Prebiotically Plausible Synthesis of N⁶-Methylcarbamoyl Adenosine

Initially, we investigated the formation of the urea-modified adenosine 2 from either 1,3-dimethylurea (DMU) or methylisocyanate (MIC) under prebiotically plausible reaction conditions (Scheme 1b). Both DMU and MIC are prebiotically plausible starting materials. MIC could be generated from CH₄ and HNCO by vacuum-UV light irradiation at 20 K.⁴² Moreover, MIC was detected in the comet 67P/Churyumov-Gerasimenko⁴³ and in the protostar IRAS 16293-2422.^{42, 44} In turn, DMU was formed by continuous flow plasma discharge experiments from gas mixtures (N₂/CO/CO₂/H₂).⁴⁵ It is also possible that DMU was produced by the combination of MIC with methylamine, which was also present on the above-mentioned comet.⁴³

In a 0.4 M HCl aqueous solution at 0 °C, the reaction of DMU with 1 equiv of NaNO₂^46-48 gave its N-nitroso derivative in 98 % yield.⁴⁹ Subsequently, the addition of adenosine to a formamide^{43, 50} or an aqueous solution, containing 50 equiv of the N-nitroso derivative of DMU, followed by heating at 70 °C afforded the N⁶-carbamoylated adenosine 2 in up to 2 % yield. The high-performance liquid chromatography—mass spectrometry (HPLC-MS) analyses of the crude reaction mixtures showed the formation of 2, as well as other adenosine derivatives having either methyl or N-methylcarbamoyl substituents, or both (Figure S1a–b). Most likely, the isomeric products of 2 had the N-methylcarbamoyl substituent at one of the ribose hydroxy groups, that is, O-methylcarbamoyl adenosine derivatives. In formamide solution at room temperature (r.t.), the reaction of adenosine with 5.5 equiv of MIC yielded 2 in a maximum amount of 9 % (Figure S2). Taken together, these results showed that two prebiotically plausible synthetic pathways could provide the urea-modified adenosine 2.

Next, we probed the stability of the N⁶-methylcarbamoyl adenosine 2 at 70 °C in aqueous solution at different pH values within the range 6–9.5. Under all these conditions, the HPLC-MS analyses, after 24 h, indicated that 2 was stable in aqueous solution. Thus, heating at 70 °C the above obtained crude reaction mixtures for 24 h in 50 mM borate buffer pH 9.5 led to the hydrolysis of some of the O-methylcarbamoyl adenosine derivatives, whereas 2 persisted in the aqueous solution (Figure S3).

Prebiotic Loading of Amino Acids onto Nucleosides

Based on our previous study on the prebiotic synthesis of amino acid modified N⁶-carbamoyl adenosine 1 from N-methylcarbamoyl amino acids (Scheme 1a),⁴¹ we envisaged that the analogous nitrosation of the urea-modified adenosine 2, followed by its conversion into the corresponding N⁶-isocyanate could also lead to the formation of 1 (Scheme 1b). First, we optimized the conditions for the reaction of 1 with Gly using HPLC-MS analyses (Table 1 and Tables S1–S4). In the first reaction step, we performed the nitrosation of 2 with 12.5 equiv of NaNO₂ in an acidic aqueous solution. Among the acids tested, 5 % H₃PO₄⁵¹ in water gave the best results, indicating that the formation of the N-nitroso derivative of 2 required rather harsh acidic conditions (pH<1.5). Next, we investigated the reaction under freezing conditions, and we found that the yield strongly improved at −20 °C for 22 h. Most likely, under these conditions, the reactants were excluded from the growing ice crystals and concentrated in the interstitial (eutectic) phase.^52-55 In the second reaction step, the thawed aqueous solution, containing initially 2 and NaNO₂, was adjusted to near neutral or basic pH. At r.t. and in the presence of 10 equiv of Gly, the loading of the amino acid onto 2 was very efficient at pH 9.5,⁵⁶ affording 1 a (g⁶A) in a remarkable 84 % yield. When the amino acid was added in the first step, the nitrosation reaction, the amino acid modified N⁶-carbamoyl adenosine was also obtained, yet in lower yield (Figure S6 and Table S5). Note that the prebiotic formation of 1 a required dramatic changes in the pH of the aqueous solution. In this respect, it is worth mentioning that the second reaction step, coupling of the amino acid, also occurred under slightly acidic conditions in 5 % yield (Table 1). We also attempted the loading of Gly using the N⁶-methylated version of 2. However, under the optimized reaction conditions, the yield of the expected nucleoside product (m⁶g⁶A) was reduced to 8 % (Figure S9).

Table 1. Optimization of reaction conditions using N⁶-methylcarbamoyl adenosine (2) and Gly.

• [a] Yields determined by HPLC analysis using the calibration curve of 1 a (Figure S10). n.d.=not detected.

To evaluate the scope of the prebiotic synthesis of amino acid modified N⁶-carbamoyl adenosine nucleosides 1, we carried out the reaction with a series of L-amino acids, having aliphatic (Ala, Pro), aromatic (Phe), neutral polar (Thr) and ionizable (Asp, Lys) side chains (Table 2).⁵⁷ Under the reaction conditions optimized above for 1 a (g⁶A), we observed, in all cases, the formation of the nucleoside products 1 b–g in yields between 61–78 % (Figures S4–S5). We also detected, in all cases, the presence of ca. 10 % adenosine and the remaining starting material 2 (Figure 1a). Interestingly, the reaction of 2 with Lys, with its α- and ϵ-amino groups, provided a selectivity of 80 : 20 for the product α-1 f (k⁶A) versus ϵ-1 f.

Table 2. Amino acid scope using N⁶-methylcarbamoyl adenosine (2) and RNA oligonucleotide (ON2).

• [a] Yields determined by HPLC analysis using the calibration curves of reference compounds (Figures S10 and S19). [b] Reactions performed in the presence of GdmCl. [c] The structure of the Gly derivative contained a nitrile group instead of a carboxylic acid. A=adenosine. n.d.=not detected.

Next, we studied the loading of an amino nitrile onto N⁶-methylcarbamoyl adenosine 2. Amino nitriles were suggested as prebiotically plausible precursors for peptides.^{11, 58-61} Indeed, the reaction of 2 with the nitrile derivative of Gly, that is, Gly_CN, generated the nucleoside product 1 h (g_CN⁶A) in 26 % yield (Table 2).

Encouraged by the reaction of 2 with amino acids, we explored the loading of Gly with other urea-modified nucleosides: N²-methylcarbamoyl guanosine 3 and N⁴-methylcarbamoyl cytidine 4 (Table 3). We found that the addition of Gly to an aqueous solution of 3 at pH 9.5, after its nitrosation, afforded 5 (g²G) in 78 % yield (Figure 2a). Similarly, the combination of 4 with Gly gave 6 (g⁴C) in 58 % yield (Figure S8). These results suggested that the putative RNA-peptide synthesis might not be limited to RNAs containing the amino acid modified N⁶-carbamoyl adenosine 1, although the non-canonical nucleosides 5 (g²G) and 6 (g⁴C) were so far not found in contemporary RNAs.⁶²

Table 3. Base scope using different nucleosides and RNA oligonucleotides.

• [a] Yields determined by HPLC analysis using the calibration curves of reference compounds (Figures S10, S11 and S19). [b] Reactions performed in the presence of GdmCl. A=adenosine; C=cytidine, G=guanosine, U=uridine and X=xanthosine.

All together, these data showed that N-methylcarbamoyl nucleosides, 2–4, could be efficiently loaded with amino acids and amino nitriles in aqueous solution under prebiotically plausible conditions to form amino acid modified N-carbamoyl nucleosides.

Prebiotic Loading of Amino Acids onto RNA

We investigated whether the loading of amino acids was possible when N⁶-methylcarbamoyl adenosine 2 was incorporated into RNA. For these experiments, and the subsequent RNA-based amino acid transfer, we synthesized the phosphoramidite derivative of 2 in four synthetic steps with a 35 % overall yield (Scheme S19). Then, we used solid-phase RNA synthesis to incorporate the urea-modified nucleoside at the 5′-end of the 11-mer homo-A RNA strand ON2 (Table 2).

We started with the homo-A RNA strand ON2 and analyzed the loading of a series of amino acids. For example, when we treated ON2 with 500 equiv of NaNO₂ in 5 % H₃PO₄ at −20 °C and then, we added, at r.t., an excess of Ala, followed by basification to pH 9.5, the HPLC chromatogram of the crude reaction mixture, as well as the matrix-assisted laser desorption/ionization—time-of-flight (MALDI-TOF) mass spectrum of the purified compound revealed the formation of the RNA strand ON1g with a terminal a⁶A nucleotide (Figure S14). Next to the remaining starting material, we noted the formation of three major side products. Using MALDI-TOF MS, we assigned the degradation products to 10-mer and 11-mer RNA strands, A₁₀ and A₁₁, that had lost the terminal urea-modified nucleoside and the N⁶-methylcarbamoyl substituent, respectively. We also detected a 10-mer RNA strand with a terminal a⁶A nucleotide, A₉-a⁶A. It seems that the nitrosation reaction provoked the partial degradation of the RNA strands. However, the addition of 100 mM of guanidinium chloride (GdmCl) salt reduced the side reactions dramatically (Figure 1b), potentially by formation of salt-bridges between the (nitrosated)^{63, 64} guanidinium cation and the phosphodiester backbone.⁶⁵ Other salts, such as NaCl and NaClO₄, had little effect on the RNA degradation (Figure S14).

In order to further prove the loading of Ala onto the RNA strand ON2, we performed an enzymatic digestion of the isolated RNA product ON1g into the nucleosides, and analyzed the digest with the help of the corresponding nucleoside reference compounds by HPLC-MS.⁶⁶ We treated ON1g with a nucleoside digestion mixture at 37 °C for 2 h. Analysis of the obtained nucleoside mixture showed two peaks corresponding to 1 g (a⁶A) and adenosine in a ca. 1 : 10 ratio (Figure 1b), which was in agreement with the nucleotide composition of the digested RNA strand.

In the presence of GdmCl and for all the amino acids studied, we observed the formation of the RNA strands ON1a–g in 10–23 % yield (Table 2 and Figures S15–S16). It is worth mentioning that the reaction of Lys with the RNA strand ON2 yielded predominantly the α-regioisomer, α-ON1f, of the k⁶A nucleotide. For the loading with the amino nitrile Gly_CN, we did not detect the expected RNA product ON1h.

In analogy to the loading of amino acids onto different nucleosides, we evaluated the loading of Gly with three 11-mer RNA strands, ON2’–ON4 (Table 3), having all four canonical nucleotides, but bearing distinct N-methylcarbamoyl nucleotides at the 5′-end. The RNA strands were synthesized from the corresponding phosphoramidite derivatives of 2–4 (Schemes S19–S21). In all cases, the reactions of the three RNA strands with Gly, under the conditions described above with GdmCl, gave the respective products, ON1a’, ON5–ON6, in 5–35 % yield (Table 3, Figure 2b and Figure S12). MALDI-TOF MS analyses of the obtained RNA products showed the successful loading of Gly onto the parent RNA strands. In addition, we detected, after the enzymatic digestion of the RNA products, the partial conversion of guanosine to xanthosine by nitrosative deamination.^{67, 68}

Moreover, we investigated the loading of Phe onto an RNA strand ON2’’ bearing two N⁶-methylcarbamoyl adenosine nucleotides, placed at the 5′-end and an internal position of the sequence (Scheme S8). Using similar reaction conditions to those described above, we observed the formation of the double-loaded RNA strand ON1c’ as the major species in the crude reaction mixture (Figure S13).

Collectively, these data showed that RNA strands containing N-methylcarbamoyl nucleotides at either terminal or internal positions, or both, could be loaded with amino acids in a sufficient extent to perform RNA-based amino acid transfer (see below).

RNA-Based Amino Acid Transfer Following the Loading Reaction

Finally, we wanted to demonstrate that the prebiotic loading of an amino acid onto an RNA strand, functionalized with a urea-modified adenosine nucleotide at the 5′-end, could undergo RNA-based amino acid transfer (Figure 3a). The loading reaction provides a terminal amino acid modified N⁶-carbamoyl adenosine (donor) as the basis for the synthesis reaction. For the amino acid transfer, we hybridize the obtained donor RNA strand with a complementary acceptor RNA strand, containing a 5-aminomethyl uridine at the 3′-end, forming a duplex. Activation of the carboxylic acid in the donor RNA strand leads to the formation of a hairpin intermediate, which is opened at elevated temperatures by urea cleavage, transferring the loaded amino acid to the acceptor RNA strand. The obtained acceptor RNA strand can be involved in subsequent reactions, enabling peptide growth on RNA.¹¹ Since the nitrosation reaction converts guanosine to xanthosine, we employed a binary code of adenosine and uridine nucleotides in the initial RNA strand ON7. We also selected Phe for the experiment because of its high reaction efficiency (Table 2).¹¹

In this experiment, the individual reaction steps were monitored by HPLC (Figure 3b). In the first step, the prebiotic loading of Phe onto the RNA strand ON7, under the reaction conditions described above with GdmCl, gave the donor RNA strand ON8 in ca. 38 % yield. In the second step, we combined the isolated donor RNA strand ON8 with an equimolar amount of the acceptor RNA strand ON9 in 100 mM 2-(N-morpholino)ethanesulfonate (MES) buffer pH 6 and 1 M NaCl. The RNA strand ON9 contained 2′-methoxy nucleotides, which are found in contemporary ribosomal RNA (rRNA)⁶⁹ and minimize degradation. The addition of 4-(4,6-dimethoxy-1,3,5-triazin-2-yl)-4-methyl-morpholinium chloride (DMTMM⋅Cl) to the solution mixture of ON8 and ON9 at 15 °C led to the formation of the hairpin intermediate ON10 in ca. 40 % yield. Since the melting temperature of the reactive duplex ON8⋅ON9 was ca. 35 °C in the buffered aqueous solution (Figure S20a–b), it was assembled almost quantitatively in the reaction mixture. As reported previously, the activation of the carboxylic acid could also be performed prebiotically with methylisonitrile (MeNC) and 4,5-dicyanoimidazole (DCI).^{11, 70} In the third step, we warmed the aqueous solution of the hairpin intermediate ON10 to 90 °C at pH 6 to cleave the urea bond,^{71, 72} which furnished the RNA-peptide chimera ON11 in ca. 48 % yield. In contrast to our previous experiments with donor RNA strands containing amino acid modified N⁶-methyl-N⁶-carbamoyl adenosine at the 5′-end,¹¹ the absence of the N⁶-methyl substituent in ON8 prevented the formation of a hydantoin side-product in the urea cleavage step.⁷³

Finally, we performed the loading of Phe and the RNA-based amino acid transfer in a one-pot experiment (Figure 3c). The crude reaction mixtures were filtered after each step to remove the excess of salts and low-molecular weight compounds but retaining the RNA strands. Indeed, analysis of the one-pot experiment by HPLC showed the formation of the RNA-peptide chimera ON11 in ca. 11 % overall yield (Figure 3c), which was in line with the result obtained for the three individual reaction steps (ca. 7 %).