The Role of Sugar-backbone Heterogeneity and Chimeras in the Simultaneous Emergence of RNA and DNA

Abstract

Hypotheses of the origins of RNA and DNA are generally centered on the prebiotic synthesis of a pristine system (pre-RNA or RNA), which gives rise to its descendent. However, a lack of specificity in the synthesis of genetic polymers would likely result in chimeric sequences; the roles and fate of such sequences are unknown. Here we show that chimeras, exemplified by mixed TNA-RNA and RNA-DNA oligonucleotides, preferentially bind to, and act as templates for, homogeneous TNA-, RNA- and DNA-ligands. The chimeric-templates can act as a catalyst, mediating the ligation of oligomers to give homogenous-backbone sequences, and the regeneration of the chimeric templates potentiates a scenario for possible cross-catalytic cycle with amplification. This process provides a proof-of-principle demonstration of a heterogeneity-to-homogeneity scenario while giving credence to the idea that DNA could appear concurrently with RNA instead of being its later descendent.

Graphical Abstract

The RNA World hypothesis proposes the emergence of self-replicating and catalytic RNA giving rise later to proteins and DNA (Fig. 1b, middle).^1,2 Models posit the existence of a genetic polymer – whether RNA or its precursor – with a homogeneous backbone that transitions to its homogenous-backbone successor^1,3–10 This transition is proposed to occur despite the difficulties^2,11–14 associated with generation of the pristine oligomers using prebiotic chemistry,^15,16 and the challenge of replacing one genetic polymer with another^2,17–21 in the absence of any sophisticated discrimination mechanism during the transition in a pre-biological world.^13,22 However, there is a growing realization^23–25 that most prebiotic pathways^26,27 would lead to nucleic acid oligomers consisting of mixed backbone units.^14,17,19,28 In this context, RNA containing a mixture of 2′,5′-and 3′,5′-linkages,^29,18,19 and chimeric RNA-DNA systems,^17,21 have been investigated (and it has been shown that these types of backbone-heterogeneity compromise aptamer function^17–19), and we have shown that RNA-DNA chimeras consistently form weaker duplexes.¹⁴ Though chimeric RNA-DNA genomes are known in extant biology³⁰ and such chimeras containing nonheritable-backbone-heterogeneity have been postulated to be useful in the emergence of functional nucleic acids,^17,19 questions have been raised about their role as enhanced-templates for replication^17,31 generating polymers with homogeneous-backbones¹⁴. For pre-RNA to RNA transitions, Orgel has speculated two extreme possibilities using TNA (Fig. 1a)³² as an example: (a) an all TNA-organism converting to all RNA-organism, and (b) a gradual replacement of TNA residues by RNA residues within the oligomeric system.³³ The second scenario leads to a continuous-pathway from TNA to RNA, via chimeric sequences.³³ We have proposed a heterogeneity-to-homogeneity scenario³⁴ for the emergence of RNA and DNA^13,14, and argued that based on certain criteria such as the stability and functional advantages inherent to homogeneous-backbone-polymers, their emergence would be a natural consequence even when starting from a mixture of its constituent building blocks (Fig. 1b, top and bottom).¹³ A demonstration that chimeric TNA-RNA sequences (TRNA, Fig. 1b, top) or RNA-DNA sequences (RDNA, Fig. 1b, bottom) can enable the non-enzymatic emergence of homogeneous backbone oligonucleotide (RNA or DNA) starting from mixtures of chimeric sequences would provide support to the heterogeneity-to-homogeneity scenario.¹³

Figure 1. The prebiotic clutter generated heterogeneity-to-homogeneity scenario versus the biology inspired paradigm of replacing one homogeneous genetic system with its homogeneous genetic successor.

(a) Constitutional formula representation of the three oligonucleotide building blocks investigated in this study. (b) Three possible scenarios for the emergence of RNA and DNA from prebiotic chemistry. Middle: the classical RNA world concept where the formation of a pristine and homogeneous RNA (or pre-RNA) leads to its homogeneous-backbone successor DNA (or RNA). Top: a heterogeneous mixture of TNA (pre-RNA) and RNA forming chimeric TRNA sequences that transition to homogenous RNA, which then gives rise to DNA. Bottom: a heterogeneous RNA-DNA mixture progressing/co-evolving via chimeric RDNA sequences directly to homogeneous RNA and DNA simultaneously.

Results

TNA-RNA chimeric sequences function as templates for RNA ligands.

We selected TNA³² – a Watson-Crick base-pairing system able to cross-pair with RNA^32,35– as a model pre-RNA polymer,¹³ based on the prebiotic availability of the sugars^27,36–39 (Fig. 1a). We investigated TNA-RNA chimeric sequences (TRNA) which exhibited peculiar base-pairing properties even though TNA formed strong and stable duplexes with complementary RNA strands (Supplementary Table 1 & 2).³² First, in general TRNA formed weaker duplexes compared to the unmodified strands. Second, based on which sugar (threose or ribose) unit contained a purine (A) or pyrimidine (T), TRNA demonstrated unpredictable duplex stabilities (Fig. 2a). Unexpectedly, TRNA non-self-complementary strands which showed weak affinity for each other (Fig. 2a, entry 7), formed stronger duplexes with the corresponding complementary RNA (or TNA) sequences (Fig. 2a, entries 6 and 8), a behavior which was general for sequences containing all four nucleobases (Supplementary Table 3 and Figs. 7–13).

Figure 2. The preferential association with, and ligation of homogeneous ligands by, chimeric TRNA template over chimeric ligands.

(a) Thermal stability of TRNA chimeric duplexes in 1 M NaCl, 10 mM Na₂HPO₄, 100 μM EDTA, pH 7.2; a = [5μM], b = [2μM] duplex concentration; c = Entry 7, no clear sigmoidal transition in UV-thermal melt was observed. (b) Comparison of the rate of ligation reaction at 4°C of homogeneous-RNA (R_L1+R_L2) and heterogeneous TNA-RNA ligands (C_L1+C_L2) on heterogeneous TNA-RNA template, C_T1. (c) EDC-mediated ligation reaction at 4 °C of mixture of homogeneous-RNA (R_L1, R_L2) and chimeric TNA-RNA ligands (C_L1, C_L2) using TRNA chimeric sequence (C_T1) as template. (d) comparison of the amounts of products R_P1 and C_P1 produced in the reaction mixture in 2c; see supplementary Fig. 24, for conditions. A, T = RNA; T = DNA; a, t = TNA. Line in graph (2b) is drawn as guide to indicate the trend and is not a mathematical curve fitting. % yields are calculated with respect to the template C_T1. Experiments were run in triplicate and the error range is less than ± 5%; error bars represent standard deviation.

The preferential association of chimeric TRNA sequences with homogeneous RNA (or TNA) sequences (Fig. 2a, entries 6, 8) implied that chimeric-sequences could act selectively as templates for the non-enzymatic ligation of homogeneous-sugar-backbone ligands, and thereby facilitate the emergence of homogeneous-backbone-oligomer (e.g. RNA), starting from a mixture of oligonucleotides. To test this proof-of-concept, we employed the widely used water soluble 1-ethyl-3-(3-dimethylaminopropyl)carbodiimide (EDC) mediated ligation conditions⁴⁰ of homogeneous-RNA ligands templated by TRNA-chimeric- and RNA-templates, and compared it with ligation of the chimeric TRNA ligands (Fig. 2b). The 3′-NH₂ modified TNA-ligand⁴¹ and 3′-NH₂-deoxynucleotide (T^NH2) terminated RNA-ligand⁴² was used to conduct the ligation-reaction within reasonable time-frame, since the corresponding TNA-3′-OH and RNA-3′-OH residues react very slowly (Supplementary Figs. 14–17). The single phosphoramidate linkage was shown to have no special effect on duplex stability (Supplementary Fig. 12). The reactions were monitored by anion-exchange chromatography (AEC), and products were confirmed by comparison with standards and matrix assisted laser desorption/ionization time-of-flight (MALDI-TOF) mass spectrometry (Supplementary Figs. 18–28). As expected from a previous study⁴¹, the efficiency and the rate of ligation reactions paralleled the affinity (and thermal stability) of the templates for ligands, in the following order: RNA template with TNA ligands ≈ RNA template with RNA ligands ≥ TRNA chimeric template with RNA ligands >> RNA template with TRNA chimeric ligands >>>> TRNA chimeric template with TRNA chimeric ligands (Supplementary Figs. 18–23). Control reactions lacking the template(s) showed no product formation (Supplementary Figs. 25–28). We then examined the ligation behavior of the mixture of all four ligands in the presence of the chimeric TRNA template (Fig. 2c) and observed only the formation and growth of the RNA-product from homogeneous RNA-ligands, with no discernible chimeric-TRNA product from heterogeneous TRNA ligands by AEC (Fig. 2d; Supplementary Fig. 24). However, MALDI-TOF analysis of the reaction of chimeric TRNA ligands with chimeric TRNA template at 24 h did show traces of the chimeric TRNA-product (Supplementary Fig. 18). We have not investigated intensively a parallel scenario for the emergence of homogeneous TNA sequences⁴³ (due to the investment in synthesizing the various TNA 3′-NH₂-phosphoramidites), though we expect a similar propensity³² based on the observation that homogeneous-TNA ligands also were preferentially ligated by the chimeric TRNA template (Supplementary Fig. 20).

RNA-DNA chimeric templates ligate complementary RNA and DNA ligands.

The above results inspired us to investigate mixed DNA and RNA chimeric sequences based on (a) our previous studies of RNA-DNA (RDNA) chimeras¹⁴ and the plausible coexistence and coevolution of RNA and DNA in prebiotic scenarios^17,21,28,44, and (b) the ease of commercial/synthetic availability of diverse RDNA chimeric sequences. We studied a series of RDNA chimeric sequences, (Supplementary Table 4), which, again, formed stronger duplexes with complementary homogeneous RNA over corresponding complementary chimeric RDNA (Supplementary Table 5 and Supplementary Figs. 29–35). To test whether the preferential association of RDNA with RNA would also translate to selective ligation of RNA ligands (as seen in the TRNA system), we investigated the ligation behavior of a hexadecamer chimeric RDNA template (C_T2, Fig. 3a) with RNA and RDNA ligands containing 3′-NH₂ deoxynucleotide units. Ligation of RNA sequences (R_L3 and R_L4) on the chimeric RDNA template (C_T2) was not only faster than the corresponding ligation of chimeric RDNA ligands (C_L3 and C_L4 on C_T2; Fig. 3b), but was almost equal to the efficiency of RNA ligands R_L3 and R_L4 (or chimeric C_L3 and C_L4 ligands; Supplementary Fig. 39) on an RNA template, R_T2 (Supplementary Figs. 36–46).

Figure 3. Chimeric RDNA templates preferentially associate and ligate homogeneous RNA and DNA ligands over chimeric ligands.

(a) The list of homogeneous and chimeric sequences used in this study. (b) Comparison of ligation efficiency by hexadecameric (AU)-RDNA template C_T2, with RNA (R_L3, R_L4), DNA (D_L1, D_L2) and chimeric RDNA (C_L3, C_L4) ligands, showing the consistent preferential formation of homogeneous ligation products, R_P2 and D_P1 over chimeric ligation products C_P2. (c and d) Comparison of ligation efficiency by octameric (A, U/T, G, C)-RDNA template C_T4, with RNA (R_L7, R_L8), and chimeric RDNA (C_L5, C_L6) ligands, showing the influence of temperature on the preferential formation of homogeneous ligation products, R_P4 over chimeric ligation products C_P3. See supplementary figs. 50–52 for EDC-ligation reaction conditions. A, U, G, C = RNA; A, T, G, C = DNA. Lines in graph (3a) are drawn as guide indicating the trend and are not mathematical curve fittings. % yields were calculated with respect to the template C_T2 or R_T2 or C_T4 respectively. Experiments were run in triplicate and the error range is less than ± 5%; error bars represent standard deviation.

The duplex formation in octameric homogeneous and chimeric sequences containing all five canonical nucleosides again showed preferential association of homogeneous-backbone sequences with complementary chimeric templates (Supplementary Table 5). Based on this, we investigated the ligation reaction mediated by the chimeric template (C_T4) with RNA and chimeric ligands shown in Fig. 3c. The results revealed a temperature-dependent ligation behavior that was not observed in the hexadecameric-AU system (Supplementary Figs. 50–52). While at lower temperatures (4°C) there was little difference between the rate of ligation between the two systems, the rate of ligation of chimeric ligands and the amounts of products formed at higher temperature (10 and 16°C) differed considerably with preference for the ligation product from homogeneous-ligands on the chimeric template (Fig. 3d and Supplementary Fig. 52). This indicates that temperature could also control and modulate the overall dynamics and distribution of the end-products.

The trend of preferential association correlating with ligation capacity of C_T2 also extended to DNA ligands (D_L1, D_L2), in place of RNA ligands, giving rise to the homogeneous DNA product D_P1 (Fig. 3b), and was valid even when starting from a pool of mixed R_L3+R_L4+C_L3+C_L4 ligands or D_L1+D_L2+C_L3+C_L4 ligands (Fig. 3b, Supplementary Figs. 47–49). When all ligands (R_L3, R_L4, D_L1, D_L2, C_L3 and C_L4) were added to chimeric RDNA template C_T2 in a single pot, three major ligation products, R_P2 (38%), D_P1 (20%) and an RNA-DNA cross ligation product (RD_P1, 75%) were formed at 24h; no chimeric product from C_L3+C_L4 was detected (Supplementary Figs. 53–54). The nature of cross-ligation product was confirmed with appropriate control experiments and shown to be the result of D_L1-R_L4 ligation (Supplementary Figs. 55–59). Replacing the chimeric template with RNA template, under otherwise identical conditions, gave R_P2 (65%), D_P1 (12%) and 62% of RD_P1 and RD_P2 (R_L3-D_L2), indicating that RNA template also gives rise to significant cross-ligation products (Supplementary Figs. 60–61). Changing the ratios of RNA ligand (R_L3+R_L4) to DNA ligands (D_L1+D_L2) affected the product distribution (Supplementary Fig. 59) implying that generation of chimeric oligomers (along with homogeneous backbone oligomeric products) have to be reckoned with; these chimeric oligomer products should, in turn, help in the formation of homogeneous RNA and DNA ligation products. While this hypothesis is reinforced by the results in Fig. 3, it was demonstrated to be so by isolating RD_P1 and using it as a template with RNA ligands producing R_P3 efficiently in 108% yield (Supplementary Fig. 62). The above results show that from a mixed system with two different oligonucleotides (e.g. RDNA) there is indeed the possibility of the simultaneous emergence of the two respective homogeneous-nucleotide polymers (e.g. RNA and DNA).

RNA-DNA chimeric-templates are better in overcoming template-product inhibition.

The above observations suggest that chimeric-templates could provide a solution to the problem of product inhibition (Fig. 4a), where the continuous production of the product is curtailed due to the strong association of the initially formed template-product complex.^45–48 For instance, RNA-ligands R_L3 and R_L4 in the presence of the R_P2-R_P3 RNA duplex under the EDC-activation conditions showed no production of R_P2 even after 24 hours, indicative of classic product inhibition behavior; but the addition of chimeric template C_T2 led to the formation of more R_P2 within a matter of few hours (Supplementary Figs. 67–68). As outlined in Fig. 4b, if there was the adventitious presence/formation of a complementary RNA partner (R_P3, from its corresponding ligands R_L5 and R_L6) in the mixture containing the chimeric duplex (R_P2-C_T2), it would induce the formation of the stronger RNA (R_P2-R_P3) duplex. This should release the original chimeric RDNA template for another round of ligation of R_L3 and R_L4 forming more of R_P2, and result in a continuous accumulation of duplex R_P2-R_P3, with the chimeric template C_T2 taking the role of a catalyst producing more of R_P2 from its respective ligands. To test this scenario, we first conducted a step-wise addition of RNA ligands R_L3 and R_L4 to RDNA chimeric template C_T2, leading to the formation of the product R_P2 (97% in 20 h, Fig. 4c). Then, ligands R_L5 and R_L6 were added to this mixture. The formation of the second ligation product R_P3 (21% in 1 h increasing to 77% in 24 h, Fig. 4c), indicated that the in situ generated first ligation product R_P2 was indeed acting as a template (Supplementary Figs. 64–66). More encouragingly, with higher ligand ratios of R_L3 and R_L4, an increased amount of the first ligation product R_P2 (251% with respect to C_T2) and of the second ligation product R_P3 (204%) was observed after 24 hours (Fig. 4c, Supplementary Fig. 65 and Table). This indicates that the chimeric template C_T2 was indeed being released to take part in a turn-over, which in turn leads to formation of more R_P2. Pertinent control experiments confirmed the need for all components to be present for this system to operate; importantly, C_T2 itself did not serve as a template to ligate R_L5 and R_L6 and did not produce R_P3 (Supplementary Figs. 45, 66). Encouraged by these results, we set up a one-pot experiment where all components, C_T2+R_L3+R_L4+R_L5+R_L6, were mixed from the beginning and observed the concomitant production of the two RNA ligation products R_P2 and R_P3 (as efficiently as the step-wise addition experiment) (Fig. 4c). The presence of chimeric template C_T2 in a mixed-one-pot-system not only initiated the ligation process, but also acted as a turn-over intermediary down-stream, potentially enabling continuous production of R_P2 and R_P3 by mitigating the inhibition by template-product complex. This process was mainly driven by the preference of a thermodynamically stable homogeneous-backbone duplex R_P2:R_P3. Control reactions for R_L3+R_L4 or R_L5+R_L6 ligands without C_T2 template showed no observable background ligation reactions. However, when all four ligands (absent C_T2) were mixed together, 33% R_P2, 20% R_P3 and 13% cross-ligation products (probably from R_L3+R_L6 and/or R_L5+R_L4) were formed but more slowly at 24 h (Supplementary Fig. 70), as opposed to 259% of R_P2 and 191% of R_P3 with no cross-ligation products in the presence of chimeric template C_T2 (Supplementary Fig. 69). The background ligation-reactions were eliminated when ligand concentrations were lowered from 200 μM to 20 μM each; and only in the presence of 10 μM chimeric template C_T2, the formation R_P2 (83%) and R_P3 (18%) in 24 h was observed (Supplementary Figs. 73–76). Furthermore, we tested whether the presence of the complementary ligands (C_L3+C_L4, Fig. 3a) leading to the C_T2-C_P2 duplex would prevent further copying of the first two RNA ligands R_L4+R_L3 and also impact the next round of copying when all four RNA ligands R_L4+R_L3+R_L5+R_L6 are present. In both cases, in 24 h at 4°C, corresponding RNA products formed in good yields; 92% of R_P2 (with 30% of C_P2) for the first experiment and in the second scenario, 83% R_P2 and 16% R_P3, with no discernible peak for C_P2 in the chromatogram trace (Supplementary Figs. 47 and 77).

Figure 4. The beneficial role for chimeric RDNA template in overcoming the template-product inhibition based on thermodynamic stability of duplexes.

(a) The expected difference between chimeric RDNA-RNA duplex and the homogeneous RNA-RNA duplex in being able to overcome the template-product inhibition. (b) Schematic representation of the proposal that hexadecameric (AU)-RDNA template C_T2 with RNA ligands R_L3+R_L4 produces R_P2, which in the presence of R_L5, R_L6 is expected to lead to R_P3, based on the greater thermodynamic stability of the R_P2:R_P3 duplex over the R_P2:C_T2 duplex, and release the C_T2 for another round of ligation reaction. (c) Time course of the EDC-mediated-ligation experiments documenting the effect of change in ratio of ligands, and the sequential-addition of ligands R_L5+R_L6 (0 h) followed by R_L5+R_L6 (at 20 h) versus all-in-one-pot reaction on the production of R_P2 and R_P3. (d) Comparison of the amount of R_P3 formed by the homogeneous RNA template R_T2 versus chimeric RDNA template C_T2 (at 48 h) demonstrating the higher efficiency of C_T2 in mediating the formation of R_P3 by overcoming the template-product inhibition. See supplementary figs. 63–78 for EDC-ligation conditions. A, U = RNA; A, T = DNA. Lines in graph (4c) are drawn as guide indicating the trend and are not mathematical curve fittings. % yields were calculated with respect to the template C_T2 or R_T2 respectively. Experiments were run in triplicate and the error range is less than ± 5%; error bars represent standard deviation.

In order to assess the efficiency of chimeric template C_T2 versus that of the corresponding homogeneous backbone RNA counterpart R_T2, the all-in-one-pot reaction was repeated but with RNA template R_T2 in place of C_T2. In this case, as expected, the production of R_P2 at 48 h was comparable (99% for R_T2 versus 109% for C_T2); however, R_P3 formation dropped by almost half to 18% (for R_T2) when compared to 30% (for C_T2) indicating template-product inhibition by the stronger R_T2:R_P2 complex meant that R_P2 was less available for ligating R_L5+R_L6 (Fig. 4d). The advantage of C_T2 over R_T2 was more apparent when the ratio of ligands was changed to 5(R_L3+R_L4):2(R_L5+R_L6) with C_T2 producing 178% of R_P2 and 77% of R_P3 when compared to 119% of R_P2 and 43% of R_P3 with RNA template, R_T2 (Fig. 4d). This strongly suggests that C_T2 is better able to dissociate from the C_T2:R_P2 template-product complex while the RNA template R_T2 is limited by the classic R_T2:R_P2 template-product inhibition and is, therefore, unable to recycle to produce more R_P2 and R_P3. In fact, C_T2 consistently outperformed R_T2 in the production of R_P3 for all other combinations of ligand ratios (Fig 4d, Supplementary Fig. 78), indicative of the beneficial role played by chimeric templates in moving towards the emergence of homogeneous-backbone sequences. But for this to be possible, this phenomenon must hold good for other strands in terms of length and sequence diversity. Given the limitations imposed by the EDC-ligation-chemistries and analysis of the chimeric sequences involved, we set up a proof-of-principle experiment as in Fig. 4b but with octameric AUGC containing chimeric template C_T4 (Supplementary Fig. 79), since it also showed a preference for the complementary homogeneous ligands over the chimeric counterparts as seen in Fig. 3d. As expected, the chimeric template C_T4 was efficient in producing the homogeneous product R_P4 and R_P5 (Supplementary Fig. 80) overcoming the template-product inhibition even in the presence of all four ligands (R_L7+R_L8+R_L9+R_L10), paralleling the observations for the AU-based system. Thus, the ability of the chimeric template to give rise to homogeneous backbones (the heterogeneity-to-homogeneity paradigm) seems to be still operative in this RDNA-chimeric system even when shortening the length of the template and expanding the sequence diversity.

We then examined the effect of step-wise dilution (as a selection pressure) on the efficiency of the templates in overcoming the template-product inhibition, asking the question – which of the templates, chimeric-RDNA or the homogeneous-RNA would produce the ligation products more efficiently as the step-wise dilution was continued? Using the AU-system outlined in Fig. 4a we conducted a step-wise dilution experiment in parallel with templates C_T2 and R_T2 containing the complementary RNA-ligands (R_L3, R_L4, R_L5 and R_L6) where, every 24 h a portion of the reaction mixture was removed and fresh ligands and EDC were added, such that the concentrations of the ligands remained constant, but the template concentration decreased with each dilution-step (Supplementary Figs. 91–93). As seen from Fig. 5a, as the step-wise dilution was implemented at 24 h intervals, the formation of R_P2 and R_P3 was observed in both cases; while there was a concomitant drop in the product concentration (by 2 μM) at each dilution step, the amount of R_P2 and R_P3 increased to level greater than the previous value with progress of time. The amount of the first ligation product R_P2 was almost the same between the chimeric-(C_T2) and homogeneous-(R_T2) template containing vials over the first two-steps (48 h) of dilution, with C_T2 performing slightly better than R_T2 as the dilution steps were continued (72–96 h, Fig. 5b). However, there was a remarkable difference in the production of the second ligation product R_P3 with increasing step-wise dilutions; the chimeric template C_T2 outperformed the homogeneous template R_T2 in producing R_P3 by ca 250% (Fig. 5c), even as the concentration of the templates were going down with each step of dilution. A comparison of the chromatogram traces at 96 h (Fig. 5d) shows the dramatic difference and highlights the ability of C_T2 to be a superior template¹⁷ for the production of homogeneous product R_P3, demonstrating the ability of chimeric template C_T2 to better bypass the template-product inhibition and turn-over even under dilute conditions when compared to R_T2. Appropriate controls without the template showed no product formation (Fig. 5d).

Figure 5. Comparison of the efficiency between chimeric RDNA (C_T2) and RNA (R_T2) templates in producing the final ligation product R_P3 under step-wise dilution conditions, demonstrating the superior ability of C_T2 to act as a template for ligation with turn over.

(a) Production of the ligation products R_P2 and R_P3 in the stepwise-dilution (in 24 h intervals) experiment with templates C_T2 and R_T2, over a period of 96 h, containing all four ligands R_L3, R_L4, R_L5 and R_L6; the drops at 24, 48, and 72 h indicate the dilution step. (b) Time course contrast between the templates C_T2 and R_T2 for the production of the first ligation product R_P2 formed from R_L3 and R_L4. (c) Comparison of the efficiency of production of the second ligation product R_P3 (from R_L5 and R_L6) between the templates C_T2 and R_T2. (d) Chromatogram traces at 96 h after three stepwise-dilution juxtaposing the three parallel experiments in the presence of C_T2 (top trace), R_T2 (middle trace) and containing no template (bottom trace). See supplementary figs. 91–93 for EDC-ligation conditions (at 4 °C). For C_T2, R_T2, R_P2, R_P3, R_L3, R_L4, R_L5 and R_L6 see Fig. 4a. Lines in graphs (5a-c) are drawn as guide indicating the trend and are not mathematical curve fittings. % yields were calculated with respect to the template C_T2. Experiments were run in triplicate and the error range is less than ± 5%; error bars represent standard deviation.

RNA-DNA chimeric templates harbor the potential for cross-catalytic self-replication.

The promise of turnover of RNA ligation (Fig. 4b) when coupled with the observation that RDNA (C_P2) chimeric products can also be formed on the RNA template (Supplementary Fig. 39) suggested that the catalytic chimeric template (C_T2) could also be regenerated in the same reaction mixture if the corresponding chimeric ligands (C_L7+C_L8) are present (Fig. 6). If this would be possible, then the regeneration of the catalytic template C_T2 could allow for a cross-catalytic cycle to be operative, which would be expected to lead to the amplification of the homogeneous RNA product R_P2 (Fig. 6b). To test this possibility, we set up a one-pot EDC-ligation reaction with RNA ligands R_L3+R_L4 along with chimeric ligands C_L7+C_L8 in the presence of chimeric template C_T2 (Fig. 6). We observed within 1–4 h the formation of expected product R_P2 (90%), which now can act as the template for the chimeric ligands C_L7+C_L8. Indeed, by 24 h, formation of the phosphoramide-linked equivalent of C_T2 (C_T2^NH, 16%) was clearly observed, and kept increasing with time to 36% in 48 h and to 48% in 72 h. And, in parallel, the amount of R_P2 increased accordingly to 125% in 24 h, to 148% in 48 h and to 160% in 72h (Supplementary Fig. 94). This is well above the levels of R_P2 produced in the ligation reaction mediated by C_T2 in the presence of only R_L4+R_L3 and lacking the chimeric ligands (Fig. 6c), where the amount of R_P2 leveled at around 108% by 72 h. Thus, the chimeric template mediated ligation process shows potential for cross-catalytic self-replicating systems that can result in amplification of the down-stream product. Further systematic investigations are ongoing to understand the scope and limitation of this system. In all of the experiments described in this work no discernible degradation of the homogeneous or chimeric templates or products was observed (confirmed by comparing with an external standard of oligonucleotide dT₂₄ added to the samples just before analysis).

Figure 6. Experiment for testing the possibility of cross-catalytic amplification in oligonucleotide replication via regeneration of the chimeric RDNA (C_T2) template.

(a) The sequences of oligonucleotides used in this investigation; C_T2^NH is the same as chimeric template C_T2 but with a single phosphoramidate (NP) link at the ligation junction. (b) Schematic representation of the hypothesis that the presence of chimeric ligands C_L7 and C_L8 (complementary to R_P2) could induce the regeneration of the chimeric template C_T2^NH leading to further production of R_P2. The concomitant release of C_T2 also creates the potential for another round of ligation reaction. (c) Comparison of the amount of R_P2 produced from the combination of C_T2+R_L3+R_L4 (1:5:5) versus the combination of C_T2+R_L3+R_L4+C_L7+C_L8 (1:5:5:2:2), demonstrating the regeneration of chimeric template C_T2^NH along with higher and increasing production of R_P2 in the latter combination. See supplementary Fig. 94 for experimental conditions. % yields were calculated with respect to the template C_T2. Experiments were run in duplicate and the error range is less than ± 5%; error bars represent standard deviation.

Discussion

The results described in this work have confirmed experimentally the beneficial roles of chimeric sequences (backbone-heterogeneity) in nucleic acid replication, augmenting the evolution of functionality in mosaic nucleic acids¹⁷; and suggest that the nucleobase sequence-information encoded in heterogeneous-backbones can indeed be heritable for chemical evolution (similar to homogeneous-backbone systems). In these chimeric systems, there is the added advantage of (a) by-passing the template-product inhibition problem commonly encountered in the non-enzymatic replication of nucleic acids (unlike the homogeneous-backbone systems), and (b) moving towards (cross-catalytic) self-replication of the chimeric-templates, that eventually are able to assist in the transition from heterogeneity-to-homogeneity in nucleic acid systems^13,14. Whether the preference for homogeneous-backbone ligands by chimeric templates (dictated by the thermodynamic stability of duplex formation) could be a general phenomenon for oligonucleotides composed of other different sugar-backbones/nucleobases that are able to cross-pair needs further examples (such as chimeras of 2′,5′-RNA with 3′,5′-RNA)^19,49,50 to validate its scope and limitations.

For the work described here, however, there are some issues still to be addressed: firstly, the use of EDC-mediated-ligation combined with 3′-NH₂ modified deoxynucleotide in this proof-of-principle study is not considered to be plausibly prebiotic. To this end, we are exploring the use of other prebiotically plausible phosphorylation-activation combined with oligomerization and ligation/recombination chemistries that may be compatible with the replication conditions.^51–54 We briefly explored the use of enzymes (T4 DNA ligase and T4 RNA ligase 2) with canonical RNA, DNA and RDNA chimeric sequences, to check if ligases could be used to overcome the limitations of (a) the side reactions with chemical (EDC) activation⁵⁵ and (b) the need for synthesizing sequences with the 3′-deoxy-NH₂ modification – so that we may be able to push towards many rounds of replication and sequence analysis within a shorter time span, but have had limited success (Supplementary Figs. 95–103). We are exploring other ligases to expand the sequence-space and length parameters to overcome restrictions imposed by the EDC chemical-ligation methods.^{55, 56}

Secondly, longer homogeneous products formed in the scenario described above are unlikely to work as continuous templates and may not provide the solution when moving towards sustained replication of longer homogeneous strands relying on thermodynamic-driven effects alone. One possible solution (alluded to in this work) is that chimeric-templates can facilitate indirect replication by catalysing the accumulation of homogeneous strands. The product homogenous strands can act as information storage, but cannot be directly replicated. Therefore, other mechanisms need to be invoked to allow the transfer of information stored in the homogeneous strands.⁵⁶ One straightforward pathway consistent with the above heterogeneity-to-homogeneity scenario would be for the homogeneous RNA strands to give rise to functional ribozymes (ligase or polymerase) with the capability to take over the replication the homogeneous strands.⁵⁷ Other pathways could involve the beneficial effects provided by different classes of molecules that have not been considered in this study. For example, two other components, primordial (depsi)peptides⁵⁸ and protocells⁵⁹ should be invoked, since they would have been an important part of any prebiotic scenario; they are as elementary as, if not more than, the nucleotide building blocks.^53,60 Including them would be the next logical step to test the idea whether they could have not only aided in the transition from heterogeneity-to-homogeneity,³⁴ but could also play a role in enabling the replication of information stored in the longer homogeneous RNA and DNA strands by overcoming the slower kinetics of strand exchange in replication of homogeneous RNA and DNA strands as strand lengths increase.^61,62

Finally, in a prebiotic context, the possibility of oligomerizing on chimeric templates starting with monomeric building blocks has to be considered along-side the ligation chemistry demonstrated in this study.⁸ In our work, we were influenced by the duplex stabilities and reasoned that (a) the selectivity expressed at the ligand-template level may not translate to the level of weaker monomer-template associations and (b) based on earlier studies,^8,63 oligomerization of monomers would be biased towards G and C containing sequences (due to their stronger association) over A and U residues. Also, as argued by others,^64,65 the presence of dimers and trimers along with monomers in a prebiotic clutter may lead to selective incorporation of the higher order oligomers (dimers and trimers) over the monomers and, therefore, the ligation process may have an advantage over the oligomerization process. It would be necessary to test the limits of oligomerization with monomers in a chimeric scenario to observe what the preference is, both in terms of effect of the sugar and base residue (based on the nearest-neighboring nucleotide).^49,50

The results reported in this study have two-fold implications for the emergence of homogeneous-backbone nucleic acids. First, starting from a mixture of binary-chimeric-systems e.g. RDNA, (a possibility that is strengthened by the recent report⁴⁴ by Sutherland and co-workers on the plausibly prebiotic conversion of RNA nucleotides to DNA nucleos(t)ides), there is the potential for the simultaneous emergence of the two respective homogeneous-polymeric and communicating informational systems (RNA and DNA). This is opposed to the often-suggested sequential –RNA as the forerunner and DNA being the successor–paradigm. The successive replication cycles^42,66,67 are expected to lead, simultaneously, to the two respective strands containing the homogeneous sugar-backbone (RNA and DNA), as indicated by the results in Figs. 3–6. Therefore, if RNA and DNA could have appeared together, then there is no need for genetic takeover by the new informational system (DNA) from an older system (RNA), a suggestion that has been made implicitly and explicitly by others,^{14,17,23–25,28,68,69} since there is neither a predecessor nor a successor in this scenario. This is also true for the supposed pre-RNA to RNA transition³³; for example, there is no need for RNA being the descendent of TNA, when TRNA can simultaneously give rise to TNA and RNA. Second, the generality of this phenomenon–exemplified by RDNA and TRNA chimera systems–lends experimental credence to a point that is implied in Fig. 1, and one that has been discussed before;^13,21,28 namely, a clean and directed prebiotic synthesis of a nucleotide building block of a particular oligonucleotide (e.g. TNA or RNA or DNA) is not an absolute requisite for a homogeneous-backbone nucleic acid like RNA to emerge. In other words, as is suggested in Fig. 1, the appearance of system with homogeneous nucleotide-backbone repeat units can be achieved at the emergent level of a replicating polymer.³⁴ Therefore, a mixture of diverse nucleotides can, via the formation of mixtures of oligonucleotides and the ensuing emergent property of template-mediated ligation, tend towards homogeneous-nucleotide backbone systems.¹³ This process can include alternative linker units and alternative nucleobases,^10,17,19,70 and chirality of the building blocks⁷¹. Which means, the appearance of a homogeneous-backbone homochiral polymer with a set of uniform building blocks from a prebiotic mixture is a natural outcome of chemical evolution,¹⁴ without the need for invoking the predecessor-successor models of extant biology.^34,68,72