Enzymatic Reconstitution and Biosynthetic Investigation of the Bacterial Carbazole Neocarazostatin A

Tricyclic carbazole is an important scaffold in many naturally occurring metabolites, as well as valuable building blocks. Here we report the reconstitution of the ring A formation of the bacterial neocarazostatin A carbazole metabolite. We provide evidence of the involvement of two unusual aromatic polyketide proteins. This finding suggests how new enzymatic activities can be recruited to specific pathways to expand biosynthetic capacities. Finally, we leveraged our bioinformatics survey to identify the untapped capacity of carbazole biosynthesis. N A 1 is a bacterial metabolite first isolated from the culture of Streptomyces sp. GP38 in 1991 and later from the soil bacterium Streptomyces sp. MA37, which consist of a tricyclic nucleus with two benzene rings (rings A and C) fused with a pyrrol ring (ring B) (Figure 1). This compound exhibits potent antioxidant activity and displays considerably lower IC50 values for inhibition of lipid peroxidation than those of the free radical scavengers, such as butylhydroxytoluene and the drug flunarizine. There have been considerable interests among medicinal chemists to develop 1 due to its pharmacological potential. Neocarazostatin A 1 belongs to a group of simple carbazole alkaloids (CAs) with aliphatic side chains. Considering the oxidative status of ring A, this group of CAs can be categorized into four subgroups: indole-fused dihydroxyl-type CAs including 1 as a representative, indole-fused monohydroxyltype CAs such as carazostatin, indole-fused ortho-quinone CAs such as carquinonstatin A, and nonaromatic type CA (Figure

N eocarazostatin A 1 is a bacterial metabolite first isolated from the culture of Streptomyces sp. GP38 in 1991 and later from the soil bacterium Streptomyces sp. MA37, which consist of a tricyclic nucleus with two benzene rings (rings A and C) fused with a pyrrol ring (ring B) ( Figure 1). 1,2 This compound exhibits potent antioxidant activity and displays considerably lower IC 50 values for inhibition of lipid peroxidation than those of the free radical scavengers, such as butylhydroxytoluene and the drug flunarizine. 2 There have been considerable interests among medicinal chemists to develop 1 due to its pharmacological potential. Neocarazostatin A 1 belongs to a group of simple carbazole alkaloids (CAs) with aliphatic side chains. Considering the oxidative status of ring A, this group of CAs can be categorized into four subgroups: indole-fused dihydroxyl-type CAs including 1 as a representative, indole-fused monohydroxyltype CAs such as carazostatin, indole-fused ortho-quinone CAs such as carquinonstatin A, and nonaromatic type CA ( Figure  1).
In our previous studies, we have delineated the functions of five key enzymes encoded in the biosynthetic gene cluster (BGC) of 1 (nzs), including the phytoene-synthase-like prenyltransferase NzsG and the P450 hydroxylase NzsA, 1 a thiamine diphosphate-dependent enzyme NzsH, 3 a freestanding acyl carrier protein (ACP) NzsE and a classical βketoacyl−acyl carrier protein synthase III NzsF. 4 Here, we demonstrated that in vitro reconstitution of NzsJ and I, together with NzsH and other necessary substrates and cofactors, enables the formation of the A ring of 1, which is spontaneously oxidized into ortho quinone-containing carba-zole, a similar observation discovered by Kobayashi and coworkers in parallel to our study. 5 Isotopic labeling studies demonstrate that one of the oxygens in the A ring is derived from water. Moreover, comparative genomics and a global network analysis of sequence similarity of nzsH, J, and I suggested that these gene homologues are clustered, and the occurrence of these BGCs in the bacterial kingdom suggests important biological functions in these organisms.
Our previous studies suggested that both nzsJ and nzsI genes are essential for the production of 1, and no obvious intermediates were accumulated in both nzsJ and I knockout mutants. 1 Bioinformatics analysis suggested that NzsJ is a putative FabH-like 3-ketoacyl-ACP synthase (KAS) III and that NzsI is a structural homologue of aromatase/cyclases (ARO/CYCs) that involve type II aromatic polyketide (PKS II) biosynthesis, strongly suggesting that Streptomyces sp. MA37 recruits new enzyme activities from the PKS II assembly line to the CA biosynthetic gene clusters to enable the evolution of biosynthetic capacity. During the manuscript preparation, reconstitution of the biosynthetic pathway of carquinostatin A, a structural homologue of 1, was reported 5 with the structural insights of one of the key biosynthetic enzymes, CqsB2, the homologue of NzsI with high sequence identity (80%).
In this study, we provide a full account of our work and show that NzsJ and NzsI catalyze the formation of the A ring of 1. To this end, we set out in vitro enzymatic assays. Overexpression of NzsJ and NzsI in S. lividans allowed us to isolate and purify the recombinant protein to near homogeneity with the estimated molecular weight of 35 kDa and 28 kDa, respectively, as determined by SDS-PAGE analysis ( Figure S1). Considering the instability of α-hydroxyl acyloin 3 from NzsH (the structure of 3 has been revised here compared to our previous structural interpretation 3 ), we first performed a one-pot reaction by incubating NzsJ (1 μM) with the enzymatic systems of NzsH and NzsE we've established before to generate the acyloin and 3-hydroxybutyryl-NzsE in situ, respectively; two new compounds with mass-to-charge ratios (m/z) of 290.1382 were formed as observed in the LC-HR-ESIMS analysis. N-Acetylcysteamine (SNAC) is commonly used as a simplified synthetic mimic of the reactive biosynthetic intermediates, such as acyl-ACP in the biosynthetic studies. 6,7 To isolate new compounds for structural elucidation, we chemically prepared R-3-hydroxybutyryl-SNAC ( Figures S2−4). Upon incubation of NzsJ with R-3hydroxybutyryl-SNAC and the enzymatic system of NzsH, the target molecules with identical molecular weights and UV absorption but different retention times (compound 5 and 6) were observed in the HR-ESIMS and HPLC analyses ( Figure  2A, Figure S5). Interpretation of NMR spectral and HR-ESIMS data suggested that 5 is likely to be an indole-acetyl ester despite the presence of the impurity (Figures S6−8). Compound 6 was proposed to be the isomer of 5.
To further confirm the structures of 5 and 6, we added NaBH 4 into the enzymatic mixture at the end of the assay, resulting in the appearance of two new ions, 7 and 8, with both of which have m/z of 292.1540 as observed in the HR-ESIMS ( Figure S9), further confirming the presence of only one ketone functional group. HR-ESLMS data for compounds 7 and 8 indicated a molecular formula of C 16 H 21 NO 4 , suggesting 7 degrees of unsaturation. The structure of 8 was proposed to be the reduced product of 5 by the interpretation of NMR spectral data (Figures S10−13, Table S1) coupled with theoretical calculations (Figures S14−16). Compound 7 was deduced to be the isomer of 8.
Analysis of the structures of 5, 6 and their derivatives, 7 and 8, led to speculation that 5 or 6 may be the substrate of NzsI. Surprisingly, an enzymatic assay of NzsI with the isolated 5 and 6 resulted in no new product, strongly suggesting that 5 and 6 were rearranged enzymatic products, a similar observation of recent parallel report. 5 This led us to speculate that 4 is the bonafide product of NzsJ that underwent a spontaneous α-ketol rearrangement reaction ( Figure 2B). Such a rearrangement is also observed in the interconversion from the dibenzo[b]fluorene skeleton to a benzo[g]chromene in the synthetic experiments 8 and the biogenetic interconversions between prekinamycin and isoprekinamycin 9 ( Figure S17). Density functional theory (DFT) calculations also confirmed that the structure of 4 has a higher energy than that of 5 or 6 (Table S2). It is noteworthy that no ions with a 4 Da increase were observed, strongly suggesting that the predicted intermediate 4 was not accumulated and underwent immediate rearrangement.
To investigate the roles of both NzsJ and NzsI, we performed a one-pot reaction of NzsI in the presence of NzsJ, 3-hydroxybutyryl-SNAC, and the NzsH system. A new product (9) was generated with m/z of 270.1125, as observed

The Journal of Organic Chemistry
Note in the HR-ESIMS analysis ( Figure 3A, Figure S18). In a control assay with boiled NzsI and in other control assays lacking either the NzsH system or NzsJ or SNAC substrate, the formation of 9 was not observed.
To confirm the structure of 9, a large-scale biotransformation reaction was carried out to afford the isolation of 9 (8 mg). The inspection of 1D and 2D NMR spectral data demonstrated that 9 is an ortho-quinone carbazole (Figures S19−22). Intriguingly, bioinformatics analysis suggested that NzsI is not an oxidative enzyme, leading us to speculate 10 is the bona fide enzymatic product of NzsI but was oxidized spontaneously ( Figure S23A). The presence of dithiothreitol (DTT, 1 mM) in the biotransformation of NzsH/J/I indeed extended the existence of the ortho-dihydroxyl compounds 10 as observed in HPLC and HR-MS analyses ( Figure S23B,C).
This led us to propose that NzsI may install the hydroxyl group at the C4 of 4 from one water molecule. To verify this hypothesis, we carried out the biotransformation using isotopically enriched water. An assay of NzsI with the substrates in the buffer system enriched with H 2 18 O (15%) was performed. Indeed, isotopically enriched 9a ([M+2] + : 18%) was observed ( Figure 4A,B), as determined by HR-ESIMS analysis and by comparison with the MS pattern of 9 ( Figure S24). To further confirm that the hydroxyl group is indeed located at the C4 position in the A ring, a large-scale biotransformation reaction was performed to afford the isolation the mixture of 9 and isotopically enriched 9a. The inspection of 13 C{ 1 H} NMR of the resultant product demonstrated the 18 O-induced isotope shifts at the ketone of C4 of 9 ( Figure 4C, Figure S25). The magnitude of this isotope shift is 0.04 ppm for the C4 attached to 18 O, consistent with the previous observation. 10 Taken together, these experiments confirmed that NzsI utilizes the acid−base chemistry to catalyze the cyclization, followed by nucleophilic aromatization of water to afford the catechol moiety of carbazole metabolites, strongly suggesting that NzsI is a new type of cyclases/aromatases.
Based on the oxygen pattern in the A ring, this group of bacterial carbazole metabolites can be categorized into four subgroups: monohydroxyl type, dihydroxyl type, orthoquinone type, and nonaromatic type ( Figure S26). We propose that the A ring of the carbazole is assembled by NzsH, NzsJ, and NzsI or corresponding enzyme homologues. The assembly of the A ring is initiated by the carboligation of 2 and pyruvate in a reaction catalyzed by NzsH or its homologues to afford the corresponding α-keto acid 3. Depending on the bioavailability of the Acyl-CoA species in the producing strains, an unidentified acyltransferase will catalyze the transthiolation reaction to load Acyl-CoA into NzsE or its homologues to generate acyl-tethered thioesters. NzsJ or its homologues will utilize both 3 and acyl-tethered thioesters to mediate the decarboxylation-driven retro-aldol reaction to generate the intermediate, followed by cyclization via abstraction of the acidic proton at the C4 position, spontaneous dehydration, and water-based nucleophilic aromatization to yield the catechol motif. Methylation or transamination followed by acylation would result in dihydroxyl type carbazoles. Reduction on the key catechol-type intermediate will generate the nonaromatic type. Currently, only one metabolite without an aromatic feature in the A ring has been isolated from Streptomyces sp. BCC26924. 11 If a reduction is involved in the NzsI-mediated reaction, the hydride-based nucleophilic aromatization would afford the monohydroxyl type of carbazole metabolites.
The presence of NzsI homologues in the bacterial kingdom was investigated. A blast search using NzsI as the query in the NCBI database allowed identification of a collection of NzsI-  The Journal of Organic Chemistry Note like homologues that share a high amino acid sequence identity (43%−80%). The corresponding genes are found in the genome of microorganisms, including the Gram-positive actinomycetes, the well-studied S. cattleya, the Gram-negative bacteria including the thermophilic bacterium Legionella gratiana, the soil-dwelling myxobacterium Sorangium cellosum, and the cyanobacterium Scytonema tolypothrichoides ( Figure 5, Table S3). Strikingly, adjacent to these nzsI-like genes are nzsH-like and nzsJ-like genes in some cases (Table S3). In silico analysis also indicated that these NzsJ-like open reading frames (ORFs) also share a high amino acid sequence identity (24%− 73%) to NzsJ (Table S3). Furthermore, MEME prediction 12 demonstrated that all of the NzsI-like ORFs share highly conserved motifs with NzsI ( Figure S27), whereas all of the NzsJ-like ORFs share the conserved catalytic triad C−H−N with NzsJ and other KAS III proteins ( Figure S28), suggesting that these ORFs should process the same chemical reactions as those for NzsI and NzsJ, respectively. However, analysis of the genes in the close proximity of these orfs suggested that these identified gene clusters encode completely different sets of auxiliary enzymes that were predicted to modify the CA ring system or the fatty acid chains. It is likely that these strains have the potential to produce not-yet-discovered new CA-like metabolites.
In conclusion, in vitro reconstitution of two unusual enzymes, NzsJ and NzsI, together with NzsH and other necessary substrates and cofactors, demonstrate the assembly of the A ring moiety of the bacterial carbazole metabolite 1. Isotopic labeling studies demonstrated that the hydroxyl group at C4 position of the A ring is originated from one water molecule. Bioinformatics analysis further uncovered that the homologues of NzsI are widespread in bacteria associated with NzsH and NzsJ homologues, suggesting that carbazole-type metabolites are ubiquitous in the bacterial kingdom.

■ EXPERIMENTAL SECTION
General Materials and Methods. All chemicals and solvents were obtained from Sigma-Aldrich except where noted. Oligonucleotide primers were synthesized by GenScript (Nanjing, China). The DNA sequencing of PCR products and all constructive plasmids was performed by TSINGKE Biological Technology (Wuhan, China). Restriction endonucleases and T4 ligase were purchased from NEB. High-fidelity DNA polymerase KOD was purchased from TOYABO. A typical PCR reaction contained 10−100 ng of the DNA template, 0.5 μM of each primer, 2 μL of 10× KOD buffer, and 0.3 μL of highfidelity KOD polymerase. Thermocycling was carried out in Bio-Rad C1000 thermocycler. E. coli DH10B was used as a cloning host. E. coli ET12567/pUZ8002 was used for intergeneric conjugation between E. coli and Streptomyces. E. coli DH10B, and E. coli ET12567/pUZ8002 were cultured in Luria−Bertani (LB, tryptone 1%, yeast extract 0.5%, NaCl 1%) or LB agar medium at 37°C. S. lividans TK24 and S. lividans 1326 were grown on MS agar plates (soybean 2%, D-mannitol 2%, agar 2%) at 28°C for sporulation and in the YEME medium (glucose 1%, tryptone 0.5%, yeast extract 0.3%, malt extract 0.3%, sucrose 10.3%) for protein overexpression. The MS medium with additional 10 mM MgCl 2 was used for intergeneric conjugation. HPLC analysis was carried out on a Shimadzu (Kyoto, Japan) HPLC instrument equipped with a degasser (DGU-20A3), an autosampler (SIL-20A), a column oven (CTO-20A) and two pumps (LC-20AT), and Phenomenex columns (C18, 5 μm) 250 mm × 4.6 mm. HPLC conditions for analysis of the one-pot reaction: the C18 column was pre-equilibrated with 20% B and developed at a flow rate of 0.8 mL/ min, 0−20 min, a linear gradient from 80% A to 20% A; 20−25 min, a linear gradient from 80% A to5% A; 25−27 min, constant with 5% A; 27−30 min, a linear gradient to 80% A; 30−35 min, constant with 80% A; UV absorption was monitored at 280 nm. Solvent A was 0.1% formic acid in H 2 O, and solvent B was 0.1% formic acid in CH 3 CN. HR-ESI-MS analysis was carried out in the positive ion mode by using a Thermo Scientific LTQ XL Orbitrap mass spectrometer equipped with a Thermo Scientific Accela 600 pump (Thermo Fisher Scientific Inc.). Each of the LC conditions were described as above. All MS analysis parameters were as follows: 45 V capillary voltage, 45°C capillary temperature, auxiliary gas flow rate 10 arbitrary units, sheath gas flow rate 40 arbitrary units, 3.5 kV spray voltage, and 50−1000 Amu mass range (maximum resolution 30 000).
Chemical Synthesis of 3-Hydroxybutyryl-SNAC. 3-Hydroxybutyryl-SNAC was synthesized as a simplified mimic of the reactive biosynthetic intermediate, which can be recognized by NzsJ in vitro. Construction of NzsJ and NzsI Overexpression Plasmid. The genes nzsJ were amplified from S. sp. MA37 genomic DNA using the primer pair His-NzsJ-F and His-NzsJ-R (Table S4) by high-fidelity PCR. The PCR-amplified products were purified and then cloned into a NdeI/HindIII cleaved expression vector pWDYHS01 by using an InFusion HD cloning kit (Clontech) to yield pWDY900. pWDY900 was then transformed into S. lividans TK24, and the resulting colonies were selected by the aac(3)IV resistant marker and confirmed by PCR to obtain the recombinant strain WDY900. The strain WDY900 was used for overexpression of the His-tagged NzsJ in Streptomyces. The genes nzsI were PCR amplified from MA37 genomic DNA using the primers His-NzsJ-F/R and His-NzsI-F/R (Table S5). The PCR products were purified from an agarose gel and then cloned into an NdeI/HindIII linearized vector, pWDYHS01, using an InFusion HD cloning kit (Clontech). The ligation mixture was transformed into competent E. coli DH10B cells by heat shock. The recovery culture was plated on LB agar containing apramaycin (50 μg/mL) to screen Figure 5. Sequence similarity network analysis of homologues of NzsH, NzsI, and NzsJ. Each node in the network represents a homologous protein, and each edge represents the pairwise connection between two proteins with a blastP E value <1 × 10 −10 .

The Journal of Organic Chemistry
Note the positive clones, and then the positive assembly was confirmed by digestion and sequencing.
Expression and Purification of NzsI and NzsJ. E. coli ET12567/pUZ8002 was first transformed with the verified plasmids and then conducted general intergeneric conjugation with the S. lividans TK24 or 1326 to yield the protein overexpressed strain WDY900 and WDY901 (Table S6), respectively. The overexpression strain WDY900 or WDY901 was inoculated into 50 mL of the TSBY medium supplemented with 50 μg/mL apramycin and grown at 28°C with shaking at 200 rpm for 2 days. The preculture was transferred to 500 mL of the YEME medium supplemented with 50 μg/mL apramycin at 28°C with shaking at 200 rpm for 2 days. Protein expression was induced with the addition of 0.5 mM (final concentration) thiostrepton. After an additional 60 h of incubation, the cells were harvested by centrifugation at 5000g for 20 min at 4°C. The cell pellets collected by centrifugation were resuspended in icecold lysis buffer (20 mM Tris-HCl, 300 mM NaCl, 10 mM imidazole, and 10% (v/v) glycerol, pH 8.0) and disrupted by a high-pressure homogenizer machine (Avestin, EmulsiFlex-C3). Cell debris was removed via centrifugation (4°C, 12 000g, 30 min), and the supernatant was filtered before loading onto a 5 mL HisTrapHP column (GE Healthcare). The HisTrapHP column was washed by six concentration-step elution buffers, with 15 mL for each step (20 mM Tris-HCl, 300 mM NaCl, 10% (v/v) glycerol, along with 20 mM, 50 mM, 100 mM, 150 mM, 200 mM, or 300 mM imidazole, pH 8.0). The elution fractions were checked by SDS-PAGE, and then the desired elution fractions were concentrated using centrifugal filter units (Milipore, 10 000 MWCO, Merck). The concentrated protein solution was subsequently desalted using a PD-10 column (GE Healthcare) pre-equilibrated with the elution buffer (20 mM Tris-HCl, 100 mM NaCl, and 10% (v/v) glycerol, pH 8.0). The desalted effluent was centrifuged again and was then aliquoted, flash-frozen in liquid nitrogen, and stored at −80°C. SDS-PAGE and spectrophotometric analysis (Bradford assay kit, Promega, BSA used as standard, absorbance at 595 nm) were used to check the purity and concentration of the protein.
In Vitro Enzymatic Assay of the NzsH/J One-Pot Reaction. The NzsH/J one-pot reaction was performed by incubating 1 μM NzsJ and 1 mM 3-hydroxybutyryl-SNAC with the enzyme systems of NzsH. The reaction mixture was incubated at 30°C for the scheduled time and quenched by an equal volume of ice-cold methanol. The clarified supernatant was obtained by centrifugation at 13 000g for 30 min and then subjected to HPLC and HR-ESIMS analysis. The NzsH and boiled NzsH/J added mixture were used as control reactions.
In Vitro Enzymatic Assay of the NzsH/J/I One-Pot Reaction. The NzsH/J/I one-pot reaction was performed by adding 1 μM NzsI in the mixture of the NzsH/J one-pot system, as described above. The mixture was incubated at 30°C for the scheduled time and quenched by an equal volume of precooled MeOH. The clarified supernatant was obtained by centrifugation at 13 000g for 30 min and then subjected to HPLC and HR-ESIMS analysis. The NzsH, boiled NzsH/J/I added, and NzsH/J one-pot mixture were all used as control reactions. The NzsH/J/I reactions with DTT were performed by adding 1 mM DTT in the one-pot system.
Isotopic Labeling Experiments. To verify that the oxygen attached to C4 of 9 (1-(2-hydroxypropyl)-2-methyl-carbazole-3,4(9H)-dione) was derived from water, the NzsH/J/I one-pot reaction was performed by incorporating 15% 18 O-water. The labeled compound 9a was obtained with theoretically consistent mass data on the HR-ESIMS analysis. We then scaled up the enzymatic reaction to 50 mL with a 15% incorporation ratio of 18 O-water in order to accumulate 9a for 13 C{ 1 H} NMR spectra analysis. The resulting incorporation ratio was calculated by integrating the peak area of 9a on the mass chromatogram as follows: 18 O incorporation ratio % = peak area 9a/(peak area 9a + peak area 9) × 100.
Structural Characterization of Reaction Products. The compound 8 (3,5-dihydroxyhexan-2-yl 2-(1H-indol-3-yl) acetate) was isolated and purified from the enzymatic reaction, evaporated to dryness, and resolved in CD 3 OD for structure characterization on NMR analysis. The compound 9 was isolated and purified from the enzymatic reaction, evaporated to dryness, and resolved in DMSO-d 6 for structure characterization on NMR analysis. The NMR data of the compound was collected on a Bruker Avance 600 MHz NMR spectrometer.  Sequence Similarity Network Analysis. The homologues of NzsH, NzsI, and NzsJ from different microorganisms were obtained by blastp search by using NzsH, NzsI, and NzsJ as query sequences, respectively. Adjacent to these nzsI-like genes, there are nzsH-like and nzsJ-like genes located in some gene clusters (Table S3). The proteins of these nzsI-like, nzsH-like, and nzsJ-like genes were selected for further network analysis. The network was constructed by an all-by-all blastp comparison of each sequence against each other sequence and was generated using the EFI-Enzyme Similarity Tool. 13 Sequence similarity networks were visualized in Cytoscape 3.6.1 14 to illustrate the distribution of NzsH, NzsI, and NzsJ cluster in microorganisms. The nodes were arranged by using the yFiles organic layout with manual adjustment. Accession numbers were listed in Table S3. The alignment identity of NzsI homologues was visualized with a heat map.
Sequence Alignment Analysis. The multiple sequence alignment of NzsJ, KAS III proteins, KAS III-like proteins, and NzsJ homologues were performed by ClustalX2 15 and visualized by ESPript 3.0. 16 To determine conserved motifs in 20 NzsI-like sequences analyzed by blast and cluster mining, MEME 12 search was performed with the following parameters: mod = zoops, nmotifs = 30, minw = 6, and maxwidth = 50.
Computational Methods for Theoretical Calculation. HR-ESIMS data for compound 8 indicated a molecular formula of C 16 H 21 NO 4 , suggesting 7 degrees of unsaturation in the structure of compound 8. Detailed analysis of 1 H, 13 C{ 1 H}, COSY, HSQC, and HMBC data enabled the construction of the spin systems. A good correlation (r 2 = 0.9984) of predicted 13 C NMR versus experimental 13 C{ 1 H} NMR data suggested the structure is possibly correct. 17 1D and 2D NMR data were inputted to the ACD/Laboratories Structure Elucidator and all possible structures calculated within an average difference of 4 ppm difference between calculated and experimental chemical shifts, yielding 48 possible structures. The top 5 calculated structures were calculated by the Structure Elucidator with 13 C chemical shift deviations between the experimental and the predicted HOSE-code (dA), 18 Artificial Neural Net (dN), and Incremental Method. After ranking using the Neural Network Match Factor, the resulted number one candidate is inconsistent with the predicted structure of 8.

The Journal of Organic Chemistry
Note ■ ASSOCIATED CONTENT

* S Supporting Information
The Supporting Information is available free of charge at https://pubs.acs.org/doi/10.1021/acs.joc.9b02688.
Additional tables of primers, strains, and protein accession numbers; NMR, HR-ESIMS, and UV spectra of reactions and products; sequence analysis of NzsJ and NzsI; DFT computation data and parameters (PDF) ■ AUTHOR INFORMATION Corresponding Authors *E-mail: h.deng@abdn.ac.uk. *E-mail: yu_yi@whu.edu.cn.