Cyanide as a primordial reductant enables a protometabolic reductive glyoxylate pathway

Investigation of prebiotic metabolic pathways is predominantly based on abiotically replicating the reductive citric acid cycle. While attractive from a parsimony point of view, attempts using metal/mineral-mediated reductions have produced complex mixtures with inefficient and uncontrolled reactions. Here we show that cyanide acts as a mild and efficient reducing agent mediating abiotic transformations of tricarboxylic acid intermediates and derivatives. The hydrolysis of the cyanide adducts followed by their decarboxylation enables the reduction of oxaloacetate to malate and of fumarate to succinate, whereas pyruvate and α-ketoglutarate themselves are not reduced. In the presence of glyoxylate, malonate and malononitrile, alternative pathways emerge that bypass the challenging reductive carboxylation steps to produce metabolic intermediates and compounds found in meteorites. These results suggest a simpler prebiotic forerunner of today’s metabolism, involving a reductive glyoxylate pathway without oxaloacetate and α-ketoglutarate—implying that the extant metabolic reductive carboxylation chemistries are an evolutionary invention mediated by complex metalloproteins. It’s unclear how protometabolic reactions emerged and evolved into extant metabolic pathways such as the tricarboxylic acid cycle. Now, it has been shown that cyanide acts as a mild and efficient reducing agent, mediating abiotic transformations of tricarboxylic acid intermediates and derivatives.

R esearch into the emergence of protometabolism as the primary driver of the origin of life is guided by the hypothesis that a non-enzymatic version of the reductive citric acid (r-TCA, also known as the reductive tricarboxylic acid or reverse Krebs) cycle functioned on the early Earth (cycle denoted by pink arrows, Fig. 1a) 1,2 . Recently it has been shown that the 'irreversible' citrate synthase reaction can also proceed in the reverse direction in certain anaerobic organisms and that the presence of large amounts of CO 2 enables this reversal as well [3][4][5] , further strengthening the case for a primordial version of the r-TCA cycle. However, there is also the 'lesser known' glyoxylate shunt pathway 6,7 , which involves the retro-aldol reaction of isocitrate to form glyoxylate and succinate, which are then incorporated into the oxidative tricarboxylic acid (TCA) pathway (also known as the citric acid cycle or Krebs cycle), completely bypassing α-ketoglutarate (green arrows, Fig. 1a). Though this oxidative glyoxylate shunt has been proposed as a precursor to the modern oxidative TCA cycle 8 , the focus still has remained on the reductive pathways due to the ancientness of the r-TCA pathway and the anoxic early Earth environment 2,9 . A simplified version of the extant r-TCA cycle, the proposed primordial r-TCA cycle is a metal-mediated carbon-fixation pathway that is able to reductively fix CO 2 by reacting with small organic molecules to produce higher-order bioorganics [10][11][12][13] . While the merits and disadvantages of this approach have been discussed from many angles 1,14 , the experimental demonstration of viable metal-mediated CO 2 incorporation and reductive prebiotic TCA transformations has been challenging 11 . The reactions are difficult to control and produce low-yielding complex mixtures via the decomposition of the intermediates formed with no appreciable accumulation of the components of the TCA pathway 11 . As an alternative to CO 2 as the primary organic source, cyanide has been investigated as a primary prebiotic source molecule for carbon and nitrogen compounds and shown to form a plethora of biologically relevant building blocks [15][16][17][18] . Thus, cyanide has the additional potential to react further with downstream products to form nitriles, enabling further chemical transformations [17][18][19][20][21] . Herein we investigate the role of cyanide-mediated transformations and show that cyanide can act as a reductant 22 that not only enables efficient transformation of TCA intermediates, but also makes possible a reductive glyoxylate shunt as a hypothetical abiotic pathway that could have operated on early Earth.
We hypothesized, based on previous work 23,24 , that cyanide could play the role of a reductant 22 in the emergence of protometabolic pathways by directly interacting with molecules of the TCA pathway to form the corresponding nitriles. The hydrolysis of these nitrile intermediates could be harnessed to enable otherwise difficult reactions-such as the reduction of unsaturated functional groups (carbonyl or olefinic) and aldol condensations of unactivated carboxylic acids. In this context we were guided by a combination of the inventory of the carboxylic acid and metabolic precursors detected in carbonaceous meteorites 25 and cyanide interactions with α-, βand γ-ketoacids to yield a suite of hydroxy acids 24 . Specifically, the cyanide-mediated dimerization of glyoxylate 1 to form tartrate 2 at pH ≈ 5 suggested a general paradigm of an overall reduction of a carbonyl bond to an alcohol through a pathway involving (1) cyanide addition to a carbonyl group to form cyanohydrin, followed by (2) its subsequent hydrolysis to α-hydroxy acid, followed by (3) decarboxylation (Fig. 1b) 24 . This addition-hydrolysis-decarboxylation sequence of reactions can be an alternative prebiotic-reduction pathway of other α-ketoacids and unsaturated functional groups in the TCA cycle (Fig. 1a).
Investigation of prebiotic metabolic pathways is predominantly based on abiotically replicating the reductive citric acid cycle. While attractive from a parsimony point of view, attempts using metal/mineral-mediated reductions have produced complex mixtures with inefficient and uncontrolled reactions. Here we show that cyanide acts as a mild and efficient reducing agent mediating abiotic transformations of tricarboxylic acid intermediates and derivatives. The hydrolysis of the cyanide adducts followed by their decarboxylation enables the reduction of oxaloacetate to malate and of fumarate to succinate, whereas pyruvate and α-ketoglutarate themselves are not reduced. In the presence of glyoxylate, malonate and malononitrile, alternative pathways emerge that bypass the challenging reductive carboxylation steps to produce metabolic intermediates and compounds found in meteorites. These results suggest a simpler prebiotic forerunner of today's metabolism, involving a reductive glyoxylate pathway without oxaloacetate and α-ketoglutarate-implying that the extant metabolic reductive carboxylation chemistries are an evolutionary invention mediated by complex metalloproteins. malate 4a quantitatively (Fig. 1a). The spontaneous hydrolysis of the cyanide moiety at room temperature is consistent with previous observations of 1,5-anchimeric assistance provided by the β-carboxyl group 24 . Heating at 50 °C hydrolysed amide 4a to 2-carboxymalate 4b, which underwent slow decarboxylation that was accelerated by Zn 2+ to produce malate 5 as the major product (Extended Data Fig. 1). This conversion, mimicking the reductive step of oxalacetate to malate in the r-TCA pathway, also works at moderate pH values (5)(6) and various concentrations (100-500 mM) in good yields (68-75% conversion; Supplementary Table 1 and Supplementary Figs. 1-4). A reaction at 10 mM of 3 in the absence of Zn 2+ gave greater than 90% conversion to 4a + 4b ( Supplementary Fig. 5) suggesting that the efficient trapping of oxaloacetate 3 by cyanide is responsible for mitigating the side reaction of the decarboxylation of 3 to pyruvate 16.
We then turned to the more challenging reduction of the double bond of fumarate 6 to succinate 10. We observed that the polarization of the C=C double bond of fumarate as the monoanion enabled cyanide addition 24,26 (Supplementary Table 2). Reacting cyanide with 6 at pH ≈ 4 and 80 °C led to 40% of 10 (Extended Data Fig. 2a and Supplementary Figs. [6][7][8]. Even K 4 Fe(CN) 6 acted as a source of cyanide to enable the conversion of fumarate to succinate under similar conditions (21%; Supplementary Fig. 9). We discovered another pathway to succinate 10 through nitrile and cyanide chemistry, starting from the hydrolysis products of fumaronitrile 17 , fumaramide 7 and its half-amide derivative 8. In the presence of cyanide, fumaramide 7 was partially hydrolysed to produce 8 resulting in concomitant addition to the C=C bond of 8 (Extended Data Fig. 2b Lactone 26a Mixture of products and intermediates Tricarballylic acid is one of the poly-carboxylic acids detected in meteorites 25 , reinforcing the potential role of nitriles in prebiotic chemistry 16 . Transformation of tricarballylic acid to citrate 14 on meteorites has been proposed 25,27 . As a proof of principle, we reacted tricarballylic acid with an oxidant (K 2 S 2 O 8 ) 28 and observed the formation of citrate 14 (33%; Supplementary Figs. 17 and 18). Reaction of cyanide with α-ketoglutarate 11 and pyruvate 16 formed the corresponding cyanohydrins 18 and 17, respectively, but did not proceed further 24 , suggesting that these cyanohydrins could act as 'stable' reservoirs and regenerate these α-ketoacids 29 . This observation when compared to the cyanide-induced transformation of oxaloacetate 3 to malate 5 and that of fumarate 6 to succinate 10 (Fig. 1a) alleviates a major concern 14 that these same abiotic processes would interfere by irreversibly reducing pyruvate and α-ketoglutarate to their corresponding α-hydroxy acids. Taking a cue from the cyanide reaction with α-ketoglutarate 11, we reinvestigated 24 the reaction of cyanide with β-ketoglutarate 19 (ref. 25 ; Extended Data Fig. 3). This reaction without any pH adjustment at room temperature produced citric acid amide 19a, which on heating resulted in citrate 14 (65%) along with citramalate 19b (Extended Data Fig. 3  Cyanide-mediated transformations of α-ketoacid analogues of TCA cycle give unexpected non-oxidative decarboxylation. A recent report described the formation of α-ketoacid analogues of the TCA cycle from simply mixing pyruvate and glyoxylate without the addition of transition metals (Fig. 2) 30 . Since these were non-canonical α-ketoacid analogues, we wondered if their interaction with cyanide would follow the same trends we observed with the intermediates of the TCA cycle (Fig. 1a). Therefore, we investigated the interaction of these α-ketoacid analogues with cyanide under the conditions described in Fig. 1 Fig. 2, Extended Data Fig. 4 and Supplementary Figs. 39 and 40). The formation of succinate indicates that the addition of cyanide to the keto-carbonyl group of 21 (instead of addition at the conjugated double bond) has facilitated a decarboxylation that leads to the simultaneous reduction of the double bond (Fig. 2), bypassing the formation of fumarate 6 (which is formed by the oxidative decarboxylation of 21 (ref. 30 )). Similarly, when cyanide was reacted with aconitoyl formate 24, under basic (or acidic) conditions, formation of tricarballylic acid 15 was observed by the same mechanism (26%; Supplementary  Figs. [41][42][43][44][45][46]. This cyanide-catalysed non-oxidative decarboxylation leading to an overall reductive transformation of 21 and 24 indicates that the structural similarity of the extended double bond conjugation with the keto group in 21 and 24 is responsible for deviating from a simple cyanohydrin adduct formation.
The results described so far are notable in that they enable transition-metal-free, efficient and selective reductive transformations of TCA intermediates, and they address some of the problems raised in arguing against the existence of prebiotically plausible catalysts that would be available on early Earth to facilitate such selective transformations for an abiotic r-TCA-type pathway to operate 14,31 .
Here, cyanide acts as the stoichiometric reductant, with the selection pressure coming from the substrates themselves: only those that have a 1,5-disposition with the carboxylate or the hydroxyl oxygen are transformed, while the rest are 'unreactive' . This transformation is instructive of an intrinsic selection where "the instruction required for a preferential transformation comes from within the reaction system itself ", without the need for an external catalyst 32 .
Towards a protometabolic reductive glyoxylate pathway. The results reported above, however, by themselves do not provide support for the functioning of the r-TCA cycle in a prebiotic setting. There is still the question of whether reproducing the abiotic r-TCA cycle itself is necessary or whether there could be simpler protometabolic pathways that could provide the framework of reactions and molecules for evolution towards the r-TCA pathway. The formation of carboxysuccinate 9, the intermediate formed from fumarate 6 on the way to making succinate 10 (Fig. 1a), provided an opportunity to explore this question. We observed that carboxysuccinate 9 could undergo a condensation reaction with glyoxylate 1, whereas succinate did not (Fig. 3a) 33,34 . Thus, reaction of carboxysuccinate 9 with glyoxylate 1 led to a mixture of carboxy isocitrate 27 and diastereomers of isocitrate 12 (and their corresponding lactones; Supplementary Figs. [47][48][49][50][51]. Further heating at 50 °C afforded 93% of the threo-and erythro-diastereomers of isocitrate 12 and its corresponding lactones (Fig. 3a and Extended Data Fig. 5). The highly efficient and near-stoichiometric conversion of 9 to 12 is important in the context of the (oxidative) glyoxylate shunt pathway 7 (Fig. 1) wherein isocitrate 12 is fragmented to form glyoxylate 1 and succinate 10 (eventually forming citrate) but without the involvement of α-ketoglutarate 11. What we have in Fig. 3a is an equivalent bypass reaction, but in the opposite (reductive) direction, where isocitrate 12 is formed from combining glyoxylate 1 and carboxysuccinate 9.  For the glyoxylate cycle to run abiotically in reverse, glyoxylate 1 would have to react with succinate 10 to form 12; such a reaction has been achieved enzymatically where the isocitrate lyase reaction is reversible both in vitro and in vivo 35 . However, the α-CH proton of succinate 10 is not acidic enough to enable the non-enzymatic condensation reaction with 1, whereas the α-CH proton of 9 is (Fig. 3a) 36 . These experimental observations suggest that a reductive glyoxylate cycle (Fig. 5) may be possible, thus avoiding the challenging (and yet to be demonstrated) abiotic reductive carboxylation steps (10 to 12) of the r-TCA cycle 14 . Incidentally, the extant oxidative glyoxylate shunt pathway also avoids the difficult reductive carboxylation steps and has been hypothesized 8 to be an evolutionary precursor of the oxidative TCA cycle; however, it still has to proceed via oxaloacetate 3, which is prebiotically not a viable molecule due to its instability 37,38 . However, the hypothesized reductive glyoxylate pathway in this work avoids 3, and thus, we began to study its feasibility by considering the end products of the cyanide-mediated chemistry, malate 5, succinate 10 and isocitrate 12 (Fig. 1).
Among these compounds, only 5 and 12 can dehydrate to form fumarate 6 and aconitate 13, respectively (Fig. 3a); however these dehydration steps proceed with low yields under forcing conditions 14,39,40 . We found that dehydration of 5 to 6 under a wet-dry cycling protocol, with protic (ammonium) salts capable of donating protons, at 80 °C, was possible (≥70%; Supplementary Table 4 and Supplementary Figs. 52 and 53). Formation of 13 by dehydration of 12 and its lactones (produced via a reaction of 9 and 1; Fig. 3a) required heating with Zn 2+ , forming the cis-and trans-aconitate 13 (14%, ~1:3), but it also produced succinate 10 (25-30%); this result was also observed when starting from authentic isocitrate 12 (Supplementary Table 5 and Supplementary Figs. 54-57). The detection of succinate 10 from the retro-aldol of 12 is important because it is reminiscent of the retro-aldol reaction of isocitrate that returns glyoxylate and succinate in the biological glyoxylate cycle (Fig. 1) without the loss of CO 2 and circumventing α-ketoglutarate 11. Though glyoxylate 1 was formed, it was converted, due to a competing Cannizzaro reaction 23 , to glycolate 32. The counterpart oxalate 31 was not observed in NMR since it precipitated in the presence of Zn 2+ (Supplementary Figs. 59 and 60). The production of glycolate as a stable end product in these reactions may be noteworthy in light of the recent findings 41 Supplementary Fig. 58) and diastereomeric mixtures of isocitrate (~33%). Attempted hydration of 13 under the mild acidic conditions led only to isomerization between cis-and trans-aconitate 39,40 . Screening of various conditions showed that Zn 2+ at pH 5.0 converted 13 to traces of citrate 14 and 55% of isocitrate 12 (Supplementary Table 6 and Supplementary Fig. 61).
We have previously shown that malonate 28 reacts readily with glyoxylate 1 to form 3-carboxymalate 29 (ref. 43 ), which is an isomer of 4b (Fig. 1a) obtained from cyanide reaction of oxaloacetate 3. In light of the observations in Figs. 1a and 3a, we reconsidered whether malate 5 and fumarate 6 can be formed directly from 29, thus bypassing the need for oxaloacetate 3, which has been difficult to form in appreciable amounts since it is unstable 37 in a prebiotic context 38 . Reacting 28 with 1 at room temperature produced a 1:1 mixture of 3-carboxymalate 29 and its dehydrated product, carboxyfumarate 30 (Fig. 3b). Heating this mixture of 29/30 at 50-80 °C (with Zn 2+ ) gave malate (up to 18%), with a competing retro-aldol reaction back to starting materials 28 and 1. Since carboxyfumarate 30 can potentially be reduced by cyanide via the addition-hydrolysis-decarboxylation process (Fig. 1), we added cyanide to the reaction mixture containing 29 and 30 and observed the formation of intermediates, all of which converted via carboxysuccinate 9 predominantly to succinate 10 (40%; Fig. 3b and Supplementary  Figs. 65-68). Encouraged by this observation, we investigated the possibility of a one-pot reaction and mixed glyoxylate 1 and three equivalents malonate 28 at pH 5 for four days at room temperature, and then added NaCN and warmed the reaction to 50 °C and continued the addition of glyoxylate. Monitoring by 1 H NMR showed the appearance of intermediates, finally leading to succinate 10 and isocitrate 12 as major products (Extended Data Fig. 6). Formation of isocitrate in this reaction mixture was corroborated by spiking and electrospray ionization high-resolution mass spectrometry showing the peaks corresponding to 12 and its lactone.
Homocitrate is important given its crucial chelating role in nitrogen-fixing enzymes 46,47 . Tricarballylic acid has been identified as a fermentation product of the reduction of trans-aconitate 48 , and citramalate 19b has been detected as part of the metabolic network in some microorganisms 49 . Malonate and tartrate have been identified in meteorites 50 -suggesting that other compounds formed through this chemistry may also be present in meteorites, which raises the intriguing possibility of these compounds being available as part of a chemical inventory elsewhere in the universe 51,52 . Thus, expanding the reaction palette of α-ketoacids with cyanide to include malonate and malononitrile (as an acetate equivalent, akin to acetyl-coenzyme A) produces a series of molecules (Figs. 1a and 3) that are of interest to the prebiotic 25 and biotic chemical inventory.

Discussion
The tricarboxylic acids-isocitrate 12, aconitate 13 and citrate 14-seem to be natural destinations for the cyanide-or malonate-mediated reaction pathways starting from glyoxylate, with isocitrate 12 predominating. The accumulation of isocitrate 12 (as opposed to citrate 14) could be of consequence since isocitrate can fragment to glyoxylate and succinate, enabling the closing of the cycle of reactions by returning to the starting ketoacid 1 (Fig. 3b). While we did observe the retro-aldol reaction of isocitrate to succinate, the glyoxylate formed was converted to oxalate and glycolate under these conditions due to a competing Cannizzaro reaction 23 . How this can be addressed in a prebiotically robust manner leading to sustainable cycles in the single-pot scenario remains to be seen. One plausible way is to transform the glycolate itself, as has been shown by recent work, where glycolate has been converted to a slew of TCA cycle products 41,42 . Another possibility is for the products in the reductive glyoxylate pathway to be transformed to a relatively stable starting material, which can be used to restart the sequence of reactions.
One way to achieve this transformation is to treat the mixture of products of the reductive glyoxylate pathway with prebiotically available H 2 O 2 (ref. 53 ), which will convert them to malonate as the final product 54  restarted by introduction of fresh glyoxylate to the malonate-rich environment (Fig. 5b). The reactions described here operate over a broad pH range (3)(4)(5)(6)(7)(8)(9) and temperature range (room temperature to 80 °C), with decarboxylative reductions and dehydrations occurring more efficiently in the acidic range, and the aldol reactions occurring in the neutral to basic range. This suggests a potential for selective control of the reaction mixture by simply changing some of these parameters, to allow some reactions to proceed while others do not. Nevertheless, such attempts at closing the cycle are only a proof-of-principle sequence of reactions, which is not the same as the robust self-sustaining r-TCA cycle in biology, affirming the need to search for prebiotic organocatalysts for an abiotic cycle to operate efficiently and continually, if at all. The reductive glyoxylate shunt (pink-coloured pathway) proposed in Fig. 5 is the 'opposite' of the oxidative glyoxylate shunt in extant biology. While the observations of the reversed oxidative TCA cycle 3-5 argue against the oxidative glyoxylate shunt as a precursor to the TCA cycle 8 and have implications for the ancientness of these pathways in ancestral life forms, it would be prudent not to extend the ramifications all the way to prebiotic chemistry 55 . Similarly, the striking resemblance of the chemical structures in Figs. 1 and 3 with those in modern metabolism should not be over-interpreted to make a direct connection, since extrapolating prebiotic chemistry to biological process for the sake of parsimony 56 could be misleading 57 . However, when inspecting the TCA cycle superimposed with the oxidative glyoxylate shunt pathway (black arrows in Fig. 5a), a 'third cycle' , the 'reductive' glyoxylate pathway, becomes apparent (pink-coloured arrows in Fig. 5a). And, remarkably, starting from glyoxylate and malonate, the chemistries mediated by cyanide proceed by the same path, via fumarate and carboxysuccinate 9, culminating in the formation of isocitrate (Fig. 5b), while bypassing many of the central intermediates of the r-TCA cycle (such as oxaloacetate and α-ketoglutarate). This congruence points to an alternate reductive glyoxylate pathway as a primordial possibility, not involving CO 2 explicitly as a reactant (Fig. 5a). A self-sustaining r-TCA cycle involving the formidable reductive carboxylation steps of converting succinate via α-ketoglutarate to isocitrate and the abiotically 'unstable' oxaloacetate may be an evolutionary development that required organocatalysts such as proteins (or perhaps ribozymes 58 ). The advent of proteinaceous catalysts could enable the transformation of the compounds accumulated from the reductive glyoxylate pathway while managing the deleterious effects of transition metals such as iron on the abiotic stability of bioorganic molecules 59 . Lastly, the chemistry outlined here depends on the prebiotic availability of glyoxylate and malonate, for which there is some support 30,[41][42][43]54,60,61 . It is plausible that as the experimental demonstration grows for the central role of glyoxylate and malonate abiotic chemistry, more efforts to put constraints on the prebiotic provenance of glyoxylate and malonate will become important 41,42 .

Online content
Any methods, additional references, Nature Research reporting summaries, source data, extended data, supplementary information, acknowledgements, peer review information; details of author contributions and competing interests; and statements of data and code availability are available at https://doi.org/10.1038/ s41557-021-00878-w. The presence of cyanide at pH 9 first hydrolyzes 7 to the half amide 8, which then undergoes the addition of cyanide followed by hydrolysis and decarboxylation to form succinate 10. Slight shift of succinate peak in NMR spectrum (C, after 11 days) is due to change in pH.