Thiophosphate photochemistry enables prebiotic access to sugars and terpenoid precursors

Over the past few years, evidence has accrued that demonstrates that terrestrial photochemical reactions could have provided numerous (proto)biomolecules with implications for the origin of life. This chemistry simply relies on UV light, inorganic sulfur species and hydrogen cyanide. Recently, we reported that, under the same conditions, reduced phosphorus species, such as those delivered by meteorites, can be oxidized to orthophosphate, generating thiophosphate in the process. Here we describe an investigation of the properties of thiophosphate as well as additional possible means for its formation on primitive Earth. We show that several reported prebiotic reactions, including the photoreduction of thioamides, carbonyl groups and cyanohydrins, can be markedly improved, and that tetroses and pentoses can be accessed from hydrogen cyanide through a Kiliani–Fischer-type process without progressing to higher sugars. We also demonstrate that thiophosphate allows photochemical reductive aminations, and that thiophosphate chemistry allows a plausible prebiotic synthesis of the C5 moieties used in extant terpene and terpenoid biosynthesis, namely dimethylallyl alcohol and isopentenyl alcohol.


Supplementary Tables
Supplementary Table 1  5 Supplementary  In an Eppendorf tube, Na3PSO3.xH2O (the purity and water content were predetermined and accounted for, 1 equiv. or 1.5 equiv.) was dissolved with degassed 10% D2O in H2O (1 mL) and the pH was adjusted to 6.5 with degassed HCl. The volume was made up to 2 mL with degassed 10% D2O in H2O, glycolonitrile 1 (0.040 mmol, 4.0 µL) was added and the solution was transferred to a quartz cuvette and sealed. The reaction was irradiated for the desired amount of time after which it was analysed by 1 H NMR spectroscopy.

d/ppm
Supplementary Fig. 2 Photochemic al reduction of glycolo nitrile 1 using thiophosphate. A -1 H NMR Spectrum of the reaction according to Procedure 1 (1 equiv. Na3PSO3) after 1 h irradiation; B -1 H NMR Spectrum of the reaction according to Procedure 1 (1.5 equiv. Na3PSO3) after 1 h irradiation; C -As spectrum A, but NaSH used in place of Na3PSO3. The notation X(h) signifies the hydrate of an aldehyde, X. There is a similar amont of acetaldehyde 3 and lactonitrile 5 in spectrums A and B, but spectrum C is almost devoid of these compounds. In spectrums A and B a small amount of EtOH (annotated in spectrum B) has started to form (~ 2% yield in both spectra), but there is a signal almost superimposed on the EtOH triplet at 1.10 ppm which corresponds to isopropanol S1 (d, J = 6.2, 6H). We later identified that this came from the reduction of acetone, an impurity in several commercial sources of 1, see Supplementary Fig. 3. A small signal for the residual acetone can be seen in spectrums A and

3(h) OH
OH OH 2(h) 6 OH HO B, just to the right of acetaldehyde 3 (at 2.15 ppm), and clearly in spectrum C where reduction to S1 has not yet taken place. The signals at 3.30 ppm and 2.44 ppm are due to MeOH and thioacetamide, respectively (although thioacetamide can be further reduced to acetaldehyde 3 the reduction of thioamides takes place much more slowly than cyanohydrins (see Supplementary Fig. 48), hence we believe the production of 3 comes mainly from deoxygenation of 2, see also Supplementary Fig. 4). Yields estimated by addition of disodium fumarate (50 mM solution, 50 µL into 450 µL of crude reaction) as a standard and relative integration of 1 H NMR signals. trace -a yield of 1% or less a -yield of 3 constitutes carbonyl + hydrate d/ppm Supplementary Fig. 3 Photochemical reduction of commercial glycolonitrile 1 or 1 formed in situ. A -Formaldehyde ~ 37% (0.040 mmol, 3.0 µL) and KCN (0.043 mmol, 2.8 mg) were dissolved in degassed 10% D2O in H2O (1 mL) and Na3PSO3 (1 equiv.) was added. The pH was adjusted to 6.5, volume made up to 2 mL with degassed 10% D2O in H2O then the solution was irradiated for 1 h before a 1 H NMR spectrum (spectrum A) of the crude reaction was acquired; B -1 H NMR Spectrum of the reaction according to Procedure 1 (1 equiv. Na3PSO3) after 1 h irradiation; C -Acetone (0.020 mmol, 1.5 µL) and Na3PSO3 (0.040 mol, 8 mg) were dissolved in degassed 10% D2O in H2O (1 mL) and Na3PSO3 (1 equiv.) was added. The pH was adjusted to 6.5 and the volume made up to 2 mL with degassed 10% D2O in H2O before being irradiated for 1 h. A 1 H NMR spectrum of the crude reaction (spectrum C) was then acquired.

Supplementary
In the reduction of 1 which was made 'in house' (spectrum A) no isopropanol S1 was observed after 1 h reaction (large singlet at 3.30 ppm is due to MeOH in commercial formaldehyde) whereas in the reduction of the commercial sample which contained acetone (spectrum B) the characteristic doublet of S1 (1.10 ppm) is present. The photochemical reduction of acetone (spectrum C) using Na3PSO3 is quantitative, giving S1 as the sole product.  6 OH HO d/ppm Supplementary Fig. 5 Extended photochemical reduction of glycolonitrile 1 using thiophosphate. A -1 H NMR Spectrum of the reaction according to Procedure 1 (1 equiv. Na3PSO3) after 2 h irradiation cf. Supplementary Fig. 2, spectrum A. In this spectrum, glycolaldehyde 2 was present in ~ 20% yield, glyceronitrile 4 in ~ 8% yield, acetaldehyde 3 (hydrate and aldehyde) in ~ 5% yield, lactonitrile 5 in ~ 1% yield, ethylene glycol 6 in ~ 1% yield and EtOH in ~ 3% yield; B -1 H NMR Spectrum of the reaction according to Procedure 1 (1.5 equiv. Na3PSO3) after 2 h irradiation, cf. Supplementary Fig. 2, spectrum B. In this spectrum 2 had been produced in ~ 22% yield, 4 in ~ 7% yield, 3 in ~ 6% yield, 5 in ~ 1% yield, 6 in ~ 2% yield and EtOH in ~ 6% yield (cf. ~ 2% yield of EtOH after 1 h irradiation, see main text, Supplementary Fig. 2 and Supplementary Table 1). The signals at 3.30 ppm and 2.44 ppm are due to MeOH and thioacetamide, respectively (although thioacetamide can be further reduced to acetaldehyde 3 the reduction of thioamides takes place much more slowly than cyanohydrins (see Supplementary Fig. 48), hence we believe the production of 3 comes mainly from deoxygenation of 2, see Supplementary Fig. 4). Yields estimated by addition of disodium fumarate (50 mM solution, 50 µL into 450 µL of crude reaction) as a standard and relative integration of 1 H NMR signals.

3(h)
OH OH OH 2(h) 6 OH HO d/ppm Supplementary Fig. 6 Reduction of glycolonitrile 1 in the presence and absence of PO4 3-. A -As Supplementary Fig. 2, Spectrum A ( 1 H NMR Spectrum of the reaction mixture after 1 h irradiation following Procedure 1); B -As Spectrum A but with the inclusion of NaH2PO4 (20 mM). Interestingly, in the presence of phosphate buffer (spectrum B) formation of thioacetamide (singlet at 2.42 ppm) was suppressed. Yields estimated by addition of disodium fumarate (50 mM solution, 50 µL into 450 µL of crude reaction) as a standard and relative integration of 1 H NMR signals -see Supplementary Table 1 Fig. 7 Attempted one pot synthesis of glyceraldehyde 7 from glycolonitrile 1. A -1 H NMR Spectrum of the crude reaction mixture after 1 h reaction according to Procedure 1 using 2 equivalents of Na3PSO3 and with the addition of KCN (0.040 mmol, 2.6 mg, 1 equiv.), initial pH of the reaction was 6.5. Yields estimated by addition of disodium fumarate (50 mM solution, 50 µL into 450 µL of crude reaction) as a standard and relative integration of 1 H NMR signals (see Supplementary Tables 2.1 and 2.2); B -As spectrum A, spiked with serine nitrile 8, made as follows: NH4Cl (54 mg, 1.00 mmol) was dissolved in H2O (0.7 mL) then glycolaldehyde 2 (6 mg, 0.100 mmol) and KCN (8 mg, 0.120 mmol) were added. The pH was adjusted to 9.1 and D2O (0.1 mL) added and volume adjusted to 1.0 mL with H2O. The solution was left at room temperature for 4 days; C -1 H NMR Spectrum of 8.  Glyceronitrile 4 Lactonitrile 5

Serine nitrile 8
Glyceraldehyde 7 Glyceraldehyde cyanohydrin 10 Glycerol 11 Ethylene glycol 6 trace trace a -yields based on carbon feedstock(s) trace -a yield of 1% or less d/ppm Supplementary Fig. 8 One pot synthesis of glyceraldehyde 7 from glycolonitrile 1 and HCN. A -A solution of KCN (0.040 mmol, 2.6 mg), NaH2PO4.2H2O (0.120 mmol, 19 mg) and Na3PSO3 (the purity and water content were predetermined and accounted for, 0.110 mmol) were dissolved in degassed 10% D2O in H2O (1 mL) and pH adjusted to 6.5. Glycolonitrile 1 (0.030 mmol, 3.0 µL) was added, the volume made up to 2 mL with degassed 10% D2O in H2O then the solution irradiated. After 1 h a 1 H NMR spectrum of the crude reaction mixture (spectrum A) was acquired; B -As spectrum A, spiked with a commercial sample of 7; C -As spectrum B, spiked with a commercial sample of glycolaldehyde 2; D -As spectrum C, spiked with an authentic sample of glyceraldehyde cyanohydrin 10; E -As spectrum D, spiked with a commercial sample of glycerol 11. Yields estimated by addition of disodium fumarate (50 mM solution, 50 µL into 450 µL of crude reaction) as a standard and relative integration of 1 H NMR signals. Yields based on maximum product concentration possible from HCN + 1, due to the fact that HCN can also be reduced and homologated to 1, thus the maximum concentration of C2 and C3 compounds are 25 mM and 16.7 mM, respectively. One pot synthesis of glyceraldehyde 7 from glycolonitrile 1 and HCN and conversion to dihydroxyacetone 12. A -As Supplementary Fig. 8, spectrum A; B -1 H NMR Spectrum of an authentic sample of glyceraldehyde cyanohydrin 10; C -To a portion (1.5 mL) of the reaction mixture from which spectrum A was taken, was added H2O (4.5 mL) and N2 bubbled gently through the solution for 48 h (final volume ~ 2 mL). A 1 H NMR Spectrum (spectrum C) was then acquired; D -1 H NMR Spectrum of a commercial sample of 7; E -As spectrum D after being left for 2 weeks at room temperature. Although there appears to be far more dihydroxyacetone 12 than 7 in spectrum E, the signal at 4.88 ppm (doublet) from 7 is attenuated by the NMR experiment which suppresses the H2O/HOD signal. The yield was calculated by addition of an external standard and relative integration of the 1 H NMR signals (in the case of 7(h), the signals at 3.67 ppm gave clear integration). Yields based on maximum product concentration possible from HCN + 1, due to the fact that HCN can also be reduced and homologated to 1, thus the maximum concentration of C2 and C3 compounds are 25 mM and 16.7 mM, respectively. Fig. 10 One pot synthesis of glyceraldehyde 7 from HCN. A solution of KCN (7 mg, 0.100 mmol), Na3PSO3.xH2O (the purity and water content were predetermined and accounted for, 0.040 mmol) and NaH2PO4.2H2O (16 mg, 0.100 mmol) in degassed 10% D2O in H2O (2 mL) at pH 6.5 were irradiated for 1.25 h and a 1 H NMR spectrum of the crude reaction was acquired. To assist identification of 7, the remaining 1.55 mL of the crude reaction was adjusted to pH 7.0, diluted to 8 mL with H2O and N2 bubbled gently through the solution for 4 d (final volume ~ 1 mL) before the crude reaction examined by 1 H NMR spectroscopy (spectrum B). C -A 1 H NMR spectrum of a commercial sample of 7; D -A 1 H NMR spectrum of a commercial sample of glycolaldehyde 2. In spectrum A there was ~ 6% of glyceronitrile 4, ~ 3% of serine nitrile 8, ~ 5% of glyceraldehyde cyanohydrin 10, based on HCN. The maximum possible yield of 2 or 7 is equivalent to half, or one third, the molarity of the initial HCN concentration, respectively. Yields estimated by addition of disodium fumarate (50 mM solution, 50 µL into 450 µL of crude reaction) as a standard and relative integration of 1 H NMR signals.  Fig. 11 Retention of glycolonitrile 1 after evaporation. To a solution of 1 (55% wt., 20.0 µL, 0.100 mmol) in D2O was added disodium fumarate solution (50 mM, 200 µL) and the pD was altered to ~ 7, then a 1 H NMR spectrum (spectrum A) was acquired without solvent suppression. The volume was adjusted to 1 mL then N2 was bubbled through the solution at ambient temperature overnight until a solid residue remained. The residue was dissolved in D2O and a 1 H NMR spectrum of the crude mixture (spectrum B) was acquired. In spectrum A, integration of fumarate:1 is 1:7.7. In spectrum B, integration of fumarate:1 is 1:0.8, ~ 10% of 1 remains. Also see Supplementary Fig. 10, spectrum B, where ~ 90% of H2O had been evaporated, but 1 remained in approximately the same concentration as before evaporation of H2O. Fig. 12 One pot synthesis of the tetroses 13 from glycolonitrile 1 and HCN. A -A solution of KCN (0.040 mmol, 2.6 mg), NaH2PO4.2H2O (0.120 mmol, 19 mg) and Na3PSO3 (the purity and water content were predetermined and accounted for, 0.110 mmol) were dissolved in degassed 10% D2O in H2O (1 mL) and pH adjusted to 6.5. Glycolonitrile 1 (0.030 mmol, 3.0 µL) was added, the volume made up to 2 mL with degassed 10% D2O in H2O then the solution irradiated. After 1 h, a 1 H NMR spectrum of the crude reaction mixture (spectrum A) was acquired; B -As spectrum A, spiked with a commercial sample of threose 13-t; C -As spectrum B, spiked with a commercial sample of erythrose 13-e; D -1 H NMR Spectrum of a commercial sample of 13-t in phosphate buffer at pH 7.0; E -1 H NMR Spectrum of a commercial sample of 13-e in phosphate buffer at pH 7.0 (a pure commercial sample of 13-e could not be obtained, hence the extraneous signals observed in spectrum E). To quantify the C2 -C4 sugar products, the reaction was repeated and oxime derivatives made to assist integration, see Supplementary Fig. 13. (1 mL) and pH adjusted to 6.5. Glycolonitrile 1 (0.030 mmol, 3.0 µL) was added, the volume made up to 2 mL with degassed 10% D2O in H2O then the solution irradiated. After 1 h a 1 H NMR spectrum of the crude reaction mixture (spectrum A) was acquired; B -To a portion (900 µL) of the crude reaction mixture was added fumarate solution (pH 7, 50 mM, 100 µL) and H2O (8 mL), then N2 was bubbled through the solution for 3 d and spectrum B obtained. This was done in order to remove HCN from the system and allow clean oxime formation; C -To the NMR sample from B was added NH2OH solution (2 M, pH 6.2, 100 µL) and spectrum C obtained; D -As spectrum C, after the addition of an authentic standard of erythrose oxime 13-e(ox); E -As spectrum D, after the addition of an authentic standard of threose oxime 13t(ox); F -As spectrum E, after the addition of an authentic standard of glyceraldehyde oxime 7(ox); G -As spectrum F, after the addition of an authentic standard of glycolaldehyde oxime 2(ox). The singlet at 6.4 ppm in all spectra is due to fumarate, added as an external standard. Assuming no loss of sugars during the removal of HCN, the yields of sugars in spectrum A was ~ 8% glycolaldehyde 2, ~ 8% glyceraldehyde 7 and ~ 19% tetroses 13. Although there is a higher concentration of 2, and 7 and 13 are present in similar concentrations, the maximum concentration of each product varies due to the fact that HCN can also be reduced and homologated to 1, thus the maximum concentration of 2, 7 and 13, is 25 mM, 16.7 mM and 10 mM, respectively. Erythrose 13-e and threose 13-t were formed in 0.9:1 ratio, as determined by the integration of the C2-H signals (4.15 -4.24 ppm). In spectrum A, there was ~ 4% C1 (MeOH + formate), ~ 27% C2 (glycolaldehyde 2, glyceronitrile 4, ethanolamine 9, ethylene glycol 6, acetaldehyde 3 and its addition products 3(h), 3(t) and 5, MeCN, acetate and EtOH), ~ 22% C3 (glyceraldehyde 7 (~ 8%, this was calculated from Supplementary Fig. 13, spectrum C, as the C2-H signal of glyceraldehyde cyanohydrin 10 did not give a clear integration), glyceraldehyde cyanohydrin 10, glycerol 11 and 1,3-dihydroxypropane) and ~ 19% C4 compounds (tetroses 13 and their cyanohydrins 15. By comparison to Spectrum C, which gave good resolution of the NMR spectroscopy signals and a yield of ~ 19% of 13, there was ~ 4% of the tetrose cyanohydrins 15 in spectrum A i.e. ~ 15% of 13. The reduced tetroses (erythritol and threitol), if present, could not be observed in the crude reaction mixture by routine 1 H NMR spectroscopy). Glyceraldehyde 7

C4
Tetroses 13 Tetrose cyanohydrins 15 ~19% Summary of the products identified in each compound class (C1-C4) in Supplementary Fig. 13, spectrum A, and the total yield for all compounds identified in each class. Also see Supplementary Fig. 13, caption. Cyanohydrins were counted in the same class of compounds as the parent aldehyde e.g. glyceronitrile 4 was counted as a C2 compound. This was to avoid misrepresentation of the overall yields for each stage of reduction i.e. an aldehyde belonging to the class of compounds Cn is converted to Cn+1 when cyanide adds in situ, artificially depleting the yield of Cn and inflating the yield of Cn+1 compounds. Glycolonitrile 1 was not included in the products, as it was not possible to know how much was synthesised in situ from HCN. Yields based on 1 and HCN. and Na3PSO3 (the purity and water content were predetermined and accounted for, 0.200 mmol) were dissolved in degassed 10% D2O in H2O (1 mL) and pH adjusted to 6.5. Glycolonitrile 1 (0.015 mmol, 3.0 µL) was added, the volume made up to 2 mL with degassed 10% D2O in H2O then the solution was irradiated. After 2.5 h, a 1 H NMR spectrum of the crude reaction mixture (spectrum A) was acquired; B -As spectrum A, spiked with a commercial sample of ribose 14-r; C -As spectrum B, spiked with a commercial sample of arabinose 14a; D -As spectrum C, spiked with a commercial sample of xylose 14-x; E -As spectrum D, spiked with a commercial sample of lyxose 14-a; F -As spectrum E, spiked with a commercial sample of threose 13-t; G -As spectrum F, spiked with a commercial sample of erythrose 13e. Na3PSO3 (the purity and water content were predetermined and accounted for, 0.200 mmol) were dissolved in degassed 10% D2O in H2O (1 mL) and pH adjusted to 6.5. Glycolonitrile 1 (0.015 mmol, 3.0 µL) was added, the volume made up to 2 mL with degassed 10% D2O in H2O then the solution irradiated. After 2.75 h a 1 H NMR spectrum of the crude reaction mixture (spectrum A) was acquired with fumarate added as a standard; B -To an aliquot (900 µL) of the crude reaction mixture was added fumarate (100 µL, 50 mM), as a standard, and H2O (7 mL), then excess HCN was removed by bubbling N2 through the solution for 2 d, after which the volume was ~ 0.5 mL. A 1 H NMR spectrum of the solution (spectrum B) was then acquired. The yields of 2, 7, 13 and 14 in spectrum B were ~ 2%, 2%, 11% and 29%, respectively. Ribose 14-r, lyxose 14-l, arabinose 14-a and xylose 14-x were present in a ratio of ~ 1:1.2:1.2:1. Due to signals from 14-r and 14-l (~ 4.78 ppm) being in close proximity to the H2O/HOD peak, and that the solvent suppression NMR experiment supressed these signals significantly, coupled with signal overlap between 14-r and the tetroses 13 (5.17 ppm), the yields of 14-r and 14-l were calculated using the peak corresponding to the major cyclic form of each sugar and its relative abundance to the minor forms of each sugar, as observed and measured for the pure sugars in phosphate buffer (pD 7, Supplementary Fig. 15, spectra C and D). The calculated integration for the peak of 14-r at 5.17 ppm could then be deducted from the yield of 13. An added complication in the calculation of these yields comes from the fact that HCN can be reduced under the reaction conditions, and after hydrolysis and addition of further HCN, B C D E F A generates the starting material, glycolonitrile 1. Thus, the maximum possible concentration of 2, 7, 13 and 14 would be 30 mM, 20 mM, 15 mM and 12 mM, respectively, and consequently the yields of each product were calculated from these values. In spectrum A, there was ~ 1% C1 (MeOH + formate), ~ 8% C2 (glycolaldehyde 2, ethanolamine 9, ethylene glycol 6, MeCN, acetate and EtOH), ~ 11% C3 (glyceraldehyde 6, glycerol 11 and 1,3dihydroxypropane), ~ 11% C4 (tetroses 13) and ~ 29% C5 compounds (pentoses 14). Glycolaldehyde 2 and glyceraldehyde 7 yields were calculated from spectrum B, as integration of their cyanohydrins in spectrum A was not possible. As solvent suppression was used to acquire the 1 H NMR spectra, these yields are thought to be conservative due to some attenuation of the signals most proximal to the HOD/H2O peak; C -1 H NMR spectrum of an authentic sample of ribose and NaD2PO4 in D2O at pD 7.0; D -1 H NMR spectrum of an authentic sample of lyxose and NaD2PO4 in D2O at pD 7.0; E -1 H NMR spectrum of an authentic sample of arabinose and NaD2PO4 in D2O at pD 7.0; F -1 H NMR spectrum of an authentic sample of xylose and NaD2PO4 in D2O at pD 7.0.

C4
Tetroses 13 ~11% C5 Pentoses 14 ~29% Summary of the products identified in each compound class (C1-C5) in Supplementary Fig. 15, spectrum A, and the total yield for all compounds identified in each class. Also see Supplementary Fig. 15, caption. Cyanohydrins were counted in the same class of compounds as the parent aldehyde e.g. glyceronitrile 4 was counted as a C2 compound. This was to avoid misrepresentation of the overall yields for each stage of reduction i.e. an aldehyde belonging to the class of compounds Cn is converted to Cn+1 when cyanide adds in situ, artificially depleting the yield of Cn and inflating the yield of Cn+1 compounds. Glycolonitrile 1 was not included in the products, as it was not possible to know how much was synthesised in situ from HCN. Yields based on 1 and HCN. (1 mL) and pH adjusted to 6.5. The volume was made up to 2 mL with degassed 10% D2O in H2O, then the solution irradiated. After 3 h, a 1 H NMR spectrum of the crude reaction mixture (spectrum A) was acquired; B -As spectrum A, spiked with commercial arabinose 14a; C -As spectrum B, spiked with commercial ribose 14-r; D -As spectrum C, spiked with commercial lyxose 14-l; E -As spectrum D, spiked with commercial xylose 14-x; F -As spectrum E, spiked with commercial threose 13-t; G -As spectrum F, spiked with commercial erythrose 13-e; H -From the crude reaction mixture, a portion (900 µL) was removed and fumarate solution (pH 7, 50 mM, 100 µL) and H2O (8 mL) were added. N2 Was bubbled through the solution for 2 d to remove HCN, then spectrum H was obtained. Yields of the sugars were then determined. In spectrum H, there was ~ 1% glycolaldehyde 2, ~ 2% glyceraldehyde 7, ~ 5% tetroses 13, ~ 19% pentoses 14 (14-r:14-a:14-l:14-x, 1:1.3:1.3:1.1). Due to signals from 14-r and 14-d (~ 4.78 ppm) being in very close proximity to the H2O/HOD peak, and that the solvent suppression NMR experiment supressed these signals significantly, coupled with signal overlap between 14-r and the tetroses 13 (5.17 ppm), the yields of 14-r and 14-l were calculated using the peak corresponding to the major cyclic form of each sugar and its relative abundance to the minor forms of each sugar, as observed and measured for the pure sugars in phosphate buffer (pD 7, Supplementary Fig. 15, spectra C and D). The calculated integration for the peak of 14-r at 5.17 ppm could then be deducted from the yield of 13. As solvent suppression was used to acquire the 1 H NMR spectra, these yields are thought to be conservative due to some attenuation of the signals most proximal to the HOD/H2O peak. Although the signals in the spectra look comparable in intensity, five cyanide molecules are required to form the pentoses 14, thus the maximum yield is one fifth the initial concentration of HCN. The same considerations are taken for the other sugars. In spectrum H, there was ~ 3% C1 (formate), ~ 9% C2 (glycolaldehyde 2, ethanolamine 9, ethylene glycol 6, thioacetamide, MeCN, lactonitrile 5, acetaldehyde hydrate 3(h), acetate and EtOH (MeCN, lactonitrile 5, acetaldehyde hydrate 3(h) and EtOH calculated from spectrum A), ~ 7% C3 (glyceraldehyde 7, glycerol 11 and 1,3-dihydroxypropane), ~ 6% C4 (tetroses 13) and ~ 19% C5 compounds (pentoses 14).

d/ppm
Supplementary Fig. 17 A -A solution of KCN (0.090 mmol, 6 mg), NaH2PO4.2H2O (0.150 mmol, 23 mg) and Na3PSO3 (the purity and water content were predetermined and accounted for, 0.200 mmol) were dissolved in degassed 10% D2O in H2O (1 mL) and pH adjusted to 6.5. Glycolonitrile 1 (0.015 mmol, 3.0 µL) was added, the volume made up to 2 mL with degassed 10% D2O in H2O then the solution irradiated. After 2 h 40 min, a portion (1 mL) was removed, diluted with H2O (7 mL) then N2 bubbled through for 2 d, after which the volume was ~ 1 mL. An aliquot was removed, D2O added and a 1 H NMR spectrum (spectrum A) acquired; B -A 1 H NMR spectrum of a roughly equimolar mixture of glycolaldehyde 2, glyceraldehyde 7, threose 13-t (erythrose 13-e was not included due to the high concentration of impurities present in commercial samples, 30 -40%), the pentoses 14, ethylene glycol 6 and glycerol 11 in NaD2PO4 buffer in pure D2O at pD 7; C -A solution of KCN (0.110 mmol, 7 mg), NaH2PO4.2H2O (0.150 mmol, 23 mg) and Na3PSO3 (the purity and water content were predetermined and accounted for, 0.200 mmol) were dissolved in degassed 10% D2O in H2O (1 mL) and pH adjusted to 6.5. The volume was made up to 2 mL with degassed 10% D2O in H2O, then the solution irradiated. After 3 h, a portion (1 mL) was removed, diluted with H2O (7 mL) then N2 bubbled through for 2 d, after which the volume was ~ 1 mL. An aliquot was removed, D2O added and a 1 H NMR spectrum (spectrum C). Disregarding relative intensities, the majority of signals are accounted for by the mixture represented by spectrum B. Ethanolamine 9 was identified subsequently, the corresponding triplet is not clearly visible but is found at 3.63 ppm. Na3PSO3 (the purity and water content were predetermined and accounted for, 0.200 mmol) were dissolved in degassed 10% D2O in H2O (1 mL) and pH adjusted to 6.5. Glycolonitrile 1 (0.015 mmol, 3.0 µL) was added, the volume made up to 2 mL with degassed 10% D2O in H2O then the solution irradiated. After 2.5 h, a 1 H NMR spectrum of the crude reaction mixture (spectrum A) was acquired; B -As spectrum A, after spiking with a commercial sample of galactose; C -As spectrum B, after spiking with a commercial sample of glucose; D -As spectrum C, after spiking with a commercial sample of mannose. As there was effectively no selectivity for the diastereoselective formation of one tetrose or pentose over another, it was assumed no selectivity would be observed in the potential formation of hexoses, hence glucose, galactose and mannose were used as representative examples.
Supplementary Fig. 19 Competition experiment of a C2, C4 and C5 sugar for HCN. To facilitate observation and quantification of the system, glyceraldehyde 7 was omitted from the mixture and only one tetrose and one pentose were added (threose 13-t and xylose 14-x were chosen as they gave resolved signals in the 1 H NMR spectrum of the mixture). A -NaD2PO4 (ca. 0.200 mmol, 26 mg) Was dissolved in D2O (1.4 mL) and the pD adjusted to 6.6 with NaOD. Glycolaldehyde 2 (0.100 mol, 6 mg), 13-t (≥ 60%, 0.100 mol, 20 mg), 14-x (0.100 mmol, 15 mg) and sodium succinate (added as an internal standard, 0.050 mmol, 8 mg, not shown) were added and a 1 H NMR spectrum acquired (spectrum A); B -As spectrum A, 15 min after the addition of KCN (0.100 mmol, 6 mg), re-adjustment of the pD to 6.9 and the volume made up to 2 mL with D2O; C -As spectrum B after 16 h. Using succinate as a reference, in spectrum A the integration of 2:13-t:14-x was 1  Fig. 20 Pentoses 14 remain after evaporation to dryness. A -A solution of KCN (0.090 mmol, 6 mg), NaH2PO4.2H2O (0.150 mmol, 23 mg) and Na3PSO3 (the purity and water content were predetermined and accounted for, 0.200 mmol) were dissolved in degassed 10% D2O in H2O (1 mL) and pH adjusted to 6.5. Glycolonitrile 1 (0.015 mmol, 3.0 µL) was added, the volume made up to 2 mL with degassed 10% D2O in H2O then the solution irradiated. After 2.75 h an aliquot (450 µL) of the reaction was removed and succinate (pH 7, 50 mM, 50 µL) was added, then a 1 H NMR spectrum was acquired (spectrum A); From the remaining reaction, a portion (900 µL) was removed, succinate (pH 7, 50 mM, 100 µL) added and the sample was diluted with H2O (~ 7 mL). N2 Was bubbled through the solution until a dry residue remained (~ 3 d). The residue was dissolved in D2O and spectrum B was acquired; C -As spectrum B, spiked with a commercial sample of ribose 14-r; D -As spectrum C, spiked with a commercial sample of arabinose 14-a; E -As spectrum D, spiked with a commercial sample of lyxose 14-l; F -As spectrum E, spiked with a commercial sample of xylose 14-x. Competition experiment showing the stability of pentoses 14 to photochemical reduction. The system was simplified to include only glycolaldehyde 2, threose 13-t and xylose 14-x, as they gave resolved signals in the 1 H NMR spectrum of the mixture and eased quantification. Glycolaldehyde 2 (18 mg, 0.300 mmol) and xylose 14-x 45 mg, 0.300 mmol) were dissolved in H2O (0.5 mL) and threose 13-t syrup (≥ 60%) was added until roughly equimolar (as judged by removing several µL, diluting with D2O and obtaining a 1 H NMR spectrum). Sodium succinate (10 mg, 0.062 mmol) was added as a standard and the volume adjusted to 2 mL with H2O. From this stock solution, a portion was removed (0.5 mL), NaH2PO4.2H2O (31 mg, 0.200 mmol) and Na3PSO3.xH2O (purity and water content predetermined and accounted for, 0.075 mmol) were added, then the pH was adjusted to ~ 7 and the volume increased to 2 mL. The solution was irradiated and inspected at timepoints by 1 H NMR spectroscopy. A -1 H NMR Spectrum of the starting mixture, singlet at 2.3 ppm due to the added succinate standard; B -As spectrum A, after 0.5 h of irradiation; C -As spectrum A, after 1 h of irradiation. Note the large singlet in spectra B and C at 3.6 ppm, which is due to ethylene glycol 6. In spectrum A, the relative integration of 2:13-t:14-x (referenced to succinate) was 3.4:4.1:4.3. In spectrum C, the relative integration of 2:13-t:14-x (referenced to succinate) was 1.6:3.0:4.4. Therefore, 47% of 2 survives, 73% of 13-t survives and all of 14-t survives.   Fig. 22 Differing rates of photochemical destruction of open-chain, predominantly cyclic and almost exclusively cyclic sugars. To facilitate observation and quantification of the system, glyceraldehyde 7 was omitted from the mixture and only one tetrose and one pentose were added (threose 13-t and xylose 14-x were chosen as they gave resolved signals in the 1 H NMR spectrum of the mixture). Glycolaldehyde 2 (18 mg, 0.300 mmol) and xylose 14-x 45 mg, 0.300 mmol) were dissolved in H2O (0.5 mL) and threose 13-t syrup (≥ 60%) was added until roughly equimolar (as judged by removing several µL, diluting with D2O and obtaining a 1 H NMR spectrum). Sodium succinate (~ 10 mg, ~ 0.062 mmol) was added as a standard, and the volume adjusted to 2 mL with H2O. From this stock solution, a portion was removed (0.5 mL), NaH2PO4.2H2O (31 mg, 0.200 mmol) added, then the pH was adjusted to ~ 7 and the volume increased to 2 mL. The solution was then irradiated for 2 h. A -1 H NMR Spectrum of the starting mixture, singlet at 2.3 ppm due to the added succinate standard; B -As spectrum A, after 2 h of irradiation. In spectrum A, the relative integration of 2:13-t:14-x (referenced to succinate) was 3.4:4.1:4.3. In spectrum B, the relative integration of 2:13-t:14-x (referenced to succinate) was 2.6:3.7:4.4. Therefore, 76% of 2 survives, 90% of 13-t survives and all of 14-t survives. A solution of KCN (0.090 mmol, 6 mg), NaH2PO4.2H2O (0.150 mmol, 23 mg) and Na3PSO3 (the purity and water content were predetermined and accounted for, 0.200 mmol) were dissolved in degassed 10% D2O in H2O (1 mL) and pH adjusted to 6.5. Glycolonitrile 1 (0.015 mmol, 3.0 µL) was added, the volume made up to 2 mL with degassed 10% D2O in H2O then the solution irradiated. A -1 H NMR Spectrum of the crude reaction mixture following the preceding procedure after 2.5 h, succinate added as a reference (not shown); B -As spectrum A, after 5 h irradiation. As can be seen, after 5 h irradiation, the reaction mixture is almost devoid of glycolaldehyde 2 and glyceraldehyde 7. Although the tetroses 13 have diminished somewhat in intensity after 5 h reaction compared to 2.5 h, the abundance of pentoses 14 have actually increased -in spectrum A there is ~ 14% of 14 and in spectrum B there is ~ 21%. X is an unidentified compound(s). Na3PSO3 (the purity and water content were predetermined and accounted for, 0.600 mmol) were dissolved in degassed 10% D2O in H2O (2 mL) and pH adjusted to 6.5 with degassed 6 M HCl. Glycolonitrile 1 (0.045mmol, 4.5 µL) was added, the volume made up to 6 mL with degassed 10% D2O in H2O and the solution was divided into 3 cuvettes before being irradiated for 2.75 h. The contents of the cuvettes were combined and N2 was bubbled through the solution for 18 h, at which point ~ 0.9 mL remained. A portion of this crude reaction mixture (450 µL) was then removed and succinate (pH 7, 50 µL, 50 mM) added before a 1 H NMR spectrum was acquired (spectrum A). To the NMR tube was charged NH2CN (15 mg, 0.357 mmol) and, after mixing, was heated to 60 °C for 2 h. The tube was removed from the oil bath and spectrum B acquired; C -As spectrum B, spiked with an authentic sample of riboaminooxazoline 16-r; D -As spectrum C, spiked with an authentic sample of xyloaminooxazoline 16-x; E -As spectrum D, spiked with an authentic sample of arabino-aminooxazoline 16-a; F -As spectrum E, spiked with an authentic sample of lyxoaminooxazoline 16-l. Ribo-Aminooxazoline was formed in ~ 1% yield based on the initial concentration of 1 and HCN.  Fig. 24; B -Following the procedure outlined in Supplementary Fig. 24, to the remaining concentrated, crude reaction mixture (~ 450 µL), CaCN2 (~ 85%, 28 mg, 0.300 mmol) was added, then the suspension was sealed and heated to 60 °C for 2 h with stirring. The suspension was briefly centrifuged, then the supernatant was removed and examined by 1 H NMR spectroscopy -spectrum B; C -As spectrum B, spiked with an authentic sample of ribo-aminooxazoline 16-r; D -As spectrum C, spiked with an authentic sample of xylo-aminooxazoline 16-x; E -As spectrum D, spiked with an authentic sample of arabino-aminooxazoline 16-a; F -As spectrum E, spiked with an authentic sample of lyxo-aminooxazoline 16-l. and Na3PSO3 (the purity and water content were predetermined and accounted for, 0.600 mmol) were dissolved in degassed 10% D2O in H2O (2 mL) and pH adjusted to 6.5 with degassed 6 M HCl. The volume was made up to 6 mL with degassed 10% D2O in H2O and the solution was divided into 3 cuvettes before being irradiated for 3 h. The contents of the cuvettes were combined and N2 was bubbled through the solution for 18 h, at which point ~ 0.6 mL remained. A portion of this crude reaction mixture (450 µL) was then removed and succinate (pH 7, 50 µL, 50 mM) added before a 1 H NMR spectrum was acquired (spectrum A). To the NMR tube was charged NH2CN (15 mg, 0.357 mmol) and, after mixing, was heated to 60 °C for 2.5 h. The tube was removed from the oil bath and spectrum B acquired; C -As spectrum B, spiked with an authentic sample of ribo-aminooxazoline 16-r. Yield of ribo-aminooxazoline 16-r ~ 1% starting from HCN.

C B
A

Procedure 2
Dihydroxyacetone 12 or glyceraldehyde 7 (0.030 mmol, 2.7 mg), Na3PSO3.xH2O (purity and water content predetermined and accounted for, 2 equiv.) and NaH2PO4.2H2O (if used, 0.060 mmol, 9 mg) were dissolved in an Eppendorf tube with degassed 10% D2O in H2O (1 mL), and the pH was adjusted to 6.5 with degassed HCl. The volume was made up to 2 mL with degassed 10% D2O in H2O, and the solution was transferred to a quartz cuvette and sealed.
The reaction was irradiated for the desired amount of time, after which it was analysed by 1 H NMR spectroscopy.

d/ppm
Supplementary Fig. 27 Photochemical reduction of dihydroxyacetone 12 to glycerol 11 in the absence of PO4 3-. A -1 H NMR Spectrum of the reaction according to Procedure 2 after 0.5 h using 12 as the starting material and no added phosphate; B -As spectrum A, spiked with a commercial sample of acetone 17 (2.15 ppm); C -As spectrum B, spiked with a commercial sample of 11; D -As spectrum C, spiked with a commercial sample of isopropanol S1; E -1 H NMR Spectrum of an authentic sample of propan-1,2-diol. The fact that only 5% of propan-1,2-diol was formed as compared to 22% of isopropanol shows that a-dehydroxylation predominates over 1,2-reduction as the preferred, one electron reduction pathway for hydroxyacetone. Yields estimated by addition of disodium fumarate (50 mM solution, 50 µL into 450 µL of crude reaction) as a standard and relative integration of 1 H NMR signals. (1 mL) was adjusted to pH 6.5. The volume was made up to 2 mL using degassed 10% D2O in H2O and 1 (0.020 mmol, 2.0 µL) was added. The reaction was then irradiated for 2.5 h and a 1 H NMR spectrum was acquired of the crude reaction mixture (spectrum A); B -1 H NMR Spectrum of a commercial sample of glycerol. Yields estimated by addition of disodium fumarate (50 mM solution, 50 µL into 450 µL of crude reaction) as a standard and relative integration of 1 H NMR signals.
Glycerol 11 (4.6 mg, 0.050 mmol) and acrylonitrile (if used, 0.200 mmol, 11 mg, 13.1 µL) were dissolved in degassed formamide (1 mL) then added to HPSO3 2in an Eppendorf tube, sealed and either left at room temperature or heated to 70 °C. Samples were removed at the desired time points, diluted with D2O and examined by 31 P NMR spectroscopy. d/ppm Supplementary Fig. 32 Phosphorylation of glycerol 11 using PSO3 3and acrylonitrile. A -31 P NMR Spectrum after 3 h of reaction carried out according to Procedure 3 at 70 °C (glycerol-1phosphate 19 ~ 30% yield and glycerol-2-phosphate 20 in ~ 11% yield); B -As spectrum A, after being spiked with a commercial sample of glycerol-1-phosphate 19; C -As spectrum B, after being spiked with a commercial sample of glycerol-2-phosphate 20; D -As spectrum C, after being spiked with commercial pyrophosphate. The other minor signals from 3.5 ppm to 5.0 ppm are thought to be bis-and tris-phosphorylated glycerol. We then attempted to determine what the signal at -3.94 ppm corresponded to, see Supplementary Fig. 33. Yields based on 11 and determined by relative integration of the 31 P NMR signals.   Fig. 33 Phosphorylation of glycerol 11 using PSO3 3and acrylonitrile. A -31 P NMR Spectrum after 3 h of reaction carried out according to Procedure 3 at 70 °C; B -As spectrum A, after being adjusted to pH ~ 13. The peak at -3.94 ppm was first assumed to be symmetrical monothiopyrophosphate, resulting from the addition of one molecule of thiophosphate through the sulfur anion onto another molecule of activated thiophosphate with loss of RS -, but the chemical shift does not match that reported for symmetrical monothiopyrophosphate (~ 15 ppm at pH = 13 45 ); C -As spectrum B, except a 31 P-1 H coupling experiment was run. As the signal is not split in this spectrum, it would appear this species is not attached to glycerol or acrylonitrile, but to rule out glycerol-1,3-cyclic phosphate unequivocally the standard was made and spiked into the NMR sample; D -As spectrum C, after being spiked with a mixture of glycerol-1,3-cyclic phosphate (-3.65 ppm) and glycerol-1,2-cyclic phosphate (18.5 ppm, see Reference 1 for synthesis); E -As spectrum A, but heating was continued for 20 h whereupon the signal at -3.94 ppm had almost been consumed and the peaks corresponding to orthophosphate and pyrophosphate had increased. Additionally, the compound does not seem to be related to acrylonitrile as it appears to form in the absence of acrylonitrile ( Supplementary Fig. 37). Slight changes in chemical shift are due to minor variations in pH upon addition of the standards. The peak at -7.15 ppm is due to pyrophosphate. Other minor signals from 3.5 ppm to 5.0 ppm are thought to be bis-and tris-phosphorylated glycerol. Yields based on 11 and determined by relative integration of the 31 P NMR signals. 3)], suggesting this is 2-cyanoethyl S-thiolodiphosphate -the product of S-2-cyanoethyl thiophosphate adding to another activated phosphate species -and shows good correlation with the 31 P NMR data of a related structure. 46 The other pair of doublets in spectrum A [d 30.1 (J = 31.7), -6.95 (J = 32.0)] we assign to unsymmetrical monothiopyrophosphate, although it shows a slight perturbation from the reported 31 P NMR data, 45 the literature data was acquired under different conditions (pure D2O, pH 13.1); C -As spectrum A, after 1 week of reaction according to Procedure 3 (glycerol-1-phosphate 19 ~ 25% yield and glycerol-2phosphate 20 in ~ 6% yield); D -As spectrum C, after being spiked with a commercial sample of 19; E -As spectrum D, after being spiked with a commercial sample of 20. The other minor signals from 4.0 -4.8 ppm are thought to be bis-and tris-phosphorylated glycerol. Yields based on 11 and determined by relative integration of the 31 P NMR signals. Dibasic thiophosphate was prepared according to General procedure 1 (0.050 mmol or 0.100 mmol). Glycerol 11 (4.6 mg, 0.050 mmol) was dissolved in degassed formamide (1 mL) and added to the thiophosphate. Ferricyanide (2 equiv. relative to thiophosphate) was then added, and the reaction agitated before a sample was removed, diluted with D2O and examined by 31 P NMR spectroscopy. A -31 P NMR Spectrum after 7 h reaction using 100 mM thiophosphate (~ 20% of glycerol-1-phosphate 19 and ~ 6% of glycerol-2-phosphate 20 for 50 mM PSO3 3-(not shown) or ~ 30% of 19 and ~ 8% of 20 for 100 mM PSO3 3-); B -As spectrum A, after being spiked with a commercial sample of 19; C -As spectrum B, after being spiked with a commercial sample of 20; D -As spectrum C, after being spiked with a commercial sample of pyrophosphate. The other minor signals from -0.1 ppm to 0.6 ppm are thought to be bis-and tris-phosphorylated glycerol. Ferrocyanides (Fe(CN6) 4-) are heavily implicated in our scheme, 1,8,9,10,36 and these complexes are easily oxidised to ferricyanides (Fe(CN6) 3-) by UV light. 47 Yields based on 11 and determined by relative integration of the 31 P NMR signals.  Fig. 37 Phosphorylation of glycerol 11 by irradiation of PSO3 3-. Dibasic thiophosphate was prepared according to General procedure 1 (0.100 mmol). Glycerol 11 (4.6 mg, 0.050 mmol) was dissolved in degassed formamide (1 mL), added to the thiophosphate and then the solution was transferred to a quartz cuvette. The reaction was sealed and irradiated for the desired amount of time before a sample was removed, diluted with D2O and examined by 31 P NMR spectroscopy. A -31 P NMR Spectrum after 15 h of reaction (~ 28% of glycerol-1-phosphate 19 and ~ 8% of glycerol-2-phosphate 20); B -As spectrum A, after being spiked with a commercial sample of 19; C -As spectrum B, after being spiked with a commercial sample of 20. The other minor signals from 1.8 ppm to 3.5 ppm are thought to be bis-and tris-phosphorylated glycerol. The compound which gives a signal at -3.94 ppm has been discussed in Supplementary Fig. 33. The pair of doublets (33.7 ppm and -9.85 ppm, J = 28.5) are thought to be due to unsymmetrical monothiopyrophosphate and correlate reasonably well with the literature values (30.8 ppm and -5.98 ppm, J = 30, in D2O, pH = 13.1) 45 , there may be some discrepancy due to the difference in pH and that our samples contained 20-30% formamide. The peak at -7.95 ppm is due to pyrophosphate. The chemical shift was altered slightly by a change in pH upon addition of the standards. Yields based on 11 and determined by relative integration of the 31 P NMR signals.  To an Eppendorf tube was charged NH4Cl (8 mg, 0.150 mmol), NaH2PO4.2H2O (8 mg, 0.050 mmol) and Na3PSO3.xH2O (purity and water content predetermined and accounted for, 0.100 mmol), then degassed 10% D2O in H2O (1 mL) was added. The pH was adjusted to 7.0, the volume was made up to 2 mL with degassed 10% D2O in H2O and glycolonitrile 1 (4.9 mL, 0.050 mmol) was added. The solution was transferred to a quartz cuvette, sealed and the reaction was irradiated for 1.5 h.
From this solution, 1 mL was removed and added to an Eppendorf tube containing cyanamide (2 mg, 0.050 mmol), which was then sealed and heated to 50 °C for 20 h. An aliquot (450 µL) was removed and transferred to an NMR spectroscopy tube and an aqueous solution of fumarate (50 mM, 50 µL) was added as an internal standard.  Although the yield of 9 was low, its accumulation may be expected given the high boiling point, unless undergoing other reactions. This is in line with Orgel's synthesis of oligonucleotides from ribo-nucleoside-2ʹ,3ʹ-cyclic phosphates, which requires the inclusion of 9 and evaporation of H2O to the point of dryness 25 . When the prebiotic synthesis of 2-aminooxazole 21 was first reported from the condensation of 2 with cyanamide, a mechanism which essentially follows the route depicted in box a was suggested. 26 However, after it was shown that 2-aminoimidazole 21 can be formed by inclusion of NH4Cl with 2 and cyanamide it was clear that a different route must be available which operates in tandem with, or instead of, that shown in box a. 48 Similarly, to form 2-aminothiazole 23, the mechanism in box a will not suffice. There are two possible alternative mechanisms which are outlined in box b, the difference being that either the nucleophile (XHn) adds first (blue pathway) or cyanamide adds first (magenta pathway). Following the blue pathway from S5, H2O is expelled to give the C=XHn-2 bond in S6, this compound can undergo a Lobry de Bruyn-Alberda van Ekenstein rearrangement in the case that XHn = OH2 and an Amadori rearrangement in the case that XHn = NH3 to give S7 (see below for XHn = SH2). Addition of cyanamide then allows cyclisation of XHn-1 onto the nitrile carbon of S8 and after dehydration and aromatisation results in 21/22/(23). Following the magenta pathway, cyanamide adds to 2 giving S2 and this species extrudes H2O to give S9 which undergoes Amadori rearrangement to aldehyde S10. Addition of the nucleophile XHn then gives S11 from which XHn-1 or C1-OH can cyclise onto the nitrile carbon, eventually leading to 21/22/23. Whilst the blue pathway may be reasonable in the case of 22, to arrive at 23 via this route would require the formation of a thiocarbonyl using nothing more than general acid-base catalysis, which seems unlikely considering the usual requirement for strong Bronsted/Lewis acid catalysis. and 29 from ethanolamine 9. box a -the reductive methylation of ammonia provides 24 (this work, see main text) and we have previously shown how simple amines such as 24 can be converted into isonitriles e.g. 25, by ferrocyanide and nitroprusside, also expected products of the cyanosulfidic scenario we have discussed before. 1,7 We, 7,8 and others, 30 have shown how isonitriles can be activated for nucleophilic attack by nucleotides S13 by an aldehyde or mildly acidic conditions. The imidoyl phosphate intermediate S14 is unstable and can be subject to attack by imidazoles, such as 2-aminoimidazole 22, giving phosphoroimidazolides S15. These can be 5'-mononucleotides, activated for incorporation into a primer, 29 or a 5'terminus of an oligonucleotide activated for ligation; 30 box b -reductive methylation of ethanolamine 9 sequentially gives 28 and 29, further addition of formaldehyde 27 and reduction converts more 9 to 28 and 28 to 29, potentially giving 29 exclusively. The amino alcohols 9, 28 and 29 can then be phosphorylated with Na3PO3.  Supplementary Fig. 44 Prebiotic synthesis of methylamine 24. A -1 H NMR Spectrum of a reaction carried out according to Procedure 7 after 2 h irradiation at pH 9.2; B -As spectrum A, spiked with a commercial sample of 24; C -1 H NMR Spectrum of a reaction carried out according to Procedure 7 after 2 h irradiation at pH 7.0. Singlets observed at d 3.28 ppm in all spectra are due to MeOH which is present in commercial formaldehyde 27 as a stabiliser and also results from direct reduction of 27.  a -Reactions containing PO4 3were run at an initial pH of 7, otherwise at an initial of pH 9.2

Procedure 8
Ethanolamine 9 (0.060 mmol, 3.6 µL), formaldehyde 27 (37% wt., 0.180 mmol, 13.5 µL) and Na3PSO3.xH2O (purity and water content predetermined and accounted for, 0.240 mmol) were dissolved in an Eppendorf tube in degassed 10% D2O in H2O (1 mL), then the pH was adjusted to 7.0. The volume was made up to 2 mL with degassed 10% D2O in H2O and the solution was transferred to a quartz cuvette, sealed and the reaction was irradiated for 4 h. A 1 H NMR spectrum was then collected.
Formaldehyde 27 (37% wt., 0.040 mmol, 3.2 µL) and Na3PSO3.xH2O (purity and water content predetermined and accounted for, 0.080 mmol) were added and the pH re-adjusted to 7.0. Irradiation continued a further 2 h before another 1 H NMR spectrum was collected.
Formaldehyde 27 (37% wt., 0.040 mmol, 3.2 µL) was added and irradiation continued for a further 2 h after which time a 1 H NMR spectrum was collected. according to Procedure 8 after 2 h irradiation after the 2 nd addition of formaldehyde 27; C -1 H NMR Spectrum of a reaction carried out according to Procedure 8 after 2 h irradiation after the 3 rd addition of 27; D -As spectrum C, spiked with an authentic sample of 28; E -As spectrum D, spiked with a commercial sample of 29.

Procedure 9
The aminoalcohol 9, 28 or 29 (0.050 mmol) and Na3PSO3.xH2O (purity and water content predetermined and accounted for, 0.050 mmol) were dissolved in an Eppendorf tube in degassed H2O (0.8 mL) and the pH was adjusted to 7.0 -7.2 with degassed HCl. The solution was frozen in liquid N2 and lyophilised.
The solid was dissolved in degassed formamide (1 mL) and potassium ferricyanide (0.100 mmol, 33 mg) was added. The reaction was then agitated and a sample was taken at the desired timepoint and diluted with D2O. NMR Spectroscopy was then used to analyse the reaction.

d/ppm
Supplementary Fig. 46 Phosphorylation of ethanolamine 9. A -31 P NMR Spectrum of a reaction after 6.5 h carried out according to Procedure 9 using 9 as the nucleophile; B -Same sample as in spectrum A but this NMR spectrum was acquired with 1 H-31 P coupling (peak at 0.34 ppm splits into a triplet, indicating coupling to a CH2 group); C -As spectrum A, spiked with a commercial sample of O-phosphorylethanolamine 30.  Supplementary Fig. 47 Synthesis of N-methylethanolamine phosphate 31 and N,Ndimethylethanolamine phosphate 32 from N-methyl ethanolamine 28 and N,N-dimethyl ethanolamine 29, respectively. A -31 P NMR Spectrum of a reaction after 6.5 h carried out according to Procedure 9 using 28 as the nucleophile; B -Same sample as in spectrum A, NMR spectrum acquired with 1 H-31 P coupling (peak at 0.20 ppm splits into a triplet, indicating coupling to a CH2 group); C -31 P NMR Spectrum of a reaction after 6.5 h carried out according to Procedure 9 using 29 as the nucleophile; D -Same sample as in spectrum C, NMR spectrum acquired with 1 H-31 P coupling (peak at 0.03 ppm splits into a triplet, indicating coupling to a CH2 group).

Procedure 10
Glycolonitrile 1 (0.030 mmol, 3.0 µL) and Na3PSO3.xH2O (purity and water content predetermined and accounted for, 0.120 mmol) were dissolved in an Eppendorf tube with degassed 10% D2O in H2O (1 mL), then the pH was adjusted to 6.5 using degassed HCl. The volume was made up to 1.5 mL with degassed 10% D2O in H2O, the reaction was sealed and heated to 65 °C for 20 h and a 1 H NMR spectrum was acquired ( Supplementary Fig. 48,  spectrum A).
The NMR spectrum sample was recombined with the mother liquors and Na3PSO3.xH2O (purity and water content predetermined and accounted for, 0.120 mmol) was added. The pH was re-adjusted to 6.5 and the solution was sealed and transferred to a quartz cuvette. The reaction was irradiated for the desired amount of time after which it was analysed by 1 H NMR spectroscopy.

Procedure 11
Potassium cyanide (6.5 mg, 0.100 mmol), acetone 17 (3.7 µL, 0.050 mmol) and Na3PSO3.xH2O (purity and water content predetermined and accounted for, 0.250 mmol) were dissolved in an Eppendorf tube with degassed 10% D2O in H2O (0.6 mL), then the pH was adjusted to 6.5 using degassed HCl. The volume was made up to 1 mL with degassed 10% D2O in H2O, the reaction was sealed and heated to 50 °C for 4 h. A 1 H NMR spectrum of the crude reaction was collected ( Supplementary Fig. 49, spectrum A). The pH was adjusted to 6.5 and the solution sparged with N2 for 2 h before a 1 H NMR spectrum was collected ( Supplementary Fig.   49, spectrum B). The volume was re-adjusted to 1 mL with degassed 10% D2O in H2O and transferred to a quartz cuvette, then the reaction was irradiated for 4 h. A 1 H NMR spectrum was collected ( Supplementary Fig. 49, spectrum C) then KCN (2 mg, 0.031 mmol) was added to the NMR sample and the pH adjusted to 6.5, after which a further 1 H NMR spectrum was collected ( Supplementary Fig. 49, spectrum D). OH C -As spectrum B after irradiation for 4 h; D -As spectrum C after addition of HCN; E -As spectrum D after spiking with synthetically prepared 43 (singlet at 3.29 ppm due to residual MeOH from the preparation of 43, see Synthetic procedures to make standards (p. 81) and Supplementary Fig. 67); F -1 H NMR Spectrum of synthetically prepared 43 (singlet at 3.29 ppm due to residual MeOH from the preparation of 43, see Synthetic procedures to make standards (p. 81) and Supplementary Fig. 67).

Procedure 12
Potassium cyanide (14 mg, 0.220 mmol) was dissolved in a solution of a-hydroxyisobutyraldehyde 38 (2 mL, ca. 100 mM, as prepared in Synthetic procedures to make standards, p. 75) and Na3PSO3.xH2O (purity and water content predetermined and accounted for, 1.2 mmol was added). The pH was adjusted to 6.5 using degassed HCl (final volume ca. 2.2 mL) and the reaction was sealed and heated to 50 °C for 2 days at which point a 1 H NMR spectrum was collected ( Supplementary Fig. 50, spectrum B). Of this reaction mixture, a portion was taken (0.5 mL) and dissolved in degassed 10% D2O in H2O (1 mL) and Na3PSO3.xH2O (purity and water content predetermined and accounted for, 0.200 mmol) was added. The pH was adjusted to 6.5 then the volume made up to 2 mL with degassed 10% D2O in H2O. The resulting solution was transferred to a quartz cuvette, sealed and irradiated for 12 h. d/ppm Supplementary Fig. 50 Synthesis of 3-methyl-butane-1,3-diol 39. A -1 H NMR Spectrum of the cyanohydrin formed after addition of HCN to a-hydroxyisobutyraldehyde 38 (peak at 3.33 ppm due to MeOH present in the initial aldehyde 38 solution, see Synthetic procedures to make standards, p. 81 and Supplementary Figs. 65 and 67); B -1 H NMR Spectrum of a reaction carried out according to Procedure 12 after heating for 2 days (peak at 3.33 ppm due to MeOH   Fig. 54 Phosphorylation of dimethylallyl alcohol 46. Dibasic thiophosphate was prepared according to General procedure 1 (0.050 mmol or 0.100 mmol). Dimethylallyl alcohol 46 (5.1 µL, 0.050 mmol) was dissolved in degassed formamide (1 mL) and added to the thiophosphate. Potassium ferricyanide (66 mg, 0.200 mmol) was then added and the reaction agitated for the desired time before a sample was removed, diluted with D2O and examined by 31 P NMR spectroscopy. A -31 P NMR Spectrum of a reaction after 7 h reaction; B -As spectrum A spiked with a sample of dimethylallyl phosphate 41 (see Synthetic procedures to make standards, p.82 and Supplementary Fig. 71 Fig. 55 Phosphorylation of dimethylallyl alcohol 46. Dibasic thiophosphate was prepared according to General procedure 1 (0.100 mmol). Dimethylallyl alcohol 46 (5.1 µL, 0.050 mmol) was dissolved in degassed formamide (1 mL), added to the thiophosphate and then the solution was transferred to a quartz cuvette. The reaction was sealed and irradiated for the desired amount of time before a sample was removed, diluted with D2O and examined by 31 P NMR spectroscopy. A -31 P NMR Spectrum of a reaction after 5 h; B -As spectrum A spiked with a sample of 41 (see Synthetic procedures to make standards, p.82 and Supplementary Fig. 71 (50 mM) in degassed 10% D2O in H2O at pH 6.5 and ambient temperature. A -t = 0, 99% thiophosphate remains; B -t = 7 days, 75% thiophosphate remains; C -t = 13 days, 57% thiophosphate remains; D -t = 19 days, 45% thiophosphate remains. The rate of hydrolysis is actually much slower than the reported rate, 49 but this is likely due to the use of a molybdate complex used to quantify orthophosphate production which can undergo reaction with sulfides to form sulfido complexes. 50 The reagent can also be reduced by sulfides giving erroneous measurements. The suspension was sealed and gently agitated after the desired amount of time a sample was removed, centrifuged and examined by 31 P NMR spectroscopy. A -Quantitative 31 P NMR spectrum of the reaction after 3 d; B -Quantitative 31 P NMR spectrum of the reaction after 6 d; C -Quantitative 31 P NMR spectrum of the reaction after 12 d. The singlet at 32.1 ppm corresponds to thiophosphate (spike not shown). The percentage of the total of the integrals of the signals in the 31 P NMR spectra above that accounts for thiophosphate was as follows: 6% in spectrum A, 13% in spectrum B and 15% in spectrum C. For discussion of some of the other species present, see Supplementary Fig. 59. Note: the endpoint for accumulation of thiophosphate was not measured. Supplementary Fig. 59 NaSH.xH2O (25 mg, 0.267 mmol) was dissolved in degassed 10% D2O in H2O (1 mL) in an Eppendorf and Fe3P (300 mg, 0.151 mmol) was added. The suspension was sealed and gently agitated for 6 d, centrifuged and the supernatant examined by 31 P NMR spectroscopy, which gave spectrum A. B -As spectrum A, but the NMR spectrum was acquired with 31 P-1 H coupling. In spectrum A, the singlet at -5.48 ppm is due to pyrophosphate; the doublet of doublets (-3.89 (J = 17.1), -4.88 (J = 17.1)) is tentatively assigned to the phosphite-phosphate anhydride (isohypophosphate) given the 31 P signals disappear upon coupling to H nuclei. Although the literature values for the chemical shifts differ (-4.4 and -9.6 ppm), those studies were conducted at acidic pH 52 ); the singlet at 3.66 ppm corresponds to phosphite; singlet at 4.76 ppm corresponds to orthophosphate; the singlet at 32.1 ppm corresponds to thiophosphate; whilst the other major signals could not be definitively assigned, it is worth noting that the phosphorus atoms are not attached to hydrogen -although the signal at 87.6 ppm appears to be split, it actually aligns with the right-hand, apparent doublet in spectrum B. We also note that species of the type dithiphosphate (PO2S2 3-) and trithiophosphate (POS3 3-) can be observed in the range ~ 60 -90 ppm and 85 -101 ppm, respectively, depending on pH. 53 Fig. 60 Corrosion experiment of Fe3P (a commercial surrogate for schreibersite 51 ) by water containing HS -/H2S at neutral pH. NaSH.xH2O (12 mg, 0.129 mmol) was dissolved in degassed 10% D2O in H2O (0.6 mL) in an Eppendorf and the pH was adjusted to 7.2 with degassed HCl. Fe3P (300 mg, 0.151 mmol) was added and the suspension was sealed and gently agitated for the desired amount of time. The suspension was briefly centrifuged and the pH checked, then a sample was removed and examined by 31 P NMR spectroscopy. A -31 P NMR spectrum of the supernatant after 7 d, the pH was ~ 8; B -31 P NMR spectrum of the supernatant after 4 weeks, the pH was ~ 11.4. In spectrum B, the broad signal at 32 ppm corresponds to thiophosphate, which integrates for ~ 10% of the all the signals in the 31 P NMR spectrum.

Procedure 14
NaH2PO4.2H2O (if used, 2 mg, 0.013 mmol), KCN (6 mg, 0.090 mmol), NaSH.xH2O (> 60%, 8 mg, 0.090 mmol), HPO3Na2.5H2O (if used, 9 mg, 0.040 mmol) and hypophosphorous acid (if used, 50% wt. 8.3 µL, 0.080 mmol) were dissolved in degassed 10% D2O in H2O (1 mL) and the pH adjusted to 6.5. Glycolonitrile 1 (55% wt. 3.0 µL, 0.030 mmol) was added, the volume was made up to 2 mL with degassed 10% D2O in H2O before the solution was transferred to a cuvette and irradiated. d/ppm Supplementary Fig. 61 The photochemical reduction of nitriles by HSin the presence and absence of hypophosphite, phosphite and phosphate. A -1 H NMR Spectrum of the crude reaction mixture after 1 h according to Procedure 14 when no phosphorus species were included in the reaction, ~ 4% of glyceronitrile 4 was present; B -1 H NMR Spectrum of the crude reaction mixture after 1 h according to Procedure 14 with the inclusion of phosphorus species, ~ 26% of glyceronitrile 4, ~ 16% of glyceraldehyde cyanohydrin 10, ~ 3% ethylene glycol 6 and ~ 3% glycerol 11 were present; C -1 H NMR Spectrum of glyceraldehyde cyanohydrin 10; D -As spectrum A with the inclusion of NaH2PO4.2H2O. As can be seen by comparing spectra A and D, the rate enhancement observed in spectrum B must be due to the inclusion of the reduced phosphorus species. The ratio of HPO3 2-:H2PO2 -:PO4 3is the same as that reported by Bryant and Kee for the anoxic corrosion of schreibersite under conditions of UV irradiation, and thus seem most relevant to our geochemical scenario. 13 d/ppm Supplementary Fig. 62 The photochemical reduction of nitriles by HSin the presence of hypophosphite, phosphite and phosphate. A -31 P NMR Spectrum of the same sample as Supplementary Fig. 61, spectrum B; B -As spectrum A, but the NMR spectrum was acquired with 31 P-1 H coupling. The presence of thiophosphate (PSO3 3-) and thiophosphite (HPSO2 2-) is clear, and as these compounds cannot be formed in the above experiments where the rate of reduction of glycolonitrile is slow ( Supplementary Fig. 61, spectra A and D) it is presumably their rapid formation and photolysis which provides the observed rate enhancement.

Synthetic procedures to make standards
a-Hydroxyisobutyraldehyde 38 54 The synthesis follows that outlined in Ref. 54 -10  0  10  20  30  40  50  60  70  80  90  100  110  120  130  140  150  160  170  180  190  200 Supplementary Fig. 67 Signal at 3.29 ppm due to MeOH liberated in the acetal hydrolysis, see procedure.   Supplementary Fig. 71. A -1 H NMR Spectrum of the product from the procedure used to prepare dimethylallyl phosphate 41 (see Synthetic procedures to make standards, p. 76); B -As spectrum A, after the addition of MgCl2 (1.5 mg) and NaCl (6 mg), pH adjusted to 8.2 and alkaline phosphatase solution added (1000 units, 3 µL). Incubated for 18 h at 37 °C. Alkaline phosphatase cleaves any monosubstituted phosphate group i.e. ROPO3 2-, yielding orthophosphate and the hydroxy R-group, ROH. The complex set of signals 3.47 -3.76 ppm are due to glycerol present in the commercial phosphatase solution; C -As spectrum B, after spiking with commercial dimethylallyl alcohol 46. As dimethylallyl pyrophosphate is absent from the 31 P NMR spectrum (see Supplementary Fig. 72), we assigned the compound in spectrum A to dimethylallyl phosphate 41. The doublet which is seen at 3.4 ppm in spectrum A is due to methyl phosphate and resulted from the phosphorylation of MeOH which was present in commercial formamide that was used for the phosphorylation reaction. After phosphatase treatment MeOH is regenerated, see spectrum B, singlet at 3.30 ppm.  Supplementary Fig.  71, spectrum A; B -Same sample as spectrum A, but run with 1 H-31 P coupling; C -31 P NMR Spectrum of the same sample as in Supplementary Fig. 71, spectrum B. In spectrum A, the singlets at 0.30 and -9.60 ppm are due to orthophosphate and pyrophosphate, respectively (also confirmed by sample spiking). The singlet at 2.25 ppm is due to methyl phosphate (see Supplementary Fig. 71 legend) and in spectrum B is split into a quartet (see Supplementary  Fig. 74 for more clarity). We assign the singlet at 0.96 ppm to dimethylallyl phosphate 41.  Fig. 73 A -1 H NMR Spectrum of the product of the procedure used to prepare isopentenyl phosphate 42 (see Synthetic procedures to make standards, p. 76); B -As spectrum A, after the addition of MgCl2 (1.5 mg) and NaCl (6 mg), pH adjusted to 8.2 and alkaline phosphatase solution added (1000 units, 3 µL). Incubated for 18 h at 37 °C. Alkaline phosphatase cleaves any monosubstituted phosphate group i.e. ROPO3 2-, yielding orthophosphate and the hydroxy R-group, ROH. The complex set of signals 3.47 -3.76 ppm are due to glycerol present in the commercial phosphatase solution; C -As spectrum B, after spiking with commercial isopentenyl alcohol 45. In spectrum A, we assign the major product as isopentenyl phosphate 42 [alkene signals obscured by HOD, 3.80 (m, 2H), 2.25 (t, J = 6.9, 2H), 1.68 (s, 3H)] and the minor compound as isopentenyl pyrophosphate S16 [alkene signals obscured by HOD, 3.97 (q, J = 6.7, 2H), 2.30 (t, J = 6.5, 2H), 1.68 (s, 3H. The formation of S16 was indicated by the conversion of the signals into isopentenyl alcohol 45 after phosphatase treatment and by 31 P NMR data (see Supplementary Fig. 74). The doublet which is seen at 3.38 ppm in spectrum A is due to methyl phosphate and resulted from the phosphorylation of MeOH which was present in commercial formamide that was used for the phosphorylation reaction. After phosphatase treatment MeOH is regenerated, see spectrum B, singlet at 3.30 ppm.  Supplementary Fig.  73, spectrum A; B -Same sample as spectrum A, but run with 1 H-31 P coupling; C -31 P NMR Spectrum of the same sample as in Supplementary Fig. 73, spectrum B. In spectrum A, the singlets at 1.9 and -7.70 ppm are due to orthophosphate and pyrophosphate, respectively (also confirmed by sample spiking, not shown). The singlet at 4.37 ppm is due to methyl phosphate (see Supplementary Fig. 73 legend) and in spectrum B is split into a quartet. We assign the singlet at 3.10 ppm to isopentenyl phosphate 42. The two doublets that are coupling to each other [-7.10 ppm (J = 21.1) and -10.51 ppm (J = 21.1)] we ascribe to isopentenyl pyrophosphate S16, as the up-field signal is split in the 1 H-31 P coupling NMR experiment (spectrum B) but the down-field signal is not and the compound is digested by phosphatase (spectrum C) returning isopentenyl alcohol 45 ( Supplementary Fig. 73). Control experiments in 'oxygenated' solvents d/ppm Supplementary Fig. 75 Photochemical reduction of glycolonitrile 1 using Na3PSO3 in H2O/D2O and degassed H2O/D2O. A -1 H NMR Spectrum of the reaction according to Procedure 1 (1 equiv. Na3PSO3) after 1 h irradiation in 10% D2O in H2O which was degassed; B -1 H NMR Spectrum of the reaction according to Procedure 1 (1 equiv. Na3PSO3) after 1 h irradiation in 10% D2O in H2O which was not degassed.

Supplementary Discussion 2
The discovery that thiophosphate is produced during the oxidation of reduced phosphorus species by H2S/HSand UV light, 14 or by aqueous alteration of schreibersite by HS -/H2S containing water (vide infra), is indicative that PSO3 3should have been accessible in some locations on Hadean Earth, most likely in the vicinity of meteorite impacts/debris. We previously pointed out that thiophosphate undergoes hydrolysis with a half-life of ~ 17 days at neutral pH and ambient temperature, 15 contrary to reported hydrolysis data, 49 which we have now confirmed using quantitative 31 P NMR (Supplementary Fig. 56). Dittmer et al. also reported that PSO3 3is more stable at alkaline pH, which we also found, observing little degradation after 6 weeks at pH 12 ( Supplementary Fig. 57). When Fe3P (a schreibersite surrogate) 51 was allowed to react with a solution of NaSH, PSO3 3could be observed by 31 P NMR spectroscopy within 3 days and accumulated further over the remainder of the experiment (Supplementary Figs. 58 and 59). Furthermore, when a solution of NaSH was adjusted to pH ~ 7 and allowed to react with Fe3P, the pH of the solution rose during the corrosion process, and after 4 weeks had reached pH ~ 11.4, PSO3 3again being present by 31 P NMR spectroscopy and accounting for ~ 10% of the soluble phosphorus species ( Supplementary Fig. 60).
We note that oldhamite (and niningerite) is prevalent in enstatite chondrites 55 and is frequently associated with schreibersite and perryite, 56,57 often in similar abundance. 58 Therefore, the rapid weathering of oldhamite, 59 forming Ca(OH)2 and H2S, opens the possibility for the leaching of thiophosphate (and/or thiophosphite) directly from meteoritic debris at alkaline pH. Additionally, the photochemical oxidation of hypophosphite by HS -/H2S at neutral pH has been estimated to be > 50% complete within 1 -10 days on primitive Earth, 60 and the oxidation has been shown to be unperturbed at alkaline pH, again forming PSO3 3in the process. 14 This data, coupled with the fact that PSO3 3is a stable solid, suggests a window of opportunity for the accumulation of PSO3 3-, in some locations if not globally, on primitive Earth was viable. Importantly, it must be noted that PSO3 3can be made (from a mixture of H2PO2 -, HPO3 2-, PO4 3and HS -) and used in situ, and consequently the accumulation of PSO3 3per se for photochemical reductions is not required, and its rate of hydrolysis becomes a moot point ( Supplementary Figs. 61 and 62, cf. Supplementary Fig. 8). Although crude, an estimate for the rate of reduction on early Earth can be made from the previous study by Rimmer et al. 61 where the Rayonet photochemical reactor was estimated to be five orders of magnitude more intense than the young Sun, hence the photochemical reductions performed here would have taken place on the timescale of months to years rather than tens of minutes. However, we take this as an absolute lower limit as only wavelengths 252-256 nm are scaled to those reaching primitive Earth from the young Sun, and in reality, broadband emission from the young Sun would be expected to irradiate Earth's surface for all wavelengths > ~ 200 nm. 61,62