Phosphorylation of nucleosides by P-N bond species generated from prebiotic reduced phosphorus sources

Abstract

P-N species e.g., amidophosphates readily phosphorylate organics, thereby overcoming the so-called ‘Phosphate Problem’. However, the formation of amidophosphates by plausible early Earth geochemical pathways is limited. We herein show that ammonolysis of the prebiotically plausible dimer of phosphite, pyrophosphite, readily affords amidophosphite, the monomeric P-N derivative of phosphite. Amidophosphite then undergoes spontaneous oxidation to form monoamidophosphate (MAP) and diamidophosphate (DAP) at room temperature (yields of the inorganic P-N species up to 48%). Oxidation of amidophosphite is promoted by O₂, H₂O₂, ClO⁻ and by UV light irradiation (365 nm). Both amidophosphite and MAP and the crude reaction mixture react with nucleosides to form nucleotides with both phosphate and H-phosphonate (yields up to 65%) at 80 °C in the presence of urea, showing that monoamidated phosphorus compounds also willingly promote prebiotic reactions. This observation expands the range of P-N phosphorylating agents that can play a role in the chemical evolution of prebiotic molecules on the early Earth.

The challenge of the ‘Phosphate Problem’ in prebiotic chemistry limits the formation of key phosphorylating agents. Here, the authors demonstrate that ammonolysis of pyrophosphite yields amidophosphite, which oxidizes to monoamidophosphate, facilitating nucleotide formation from nucleosides.

Introduction

Phosphorus (P, hereafter) plays a vital role in the genetic framework, cellular architecture, and metabolic processes of extant life, and presumably was important for the origin of life^1–4. Amides of phosphates such as monoamidophosphate (MAP) and diamidophosphate (DAP) have been shown to facilitate phosphorylation reactions. These reactions yield an array of phosphorylated molecules^5–8, which could have been significant in the primordial synthesis of the evolutionary precursors of biomolecules on the early Earth. In contrast, most phosphorylation studies in prebiotic chemistry have employed phosphorus minerals found commonly on Earth’s surface as P sources. These are mainly orthophosphate minerals where phosphate is bonded with divalent metals such as Ca^{2+ 2}, which exhibit low solubility and reactivity. However, phosphate solubility may have been enhanced via reactions with carbonate under alkaline conditions⁹ or ferruginous mineralization¹⁰.

As an alternative to phosphate, reduced oxidation state phosphorus (reduced P, hereafter) compounds such as phosphite may have served in prebiotic reactions. Phosphite is generated through meteorite corrosion^11–13, and by terrestrial processes including lightning^14,15, serpentinization¹⁶, hydrothermal reactions¹⁷, phosphorus biogeochemical cycling¹⁸, by marine planktonic phosphate reduction¹⁹, as well as iron redox geochemistry²⁰, and extraterrestrial impacts^21,22. The plausibility of reduced P compounds being present on the prebiotic Earth is further supported by the discovery of phosphonic acids in the Murchison meteorite²³, and the presence of phosphite in early Archean (3.52 billion year old) marine carbonates²⁴. It is estimated that the total mass of reduced P delivered to the early Earth during the late accretion period (e.g., from 4.50 Ga to 3.50 Ga) was around 10²⁰ kg²⁵.

The phosphide Fe₃P, which is an analog to the meteoritic mineral schreibersite, produces MAP and DAP by reaction with aqueous NH₄OH²⁶. MAP and DAP are also formed by hydrolysis of P₄O₁₀, and P-N bonds form via the ring opening of trimetaphosphate (which releases about 40 kJ/mol). In iron phosphide, the energy of hydrolysis and oxidation also exceeds 30 kJ/mol (Table 1). Below the 30 kJ/mol threshold, no amidation occurs. Using this free energy as a guide, we investigated how the dimer of the phosphite ion, termed pyrophosphite, may also afford amido-P molecules, as the hydrolysis of pyrophosphite is about 40 kJ/mol²⁷.

Table 1.

Free energies of hydrolysis of prebiotic phosphorylation agents^a

Name/formula/reaction	ΔG0rxn (kJ/mol)	Ref.
P4O10 + 6H2O = 4H3PO4	−420	26
Schreibersite oxidation, Fe3P + 4H2O + 4H+ = HPO42– + 3Fe2+ + 5.5H2	−285	26
MAP hydrolysis, H2NPO3H– + H2O = HPO42– + NH4+	−30	26
DAP hydrolysis, (H2N)2PO2– + 2H2O = H1.5PO41.5– + 2NH4+	−70	26
Pyrophosphate hydrolysis, HP2O73–	−19	67
Trimetaphosphate (ring-opening), HP3O92–	−40	68

^aThe reaction of hydrolysis means here the addition of water to form the products listed in the table and is generally given at pH 7, except for the P₄O₁₀ reactions. Modified from ref. ¹¹.

Pyrophosphite is readily formed in environments where phosphite is present and undergoes drying cycles²⁷. Despite its high energy of hydrolysis, pyrophosphite forms more readily than pyrophosphate, largely due to the higher solubility of the phosphite ion compared to the phosphate ion. For example, pyrophosphite has been produced through incubating meteoritic fluids²⁸, under relatively low temperatures, and, if phosphite is heated with urea, pyrophosphite can be produced at yields of >80%²⁹.

Inspired from previous findings namely the (1) ubiquity and high solubility of reduced P compounds on the early Earth^12–20,30, (2) prebiotic syntheses of high energy condensed reduced P compounds (e.g., pyrophosphite [PPi(III)])^3,27–29, and (3) generation of P-N compounds from the reaction of iron phosphide (Fe₃P) and aqueous ammonia²⁶, we explored further prebiotic pathways for the generation of reactive P-N compounds from reduced P sources that have not been considered previously.

We present here the production of the inorganic P-N compound amidophosphite (amide-derivative of phosphite) by the ammonolysis of pyrophosphite [PPi(III)] under ambient conditions. We also show amidophosphite autoxidizes at room temperature and under sealed oxic conditions to MAP and even DAP. We acknowledge that many of these conditions are not prebiotic but focus on establishing chemical pathways. We further show the availability, stability and generation of such P-N species under various prebiotically relevant geological conditions such as more moderate pH, lower ammonium concentrations, and through oxidizing conditions that do not rely on O₂. We then demonstrate that the crude reaction mixtures containing P-N molecules are capable of phosphonylating and phosphorylating organic molecules. Significantly, we show that MAP by itself is capable of phosphorylating nucleosides to form nucleotides.

Results

Inorganic production of amido-P compounds

After confirming that the starting P sources did not show any impurities (Figs. S1 and S2), pyrophosphite [PPi (III)] was synthesized from phosphite (Figs. S3 and S4), followed by the study of the generation of the inorganic P-N species. When pyrophosphite (in solid state) was added to a solution of NH₄OH (14 N), it initially generated amidophosphite as the only amidated phosphorus molecule (Figs. S5 and S6)³¹. After 24 h, the amidophosphite was partially oxidized to monoamidophosphate (MAP) and, to a much lesser extent, diamidophosphate (DAP) (Figs. 1 and S7, Scheme 1 and Table S1). With the addition of ammonium hydroxide to pyrophosphite, we observed the formation of amidophosphite within 30 min with a conversion of around 22–23% which continued to rise for 24 h and consistently reached a maximum of 44–47%. The yields (based on the relative integration) of the P-N compounds ranged from 0.5–48%. Interestingly, the amidophosphite concentration declined over time, with a concomitant increase in MAP, which is attributed to the oxidation of amidophosphite. The reduced P compounds oxidized with a half-life of about 15 days under air at pH 13.5 and at room temperature. Ammonolysis of pyrophosphate (the oxidized analog of pyrophosphite) with 14 N NH₄OH, on the other hand, only showed the formation of orthophosphate (Fig. S8). Both MAP and DAP were also confirmed through spiking with authentic samples of the respective ions (Figs. S9 and S10). The experiment was repeated with a ¹⁵N labeled NH₄OH solution, which confirmed the formation of P-N compounds (Fig. S11).

Fig. 1. Time progression ³¹P-NMR {H-coupled} spectra showing the formation of P-N compounds over time.

The reaction was carried out by adding 4 ml of 14 N ammonium hydroxide (pH = 14) to a clean glass vial (20 ml capacity) containing powdered (0.2 g) pyrophosphite and a clean magnetic stirring bar. The (Teflon capped) sealed reaction vial was immediately stirred at room temperature under air. Inset is ×1.5 magnification (left) and ×2 magnification (right).

Scheme 1. Routes to form various P-N, P-O, P-O-P and P-H species obtained*.

*Reaction pathway to form P-N compounds by pyrophosphite ammonolysis with 14 N NH₄OH to form various P-N species and various other inorganic P-compounds.

The ammonolysis of pyrophosphite was rapid in the NH₄OH solution (Fig. 1) due to the high pH (14). Both pyrophosphite and pyrophosphate are stable near neutral pH, and both hydrolyze rapidly under low pH conditions (Fig. S12). However, pyrophosphite is much less stable than pyrophosphate under high pH conditions, where its rate of hydrolysis is ~10¹⁰ times faster³². Given the pH sensitivity of pyrophosphite hydrolysis, we investigated the formation of P-N compounds under pH conditions <14, and observed that even at pH 8 pyrophosphite still yields amidophosphite (~1% after 30 min) and over time both MAP and DAP (2–4% after 24 h), albeit with less ammonolysis and hydrolysis of pyrophosphite (Fig. S12 and Table S2). We observed that the generation of the P-N compounds was successful under ambient conditions, with pH >7 and was best at pH >10. The total yields of P-N species at various pH values were as follows; 1% at pH = 8, 21% at pH = 9 and 28% at pH = 11 and 22.5% at pH = 14. These reactions were studied within 30 min of adding NH₄OH (with pH controlled by addition of HCl) to the (powdered) pyrophosphite. However, after 24 h higher yields were observed with the solution having highest pH, e.g., 47% at pH = 14, compared to 37% at pH = 9, and 38% at pH = 11 (day 1 or 24 h) (Table S2). Furthermore, the amount of P-N bonds from pyrophosphite generally exceeds the molar fraction of NH₃ (vs. H₂O, Fig. 2), and these P-N bonds are formed through ammonolysis and not oxidation. Amidated phosphorus compounds are produced even at 0.2 M of ammonia/ammonium, though the pH begins to approach neutrality at these lower concentrations and hence pyrophosphite becomes more stable toward hydrolysis (Fig. S13 and Table S3). The formation of MAP from pyrophosphite via amidophosphite suggests an oxidation step.

Fig. 2. Investigation of N incorporation in pyrophosphite.

a The fraction of P-H bonds in solution vs. mole fraction of N in solution stays roughly constant, showing little oxidation of phosphite, b the fraction of P-N bonds in solution vs. the mole fraction of N in solution increases with increasing N in solution, and c the fraction of P-H stays roughly constant even though the number of P-N bonds increases.

This ease of oxidation of amidophosphite was surprising, as the phosphite ion—despite bearing a reduced oxidation state relative to phosphate—is quite stable to oxidation even under air (e.g., Herschy et al.²⁰ demonstrated only 0.1% oxidation in water sealed under air after 5 years)²⁰. We sought to identify the source of the oxidant in this reaction, by redoing the reaction under an N₂ atmosphere. Notably, the oxidation of amidophosphite to MAP is greatly reduced in the absence of O₂, demonstrating that amidophosphite is much more susceptible to oxidation than phosphite.

When ¹⁸O labeled H₂O was used under ambient conditions (in air and at room temperature), the MAP that formed did not incorporate ¹⁸O indicating that the amidophosphite instead likely reacted with adventitious oxygen (Fig. S14). This was also supplemented by observations that under anoxic conditions, we obtained only 0.01% of MAP, even though 41% amidophosphite was present in the reaction mixture (Fig. S15 and Table S4). With the awareness that amidophosphite is more susceptible to oxidation by air than phosphite, we investigated the proclivity of oxidation of amidophosphite under N₂ in the presence of other oxidants (Fig. S16). We observed that both hydrogen peroxide and hypochlorite promoted the oxidation of amidophosphite to MAP under N₂ (Fig. 3) while no oxidation occurred with sulfate, Fe³⁺ or nitrate as oxidants (Fig. S17 and Table S5).

Fig. 3. ³¹P-NMR {H-coupled} spectra showing the formation of MAP and DAP by the oxidation of amidophosphite produced by the ammonolysis pyrophosphite by hypochlorite and peroxide under anoxic conditions.

In each reaction, 4000 µl, 14 N NH₄OH and 1.05 mmoles of pyrophosphite were used. The initial molarities and volumes of the respective oxidants are (a) No additive, (b) 0.70 M ClO⁻ (300 µl), (c) 0.70 M ClO⁻ (700 µl), (d) 4.9 M H₂O₂ (300 µl) and (e) 4.9 M H₂O₂ (700 µl). For the complete figure see SI Fig. S18). Inset is ×1.5 magnification (left) and ×3 magnification (right).

Under anoxic conditions (N₂ atmosphere), H₂O₂ oxidized amidophosphite to about 7% MAP after 5 days (200 µl, 4.9 M H₂O₂ solution). Adding more H₂O₂ to the solution negatively impacted the stability of the P-N compounds, as there was a competition between oxidation of amidophosphite to MAP vs. hydrolysis of amidophosphite to phosphite as the pH decreased. Hypochlorite also assisted with the oxidation of amidophosphite to MAP (1–1.6%) (Figs. 3 and S18 and Table S6).

We studied if the oxidation of amidophosphite to MAP and DAP could also be favored under air and by adding H₂O₂ as an additional oxidizing agent. We observed that adding 100 µl of 4.9 M H₂O₂ solution to the reaction mixture and consistent stirring under ambient (sealed) conditions for 10 days formed about 14% MAP and 0.5% DAP. These results are comparable to results without H₂O₂, where we observed 26% MAP and 0.6% DAP. When a solution containing pyrophosphite and ammonium hydroxide was directly exposed to a UV light source of 365 nm under N₂, amidophosphite oxidized to MAP (2.5%) and DAP (0.5%). We also observed polyphosphates (mainly including triphosphates, 1–1.5%) along with phosphite (51%) and phosphate (5.2%) (Figs. S20–S22). The overall yield of P-N compounds was 40%, including amidophosphite. When exposed to a higher frequency (245 nm) UV light, amidophosphite (8.5%) and phosphite (89%) were the major products but MAP and DAP were not observed (Fig. S23 and Table S8, entry 1 and 2) after 20 days (Fig. S19 and Table S7). ³¹P-NMR analysis of UV experiments at 3 and 6 weeks (Fig. S22), showed that the yield of MAP in both samples was significantly higher (e.g., 22% after 3 weeks and 20% after 6 weeks) with the yield of DAP to be consistently in the range of 0.4–0.5%. The overall yield of P-N compounds in this sample over the time span of 3 to 6 weeks was 21–24%. This also showed that the P-N compounds under alkaline pH (≈14) were quite stable in the solution (Table S8, entry 3 and 4).

Pyrophosphite did not hydrolyze significantly when exposed to 365 nm UV light under anoxic conditions in water at pH of 5.5–6 (without adding any NH₄OH) (Fig. S24 and Table S8, entry 5 and 6) (pyrophosphite 87%) with only 5.5% phosphite and 2.5% of orthophosphate, showing pyrophosphite is stable at lower pH. Interestingly, when a similar reaction was carried out at a pH of around 14, about 77% phosphite, 20% orthophosphate, 2.6% polyphosphates and 0.4% an unknown peak ‘PX’ (possibly peroxophosphate) were obtained. Pyrophosphite seemed to have completely been hydrolyzed at higher pH (Fig. S24 and Table S8, entry 5 and 6). These results were also consistent with our pH studies (Fig. S12 and Table S2). This observation was further strengthened by the comparative experiments of ammonolysis (pH = 14) vs. hydrolysis (pH = 5.5–6) of pyrophosphite under ambient conditions (Fig. S25 and Table S9). These experiments demonstrate that the formation of amido-P compounds occurs across a range of pH conditions and varied ammonia concentrations, and while the oxidation is promoted by O₂, other oxidants and UV irradiation can also affect this oxidation. These results hint that the formation of amido-P compounds, including amidophosphates, should be relatively robust across a range of prebiotic environments.

As further evidence of the robustness of this pathway, we also investigated the ammonolysis of pyrophosphite in simulated saltwater with oceanic concentrations of salts (Na, K, Mg, and Ca), under anoxic conditions. The formation of amido-P compounds still readily occurs, including the formation of a small amount of MAP (Fig. S26). Under anoxic conditions, when the generation of the P-N compounds was studied in the presence of ocean salts, the detectable products included amidophosphite (21%), phosphite and even (0.7%) MAP, suggesting the ease of formation and relative stability of the P-N compounds as opposed to orthophosphate (Fig. S26 and Table S4). We also investigated if other sources of phosphite may afford amido-P compounds. We observed that the heating of CaHPO₃ or a synthetic analog of schreibersite, Fe₃P, with urea, followed by reaction with NH₄OH solution also affords amido-P compounds, including MAP and even DAP (Fig. 4).

Fig. 4. Various sources of reduced P were used as starting material to form pyrophosphite.

Pyrophosphite was obtained generally by heating the aqueous solutions of equimolar (0.1 M) phosphorus acid and sodium hydroxide at 200 °C, in a tube furnace, in an atmosphere of N₂. Hypophosphite was reacted with two equivalents of H₂O₂ and one of FeCl₂ in water to form phosphite alongside other oxidation (inorganic) P products. The aqueous solutions of all sources of phosphite were heated at 78–80 °C, un-sealed, in the presence of urea to form pyrophosphite.

The yields of P-N compounds reached 48% when Fe₃P was used as a P source (Figs. S27–S30 and Table S10). Another route investigated was the oxidation of hypophosphite³³, as a starting material. The reaction products were heated with urea^34,35, and subsequently subjected to ammonolysis, generating (15%) P-N compounds (Figs. S31 and S32, Table S11 and Scheme S1).

Phosphorylation and H-phosphonylation of nucleosides

Having demonstrated the formation of P-N compounds from various P-sources, we then investigated the reactivity of the P-N compounds produced from pyrophosphite: amidophosphite and MAP. While the phosphorylation potential of DAP has been investigated in this context³⁶, the reactions of amidophosphite and MAP have not been studied. We chose ribonucleosides and deoxyribonucleosides as the test system to compare the reactivity of MAP and amidophosphite with that of DAP, since the DAP-mediated phosphorylation reactions of nucleosides have been well studied^6–8. We also included orthophosphate and pyrophosphate to check the phosphorylation activity and because MAP is known to convert to orthophosphate and pyrophosphate under various conditions^7,37. In addition, we studied the H-phosphonylation reactions of pyrophosphite and phosphite to compare with the crude reaction mixture containing amidophosphite and phosphite that was generated from the reaction of pyrophosphite with ammonia.

These reagents were contrasted to DAP under the same reaction conditions. All reactions were carried out with urea as an additive. The reactions were monitored with ¹H and ³¹P-NMR and mass spectrometry. Where possible authentic commercial standard nucleotides were used to confirm product identity by comparison/spiking (Figs. S33–S40). The conversion % was calculated based on the ¹H-NMR integration of signals of starting nucleoside versus the nucleotides after the deconvolution process using MestReNOva³⁸.

We first studied MAP (Fig. S41, for ³¹P-NMR of pure MAP, DAP, phosphate and pyrophosphate) as a phosphorylation agent (urea, 80 °C) with five canonical ribonucleosides (U, C, A, G and I). Using the integral values of the ¹H-NMR signals of the various phosphorylated products that could be identified vis-á-vis the signal of the starting nucleoside, we observed that pyrimidines had higher efficiency of phosphorylation compared to purines (Table 2). Uridine, guanosine and inosine showed similar % conversion to the corresponding nucleotides. The low phosphorylation conversion of certain nucleosides (e.g., guanosine) could be attributed to the low solubility of purines, however inosine is phosphorylated efficiently with DAP (Table 2).

Table 2.

Ribo and deoxyribonucleoside phosphorylation using different P compounds^a

Entry
	Nucleobase	Ribonucleoside total phosphorylation (%)b
		DAP (avg. 60%)	MAP (avg. 51%)	Phosphate (avg. 39%)	Pyrophosphate (avg. 45%)
1	Cytosine (C) (avg. 69%)	77	67	75	59
2	Uracil (U) (avg. 50%)	73	55	43	29
3	Adenine (A) (avg. 35%)	42	56	14	28
4	Guanosine (G) (avg. 28%)	24	34	14	42
5	Hypoxanthine (I) (avg. 61%)	85	45	48	67
Entry
	Nucleobase	Deoxyribonucleoside total phosphorylation (%)c
		DAP (avg. 61%)	MAP (avg. 56%)	Phosphate (avg. 36%)	Pyrophosphate (avg. 56%)
6	Cytosine (dC) (avg. 61%)	59	53	45	87
7	Thymine (dT) (avg. 62%)	79	80	19	72
8	Adenine (dA) (avg. 41%)	46	52	24	44
9	Guanosine (dG) (avg. 44%)	62	40	56	21

^aConditions: Deoxy or ribonucleoside (0.05 mmol, 1 equiv), P-Compound (5 equiv), urea (3 equiv), and 100 μL of water. The reactions were carried out at 80 °C for 3 days uncapped.

^bTotal ribonucleoside phosphorylation yield refers to the sum of 2’,3’-cyclic phosphate + 2’-phosphate + 3’-phosphate + 5’-phosphate of ribonucleosides.

^cTotal deoxyribonucleoside phosphorylation yield refers to the sum of 3’-phosphate + 5’-phosphate and 3’,5’-bis phosphate of deoxyribonucleosides. Percent conversion yields were determined by ¹H-NMR (obtained after deconvolution of the spectrum).

Table 2 shows a comparison of the phosphorylation outcomes for the four phosphorylation agents used in this study, under the same conditions. In general, ribonucleosides had lower conversion compared to deoxyribonucleosides with some individual exceptions. Furthermore, paralleling the previously observed pyrimidine versus purine dichotomy in phosphorylation^6,7, pyrimidine phosphorylation was higher (avg. 50–69%) compared to purines (avg. 28–35%)—with hypoxanthine being the exception (avg. 61%).

The total phosphorylation of uridine was around 55% (Table 2). Among the products the 2’,3’-cyclic phosphate derivative was dominant (38%) while 5’-UMP was the second major product (25%, Figs. S42–S45). The total phosphorylation of cytidine was 67%, and included 5’-CMP, 2’,3’-cyclic phosphate cytidine and 2’-CMP (Fig. 5). There was no evidence of 3’-CMP as judged by ¹H-NMR (Figs. 5 and S46–S49) and by comparison to known value in literature^7,8. In the purine series, adenosine, guanosine and inosine were phosphorylated with 56%, 34% and 45% conversion (Figs. S50–S61) with 2’-AMP, 3’-GMP and 3’-IMP being the most abundant species respectively in each of the series.

Fig. 5. ¹H-NMR spectra of crude reaction of cytidine and the phosphorylating agents after 3 days.

(a) t = 15 min of reaction, (b) MAP, (c) DAP, (d) Phosphate and (e) Pyrophosphate. (For conditions see Table 2 and for full spectra see Fig. S46).

In all the above reactions an extra signal around 19 ppm in the ³¹P-NMR was observed, close to the signal of the 2’,3’-cyclic monophosphate; this signal was the most intense in cytidine compared to the other nucleosides. This signal was assigned as the bisphosphate (2’,3’-cyclic and the phosphate in the 5’-position) of cytidine and was identified by the mass spectral data (m/z found 384.00511, Fig. S49) and the observation in ³¹P-NMR of a signal around 20.38 and 3.85 ppm. We also observed 5’-phosphate of 2’,3’-cyclic phosphate of uridine based on the signal at 7.61 ppm in the ¹H-NMR and peaks at 19.83 and 3.79 ppm in the ³¹P-NMR (Figs. S42 and S43) and mass spectra (m/z found 384.98929, Fig. S45). For adenosine, guanosine and inosine the spectra were too complex to see the signal of the 2’,3’-cyclic and the phosphate in the 5’- position, but ³¹P-NMR (Figs. S51, S55 and S59, respectively) and mass spectra (Figs. S53, S57 and S61, respectively) showed the bisphosphate of ribonucleosides.

However, the formation of the 2’,3’-cyclic phosphate derivative was surprising since MAP had only one amino group as opposed to DAP—which has two amino groups and was thought to be necessary for the second intramolecular phosphorylation step to form the cyclic phosphate. That MAP itself could phosphorylate nucleosides with slightly less efficiency than DAP (under identical conditions, Table 2, entries 4–5) was gratifying since it is more easily formed in prebiotic environments via the processes shown above and by other work³⁹. That MAP also forms 2’,3’-cyclic phosphate suggests that the initially formed uridine 2’- or 3’-monophosphate from the reaction of MAP with uridine, which then cyclizes further to form 2’,3’-cyclic-UMP in the presence of urea (or MAP) under these reaction conditions.

We used 3’-UMP as proof of principle for the 2’,3’-cyclic phosphate formation. We heated the 3’-UMP in water for 3 days at 80 °C and saw no evidence (based in ¹H and ³¹P-NMR) of cyclization, similar result was obtained when the reaction only was run with MAP. Nevertheless, when the reaction mixture was 3’-UMP and urea (no MAP), the cyclization took place. Based on ¹H-NMR the conversion was about 67% (Figs. S62 and S63). Urea plays a pivotal role in MAP phosphorylation for the formation of 2’,3’-cyclic phosphates. We also observe other 2’,3’-cyclic phosphate products as seen by the extra peaks in ³¹P-NMR around 19–20 ppm (Fig. S63). These could be assigned to the cyclic phosphate of uridine dimer (UpU) and cyclic phosphate of 5’-uridine phosphate. Such assignments are corroborated by the observation of corresponding peaks in the mass spectra (Fig. S64).

Next, we investigated the reactions of deoxyribonucleosides—deoxycytidine (dC), deoxythymidine (dT), deoxyadenosine (dA) and deoxyguanosine (dG)—using MAP for phosphorylation. For deoxythymidine, the most abundant was the 3’,5’-deoxythymidine bisphosphate, followed by the 3’-dTMP and finally 5’-dTMP (Figs. S65–S68). However, for the deoxycytidine the total phosphorylation was 52% (Figs. S69–S72). Deoxyadenosine and deoxyguanosine had similar phosphorylation conversions (52% and 40%, respectively, Figs. S73–S80). Interestingly, in all experiments, the most abundant compound was the 3’,5’-bisphosphate deoxyribonucleoside (Table 2).

In all the above reactions we observed the formation of pyrophosphate (PPi) and orthophosphate (Pi) as side products. Therefore, we investigated the reactions of these two with all five ribonucleosides (U, C, A, G and I) under the same reaction conditions (urea and 80 °C) to understand their reactivity and contribution towards phosphorylation. Overall, the average phosphorylation was 42–49% for PPi while for the phosphate it was 21–33% with deoxyribonucleosides generally providing lower conversion when compared to ribonucleosides. In both cases, in general pyrimidines had relatively higher conversion (and they more easily form the 2’,3’-cyclic phosphates) when compared to purines (major products are monophosphates) (Table 2 and Figs. S44, S48, S52, S56 and S60).

Based on these values, of the two, pyrophosphate seemed to behave like MAP (Table 2) in terms of overall efficiency. However, the hydrolysis of pyrophosphate was less when compared to MAP, according to the abundance of the signal in ³¹P-NMR. While MAP phosphorylated the nucleoside, it also then is hydrolyzed to phosphate and pyrophosphate after the 3rd day, whereas pyrophosphate enabled phosphorylation without its own hydrolysis process.

Among the reagents MAP and pyrophosphate are better than inorganic phosphate in terms of overall phosphorylation efficiency (avg. 46/52% and 42/60% respectively), approaching the proficiency of DAP (avg. 62/48%) under these reaction conditions. The main difference between the various P-reagents is in their efficiency and distribution of products. DAP-mediated phosphorylation produces 2’,3’-cyclic phosphate ribonucleoside while MAP favors the 5’- and the 3’-monophosphate ribonucleoside.

Not surprisingly, urea as an additive is needed for phosphorylation of ribo- and deoxyribonucleosides with some of these reagents. When we conducted the reaction of uridine with the phosphorylating reagents without urea, only MAP and DAP gave some phosphorylation (10% and 22%, respectively), while phosphate and pyrophosphate did not (Figs. S81 and S82). This observation clearly emphasizes the enabling role of nitrogenous compounds and the importance of P-N bonds for phosphorylation when using phosphate-derived compounds.

Since reduced P-compounds were produced in the above investigations, we also studied the corresponding reactions of phosphite, amidophosphite (from the crude reaction mixture) and pyrophosphite with ribo- and deoxyribonucleosides. Utilizing the same reaction conditions of urea as an additive at 80 °C, we monitored the H-phosphonylation of nucleosides by ¹H- and ³¹P-NMR. The phosphonylation efficiency (as judged by the % conversion in ¹H NMR), was again dependent on the nature of the nucleoside and the reagent (Table 3).

Table 3.

Ribo and deoxyribonucleoside phosphonylation using different P compounds^a

Entry
	Nucleobase	Ribonucleoside total phosphonylation (%)b
		Phosphite (avg. 50%)	Amidophosphite (avg. 77%)	Pyrophosphite (avg. 65%)
1	Cytosine (C) (avg. 51%)	49	67	39
2	Uracil (U) (avg. 84%)	87	85	80
3	Adenine (A) (avg. 72%)	54	79	83
4	Guanosine (G) (avg. 43%)	13	69	47
5	Hypoxanthine (I) (avg. 71%)	51	87	77
Entry
	Nucleobase	Deoxyribonucleoside total phosphonylation (%)c
		Phosphite (avg. 48%)	Amidophosphite (avg. 61%)	Pyrophosphite (avg. 35%)
1	Cytosine (C) (avg. 46%)	55	48	36
2	Thymine (T) (avg. 75%)	76	77	74
3	Adenine (A) (avg. 44%)	50	65	18
4	Guanosine (G) (avg. 27%)	13	54	14

^aConditions: Deoxy or ribonucleoside (0.05 mmol, 1 equiv), P-Compound (5 equiv), urea (3 equiv), and 100 μL of water. The reactions were carried out at 80 °C for 3 days uncapped.

^bTotal ribonucleoside phosphonylation yield refers to the sum of 2’,3’-cyclic phosphate + 2’-H-phosphonate + 3’-H-phosphonate + 5’-H-phosphonate of ribonucleosides.

^cTotal deoxyribonucleoside phosphonylation yield refers to the sum of 3’-H-phosphonate + 5’-H-phosphonate and 3’,5’-bis phosphite of deoxyribonucleosides. Percent conversion yields were determined by ¹H NMR (obtained after deconvolution of the spectrum).

Overall, amidophosphite gave the best conversion of the nucleosides with an avg. 77% in the ribo-series and 61% in the deoxyribo-series, followed phosphite and pyrophosphite. There was no stark difference between the purines and pyrimidines. Though among the pyrimidines, cytosine gave lower conversion yields compared to uracil and thymine while guanine was consistently lower than adenine (Table 3). Many of the H-phosphonate products of nucleosides are not known or marginally described and not available as standards. Therefore, we had to rely on the combination of ¹H- and {H-coupled} ³¹P-NMR to assign structures of the products (Figs. S83 and S84). While some of the structures could be ascertained with a degree of certainty, the identity of other H-phosphonate products is not certain. Most of the assignments were done by comparison of the chemical shift in ¹H-NMR of the authentic nucleotides (example 5’-NMP with the corresponding 5’-H-phosphonate).

In the case of the deoxyribonucleosides since the number of possible sites for reaction (compared to ribonucleosides) are limited, it was relatively straightforward to identify the products. For example, we could identify the products from the reaction of deoxycytidine with pyrophosphite (Fig. 6). In the ³¹P-NMR {H-coupled} we observed two triplet signals that were assigned to the 5’-H-phosphonate deoxycytidine, since the triplets are due to the coupling with the 5’-CH₂- and there are two triplets resulting from the coupling of P–H. The two doublets are for the 3’-H-phosphonate (doublet for the coupling of the 3’-CH and from the P-H). Finally, the two singlets belong to the phosphite compound that is formed after the hydrolysis of the pyrophosphite (two triplets in the t = 15 min spectra). In a similar fashion we could assign the structures in all the deoxy series (Table 3 and Figs. S110–S125).

Fig. 6. Reaction mixture of deoxycytidine (dC) (0.05 mmol, 1 equiv), pyrophosphite (5 equiv), urea (3 equiv) and 100 μL of water at 80 °C after 3rd day.

The reactions were carried out at 80 °C for 3 days uncapped. a ¹H NMR of reaction crude, red color means dC (starting material), the signals of products are close to the dC signal. b ³¹P NMR {H-Coupled} of the same reaction, after the 3rd day, pyrophosphite was hydrolyzed to phosphite and oxidized to phosphate, the products are 5’-H-phosphonate deoxycytidine, 3’-H-phosphonate deoxycytidine and the 3’,5’-H-diphosphonate deoxycytidine.

The product assignments in the ribonucleoside series were complicated by the overlapping signals in the ¹H-NMR. We relied on the ¹H-coupled ³¹P-NMR to ascertain whether the P-signals observed were indeed attached to the nucleoside—based on the splitting pattern, a doublet when attached to the 2’- or 3’- position and a triplet at the 5’-position. In the pyrimidine series, the H-phosphonate products from uridine and cytidine differ in the sense that in the case of cytidine the major product was the 5’-H-phosphonate while with uridine the major product was the 2’-H-phosphonate, based in the comparison of chemical shift in ¹H-NMR of the 2’-UMP (Figs. S90–S94). Reactions were also analyzed by mass spectrometry to corroborate the NMR interpretations. In the case of uridine two major peaks were observed corresponding to a mono H-phosphonate (the 5’-, 3’- or 2’-H-phosphonate) and the other corresponded to the bis-H-phosphonate (Fig. 7). The purines reactivity (adenosine and inosine) is similar with each other in terms of phosphonylation conversion (up to 88%), whereas guanosine phosphonylation is lower (average 39%) due to poor solubility.

Fig. 7. Mass spectra of the crude reaction of uridine (0.05 mmol, 1 equiv), amidophosphite (crude reaction mixture) (5 equiv), urea (3 equiv) and 100 μL of water after 3 days at 80 °C.

The reactions were carried out at 80 °C for 3 days uncapped. The exact mass of the 5’-H-phosphonate is 307.0337 m/z [M-H]⁻, found 307.0401 m/z [M-H]⁻, and the exact mass of the 2’ or 3’,5’-H-bisphosphonate is 371.0051 m/z [M-H]⁻, found 371.0113 m/z [M-H]−.

Unexpectedly, we observed the signal (³¹P-NMR) that corresponds to 2’,3’-cUMP in 20.02 ppm which was also corroborated by the ¹H-NMR spectra 7.55 ppm (Figs. S85 and S86). Except for cytidine, both pyrimidines and purines showed the formation of the 2’,3’-cyclic phosphate product based on the ³¹P-NMR {H-decoupled} (Figs. S91, S96, S101 and S106).

A possible explanation is the after the 2’- or 3’-H-phosphonate is formed, it is followed by an oxidation-cyclization step yielding the uridine cyclic phosphate. We can assume that this behavior is due to the role of urea, because in the uridine and MAP reaction, the 3’-UMP was heated under the same conditions just in presence of urea, we observed the cyclization (Figs. S62 and S63). As indicated above deoxyribonucleosides also were investigated. Again, deoxyguanosine was the only nucleoside with low conversion (around 27%), while the average conversion for other deoxyribonucleosides was higher (up to 75%).

We observed the 5’-and the 3’-H-phosphonates, as well the 3’,5’-bis-H-phosphonates (Figs. S110–S125). Similarly, when urea was absent the phosphonylation reactions did not work well. For example, uridine reacting with amidophosphite and pyrophosphite gave less than 12% of conversion (Figs. S126 and S127). All of the above observations prompted the question whether we can carry a one-pot reaction by mixing phosphite, ammonia source (ammonium hydroxide), urea, uridine with air as oxidant at 80 °C to test whether direct the H-phosphonylation and phosphorylation can be achieved. And when we carried out this reaction we identified 5’- (20%), 3’- (22%), 2’-H-phosphonate (5%), and 2’,3’-cUMP (13%)—for a total H-phosphonylation and phosphorylation of 61% (Figs. S128 and S129). This again suggests the enhancement of phosphonylation is possible in a single reaction in the presence of N-containing additives (such as urea and ammonia). Overall, the reduced P-compounds, amidophosphite, phosphite and pyrophosphite provided better conversion to nucleotides when compared to the phosphorylation results from the corresponding oxidized versions of MAP, Pi and PPi from Table 1.

Discussion

The P-N derivatives of orthophosphates such as DAP (diamidophosphate) and MAP (monoamidophosphate) are significant to expanding the plausible prebiotic phosphorylation reactions on the early Earth, however, plausible geochemical pathways to form DAP and MAP and other P-N compounds remain scarce. In fact, the only prebiotic plausible pathway to form DAP and other amino derivatives of phosphates/phosphite is observed when Fe₃P (the mineral schreibersite), condensed phosphates (such as P₄O₁₀) and reduced P compounds (phosphite) that could be available on the Earth are exposed to aqueous ammonia solutions²⁶.

The schreibersite ammonolysis pathway is strictly dependent on pH, concentration of aqueous ammonia and anaerobic conditions²⁶, and requires time (between 1–2 weeks) to form P-N compounds. Nonetheless, the efficient conversion of P species, including phosphate and phosphite into their N-derivatives (e.g., DAP, MAP and amidophosphite) motivated us to find more energetic, facile and efficient pathways to form reactive P-N species. The proposed mechanism of the reaction involves important steps such as ammonolysis, amidation and oxidation of various P-O and P-N species (Scheme S2).

In our study, the total incorporation of nitrogen in the P-N species calculated from ³¹P-NMR was found to exceed or match the percentage of dissolved nitrogen available in the ammoniacal solution²⁶. These reactions would have been plausible on the early Earth due to the ubiquity of the reduced P sources and ammonia (see discussion below). Furthermore, given the role of ammonia-water mixtures in building moons in the outer solar system^40,41, nitrogenous P species could have played significant roles in the origin of life on those bodies⁴².

We are currently testing the possibility that other ammonia sources—namely ammonium carbonate—could serve to form P-N compounds. In the current study, we employed pyrophosphite [PPi(III)], as a starting material. This compound is an analog of pyrophosphate, but has only two ionizable protons compared with the four in pyrophosphate (PPi(V)]³². It has also been suggested as an important energy currency molecule and an important prebiotic phosphorylating agent in prebiotic chemistry⁴³. Pyrophosphite is readily formed from the dehydration of phosphite. Phosphite readily leaches from calcium phosphite (contrast to the lack of leaching of phosphate from calcium phosphates), which can be sourced from natural glasses called fulgurites⁴⁴. Similarly, a significant source of phosphite would have been supplied to the early Earth by meteorites during the heavy bombardment period^45–47. Moreover, considering Hadean ocean was iron rich^48,49, iron phosphite could possibly have been a precipitate³⁴.

The generation of P-N compounds was almost immediate and once amidophosphite is formed, the formation of MAP and DAP proceeds smoothly (see Scheme S2). We note that most of the experiments explored above focus not on prebiotic chemistry but on establishing the synthesis and evolution pathways of these molecules. We investigated several individual parts of these reactions to establish if these reactions may be broadly relevant to prebiotic chemistry. For one, the reactions work without O₂, by using various oxidants such as peroxide and hypochlorite that could be relevant to the early Earth. Hydrogen peroxide could possibly have existed on the early Earth (and other rocky planets)³³ possibly by the photolysis of atmospheric water^50,51 or ice^52–54. Other abiotic pathways in the Archean involve the abrasion of quartz surfaces and the formation of certain surface-bound radicals that could possibly oxidize H₂O to H₂O₂ and O₂⁵⁵. Likewise, hypochlorite is known to exist on the rocky planets such as Mars. Studies have shown the formation of hypochlorite by gamma rays mediated decomposition of perchlorates in a CO₂ atmosphere⁵⁶.

This suggests that MAP and DAP form under anoxic conditions in the presence of oxidants that could have been widespread on the early Earth before the Great Oxygenation Event, e.g., ca. 2.5 billion years ago⁵⁷. In lieu of oxidants, ultraviolet radiation may have promoted MAP and DAP from amidophosphite and its precursors. It is generally held that the early Sun produced a far larger amount of radiation in the UV region than at present. Consequently, the early Earth surface was exposed to both UVC radiation (<280 nm) and higher doses of UVB (280–315 nm) compared with the present terrestrial surface^58–62. This suggests a broad wavelength range for the prebiotic photochemical reactions to be inherently nonlinear⁵⁹. Our experiments used a 365 nm wavelength UV light source and show that photochemical reactions produce the phosphorylating agents of from reduced P.

We employed high concentrations of ammonia at high pH in these experiments and found P-N production directly related to the concentrations of ammonia. We tested lower ammonia concentrations (and correspondingly lower pHs), finding that P-N bonds do form, but at a lower total fraction due to less hydrolysis of pyrophosphite. Hence for these reactions to be prebiotically relevant, higher ammonia would be necessary. As examples, early ocean hydrothermal systems could have sustained sources of ammonia⁶³. Ammonia was also likely supplied to the early Earth by meteorites. The meteorite Grave Nunataks (GRA) 95229 releases free ammonia in excess upon hydrothermal treatment⁶⁴. In spite of these scenarios, there could be reasonable objection that ammonia may not be available at such high concentrations in a prebiotic environment. By the same token, urea (or close relatives such as cyanamide etc.), which is widely used for prebiotic phosphorylation with inorganic phosphates, could be subject to the same objections, since urea (or cyanamide) is derived from reduced nitrogen sources. If the prebiotic availability of urea or cyanamide (in the concentrations needed) were deemed necessary, then it is not unreasonable to expect that other reduced N-species (such as ammonia) may have been present as well. The mono P-N compounds generated in this study were shown to phosphorylate and H-phosphonylate nucleosides to form their corresponding nucleotides and H-phosphonates in the presence of urea, with good conversion yields (average of 54% and 69% respectively), and did so even in the absence of urea.

The significance of this observation is twofold. First, this P-transfer efficiency has been achieved with a mono P-N species such as monoamidophosphate and monoamidophosphite, which has to be contrasted with the results from diamidophosphate which has two P-N bonds³⁶. Second, the conversion yields with the mono P-N compounds is an improvement when compared with the corresponding P-O compounds (average conversion of 44% and 50% respectively in this study and other investigations)^29,35,36. Furthermore, it is only the P-N containing compounds that show any significant phosphorylation in the absence of activators (such as urea), which is not surprising since the P-N bond itself is activated for transferring the P-moiety to the acceptor (here the nucleoside)—again highlighting the need for co-occurrence of P- and reduced species for prebiotic phosphorylation chemistries.

The above observations suggest that P-N compounds and the chemistry of P-N bonds in the prebiotic context can offer solutions that need to be explored and exploited as to how they could have helped in (or, at the very least expanded) the transition from abiotic phosphorylation processes to the types of phosphorylation chemistries that are amenable to biochemical pathways. In this context, it has been pointed out³⁹ that extant biology still utilizes mono P-N bonds for phosphorylation in various contexts such as energy currency (converting ADP to ATP) and functional regulation (enzyme catalysis).

The prebiotic geochemical origin of P–N compounds remain largely unexplored, and application of such compounds from the context of prebiotic phosphorylation chemistry remains somewhat underutilized even though these compounds have shown tremendous promise in overcoming the so-called “water problem” in prebiotic chemistry^2,4,11. We have attempted to identify routes to P-N molecule formation, and have contrasted P-N molecule reactivity with PO molecules. We note that this does not diminish prior phosphorylation experiments with phosphate and urea⁶⁵, but serves to provide new and expanded routes to phosphorylation that may be more relevant with a broader repertoire of P compounds.

We hereby have shown a route to form inorganic P-N compounds of potential prebiotic relevance (specifically MAP), utilizing various sources of potentially prebiotically relevant P compounds under early Earth conditions, and have also demonstrated the phosphorylation and phosphonylation capabilities of nucleosides by using such P-N compounds under early Earth conditions. We focused exclusively on nucleosides, but other possible prebiotic molecules can be phosphorylated/phosphonylated as well. While the organophosphates (nucleotides) in our study are indeed of direct relevance to the prebiotic formation of the oligomers for RNA and possibly DNA, their organophosphite derivatives also hold significance, as it is likely that such phosphite derivatives would have faced conducive oxidation mechanisms on the early Earth to have converted into their phosphate counterparts⁶⁶.

Methods

Starting P sources did not show any impurities (Figs. S1 and S2). Pyrophosphite [PPi (III)] was synthesized from phosphite (Figs. S3 and S4). Phosphite was also sourced from geologically reasonable materials such as CaHPO₃, Fe₃P and FeHPO₃ as well as the Fenton chemistry of hypophosphite to investigate effects of counterions on ammonolysis (SI, Experimental procedures 3.0 and 3.1). Heating an aqueous solution of a phosphite source in the presence of additives such as urea at 80 °C, formed pyrophosphite (SI, Experimental procedure 2.0).

Pyrophosphite was ammonolyzed at room temperature and sealed under air (SI, Experimental procedure 2.1). The generation of the reduced P-N species amidophosphite and its subsequent oxidation to oxidized P-N compounds including MAP and DAP were studied on timescales from hours to weeks and was monitored by using ³¹P-NMR, where the splitting patterns of the P-N molecules were observed by utilizing ¹⁵N labeled NH₄OH (SI, Experimental procedure 2.3). MAP and DAP were further confirmed by spiking with authentic samples (SI, Experimental procedure 2.2). Furthermore, mechanistic studies of ‘O’ insertion into amidophosphite to form MAP and DAP were investigated by use of isotopically labeled ¹⁸O water (SI, Experimental procedure 2.4). The effects of NH₄OH concentration (SI, Experimental procedure 2.5), pH (SI, Experimental procedure 2.6) varied oxidants, (anoxic) oceanic conditions, anoxic vs. oxic conditions, UV light (245 nm and 365 nm) was also monitored to understand the formation of P-N

compounds and other reaction products (SI, Experimental procedures 3.0–3.7). Blank reactions comparing the ammonolysis and hydrolysis reactions of pyrophosphite were also studied (SI, Experimental procedure 2.1). Furthermore, ammonolysis of pyrophosphate was also attempted (SI, Experimental procedure 3.8).

The yields (%) of various inorganic P-O, P-O-P, P-N and P-H compounds and the progression of the inorganic reactions were based on the dissolved P in the solution and by the peak integration method in the ³¹P-NMR as previously described³⁰. The ³¹P-NMR was acquired in both H-coupled and decoupled modes. Inorganic P-N, P-O, P-H, and P-O-P species were identified based on their chemical shift values and the peak splitting patterns (if any).

Prebiotic phosphorylation and prebiotic H-phosphonylation reactions of canonical nucleosides (both deoxyribose and ribonucleosides) were also assessed using MAP, DAP, phosphate and pyrophosphate, and phosphite, pyrophosphite, and amidophosphite in the presence of urea at 80 °C, under a ‘warm (pH = 5–13) evaporating pool’s scenario (Experimental procedures 4.1–4.9). The reactions were monitored by ¹H and ³¹P-NMR and in some cases, by mass spectrometry (ESI as described previously)⁸. Percent conversion (%) to the respective nucleotide(s) was obtained by comparing the integration of the signal in ¹H-NMR for the starting nucleoside and the corresponding phosphorylated and phosphonylated products using deconvolution for resolving overlapping peaks where required⁸. For the pyrimidines, the H−C(6) of the nucleobase, while for the purines, the anomeric H−C(1′) signals were used, as described in previous studies⁸.

Author contributions

M.G. designed and performed the inorganic synthetic reactions of the P-N compounds and their analyses as well as prepared the draft, H.A.C. studied the H-phosphonylation and phosphorylation reactions of nucleosides with various inorganic P-N species and other P compounds and their chemical analyses as well prepared the draft, R.K. and M.A.P. conceived the idea, acquired funding, supervised the research and prepared the draft.

Peer review

Peer review information

Communications Chemistry thanks the anonymous reviewers for their contribution to the peer review of this work. Peer review reports are available.

Data availability

All the experimental including ³¹P-NMR {H-decoupled and H-coupled} of crude reactions of pyrophosphite, ¹H-NMR and ³¹P-NMR spectra of crude reactions of phosphorylation and phosphonylation, and mass spectra of the crude reaction (PDF) were provided in the ESI. Data availability. Reasonable requests for the raw data files in another format should be directed to a corresponding authors.

Competing interests

The authors declare no competing interests.

Supplementary information

The online version contains supplementary material available at 10.1038/s42004-025-01577-0.