Is cyclopropyl methyl carbanion also exceptionally stable

Characterization of a two-component cyclopropanase system

We have previously showed that the radical SAM enzyme C10P is essential for the formation of the cyclopropane moiety in 1 (Fig. 2d), and inactivation of c10P produced a compound 6 (Fig. 2i)29. Subsequently, gene deletion mutations led us to the identification of a SAM-dependent methyltransferase gene c10Q that is also required for the production of 1 (Fig. 2e and Supplementary Fig. 2). Intriguingly, this ∆c10Q mutant bears the same metabolite profile as that of the ∆c10P mutant, that is, mutation of c10Q also led to accumulation of the compound 6 in the fermentation broth (Fig. 2e). Moreover, introduction of the c10Q gene in trans into the ∆c10Q mutant restored the biosynthesis of 1 (Fig. 2f). These genetic experiment results inspired us to surmise that C10P and C10Q may work together to convert the compound 6 to 1.

To test this hypothesis, we first expressed c10P and c10Q in E. coli, and purified the recombinant enzymes as His6-tagged fusion proteins, respectively (Supplementary Fig. 3). Despite many efforts we could not obtain the C10P proteins with high purity. Fortunately, using the C10P sequence as a query we retrieved a cryptic BGC from Shewanella woodyi ATCC 51908 (GenBank accession number NC_010506) that shows high homology and synergy with that of 1 (Supplementary Fig. 4). We then cloned and integrated the Swoo_2002 gene (the c10P counterpart) into the chromosome of the ∆c10P mutant, and found that Swoo_2002 could function as a C10P surrogate (Fig. 2g, h). Subsequent heterologous overexpression of Swoo_2002 in E. coli led to obtaining relatively pure recombinant protein (Supplementary Fig. 3). After anaerobic reconstitution, Swoo_2002 contained 4.29 ± 0.12 of iron and 4.70 ± 0.14 of sulfide per polypeptide; its UV–visible absorption spectra revealed an A280/A420 ratio of 3.4:1 and an apparent decrease of A420 upon dithionite reduction (Supplementary Fig. 3). These evidences are well consistent with the existence of one [4Fe-4S] cluster per polypeptide of HemN-like radical SAM enzymes31,32. As anticipated, in vitro enzymatic assays showed that only the two-component system, composed of reconstituted C10P (or reconstituted Swoo_2002) and C10Q, could catalyze the formation of 1 using the compound 6 as substrate in the present of SAM and sodium dithionite under strictly anaerobic conditions, whereas all other control experiments did not generate any unexpected products (Fig. 2j–p and Supplementary Fig. 5).

Compound 7 in the complete enzymatic reactions attracted our attention (Fig. 2o, p). Next, we isolated sufficient quantity of this compound for structural characterization from large-scale enzymatic assays. High-resolution mass spectrometry (HR-MS) and nuclear-magnetic resonance (NMR) spectroscopy revealed that 7 is a methylated derivative of 6 at the C-11 position (Fig. 2i and Supplementary Figs 6–8). The structure of 7 prompted us to question whether it is a biosynthetic intermediate. Subsequent enzymatic assays, however, showed that 7 could not be converted to 1 (Supplementary Fig. 9), which suggests that 7 may be an off-pathway product. In agreement with this observation, 7 is only detected under the in vitro enzymatic conditions but not in any fermentation cultures of related gene deletion mutants (Fig. 2d, e). We will discuss the relationship between 7 and 1 below.

These results also indicated that the radical SAM enzyme and the methyltransferase worked very closely. Indeed, isothermal titration calorimetry (ITC) studies confirmed that there exists an obvious interaction between Swoo_2002 and C10Q (KD = 8.0 μM) (Supplementary Fig. 10). Although we tried many different enzymatic reaction conditions, we could not improve the production of 1 (Supplementary Figs 11–15). Collectively, our genetic and biochemical results demonstrate that a two-component cyclopropanase system, composed of a radical SAM enzyme and a methyltransferase, catalyzes the formation of the cyclopropyl moiety in biosynthesis of 1.

Mechanistic studies of the HemN-like radical SAM enzyme

Bioinformatics analysis revealed that the C10P (and Swoo_2002) protein belongs to a HemN-like coproporphyrinogen III oxidase that is a radical SAM enzyme (Supplementary Fig. 16). Radical SAM superfamily proteins generally contain a highly conserved CxxxCxxC motif that coordinates a [4Fe-4S] cluster for binding and reductive cleavage of SAM33. As expected, mutagenesis experiments confirmed that the conserved cysteine motif of Swoo_2002 (C57A) is essential for the catalysis (Supplementary Fig. 17). Previous studies have revealed that there are two bound SAM molecules in the E. coli HemN crystal structure32. One molecule of SAM (SAM1) juxtaposed in close proximity to the [4Fe-4S] cluster as in other radical SAM enzymes is harnessed to yield a highly reactive dAdo radical34,35, whereas the function of the second SAM (SAM2) remains elusive. Multiple sequence alignment revealed that the two SAM-binding motifs are highly conserved within HemN-like radical SAM enzymes (Supplementary Fig. 16). Subsequent mutation of both motifs from Swoo_2002 proved that they are also required for the activity (Supplementary Fig. 18). During our enzymatic cyclopropanation reactions, we simultaneously observed the formation of 5′-deoxyadensine (5′-dA) and SAH (Fig. 3a–e and Supplementary Fig. 19). We then carried out a time-course analysis of the concentration changes of 1, 6, 7, 5′-dA, and SAH as the reaction proceeds. The results showed that 5′-dA and SAH were always produced at 1:1 stoichiometry and that the amount of 5′-dA (or SAH) was approximately equivalent to that of 1 plus 7 (Fig. 3f). These evidences may suggest that two molecules of SAM are consumed to yield one molecule of 5′-dA and one molecule of SAH during a single turnover with one SAM as a radical initiator and the other one as a methyl donor. This is consistent with the two bound molecules of SAM located in the HemN-like proteins32.

Next, we probed possible enzymatic intermediates during the cyclopropanation process from 6 to 1. Fortunately, we were able to detect a small peak (8) by high performance liquid chromatography (HPLC) analysis only from the highly concentrated complete enzymatic reactions via lyophilization (Fig. 3a–e and Supplementary Fig. 20). HR-MS analysis revealed that 8 exhibited molecular ions at m/z = 1088.3761 ([M]+) and 1110.3763 ([M-H + Na]+) (Fig. 3g). Further MS/MS analysis showed that the fragment ions produced included 250.0945, 384.1216, 704.2466, and 838.2749 (Fig. 3i), all of which are well-matched with several predicted fragmentation patterns of a covalent adduct formed between the substrate 6 and SAM (Fig. 3j).

The compound 8 may be a key intermediate during cyclopropanation, suggesting that the radical SAM enzyme probably mediates an addition of a SAM methylene radical to the C-11 position of the substrate 6. To verify this proposal, we used CD3-SAM instead of SAM for the complete enzymatic assays. The produced 5′-dA had a mass increment of +1 Da ([M + H]+ = 253.1084 for D-5′-dA) (Fig. 4b and Supplementary Fig. 21), consistent with the proposal that the dAdo radical generated from the first molecule of CD3-SAM abstracts a deuterium atom from the methyl group of the second molecule of CD3-SAM to give D-5′-dA and a SAM methylene radical. The intermediate 8 showed a mass shift of +2 Da ([M]+ = 1090.3809 and [M-H + Na]+ =1112.3801 for D2-8), which further confirms that it is derived from SAM (Fig. 3h). Both the generated 1 and 7 showed a mass increment of +2 Da ([M + H]+ = 706.2518 for D2-1 and 706.2519 for D2-7) suggesting that only two deuterium atoms from CD3-SAM were incorporated into the products (Fig. 4a–f). In addition, when the enzymatic assays were performed in D2O instead of H2O, 1 and 7 showed a mass shift of +1 Da (Fig. 4g–j); subsequent 2H NMR analysis of isolated D-7 revealed that the chemical shifts of deuterium were identical to those corresponding proton signals of C-12 and the methyl group at C-11 (C-11M) in 7 produced using H2O (Fig. 4k, l). This result suggests that the deuterium atom either located in the C-12 or C-11M position. Collectively, these labeling experiments implicate that the formation of 8 requires the participation of a SAM methylene radical and involves a carbon radical at C-12 (the intermediate 9) that abstracts a solvent-exchangeable proton, and that the off-pathway compound 7 may be produced from 8 through 10 via isomerization (Fig. 5).

Site-mutagenesis analysis of the methyltransferase

The conversion of 8 to 1 requires an intramolecular cyclization reaction with elimination of SAH as a co-product, which resembles an intermolecular methyltransferase activity. Based on our in vivo and in vitro results, this cyclization reaction is probably catalyzed by C10Q. Consistent with this proposal, we did not observe any SAM molecules in the purified C10Q protein (Supplementary Fig. 22). We initially attempted to isolate a little amount of 8, but it was so unstable that we were not able to obtain sufficient amount for enzymatic assays. Next, we investigated C10Q through mutagenesis experiments. Previous reports have revealed that the plant O-methyltransferases ChOMT and IOMT each employ a conserved histidine (His) residue to abstract a proton from the methyl acceptor group of substrate36. Subsequent multiple sequence alignment identified several conserved motifs from C10Q, such as a variant of the SAM-binding motif DxGxGxG (DxGxNxG for C10Q) and a conserved His residue likely for activation of the methyl acceptor group (Supplementary Fig. 23). As expected, mutation of the SAM-binding motif (C10Q D61A) completely eliminated its activity (Supplementary Fig. 24), while mutation of the conserved His (H138A) only led to production of 7 but not 1(Fig. 2q). Thus, the C10Q His-138 residue likely serves as a critical catalytic base responsible for deprotonation of the phenolic hydroxyl group at C-6 of the intermediate 8, and this will promote an intramolecular SN2 reaction to yield 1 with SAH as a leaving group.

Taken together, on the basis of our identification of the intermediate 8 and labeling experiments, we propose a catalytic mechanism for the cyclopropanation process (Fig. 5). Reductive cleavage of the first molecule of SAM1 from the HemN-like radical SAM enzyme yields a highly reactive dAdo radical, which will abstract a hydrogen atom from the activated methyl group of the second molecule of SAM2. A SAM methylene radical is thus produced and then adds to the C-11 position of the substrate 6 to generate a radical intermediate 9. This radical species abstracts a solvent-exchangeable proton to produce the intermediate 8. Subsequently, the His-138 residue from C10Q likely deprotonates the phenolic hydroxyl group (C-6) of 8, which triggers the intramolecular SN2 cyclopropanation to yield 1 with SAH as a co-product. Besides, the intermediate 8 may be non-enzymatically converted to the intermediate 10 containing an exocyclic double bond via elimination of SAH, followed by rapid and thermodynamic driving isomerization to give a methylated off-pathway compound 7.