Harnessing Chemical Energy for the Activation and Joining of Prebiotic Building Blocks

Abstract

Life is an out of equilibrium system sustained by a continuous supply of energy. In extant biology, the generation of the primary energy currency, adenosine 5’-triphosphate (ATP), and its use in biomolecules synthesis require enzymes. Before their emergence, alternative energy sources, perhaps assisted by simple catalysts, must have mediated the activation of carboxylates and phosphates for condensation reactions. Here we show that the chemical energy inherent to isonitriles can be harnessed to activate nucleoside phosphates and carboxylic acids through catalysis by acid and 4,5-dicyanoimidazole under mild aqueous conditions. Simultaneous activation of carboxylates and phosphates provides multiple pathways for the generation of reactive intermediates, including mixed carboxylic acid-phosphoric acid anhydrides, for the synthesis of peptidyl-RNAs, peptides, RNA oligomers and primordial phospholipids. Our results indicate that unified prebiotic activation chemistry could have enabled the joining of building blocks in aqueous solution from a common pool and enabled the progression of a system towards higher complexity foreshadowing today’s encapsulated peptide-nucleic acid system.

Short RNA oligomers or peptides could have formed on early Earth in dry state^{1, 2} or solution phase reactions^{3, 4} following the activation of nucleotide phosphates or amino acid/peptide carboxylates, respectively. While these reactions have been performed separately, the simultaneous synthesis of RNA oligomers and peptides under dry state conditions starting from mixtures of appropriate building blocks (monomers and short oligomers) has not yet been reported. In aqueous solution, high-energy molecules, such as cyanamide^{5, 6} and carbonyl sulfide^{7, 8}, have been shown to drive condensation reactions of nucleotide phosphates and amino acids, respectively. However, although it can be released by outgassing due to geothermal activity, it is difficult to envisage how a reactive molecule like carbonyl sulfide could have accumulated on the surface of early Earth. Conversely, the kinetic stability of cyanamide that potentially allows its accumulation also makes it a sluggish activating agent. Although some form of catalysis might offer a solution to this problem^{9, 10}, no combination of prebiotically plausible high-energy molecule and catalyst has been found to simultaneously activate phosphates and carboxylates in solution phase. Whilst dry state and solution phase production of oligopeptides and oligonucleotides may have sufficed initially, solution phase RNA ligation/replication, peptide formation and peptidyl-RNA chemistry must have taken place to enable the eventual transition to biology according to the principle of continuity¹¹.

Methyl isonitrile has been detected in interstellar space¹², while Eschenmoser et al. reported that it can also be produced by thermolysis of iminodiacetonitrile, which derives in short measure from hydrogen cyanide^{13, 14}. We have previously described an early Earth geochemical model that leads to methyl isonitrile-iron(II) complexes from ferrocyanide ([Fe(CN)₆]^4-), nitric oxide and methylamine¹⁵. Methyl isonitrile could have been stored in the form of persistent Fe(II)-complexes, such as[Fe(CN)₅(CNCH₃)]^3-, and released by UV irradiation¹⁵. Isonitriles can activate phosphate groups upon reaction with aldehydes or imines^{15, 16}, however, the activation of carboxylates using similar chemistry leads, via intramolecular acyl transfer, to esters and amides, respectively^{17, 18}. Accordingly, we wondered if the chemical energy of methyl isonitrile could be harnessed, under possible aldehyde- and imine-free conditions, to allow the simultaneous activation of prebiotic phosphate and carboxylate building blocks.

Results

Simultaneous activation of adenosine 3'-monophosphate and carboxylates, catalysis by 4,5-dicyanoimidazole (DCI) and stereoselective acyl transfer

It has previously been shown that the hydrolysis of methyl isonitrile (1, pK _aH ≈ 1) to N-methylformamide 2 is accelerated by carboxylic acids^{19, 20}. We therefore wondered if acid catalysis could induce the addition of carboxylate and phosphate nucleophiles to 1. A model system was set up by mixing methyl isonitrile 1, adenosine 3'-monophosphate (A3'P, 3), and individual members selected from a library of acids, including carboxylic and amino acids, under mildly acidic aqueous conditions (Fig. 1A). Acetic acid (AcOH, pK _a = 4.8) promoted the formation of adenosine 2',3'-cyclic monophosphate (A>P, 4, yield of 4 was 56% vs 12 % for blank after 24 h, Table 1, Entry 1 and 2, Supplementary Fig. 1 and 2), indicative of phosphate activation^{5, 9, 10, 16}. Free amino acids, such as glycine (Gly, pK _a = 2.3) exerted no obvious rate enhancement (11% yield of 4, Table 1, Entry 3, Supplementary Fig. 3). The N-protected amino acid, N-acetyl-L-alanine (Ac-L-Ala, pK _a = 3.8), on the other hand, not only accelerated the formation of A>P 4 (in 18% yield), but also formed adenosine-2'-(N-acetyl-L-alanyl)-3'-monophosphate 5-L-a, albeit in a yield of only 5 %(Table 1, Entry 4, Supplementary Fig. 4 and 5). The formation of 5-L-a proceeds via a mixed carboxylic acid-phosphoric acid anhydride intermediate 6 ^{21, 22}, which can be observed by NMR spectroscopy by running the reaction at pH 4 (Table 1, Entry 5, Supplementary Fig. 6 and 7) and along with the concurrent formation of 4 is indicative of simultaneous activation of carboxylate and phosphate.

Fig. 1. Joining of prebiotic building blocks driven by common activation chemistry in aqueous solution.

a) Model reaction of A3'P and carboxylates upon activation. The mixed anhydride intermediate 6 can give either 4 or 5. b) Model reaction of A5'P and carboxylates upon activation. The mixed anhydride 10 is reminiscent of structurally analogous intermediates common to peptide and phospholipid biosynthesis. c) Formation of peptides upon activation. d) Simultaneous activation of carboxylates and phosphates under unified activation conditions. Carboxylates (aliphatic, peptidic and amino acids) are highlighted in blue, the catalyst, DCI 7, is shown in orange and the activating reagent, methyl isonitrile 1, is shown in green. Ade: Adenine.

Table 1. The yields of 5 from different carboxylic acids and A3'P 3, activated by methyl isonitrile 1 with different concentrations of DCI 7.

Entry	Carboxylic acid	DCI 7(mM)	pH	Time (h)	Yielda of 5(%)	Yielda of 4(%)
1	-	0	5.2	12	n. d.b	12
2	Acetic acid	0	5.2	12	n. d.	56
3	Glycine	0	5.2	12	n. d.	11
4	Ac-L-Ala	0	5.2	12	5-L-a: 5; 5-D-a: <1	18
5	Ac-L-Ala	0	4.0	12	5-L-a: 2; 5-D-a: <1	20
6	Ac-L-Ala	5	5.2	12	5-L-a: 10; 5-D-a: <1	20
7	Ac-L-Ala	25	5.2	12	5-L-a: 15; 5-D-a: <1	20
8	Ac-L-Ala	100	5.2	12	5-L-a: 29; 5-D-a: 2	23
9	Ac-L-Ala and Ac-D-Alac	100	5.2	12	5-L-a: 21; 5-D-a: 5	25
10	Ac-D-Ala	100	5.2	12	5-L-a: 4; 5-D-a: 8	27
11	Acetic acid	100	5.2	12	32	47
12	Ac-L-Ala	100	4.0	12	5-L-a: 20; 5-D-a: <1	17
13	Gly-Glyd	100	4.0	12	27	55
14	Gly-Glyd	100	5.2	12	6	68
15	Gly-Glyd	0	4.0	7	1	63
16	Gly-Gly-Glyd	100	4.0	12	34	43
17	L-Ala-L-Alad	100	4.0	8	51e	37
18	D-Ala-D-Alad	100	4.0	8	12e	61
19	Formyl-Gly-Glyf	100	4.0	9	31	31
20	Formyl-Gly-Glyf	100	5.2	9	28	47

Data reported are representative of distinct samples, n = 2 replicates.

^aConditions used for the formation of 5: A3'P 3 (10 mM), methyl isonitrile 1 (100 mM), carboxylic acid (100 mM), DCI 7 as indicated, at 23 °C unless specified. Yields were calculated based on ¹H-NMR spectroscopy;

^bProduct not detected;

^cAc-L-Ala (50 mM) and Ac-D-Ala (50 mM);

^dmethyl isonitrile (200 mM);

^eTotal yield;

^f N-Formyl-Gly-Gly (200 mM);

Encouraged by these findings, we investigated whether nucleophilic catalysis might also contribute to carboxylate activation and transfer on a wider range of substrates, including free and protected amino acids, peptides and aliphatic acids of different chain length. Based on the well-established nucleophilic catalysis of acyl transfer reactions by imidazole^{23, 24}, we tested several prebiotically plausible azoles, including imidazole²⁵ (pK _aH = 7), 5-aminoimidazole-4-carboxamide²⁶(AICA, pK _aH = 5.2), 2-aminoimidazole²⁷(2-AI, pK _aH = 8.5) and 4,5-dicyanoimidazole (DCI, 7, pK _a = 5.2). The latter azole is generated in a yield of 21 % alongside adenine as a product of cyanide polymerisation in formamide²⁸ (Fig. 2, and Supplementary Fig. 8). In the reaction with Ac-L-Ala, DCI 7 increased the yield of 5-L-a (6-fold after 12 h, Table 1, Entries 6 to 8) relative to the reaction in the absence of an azole (Supplementary Fig. 9 to 13). Remarkably, when a 1:1 mixture of Ac-L-Ala and Ac-D-Ala was reacted with 1, A3'P 3 and 7, acyl adducts 5-L -a and 5-D-a were observed in a 4:1 ratio after 12 h (Table 1, Entry 9, Supplementary Fig. 14). When Ac-L-Ala and Ac-D-Ala were separately mixed with 1, 3 and 7, both 5-L-a and 5-D-a were again produced, indicating racemisation during the activation chemistry (Table 1, Entries 6, 9 and 10; Supplementary Fig.15 and 16). No 2'-acyl adducts could be observed when free amino acids were used as substrates, possibly due to the reduced nucleophilicity of their carboxylates (pK _aH~ 2.2). In the reaction of A3'P 3 and acetic acid with 1 and 7, 2'-acetyl adduct 5b was observed in 32 % yield after 12 h, similar adducts been shown to be able to play a key role in abiotic RNA proofreading^{22, 29}, (Table 1, Entry 11, Supplementary Fig. 17). Furthermore, Ac-L-Ala, di- and tripeptides activated by the same chemistry gave good yields of 2'-acyl adducts 5 at pH 4 in the presence of 7 (Table 1, Entries 12 to 20, Supplementary Fig. 18 to 25).

Fig. 2. Proposed reaction scheme of the activation chemistry.

From top to bottom, four distinct chemical stages are highlighted in different colors. The first stage encompasses the prebiotic synthesis of building blocks, activating reagents, catalysts and other species. The second stage is activating reagent dependent, and the activating reagent is covalently attached to carboxylates or phosphates in the form of intermediates 11, 12 and 13. A dashed boxed includes two potential alternative pathways by which the activation chemistry mediated by methyl isonitrile 1 may be initiated. The third stage is activating reagent independent and is characterized by common intermediates such as 6, 8, 10 and 14. The formation of these intermediates could in principle be driven by alternative activating reagents. The fourth stage involves the condensation of various species to form oligopeptides, peptidyl-RNAs, RNAs and primordial phospholipids. Pathways involving catalysis by DCI 7 are highlighted in orange, and bold arrows indicate steps preferentially taken. Carboxyl moieties including aliphatic carboxylic acids, N-acyl amino acids and peptides, are highlighted in blue. All the activated carboxylates and phosphates are prone to hydrolysis but these reactions are omitted for the sake of clarity.

Simultaneous activation of adenosine 5'-monophosphate and carboxylates, formation of mixed anhydrides and peptides

The stereochemical results obtained with both enantiomers of Ac-Ala can be explained by the intermediacy of 5(4H)-oxazolone 8a (Fig. 2), which is known to undergo racemization via an aromatic tautomer 8a' ^30–32 and was further confirmed by an H/D exchange experiment (Supplementary Fig. 26). Interestingly, Tamura and Schimmel have previously reported a 4:1 L- over D-enantiomer preference in the transfer of (N-acyl) aminoacyl groups from the 5'-phosphate of a DNA strand to the 2',3'-diol of an abutting RNA strand in a templated system, using a mixed anhydride pre-synthesised by conventional procedures^{33, 34}(Supplementary Fig. 27). The observation of mixed anhydrides of 3'-phosphates 6 suggested that the chemistry might enable the synthesis of mixed anhydrides of 5'-phosphates, and this could provide a link between prebiotic activation chemistry and the Tamura and Schimmel chemistry and suggest an underlying reason for the relative stereochemistry of peptides and nucleic acids in biology. For simplicity, and to allow analysis by NMR spectroscopy, we employed adenosine 5'-monophosphate (A5'P, 9, Fig. 1B) as a proxy for an oligonucleotide-5'-phosphate. We thus investigated the reaction of carboxylic acids and methyl isonitrile 1 at pH 5.2 with A5'P 9. In the absence of DCI 7, 25 % of the 5'-mixed anhydride 10a was produced after 24 h when Ac-L-Ala was used (Table 2, Entry 1, Supplementary Fig. 28). Unexpectedly, the yield of 10a decreased as the concentration of DCI 7 was increased, although the reactions were also accelerated, based on the reduced time needed to reach maximum yield (Table 2, Entries 1 to 5, Supplementary Fig. 29 to 31, Supplementary Table 1). A pre-synthesised mixed anhydride 10 was incubated with different concentrations of DCI under similar conditions, which clearly showed that DCI catalysed the hydrolysis of mixed anhydride 10 (Supplementary Fig. 32). No stereoselectivity was observed in the reaction of A5'P 9 with Ac-L-Ala and Ac-D-Ala (Table 2, Entries 1, 5 and6, Supplementary Fig 33 to 35). The reaction of acetic acid with A5'P 9 in the absence of 7 gave only 2% of mixed anhydride 10b after 24 h (Table 2, Entry 7, Supplementary Fig 36, Supplementary Table 2). When unprotected di- or tripeptides were added to a solution containing A5'P 9 and methyl isonitrile 1, very low yields of mixed anhydrides were detected by³¹P-NMR spectroscopy (less than 3% yield, data not shown). We reasoned that aminolysis of activated peptides, or the intramolecular formation of diketopiperazines in the case of dipeptides³⁵, reduced the yields of mixed anhydride. In fact, oligopeptide formation was observed by¹ H-NMR and confirmed by mass spectrometry(up to 12-mers, Supplementary Fig. 37). An analogous reaction of A5'P 9 with N-formylglycylglycine, a model N-terminal protected peptide, was performed and a 19 % yield of mixed anhydride was observed after 36 h (Table 2, Entry 8, Supplementary Fig. 38), further supporting our hypothesis. To quantify the efficiency of peptide formation⁴, Ac-L-Ala (100 mM), glycine (50 mM), methyl isonitrile (200 mM) were mixed together at pH 4 or 5.2 and incubated at 23°C. The yield of the major peptide product Ac-Ala-Gly reaches 64 % after 24 hat pH 4 based on ¹H-NMR spectroscopy, and a yield of 59% was obtained after 10 days for the same reaction at pH 5.2(Supplementary Fig 39 to 48).

Table 2. The yields of 10 from different carboxylic acids and A5'P 9, activated by methyl isonitrile 1 with different concentrations of DCI 7.

Entry	Carboxylic acid	DCI 7 (mM)	pH	Time (h)	Yielda of 10 (%)
1	Ac-L-Ala	0	5.2	24	10-L-a: 18; 10-D-a: 7
2	Ac-L-Ala	5	5.2	24	10-L-a: 14; 10-D-a: 4
3	Ac-L-Ala	25	5.2	12	10-L-a: 9; 10-D-a: 2
4	Ac-L-Ala	100	5.2	6	10-L-a: 5; 10-D-a: 1
5	Ac-D-Ala	0	5.2	24	10-L-a: 7; 10-D-a: 17
6	Ac-L-Ala and Ac-D-Alab	0	5.2	24	10-L-a: 13; 10-D-a: 15
7	Acetic acid	0	5.2	24	2
8	Formyl-Gly-Gly	0	5.2	36	19

Data reported are representative of distinct samples, n = 2 replicates.

^aConditions used for the formation of 10: A5'P 9 (10 mM), methyl isonitrile 1 (100 mM), carboxylic acid (200 mM), at 23 °C unless specified. Yields were calculated based on ³¹P-NMR spectroscopy;

^bAc-L-Ala (100 mM) and Ac-D-Ala (100 mM).

Formation of peptides and peptidyl-RNAs in a unified system

We wondered how the reactions described above might interfere and interact when operated together in a more complex system. Reaction mixtures with A3'P 3 (10 mM), A5'P 9 (10 mM), Ac-L-Ala (100 mM), glycine (50 mM), DCI 7 (20 mM) and methyl isonitrile 1 (200 mM) at pH 4 or 5.2 (Fig. 1C, Supplementary Fig. 49 and 50) were monitored by ¹H- and ³¹P-NMR spectroscopy at 23°C. At pH 4, the yields of 2'-acyl adducts 5a and mixed anhydride 10a were 10 % and 8 %, respectively, after 12 h and the yield of peptide Ac-Ala-Gly was 60 % after 24 h, these yields being comparable to those obtained in separate reactions(Supplementary Fig. 51 to 56). Notably, however, two new species were identified in yields of 4 % and 2 %after 16 h by NMR spectroscopy, which were shown to be the 2'-Ac-Ala-Gly adduct of A3'P 5c and the Ac-Ala-Gly mixed anhydride of A5'P 10c, respectively (Supplementary Fig. 57 to 60). At pH 5.2, the yields of 5a and 10a were 19 % and 7 %, respectively, after 12 h, and the yield of Ac-Ala-Gly was 50 % after 10 days (Supplementary Fig. 61 to 62). These results suggest that common activation chemistry could not only have brought about the activation and joining of carboxylate and phosphate building blocks in solution, but it could also have driven the system towards higher order complexity.

Network of activation chemistry pathways driven by methyl isonitrile

Based on the results above, a reaction scheme for the activation chemistry was proposed (Fig. 2). In the first step, imidoyl carboxylates 11 and imidoyl phosphates 12 and 13 are formed following specific/general acid catalysed attack of carboxylates or phosphates on the isonitrile (Dashed box in Fig. 2). In later steps, mixed anhydrides 6 and 10 are generated, via the putative oxazolone 8a/8a' and N-acyl dicyanoimidazole intermediates 14, followed by thermodynamically and/or kinetically favored pathways to 4, 5, and peptides, respectively. Additional experiments were run to support our reaction scheme and the formation of the key intermediate 14, connecting 6, 8 and 10, by DCI catalysis (Fig. 2). The reaction of A3'P 3 (10 mM) with premade 2-methyl-4(4-methoxybenzyl)-5(4H)-oxazolone³¹ (50 mM) gave 5 and 6 more rapidly in the presence of DCI 7 (Pathway 8➔6➔5 and 8➔14➔6➔5 in Fig. 2, Supplementary Fig. 63) than in its absence. Imidazoles are known to catalyse the hydrolysis of mixed anhydrides via nucleophilic catalysis²³, and this is consistent with our observation that DCI 7 catalyses both the formation and hydrolysis of 10a, presumably via equilibrium with N-acyl dicyanoimidazolide 14 (Pathway 10➔14 and 14➔10 in Fig. 2). Interestingly, the role of DCI 7 in enhancing the formation of 2'-acyl adducts 5 can also result from the ability of DCI as a good nucleophile to take any activated carboxylate in the system to form N-acyl dicyanoimidazole 14, and 14 could also be in equilibrium with 3'-mixed anhydride 6. But the favourable intramolecular acyl transfer reaction of 6 drives the equilibrium towards the 2'-acyl adduct 5.

To support the proposed DCI-catalysed equilibrium between 10 and 6, proceeding via 14 (Pathway 10➔14➔6➔5 as shown in Fig. 2), the reaction of premade mixed anhydride 10e (~20 mM), A3'P (20 mM) and DCI 7 (0 mM, 5 mM or 20 mM) in MES buffer (250 mM, pH 5.2) was monitored by³¹ P-NMR spectroscopy (Supplementary Fig. 64). When 5 or 20mM of DCI was used the yields of A>P 4 after 12 h were 4 and 6%, respectively, while the yields of ester 5i were 5 and 10%, respectively. In the absence of DCI, only 4 was observed in 2% yield after 12 h (Supplementary Fig. 64). These results show that DCI catalyses the transfer of acyl groups from A5'P to A3'P in a concentration-dependent manner. This is best explained by nucleophilic catalysis proceeding via the acyl-dicyanoimidazolide intermediate 14 ^23,24. Reaction of A5'P 9 (10 mM), Ac-Gly (50 mM), DCI 7 (5 mM)and 1-ethyl-3-(3-dimethylaminopropyl)carbodiimide (EDC, 50 mM) in MES buffer (250 mM, pH 6) to give mixed anhydride 10e was monitored by ³¹P-NMR spectroscopy and revealed that both the formation (Pathway 8➔10 and 8➔14➔10 in Fig. 2, Supplementary Fig. 65) and the hydrolysis (Pathway 10➔14 in Fig. 2) of mixed anhydride 10e are catalysed by DCI 7. These results suggest that 6, 8, 10 and 14 are common intermediates/products and that the pathways leading to these species do not require a specific activating reagent.

Formation of primordial phospholipids

With this mechanism in mind, we reasoned that this chemistry could potentially be used to generate phospholipids. Thus, glycerol-2-phosphate³⁶ 15 was mixedwith methyl isonitrile 1 and DCI 7 in the presence of carboxylic acids of different chain length. Mono-acyl glycerol-2-phosphates and mono-acyl glycerol-2,3-cyclic phosphates were detected (19% and 4% yield, respectively, for decanoic acid) by ³¹P-NMR spectroscopy (Supplementary Fig. 66). Importantly, we and others have previously shown that the phospholipids derived from octanoic acid and decanoic acid are capable of self-assembling into vesicles^{37, 38}.

Template-directed ligation of oligonucleotides

Finally, we wondered if this activation chemistry could drive template-directed RNA ligation in solution. Oligonucleotides collectively constituting a nicked duplex were mixed with methyl isonitrile 1, DCI 7, divalent metal ions (Mn²⁺, Zn²⁺, Mg²⁺ or Fe²⁺), and N-methylimidazole at pH 6 (Supplementary Table 3) and the mixture was incubated at 23°C. The best yield of the expected 15nt ligation product was obtained with Mn²⁺– 45% after 5 days and 72% after 15 days. The yield of the ligation dropped to less than 1 % and 10 % after 5 days when the same reaction was performed at pH 4 and pH 5 respectively. A 16nt side product was proven to be the 5',5'-pyrophosphate of the 5'-phosphorylated ligator oligonucleotide (Supplementary Fig. 67).

Discussion

We have previously shown that RNA, peptide and lipid building blocks can be synthesised by a common cyanosulfidic chemistry from hydrogen cyanide^{36, 39}. Herein, we describe how a combination of methyl isonitrile 1 and 4,5-dicyanoimidazole 7, which both derive from cyanide, allow the activation of multiple prebiotic building blocks in solution (Fig. 2). Recently, diamidophosphate has been reported to phosphorylate and enable the condensation of amino acids and nucleotides, respectively³⁷. However, low water activity and paste-like conditions are required for the formation of the products and mixed species such as peptidyl-RNA adducts are not observed. In our system, simultaneous activation of peptides, RNA and amino acids gives higher order peptides and peptidyl-RNA in aqueous solution. Invoking pH cycling within the system⁴⁰, unified prebiotic activation chemistry could have potentially supported the synthesis of longer peptides, peptidyl-RNA and RNA in aqueous solution at room temperature. Primitive peptidyl-RNAs might have possessed catalytic properties analogous to modern ribonucleoproteins⁴¹, and therefore represent species worthy of consideration in either a coevolving peptide-RNA world or a more nuanced variant of the RNA world scenario. In addition, the observed preferential transfer of L- over D-N-acyl amino acids and peptides suggests that activation chemistry could have potentially played a role in the stereoselective RNA-templated synthesis of peptides on early Earth. If the absolute stereochemistry of nucleotides could have been fixed⁴², peptide synthesis based on RNA could have operated with some degree of stereocontrol. Furthermore, the activation of glycerol-2-phosphate in the presence of fatty acids gives primordial phospholipids, which could enable the transition from fatty acid to phospholipids vesicles⁴³. Taken together these results suggest that unified activation chemistry could have potentially supported the synthesis of the functional polymers and membrane-forming components required for life, while simultaneously underpinning their intimate association in a compartmentalised RNA-peptide system⁴⁴.

Methods

General Methods

Reagents and solvents were obtained from Acros Organics, Alfa Aesar, Santa Cruz Biotechnology, Sigma-Aldrich, SYNTHON Chemicals GmbH & Co. KG and VWR International, and were used without further purification unless otherwise stated. The 7nt RNA oligonucleotide (7nt primer, 5'-GAGAACC-3'), 8nt RNA oligonucleotide (8nt ligator, 5'-/phos/UACUGGCA/3Cy3Sp/-3') and 13nt RNA oligonucleotide (13nt template, 5'-/CCAGUAGGUUCUC/-3') were purchased in HPLC-purified Na⁺ form from Integrated DNA Technologies. 8nt RNA oligonucleotide (pUACUGGCA) was synthesised using an ÄKTA™ oligopilot™ plus 10 (GE Healthcare). All photochemical reactions were carried out in Norell Suprasil quartz NMR tubes purchased from Sigma-Aldrich using Hg lamps with principal emission at 360 nm in a Rayonet photochemical chamber reactor RPR-200, acquired from The Southern New England Ultraviolet Company. A MettlerToledo SevenEasy pH Meter S20 combined with a ThermoFisher Scientific Orion 8103BN Ross semi-micro pH electrode was used to measure and adjust the pH to the desired value. ¹H-, ³¹P-, and ¹³C-nuclear magnetic resonance (NMR) spectra were acquired using a Bruker Ultrashield 400 Plus or Bruker Ascend 400 operating at 400.13, 161.97, and 100.62 MHz, respectively. Samples consisting of H₂O/D₂O mixtures were analysed using HOD suppression to collect ¹H-NMR spectroscopy data. The notations s, d, p and m represent the multiplicities singlet, doublet, quintet and multiple, respectively. Chemical shifts (δ) are shown in ppm. Mass spectra were acquired on an Agilent 1200 LC-MS system equipped with an electrospray ionization (ESI) source and a 6130 quadrupole spectrometer (LC solvents: A, 0.2 % formic acid in H₂O – B, and 0.2 % formic acid in acetonitrile). Gel electrophoresis: 20 % polyacrylamide, 8 M urea gels (0.75 mm thick, 20 cm long) were typically run at 15 W in TBE buffer. Fluorescence imaging was performed using an Amersham Typhoon imager (GE Healthcare) and quantified using Image Quant TL software (version 7.0). Oligonucleotide concentrations were determined by UV absorbance at 260 nm using a NanoDrop® ND-1000 spectrophotometer.

Prebiotic synthesis of DCI 7

KCN (0.46 mmol, 30 mg, ¹³C-labelled or natural abundance) was mixed with ammonium formate (0.06 mmol, 4 mg) in formamide (1 mL) and the mixture was heated at 165 °C for 2 h²⁸. The solvent was evaporated under reduced pressure and the residue was extracted with hot water following evaporation, the product mixture was analysed by ¹H-NMR and ¹³C-NMR spectroscopy (10% D₂O in H₂O). The formation of DCI 7 was confirmed by spiking the crude mixture with authentic material (purchased from Sigma-Aldrich) and recording the ¹H-NMR spectrum, and also by comparing the ¹³C-NMR spectrum of the crude mixture with that of the authentic material.

Chemical synthesis of N-formyl-glycylglycine (formyl-Gly-Gly)⁴⁵

Glycyl-glycine (15.1 mmol, 2 g) was stirred in formic acid (35 mL), and heated at 55 °C while adding acetic anhydride (13.5 mL) dropwise. The solution was stirred at room temperature for 1 hour then ice-water (12 mL) was added, and the solution was concentrated under reduced pressure. The residual solution was left in a fridge at 4 °C overnight whereupon crystals formed at the bottom of the flask. The crystals were collected by filtration, washed with diethyl ether, and dried overnight using a desiccator to give a white solid (810 mg yield: 31 %). ¹H-NMR (400 Hz, D₂O) δ= 8.2 (s, 1H), δ 4.1 (s, 2H), δ 4.0 (s, 2H); ¹³C-NMR (101 Hz, D₂O) δ= 173.3, 171.6, 164.9, 41.1, 41.0.

Chemical synthesis of N-acetyl-alanylglycine (Ac-Ala-Gly)⁴⁶

Alanyl-glycine (1 mmol, 146 mg) was dissolved in a minimum amount of saturated NaHCO₃ solution. Twice molar equivalents of acetic anhydride were added. After 30 min at room temperature, Na⁺ was removed by H⁺-formed Dowex-50 column. The solution was lyophilized to give a white powder (170 mg yield: 83 %). ¹H-NMR (400 Hz, D₂O) δ= 8.22 (s, 1H), 8.04 (s, 1H), 4.27 (p, J = 7.1 Hz, 1H), 3.84 – 3.67 (m, 2H), 1.97 (s, 3H), 1.32 (d, J = 7.2 Hz, 3H); ¹³C-NMR (101 Hz, D₂O) δ=175.59, 175.09, 174.17, 49.74, 42.84, 21.83, 16.80.

Standard procedure for carboxylate and phosphate activation in reactions containing A3'P 3

An aqueous solution (0.5 mL, H₂O/D₂O, 9:1) of A3'P 3 (10 mM), carboxylic acid (100 mM, if required) and heterocyclic compound (100 mM, if required) was adjusted to the corresponding pH value with HCl (1 M) and NaOH (1 M) solutions. Methyl isonitrile (3 μL, final concentration to 100 mM) was then added and the reaction was analysed by ¹H- and ³¹P-NMR spectroscopy at 23 °C. The pH had changed only slightly (< 0.2) by the time the isonitrile was totally consumed.

Standard procedure for carboxylate and phosphate activation in reactions containing A5'P 9

An aqueous solution (0.5 mL, H₂O/D₂O, 9:1) of A5'P 9 (10 mM), carboxylic acids (200 mM) and heterocyclic compounds (100 mM, if required) was adjusted to the desired pH value with HCl (1 M) and NaOH (1 M) solutions. Methyl isonitrile (3 μL, final concentration to 100 mM) was then added and the reaction was analysed by ¹H- and ³¹P-NMR spectroscopy at 23 °C. The pH had changed only slightly (< 0.2) by the time the isonitrile was totally consumed.

Standard procedure for carboxylate and phosphate activation in reactions containing A3'P 3 and A5'P 9

An aqueous solution (0.5 mL, H₂O/D₂O, 9:1) of A3'P 3 (10 mM), A5'P 9 (10 mM), Ac-L-Ala (100mM), glycine (50 mM) and DCI (20 mM) was adjusted to pH 4 or 5.2 with HCl (1 M) and NaOH (1 M) solutions. Methyl isonitrile (6μL, final concentration to 200 mM) was then added and the reaction was analysed by ¹H- and ³¹P-NMR spectroscopy at 23 °C. The pH had changed only slightly (< 0.2) by the time the isonitrile was totally consumed.

Preparation of adenosine-2'-N-Acetyl-L(or D)-alanine-3'-monophosphate (L-standard or D-standard)

1,1’-Carbonyldiimidazole (8.9 mg, 55 μmol) was added to a suspension of Ac-L-Ala or Ac-D-Ala (6.5 mg, 50 μmol) in acetonitrile (50 μL) and the mixture was vortexed for 5 min at room temperature. The resulting solution, containing the acyl imidazolide derivative of N-acetyl-L(or D)-alanine, was then added to a solution of A3'P 3 (9.8 mg, 25 μmol) in H₂O/D₂O (9:1, 450 μL). The mixture was quickly analysed by ¹H- and ³¹P-NMR spectroscopy to confirm that the desired 2'-acylated product had formed. The mixture was then used immediately to spike other experiments.

Chemical synthesis of 10

EDC (76.4 mg, 400 μmol) was added to 2 mL of solution with A5'P (8 mg, 200 μmol), Ac-Gly(47mg, 400 μmol) and DCI 7 (2.4 mg, 20 μmol) at pH 6.0, room temperature. The pH of the mixture was kept around 6 by adding 5 M HCl solution. After 2 h, the reaction was titrated to pH 4 and quenched by adding 40 mL of cold NaClO₄ solution (50 mMNaClO₄ in acetone). The precipitate was collected by centrifugation, washed with cold acetone and dried under a stream of nitrogen. The yield of the product 10e (40 to 70%) was checked by ³¹P-NMR spectroscopy at pH 5 in H₂O/D₂O.

DCI catalysed hydrolysis of 10e

Synthesised 10e was dissolved in 1 mL water at pH 5. 150 μL of this solution, 200 μL of MES buffer (500 mM, pH 5.2), 50 μL D₂O and corresponding volume of DCI solution (500 mM, pH 5.2) were mixed together to give final concentrations of DCI 0 mM, 10 mM and 50 mM. Water was added to a final volume of 500 μL if needed. The pH was checked to be 5.2 otherwise adjusted with HCl (1 M) and NaOH (1 M) solutions. The reactions were then monitored by ³¹P-NMR spectroscopy at 23 °C.

Formation of oligopeptides from a dipeptide under activation chemistry

An aqueous solution (0.5 mL, H₂O/D₂O, 9:1) of A5'P 9 (10 mM), L-Ala-L-Ala (100 mM), and DCI (100 mM) was adjusted to pH 5.2 with HCl (1 M) and NaOH (1 M) solutions. Methyl isonitrile (3 μL, final concentration to 100 mM) was then added and the mixture was incubated at room temperature (white precipitate was formed during the incubation). After 3 days, the precipitate was removed by centrifugation. The supernatant was diluted 100-fold with H₂O and analysed by mass spectrometry (ESI, positive ion mode).

Formation of dipeptide Ac-Ala-Glyin reactions of Ac-Ala and glycine under activation chemistry

An aqueous solution (0.5 mL, H₂O/D₂O, 9:1) of A5'P 9 (0 or 10 mM), Ac-L-Ala (100 mM), glycine (50 mM) and DCI (0, 20 or 50 mM) was adjusted to pH 5.2 or 4 with HCl (1 M) and NaOH (1 M) solutions. Methyl isonitrile (6μL, final concentration to 200 mM) was then added and the mixture was monitored by NMR spectroscopy at 23 °C.

General procedure for desalting of RNA by ethanol precipitation

Oligonucleotides were desalted by addition 2 M imidazole nitrate solution (pH 6.2, 1/10 volume), followed by a 3 M sodium acetate solution (pH 5.2, 1/10 volume) and absolute ethanol (to a final concentration of 75 % (v/v)). The resulting mixture was kept at-20°C for 3 h and then centrifuged for 30 min at 16000 ×g. The supernatant was removed and the pellets were washed with 75 % (v/v) aqueous ethanol, before additional centrifugation (10 min at 16000 ×g). The resulting pellets were air dried before being re-dissolved in water.

General procedure for oligonucleotide ligation reactions

An aqueous solution of 7nt primer (20 μM, 0.5 μL of 400 μM stock solution), 8nt ligator RNA (10 μM, 0.5 μL of 200 μM stock solution), 13nt template RNA (10 μM, 0.5 μL of 200 μM stock solution), DCI(0 or 100 mM, 0 or 4 μL of 250 mM stock solution at pH 6), 1-methylimidazole (0 or 100 mM, 0 or 1 μL of 1 M stock solution at pH 6), divalent metal ion (Mn²⁺, Mg²⁺, Fe²⁺ or Zn²⁺, 0 or 10 mM, 0 or 1 μL of 100 mM stock solution at pH 6) was added nuclease-free water to 8 μL, then methyl isonitrile (400 mM, 2 μL of a 2 M aqueous stock solution) was added and the reaction was kept at room temperature. Aliquots of 3.0 μL were taken at the indicated time points and desalted by ethanol precipitation. For each aliquot, the resulting pellet was re-dissolved in 3.0 μL of nuclease-free water, and 1.0 μL of the resulting solution was mixed with 4.0 μL of loading dye (90 % (v/v) formamide, 5 % (v/v) glycerol, 24 mM EDTA, Orange G). The resulting mixture was analysed by gel electrophoresis.

Pyrophosphate formation under activation chemistry

An aqueous solution (0.5 mL, H₂O/D₂O, 9:1) of pUACUGGCA(1mM), DCI (100 mM), MgCl₂ (10mM) and N-methylimidazole (100mM) was adjusted to pH 6 with HCl (1 M) and NaOH (1 M) solutions. Methyl isonitrile (12μL, final concentration to 400mM) was then added and the mixture was monitored by NMR spectroscopy at 23 °C.

Standard procedure for reactions of glycerol-2-phosphate 15 or A3'P 3 with different aliphatic acids activated by methyl isonitrile 1

A solution (0.5 mL, formamide/D₂O, 9:1) of glycerol-2-phosphate 15 (10 mM) or A3'P 3 (10 mM), aliphatic acids (100 mM of decanoic acid, octanoic acid, butyric acid or acetic acid), and DCI (100 mM) was adjusted to pH 5.2. Methyl isonitrile (3 μL, final concentration to 100 mM) was then added and the reactions was analysed periodically by ¹H- and ³¹P- NMR spectroscopy at 23 °C.

Data availability

All data generated or analysed during this study are included in the manuscript and the Supplementary Information.