2.1 Enzymes, nucleotides, and oligonucleotides

T7 RNA polymerase was purified from Escherichia coli strain BL21 harboring plasmid pAR1219 (kindly provided by Dr. Smita S. Patel, Robert Wood Johnson Medical School, Piscataway, NJ, USA). T4 polynucleotide kinase and calf intestine phosphatase were purchased from New England Biolabs (Beverly, MA, USA). Nucleoside 5′-triphosphates were purchased from Roche (Mannheim, Germany), and [γ-³²P]ATP (6 μCi/pmol) was obtained from Amersham Biosciences (Piscataway, NJ, USA). Oligodeoxynucleotides were synthesized by Bioneer (Daejeon, Korea) and purified using denaturing polyacrylamide gel electrophoresis (PAGE).

2.2 Preparation of ribozyme and RNA substrates

The ribozymes T and T′ and the RNA substrates A, A′, B, and B′ were prepared by in vitro transcription as described previously [20]. Briefly, the transcription mixture contained 15 mM MgCl₂, 2 mM spermidine, 5 mM dithiothreitol, 50 mM Tris (pH 7.5), 0.4 μM DNA template, 0.8 μM annealing oligoDNA (sequence 5′-GGACTAATACGACTCACTATA-3′, the T7 promoter sequence is shown in bold face), 2 mM each of the four NTPs, and 30 units/μl T7 RNA polymerase. The mixture was incubated at 37 °C for 2 h, and then quenched with an equal volume of gel-loading buffer (15 mM EDTA and 8 M urea). The synthesized RNAs were purified by denaturing PAGE, and desalted using a C18 Sep-Pak^® cartridge (Waters, Milford, MA, USA). The substrate RNAs A and A′ were 5′ labeled using T4 polynucleotide kinase and [γ-³²P]ATP, following removal of the 5′ triphosphate with calf intestine phosphatase; then the RNAs were gel-purified and desalted.

2.3 RNA-catalyzed reaction

RNA ligation reactions were performed in a mixture containing 2 μM ligation substrates (5′-labeled A and/or A′, unlabeled B and/or B′), 1 μM RNA ligase (unlabeled T and/or T′), 25 mM MgCl₂, and 50 mM N-[2-hydroxyethyl]-piperazine-N′-[3-propanesulfonic acid] (EPPS, pH 8.5). The mixtures were incubated under isothermal (23 °C) or alternating (periodic heat pulse, 55 °C for 2 min) temperature conditions. The reactions were initiated by mixing equal volumes of two solutions, one containing the substrates and the other containing the ribozymes. Aliquots were removed at various times and quenched by the addition of an equal volume of gel-loading buffer. The reaction products were resolved by 12% PAGE containing 8 M urea and quantitated using a Cyclone storage phosphor system (Packard, Meriden, CT, USA).

2.4 Kinetic analysis

The fraction of the substrates that had reacted at each time point was determined, and the time-dependent accumulations of the ligated products were fit to an exponential function: fraction reacted = a(1 − e^−kt ), where a and k are the amplitude and rate of exponential increase of the products, respectively. The initial rate of reaction was obtained by multiplying a and k. The data for cross-catalytic replication were fit to a logistic equation for sigmoid growth: fraction reacted = a/(1 + b e^−kt ), where a is the maximum extent of reaction, k is the rate of growth, and b is a floating parameter for fitting. All of the kinetic results represent the averages of at least three replicate experiments and are presented as means ± S.E.

3.1 Design of cross-replicating ligase ribozymes and their substrates

Amplification of the RNA molecule under alternating temperature conditions resembles the general scheme of the polymerase chain reaction. As illustrated in Fig. 1 , the RNA ligation reaction was adapted instead of a polymerization reaction to copy the complementary strand of the template RNA. Note that to overcome the rate-limiting product dissociation, while allowing for efficient substrate binding, periodic heat pulses were applied in the ligase ribozyme reaction system. Starting from the self-replicating ribozymes and their substrates constructed previously [20], two ligase ribozymes serving as templates for amplification (T and T′) were redesigned for use in the amplification process. The ligase ribozyme contains the catalytic core sequence of the R3C ligase ribozyme [18] that is crucial for the ligation reaction (Fig. 2 , shaded sequences). Each template ligase was designed to recognize and ligate its cognate substrate pair (A′ and B′ for T; A and B for T′), thereby generating the opposite template.

Compared to the previous ligase ribozyme constructs [20], the terminal part of the substrate binding arm that determines substrate specificity (i.e., the P3 domain) was modified to enhance the stability of the ligase template/substrate pair by introducing one G–C pair into the P3 stems (Fig. 2, shown in bold face). Under alternating temperature conditions with a heat pulse, the ligated product (opposing template) was easily dissociated, without limiting the rate of the entire catalytic cycle. Thus, the stability of the P3 stem was improved over the ligase/substrate pair reported previously to allow for tight binding of the cognate substrates. The length difference required to distinguish the ligated products and catalytic activity in the four substrate reaction system was optimized previously [20]. Consequently, the only changes made to the ligase construct involved the P3 stems and optimization of the design of the template and corresponding substrates, as shown in Fig. 2.

3.2 RNA ligation reactions with template ligase ribozymes

Each pair of substrates was tested for ligation with either the corresponding template or the mismatched template at 23 °C (Fig. 3 A). Previously, the R3C ligase ribozyme was investigated for RNA substrate ligation with varying concentrations of the RNA ligase [18]. The initial rate of the ligation reaction increased linearly as the initial concentration of RNA ligase increased, implying that binding of substrates to the template RNA ligase follows first-order reaction kinetics. Based on the previous reaction conditions [19, 20], we used 2 μM RNA substrates and 1 μM RNA template ligase as the standard reaction. In the presence of ligase template T′, substrates A and B were ligated efficiently. However, the mismatched template ligase T showed no ligation activity for substrate A and B. Similarly, A′ and B′ were only ligated efficiently in the presence of template T. Time course plots of the template-directed ligation reactions were fit to an exponential function, giving an amplitude of 0.19 ± 0.008 and 0.30 ± 0.011 and an exponential rate of 0.017 ± 0.0019 min⁻¹ and 0.022 ± 0.0023 min⁻¹ for reactions catalyzed by T′ and T, respectively (Fig. 3B). In the presence of 2 μM RNA substrates (A and B, or A′ and B′), a linear relationship was observed between the starting concentration of template ligase and initial reaction rate (data not shown).

Since the ligated product (T · T′) is readily dissociated at high temperature, one of the ligation reaction mixtures (A and B ligation by T′) was subjected to alternating temperature conditions. At 30 min intervals, the reaction vessel incubated at 23 °C was transferred to another bath at a higher temperature (55 °C) for 2 min and then returned to the 23 °C bath—similar to a conventional PCR temperature profile. Conditions for this periodic heat pulse (55 °C for 2 min) were chosen to ensure that RNAs in the reaction were not degraded by the high temperature, and that the ligated product complex was fully dissociated. The heat pulse process was repeated over the entire time course. Progressive accumulation of the ligated product (T) occurred without saturation, and exceeded the amount of product that was obtained by performing the reaction at constant temperature (Fig. 3B). To address whether the heat pulse strategy can provide conditions for the unlocking of misfolded ribozyme–substrate complexes instead of efficient product dissociation, the ligase reaction mixture was initially heated at 55 °C for 2 min and then transferred to a vessel incubated at 23 °C. This initial heat treatment did not enhance the yield of ligation reaction products, and the kinetic properties of the reaction were unchanged (data not shown). Thus, periodic heat treatment contributes mainly to the dissociation of the RNA ligation product complex, and does not affect the initial RNA folding state that may influence a competent ligation reaction. However, we do not know why the average extent of the ligation reactions only reached up to 20–30% of the total substrates present in the reaction mixture (Fig. 3B, isothermal reactions), which reflects the fraction of competent RNA ligase–substrate complexes in a reaction. Since the binding affinity of the P3 arms in the construct increased, the entire ligase reaction cycle is likely limited by the product dissociation step, rather than by substrate binding. The heat pulsed reactions thus yielded more products by taking advantage of the forced product dissociation at the higher temperature. Therefore, alternating temperature conditions is an appropriate means of enhancing the turnover rates of RNA ligations by the template ligase ribozyme in a reaction.

At a constant temperature of 23 °C, the template-directed RNA ligation reaction was also performed using all four substrates: [5′-³²P]-labeled A and A′ and unlabeled B and B′, each present at 2 μM in the absence or presence of a ligase ribozyme (either T or T′ templates, 1 μM). In this scheme, four ligation products may form, which can be resolved by gel electrophoresis: A–B (equivalent to T) is 66 nucleotides (nt), A′–B′ (equivalent to T′) is 78 nt, A′–B is 74 nt, and A–B′ is 70 nt. Ligation products were not observed in the absence of ligase template, whereas the addition of each ligase (T or T′) resulted in a large increase in the formation of directed ligation products (Fig. 3C). At constant temperature, each ligase template catalyzed the ligation of predominantly the corresponding pair of substrates over the other possible pair. However, additional ligation products resulted from mismatched substrates. As shown in Fig. 3C, template ligase T assisted A–B′ ligation and T′ catalyzed A′–B ligation to some extent.

We compared the newly constructed ligase templates to those reported previously [20] with respect to the ligation of mismatched substrates. Both the old and new ligase ribozymes catalyzed ligations of mismatched substrates; A and B′ were ligated in the presence of T, while A′ and B were ligated in the presence of T′ (Fig. 3D). Thus, the ligation reaction appears to tolerate mismatches within the P3 stem that binds the longer substrate (A or A′), but not within the P3 stem that binds the shorter substrate (B or B′) (see inset of Fig. 3D). The new version of the ligase templates used in this study caused a decrease in the observed mismatch ligation. This result indicates that the increased binding affinity in the P3 stem provided better specificity to the template ligases. Stabilization of the directed ligation reaction complex over the mismatched ligation reaction is likely attributable to the enhanced specificity and product yield of the newly designed template ligases.

3.3 Cross-catalytic amplification of RNA templates under alternating temperature conditions

The template-directed RNA ligation reaction was performed by applying unlabeled T and T′ templates (1 μM each) to the substrates (³²P-labeled A and A′, and unlabeled B and B′, 2 μM each) in a reaction mixture at a constant 23 °C. Each pair of substrates (A′ and B′, A and B) were ligated following an exponential kinetic pattern (Fig. 4 , isothermal reaction). A hybrid product from A and B′ ligation was also formed. The yield of major ligation products (T and T′) was roughly 5- to 10-fold lower in the presence of all four substrates compared to the reactions carried out in the presence of only two matched substrates. This was likely due to the non-productive RNA complex formation and the promiscuous reactions that occur in the presence of all four substrates (Fig. 3C), which divert materials from the directed ligation reaction. Since four RNA substrates were present, which could have potentially paired with each cognate partner in the reaction mixture, a significant portion of the RNA substrates could form non-productive dimeric complexes, such as A · B′ and A′ · B, or a quaternary complex (A · B′ · A′ · B). In previous studies, such complexes existed where perfect base-pairing between substrates was observed in a gel-shift assay [18]. Furthermore, a stable T · T′ complex could have formed prior to the reaction due to the presence of T and T′ in the same reaction mixture, causing an initial decrease in the ligase available. Thus, these non-productive complexes that formed at the beginning of the reaction decreased the number of competent (A · B · T′ or A′ · B′ · T) reaction complexes in directing ligation reactions.

The same ligation reaction was repeated under alternating temperature conditions employing heat pulses (Fig. 4A). Ligation products were detected by the formation of ³²P-labeled T and T′ that resulted from the ligation of A and B, and A′ and B′, respectively. The formation of ligation products increased linearly over 6 h under alternating temperature conditions (Fig. 4B). In contrast, template-directed ligation performed at constant temperature showed no accumulation of the ligation products after 2 h of reaction (Fig. 4B). This suggests that the heat pulses provide a condition for efficient dissociation of the product (T · T′) and enable the newly formed RNA template ligases to commit to the next cycle of the reaction. Due to the increased binding affinity of the P3 stems in the ligase–product complex, product dissociation likely limits the rate of subsequent steps. Thus, alternating the temperature conditions provides a means to not only enhance the dissociation rate at the high temperature, but also allow efficient binding of the cognate substrates by the template ligase at the ambient temperature. Note that the benefits of using alternating temperature conditions also contributed to amplification of the hybrid template ligase (T _h) resulting from the ligation of A and B′, which involved partial mismatches within the P3 stem (Fig. 4B).

The template-directed ligation reaction with four substrates was more closely investigated under both isothermal and alternating temperature conditions to address whether the cross-catalytic product indeed leads to additional copies of the starting template. To distinguish the cross-catalytic product from the preferential ligation product resulting from the complement template, one template ligase (T′) instead of two was added to all four substrates in a common reaction mixture. The four substrates, each at 2 μM, were incubated at 23 °C in the presence or absence of template ligase T′. In the absence of template, ligated products were rarely detected (Fig. 3C). Addition of 1 μM T′ resulted in a significant increase in the formation of T, with a yield of 5% after 6 h (Fig. 5 , isothermal reaction). Under alternating temperature conditions (Fig. 5, heat pulsed reaction), formation of T significantly increased compared to the amount of T formed at the constant temperature. Most importantly, under the alternating temperature conditions, reactions employing unlabeled T′ mixed with all four substrates (³²P-labeled A and A′, and unlabeled B and B′) led to the formation of labeled T′, catalyzed by the newly generated ligation product (template ligase T) that was not initially present (Fig. 5A). This strongly suggests that the cross-catalytic amplification, as illustrated in Fig. 1, is substantiated in this ligase system under alternating temperature conditions. Accumulation of ligation product (A′ and B′ ligation in the presence of T′; Fig. 5B) that is the same as the template ligase itself (T′) exhibits sigmoidal kinetics, which is indicative of cross-catalytic amplification.

The cross-catalytic amplification activity, however, was not observed under isothermal reaction conditions. This suggests that the ligation product (T) was not dissociated from the template ligase (T′) due to high affinity in the P3 stem, leading to the failure of newly formed template ligase to commit to the next reaction cycle. Instead, under isothermal reaction conditions, the product of A′ and B ligation, which is equal in length to the hybrid template (74 nt), accumulated slightly in the presence of template ligase T′, as observed in the mismatched ligation reaction (shown in Fig. 3C). This mismatched ligation reaction occurs under isothermal conditions, but is not observed under alternating temperature conditions. Under the alternating temperature conditions, A′ and B ligation with template ligase T′ seemed to be less efficient than the cross-catalytic ligation reaction because the template ligase had to compete for common substrates with a much slower rate of A′ and B ligation due to a partial mismatch within the P3 stem at the 3′ end of the ribozyme.

The present study demonstrates cross-catalytic amplification of the starting molecule in a reaction system involving two ribozymes, which catalyzed each other's synthesis from a total of four RNA substrates under alternating temperature conditions. Given that the rate of the chemical step (i.e., the phosphodiester bond formation between two RNA substrates) is the same among the template ligase ensemble, the efficiency and specificity of the two component reactions are dictated mainly by two reaction steps: substrate binding and product dissociation. These steps are largely governed by the binding affinity in the P3 stem of the ligase ribozyme constructs. An increase in binding affinity in the P3 stem enhances the stability of ribozyme substrates, but also decreases the rate of product dissociation. Thus, the optimal binding affinity that balances these two opposing effects is required for survival of the fittest in the ligase ribozyme replication system. This problem can be circumvented by using the heat pulse method, which provides an artificial condition for monitoring the cross-catalytic replication reaction in a reasonable timescale.

Similar to the heat cycling concept used in this paper, a thermosynthesis system in which biological free energy is harnessed from thermal cycling was proposed to unify the hypothesis of the RNA world [21]. In the thermosynthesis system, thermal cycling convection endows the thermodynamic benefit of free energy to the primordial molecule, enabling it to perform several cycles of catalysis. In this regard, a temperature-fluctuating convection system might be hypothesized to exist in a primordial environment near hot springs and/or hydrothermal vents occurring within volcanic areas of a seabed [22, 23]. Previously, based on the rRNA sequences of bacteria, the last common ancestor of all living organisms has been proposed to have lived in a volcanic hot spring [24]. Thus, one may speculate that the RNA world thrives in those ecosystems, which appears to be the first event toward chemical evolution.