal 〈FRET〉	τ [s]
Figure 2b		DNA	51	50	67	61	12	0.81	0.14	10 ± 4
	I	DNA	51	150	82	61	12	1.00	0.21	3 ± 1
	II	RNA	51	150	88	61	12	1.00	0.41	8 ± 4
Figure 4a	Ill	DNA	24	500	43	20	9	0.98	0.21	14 ± 12
	IV	RNA	24	500	45	38	15	0.76	0.29	3 ± 1

We studied the effects of atmospheric pressure on the dynamics of the microfluidic watercycle and compared the microscale watercycle between 1 bar and 0.2 bar in Figure 5.

Figure 5. Faster water cycle at reduced pressure. Comparison of the critical droplet size.

(a) defined by the maximal size achievable by a condensation droplet before precipitation is observed. The lifetime (b) was measured by the time between a droplet nucleation and its precipitation into the water phase. Both (a) and (b) were measured at ambient pressure (1 bar) or 0.2 bar. c) and d) Kinetics of FRET over time at the gas-water interface over similar time span at ambient pressure (c) or 0.2 bar (d). Experimental conditions were: 50 mM NaCl, 10 mM Tris, 1 mM EDTA, pH 7.5, DNA 5 μM, a temperature gradient of 12°C between 55°C and 67°C. e) Representative snapshots of the microscale water cycle at 0.2 bar. Condensation droplets were seen as shiny spheres above the gas-liquid interface. A comparison with ambient pressure is shown in Movie 4 in the Supporting Information.

The lower pressure did not significantly alter the average size of the condensing droplets (Figure 5a). Rather, it increased their growth speed by a factor of about 4. This effect was exploited in our experiments before to obtain more frequent strand separation events. The statistics was obtained from analyzing white light movies shown in Figure 5e.

Conclusion

Our study demonstrates autonomous salt oscillations denature double-stranded oligonucleotides under mild conditions. Microscale water cycles in pores of rock, subjected to a temperature difference are ubiquitous conditions to implement continuous salt oscillations. As water evaporates on the warm side continuously and drops back from precipitated pure water droplets, DNA at the water-air interface is molten periodically. Under these conditions, DNA (or RNA) denaturation was observed at temperatures well below the T_m of the initial solution. For the conditions of Figure 2, we estimated that about 50 % of the water volume is subjected to the salt dilution per one hour of the experiment.

The finding is significant for the emergence of molecular evolution in the context of non-enzymatic chemical reactions for DNA or RNA. They often do not tolerate high temperatures due to hydrolysis, but require cyclic strand separation events for exponential replication. For example, ligation activated by EDC- or imidazole-activated polymerization reactions do not tolerate high temperatures^[18,19] but at the same time require DNA or RNA strands to be separated to reshuffle the strands for the next replication cycle. Future experiments will focus on combining the setting with the possibility to perform non-enzymatic replication or ligation reactions at moderate temperatures, so that the separation of the DNA or RNA strands will not be induced by heat but will be provided by salt fluctuations at the gas–water interface.