Skip to main content
NCBI home page
NIHPA Author Manuscripts logoLink to NIHPA Author Manuscripts
. Author manuscript; available in PMC: 2022 Jun 21.
Published in final edited form as: J Am Chem Soc. 2022 Apr 28;144(18):8257–8266. doi: 10.1021/jacs.2c01924

Mechanism of methyldehydrofosmidomycin maturation: Use olefination to enable chain elongation

Xiaojun Li 1, Shan Xue 2, Yisong Guo 2,*, Wei-chen Chang 1,*
PMCID: PMC9210497  NIHMSID: NIHMS1816724  PMID: 35482829

Abstract

Utilization of mononuclear iron- and 2-oxoglutarate-dependent (Fe/2OG) enzymes to enable C-H bond functionalization is a widely used strategy to diversify structural complexity of natural products. Besides those well-studied reactions including hydroxylation, epoxidation and halogenation, in the biosynthetic pathway of dehydrofosmidomycin, an Fe/2OG enzyme is reported to catalyze desaturation, alkyl chain elongation along with demethylation in which trimethyl-2-aminoethylphosphonate is converted to methyldehydrofosmidomycin. How this transformation takes place is largely unknown. Herein we characterized the reactive species, revealed the structure of reaction intermediate, and used mechanistic probes to investigate the reaction pathway and mechanism. These results led to elucidation of a two-step process in which the first reaction employs a long-lived Fe(IV)-oxo species to trigger C=C bond installation. During the second reaction, the olefin installed in situ enables C-C bond formation that is accompanied with a C-N bond cleavage and hydroxylation to furnish the alkyl chain elongation and demethylation. This work expands the reaction repertoire of Fe/2OG enzymes by introducing a new pathway to the known C-C bond formation mechanisms utilized by metalloenzymes.

Graphical Abstract

graphic file with name nihms-1816724-f0001.jpg

Introduction

Alkyl chain elongation is one of the essential processes to build organic molecules. A common chain-elongation strategy utilized in the biosynthesis of fatty acids and polyketides undergoes Claisen-type condensation (Scheme 1a).1 An enolate anion serves as a common intermediate to react with a carboxylic acid derivative in which the C−C bond is installed.1–2 In addition to the aforementioned strategy, metalloenzymes with diverse metal-cofactors, e.g., heme and nonheme iron, are known to catalyze Csp3-Csp3 and Csp3-Csp2 bond formation.3–4 To enable the C−C bond forming reactions, the substrate radical generated via the hydrogen atom abstraction or the subsequent cation is captured by an olefin or an aromatic ring (Scheme 1b). Recently, in the biosynthetic pathway of dehydrofosmidomycin, a mononuclear iron- and 2-oxoglutarate-dependent (Fe/2OG) enzyme, DfmD, is reported to catalyze the conversion of trimethyl-2-aminoethylphosphonate (1) into methyldehydrofosmidomycin (3). Observation of formaldehyde (CH2O) as the co-product using an isotope tracer experiment further establishes the reaction sequence wherein 3 is produced via the hemiaminal group degradation of 2 (Scheme 1c).5 While several members in this enzyme family affect C−C bond formation using the radical or the cation acceptor, i.e., a double bond, appended on the substrate.6–12 the substrate of DfmD does not contain such a group. Furthermore, examples of using an Fe/2OG enzyme to catalyze chain elongation, desaturation and hydroxylation are very rare.

Scheme 1.

Scheme 1.

Open in a new tab

(a) The Claisen-type condensation is a commonly used strategy to enable alkyl chain elongation found in natural product biosynthesis. (b) Metalloenzymes employ an oxidative strategy to affect C−C bond formation. For example, in the KabC and DPS catalyzed reactions,7, 10, 12–13 following hydrogen atom abstraction step triggered by an iron-oxo species, a preinstalled olefin or a benzene ring is used to capture the radical or the subsequent cation to furnish C−C bond formation. The hydrogen atom abstraction site is labeled with an asterisk (*). (c) In the DfmD catalyzed reaction, trimethyl-2-aminoethylphosphonate (1) is converted to 2. Degradation of 2 results in formaldehyde and methyldehydrofosmidomycin (3) formation.

In the vast majority of metalloenzyme-catalyzed C−C bond forming reactions,3–4 the reaction includes cleavage of two C−H bonds with concomitant formation of a C−C bond, thus the overall transformation is a two-electron oxidation of the substrate. However, based on structures of 1 and 2, the DfmD-catalyzed reaction involves a four-electron oxidation process. To unravel the reaction pathway, we captured the reactive species, characterized the reaction intermediate, and used substrate analogs and isotopologues as mechanistic probes to elucidate the plausible mechanism. Taken together, these results provide experimental evidence to support that the DfmD-catalyzed reaction involves a two-step process. As shown in the Scheme 1a, in the first reaction, an Fe(IV)-oxo species triggers hydrogen atom abstraction to furnish olefin intermediate (4). In the second step, the reaction initiates from N-Me hydrogen atom abstraction of 4. The C=C bond installed in situ serves as a radical acceptor to enable chain elongation which is accompanied by hydroxylation on another methyl group. Departure of a formaldehyde from 2 results in 3. Notably, the kinetic and spectroscopic studies, along with in vitro assays revealed that the quaternary ammonium cation of the substrate is critical for the reaction efficiency and selectivity wherein a tertiary amine analog (8) changes the reaction from desaturation into hydroxylation.

Results

In vitro assays establish the DfmD-catalyzed reaction involves a two-step process.

Typical Fe/2OG enzymes utilize a short-lived ferryl species, a.k.a., Fe(IV)-oxo, generated at the expense of 2OG and O2 to trigger hydrogen atom abstraction.14–16 In hydroxylation and halogenation, the substrate radical is captured by Fe(III)-OH or Fe(III)-halide to furnish C−O or C−halide bond and regenerate the Fe(II) species.14, 17–18 Therefore, the overall reaction is a two-electron oxidation of the substrate. Notably, since conversion of 1 to 2 is a four-electron oxidation process, in principle, two equivalents of 2OG and O2, are involved and the reaction should include an intermediate product. We began our study by expressing N-His6-tagged DfmD in E. coli. To help accumulate the intermediate, equal molar quantities of 1 and 2OG were used in the DfmD-catalyzed reaction. The reactions were monitored by liquid chromatography coupled mass spectrometry (LC-MS). As shown in Figure 1, consumption of 1 with concomitant formation of 3 were observed. Importantly, a peak with an m/z value of −2 Da (m/z 168.1 166.1, electrospray ionization, positive mode (ESI+)) relative to the substrate was also detected. This peak may represent the intermediate produced under limiting 2OG condition. Alternatively, it could originate from a shunt reaction. Several Fe/2OG enzymes including AsqJ, KabC, NapI and SyrB2,10, 13, 19–21 have been reported to produce side-products under in vitro conditions.

Figure 1.

Figure 1.

Open in a new tab

(Left) LC-MS analysis of the DfmD-catalyzed reaction. Consumption of 1 (panel c), with concomitant formation of 3 (panel a) and a new peak which has a −2 Da to the substrate (panel b) were detected. Synthetic standard of 3 and 5 are shown in panels a and b (green traces), respectively. (Right) Using 31P-NMR to track the DfmD-catalyzed reactions. Under limiting 2OG condition (1:2OG = 1:1), two product peaks (δ = 1.9 and 5.7 ppm) were produced (trace e). Addition of DfmD and 2OG to the filtrate resulted in complete consumption of both 1 and the peak at 1.9 pm along with formation of 3 (trace d). Traces a-c are standards of 5, 3 and 1. The chemical shifts of all peaks are referenced to phosphate (δ = 0 ppm). Notably, mechanistic probe 5 and the reaction intermediate have different retention time and 31P chemical shift.

To distinguish these mechanistic hypotheses, 31P-NMR was used to track the DfmD-catalyzed reactions. Under limiting 2OG condition, i. e., 1:2OG = 1:1, decrease of 1 (chemical shift (δ) of 11.9 ppm) and formation of two peaks were observed (Figure 1, NMR, trace e). The peak with the chemical shift of 5.7 ppm is in accordance with the product standard (3). Thus, the other peak (δ of 1.9 ppm) is most likely associated with the new peak with an m/z value of 166.1 identified in the LC-MS. To investigate the relevance of this peak to 3 formation, the protein was removed by filtration. DfmD, 2OG and O2 were then introduced to the filtrate. Disappearance of 1 and the peak at 1.9 ppm along with accumulation of 3 were detected (trace d). Complete consumption of the peak at 1.9 ppm at expense of 2OG and O2 clearly establishes its role as an on-pathway intermediate, and suggests that the DfmD reaction involves a two-step process. In addition, observation of 3, but not 2, in the 31P-NMR is consistent with the literature5 and suggests that 2 is unstable and readily decomposes to 3 and formaldehyde.

Using substrate analogs as mechanistic probes to investigate the reaction pathway and to elucidate the structure of the intermediate in situ. As inferred above, the m/z shift of −2 Da to the substrate implies the structure of the intermediate contains one degree of unsaturation. Besides DfmD, γ–butyrobetaine dioxygenase (BBOX) has been reported to effect methyl group insertion.22 The native reaction catalyzed by BBOX is hydroxylation in which butyrobetaine is converted to carnitine. Interestingly, in the presence of the substrate analog, trimethylhydrazine propionate (6), BBOX is able to catalyze C−C bond installation (Scheme 3a).22–23 Following hydrogen atom abstraction, cleavage of the N−N bond and a 1,2-hydrogen migration of the resulting trimethylammonium cation radical affords a carbon radical species. The reaction then follows C−C bond formation, 1,5-hydrogen migration and hydroxylation. Degradation of the hemiaminal moiety results in formaldehyde and 3-amino-4-(methylamino) butanoate (7) formation.22–23 If DfmD deploys a similar strategy, the reaction likely initiates from C1-H activation. A quaternary ammonium cation may induce the cleavage of C1-C2 bond and results in trimethylammonium cation radical formation. Subsequent to 1,2-hydrogen migration, the identical radical species as proposed in the BBOX reaction triggers C−C bond formation. Hydroxylation and dehydration takes place to afford 5 (Scheme 3b). Analog 5 was synthesized and used as a mechanistic probe to test this hypothesis. If 5 serves as an intermediate product, one would expect that it can be converted to 3. Under the current conditions, neither substrate 5 consumption nor product 3 formation could be detected by LC-MS (Figure S1). Additionally, the 31P-NMR chemical shift and the retention time of 5 and the intermediate identified in the enzymatic reaction are different (Figure 1). While the cis-isomer of 5 was not investigated, we anticipate it has a similar 31P-NMR chemical shift as of 5. Taken together, these observations are less consistent with 5 as the intermediate and imply that DfmD employs a different, yet unidentified strategy to affect alkyl chain elongation.

Scheme 3.

Scheme 3.

Open in a new tab

(a) BBOX catalyzes hydroxylation to convert butyrobetaine to carnitine. In the presence of 6, BBOX triggers methyl group insertion to enable 7 formation. The reaction is proposed to undergo C3-H activation, cleavage of N−N bond followed by 1,2-hydrogen atom migration to afford a carbon radical species. Following C−C bond installation, 1,5-hydrogen atom migration and hydroxylation, decomposition of the resulting hemiaminal produces 7 and formaldehyde. (b) Analogous to the BBOX-catalyzed reaction, a similar reaction can be envisioned in the DfmD-catalyzed reaction. Following C1-H activation and 1,2-hydrogen atom migration, a carbon radical species triggers C−C bond formation. The reaction follows hydroxylation and dehydration to afford 5.

To establish the DfmD reaction pathway, it is of critical importance to characterize the chemical structure of this intermediate. Based on the 31P-NMR, it only accumulates to less than 20% under the limiting 2OG condition. Increasing 2OG results in its complete consumption. Thus, isolation of this intermediate is challenging. Indeed, several attempts using large-scale enzymatic reactions failed to accumulate this intermediate. To overcome this obstacle, we decided to elucidate the structure in situ by using the 13C-labeled substrate analogs. Compounds 1,2-13C2-1 and N- 13C-1 were prepared using 13C-enriched dibromoethane and iodomethane according to the procedure described in the supporting information. Using equal molar quantities of 1,2-13C2-1 (δ of 25.3 and 66.0 ppm, Jc-p, c-c = 129 and 35 Hz, Jc-c, c-p = 35 and 3 Hz in 13C-NMR) and 2OG, two sets of new peaks were detected with chemical shifts of 126.5/143.4 and 130.7/137.5 ppm in 13C-NMR (Figure 2). Increasing 2OG resulted in decrease of peaks at 25.3/66.0 and 126.5/143.4 ppm with concomitant increase of peaks at 130.7/137.5 ppm. Therefore, the peaks with chemical shifts of 130.7/137.5 ppm are associated with the final product (3). On the other hand, peaks at 126.5/143.4 ppm represent the intermediate which has the 31P chemical shift of 1.9 ppm and the m/z value of 166.1. The peak at 126.5 ppm appears as a doublet of doublet with coupling constants of 153 and 73 Hz. The larger coupling originates from C1−P coupling while the smaller coupling is associated with C−C nuclei interaction. The peak at 143.4 ppm also appears as a doublet of doublet with a coupling constant of 73 (C−C) and 14 (C2−P) Hz, respectively. While the stereochemistry of the C=C bond, i.e., trans vs. cis, remains to be determined, based on the chemical shift and the coupling pattern, the intermediate (4) is confirmed to have a double bond installed between C1 and C2. Notably, in comparison with 1,2-13C2-1 standard, slight 13C-chemical shift difference is likely associated with the pH change (D2O vs. Tris-buffer pH 7.5.)

Figure 2.

Figure 2.

Open in a new tab

Using NMR and isotope-labeled analogs to characterize the structure of the intermediate in the DfmD-catalyzed reactions in situ. (a) When equal molar quantities of 1,2-13C2-1 and 2OG were used, two sets of peaks with chemical shifts of 126.5/143.4 and 130.7/137.5 were produced (bottom trace). Peaks with chemical shifts of 25.3 and 66.0 ppm are associated with the substrate. Through increasing 2OG, peaks at 25.3/66.0 and 126.5/143.4 decreased with a concomitant increase of peaks at 130.7/137.5 ppm (top trace). (b) Use of N-13C-1 (53.9 ppm) as the substrate resulted in five peaks observed in 13C-NMR (bottom trace). Decrease of peaks at 55.7 ppm and 53.9 ppm, along with increase of peaks at 33.9, 52.6 and 82.2 ppm were detected at a higher molar ratio of 2OG to N-13C-1 (3/1).

Besides, we also carried out the reaction using N-13C-1 where 13C is labeled on one of the three N-methyl groups. Under equal molar quantities of N-13C-1 and 2OG, five peaks were observed in 13C-NMR (Figure 3). Through increasing 2OG, two peaks (δ of 53.9 and 55.7 ppm) were decreased along with accumulation of other three peaks (δ of 33.9, 52.6 and 82.2 ppm). The substrate has the chemical shift of 53.9 ppm. Therefore, the peak at 55.7 ppm originates from the intermediate 4, which is produced under a limiting 2OG condition. Based on the chemical shift and C-H coupling pattern (a quartet under C-H coupling mode, Figure S2), it is evident that three protons are retained on the methyl group. In addition, since only one of the three N-methyl groups is 13C-labeled, 13C thus can be located at N-Me of 3, C3 of 3 or formaldehyde (as the hydrate form) with chemical shifts of 33.9, 52.6 and 82.2 ppm, respectively. The 13C chemical shifts are consistent with the synthetic standard of 3. Due to the long-range C3−P nuclei interaction, the C3 (δ of 52.6 ppm) of 3 appears as a doublet with a coupling constant of 21 Hz.

Figure 3.

Figure 3.

Open in a new tab

Optical absorption spectra of the DfmD-catalyzed reactions monitored by SF-Abs. (Left) Selected absorption spectra of the DfmD•Fe(II)•2OG•1 complex mixed with the oxygenated buffer. The inset shows the kinetic traces at 330 nm of mixing DfmD•Fe(II)•2OG•1 complex (black) or the DfmD•Fe(II)•2OG•8 complex (purple) with saturated O2. (Right) The kinetic traces (330 nm) of mixing the DfmD•Fe(II)•2OG•1 complex (green) or the DfmD•Fe(II)•2OG•1,2-2H4-1 complex (blue) under the limiting O2. The green (for 1) and blue (for 1,2-2H4-1) solid lines are kinetic simulations using the kinetic model described in the SI.

Evidence that an atypical long-lived ferryl (Fe(IV)-oxo) species enables desaturation.

While several Fe/2OG enzyme catalyzed C=C bond formations have been reported in natural product biosynthesis,24–26 introduction of an olefin group adjacent to the phosphonate is rare. To identify the reactive species, transient kinetics studies using stopped-flow absorption spectroscopy (SF-Abs) was employed. The DfmD•Fe(II)•2OG•1 complex exhibits a broad absorption band centered at ~ 510 nm, which represents the typical metal-to-ligand charge transfer band due to the bidentate binding of 2OG to the Fe(II) center (Figure S3 and SI for detailed discussion).27–28 Upon rapid mixing of O2 (in the form of a O2-saturated buffer) with the quaternary DfmD•Fe(II)•2OG•1 complex, the optical absorption across the entire UV and visible range increased and reached to a maximum after ~ 1 s (Figures 4, S4, and S5). This broad feature persisted for more than 100 s, and then slowly decayed. The apparent formation rate of this species is proportional to O2 concentration (~ 0.22 mM and ~ 0.90 mM O2 vs. 0.27 mM Fe(II) loaded DfmD) (Figure S5), thus indicates this species most likely represents an oxygen-derived reactive intermediate, i.e, an iron-oxo species. Next, we carried out a stopped-flow experiment using deuterated substrate analog 10 to establish the role of this long-lived species under limiting O2 condition (~ 0.22 mM O2 vs. 0.27 mM DfmD•Fe(II)•2OG•substratae complex). In comparison with 1, an increased amplitude with concomitant slower decay of this species was observed, which most likely reflects the deuterium kinetic isotope effect (KIE) during the hydrogen atom abstraction step (Figure 3, the right panel).

Figure 4.

Figure 4.

Open in a new tab

LC-MS of the samples quenched at different time points. At the maximum accumulation of the intermediate observed in SF-Abs (1s), neither 4 nor 3 was produced. At 100s, only 4 was observed. At a longer time point (600s), both 4 and 3 were produced. The control experiment (top trace) was carried out under limiting 2OG condition.

To demonstrate the role of the long-lived species observed in the SF-Abs, we carried out a chemical quench experiment. A solution generated by mixing DfmD•Fe(II)•2OG•1 complex with O2-saturated buffer was quenched at designated time points (see the SI for detailed procedure). At the time point of the maximal accumulation of this species (~ 1 s), no product was detected by LC-MS. At 100 s, only 4 was observed. At a longer time point (600 s) where this species was fully decayed, both 4 and 3 were detected (Figure 4). Taken together, the kinetic analysis and the chemical quench experiment reveal that this long-lived species accumulates prior to the intermediate (4) production and its decay is associated with formation of 4.

Furthermore, to verify the observed difference in SF-Abs experiments with 1 and 1,2-2H4-1 reflects the intrinsic H/D KIE, chemical quench experiments were carried out under a limiting-O2 condition. As shown in Figure 3 (the right panel), the maximal accumulation of the intermediate using 1 was achieved at ~ 10 s. However, at 10 s no product was detected in the chemical quench experiments. At 50 s and 250 s, in the decay phase of this long-lived species, a similar level of 4 was observed (Figure S11). Due to the limiting O2 condition used, the reaction was halted at the first step wherein 3 was not produced. In contrast, while an obvious formation and decay of the long-lived species was observed in by SF-Abs, 1,2-2H4-1 did not result in any product formation. These observations suggest that two pathways are involved in the decay of the long-lived species: one pathway leads to product formation (kH), the other one is an unproductive pathway (kself-decay). When native substrate 1 is used, product formation pathway dominates (kH >> kself-decay). On the contrary, the productive pathway is hampered in the presence of the deuterated substrate (1,2-2H4-1) due to the H/D isotope effect, while the unproductive pathway remains unaffected (kD << kself-decay). A kinetic simulation reveals that the decay rate constant for productive pathway in 1 (kH) is 0.025 s−1, and that for unproductive pathway (kself-decay) is 0.003 s-1. While the decay rate constant for productive pathway using 1,2-2H4-1 (kD) cannot be accurately determined (see the SI for detailed discussion), the kD should be significantly smaller than kself-decay to reflect the chemical quench experiments. Taken together, these results suggest that a hydrogen atom abstraction step is the rate-limiting step and the broad absorption feature is associated with the reactive species responsible for carrying out such a chemical step.

In the vast majority of Fe/2OG enzymes, mixing the enzyme•Fe(II)•2OG•subtrate quaternary complex with O2 typically generates a short-lived ferryl intermediate. It is very rare that a ferryl species can be accumulated in the seconds to minutes range using protiated substrate. To characterize the long-lived species observed in SF-Abs, we carried out freeze quench coupled Mössbauer measurements by rapid mixing the DmfD•Fe(II)•2OG•1 complex with an oxygenated buffer. The anaerobic DmfD•Fe(II)•2OG•1 complex exhibited a broad quadrupole doublet, which can be simulated by two major species with parameters (δ = 1.27 mm/s, |ΔEQ| = 2.68 mm/s for Species I and δ = 1.28 mm/s, |ΔEQ| = 2.09 mm/s for Species II) typical of high-spin ferrous species in a 6-coordinate ligand environment found in many Fe/2OG enzymes (Figures 5 and S6, see Tables S1-S2 for the simulation parameters). An O2 insensitive ferrous species, representing < 10% of the iron in the sample, was also observed. This species likely corresponds to the inactive enzyme. After mixing with O2, both ferrous species were consumed and a new quadruple doublet with δ = 0.24 mm/s and |ΔEQ|= 0.65 mm/s was accumulated to ~ 70% relative to the total iron in the sample (Figure 5, t = 5 s). Subsequently, a slow decay concomitant with the formation of the Fe(II) species exhibiting the identical spectroscopic features as of the initial enzyme quaternary complex were observed (Figure S6). While the Mössbauer parameters of the intermediate species are distinct from several reported ferryl intermediates (δ ~ 0.28 – 0.32 mm/s),13, 16, 20, 29–34 the parameters are almost identical to one of the two ferryl intermediates observed in Fe/2OG halogenase SyrB2 and CytC3 (δ = 0.23 mm/s and |ΔEQ| = 0.7 – 0.8 mm/s).35–37 To reveal the electronic nature of this intermediate species, we carried out variable field Mössbauer analysis (Figure S7). The spectral analysis confirms that this species is a ferryl intermediate with an S = 2 ground spin state which is characterized by a large and positive zero-field splitting parameter (D ~ 10 cm−1) and an axial 57Fe nuclear hyperfine tensor (Ax/gnβn = Ay/gnβn = −17.4 T, assuming E/D = 0). Compared to other Fe/2OG enzymes, accumulation of a long-lived ferryl species due to an extremely slow decay kinetics is very uncommon. For example, the decay rate constant of the ferryl intermediate in taurine-2OG dioxygenase (TauD) was ~ 13 s−1,15–16 while a similar ferryl intermediate in DfmD has the decay rate of ~ 0.025 s−1 in the presence of the native substrate. At 250 s, recovery of the Fe(II) state without producing significant amount of the Fe(III) species (less than 10%) is confirmed by a high-field Mössbauer measurement (Figure S8).

Figure 5.

Figure 5.

Open in a new tab

Zero field Mössbauer spectra monitoring the accumulation of the ferryl intermediate. (Left) The Mössbauer spectra (black vertical bars) and the overall spectral simulations (the solid lines) of the reaction of DfmD•Fe(II)•2OG•1 with O2 and quenched at the selected time points. The red solid lines indicate the spectral simulation of the ferryl intermediate. (Right) The Mössbauer spectra (black vertical bars) and the overall spectral simulations (the solid lines) of the reaction of DfmD•Fe(II)•2OG•8 with O2 observed at the selected time points. The blue solid lines indicate the spectral simulation of the ferryl intermediate. The simulation parameters are listed in Tables S1 – S4.

A quaternary ammonium cation is critical to affect the reaction efficiency and direction.

In addition to DfmD, two other Fe/2OG enzymes, PhnY and TmpA, have been reported to use 1,2-amino phosphonates as the substrates and catalyze hydroxylation reactions.38–39 As suggested by the X-ray structure of TmpA, an aromatic cage is recognized to interact with the quaternary ammonium group of 1 through cation-π interactions.39 Notably, replacing a methyl group with a proton, i.e. 1 8, reduces the efficiency of TmpA while maintaining the reaction outcome. To investigate the influence of methyl groups of the substrate and to reveal possible reaction mechanism, analog 8 was synthesized. As revealed in SF-Abs and FQ-Mössbauer experiments (Figures 3 and 5), 8 triggers ferryl species formation. Rapid mixing of the DfmD•Fe(II)•2OG•8 complex with O2 led to the appearance of a highly similar optical feature in SF-Abs (Figure S9), thus indicating formation of a ferryl species. While the apparent formation rate constant is much smaller (4.0 s−1 in 1 vs. 0.25 s−1 in 8, under the saturating O2 condition, Figure S5), the decay kinetics is largely unaffected (see Figure 4). Additionally, the DmfD•Fe(II)•2OG•8 complex exhibits a broad quadrupole doublet that is attributed to two major doublet with parameters that are typical of high-spin ferrous species in the Mössbauer measurements (Tables S3 and S4). Notably, the parameters of the observed Fe(II) species are different from those of DmfD•Fe(II)•2OG•1 complex. At 20 s, a ferryl intermediate with δ = 0.25 mm/s and |ΔEQ| = 0.75 mm/s was observed, which represents 45% of the total iron in the sample. Compared to the ferryl intermediate observed using 1, the magnitude of the quadrupole splitting changed from 0.65 mm/s to 0.75 mm/s. At 75 s, this ferryl intermediate decays to < 10% (Figure 5). Additionally, 8 changed the DfmD reaction from desaturation to hydroxylation where a peak with +16 Da (154.1 → 170.1, ESI+) was observed by LC-MS (Figure S10). The subtle difference of the ferryl species revealed by the Mössbauer spectroscopy implies that replacing a methyl group with a proton likely affect the substrate binding, thus influences the electronic structure of the ferryl species and the reaction outcome.

Evidence that the chain elongation initiates from N-Me hydrogen abstraction and involves another step sensitive to C-H bond cleavage.

To assess the reaction pathway leading to chain elongation, we used an internal isotope competition experiment to reveal steps sensitive to C-H bond cleavage,40–42 in which the substrate analog (N-Me-2H3-1) with one of the three N-CH3 groups replaced with the N-CD3 was used. This analog was synthesized using deuterated methyl iodide (CD3-I). As revealed by the in situ 13C-NMR experiment, the 13C labeled on the N-Me of N- 13C-1 is incorporated onto C3, N-Me of 3 and formaldehyde. Herein, we followed the number of deuterons retaining on 3 to quantify the products generated via different pathways. Depending on the site of hydrogen atom abstraction, N-Me-2H3-1 yields three products with three-, two- or non-deuteron incorporation with m/z values of 155.1, 154.1 and 152.1 (Figure 6), respectively. Specifically, the pathway involves deuterium atom abstraction at the N-methyl position results in 3,3-2[H]2-3. In contrast, 3 and N-Me-2[H]3-3 are produced when hydrogen atom abstraction takes place. If hydrogen atom abstraction at the N-Me position has a H/D isotope effect, an influence on the product distribution is anticipated. Integration of each product peak in the LC-MS chromatograms established that 3, 3,3-2[H]2-3 and N-Me-2[H]3-3 are produced in a ratio of ~ 0.13:0.02:0.85 (Figure 6). Product 3,3-2[H]2-3 accounts for ~ 2 % of the total products, which strongly reflects on the H/D KIE on the N-Me hydrogen atom abstraction step. Furthermore, the product ratio of 0.13/0.85 for 3 and N-Me-2[H]3-3 implies that another chemical step sensitive to C-H bond cleavage is likely involved in the conversion of 4 into 3.

Figure 6.

Figure 6.

Open in a new tab

DfmD-catalyzed reaction using N-Me-2H3-1 as the substrate. (Left) Three products with distinctive isotope labeling patterns are produced. (Right) LC-MS chromatograms of 3, 3,3-2[H]2-3 and N-Me-2[H]3-3 (3, 3,3-2[H]2-3 and N-Me-2[H]3-3 are produced in a ratio of ~ 0.13:0.02:0.85).

Discussion

Mononuclear iron- and 2OG-dependent enzymes catalyze, in the most cases, a two-electron oxidation of the primary substrate. In the canonical hydroxylation, following hydrogen atom abstraction, the substrate radical reacts with Fe(III)−OH to install the C−O bond.14, 24 Some Fe/2OG enzymes can catalyze C−C bond forming reactions in which the pre-installed olefin moiety or an aromatic ring are used to capture the resulting radical or cation.7, 10, 13, 24, 26 Apart from those well-studied reactions, DfmD employs a new approach to enable C−C bond formation, desaturation and demethylation.

A structural comparison of 1 and 2 reveals that the overall transformation is a four-electron oxidation, therefore, DfmD likely catalyzes two consecutive reactions. We first use LC-MS and 31P-NMR to observe such an intermediate product, 4 (Figure 1). These observations establishes a two-step reaction sequence. To overcome the low accumulation level of 4 in the current enzyme assay conditions, we use 13C-labeled analogs (1,2-13C2-1 and N- 13C-1 ) to elucidate the chemical structure of this intermediate by 13C-NMR (Figure 2). These results unambiguously establish the structure of 4, confirming that the first step of the DfmD catalysis is to install the C=C bond at C1 and C2. To trap and to analyze plausible reactive species involved in the formation of 4, we apply stopped-flow absorption spectroscopy along with freeze-quench coupled Mössbauer spectroscopy (Figures 3 and 5). A long-lived ferryl species is identified as the key species to trigger hydrogen atom abstraction. The vast majority of Fe/2OG enzymes deploy a short-lived ferryl species with a rapid formation (up to 1–2 × 105 M−1s−1) and a fast decay rate constants (in the order of seconds).14–16, 29 The ferryl species generally features a large H/D KIE (up to > 100 demonstrated in IsnB from Photorhabdus luminescens that catalyzes vinyl isonitrile formation in rhabduscin biosynthesis).43 The large H/D KIE value suggests hydrogen nuclear quantum tunneling effect contributes to the hydrogen atom abstraction step. Notably, the ferryl species deployed by DfmD exhibits distinctive kinetic behaviors, wherein the formation rate constant is only ~ 6 × 103 M−1s−1 (even under the saturating O2 condition), suggesting a sluggish reactivity towards O2. Additionally, the decay rate is also very slow with a rate constant of ~ 0.03 s−1, indicating a substantial activation barrier at the hydrogen atom abstraction step. Consequently, the ferryl intermediate can be accumulated to a nearly stoichiometric amount (Figure 3). While the observed H/D KIE ~ 3 (1 vs. 1,2-2H4-1 ) on the ferryl decay was deduced from the SF-Abs data, the corresponding product analysis using deuterated analog 1,2-2H4-1 in the chemical quench experiments did not observe any product formation under limiting-O2 condition (Figure S11), suggesting two reaction channels (productive vs. unproductive) are employed in the ferryl intermediate decay. In the presence of protium substrate, productive pathways dominates (>90%), which results in 4 formation and Fe(II) regeneration as revealed by LC-MS analysis and by Mössbauer analysis. On the other hand, due the H/D kinetic isotope effect, the use of deuterated substrate 1,2-2H4-1 significantly slows down the productive pathway, thus redirecting the reaction flux to the unproductive pathway without 4 formation.

To reveal a possible pathway deployed for the alkyl chain elongation, an internal isotope competition strategy is applied to hint at the steps that are sensitive to H/D KIE. Incubating N-Me-2H3-1 with DfmD resulted in 3, 3,3-2[H]2-3 and N-Me-2[H]3-3 formation in a ratio of ~ 0.13:0.02:0.85 (Figure 6). The product ratio clearly reflects the intrinsic H/D isotope effect, wherein ~ 2% of 3,3-2[H]2-3 suggests that the chain elongation most likely initiates via hydrogen atom abstraction at the N-Me position. Additionally, a substantial difference between 3 and N-Me-2[H]3-3 production further implies the involvement of another step sensitive to C-H bond cleavage in the chain elongation step.

Another Fe/2OG enzyme, TmpA, has been reported to catalyze hydroxylation using 1 as the native substrate. Replacement of a methyl group with a proton (1 8) resulted in the reduced catalytic activity of TmpA, while maintaining the reaction selectivity. Based on the published substrate bound TmpA structure,39 the reduced reactivity is most likely associated with the perturbation of substrate binding. A similar effect is observed by Mössbauer analysis. In the presence of 8, a ferryl intermediate was formed at a much slower formation rate (~ 6000 M−1s−1 vs. ~ 200 M−1s−1 1 vs. 8). Different from TmpA, 8 redirects the DfmD reaction outcome, wherein 8 results in hydroxylation. While the detailed mechanism with regard to the reactivity switching remains to be carefully elucidated, these results point out the importance of methyl groups (N(Me)3) to maintain the reaction efficiency and selectivity

Conclusion

Taken together, the results included in this study provide the experimental support that leads to the elucidation of reaction pathway for the DfmD-catalyzed alkyl chain elongation as illustrated in Scheme 2. In the first reaction, DfmD employs a long-lived ferryl species to enable desaturation. In the second reaction, following the N-Me hydrogen atom abstraction, capture of the presumptive radical using the olefin installed in situ enables formation of an aziridinium intermediate. Following the cleavage of C3-N bond, the reaction undergoes a 1,2-hydrogen transfer, hydroxylation to afford 2. Alternatively, a pathway includes electron transfer coupled proton removal or hydrogen atom abstraction can be envisioned. In this instance, the the resulting iminium species is likely quenched by water. Decomposition of the hemiaminal moiety of 2 takes place to generate formaldehyde and methyldehydrofosmidomycin. Notably, replacing a methyl group with a proton alters the Mössbauer parameters of the ferryl species and directs the DfmD reaction from desaturation to hydroxylation, thus implying that a potential cation-π interaction of the substrate with the surrounding protein environment might play a role in the DfmD reaction. While identification of hydrogen atom abstraction site using an internal isotope competition experiment provides important insight to elucidate plausible pathways for the chain elongation, detailed mechanisms about how DfmD can catalyze chemically divergent reactions await for further investigations. Nevertheless, the current study provides new mechanistic insights into a novel C-C bond forming reaction catalyzed by Fe/2OG enzymes.

Scheme 2.

Scheme 2.

Open in a new tab

(a) Overall reaction catalyzed by DfmD involves a two-step process. In the first reaction (1 → 4), an Fe(IV)-oxo species triggers hydrogen atom abstraction to enable 4 formation. In the second reaction (4 → 2 → 3, a methyl group insertion is accompanied by a remote hydroxylation to furnish 2. It is followed by formaldehyde departure to produce 3. (b) Possible pathways for chain elongation. Following hydrogen atom abstraction, the olefin moiety installed in situ enables an aziridinium ring formation. Cleavage of the C2−N bond followed by 1,2-hydrogen migration and hydroxylation affords 2. Alternatively, a pathway involving addition of a water to the iminium species e also envisioned to effect 2 formation.

Supplementary Material

SI

ACKNOWLEDGMENT

This work was supported by the National Institutes of Health (GM127588 to W.-c.C. and GM125924 to Y.G.)

Footnotes

ASSOCIATED CONTENT

Supporting Information. NMR spectra of all compounds, experimental details for enzymatic assays, LC-MS methods and chromatograms, stopped-flow absorption and Mössbauer spectra and parameters. This material is available free of charge via the Internet at http://pubs.acs.org.

REFERENCES

  • 1.Heath RJ; Rock CO, The Claisen condensation in biology. Nat. Prod. Rep 2002, 19, 581–596. [DOI] [PubMed] [Google Scholar]
  • 2.Lin CI; McCarty RM; Liu HW, The Enzymology of Organic Transformations: A Survey of Name Reactions in Biological Systems. Angew. Chem. Int. Ed 2017, 56, 3446–3489. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 3.Guengerich FP; Yoshimoto FK, Formation and Cleavage of C-C Bonds by Enzymatic Oxidation-Reduction Reactions. Chem. Rev 2018, 118, 6573–6655. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 4.Tang MC; Zou Y; Watanabe K; Walsh CT; Tang Y, Oxidative Cyclization in Natural Product Biosynthesis. Chem. Rev 2017, 117, 5226–5333. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 5.Parkinson EI; Erb A; Eliot AC; Ju KS; Metcalf WW, Fosmidomycin biosynthesis diverges from related phosphonate natural products. Nat. Chem. Biol 2019, 15, 1049–1056. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 6.Lloyd MD; Lipscomb SJ; Hewitson KS; Hensgens CM; Baldwin JE; Schofield CJ, Controlling the substrate selectivity of deacetoxycephalosporin/ deacetylcephalosporin C synthase. J. Biol. Chem 2004, 279, 15420–15426. [DOI] [PubMed] [Google Scholar]
  • 7.Lau W; Sattely ES, Six enzymes from mayapple that complete the biosynthetic pathway to the etoposide aglycone. Science 2015, 349, 1224–1228. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 8.Nakamura H; Matsuda Y; Abe I, Unique chemistry of non-heme iron enzymes in fungal biosynthetic pathways. Nat. Prod. Rep 2018, 35, 633–645. [DOI] [PubMed] [Google Scholar]
  • 9.Brunson JK; McKinnie SMK; Chekan JR; McCrow JP; Miles ZD; Bertrand EM; Bielinski VA; Luhavaya H; Obornik M; Smith GJ; Hutchins DA; Allen AE; Moore BS, Biosynthesis of the neurotoxin domoic acid in a bloom-forming diatom. Science 2018, 361, 1356–1358. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 10.Chekan JR; McKinnie SMK; Moore ML; Poplawski SG; Michael TP; Moore BS, Scalable Biosynthesis of the Seaweed Neurochemical, Kainic Acid. Angew. Chem. Int. Ed 2019, 58, 8454–8457. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 11.Abe I, Nonheme Iron- and 2-Oxoglutarate-Dependent Dioxygenases in Fungal Meroterpenoid Biosynthesis. Chem. Pharm. Bull 2020, 68, 823–831. [DOI] [PubMed] [Google Scholar]
  • 12.Tang H; Wu MH; Lin HY; Han MR; Tu YH; Yang ZJ; Chien TC; Chan NL; Chang W.-c., Mechanistic analysis of carbon-carbon bond formation by deoxypodophyllotoxin synthase. Proc. Natl. Acad. Sci. U.S.A 2022, 119, e2113770119. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 13.Chen TY; Xue S; Tsai WC; Chien TC; Guo YS; Chang W.-c., Deciphering Pyrrolidine and Olefin Formation Mechanism in Kainic Acid Biosynthesis. ACS Catal 2021, 11, 278–282. [Google Scholar]
  • 14.Martinez S; Hausinger RP, Catalytic Mechanisms of Fe(II)- and 2-Oxoglutarate-dependent Oxygenases. J. Biol. Chem 2015, 290, 20702–20711. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 15.Price JC; Barr EW; Hoffart LM; Krebs C; Bollinger JM Jr., Kinetic dissection of the catalytic mechanism of taurine:alpha-ketoglutarate dioxygenase (TauD) from Escherichia coli. Biochemistry 2005, 44, 8138–8147. [DOI] [PubMed] [Google Scholar]
  • 16.Price JC; Barr EW; Tirupati B; Bollinger JM Jr.; Krebs C, The First Direct Characterization of a High-Valent Iron Intermediate in the Reaction of an Alpha-Ketoglutarate-Dependent Dioxygenase: A High-Spin FeIV Complex in Taurine/Alpha-Ketoglutarate Dioxygenase (TauD) from Escherichia coli. Biochemistry 2003, 42, 7497–7508. [DOI] [PubMed] [Google Scholar]
  • 17.Bollinger JM Jr; Chang W.-c.; Matthews ML; Martinie RJ; Boal AK; Krebs C, CHAPTER 3. Mechanisms of 2-Oxoglutarate-Dependent Oxygenases: The Hydroxylation Paradigm and Beyond. In 2-Oxoglutarate-Dependent Oxygenases, 2015; pp 95–122.
  • 18.Krebs C; Galonic Fujimori D; Walsh CT; Bollinger JM Jr., Non-heme Fe(IV)-oxo intermediates. Acc. Chem. Res 2007, 40, 484–492. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 19.Einsiedler M; Jamieson CS; Maskeri MA; Houk KN; Gulder TAM, Fungal Dioxygenase AsqJ Is Promiscuous and Bimodal: Substrate-Directed Formation of Quinolones versus Quinazolinones. Angew. Chem. Int. Ed 2021, 60, 8297–8302. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 20.Dunham NP; Chang W.-c.; Mitchell AJ; Martinie RJ; Zhang B; Bergman JA; Rajakovich LJ; Wang B; Silakov A; Krebs C; Boal AK; Bollinger JM Jr., Two Distinct Mechanisms for C-C Desaturation by Iron(II)- and 2-(Oxo)glutarate-Dependent Oxygenases: Importance of alpha-Heteroatom Assistance. J. Am. Chem. Soc 2018, 140, 7116–7126. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 21.Matthews ML; Neumann CS; Miles LA; Grove TL; Booker SJ; Krebs C; Walsh CT; Bollinger JM Jr., Substrate positioning controls the partition between halogenation and hydroxylation in the aliphatic halogenase, SyrB2. Proc. Natl. Acad. Sci. U.S.A 2009, 106, 17723–17728. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 22.Leung IK; Krojer TJ; Kochan GT; Henry L; von Delft F; Claridge TD; Oppermann U; McDonough MA; Schofield CJ, Structural and mechanistic studies on gamma-butyrobetaine hydroxylase. Chem. Biol 2010, 17, 1316–1324. [DOI] [PubMed] [Google Scholar]
  • 23.Henry L; Leung IK; Claridge TD; Schofield CJ, gamma-Butyrobetaine hydroxylase catalyses a Stevens type rearrangement. Bioorg. Med. Chem. Lett 2012, 22, 4975–4978. [DOI] [PubMed] [Google Scholar]
  • 24.Islam MS; Leissing TM; Chowdhury R; Hopkinson RJ; Schofield CJ, 2-Oxoglutarate-Dependent Oxygenases. Annu. Rev. Biochem 2018, 87, 585–620. [DOI] [PubMed] [Google Scholar]
  • 25.Herr CQ; Hausinger RP, Amazing Diversity in Biochemical Roles of Fe(II)/2-Oxoglutarate Oxygenases. Trends Biochem. Sci 2018, 43, 517–532. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 26.Gao SS; Naowarojna N; Cheng R; Liu X; Liu P, Recent examples of alpha-ketoglutarate-dependent mononuclear non-haem iron enzymes in natural product biosyntheses. Nat. Prod. Rep 2018, 35, 792–837. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 27.Pavel EG; Zhou J; Busby RW; Gunsior M; Townsend CA; Solomon EI, Circular dichroism and magnetic circular dichroism spectroscopic studies of the non-heme ferrous active site in clavaminate synthase and its interaction with alpha-ketoglutarate cosubstrate. J. Am. Chem. Soc 1998, 120, 743–753. [Google Scholar]
  • 28.Ryle MJ; Padmakumar R; Hausinger RP, Stopped-Flow Kinetic analysis of Escherichia coli Taurine/Alpha-Ketoglutarate Dioxygenase: Interactions with Alpha-Ketoglutarate, Taurine, and Oxygen. Biochemistry 1999, 38, 15278–15286. [DOI] [PubMed] [Google Scholar]
  • 29.Hoffart LM; Barr EW; Guyer RB; Bollinger JM Jr.; Krebs C, Direct spectroscopic detection of a C-H-cleaving high-spin Fe(IV) complex in a prolyl-4-hydroxylase. Proc. Natl. Acad. Sci. U.S.A 2006, 103, 14738–14743. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 30.Sinnecker S; Svensen N; Barr EW; Ye S; Bollinger JM Jr.; Neese F; Krebs C, Spectroscopic and computational evaluation of the structure of the high-spin Fe(IV)-oxo intermediates in taurine: alpha-ketoglutarate dioxygenase from Escherichia coli and its His99Ala ligand variant. J. Am. Chem. Soc 2007, 129, 6168–6179. [DOI] [PubMed] [Google Scholar]
  • 31.Panay AJ; Lee M; Krebs C; Bollinger JM; Fitzpatrick PF, Evidence for a high-spin Fe(IV) species in the catalytic cycle of a bacterial phenylalanine hydroxylase. Biochemistry 2011, 50, 1928–1933. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 32.Chang W.-c.; Guo Y; Wang C; Butch SE; Rosenzweig AC; Boal AK; Krebs C; Bollinger JM Jr., Mechanism of the C5 Stereoinversion Reaction in the Biosynthesis of Carbapenem Antibiotics. Science 2014, 343, 1140–1144. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 33.Chang W.-c.; Li J; Lee JL; Cronican AA; Guo Y, Mechanistic Investigation of a Non-Heme Iron Enzyme Catalyzed Epoxidation in (−)-4’-Methoxycyclopenin Biosynthesis. J. Am. Chem. Soc 2016, 138, 10390–10393. [DOI] [PubMed] [Google Scholar]
  • 34.Chen TY; Chen J; Tang Y; Zhou J; Guo Y; Chang W.-c., Pathway from N-Alkylglycine to Alkylisonitrile Catalyzed by Iron(II) and 2-Oxoglutarate-Dependent Oxygenases. Angew. Chem. Int. Ed 2020, 59, 7367–7371. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 35.Galonic DP; Barr EW; Walsh CT; Bollinger JM Jr.; Krebs C, Two interconverting Fe(IV) intermediates in aliphatic chlorination by the halogenase CytC3. Nat. Chem. Biol 2007, 3, 113–116. [DOI] [PubMed] [Google Scholar]
  • 36.Fujimori DG; Barr EW; Matthews ML; Koch GM; Yonce JR; Walsh CT; Bollinger JM Jr.; Krebs C; Riggs-Gelasco PJ, Spectroscopic evidence for a high-spin Br-Fe(IV)-oxo intermediate in the alpha-ketoglutarate-dependent halogenase CytC3 from Streptomyces. J. Am. Chem. Soc 2007, 129, 13408–13409. [DOI] [PubMed] [Google Scholar]
  • 37.Matthews ML; Krest CM; Barr EW; Vaillancourt FH; Walsh CT; Green MT; Krebs C; Bollinger JM, Substrate-triggered formation and remarkable stability of the C-H bond-cleaving chloroferryl intermediate in the aliphatic halogenase, SyrB2. Biochemistry 2009, 48, 4331–4343. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 38.McSorley FR; Wyatt PB; Martinez A; DeLong EF; Hove-Jensen B; Zechel DL, PhnY and PhnZ comprise a new oxidative pathway for enzymatic cleavage of a carbon-phosphorus bond. J. Am. Chem. Soc 2012, 134, 8364–8367. [DOI] [PubMed] [Google Scholar]
  • 39.Rajakovich LJ; Pandelia ME; Mitchell AJ; Chang W.-c.; Zhang B; Boal AK; Krebs C; Bollinger JM Jr., A New Microbial Pathway for Organophosphonate Degradation Catalyzed by Two Previously Misannotated Non-Heme-Iron Oxygenases. Biochemistry 2019, 58, 1627–1647. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 40.Northrop DB, The expression of isotope effects on enzyme-catalyzed reactions. Annu. Rev. Biochem 1981, 50, 103–131. [DOI] [PubMed] [Google Scholar]
  • 41.Cleland WW, The use of isotope effects to determine enzyme mechanisms. J. Biol. Chem 2003, 278, 51975–51984. [DOI] [PubMed] [Google Scholar]
  • 42.Cleland WW, The use of isotope effects to determine enzyme mechanisms. Arch. Biochem. Biophys 2005, 433, 2–12. [DOI] [PubMed] [Google Scholar]
  • 43.Yu CP; Tang Y; Cha L; Milikisiyants S; Smirnova TI; Smirnov AI; Guo Y; Chang W.-c., Elucidating the Reaction Pathway of Decarboxylation-Assisted Olefination Catalyzed by a Mononuclear Non-Heme Iron Enzyme. J. Am. Chem. Soc 2018, 140, 15190–15193. [DOI] [PMC free article] [PubMed] [Google Scholar]

Associated Data

This section collects any data citations, data availability statements, or supplementary materials included in this article.

Supplementary Materials

SI
NIHMS1816724-supplement-SI.pdf (4.8MB, pdf)

RESOURCES