Structural insights into dehydratase substrate selection for the borrelidin and fluvirucin polyketide synthases
Engineered polyketide synthases (PKSs) are promising synthetic biology platforms for the production of chemicals with diverse applications. The dehydratase (DH) domain within modular type I PKSs generates an α,β-unsaturated bond in nascent polyketide intermediates through a dehydration reaction. Several crystal structures of DH domains have been solved, providing important structural insights into substrate selection and dehydration. Here, we present two DH domain structures from two chemically diverse PKSs. The first DH domain, isolated from the third module in the borrelidin PKS, is specific towards a trans-cyclopentane-carboxylate-containing polyketide substrate. The second DH domain, isolated from the first module in the fluvirucin B1 PKS, accepts an amide-containing polyketide intermediate. Sequence-structure analysis of these domains, in addition to previously published DH structures, display many significant similarities and key differences pertaining to substrate selection. The two major differences between BorA DH M3, FluA DH M1 and other DH domains are found in regions of unmodeled residues or residues containing high B-factors. These two regions are located between α3–β11 and β7–α2. From the catalytic Asp located in α3 to a conserved Pro in β11, the residues between them form part of the bottom of the substrate-binding cavity responsible for binding to acyl-ACP intermediates.
KeywordsPolyketide Dehydratase Borrelidin Fluvirucin
Acyl carrier protein
Multiple sequence alignment
Polyketide natural products are one of the largest classes of secondary metabolites, possessing vast structural and chemical diversity. The collinear biosynthetic logic of type I modular polyketide synthases (PKSs) make them a promising synthetic biology platform for the production of existing and novel compounds. The dehydratase (DH) domain within type 1 modular PKSs is responsible for the dehydration of specific C-3 hydroxyacyl-acyl carrier protein (ACP) intermediates, resulting in a corresponding enoyl-ACP . These unsaturated ACP-tethered intermediates can be either reduced by an associated enoyl reductase (ER) domain or result in the production of alkenes in even-to-odd positions in the final polyketide structure . The dehydratase domain from type 1 modular PKSs contains a canonical double hot-dog fold motif with an invariant His/Asp catalytic dyad. These structural features are well conserved and well depicted in DH domain crystal structures from erythromycin , curacin [3, 15], rifamycin  and the gephyronic acid PKSs . Despite having several crystal structures of DH domains, the structural basis for substrate selectivity and specificity is not completely understood. There is a need to understand substrate selection to enable effective design of synthetic PKSs with different DH substrates. Two main factors important for substrate selection by DH domains include the stereochemistry of the C-3 hydroxyacyl group and the chemical structure of the acyl-ACP intermediate past the C-3 hydroxyacyl position. Dehydration proceeds via a syn-coplanar elimination of water and is, therefore, sensitive to the stereochemical configuration of the C-3 hydroxyacyl-ACP substrate , which is determined enzymatically by the ketoreductase (KR) domain within the module. Thus, DH domains are tied to KR domains that precede them biochemically .
Most of the structural deviations between BorA DH M3, FluA DH M1 and other DH domains are evident in multiple loop regions connecting the α-helices and β-sheets of the core double-hotdog motif (Fig. 2e). Small loops connecting β1–β2, β5–β6, β6–β7 and β12–β13 displayed small structural deviations. However, the largest deviations were located in two loop regions connecting β7–α2 and α3–β11 (Fig. 2F, Fig. S2). This was further supported by either high B-factors within these residues or missing electron density for residues in the loop region (Fig. 2c, d). In the case of BorA DH M3, we were unable to model 11 residues between α3 and β11, suggesting a highly flexible region within the DH domain (Fig. S2A). A lack of modeled residues between α3 and β11 is also observed in the erythromycin M4, phthiocerol dimycocerosate, curacin F and the rifamycin M10 DH domains (Fig. S2C, E, G, I). Unlike the BorA DH M3, all residues between α3 and β11 were modeled in the FluA DH M1 (Fig. S2B). A closer inspection of the residues between α3 and β11 FluA DH M1 displayed partially higher B-factors. These higher B-factors between α3 and β11 FluA DH M1 are also observed in the curacin H, J, K and gephyronic acid DH domains (Fig. S2D, F, H, J). Overall, B-factor analysis and lack of electron density between α3–β11 and β7–α2 in type I DH structures suggest a highly flexible region within the domain.
The catalytic region of the DH domain contains the catalytic His and Asp dyad responsible for the dehydration reaction via a syn-coplanar elimination of water. In addition to the catalytic His and Asp, both BorA DH M3 and FluA DH M1 contain conserved Leu and Tyr residues that outline the catalytic region, the latter orienting a water molecule (Fig. S5). The catalytic His is located at the beginning of β3 and is on average less than 3.8 Å away from the Asp, located within α3. Recent findings in the fostriecin DH domain suggest DH domains utilize a single-base mechanism, where the active site His residue acts as the base to deprotonate C-2, subsequently protonating the C-3 hydroxyl group to promote C–O bond-cleavage and elimination of water . The carboxylate group of the Asp likely binds and orients the hydroxyl group of the substrate in the favored conformation .
The acyl intermediate-binding region can be defined as the residues outlining the bottom of the substrate-binding cavity, outside of the catalytic region. These residues bind to the chemical moiety past the C-3 hydroxyacyl group of acyl-ACP intermediate. The polyketide intermediate past the C-3 hydroxyacyl group can vary in chemical structure, length, and atom heterogeneity, and is dependent on both the upstream PKS module enzymatic architecture and the diverse starter-units incorporated in the nascent polyketide intermediate. The BorA DH M3 can select for a cyclopentane-carboxylate-containing intermediate (Fig. 1a), while the FluA DH M1 can select for an amide-containing dipeptidyl intermediate (Fig. 1b). Both of these substrates require interactions with residues of distinct chemical properties on the acyl-binding region of the DH domain. A thorough inspection of the polyketide acyl-binding region in BorA DH M3 and FluA DH M1 identified residues located between α3–β11 and α2–β8 (Fig. 3). Surprisingly, the same flexible loop region between α3 and β11, where residues are either missing (BorA DH M3) or have higher B-factors (FluA DH M1), form part of the acyl-binding region. Amino acid preference and multiple sequence alignment analysis suggest that most DH domains have a variant loop region between α3 and β11 (Fig. 5b, Fig. S6). Starting from the catalytic Asp in the middle of α3 to a conserved Pro at the beginning of β11, the loop region can vary in length and amino acid composition. These structural insights suggest a putative binding region responsible for selecting the diverse chemical moieties of different polyketide intermediates within DH domains.
Docking simulations of PPant-tethered substrate intermediates
In our efforts to further validate the importance of the acyl intermediate-binding region for substrate binding and identify key substrate-binding residues, we initially aimed to co-crystalize BorA DH M3 and FluA DH M1 with their corresponding starter units. These starter units included the trans-cyclopentane-dicarboxylic acid and β-alanine, in both the free acid and the N-acetyl-cysteamine thioester (SNAC)-coupled form. However, no electron density for the corresponding substrates was observed. The lack of substrate binding may be due to high specificity of both BorA DH M3 and FluA DH M1 for their native PPant-tethered polyketide intermediates 1, 4 (Fig. 1). Using an inactive form of BorA DH M3 and FluA DH M1 with the corresponding native PPant-substrates would greatly increase our efforts in obtaining a substrate–DH domain complex structure. This is supported by the inactive H/F959 PpsC DH domain complexed with a PPant-containing trans-dodec-2-enoyl-CoA substrate . Given the complexity to synthesize the native PPant-tethered borrelidin  and fluvirucin intermediates for co-crystallization experiments, we decided to conduct in silico docking as an alternative approach. Our initial docking analysis was focused on the crystal structure of FluA DH M1 and its corresponding native substrate 4. The lack of 11 unmodeled residues in the BorA DH M3 acyl intermediate-binding region would yield inconclusive in silico docking results (Fig. 3a). Therefore, we opted not to conduct substrate docking analysis with the BorA DH M3 model.
Several residues in the acyl intermediate-binding region were in the vicinity of the nitrogen-containing dipeptide moiety of 4 (Fig. 4). The variable loop region between α3 and β11 (light blue) contained polar residues His 235 and Glu 244 within 4.0 Å from the terminal amino group of 4. In addition, residues between α2 and β8 (dark blue), on the distal side of the variable loop region between α3 and β1, interacted with the dipeptide moiety of 4. Residues between Ser 185 and the conserved Tyr 189 were in close contact with 4. Surprisingly, the PPant-tethered fluvirucin intermediate 4 does not accommodate the entire catalytic region of the acyl intermediate-binding region. On average, there is a 9.5-Å distance gap between the terminal amino moiety of 4 and the bottom of the substrate-binding cavity. The bottom of the cavity contains several variant polar residues (Glu 241 and Gln 242). The gap identified between these polar residues with 4, the high B-factors associated within the invariant loop and the lack of modeled residues for similar type I DH domains suggest this structural region may adopt a distinct conformational change upon binding to the native substrate. This is further supported by the small structural conformational changes observed between the variable loop regions α3–β11 of the apo FluA DH M1 crystal structure and the energy minimized docked PPant intermediate 4-FluA DH M1 complex structure (Fig. S7). Cα superposition results in RMSD values of 2.10 Å within the 18 residues of the α3–β11 loop region.
Analysis of DH domain conservation
With the aim of identifying residues and/or structural regions that may play a role in substrate binding past the C-3 hydroxyacyl position, we looked for residues that may be easily substituted with others to accommodate the diversity of the acyl-ACP substrate intermediate. A closer inspection of DH domains showed two regions with high H values. These higher H values were present in the loop regions between β7–α2 and α3–β11 (Fig. 5b). As observed in the BorA DH M3, FluA DH M1 and other DH structures, these two regions with higher H values are located in the same region of unmodeled, high-B-factor residues in the PDB structures. Residues between α3 and β11, and within α2 form part of the acyl intermediate-binding region (Figs. 5, S6). These results further support our DH domain structural and in silico docking analysis regarding the importance of α3–β11 and α2 in substrate selection. These two variant regions between α3–β11 and β7–α2 may possess distinct amino acid properties across different DH domains and may be important for the structural architecture of the acyl intermediate-binding region. In particular, the residues between α3 and β11 and within α2 may play a direct role in substrate selectivity past the C-3 hydroxyacyl position.
Substrate selection by DH domains is attributed to both the stereochemistry of the C-3 hydroxyacyl group and the chemical structure on the acyl-ACP intermediate past the C-3 hydroxyacyl position. The recent structural and biochemical insights of various DH domains have significantly advanced our understanding of stereoselection by DH domains [3, 15, 17, 25, 26]. However, DH substrate selection pertaining to the acyl-ACP intermediate past the C-3 hydroxyacyl position remains elusive. For example, the Rif DH M10, CurK DH, and most DH domains act on a C-3 hydroxyl-ACP intermediate. Yet, the Rif DH M10 from rifamycin is selective towards a C-18, naphthoquinone-containing intermediate [17, 38] while the CurK DH acts on a shorter thiazoline, cyclopropyl-containing intermediate [3, 8]. A key question in DH substrate selectivity is what residue(s) and/or structural features within the DH domain are responsible for selection of the acyl intermediate past the C-3 hydroxyacyl position. Further structural and biochemical insights on DH domains would provide a better understanding of substrate selection past the PPant and C-3 hydroxyacyl moiety.
To further improve our current understanding of DH selectivity towards different acyl-ACP intermediates past the C-3 hydroxyacyl position, we have performed structural analysis on two chemically diverse DH domains. The first DH domain, isolated from the third module in the borrelidin PKS, is specific towards a trans-cyclopentane-carboxylate-containing polyketide substrate. The second DH domain, isolated from the first module in fluvirucin B1 PKS, accepts an amide-containing polyketide intermediate. Both DH domains maintain the canonical double-hotdog fold motif with the invariant His/Asp catalytic dyad. Surprisingly, a structural comparison of the BorA DH M3, FluA DH M1 and other type I DH domains show few significant deviations from each other. The two major differences between BorA DH M3, FluA DH M1 and other DH domains are found in regions of unmodeled residues or residues containing high B-factors. These two regions are located between α3–β11 and β7–α2. From the catalytic Asp located in α3 to a conserved Pro in β11, the residues between them form part of the bottom of the substrate-binding cavity responsible for binding to the acyl-ACP intermediates past the C-3 hydroxyacyl position. The distal region from α3 to β11 is also part of substrate-binding cavity and is located in α2–β8. The loop connecting β7–α2 is one of the most variable regions in modular type I DH domain structures. The variable β7–α2 loop region may provide flexibility to part of the substrate-binding cavity located in α2–β8 (Fig. 4). These structural insights suggest both α3–β11 and β7–α2 are important for binding of the acyl-ACP intermediate past the C-3 hydroxyacyl position. This was further supported by our in silico docking analysis of the FluA DH M1 and prediction of conserved residues through Shannon entropy measurements (Figs. 4, 5). Structural evidence for this can be observed in the PpsC DH structures in complex with trans-dodec-2-enoyl-CoA (PDB ID: 5NJI) or crotonyl-CoA ((PDB ID: 5I0K). Comparison of the α3–β11 and β7–α2 regions in both structures displays key differences. More residues were modeled in the α3–β11 of PpsC DH structure in the presence of the longer trans-dodec-2-enoyl-CoA substrate, suggesting higher stability of the α3–β11 region upon binding of a substrate that closely resembles the native acyl-ACP intermediate.
The identification of both the α3–β11 and β7–α2 regions involved in substrate selection may provide insights into rational engineering of DH domains. In our efforts to improve DH-mediated dehydration in adipic acid production using the engineered BorA2 PKS , we tested various DH domains and chimeric DH domains with α3–β11 loop swaps in vitro (Fig. S8a). Surprisingly, all chimeric DH domains containing the α3–β11 loop swaps were soluble (Fig. S8b). This suggests that a key secondary structure is maintained within the DH domain’s α3–β11 loop swap region. However, a close inspection of adipic acid production using the chimeric DH domains was inconclusive (Fig. S8c). We speculate that the lack of conclusive results may be due to both the high background of adipic acid that is likely derived from E. coli during protein purification, and low efficiency of the BorA2 PKS to produce adipic acid. This assay may not be optimal for testing chimeric DH activity. Nonetheless, generating soluble and stable chimeric DH domains is a first step towards DH engineering. Further efforts in investigating the activity of chimeric DH domains can be simplified by testing standalone chimeric DH domains containing the α3–β11 loop swaps with more naturally relevant acyl-ACP, acyl-PPant or acyl-SNAC substrates .
Our structures of the BorA DH M3 and FluA DH M1 provide further structural evidence of substrate selection by DH domains within modular type I PKSs. We present both a sequence-structural analysis between DH domains and identify significant similarities and key differences pertaining to substrate selection. The work presented here in combination with existing DH domain studies should facilitate further engineering efforts of modified PKSs that can process non-natural substrates with improved catalytic properties.
Materials and methods
Cloning of BorA DH M3 and FluA DH M1
Vector backbone template originated from JPUB_008800 plasmid (Table S1). The genes encoding for BorA DH M3 and FluA DH M1 were PCR amplified from the pDVA00936 and pTL-A01 plasmids, and cloned into a pET28a vector using Gibson assembly methodology . The j5 DNA assembly design automation software  was utilized to generate Gibson assembly primers for the BorA DH M3, FluA DH M1 and pET28a vector backbone. Primer sequences are available in Table S1. Polymerase chain reaction amplification of the BorA DH M3, FluA DH M1 and pET28a backbone was conducted using Phusion Hot Start II DNA polymerase (Thermo Fisher) using the vendor’s recommended protocol at an annealing temperature of 65 °C and 1 min extension time for the DHs and 3 min extension time for the pET28 backbone. The pET28a PCR product was Dpn1 digested (Thermo Fisher) using the vendor-recommended protocol and subjected to PCR clean up (Zymo Research). The DH genes and pET28a PCR product were ligated using Gibson assembly master mix (New England BioLabs), following the vendor-recommended protocol. Both DH constructs were designed with a thrombin-cleavable N-terminal 6 × His-tag. The strains and plasmid sequences utilized in this study are listed in Table S1. All strains and plasmid sequences may be accessed and requested through the Joint BioEnergy’s public registry (https://public-registry.jbei.org/folders/412) .
Protein expression and purification
The recombinant BorA DH M3 and FluA DH M1 containing an N-terminal 6 × His-tag were produced in BL21 (DE3) E. coli cells (Novagen). Cells containing the pET28-DH plasmids were grown to OD600 = 0.8 at 37 °C in TB medium containing 50 µg/mL kanamycin. The cell cultures were cooled to 18 °C and expression was induced using 1 mM IPTG. The cell cultures were incubated for an additional 16 h at 18 °C and harvested by centrifugation at 5525 r.c.f. for 10 min. The cell pellets were resuspended in 50 mM Hepes, pH 7.6, 10% glycerol, 10 mM imidazole, 300 mM NaCl. Resuspended cells were cooled on ice for 30 min and the cells were disrupted using sonication. The cell debris was cleared by centrifugation at 21,036 r.c.f. for 1 h. The supernatant was collected and batch bound to HisPur TM Cobalt Resin (Thermo Scientific) for 1 h at 4 °C. BorA DH M3 and FluA DH M1 were purified according to the manufacturer’s instructions using an imidazole step-gradient. Fractions containing pure protein were determined by SDS-PAGE and fractions containing either BorA DH M3 or FluA DH M1. Removal of the N-terminal 6 × His-tag was conducted by incubating the fractions containing the proteins at 14 °C for 30 h with thrombin from bovine plasma (Sigma-Aldrich) at a concentration of 2 U/mg of recombinant protein and 3.5 mM CaCl2. After thrombin digestion and assessment of N-terminal 6 × His-tag cleavage via SDS-PAGE, we dialyzed the DH domains against 50 mM Hepes, pH 7.6, 5% glycerol. Removal of thrombin and further purification of BorA DH M3 and FluA DH M1 were conducted by anion-exchange chromatography using HiTrap Q FF (GE Healthcare) according to the manufacturer’s instructions. Purified BorA DH M3 and FluA DH M1 were dialyzed against crystallization buffer, which consisted of 25 mM Hepes, pH 7.6, and 1 mM dithiothreitol (Fig. S9). Both BorA DH M3 and FluA DH M1 were concentrated to 12 mg/mL for crystallographic studies.
Crystallization and structure determination
Crystallization screening was carried out on a Phoenix robot (Art Robbins Instruments, Sunnyvale, CA) using a sparse matrix screening method . BorA DH M3 concentrated to 10 mg/mL was crystallized by sitting drop vapor diffusion in drops containing a 1:1 ratio of protein solution and 0.20 M MgCl2, 0.10 M Tris, pH 8.5, and 29% (w/v) PEG 4000. For FluA DH M1, 10 mg/mL protein was crystallized by sitting drop vapor diffusion in drops containing a 1.5:1 ratio of protein solution and 0.40 M NaCl, 0.10 M Tris, pH 8.5, and 29% (w/v) PEG 3350. For both proteins, a final concentration of 15% glycerol was added before flash freezing in liquid nitrogen.
The X-ray data sets for BorA DH M3 and FluA DH M1 were collected at the Berkeley Center for Structural Biology on beamlines 8.2.1 and 8.2.2 of the Advanced Light Source at Lawrence Berkeley National Laboratory. Diffraction data were recorded using ADSC Q315R detectors (Area Detector Systems Corporation, San Diego, CA). Processing of image data was performed using the HKL2000 suite of programs . For both proteins, phases were calculated by molecular replacement with the program Phaser , using the structure of Rif DH M10 (PDB id: 4LN9)  as a search model. Automated model building was conducted using AutoBuild [39, 40, 41] from the Phenix  suite of programs resulting in a model that was mostly complete. Manual building using Coot  was alternated with reciprocal space refinement using Phenix . Waters were automatically placed using Phenix and manually added or deleted with Coot according to peak height (3.0σ in the Fo-Fc map) and distance from a potential hydrogen bonding partner (< 3.5 Å). TLS refinement  using ten groups, chosen using the TLSMD web server , was used in later rounds of refinement. All data collection, phasing, and refinement statistics are summarized in Table S2.
In silico substrate docking
We conducted in silico substrate docking between 4 and the FluA DH M1. The natural PPant-tethered substrate 4 was initially drawn in Chemdraw (PerkinElmer) and transferred in SMILES format to the PHENIX software suite, eLBOW ligand and constraints generator . The eLBOW program generated a PDB file of the PPant-tethered substrate 4 and a constraints (.cif) file. Both files were used to dock 4 into the FluA DH M1. As a reference for substrate binding, we used the co-crystal structure of PpsC DH complexed with trans-dodec-2-enoyl-CoA (PDB ID: 5NJI) to position substrate 4 into the FluA DH M1. We utilized the program Coot to overlay the PpsC DH with chain A of the FluA DH M1 and model in substrate 4 . The entire FluA DH M1-substrate 4 complex was then energy minimized using the program UCSF CHIMERA . The lack of 11 unmodeled residues in the BorA DH M3 acyl intermediate-binding region made it difficult to conduct precise in silico docking on this target and obtain conclusive information. Therefore, we opted not to conduct substrate docking analysis with the BorA DH M3 model.
Protein structure and visualization
All of the protein structure analysis and figures were generated using UCSF CHIMERA  and PyMOL (The PyMOL Molecular Graphics System, PyMOL1.2edu1 2009, Schrödinger, LLC). Both PyMOL and the online I-TASSER server were used for PDB structural alignment analysis . Multi-sequence alignment for Fig. S6 was generated using MUSCLE  and visualized using ESPript 3.0 .
Analysis of DH sequence conservation
The Shannon entropy of aligned DH sequences was calculated at each position within a refined multiple sequence alignment (MSA). Briefly, 330 sequences derived from PKS non-loading modules containing AT, DH, KR, and ACP domains with active KR and DH domains were exported from the ClusterCAD database . Sequences contained the first residue after the AT domain to the last residue immediately preceding the downstream domain (usually a KR or ER). Sequences were then aligned using MAFFT . The MSA was then refined using the ProDy python library so that all positions within the MSA maintained at least 10% occupancy . ProDy was then used to calculate the Shannon entropy of each position within the resulting MSA. The boundaries of the DH domain within the MSA were further refined using PDB structural information.
We thank Dr. Sangeeta Nath and Dr. Samuel Deutsch at the Joint Genome Institute for providing the template pDVA00936 plasmid, from which the pET28-BorA DH M3 was cloned. We also thank Dr. Nathan A. Schnarr for providing the pTL-A01 plasmid from which the pET28-FluA DH M1 was cloned. Finally, we thank Dr. Leonard Katz and Christopher Eiben for their suggestions and helpful discussions. This work was part of the DOE Joint BioEnergy Institute (https://www.jbei.org) supported by the U. S. Department of Energy, Office of Science, Office of Biological and Environmental Research, and was part of the Agile BioFoundry (https://agilebiofoundry.org) supported by the U. S. Department of Energy, Energy Efficiency and Renewable Energy, Bioenergy Technologies Office, through contract DE-AC02-05CH11231 between Lawrence Berkeley National Laboratory and the U. S. Department of Energy. The views and opinions of the authors expressed herein do not necessarily state or reflect those of the United States Government or any agency thereof. Neither the United States Government nor any agency thereof, nor any of their employees, makes any warranty, expressed or implied, or assumes any legal liability or responsibility or the accuracy, completeness, or usefulness of any information, apparatus, product, or process disclosed, or represents that its use would not infringe privately owned rights. The United States Government retains and the publisher, by accepting the article for publication, acknowledges that the United States Government retains a non-exclusive, paid-up, irrevocable, world-wide license to publish or reproduce the published form of this manuscript, or allow others to do so, for United States Government purposes.
JFB, RPM, JHP, PDA, JDK designed the research. JFB, RPM, BP, TdR, VTB, EEKB performed wet lab experiments and MGT, TWHB performed bioinformatic analysis. JFB, RPM, HGM, NJH, PDA, and JDK wrote and edited the manuscript.
Compliance with ethical standards
Conflict of interest
J. D. K has financial interests in Amyris, Lygos, Constructive Biology, Demetrix, Napigen, and Maple Bio.
- 2.Afonine PV, Grosse-Kunstleve RW, Echols N, Headd JJ, Moriarty NW, Mustyakimov M, Terwilliger TC, Urzhumtsev A, Zwart PH, Adams PD (2012) Towards automated crystallographic structure refinement with phenix.refine. Acta Crystallogr D Biol Crystallogr 68:352–367. https://doi.org/10.1107/S0907444912001308 CrossRefPubMedPubMedCentralGoogle Scholar
- 8.Chang Z, Sitachitta N, Rossi JV, Roberts MA, Flatt PM, Jia J, Sherman DH, Gerwick WH (2004) Biosynthetic pathway and gene cluster analysis of curacin A, an antitubulin natural product from the tropical marine cyanobacterium Lyngbya majuscula. J Nat Prod 67:1356–1367. https://doi.org/10.1021/np0499261 CrossRefPubMedGoogle Scholar
- 29.Miyanaga A, Hayakawa Y, Numakura M, Hashimoto J, Teruya K, Hirano T, Shin-Ya K, Kudo F, Eguchi T (2016) Identification of the fluvirucin B2 (Sch 38518) biosynthetic gene cluster from Actinomadura fulva subsp. indica ATCC 53714: substrate specificity of the β-amino acid selective adenylating enzyme FlvN. Biosci Biotechnol Biochem 80:935–941. https://doi.org/10.1080/09168451.2015.1132155 CrossRefPubMedGoogle Scholar
- 38.Stratmann A, Toupet C, Schilling W, Traber R, Oberer L, Schupp T (1999) Intermediates of rifamycin polyketide synthase produced by an Amycolatopsis mediterranei mutant with inactivated rifF gene. Microbiology (Reading, England) 145(Pt 12):3365–3375. https://doi.org/10.1099/00221287-145-12-3365 CrossRefGoogle Scholar
Open AccessThis article is distributed under the terms of the Creative Commons Attribution 4.0 International License (http://creativecommons.org/licenses/by/4.0/), which permits unrestricted use, distribution, and reproduction in any medium, provided you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons license, and indicate if changes were made.