A prebiotically plausible scenario of an RNA–peptide world

1A central commonality of all cellular life is the translational process, in which ribosomal RNA (rRNA) catalyses peptide formation with the help of transfer RNAs (tRNA), which function as amino acid carrying adapter molecules,,. Comparative genomics suggests that ribosomal translation is one of the oldest evolutionary processes,,,, which dates back to the hypothetical RNA world–. The questions of how and when RNA learned to instruct peptide synthesis is one of the grand unsolved challenges in prebiotic evolutionary research–.

2The immense complexity of ribosomal translation demands a stepwise evolutionary process. From the perspective of the RNA world, at some point RNA must have gained the ability to instruct and catalyse the synthesis of, initially, just small peptides. This initiated the transition from a pure RNA world into an RNA–peptide world. In this RNA–peptide world, both molecular species could have co-evolved to gain increasing ‘translation’ and ‘replication’ efficiency.

3To gain insight into the initial processes that may have enabled the emergence of an RNA–peptide world, we analysed the chemical properties of non-canonical nucleosides,, which can be traced back to the last universal common ancestor and, as such, are considered to be ‘living molecular fossils’ of an early RNA world–.

4This approach, which can be called ‘palaeochemistry’, enabled us to learn about the chemical possibilities that existed in the RNA world and, therefore, sets the chemical framework for the emergence of life. In contrast to earlier investigations of the origin of translation–, we used naturally occurring non-canonical vestige nucleosides and conditions compatible with aqueous wet–dry cycles,.

5In modern tRNAs (Fig. 1a), the amino acids that give peptides are linked to the CCA 3′ terminus via a labile ester group. Some tRNAs, however, contain additional amino acids in the form of amino acid-modified nucleosides, for example, g6A (ref. ), t6A (ref. ) and m6t6A (ref. ), which are found directly next to the anticodon loop at position 37. Other non-canonical vestige nucleosides often present in the wobble position 34 are nm5U and mnm5U (refs. –).

6Close inspection of their chemical structures (Fig. 1b) suggests that if they are in close proximity (step 1), an RNA-based peptide synthesis may be able to start (step 2), which would create, via a hairpin-type intermediate, a peptide attached by a urea linkage to the nucleobase (m6)aa6A. Cleavage of the urea, (step 3) would furnish RNA with a peptide connected to a (m)nm5U (step 4). Subsequently, strand displacement with a new (m6)aa6A strand may finally enable the next peptide elongation step.

7To investigate the potential evolution of an RNA–peptide world, we synthesized two complementary sets of RNA strands, 1a–1j and 2a–2c (Fig. 2). The first set contained various m6aa6A nucleotides at the 5′ end (1a–1j) as RNA donor strands. The complementary RNA acceptor strands were prepared with an (m)nm5U nucleotide at the 3′ terminus (2a–2c). Figure 2a shows the reactions between 1a and 2a. The analytical data are presented in Fig. 2b. We hybridized 1a with 2a and activated the carboxylic acid of 1a using reagents such as EDC/Sulfo-NHS, DMTMM·Cl or methyl isonitrile (pH 6, 25 °C). In all cases we observed high yielding product formation (Fig. 2c).

8A kinetic analysis shows that the nature of the amino acid affects the coupling rate (Fig. 2d). For example, G (in 1a) couples to 2c with an apparent rate constant (kapp) of 0.1 h−1. For the amino acids L (in 1d), T (in 1e) and M (in 1h) a fourfold higher rate constant (≈0.4 h−1) was determined, and the highest rate was measured for F (in 1g) with kapp > 1 h−1. These differences establish a pronounced amino acid selectivity in the coupling reaction, probably as a result of distinct pre-organizations. We next reduced the length of the RNA donor strand to five, and finally to three, nucleotides (Supplementary Information). We detected coupling even with a trimer RNA donor strand, although it required duplex-enforcing high salt and low temperature conditions (1 M NaCl and 0 °C). The interaction of three nucleotides on the donor with the corresponding triplet on the acceptor seems to be the lower limit for productive coupling. Interestingly, this is the size of the codon–anticodon interaction in contemporary translation,.

9We next investigated coupling of the nitrile derivative of 1a (m6gCN6A, 1j) with the different acceptors 2a–2c under the recently described prebiotically plausible thiol activation conditions (DTT, pH 8, 25 °C). Here also, the coupling products were obtained within a few hours (Fig. 2c). For example, the combination of nm5U 2b with 1a gives coupling yields of 64% and 66% using EDC/Sulfo-NHS or DMTMM·Cl, respectively. Coupling of 1a and 2a, featuring a secondary amine, afforded 3a in 16% and 33% yields. The nitrile of 1j afforded yields of up to 65% after thiol activation coupling.

10We next measured the stability of the hairpin-type intermediates. For the hairpin 3a (Fig. 2a), a melting temperature (Tm) of approximately 87 °C was determined, which in comparison to the starting duplex (approximately 30 °C for 1a·2a, see Supplementary Information), proves that the peptide formation reaction generated thermally more stable structures. This could have been an advantage during wet–dry cycling under early Earth conditions.

11The discovered concept also enabled the synthesis of longer peptides. When we used 3′-vmnm5U-RNA-5′ 2c as the acceptor, we observed, on reaction with 1a–1j, peptide bond formation with up to 77% yield (Fig. 2c, d and Fig. 3a).

12We next studied the cleavage of the urea linkage and found that this reaction was possible at elevated temperatures (90 °C) in water at pH 6 (Fig. 2a, b). After 6 h, the products, m6A-containing RNA 4 and RNA 5a were formed already with a yield of 15%.

13We next investigated how the length of the generated peptides influences the coupling reaction (Fig. 3 and Extended Data Fig. 1). For this study we used synthetic 3′-peptide-mnm5U-RNA-5′ acceptor strands as starting materials (Supplementary Information). The synthesized acceptor strands were hybridized to the donor strand 1a. After carboxylic acid activation, rapid formation of elongated hairpin-type intermediates with yields between 40% and 60% was observed (Fig. 3b). We found that the coupling yields did not drop substantially with increasing peptide length, suggesting that other factors, such as the RNA hybridization kinetics, are rate limiting. In all cases, the subsequent urea cleavage (pH 4, 90 °C) affords dipeptide- to hexapeptide-decorated RNAs in 10–15% yield. These modest yields are the result of substantial RNA degradation, driven by the pH and temperature conditions that were used. The decomposition of RNA, however, can be overcome by using 2′-OMe nucleotides (see 'Stepwise growth of peptides on RNA'), which are also vestiges of the early RNA world.

14During urea cleavage we detected competing formation of hydantoin side products, depending on the pH and temperature (Fig. 3a). Under mildly acidic conditions (pH 6, 90 °C), exclusive formation of the hydantoin product, cyclic-5c, was observed. Reducing the temperature and a shift to higher acidity (pH 4, 60 °C) led to the preferential formation of the peptide product, 5c (approximately 7:1 5c:cyclic-5c ratio).

15We investigated whether longer peptides can also be generated by fragment coupling chemistry with RNA donor strands containing an already longer peptide (m6peptide6A). This is essential because an RNA–peptide world, with initially low chemical efficiency, might have been limited to the synthesis of smaller peptides. We found that the required adenosine nucleosides, containing a whole peptide attached to the N6-position, are available if the peptides that are produced by RNA degradation of the RNA–peptide chimeras, for example, can react with nitrosated N6-methylurea adenosine (Fig. 4a). When we treated N6-methylurea adenosine with NaNO2 (5% H3PO4) and added the solution to triglycine (pH 9.5), we obtained the peptide-coupled adenosine nucleoside ggg6A in approximately 65% yield. Incorporation of (m6)ggg6A into RNA and hybridization of this donor strand with a 3′-ggvmnm5U-RNA-5′ acceptor strand furnished, after coupling and urea cleavage, the RNA–peptide chimera 3′-gggggvmnm5U-RNA-5′ (53% coupling, approximately 10% cleavage; Fig. 4b, left). We could also directly transfer longer peptides. When we hybridized the 5′-m6gaggg6A-RNA-3′ donor with the 3′-agggvmnm5U-RNA-5′ acceptor, 3′-gagggagggvmnm5U-RNA-5′ was obtained as the product (56% coupling, approximately 9% cleavage; Fig. 4b, right). These experiments suggest the possibility of generating highly complex RNA–peptide chimeras with just a small number of reaction steps.

16We next investigated whether peptide growth is possible at different RNA positions simultaneously. To this end, we examined the simultaneous binding of different donor strands to one or two acceptor strands. We hybridized two donor strands (7-mer: 5′-m6g6A-RNA-3′ and 10-mer: 5′-m6v6A-RNA-3′) to a single RNA acceptor strand (21-mer) with a central gmnm5U and a 3′ terminal nm5U (Fig. 5a, left). On activation of the carboxylic acids, a GG-dipeptide was synthesized in the centre of the RNA, whereas a valine amino acid was attached to the 3′ end of the acceptor strand. In a different experiment, we hybridized an RNA donor strand (22-mer), containing both a 3′-m6g6A and a 5′-m6v6A, to two different acceptor RNAs, containing a central vmnm5U (21-mer) and a 3′ terminal vmnm5U (11-mer) (Fig. 5a, right). On activation, we observed formation of a central GV- and a terminal VV-dipeptide.

17To investigate the importance of sequence complementarity, we added two RNA donor strands of different lengths (7-mer: 5′-m6g6A-RNA-3′ and 11-mer: 5′-m6v6A-RNA-3′) to an acceptor strand with a vmnm5U at the 3′ end (11-mer: 2c) (Fig. 5b, left). On the basis of the melting temperatures of the two possible duplexes (approximately 30 °C for the 7-mer·11-mer and 59 °C for the 11-mer·11-mer, see Supplementary Information), only formation of the VV-dipeptide RNA conjugate, derived from the thermodynamically more stable duplex, was observed. Finally, we mixed two RNA donor strands of identical length (7-mer). The first contained a 5′-m6l6A and the second a 5′-m6g6A, together with two mismatches. We added this mixture to an RNA acceptor strand (11-mer: 2c) with a 3′-vmnm5U nucleotide (Fig. 5b, right). In this experiment, exclusive formation of the LV-dipeptide was found, generated from the fully complementary strands and thus the more stable duplex. Collectively, these results support that full complementarity is needed for efficient peptide synthesis.

18We finally investigated whether one-pot stepwise growth of a peptide on RNA is possible (Fig. 5c). To increase the stability of the RNA towards phosphodiester hydrolysis, as needed for this experiment, we used the RNA acceptor strand 2g, in which the contemporary canonical bases were replaced by the non-canonical 2′-OMe nucleotides: Am, Cm, Gm and Um. The strand 2g was equipped with an additional 3′-mnm5U nucleotide. For the experiment we used the same amount of donor strand for all coupling steps and performed filtration steps to remove remaining activator. After two couplings, two urea cleavages and two filtrations, we observed, by high-performance liquid chromatography (HPLC) analysis, the presence of the product 3′-ggmnm5U-RNA-5′ 7g (Fig. 5c, left). The circumvented material consuming isolation steps (Extended Data Fig. 2) enabled us to obtain the product in an overall yield of about 18%. A final, third coupling reaction with the 5′-m6f6A donor strand 1g furnished the FGG-hairpin intermediate 8g in approximately 10% overall yield (Fig. 5c, right).

19We next studied fragment condensation with the 5′-m6ggg6A-RNA-3′ donor strand and the complementary 3′-aggmnm5U-RNA-5′ acceptor strand, consisting only of 2′-OMe nucleotides. Here, coupling with approximately 50% and urea cleavage with approximately 85% generated the product 3′-gggaggmnm5U-RNA-5′, together with some of the hydantoin side product (Supplementary Information). Together these data show that, with the help of 2′-OMe nucleotides, peptides can grow on RNA in a stepwise fashion and via fragment condensation to generate higher complexity.

20The plausible formation of catalytically competent and self-replicating RNA structures without the aid of proteins is one of the major challenges for the model of the RNA world–. It is difficult to imagine how an RNA world with complex RNA molecules could have emerged without the help of proteins and it is hard to envision how such an RNA world transitions into the modern dualistic RNA and protein world, in which RNA predominantly encodes information whereas proteins are the key catalysts of life.

21We found that non-canonical vestige nucleosides–, which are key components of contemporary RNAs,, are able to equip RNA with the ability to self-decorate with peptides. This creates chimeric structures, in which both chemical entities can co-evolve in a covalently connected form, generating gradually more and more sophisticated and complex RNA–peptide structures. Although, in this study, we observe peptide coupling on RNA in good yields, the efficiency will certainly improve if we allow optimization of the structures and sequences of the RNA–peptides by chemical evolution. The simultaneous presence of the chemical functionalities of RNA and amino acids certainly increases the chance of generating catalytically competent structures. The stabilization of RNA by incorporation of 2′-OMe nucleotides significantly improved the urea cleavage yield.

22Interestingly, in the coupling step we observed large differences in the rate constants, which suggests that our system has the potential to preferentially generate certain peptides. We also found that peptides can simultaneously grow at multiple sites on RNA on the basis of rules determined by sequence complementarity, which is the indispensable requirement for efficient peptide growth.

23All these data together support the idea that non-canonical vestige nucleosides in RNA have the potential to create peptide self-decorating RNAs and hence an RNA–peptide world. The formed RNA–peptide chimeras are comparatively stable, and so it is conceivable that some of these structures learned, at some point, to activate amino acids by adenylation and to transfer them onto the ribose OH groups to capture the reactivity in structures that were large and hydrophobic enough to exclude water. This would then have been the transition from the non-canonical nucleoside-based RNA–peptide world to the ribosome-centred translational process that is a hallmark of all life on Earth today.

24The RNA donor and acceptor strands (1:1 ratio, 5 nmol of each strand) were annealed with NaCl (5 μl from a 1 M aqueous solution) by heating at 95 °C for 4 min, followed by cooling down slowly to room temperature. After that, MES buffer pH 6 (25 μl from a 400 mM aqueous solution) and NaCl (5 μl from a 1 M aqueous solution) were added to the oligonucleotide solution. Finally, carboxylic acid or nitrile activator/s (10 μl of each component from a 500 mM aqueous solution) and water (100 μl of total reaction volume) were added to the solution mixture. The peptide coupling reaction was incubated at 25 °C for 24 h. The crude reaction mixtures were analysed by HPLC and MALDI-TOF mass spectrometry.

25The hairpin-type intermediate (0.5 nmol) was diluted with MES buffer pH 6 or acetate buffer pH 4 (12.5 μl from a 400 mM aqueous solution), NaCl (5 μl from a 1 M aqueous solution) and water (50 μl of total reacion volume). The urea cleavage reaction was incubated at 60–90 °C at different time intervals. The crude reaction mixtures were analysed by HPLC and MALDI-TOF mass spectrometry.

26Any methods, additional references, Nature Research reporting summaries, source data, extended data, supplementary information, acknowledgements, peer review information; details of author contributions and competing interests; and statements of data and code availability are available at 10.1038/s41586-022-04676-3.