Introduction

Peptides and nucleic acids have universally conserved, interdependent roles in biology that suggest a fundamental link between peptides and nucleotides that may date back to the origins of life. The efficiency and simplicity of Strecker amino acid synthesis from aldehydes (1, R2 = H), hydrogen cyanide (HCN) and ammonia1 suggest that the Strecker reaction is a likely prebiotic amino acid synthesis2. The implication of HCN in both amino acid and nucleotide synthesis strongly suggests the importance of cyanide chemistry at the origins of life2. However, Strecker synthesis is highly pH dependant and is reversible under sufficiently acidic or alkaline conditions3. Moreover, if a successful synthesis of aminonitrile 2 (or amino acid 3) can be achieved under prebiotic constraints, subsequent activation is required for peptide synthesis. The (prebiotic) electrophilic activation and oligomerisation of amino acids have been achieved with mixed success4,5,6,7, but in extant biology, peptide synthesis begins with phosphorylation of amino acid 3. Intrigued by the biological relationship between amino acids and nucleotides that is manifest in the genetic code, and the prebiotic reactions of HCN (that can furnish amino acids 3, as well as nucleobases and simple sugars)2, we hypothesised that an imine-based reaction might link sugar/nucleotide phosphorylation with amino acid phosphorylation to yield a robust, neutral pH Strecker reaction in water. Specifically, we reasoned that aminonitriles might be accessed at neutral pH by using a low pKa amine (Fig. 1, EWG-NH2; EWG = electron-withdrawing group) that is deprotonated, and therefore nucleophilic, at low pH. Moreover, judicious choice of EWG-NH2 would provide a traceless masked ammonia source that is activated to undergo Strecker reaction at low pH, and subsequent unmasking of the amine moiety would liberate free 2 or an activated derivative of 2 or 3.

Fig. 1
figure 1

Masked Strecker reaction. a Strecker reactions proceed with ammonia (NH3) at high pH to furnish aminonitriles (3) (blue arrow). b Ammonia is protonated (NH4+) at neutral pH, which blocks low pH Strecker reaction with ammonia (magenta arrow). If ammonia is masked (green arrows) to yield an amine with a low pKaH, the Strecker reaction can proceed at low pH before the latent primary amine moiety is unmasked to liberate the free NH2-moiety. c Prebiotic synthesis of diamidophosphate (DAP) from cyclotrimetaphosphate (cTMP)

Full size image

Recently, we reported an α-phosphorylation controlled reaction network that gives access to all the intermediates of triose glycolysis (glyceric acid 2-phosphate, glyceric acid 3-phosphate, PEP (phosphoenol pyruvate) and pyruvate, as well as phosphoserine)8, which demonstrated that switching the order of oxidation and elimination renders triose glycolysis chemically predisposed. The key to site-selective α-phosphorylation was imine-tethered diamidophosphate (DAP)8,9,10, which renders intramolecular phosphorylation of the α-hydroxyl moieties of simple sugars highly efficient and remarkably selective, especially at low pH8.

The importance of DAP-imine chemistry in controlling prebiotic glycolysis prompted us to explore the reactivity of DAP further, and consider DAP as masked ammonia (Fig. 1) in the Strecker reaction. The most-plausible reported prebiotic DAP synthesis is achieved by cyclotrimetaphosphate (cTMP) ammonolysis9,10,11. Therefore, it is of note that Rabinowitz reported cTMP-mediated ligation of amino acids (pH 11 at room temperature)12,13,14,15. The proposed mechanism for cTMP activation involves initial nucleophilic attack of the amine moiety of amino acid 3 at the electrophilic phosphorus atom of cTMP, followed by formation of a five-membered phosphoramide intermediate15. This cyclic intermediate is unstable to isolation, but has been observed in situ for glycine and alanine16,17. The strongly alkaline conditions required for cTMP-mediated phosphorylation lead to very poor yields and various undesirable side reactions9,18,19,20,21, but we reasoned these problems might be ameliorated by using imine tethering and nucleophilic amine catalysis provide by the ammonolysis of cTMP to form amidotriphosphate and DAP8,9,10,22.

Here, we show that the reaction of aldehydes (1, R2 = H), HCN and DAP yields N-phosphorylated aminonitriles (4) at neutral pH. Compounds 4 are stable under the reaction conditions and, unlike 2, retro–Strecker reaction is not observed under acidic or alkaline conditions. Hydrolysis of N-phosphorylated aminonitriles at acidic pH affords the corresponding aminonitriles 2. Conversely, alkaline hydrolysis selectively yields N-phosphoro amino acids (5). Finally, we demonstrate that selective thiolysis of the nitrile group allows the synthesis of thioamides 6 at neutral pH. Our synthesis opens a lower pH window for Strecker chemistry, and provides a selective mechanism for amino acid N-phosphorylation23. We have also demonstrated pH-dependent hydrolysis and selective (prebiotically plausible) thiolysis of N-phosphorylated aminonitriles 4, which may open new avenues for selective peptide ligation controlled by phosphorylation.

Results

Phosphoro–Strecker reaction optimisation

We first studied the reaction of aldehydes 1a–h (50–200 mM), HCN (1.2 equiv.) and DAP (1.2–4 equiv.) in water across a broad pH range (pH 5–10) at room temperature (Fig. 2, Table 1 and Supplementary Table 1). This wide range of (prebiotically plausible) aldehydes underwent phosphoro–Strecker reaction in good yields (73%–Quant.) from their respective phosphoro–aminonitriles 4a–h (Table 1, entries 1–8; Supplementary Figs. 1–16). The optimal pH for phosphoro–Strecker reaction was pH 7 (Supplementary Fig. 17), which starkly contrasts with typical Strecker reactions that require alkaline (pH 9–10) conditions close to the pKaH of ammonia. In contrast to the phosphoro–Strecker reaction, aldehydes (1, R2 = H) react poorly with HCN (1.2 equiv.) and ammonia (4 equiv.) at neutral pH to afford low yields (6–10%) of aminonitriles (2) after 4 days (Supplementary Figs. 18–20), with a large amount of aldehyde (1) remaining trapped as the corresponding cyanohydrin (7). Consequently, the pH dependence of the Strecker reaction allows the pH-controlled preparation of 2 or 4 in a remarkably selective way. The reaction of acetaldehyde 1a, HCN (1.2 equiv.), DAP (4 equiv.) and ammonia (4 equiv.) at pH 7 for 4 days yields 4a in excellent yield (94%), alongside a low yield of aminonitrile 2a (6%). However, under alkaline conditions (pH 10) the observed yields for 4a (3%) and 2a (90%) are reversed (Supplementary Fig. 18).

Fig. 2
figure 2

Phosphoro-Strecker reaction. N-Phosphoro-aminonitriles (4) are synthesised from aldehydes (1) (R1 = alkyl or H, R2 = H), HCN and diamidophosphate (DAP) in excellent yields at pH 7. Ketones (1) (R1 ≠ H, R2 ≠ H) give α,α-disubstituted N-phosphoro-aminonitriles (4) (R2 ≠ H) in low yields alongside their respective cyanohydrins (7). N-Phosphoro-aminonitriles (4) afford their respective aminonitriles (2) upon acid hydrolysis, N-phosphoro amino acids (5) upon alkaline hydrolysis, and thioamides (6) upon thiolysis at neutral pH. See Table 1 for reaction conditions and yields

Full size image
Table 1 Phosphoro–Strecker reaction products and yields
Full size table

Aldehydes 1 (R2 = H) bearing polar and charged functional groups gave good-to-excellent yields of 4 in water (Table 1, entries 10–11, Supplementary Figs. 21–24). For example, 4-oxobutanoic acid (1j) furnished N-phosphoro-glutamic acid aminonitrile (4j) in 88% yield at neutral pH (Table 1, entry 10; Supplementary Figs. 23 and 24). Low aldehyde solubility can lead to poor Strecker reaction yields, but the yield of phosphoro–Strecker reactions for poorly water-soluble aldehydes (e.g. 1g) was enhanced by the addition of co-solvents. For example, addition of 10% DMSO or formamide improved the solubility of the aldehyde substrates and their respective cyanohydrins. However, the reaction with formaldehyde (1l), the precursor of glycine, required slow addition of 1l (Fig. 3). The general reaction conditions initially yielded cyanohydrin 7l in excellent yield (Supplementary Figs. 25 and 26), but after 6 day significant amounts of material were sequestered into various oligomers, rather than the desired product 4l. Excess 1l yielded aminal 8 (Fig. 3, Method 1 in the Supplementary Methods) as a major product. Aminal 8 has a highly characteristic 1H NMR phosphorus-coupled diastereotopic spin system [δH 5.29 ppm (dt, J = 14.1, 7.0 Hz); 4.33 ppm (dt, J = 21.7, 14.1 Hz)] (Supplementary Figs. 27 and 28) and was readily precipitated as a calcium salt. Recrystallisation and X-ray analysis (Supplementary Fig. 29, Supplementary Table 2 and Method 2 in the Supplementary Methods) unambiguously proved that 8 was a highly symmetric cage structure furnished by oligomerisation of DAP/1l (3:6). Slow addition of stoichiometric 1l to a solution of HCN and DAP at pH 7 (Method 3 in the Supplementary Methods) yielded the desired product 4l in a satisfactory 48% yield (Table 1, entry 12; Supplementary Figs. 30–32).

Fig. 3
figure 3

Formaldehyde-DAP phosphoro–Strecker reaction. a Aminal (8) is formed upon oligomerisation of formaldehyde (1l) and DAP at pH 7 and room temperature. b Slow addition of formaldehyde (1l) to a solution of DAP and HCN at pH 7 and room temperature yields phosphoro-aminonitrile 4l (48%). c Crystal structure of (8); thermal ellipsoids are shown at the 50% probability level. All hydrogen atoms, water molecules, calcium cations and atom labels are omitted for clarity (crystal packing is shown in Supplementary Fig. 28). Colour scheme: carbon = grey, nitrogen = blue, oxygen = red, phosphorus = orange

Full size image

Owing to their (retrosynthetic) relationship with proteinogenic amino acids, aldehydes (1, R2 = H) are a central focus of prebiotic chemistry. However, ketones (1, R2 ≠ H) have been shown to emerge alongside aldehydes from prebiotic networks24. Their corresponding aminonitriles (2) have been reported to form in yields similar to those formed from aldehydes at basic pH, at which conventional Strecker reactions operate25. However, cyanohydrin (7) formation from ketones and HCN is disfavoured with respect to aldehydes at neutral pH24,26. Therefore, we next studied the phosphoro–Strecker reaction of a range of ketones (1m–r, Table 1, entries 13–18). Simple ketones (1m–p) gave significantly lower yields (<19%) of α,α-disubstituted N-phosphoro-aminonitriles 2m–p (Table 1, entries 13–16; Supplementary Figs. 33–40) than the aldehyde substrates. The phosphoro–Strecker reaction of ketones was not improved by the addition of co-solvents (e.g. 10% DMSO), even when near-quantitative ketone solubility, as a mixture of 4, ketone (1m–p) and cyanohydrin (7), was observed by NMR spectroscopy. To further probe the importance of carbonyl reactivity in controlling phosphoro–Strecker reaction yields, we investigated ring-strained ketones 1q and 1r. Ketones 1q and 1r furnished improved yields of N-phosphoro-aminonitrile 2q (38%) and 2r (41%), respectively (Table 1, entries 17 and 18), which demonstrates that carbonyl stability is a key factor in controlling the phosphoro–Strecker reaction yield. Intrigued by the observed difference between ketone and aldehyde reactivity, we next investigated the competition between aldehydes and ketones in phosphoro–Strecker chemistry. Pleasingly, every combination of aldehyde/ketone (1:1, 200 mM, Method 4 in the Supplementary Methods, Supplementary Figs. 41–46 and Supplementary Tables 3–7) explored, underwent a highly selective reaction to furnish the aldehyde phosphoro-aminonitrile product (4, R2 = H). For example, a 1:1 mixture of acetaldehyde (1a) and acetone (1m) yielded 4a (70%) as the major product after 4 days (Supplementary Fig. 41) and only minimal conversion to 4m (<1%) was observed. There is a clear aldehyde selectivity in the phosphoro–Strecker reaction at neutral pH, therefore it is of note that aldehydes yield proteinogenic aminonitriles, whereas ketones yield α,α-disubstituted products that are not assigned to the universal genetic code.

Phosphoro–aminonitrile reactivity

Selective precipitation provided a facile route to isolate 4 without the use of chromatography. For example, the addition of methanol or ethanol to enhance solvent miscibility, followed by diethyl ether as an anti-solvent resulted in effective precipitation of 4 (Supplementary Figs. 47–54). With pure 4a–i and 4m in hand, we turned our attention to their reactivity. Aminonitriles (2a–k) undergo hydrolysis in water to give their respective cyanohydrins 7a–k (for example, α-aminopropionitrile 2b t1/2 = 0.5 h at 50 °C and pH 7.0)27. Under these conditions 4b underwent minimal decomposition to 7b (<5%). α,α-Disubstituted aminonitriles (2m–p) are stable at pH 7–1327,28,29, where as α,α-disubstituted phosphoro–aminonitriles 4m–p readily undergo retro–phosphoro–Strecker reaction at near-neutral pH. For example, when 4m (25–50 mM) was dissolved in D2O (pD 7.2), 1m (>90%) was liberated after 24 h at room temperature (Supplementary Fig. 55). Conversely, 4a–i were remarkably stable at pH 6–9 and room temperature; no decomposition was observed over a 24 h period, and aminonitrile 2 was only observed after long incubation times (>10 day).

Although 4a–i are highly stable at near-neutral pH, they hydrolyse under strongly acidic or alkaline conditions. Compounds 4a–f (100 mM) were incubated in acidic solution (initial pH 1.5, 50 °C, 24 h) and afforded the corresponding aminonitriles 2a–f in near-quantitative yield (Table 1, entries 1–6; Supplementary Figs. 56–59). A switch in reactivity was observed in alkaline solution.

Incubation of 4a–f (100 mM) under alkaline conditions (initial pH 13, 50 °C) led to very clean hydrolysis to afford 5 after 24 h (Table 1, entries 1–6; Supplementary Figs. 60–67) and, unexpectedly, small amounts of amino acid 3 (<10%). Whether these extreme pH conditions are relevant to prebiotic chemistry is not clear, but they clearly demonstrate a straightforward chemical methodology for the divergent synthesis of 2 and 5, and highlight the pronounced pH dependence on the reactivity of 4.

Intrigued by the stability of 4 in solution in neutral water, we next investigated (prebiotically plausible) thiolysis of 4. Thioamides 6 have been proposed as a reactive species for polymerisation to yield peptides3, but thiolysis of aminonitriles 2 is reported to yield a variety of products, including dithiopiperazines 930, rather than the corresponding thioamides 6. We strongly suspected that thiolysis of N-phosphoramide-masked nitriles 4 would prevent the formation of 9 (Fig. 2). Accordingly, we incubated 4a–i (100 mM) with hydrogen sulfide (H2S; 10 equiv., pH 9.0) and observed conversion to N-phosphoro-thioamides 10a-i, respectively (Supplementary Figs. 68–86). During thiolysis the solution pH increased to pH 11. Compound 10 was observed as the major product but, interestingly, thioamide 6 (<20%) was also observed, and N-phosphoro-amino acids 5 were detected in small amounts (<5%) after significantly longer (>30 day) incubation times. However, highly selective initial diamidophosphate hydrolysis was observed to liberate ammonia, which suggests that intramolecular nucleophilic catalysis promotes the release of ammonia (Fig. 4, black arrows). Encouraged by this observation, we decided to incubate 4a–i (50–100 mM) with H2S (10 equiv., pH 7.0), which afforded mixtures of the respective thioamides 6 (>60%) and thioamides 10. Again, the pH of these reactions increases to pH 9.0–9.2, but upon adjusting (or buffering) the solution to pH 7, thioamides 6a–i were obtained in excellent yields (Table 1, entries 1–9; and Supplementary Figs. 70–85). Importantly, this protocol represents a novel way to directly access to α-aminothioamides (6) in aqueous solution at neutral pH. Masking the amine of the Strecker product as a phosphoramide also avoids uncontrolled oligomerisation and, importantly, the formation of cyclic species (e.g. 9), which leads to a remarkably well-controlled synthesis of thioamides 6.

Fig. 4
figure 4

Proposed mechanism for the selective cleavage of the NH2 moiety of N-phosphoro-amino acid derivatives. Black arrows: Proposed intramolecular catalysis. Blue arrows: Major pathway; yields N-phosphoro-amino acid derivatives (10) (X = S) or (14) (X = O). Red arrows: Minor pathway; furnishes amino acids (3) or thioamides (6) and monoamidophosphate (MAP). Dashed arrow: Direct diamidophosphate hydrolysis (X = O) or thiolysis (X = S) is not observed

Full size image

To further shed light on the mechanism of diamidophosphate hydrolysis, 4a–f (100 mM) were incubated under alkaline conditions (initial pH 13) at room temperature. N-Phosphoro-amino acids 5 were obtained in good yields, but at room temperature these hydrolyses were sluggish (>6 day to completion), and numerous intermediates were observed. Interestingly, nitrile hydrolysis always preceded phosphoramide (P-NH2 moiety) hydrolysis to afford N-phosphoroamide 11 (Supplementary Fig. 87). Direct diamidophosphate hydrolysis/thiolysis to afford 12 was not observed. It is likely that after nitrile hydrolysis (or thiolysis) transient intramolecular amide/thioamide addition to phosphorous (to form cyclic phosphoramidate 13) promotes the release of ammonia (Fig. 4, black arrows). Cyclic intermediate 13, which is highly reminiscent of the activated five-membered phosphoramide intermediate reported by Rabinowitz (and others)12,13,14,15,16, rapidly opens to afford N-phosphoro-thioamide 10 or N-phosphoro-amide 14 (Fig. 4, blue arrows). This mechanism also explains the formation of small amounts of monoamidophosphate (MAP) and 3 or thioamide 6 under basic conditions (Fig. 4, red arrows). To test our hypothesised mechanism, we investigated the hydrolysis/thiolysis of 4 in stoichiometric competition with DAP. Rapid formation of N-phosphoro-amino acid 5 (or N-phosphoro thioamide 10) was observed, whereas DAP did not undergo hydrolysis and remained unchanged throughout the transformation of 4 to 5 or 10.

Finally, to probe the chemical continuity of the synthesis N-phosphoro aminonitriles and their subsequent reactions we investigated a one-pot synthesis and hydrolysis/thiolysis of N-phosphoro-aminonitriles 4. Compounds 4a–f (200 mM) were obtained in water at neutral pH, and the crude products were diluted (to 100 mM concentration) with acidic (pH 1.5) or alkaline solution (pH 13) and then incubated at 50 °C for 24 h to achieve complete hydrolysis. Alternatively, H2S (10 equiv.) was added to the crude compounds 4a–f (100 mM, pH 9), which where then incubated at room temperature until complete thiolysis was observed by NMR spectroscopy. The corresponding aminonitriles 2a–f, N-phosphoro-amino acids 5a–f and N-phosphoro-thioamides 10a–f were all obtained in comparable yields to the reactions performed with purified 4 (Supplementary Figs. 88–117, Supplementary Table 8).

Discussion

We demonstrate that 4 is obtained in good-to-excellent yields from the reaction of aldehydes (1a–l), HCN and DAP at neutral pH. The phosphoro–Strecker reaction provides a selective strategy for prebiotic amino acid synthesis: it is selective towards aldehydes, with poor conversion for ketone substrates, which thereby avoids the synthesis of non-biological α,α-disubstituted amino acids and leaves prebiotic ketones available to undergo other reaction pathways24,25. Furthermore, 4a–l are remarkably stable at near-neutral pH, whereas α,α-disubstituted products 4m–r degrade quickly to their corresponding ketones, HCN and DAP.

Thioamides 6 are easily obtained in near-quantitative yield at neutral pH from the reaction of 4 with H2S. The formation of 4, and the consequent desymmetrisation of diamidophosphate (DAP), promotes highly selective amidophosphate hydrolysis that cleaves the NH2 moiety and retains the amino acid moiety to selectively to furnish N-phosphoro amino acids derivatives (and not MAP). This methodology can be readily applied to the controlled synthesis of thioamides (6) to provide a mild protocol for their synthesis in neutral water, which warrants further investigation of 6 and its phosphorylated precursors in prebiotic peptide ligations. In the context of the origins of life, it is interesting to note that 4 is not only readily synthesised but also highly stable at neutral pH; this stands in stark contrast to the instability of aminonitriles 2 synthesised from ammonia. Prebiotic scenarios that result in pH oscillations and gradients can be envisaged31, and it is reasonable to consider that these oscillations/gradients may have played an important role during the origins of life. However, it is not clear that the extreme pH conditions required for the conversion of 4 into 2 or 3 would be available en route to life. More likely the efficient reactivity of H2S with the nitrile moiety of 4, which then leads to the synthesis of thioamide 10 and rapid (intramolecular) phosphoramide hydrolysis, would be crucial to the forward reactivity of 4 under geochemically plausible conditions. We are currently exploring the polymerisation, ligation and onward reactivity of 4, 6 and 10 following their efficient synthesis at near-neutral pH.

Methods

Phosphoro–Strecker reaction protocol A

Aldehyde (1a–c, 2 mmol), sodium cyanide (1.2 equiv.) and DAP (4 equiv.) were dissolved in water (10 mL) at pH 7.0. This solution was then stirred at room temperature and monitored periodically by 1H and 31P NMR spectroscopy. When the NMR spectra showed no further change in composition, ethanol (40–60 mL) was added to precipitate DAP. DAP was removed by filtration and the residual solution was concentrated in vacuo. The residue was dissolved in water/ethanol (1:2, 20 mL) and 4a–c were precipitated by addition of diethyl ether (20–40 mL) and isolated by filtration.

Phosphoro–Strecker reaction protocol B

Aldehyde 1d–i (2 mmol), sodium cyanide (1.2 equiv.) and DAP (4 equiv.) were dissolved in water (10 mL) at pH 7.0. This solution was then stirred at room temperature and monitored periodically by 1H and 31P NMR spectroscopy. When the NMR spectra showed no further change in composition, methanol/ethanol (1:4–1:6, 50–70 mL) were added to precipitate DAP. DAP was removed by filtration and the residual solution was concentrated in vacuo. The residue was dissolved in water/methanol/ethanol (1:1:2, 20 mL), and 4d–i were precipitated by addition of diethyl ether (20–40 mL) and isolated by filtration (See Supplementary Methods for compound data).

Phosphoro–Strecker reaction protocol C

Aldehyde 1a–i (2 mmol), sodium cyanide (1.2 equiv.) and DAP (4 equiv.) were dissolved in water/formamide or water/DMSO (10 mL, 9:1) at pH 7.0. This solution was then stirred at room temperature and monitored periodically by 1H and 31P NMR spectroscopy. When the NMR spectra showed no further change in composition the solution was lyophilised. The residue was dissolved in water (10 mL). DAP, and subsequently 4a–c or 4d–i, were then precipitated as described in Protocol A or Protocol B above, respectively (See Supplementary Methods for compound data).

Acid hydrolysis

Phosphoro-amino nitrile 4 (100 mM) was dissolved in H2O/D2O (9:1). The pH of the solution was adjusted to pH 1.5 by addition of 4 M HCl. The reaction mixture was heated at 50 °C. The reaction pH was monitored and periodically adjusted to pH 1.5. Once the pH of the solution had stabilised at pH 1.5, the solution was analysed by NMR spectroscopy.

Alkaline hydrolysis

Phosphoro-amino nitrile 4 (100 mM) was dissolved in H2O/D2O (9:1). The pH of the solution was adjusted to pH 13.5 by addition of 4 M NaOH. The reaction mixture was stirred at 50 °C and monitored by 1H and 31P NMR spectroscopy (See Supplementary Methods for compound data).

Thiolysis

Phosphoro–amino nitrile 4 (50–100 mM) and NaSH (10 equiv.) were dissolved in H2O/D2O (9:1) or 500 mM phosphate buffer at pH 7 or 9 and stirred at room temperature. The reaction was monitored by 1H and 31P NMR spectroscopy until complete consumption of the diamidophosphate was observed, then the excess H2S was purged with Ar at pH 7 (See Supplementary Methods for compound data).