Reduction and Condensation of Aldehydes by the Isolated Cofactor of Nitrogenase.

Isolated nitrogenase cofactors can reduce CO, CN-, and CO2 to short-chain hydrocarbons in reactions driven by a strong reductant. Here, we use activity analyses and isotope labeling experiments to show that formaldehyde and acetaldehydes can be reduced as-is or reductively condensed into alkanes and alkenes by the isolated cofactor of Mo-nitrogenase in the presence of EuII-diethylenetriamine pentaacetate (DTPA). Further, we demonstrate that aldehydes can be condensed with CO by the isolated cofactor under the same reaction conditions, pointing to aldehyde-derived species as possible intermediates of nitrogenase-catalyzed CO reduction. Our deuterium labeling experiments suggest the formation of a cofactor-bound hydroxymethyl intermediate upon activation of the formaldehyde, as well as the release of C2H4 as a product upon β-hydride elimination of an acetaldehyde-derived hydroxyethyl intermediate. These findings establish the reductive condensation of aldehydes as a previously unobserved reactivity of a biogenic catalyst while at the same time shed light on the mechanism of enzymatic CO reduction and C-C bond formation, thereby providing a useful framework for further exploration of the unique reactivity and potential applications of nitrogenase-based reactions.

Introduction

Nitrogenase is known for its key role in the global nitrogen cycle, catalyzing the ambient reduction of atmospheric N2 to bioavailable NH4+.[1−4] Recently, nitrogenase was shown to reduce C1 substrates, such as CO and CN–,[5−7] to short-chain alkanes and alkenes under ambient conditions, thereby gaining recognition as a versatile metalloenzyme capable of both Haber-Bosch[8,9] and Fischer–Tropsch[10,11] like reactions. The “conventional” Mo-nitrogenase utilizes a two-component system for catalysis, with a reductase component (Fe protein) transferring electrons—concomitant with ATP hydrolysis—to the cofactor site (M-cluster) of the catalytic component (MoFe protein) for substrate reduction (Figure S1A). Interestingly, the protein-bound M-cluster only shows a marginal activity of CO reduction that is 700–800 times lower than that of its protein-bound counterpart in the “alternative” V-nitrogenase;[7] upon extraction into an organic solvent, however, the isolated M-cluster is comparable to its V-counterpart in total carbon turnover yields when catalyzing the reduction of CO and other C1 substrates (i.e., CN– and CO2), in the presence of a strong reductant (e.g., EuII-DTPA or SmI2) (Figure S1B).[12,13] The ability of the extracted M-cluster to catalyze the reduction of C1 substrates is important, as it presents a unique opportunity for us to investigate the reactivity of isolated cofactors toward certain oxygenated carbon species, such as aldehydes, alcohols, and acetone, which cannot be directly applied to the protein-bound cofactors because of their destabilizing effects on protein structures. Knowledge in this regard will not only enable a further expansion of the catalytic repertoire of nitrogenase cofactors, but also lead to a better understanding of the mechanism of enzymatic CO reduction and C–C bond formation, as these oxygenated carbon species are potential intermediates of this nitrogenase-catalyzed reaction.[14,15]

Results and Discussion

We started out by testing if oxygenated C1 and C2 species, including aldehydes, alcohols, and acetone, could be converted by the isolated M-cluster into hydrocarbon products. (Note: no unexpected safety hazards were encountered.) Driven by EuII-DTPA (E0′ = −1.14 V at pH 8)[16] in a H2O-based reaction, the isolated M-cluster was capable of converting aldehydes (CH2O, C2H4O), but not alcohols (CH3OH, CH3CH2OH) or acetone ((CH3)2CO), into hydrocarbons (Figure A). GC-MS analysis confirmed that the hydrocarbon products were derived from the aldehyde substrates, showing the expected mass shifts and fragmentation patterns upon substitution of 13C-aldehyde for the corresponding 12C-aldehyde (Figure B,C; also see Figure S5). Notably, formaldehyde (CH2O) gave rise to C1–C4 alkanes (C2H4, C2H6, C3H6, C3H8, C4H8, C4H10), and acetaldehyde (C2H4O) gave rise to C2 and C4 alkanes (C2H4, C2H6, C3H6, C3H8, C4H8, C4H10), suggesting that aldehyde was either reduced as-is or coupled with each other into the respective products (see Figure A). Moreover, compared to CO, both formaldehyde and acetaldehyde were converted by the isolated M-cluster to hydrocarbons at considerably higher yields (see Figure A). Together, these observations firmly established aldehyde as a substrate of the isolated nitrogenase cofactor, which can undergo direct reduction or reductive condensation into hydrocarbon products.

Figure 1

Reduction of oxygenated C1 and C2 substrates by the isolated M-cluster. (A) TONs of reactions catalyzed by the isolated M-cluster, where CO or oxygenated C1 or C2 species was supplied as a substrate. TON, turnover number, calculated based on the number of reduced carbons in hydrocarbon products. (B, C) Product distributions (left) and fragmentation patterns (right) of hydrocarbons generated from the reduction of CH2O (B) and CH3CHO (C). The masses of the base peaks of products are indicated (B, C). Note that the formation of hydrocarbons from CH2O and CH3CHO was completed after ∼1 h (Figure S2); however, all assays were performed overnight to demonstrate that product formation was only detected when aldehydes (and not alcohols or ketones) were supplied as substrates. Control experiments with omissions of the reductant or the M-cluster pointed to the isolated M-cluster as the actual catalyst of the reactions of aldehyde reduction and condensation (Figure S3A,B, 1–3), which was further supported by the inability of Fe/S fragments or degraded M-clusters to enable the same reactivity (Figure S3A,B, 4–6) and a linear correlation between the integrity of the M-cluster and the activity of the M-cluster in product formation over a time period of 2 h (Figure S4).

The newly discovered reactivity of the isolated cofactor toward aldehydes provided a platform for us to use isotope labeling experiments to probe the mechanism of cofactor-based aldehyde reduction. The first question we asked was whether activation of formaldehyde (CH2O) would result in a cluster-bound hydroxymethyl intermediate via several plausible routes.[17−19] Should this be the case, the hydroxymethyl species would need to undergo a series of proton/electron transfer steps, coupled with removal of oxygen as water, to generate CH4 that contains two substrate-derived hydrogens and two solution-derived hydrogens (Figure A). To test our hypothesis, we monitored the amount of H/D labels in product CH4 when CH2O and CD2O were supplied as the respective substrates to D2O- and H2O-based reactions (Figure A, ①, ②). GC-MS analysis revealed formation of CH2D2 as the predominant species upon reduction of CH2O in a D2O-based reaction (Figure B, ①), or upon reduction of CD2O in an H2O-based reaction (Figure B, ②; also see Figure S6A). The appearance of two solution-derived hydrogens in CH4 would be consistent with the appearance of a cluster-bound hydroxymethyl intermediate as an activated C1 species in the cofactor-catalyzed reaction of aldehyde reduction (see Discussion in the Supporting Information for consideration of alternative mechanisms).

Figure 2

Activation of formaldehyde by the isolated M-cluster. (A) Expected identities of products when CH2O is supplied as a substrate to a D2O-based reaction (①) or when CD2O is supplied as a substrate to an H2O-based reaction (②), assuming CH2O is bound to the M-cluster as a hydroxymethyl group (without losing H). M, the isolated M-cluster. (B) GC-MS fragmentation patterns of CH4 formed in a D2O-based reaction, where CH2O was supplied as a substrate (①), or in an H2O-based reaction, where CD2O was supplied as a substrate (②). The masses of the base peaks of products are indicated (B).

Having tackled the question related to the identity of the activated C1 species, we then asked the question of whether the product formed upon C–C bond formation could be released via β-hydride elimination.[20] In this scenario, a partially reduced, cofactor-bound C2 species—such as a cofactor-bound hydroxyethyl group derived from acetaldehyde (C2H4O)—could undergo multiple proton/electron transfer events that are coupled with the removal of its oxygen atom as water, as well as β-hydride elimination that occurs concurrently with the formation of a C=C bond, which eventually leads to the release of C2H4 as a product (Figure A). To test our hypothesis, we monitored the amount of D labels in product C2H4 when D3- or D4-labeled C2H4O (i.e., CD3CHO or CD3CDO) was used as a substrate in an H2O-based buffer system (Figure A, ①, ②). GC-MS analysis demonstrated a predominant formation of C2HD3 (i.e., CD2CDH) when CD3CDO was used as a substrate (Figure B, ①), and the predominant formation of C2H2D2 (i.e., CD2CH2) when CD3CHO was used as a substrate (Figure B, ②; also see Figure S6B). The observation that only one β-hydrogen of C2H4O was replaced in both cases strongly pointed to a mechanism of product release via β-hydride elimination (see Discussion in the Supporting Information for consideration of alternative mechanisms).

Figure 3

Release of C2 product from the isolated M-cluster. (A) Expected identities of products when CD3CDO (①) or CD3CHO (②) is supplied as a substrate to a H2O-based reaction, assuming the CH3CHO-derived intermediate is reduced upon removal of its oxygen atom and undergoes β-hydride elimination concomitant with the formation of a C=C bond, which results in the release of C2H4 as a product. (B) GC-MS fragmentation patterns of C2H4 formed in an H2O-based reaction, where CD3CDO (①) or CD3CHO (②) was supplied as a substrate. The masses of the base peaks of products are indicated (B).

The successful identification of formaldehyde- and acetaldehyde-derived species as catalytically competent intermediates for cofactor-based hydrocarbon formation led to the question of whether these intermediates could be shared by other carbon substrates, such as CO and CN–, in the same type of reactions catalyzed by the cofactor of nitrogenase. To address this question, we examined whether the C1 or C2 aldehyde could be coupled with CO into C2 or C3 products. Excitingly, when CO was supplied in addition to formaldehyde (CH2O), there was an increase in the yield of C2 products (C2H4, C2H6), whereas when CO was supplied in addition to acetaldehyde (C2H4O), C3 products (C3H6, C3H8)—which were absent when CH3CHO was supplied as the sole substrate—were detected in the reaction (Figure A, highlights). Similar effects were observed when CN–, another known C1 substrate of the nitrogenase cofactor, was supplied in addition to CH2O or C2H4O to the reaction (Figure A, highlights). GC-MS analysis further confirmed a cross condensation between aldehyde and the C1 substrate, showing the expected mass shifts and fragmentation patterns of C2 (containing one 13C and one 12C) or C3 (containing two 13C and one 12C) products when 13CH2O or 13C2H4O (the origin of one or two 13C) was supplemented with 12CO or 12CN– (the origin of one 12C) as the substrates of the reaction (Figure B,C; also see Figure S7A,B).

Figure 4

Cross condensation of formaldehyde and acetaldehyde with CO or CN by the isolated M-cluster. (A) TONs of reactions catalyzed by isolated M-clusters, where CH2O or CH3CHO was supplied alone or in combination with CO or CN– as a substrate(s). Products generated upon cross condensation between CH2O (left) or CH3CHO (right) and CO or CN– are highlighted. TON, turnover number, calculated based on the number of reduced carbons in hydrocarbon products. (B, C) GC-MS fragmentation patterns of C2 (B) and C3 (C) hydrocarbons generated from the cross condensation between 13CH2O (B) or 13CH31CHO (C) and 12CO or 12CN–. The masses of the base peaks of products are indicated (B, C). Note that, as expected, C3 products were not detected in the activity assay (A) and GC-MS analysis (C) when CH3CHO was supplied alone as a substrate.

Strikingly, the fragmentation patterns of C2 and C3 products, particularly those of C2H4, C2H6 and C3H6, remained largely unchanged following the expected mass shift upon incorporation of 13C labels (see Figure B,C, green or blue traces vs black traces). This observation implied that these products were predominantly derived from a cross condensation between aldehyde (13CH2O or 13C2H4O) and the C1 substrate (12CO or 12CN–) instead of from a direct condensation of the C1 substrate (12CO or 12CN–) itself, as peaks arising from products with only 12C labels would have become more prominent in the latter case, resulting in alterations of the fragmentation patterns. Moreover, addition of CO or CN– seemed to decrease the formation of C1 and C2 products, respectively, via direct reduction of CH2O and C2H4O (see Figure A), which could be explained by a preferential cross condensation between aldehyde (CH2O, C2H4O) and the C1 substrate (CO, CN–) over a direct reduction of aldehyde itself. The observation of C–C bond formation via an aldehyde-derived moiety and a CO- or CN–-derived moiety was consistent with a recent study that excluded condensation between two neighboring, cofactor-bound CO moieties[21] and pointed to aldehydes as possible intermediates along the reaction pathway of CO reduction.

Interestingly, Fe-bound C1 (hydroxymethyl) and C2 (hydroxyethyl) species have been suggested by density functional theory (DFT) calculations as potential intermediates of CO2 reduction by synthetic [Fe4S4] clusters in a reaction driven by SmI2 (E0′ ≈ – 1.5 V in DMF).[22] It is appealing, therefore, to consider the plausible scenario that the cofactor of nitrogenase—a complex FeS cluster variant—also reduces CO via certain aldehyde-derived intermediates. The significant increase of hydrocarbon yields when aldehydes are supplied as substrates instead of CO seems to support this argument, as aldehydes are further reduced and, therefore, “easier” substrates to enter the CO reduction pathway.

The involvement of β-hydride elimination in the termination step of alkene formation does not come as a complete surprise, not only because of its prevalence in organometallic chemistry,[20] but also because of its implications in nitrogenase-catalyzed reactions. Previously, it was observed that ethene was predominately produced in the enzymatic CO condensation reaction catalyzed by nitrogenase,[5,6] which points to a possible product release mechanism via β-hydride elimination in this case. Recently, it was proposed that metal hydride could play a central role in the reduction of N2 by nitrogenase. In this proposal, the protein-bound M-cluster would first accumulate hydrides as a form of reducing equivalents, which can then be reductively eliminated to generate a “super reduced” state of the metal center for N2 activation.[23,24] While the involvement of a β-hydride elimination step in CO reduction seems to serve a different purpose in terminating the reaction and facilitating product release, it could act in a manner analogous to that proposed for N2 reduction by “shortcutting” the reduction cycle via a “recharge” of the metal center with one hydride for the subsequent reduction events along the CO reduction pathway.

The reductive condensation of aldehydes that directly leads to the formation of hydrocarbons is, to our knowledge, unprecedented in biochemistry. The aldehyde deformylating oxygenase (ADO), a diiron oxygenase that converts aldehyde to hydrocarbon and formate via oxidative cleavage of the aldehydic carbon, is one loosely related biological example along this line of reactivity.[25] In the context of organometallic chemistry, reactions catalyzed by transition metals seem to utilize a mechanism that is distinct from the reaction catalyzed by the M-cluster in that they typically generate alcohols by coupling aldehydes and ketones with olefins or other reactive species.[26] The formation of hydrocarbons in these reactions would require much harsher conditions.

It is important to note that a large variety of homogeneous systems that mimic the Fischer–Tropsch synthesis (FTS) have been documented in the literature.[27] Early development of these systems focused on direct reduction of CO with hydrides, such as the reduction of CpFe(CO)3+ with NaBH4, or conversion of CO to reduced C2 species, by metal hydride complexes.[28,29] In addition, substantial efforts were dedicated to development of metal clusters as FTS models based on the premise that metal clusters, particularly heterobimetallic systems, could exhibit metal surface reactivity patterns in CO activation and C–C bond coupling.[30,31] One relatively well-known example of this genre is the lanthanide-hydride tetramers, which react with CO under ambient conditions to generate ethylene.[32] This is one of the few systems capable of performing the facile conversion of CO to hydrocarbons, encompassing some of the difficult steps, such as C–O bond cleavage and product release. However, upon removal of the oxygen atom, the starting cluster is converted into lanthnide-oxide cubanes, rendering the reaction not catalytic. Another outstanding example that deals with a similar oxygen removal scenario is the terphenyl–diphosphine coordinated molybdenum complex, which is capable of scission of the C–O bond, C–C coupling, and spontaneous dissociation of the resulting C fragment. In this case, Me3SiCl is utilized to efficiently attack and remove oxygen in the presence of K+ to form reactive metal carbene species.[33,34] It is worth noting that, as observed in the transition metal-catalyzed reactions of ketone/aldehyde reduction, our M-cluster-based reaction is efficient in removing oxygen and hydroxyl because of the protic/aqueous reaction system.

Interestingly, it has been demonstrated for the industrial FTS that aldehyde can be incorporated into hydrocarbons when it is supplied along with syngas in the feedstock, which allows for more selective synthesis of desired hydrocarbon products, depending on the aldehyde additive supplied to the reaction.[35] Given the observation of aldehyde condensation with CO, a similar functionality may also be achieved by catalysts based on nitrogenase cofactors. Continued exploration of this reactivity, in combination with further mechanistic investigations into the enzymatic CO reduction and C–C bond formation, could facilitate future developments of nitrogenase-based catalysts for synthesis of valuable chemical commodities.