A Nonenzymatic Analog of Pyrimidine Nucleobase Biosynthesis.

Metabolic theories for the origin of life posit that inorganic catalysts enabled self-organized chemical precursors to the pathways of metabolism, including those that make genetic molecules. Recently, experiments showing nonenzymatic versions of a number of core metabolic pathways have started to support this idea. However, experimental demonstrations of nonenzymatic reaction sequences along the de novo ribonucleotide biosynthesis pathways are limited. Here we show that all three reactions of pyrimidine nucleobase biosynthesis that convert aspartate to orotate proceed at 60 °C without photochemistry under aqueous conditions in the presence of metals such as Cu2+ and Mn4+ . Combining reactions into one-pot variants is also possible. Life may not have invented pyrimidine nucleobase biosynthesis from scratch, but simply refined existing nonenzymatic reaction channels. This work is a first step towards uniting metabolic theories of life's origin with those centered around genetic molecules.

Thinking about the chemical origins of life is often divided along “genetics‐first” and “metabolism‐first” hypotheses. The “genetics‐first” approach assumes that self‐replicating genetic polymers, likely RNA, emerged directly from a prebiotic chemistry bearing little resemblance to life's biosynthetic pathways. From that perspective, early prebiotic chemistry is simply a tool to furnish building blocks for a later self‐organized process. A central challenge for this vision of prebiotic chemistry is the direct, robust and high‐yielding synthesis of ribonucleotides or deoxyribonucleotides by any plausible means. Experimental work along these lines has been arguably the major thrust of the field for the past 50 years, with recent experimental work from Sutherland, Carell, Powner, Benner, and Trapp.

However, without overlooking the undeniable importance of ribonucleotides and RNA in the origin of life, starting in the late 1980s conceptual difficulties with the “genetics‐first” approach prompted “metabolism‐first” proposals for the origin of life.[ , , , ] This approach considers the origin of life to have involved the spontaneous onset of a self‐organized reaction network, which was driven into existence by a thermodynamic need to relax geochemical redox gradients. The initial catalysts enabling the self‐organized reactivity would have been minerals, clays and metals, but certain organics produced by the network could additionally feed back as catalysts to reinforce the existing reactions and to enable new ones.[ , , ] Such a reaction network would have been historically continuous with biological metabolism. Furthermore, because complex systems, such as metabolism, are difficult to disrupt and rewrite, certain energetic and chemical similarities should exist with metabolism today. From this alternative perspective, ribonucleotides and their oligomers would have been produced by the prebiotic reaction network, possibly initially in small quantities, but gained prominence within it because they offered a critical catalytic or regulatory function within the existing self‐organized chemistry. Viewed from this angle, a central challenge of prebiotic chemistry is to figure out what the conditions for such self‐organized chemistry would have been. One way towards this goal is to search for conditions that recapitulate nonenzymatic variants of core metabolic reactions and processes,[ , ] including those for the synthesis of genetic molecules. Multiple, possibly very different, conditions might be found to enable any particular reaction or pathway without enzymes, and some of these conditions may have no relevance to the origin of life. However, this strategy should eventually identify conditions that enable a larger system to self‐organize. Experimental efforts in this direction have thus far focused on metal‐catalyzed analogues of carbon fixation pathways,[ , , , , ] amino acid[ , , ] and cofactor[ , ] biosynthesis, and sugar metabolism.[ , , , ] Investigations into purely chemical conditions that trigger reactions of ribonucleotide biosynthesis in a prebiotic context have been limited, as we will discuss below.

Ribonucleotides are biosynthesized in two general ways: by salvage pathways that regenerate ribonucleotides after they degrade, and by de novo pathways that build them up from scratch starting from amino acids, C1 compounds and phosphoribosylpyrophosphate (PRPP), an activated form of ribose‐5‐phosphate. The first three steps of de novo pyrimidine ribonucleotide biosynthesis are illustrated by the black arrows in Scheme 1. First, aspartate (ASP) reacts with carbamoyl phosphate (CAP) to give carbamoyl aspartate (CAA), catalyzed by carbamoyl phosphate synthetase. CAA then undergoes a dehydrative cyclization to give dihydroorotate (DHO) catalyzed by dihydroorotase. DHO is oxidized to orotate (ORO) by various electron‐accepting cofactors, catalyzed by dihydroorotate dehydrogenase. Later on in the pathway, not illustrated in Scheme 1, ORO couples with PRPP to give orotidine 5′‐monophosphate, and, in subsequent steps, the canonical pyrimidine ribonucleotides uridine‐5′‐monophosphate and cytidine‐5′‐monophosphate. The conversion of ASP to ORO is therefore a key biosynthetic bottleneck in the formation of pyrimidine ribonucleotides.

Scheme 1

The first three steps of de novo pyrimidine ribonucleotide biosynthesis convert aspartate (ASP) to orotate (ORO), as shown by the black arrows. Additional reactions observed in this report are described by blue arrows.

Much prebiotic chemistry has been described that conceptually resembles ribonucleotide biosynthesis, but little that involves precisely the same reactions and substrates. However, if metabolism is derived from self‐organized chemistry, deviations from the reactions or substrates it uses, even if they lead to a biological product, might lack relevance to prebiotic chemistry. As for reports that specifically recapitulate ribonucleotide salvage pathways, Jaber, Georgelin and co‐workers examined the nonenzymatic formation of adenosine monophosphate from adenine and phosphoribosylpyrophosphate (PRPP). Chemistry that roughly resembles de novo pyrimidine nucleobase biosynthesis has also been reported, but not for the biological substrates along the entire path.[ , , , ] Yamagata used urea, rather than CAP, and UV light, rather than a chemical oxidant, at 80–90 °C. In that study, the photooxidation of DHO to ORO was particularly challenging, reaching <1 % yield after 2.5 days and declining thereafter. It still remains unclear whether all the biological transformations of pyrimidine ribonucleobase biosynthesis can occur without enzymes and, crucially, without photochemistry, as none of the nonenzymatic analogues of the other pathways described above depend on light.[ , , , , , , , , , , , , , ] Here we examine these first three key steps experimentally and find that they all occur nonenzymatically in water at 60 °C without the need for UV light.

We began our investigations by exploring the nonenzymatic reaction between ASP, derived from the non‐enzymatic reductive amination of oxaloacetate, and CAP to give CAA. Previously, CAP and CAA have been obtained in a prebiotic context starting from cyanate or from urea (via cyanate),[ , , , ] however the use of CAP to prepare CAA, as in the biological pathway, has not yet been studied. The reaction between ASP (8.5 mM) and CAP (8.5 mM) was assayed over a pH range of 3–9 and a temperature range of 0–60 °C. Acidic conditions were found to be ineffective. The optimal conditions were found to be pH 8 and 60 °C, giving 43 % yield after 16 h, as determined by quantitative 1H NMR by integrating against dimethyl sulfone (DMS) as an internal standard. The identity of the product was additionally confirmed by LC‐QTOF‐MS analysis of the reaction mixture through comparison with an authentic sample. The use of a two‐fold excess of CAP relative to ASP gave CAA in 63 % yield after 16 h (Figure 1A). When the initial concentration of ASP was increased from 8.5 mM to 21 mM (still with a 2 : 1 ratio of CAP to ASP), the yield of CAA increased to 77 %. Full optimization details can be found in Table S1. From a mechanistic standpoint, cyanate is known to be produced, along with phosphate, by thermal fragmentation of CAP and is therefore a likely intermediate in the reaction.[ , ] In line with this proposal, the use of pH 8 phosphate buffer, rather than water adjusted to pH 8, decreased the yield of CAA to 10 % (Table S1, entry 10), which might be explained by a less favorable equilibrium between CAP and cyanate in the presence of excess phosphate.

Figure 1

Study of single reaction steps. A) 1H NMR spectrum of the products of reaction of aspartate ASP and carbamoyl phosphate CAP. δ, chemical shift. B) 1H NMR of the products of cyclization reaction of carbamoyl aspartate CAA. C) 1H NMR of the products of oxidative reaction of dihydroorotate DHO. D) 1H NMR of the products of oxidative reaction of hydantoin‐5‐acetate (HAA). Yields were determined by quantitative 1H NMR with water suppression techniques, integrating against dimethyl sulfone as an internal standard.

Encouraged by this result, we next explored the nonenzymatic dehydrative cyclization reaction of CAA to DHO. It should be noted that this reaction is endergonic and at pH 7.4 at 37 °C the enzyme catalyzed process favors CAA over DHO by a factor of 16.6 at equilibrium. Thus, low yields of DHO are to be expected under similar conditions. The reaction was assayed over a pH range of 1–8 and a temperature range of 20–60 °C. The products were quantified by 1H NMR and confirmed by LC‐QTOF‐MS in the same way as described earlier. At pH values near the lower end of this range, the reaction did not produce any DHO, but instead gave rise to the 5‐membered product, hydantoin‐5‐acetate (HAA). This result is similar to that observed in a non‐prebiotic context during the total synthesis of orotic acid. A new screen was therefore conducted across the same range of conditions but in the presence of a panel of metal salts and oxides. Zn2+, Cu2+ and Fe3+ produced a small amount of DHO at the higher pH values, with Cu2+ being the most selective for DHO over HAA. For example, the reaction of CAA with CuSO4 (40 mol %) at 60 °C, pH 8 after 16 h gave DHO in 1 % yield as the only observed cyclization product (Figure 1B). At lower pH values the DHO obtained increased, but HAA became the major product due to the acid‐catalyzed background reaction (Table S2). For example, at pH 2 in the presence of Cu2+, DHO was obtained in 3 % yield together with 25 % of HAA. Increasing the initial concentration of the starting material to 21 mM improved the yield of both products, giving 4 % DHO and 40 % HAA. We note that Cu2+ has proven an intriguing additive in a number of different prebiotic contexts.[ , ]

We next examined the oxidation of DHO to ORO. Previously, a nonenzymatic version of this reaction had only been observed in <1 % yield under direct UV irradiation. A screen was devised to assay oxidants over a pH range of 4–13 and a temperature range of 20–60 °C. At 60 °C, two optimal oxidants were found to be H2O2 and MnO2, the prebiotic plausibility of which was recently discussed.[ , ] In the case of H2O2 as oxidant, the reaction worked best at pH 9, giving 34 % yield of ORO after 16 h. In the case of MnO2 as oxidant, at pH 4, ORO was obtained in 18 % yield, together with 2 % of uracil (URA) after 16 h. Next, we studied the influence of metal additives on the oxidation. The presence of metal ions inhibited the H2O2 promoted oxidation to ORO. For example, the presence of Cu2+ reduced the yield to <3 % ORO, together with 4 % URA. However, in the case of the MnO2 promoted oxidation, a screen of metal ion additives (Fe2+, Co2+, Ni2+, Cu2+) revealed that the presence of Cu2+ (40 mol %) enhanced the outcome, giving ORO in 51 % yield together with 2 % URA (Figure 1C). The other metal co‐catalysts assayed did not display any enhancing effects (Table S3). Cu2+ is thus capable of influencing both the cyclization and oxidation steps on the nonenzymatic version of the biological path to ORO.

Given that the cyclization of CAA to DHO occurred in highest yield at lower pH, where it is accompanied by HAA as major product (Table S2), we wondered whether HAA might also be converted to ORO under similar oxidative conditions. Could HAA be oxidized to 1 under plausibly prebiotic conditions, which could then undergo a known rearrangement under alkaline aqueous conditions to give ORO? Exposing HAA to H2O2 did not result in its oxidation under the conditions we tested. In contrast, exposing HAA to MnO2 in the presence of Cu2+ (40 mol %) at 60 °C for 16 h indeed resulted in its oxidation to 1 (characteristic singlet at 5.4 ppm in 1H NMR) and to URA (5 %) (Table S4). Basifying the reaction mixture with NaOH and continued heating for a further 3 h resulted in a decrease in concentration of 1 due to its conversion to ORO (11 %) along with hydrolysis to CAA (Figure 1D). Thus, the adventitious conversion of CAA to HAA is not necessarily a dead end, as it can be funneled towards the same end‐product, ORO (Scheme 1, blue arrows).

In order to assess the feasibility of a non‐photochemical one‐pot process mimicking pyrimidine nucleobase biosynthesis, we first attempted to merge the dehydrative cyclization and oxidation steps. Heating CAA in the presence of Cu2+ at 60 °C, followed by the addition of MnO2 at 16 h and NaOH at 32 h gave ORO in 2 % yield at 35 h. It was later found that the yield could be improved to 13 % NMR yield (24 % by LC‐QTOF‐MS) of ORO by lowering the initial reaction pH, along with a 1 % yield of URA and 13 % of unreacted CAA (Figure 2A). This outcome is in agreement with the aforementioned influence of pH on the cyclization of CAA to DHO and HAA. Next, we examined a one‐pot process for the conversion of ASP and CAP to ORO. Heating ASP and CAA (1 : 2) at 60 °C at pH 8, followed by the addition of Cu2+ and HCl at t=16 h, MnO2 at t=32 h, and NaOH at t=48 h gave ORO in 1 % NMR yield (4 % by LC‐QTOF‐MS) along with 14 % CAA at t=54 h. Intermediate 1 is notably observed by 1H NMR in both of the one‐pot experiments described in Figure 2, indicating that the pathways proceeding through DHO and through HAA are both occurring. It was also possible to obtain ORO from ASP in one pot by adding all the reagents from the beginning apart from the basification of the reaction at t=16 h with NaOH, however, the yield decreased to <1 % (Table S6, entry 2). NaOH must not be added from the beginning as DHO rapidly decomposes in its presence. Control experiments indicate that the decrease in yield observed when all reagents (other than NaOH) are present from the beginning is due to the decomposition of CAP by Cu2+ (Table S1, entry 6).

Figure 2

One‐pot formation of pyrimidine bases. A) 1H NMR and LC‐ QTOF‐MS of the products of multi‐step reaction of carbamoyl aspartate (CAA). B) 1H NMR and LC‐QTOF‐MS of the products of multi‐step reaction of aspartate, (ASP) and carbamoyl phosphate (CAP). [M−H] at m/z=155.0098 was chosen as characteristic fragment for ORO, [M−H] at m/z=111.0196 was chosen as characteristic fragment for URA.

In conclusion, we have uncovered a non‐photochemical nonenzymatic analog of pyrimidine nucleobase biosynthesis in which some steps are promoted by metals. Some of the steps are inhibited by metals or by unfavorable pH values, which explains why the one‐pot version of this chemistry was helped by timed addition of metals or timed changes in pH. Although pH gradients do exist in certain natural environments, such as hydrothermal vents, we feel it is premature to place this chemistry in a specific geological location. Whether or not the particular set of conditions described here are directly relevant for the origin of life, they nonetheless constitute a proof of principle that this key biological process on the path to genetic molecules could have existed even before enzymes had evolved to catalyze it. In addition to the specific reactions of the biosynthetic pathway shown here, parallel nonenzymatic reactions not found in biology also occur but still eventually arrive at the same biological endpoint: orotate. The conversion of ASP to ORO does not necessarily require metals, especially if a potentially prebiotic alternative oxidant could be found. However, at least at 60 °C in aqueous solution, only the metal‐mediated reaction sequence follows the de novo biosynthetic pathway, passing through DHO. It is this path that carries greater explanatory value when viewed in the framework of the hypothesis that nonenzymatic reactions templated the evolution of the biological pathway.[ , , , , , , , , , ] Evolutionary refinement of the redundant branched nonenzymatic reaction network shown in Scheme 1 (both black and blue arrows) would logically result in a streamlined linear pathway like the one found in biology today (black arrows only). In line with this idea, some enzymes still catalyze an equilibrium between CAA, HAA and DHO, even though there appears to be no biological use for HAA. The present work can be viewed as a first experimental step toward incorporating genetic molecules into a self‐consistent metabolic framework for the origin of life. Preliminary experiments indicate that the N‐carbamoylation of ASP with CAP is not selective over other amino acids (Figure S7). However, the potential implications of this observation to a self‐organized prebiotic chemistry depends on many factors which are at present difficult to constrain. Future work should investigate to what extent a one‐pot nonenzymatic version of this pathway can occur without controlled changes to reaction conditions, and whether the nonenzymatic assembly of pyrimidine ribonucleotides can occur in ways paralleling their de novo biosynthesis.

Conflict of interest

The authors declare no conflict of interest.

As a service to our authors and readers, this journal provides supporting information supplied by the authors. Such materials are peer reviewed and may be re‐organized for online delivery, but are not copy‐edited or typeset. Technical support issues arising from supporting information (other than missing files) should be addressed to the authors.

Supporting Information

Click here for additional data file.