Prebiotic Phosphorylation of 2-Thiouridine Provides Either Nucleotides or DNA Building Blocks via Photoreduction

Abstract

Breakthroughs in the study of the origin of life have demonstrated how some of the building blocks essential to biology could have been formed under various primordial scenarios, and could therefore have contributed to the chemical evolution of life. Missing building blocks are then sometimes inferred to be products of primitive biosynthesis, which can stretch the limits of plausibility. Here, we demonstrate the synthesis of 2'-deoxy-2-thiouridine, and subsequently 2'-deoxyadenosine and 2-deoxyribose, under prebiotic conditions. 2'-Deoxy-2-thiouridine is produced by photoreduction of 2,2'-anhydro-2-thiouridine, which is in turn formed by phosphorylation of 2-thiouridine–an intermediate of prebiotic RNA synthesis. 2'-Deoxy-2-thiouridine is an effective deoxyribosylating agent and may have functioned as such in either abiotic or proto-enzyme-catalysed pathways to DNA, as demonstrated by its conversion to 2'-deoxyadenosine by reaction with adenine, and 2-deoxyribose by hydrolysis. An alternative prebiotic phosphorylation of 2-thiouridine leads to the formation of its 5'-phosphate, showing that hypotheses in which 2-thiouridine was a key component of early RNA sequences are within the bounds of synthetic credibility.

Non-canonical, sulfur-containing nucleobases are utilised by extant biology¹,², and exert interesting effects on the stability, base-pairing properties and replication fidelity of nucleic acid polymers³. As such–and given the key role of sulfur in established prebiotic synthetic scenarios⁴,⁵–they are potential components of a genetic system that either complemented or perhaps predated, and ultimately evolved into, enzymatically replicated canonical RNA and DNA. We recently demonstrated a viable pathway to pyrimidine ribonucleosides and ribonucleotides, via thioribonucleosides (Fig. 1)⁵. Highly crystalline ribo-aminooxazoline-1 was shown to react with cyanoacetylene 2 to give anhydronucleoside 3, which on thiolysis furnished α-2-thioribocytidine 4. 4 could be efficiently photoanomerised to β-2-thioribocytidine 5, which serves as a precursor to the canonical pyrimidine ribonucleosides cytidine 6 and uridine 7, β-2-thiouridine 8, and oligomerisable⁶ cyclic phosphates of cytidine and uridine (9 and 10, respectively).

Fig. 1. Summary of previous and present work.

Previous work demonstrated the synthesis of the canonical pyrimidine RNA nucleotides 9 and 10 alongside other RNA building blocks, such as 2-thiouridine 8. This work (boxed) demonstrates a link between 8 and a range of DNA building blocks, via prebiotic phosphorylation and photoreduction. hv, UV irradiation;P_i1, NH₄H₂PO₄/urea.

Since we had demonstrated that thioribonucleosides 5 and 8 are likely formed in primordial pyrimidine synthesis, we set out to further explore prebiotic phosphorylation conditions that might yield oligomerisable thioribonucleotides. Herein, we demonstrate that phosphorylation of 2-thiouridine 8 can either lead to an oligomerisable nucleotide or initiate a prebiotic synthesis of a deoxythionucleoside, 2'-deoxy-2-thiouridine 11, via hydrogen sulfide-mediated photoreduction as the key step. We show that 11 is a productive transglycosylating reagent in an abiotic or proto-enzymatic context; for example, in the supply of 2-deoxyribose 12, and the synthesis of a canonical 2'-deoxynucleoside, 2'-deoxyadenosine 13. These results provide a chemical connection between RNA and DNA in a prebiotic context (that is, in the absence of biosynthesis).

Results and Discussion

5' Phosphorylation

First, we attempted the phosphorylation of 2-thiouridine 8 in hot formamide⁷, but increased the concentration of phosphate and nucleoside from 100 mM to 200 mM, compared to our previous phosphorylation of cytidine (full experimental details are included in the Supplementary Information). After heating at 100 °C for 20 h, we observed the initially formed phosphorylated product, thiouridine 5'-phosphate 14, in 24% yield alongside the starting material 8 in 42% and some minor products (Fig. 2, Supplementary Fig. 39). In hot urea, ribonucleosides generally undergo initial (kinetic) 5'-phosphorylation⁸; over time, the 2',3'-cyclic phosphate dominates over the 5'-phosphate (thermodynamic control). The formation of phosphate monoesters in hot urea (and formamide) is reversible but slow, such that clear-cut kinetic or thermodynamic control is not shown by such systems. Szostak and coworkers³,⁹ have already shown that non-enzymatic RNA replication using the 2-methyl- and 2-amino-imidazole-activated derivatives of 14 proceeds with higher rate and fidelity than the corresponding activated uridine nucleotide, but until now a prebiotic provenance of 14 remained unestablished. These results show that the kinetic product of phosphorylation–the 5' nucleotide 14–is indeed accessible, and would become increasingly plausible if progressively milder prebiotically plausible phosphorylation procedures are uncovered. Our synthesis of 14 shows that it could have been available for non-enzymatic incorporation into primordial RNA sequences, and provides further impetus for the investigation of the potential role of 2-thiouridine in RNA.

Fig. 2. 5' phosphorylation of 2-thiouridine 8 in concentrated formamide solution.

Prebiotically plausible conversion of 8 into 2-thiouridine-5'-phosphate 14 in hot formamide potentially enables incorporation of 8 into primordial RNA sequences, by previously developed prebiotic activation of nucleoside phosphates and oligomerisation chemistry³,⁹.

Thioanhydride Formation

Semi-molten urea is an alternative prebiotic phosphorylation medium to hot formamide⁸, so we next used it to attempt phosphorylation of 8 (Fig. 3a). Surprisingly, no 2',3'-cyclic nucleotide derivatives were observed. A major product had been formed, but a total of four monophosphate ³¹P-NMR resonances suggested that the hydroxyl groups of this product were phosphorylated to varying degrees, complexifying the NMR spectra. The crude mixture from the phosphorylation reaction was therefore treated with alkaline phosphatase, generating a single major nucleoside product. Extensive analysis of both one- and two-dimensional NMR spectra of the dephosphorylated product suggested the formation of 2',2-thioanhydronucleoside 15a¹⁰. Previously, Hampton and Nichol synthesized a related derivative, 2',2-anhydro-uridine 16, when uridine 7 was treated with diphenylcarbonate in hot dimethylformamide (Fig. 3b). Under those conditions, intermediate uridine 2',3'-cyclic carbonate¹¹ 17a is a good substrate for ring opening at C2' by 2-O to give cyclonucleoside 16. Thus, under phosphorylation conditions, we propose cyclic phosphate 18 as the intermediate in the formation of 15 (R₁ = R₂ = H or PO₃H^-), with rapid ring opening by the nucleophilic sulfur atom ensuing to convert 18 to 15. To confirm the structure of the product and support our proposed mechanism, 2',2-anhydro-2-thiouridine 15a was unambiguously synthesized from 2-thiouridine 8 and diphenylcarbonate (Fig. 3b). The fully characterized synthetic standard was used for spiking experiments to confirm that the major product of phosphorylation of 8 (even before enzymatic dephosphorylation) was indeed 15a. After the crude reaction products from the phosphorylation were subjected to alkaline phosphatase (Fig. 3a), addition of an NMR integration standard (sodium formate) revealed that 2',2-anhydro-2-thiouridine 15a had been formed in 50% yield, accompanied by some glycosidic bond cleavage (2-thiouracil 19 (15%) and isocytosine 20 (7%)). To obtain the ratio of the anhydronucleoside to its phosphorylated derivatives before treatment with alkaline phosphatase, the crude mixture from the phosphorylation of 8 was subjected to analytical high performance liquid chromatography, which showed the ratio of the 15a: 15b: 15c was 3.7 : 1: 1 (Supplementary Figs. 57 and 58). The same phosphorylation conditions were also applied to 2-thiocytidine 5 (Fig. 3a), which afforded 2',2-anhydro-2-thiocytidine 21¹² (43% yield), cytidine 6 (8%), diaminopyrimidine 22 (16%), 2-thiocytosine 23 (8%) and cytosine 24 (8%). All structures were confirmed by spiking experiments, with synthetic standards made in house or purchased (Supplementary Figs. 32 and 35).

Fig. 3. Prebiotic and synthetic routes to (thio)anhydronucleosides.

a, Synthesis of the thioanyhydronucleosides of uridine 15 and cytidine 21 under prebiotic phosphorylation conditions. 2-Thiouridine 8 and 2-thiocytidine 5 both undergo phosphorylation to form cyclophosphate intermediate 18 shown for the uridine variant. These intermediates undergo rearrangement to give structures of type 15 and 21 respectively. 15d was not directly observed, but is inferred from the mechanism. In addition to rearrangement, the nucleosides undergo some cleavage of the glycosidic C–N bond to yield free nucleobases 19 and 23, as well as their hydrolysis/ammonolysis products 20, 22 and 24, respectively. b, Hampton and Nichol’s synthesis of 2',2-anhydrouridine 16 and our use of the same conditions to synthesize the sulfur analogue 15a as a standard. 17 a and 17b are the proposed intermediates leading to the formation of 16 and 15a, respectively. P_i2, Na₂HPO₄/NH₄Cl/NH₄HCO₃.

To further support a cyclic phosphate intermediate, we sought milder conditions for phosphorylation of 8. We chose diamidophosphate (DAP), a reagent known to generate not only the corresponding 2',3'-cyclophosphates from nucleosides, but also short oligonucleotides after prolonged reaction.¹³ Indeed, reaction of 2-thionucleoside 8 with DAP and imidazole under moist-paste conditions at room temperature resulted in formation of the cyclic phosphate 18a after one week (Fig. 4a). Monitoring the phosphorylation reaction by ³¹P, ¹H and ¹³C NMR, we observed the transformation of cyclophosphate 18a to a new species that showed characteristic NMR, electrospray ionization high-resolution mass spectrometry and ultraviolet spectral signatures of anhydronucleotide 15b (Supplementary Figs. 60–63). Furthermore, the DAP-reaction of 8 does not stop at 15b; the 3' phosphate of 15b reacts again with DAP leading to the corresponding 3'-amidodiphosphate of the 2',2-thioanhydronucleoside 25. The total conversion (measured by ¹H NMR) of 8 to 15b/25 was 78% after one month at room temperature. When 4-thiouridine 26 was subjected to the same DAP phosphorylation reaction as a control, the expected cyclophosphate 27 was formed after a few days at room temperature, but no further transformation to the corresponding 2',2-anhydride was observed (Fig. 4a; Supplementary Figs. 66, 67). In this case, cyclisation was presumably prevented due to the lower nucleophilicity of the oxygen on C2 compared with sulfur.

Fig. 4. Phosphorylation of thionucleosides under mild conditions with DAP.

a, Phosphorylation of 2-thiouridine 8 with DAP leads to the formation of thioanhydrouridine-3'-phosphate 15b and corresponding 3'-amidodiphosphate 25 via the intermediate cyclophosphate 18a. The same phosphorylation with 4-thiouridine 26 results in the cyclophosphate 27, which can undergo no further transformation. b, Phosphorylation of 2-thiocytidine 5 with DAP under slightly more forcing conditions affords thioanhydrocytidine-3'-phosphate 21b and corresponding 3'-amidodiphosphate 29, again via an intermediate, cyclophosphate 28.

We also investigated the reaction of DAP (moist paste) with 2-thiocytidine 5, which barely proceeded under the conditions employed for 8. By lowering the pH of the reaction-paste from 7 to 4 and by raising the reaction temperature to 60 °C, we were able to detect cyclic phosphate 28, and later were able to identify the corresponding 3'-phosphate and 3'-amidopyrophosphate of thioanhydrocytidine (21b/29) by ³¹P NMR (Fig. 4b; Supplementary Figs. 64, 65). An accurate conversion was difficult to determine by ¹H-NMR, but the combined yield for 28, 21b and 29 was estimated at slightly greater than 10%. The greater reactivity of 2-thiouridine 8 towards DAP compared to 2-thiocytidine 5 likely stems from its greater solubility and pK_a near neutrality (8.0, N3–H)².

C2' Reduction

Our syntheses of thioanhydronucleosides 15 and 21 from 2-thionucleosides 8 and 5 are the first prebiotically plausible routes to nucleoside derivatives containing a sulfur atom connected to C2'. Such compounds are of significant interest in the origins-of-life field, because the potential conversion of a C–S to a C–H bond (a chemically facile process¹⁴,¹⁵) would result in the synthesis of 2'-deoxynucleosides – components of DNA. In modern biology, ribonucleotide reductases convert ribonucleotides to deoxyribonucleotides¹⁶. This enzymatic reduction proceeds via a radical mechanism with a dithiol as the stoichiometric reductant. We wondered if a mechanism involving sulfur-based reductants might have led to the formation of 2'-deoxyribonucleosides/tides on the early earth, in the absence of complex enzymes, from 2'-thio-functionalised ribonucleosides. Nagyvary and coworkers¹⁷–¹⁹ first experimentally investigated this idea by synthesizing 2'-thiocytidine (using an enzyme) and studying a variety of reduction procedures, including photoreduction. However, the limited published characterization data for the supposed 2'-thiocytidine substrate have since been pointed out as apparently inconsistent with the structure²⁰. Recently, Powner and coworkers²¹ revisited the idea with a proposal to modify the prebiotically plausible pyrimidine synthesis⁵,²² by including a sulfur atom attached to C2' from the outset. This route proved unsuccessful because the sulfur-containing building blocks exhibited different reactivity to analogues from the pyrimidine synthesis. Indeed, while a route from pure ribose or arabinose to deoxyribose has recently been established²³, a plausible abiotic synthesis of 2'-deoxynucleosides themselves – whether by desulfurization or other means – has not yet been reported. 2',2-thioanhydronucleosides 15a and 21a were therefore subjected to ultraviolet irradiation at 254 nm in the presence of aqueous hydrogen sulfide as a reductant (Fig. 5a). Flux measurements from the same experimental setup have been carried out within our group previously, from which we determined a flux rate of 2.5×10¹⁶ cm^-2s^-1 (19 mWcm^-2) for the 254-nm-centred emission²⁴. Inspection of ¹H-NMR spectra showed that, under these conditions, thioanhydrocytidine 21a gave 2-thiocytosine 23 in 29% yield after 3 h of irradiation. However, thioanhydrouridine 15a afforded two new nucleoside derivatives as well as thiouracil 19 as major products. Promisingly, one nucleoside showed two 2'-H signals shifted upfield (2.64 ppm and 2.32 ppm) in the ¹H NMR spectrum (Supplementary Fig. 38), strongly suggesting that a deoxyribosyl moiety was present. Comparison of NMR spectra with literature values for deoxyribosyl pyrimidines suggested the formation of 2'-deoxy-2-thiouridine 11. We subsequently verified this assignment by synthesizing 11 via a literature route²⁵ and using this material to spike our samples (Supplementary Fig. 40). We speculate that the second nucleoside product is 2'-α-thio-2-thiouridine 30, but attempts to isolate and characterize this compound have so far been thwarted by its facile decomposition to 2-thiouracil 19 as the only identifiable product. The instability of the glycosidic bonds of 2'-thioribonucleosides has previously been documented²⁶. Addition of the NMR integration standard, pentaerythritol, to the crude mixture indicated yields of 33, 25 and 28% for 11, 19 and presumed 30, respectively. The 5'- and 3'-phosphates of 2',2-anhydro-2-thiouridine (15b and 15c, respectively) undergo the same photoreduction, in similar yields to 15a (Supplementary Figs. 41 and 42).

Fig. 5. Photoinduced reduction of thioanhydrouridine with hydrosulfide (HS^–) and its proposed mechanism.

a, Photoreduction of 15a gives the major product 2'-deoxy-2-thiouridine 11, as well as 2-thiouracil 19, which is possibly formed via facile nucleobase loss from proposed intermediate 30 (see ref. ²⁶). Nucleotides 15b and 15c–minor products of the phosphorylation of 8 in Fig. 3–undergo photoreduction under the same conditions, in similar yields (Supplementary Figs. 41 and 42). b, Our proposed mechanism is initiated by photodetachment of an electron from aqueous hydrosulfide to generate a solvated electron. Reduction of 15a by this electron generates radical anion 31, which undergoes C–S bond homolysis to generate stabilised radical anion 32. 32 can react with a hydrogen atom donor or combine with a HS• radical to give major isolated product 11 or proposed intermediate 30, respectively, which we observed in crude spectra but were unable to isolate.

A possible mechanism for this remarkable photoreduction is proposed in Figure 5b. Separate control reactions in the dark with H₂S and irradiation without H₂S gave no reduction products nor 30, suggesting that initiation is by an electron released from hydrosulfide under ultraviolet irradiation⁴,²⁷,²⁸, rather than homolysis of the C–S bond of 15a. Transfer of this electron to thioanhydrouridine 15a to form the radical anion 31, and subsequent homolysis of the C–S bond would give intermediate 32. The newly formed C2' radical can abstract a hydrogen atom from H₂S (pictured) or react with a hydrogen atom (not pictured), to give 11. Alternatively, reaction with an HS^ߦ radical equivalent (HS^- with loss of an electron, or H₂S₂ or a HS^ߦ radical, generated in previous steps) forms 2'-thionucleoside 30. We conjecture that 21 probably does not undergo this reduction because the first step, were it to occur, would lead to a radical anion intermediate with less resonance stabilization than 31.

Syntheses of deoxyribose and β-2'-deoxyadenosine

Since our synthesis of 2'-deoxyribonucleoside 11 takes place under the same conditions under which amino acid⁴ and sugar²⁹ precursors and RNA pyrimidines⁵,²² are formed, we immediately questioned whether DNA building blocks such as 11, and related (canonical) derivatives, may also have been present at the advent of life. In an attempt to form the canonical 2'-deoxyribonucleoside, deoxyuridine, we investigated the hydrolysis of 11 under the same conditions under which β-2-thioribocytidine 5 is converted to cytidine (Fig. 1)⁵. However, in the presence of phosphate buffer, 11 is cleanly hydrolysed to 2-deoxyribose 12 and thiouracil 19. A detailed study of the hydrolysis of 2'-deoxy-2-thiouridine 11 at 60 °C revealed half-lives of 32 and 31 hours for 2'-deoxy-2-thiouridine 11 in 0.27M phosphate (pH 7.0) and acetate (pH 4.1) buffers, respectively, and 28 hours in unbuffered solution (Fig. 6a and Supplementary Figs. 43–50). The complete and clean conversion of 2'-deoxy-2-thiouridine 11 to 2-thiouracil 19 and deoxyribose 12 suggests that 11 may have been a prebiotic source of deoxyribose, both in abiotic reactions and/or for early enzymes, akin to the use of activated ribose in modern phosphoribosyltransferases³⁰–³². This reactivity of 11 presented an opportunity for a plausible synthesis of purine deoxynucleosides, via transglycosylation³³–³⁵. Indeed, heating of 11 with excess adenine 33 in the dry state³⁶ at 100 °C for 31 h led to the formation of both alpha and beta isomers of 2'-deoxyadenosine 13, in 6% and 4% yield respectively, accompanied by nucleobase loss to give 2-thiouracil 19 in 72% yield (Fig 6b and Supplementary Figs. 54 and 55). Notably, replacement of 11 with 2-deoxyribose 12 led only to a complex mixture that did not contain β-2'-deoxyadenosine (Supplementary Fig. 56). Such a demonstration of a plausible abiotic pathway stemming from the ribo-pyrimidine synthesis to DNA is significant, and further efforts are underway in our laboratory to discover more favourable transglycosylation conditions. Additionally, if promiscuous proto-enzymes with ribosylase activity existed, 11 may have been an efficient source of deoxyribose in an early biosynthesis of deoxyribonucleotides via transglycosylation. Thus, there are various plausible pathways by which the C2'-reduced nucleoside 11 could have facilitated the chemical progression of RNA into DNA.

Fig. 6. Hydrolysis and transglycosylation reactions of 2'-deoxy-2-thiouridine 11.

a, Hydrolysis of 11 at different pH values gives deoxyribose 12 and the corresponding nucleobase, 2-thiouracil 19. b, Transglycosylation in the dry state with adenine 33 gives a mixture of the alpha– and beta– stereoisomers of the canonical deoxynucleoside, deoxyadenoisine 13, as well as nucleobase loss to yield 19.

Conclusions

Our investigation into the phosphorylation of prebiotic RNA pyrimidine intermediate 8 has resulted in several significant discoveries. We have shown that 5'-phosphoryl-2-thiouridine 14 can be obtained from kinetically selective phosphorylation of 8, demonstrating its viability as a potentially important primordial non-canonical component of nucleic acids. Other phosphorylation conditions lead to the until-now elusive prebiotic functionalization of the C2' position of a nucleoside with sulfur. Moreover, we successfully reduced such a sulfur-containing derivative to a non-canonical 2'-deoxyribonucleoside, 11, using a prebiotically plausible, hydrogen sulfide-mediated photoreduction. This photochemical reaction emulates the critical biochemical conversion of ribonucleotides to 2'-deoxyribonucleotides, but without the involvement of any complex enzymes. Finally, we have demonstrated that this deoxyribonucleoside reacts with adenine to form β-deoxyadenosine 13b, a canonical DNA purine deoxynucleoside. Thus, 2-thioribonucleosides are key intermediates in a synthesis of both RNA and DNA building blocks, providing a long sought-after prebiotic link between DNA and its biological and chemical progenitor, RNA.

Supplementary Material

Supplementary information is available for this paper at https://doi.org/10.1038/s41557-019-0225-x.