Crystal structure of the erythromycin polyketide synthase dehydratase

The dehydratases (DHs) of modular polyketide synthases (PKSs) catalyze dehydrations that occur frequently in the biosynthesis of complex polyketides, yet little is known about them structurally or mechanistically. Here, the structure of a DH domain, isolated from the fourth module of the erythromycin PKS, is presented at 1.85 Å resolution. As with the DH of the highly related animalian fatty acid synthase, the DH monomer possesses a double-hotdog fold. Two symmetry mates within the crystal lattice make a contact that likely represents the DH dimerization interface within an intact PKS. Conserved hydrophobic residues on the DH surface indicate potential interfaces with two other PKS domains, the ketoreductase and the acyl carrier protein. Mutation of an invariant arginine at the hypothesized acyl carrier protein docking site in the context of the erythromycin PKS resulted in decreased production of the erythromycin precursor 6-deoxyerythronolide B. The structure elucidates how the α-hydrogen and β-hydroxyl group of a polyketide substrate interact with the catalytic histidine and aspartic acid in the DH active site prior to dehydration.     

Architectures of Whole-Module and Bimodular Proteins from the 6-Deoxyerythronolide B Synthase

The 6-deoxyerythronolide B synthase (DEBS) is a prototypical assembly line polyketide synthase produced by the actinomycete Saccharopolyspora erythraea that synthesizes the macrocyclic core of the antibiotic erythromycin 6-deoxyerythronolide B. The megasynthase is a 2-MDa trimeric complex composed of three unique homodimers assembled from the gene products DEBS1, DEBS2, and DEBS3, which are housed within the erythromycin biosynthetic gene cluster. Each homodimer contains two clusters of catalytically independent enzymatic domains, each referred to as a module, which catalyzes one round of polyketide chain extension and modification. Modules are named sequentially to indicate the order in which they are utilized during synthesis of 6-deoxyerythronolide B. We report small-angle X-ray scattering (SAXS) analyses of a whole module and a bimodule from DEBS, as well as a set of domains for which high-resolution structures are available. In all cases, the solution state was probed under previously established conditions ensuring that each protein is catalytically active. SAXS data are consistent with atomic-resolution structures of DEBS fragments. Therefore, we used the available high-resolution structures of DEBS domains to model the architectures of the larger protein assemblies using rigid-body refinement. Our data support a model in which the third module of DEBS forms a disc-shaped structure capable of caging the acyl carrier protein domain proximal to each active site. The molecular envelope of DEBS3 is a thin elongated ellipsoid, and the results of rigid-body modeling suggest that modules 5 and 6 stack collinearly along the 2-fold axis of symmetry.     

Comprehensive analysis of distinctive polyketide and nonribosomal peptide structural motifs encoded in microbial genomes

We developed a highly accurate method to predict polyketide (PK) and nonribosomal peptide (NRP) structures encoded in microbial genomes. PKs/NRPs are polymers of carbonyl/peptidyl chains synthesized by polyketide synthases (PKS) and nonribosomal peptide synthetases (NRPS). We analyzed domain sequences corresponding to specific substrates and physical interactions between PKSs/NRPSs in order to predict which substrates (carbonyl/peptidyl units) are selected and assembled into highly ordered chemical structures. The predicted PKs/NRPs were represented as the sequences of carbonyl/peptidyl units to extract the structural motifs efficiently. We applied our method to 4529 PKSs/NRPSs and found 619 PKs/NRPs. We also collected 1449 PKs/NRPs whose chemical structures have been determined experimentally. The structural sequences were compared using the Smith-Waterman algorithm, and clustered into 271 clusters. From the compound clusters, we extracted 33 structural motifs that are significantly related with their bioactivities. We used the structural motifs to infer functions of 13 novel PKs/NRPs clusters produced by Pseudomonas spp. and Burkholderia spp. and found a putative virulence factor. The integrative analysis of genomic and chemical information given here will provide a strategy to predict the chemical structures, the biosynthetic pathways, and the biological activities of PKs/NRPs, which is useful for the rational design of novel PKs/NRPs.     

From genetic diversity to metabolic unity: studies on the biosynthesis of aurafurones and aurafuron-like structures in myxobacteria and streptomycetes

The myxobacterial polyketide secondary metabolites aurafuron A and B were identified by genome mining in the myxobacterial strain Stigmatella aurantiaca DW4/3-1. The compounds contain an unusual furanone moiety and resemble metabolites isolated from soil-dwelling and marine actinobacteria, a fungus and mollusks. We describe here the cloning and functional analysis of the aurafuron biosynthetic gene cluster, including site-directed mutagenesis and feeding studies using labeled precursors. The polyketide core of the aurafurones is assembled by a modular polyketide synthase (PKS). As with many such systems described from myxobacteria, the aurafuron PKS exhibits a number of unusual features, including the apparent iterative use of a module, redundant modules and domains, a trans acting dehydratase and the absence of a terminal thioesterase domain. Four oxidoreductases are encoded within the gene locus, some of which likely participate in formation of the furanone moiety via a Baeyer-Villiger type oxidation. Indeed, inactivation of a gene encoding a cytochrome P₄₅₀ monooxygenase completely abolished production of both compounds. We also compare the complete gene locus to biosynthetic gene clusters from two Streptomyces sp., which produce close structural analogues of the aurafurones. A portion of the post-PKS biosynthetic machinery is strikingly similar in all three cases, in contrast to the PKS genes, which are highly divergent. Phylogenetic analysis of the ketosynthase domains further indicates that the PKSs have developed independently (polyphyletically) during evolution. These findings point to a currently unknown but important biological function of aurafuron-like compounds for the producing organisms.     

Identification and characterization of the biosynthetic gene cluster of divergolides from Streptomyces sp. W112

Divergolides are a group of structurally unprecedented ansamacrolactam antibiotics with antibacterial and antitumor activities. A biosynthetic gene cluster predicted to encode the biosynthesis of divergolides was cloned and sequenced from endophytic Streptomyces sp. W112. The gene cluster of divergolides (div) spans a DNA region of 61-kb and consists of 20 open reading frames (ORFs) that encode polyketide synthases (PKSs), enzymes for the synthesis of AHBA and PKS extender units, and post-PKS modifications, proposed regulators, and putative transporters. Disruption of the AHBA synthase gene (divK) completely abolished the production of divergolides proved its involvement in the biosynthesis of divergolides. Bioinformatics analysis suggested that the regulatory gene div8 in div gene cluster might encode a positive regulator for the biosynthesis of divergolides. Constitutive overexpression of div8 improved the production of divergolides E, implying that div gene cluster maybe responsible for the biosynthesis of divergolides. These findings set the stage for fully investigating the biosynthesis of divergolides and rational engineering of new divergolide analogs by genetic modifications, and pave the way to further improve the production of divergolides.     

Alkylresorcylic acid synthesis by type III polyketide synthases from rice Oryza sativa

Alkylresorcinols, produced by various plants, bacteria, and fungi, are bioactive compounds possessing beneficial activities for human health, such as anti-cancer activity. In rice, they accumulate in seedlings, contributing to protection against fungi. Alkylresorcylic acids, which are carboxylated forms of alkylresorcinols, are unstable compounds and decarboxylate readily to yield alkylresorcinols. Genome mining of the rice Oryza sativa identified two type III polyketide synthases, named ARAS1 (alkylresorcylic acid synthase) and ARAS2, that catalyze the formation of alkylresorcylic acids. Both enzymes condensed fatty acyl-CoAs with three C₂ units from malonyl-CoA and cyclized the resulting tetraketide intermediates via intramolecular C-2 to C-7 aldol condensation. The alkylresorcylic acids thus produced were released from the enzyme and decarboxylated non-enzymatically to yield alkylresorcinols. This is the first report on a plant type III polyketide synthase that produces tetraketide alkylresorcylic acids as major products.     

Identification and characterization of multiple curcumin synthases from the herb Curcuma longa

Curcuminoids are pharmaceutically important compounds isolated from the herb Curcuma longa. Two additional type III polyketide synthases, named CURS2 and CURS3, that are capable of curcuminoid synthesis were identified and characterized. In vitro analysis revealed that CURS2 preferred feruloyl-CoA as a starter substrate and CURS3 preferred both feruloyl-CoA and p-coumaroyl-CoA. These results suggested that CURS2 synthesizes curcumin or demethoxycurcumin and CURS3 synthesizes curcumin, bisdemethoxycurcumin and demethoxycurcumin. The availability of the substrates and the expression levels of the three different enzymes capable of curcuminoid synthesis with different substrate specificities might influence the composition of curcuminoids in the turmeric and in different cultivars.     

Structure of the Human Fatty Acid Synthase KS-MAT Didomain as a Framework for Inhibitor Design

The human fatty acid synthase (FAS) is a key enzyme in the metabolism of fatty acids and a target for antineoplastic and antiobesity drug development. Due to its size and flexibility, structural studies of mammalian FAS have been limited to individual domains or intermediate-resolution studies of the complete porcine FAS. We describe the high-resolution crystal structure of a large part of human FAS that encompasses the tandem domain of β-ketoacyl synthase (KS) connected by a linker domain to the malonyltransferase (MAT) domain. Hinge regions that allow for substantial flexibility of the subdomains are defined. The KS domain forms the canonical dimer, and its substrate-binding site geometry differs markedly from that of bacterial homologues but is similar to that of the porcine orthologue. The didomain structure reveals a possible way to generate a small and compact KS domain by omitting a large part of the linker and MAT domains, which could greatly aid in rapid screening of KS inhibitors. In the crystal, the MAT domain exhibits two closed conformations that differ significantly by rigid-body plasticity. This flexibility may be important for catalysis and extends the conformational space previously known for type I FAS and 6-deoxyerythronolide B synthase.     

Structural analysis of trisialylated biantennary glycans isolated from mouse serum transferrin: Characterization of the sequence Neu5Gc(α2-3)Gal(β1-3)[Neu5Gc(α2-6)]GlcNAc(β1-2)Man

Five variants of mouse serum transferrin (mTf, designated mTf-I to mTf-V) with respect to carbohydrate composition have been isolated by DEAE-cellulose chromatography in the following relative percentages: mTf-I: 0.55; mTf-II: 0.79; mTf-III: 71.80; mTf-VI: 21.90 and mTf-V: 4.96. The primary structures of the major glycans from mTf-III and mTf-IV were determined by methylation analysis and ¹H-nuclear magnetic resonance (NMR) spectroscopy. All glycans possessed a common trimannosyl-N,N′-diacetylchitobiose core. From the glycovariant mTf-III two isomers of a conventional biantennary N-acetyllactosamine type were isolated, in which two N-glycolylneuraminic acid (Neu5Gc) residues are linked to galactose either by a (α2-6) or (α2-3) linkage. A subpopulation of this glycovariant contains a fucose residue (α1-6)-linked to GlcNAc-1. The structure of the major glycan found in variant mTf-IV contained an additional Neu5Gc and possessed the following new type of linkage: Neu5Gc(α2-3)Gal(β1-3)[Neu5Gc(α2-6)]GlcNAc(β1-2)Man(α1-3). In addition to this glycan, a minor compound contained the same antennae linked to Man(α1-6). In fraction mTf-V, which was found to be very heterogeneous by ¹H NMR analysis, carbohydrate composition and methylation analysis suggested the presence of tri′-antennary glycans sialylated by Neu5Gc α-2,6- and α-2,3-linked to the terminal galactose residues. In summary, mTf glycans differed from those of other analyzed mammalian transferrins by the presence of Neu5Gc and by a Neu5Gc(α2-6)GlcNAc linkage in trisialylated biantennary structures, reflecting in mouse liver, a high activity of CMP-Neu5Ac hydroxylase and (α2-6)GlcNAc sialyltransferase.     

Structural and functional analysis of two glutamate racemase isozymes from Bacillus anthracis and implications for inhibitor design

Glutamate racemase (RacE) is responsible for converting l-glutamate to d-glutamate, which is an essential component of peptidoglycan biosynthesis, and the primary constituent of the poly-gamma-d-glutamate capsule of the pathogen Bacillus anthracis. RacE enzymes are essential for bacterial growth and lack a human homolog, making them attractive targets for the design and development of antibacterial therapeutics. We have cloned, expressed and purified the two glutamate racemase isozymes, RacE1 and RacE2, from the B. anthracis genome. Through a series of steady-state kinetic studies, and based upon the ability of both RacE1 and RacE2 to catalyze the rapid formation of d-glutamate, we have determined that RacE1 and RacE2 are bona fide isozymes. The X-ray structures of B. anthracis RacE1 and RacE2, in complex with d-glutamate, were determined to resolutions of 1.75 A and 2.0 A. Both enzymes are dimers with monomers arranged in a "tail-to-tail" orientation, similar to the B. subtilis RacE structure, but differing substantially from the Aquifex pyrophilus RacE structure. The differences in quaternary structures produce differences in the active sites of racemases among the various species, which has important implications for structure-based, inhibitor design efforts within this class of enzymes. We found a Val to Ala variance at the entrance of the active site between RacE1 and RacE2, which results in the active site entrance being less sterically hindered for RacE1. Using a series of inhibitors, we show that this variance results in differences in the inhibitory activity against the two isozymes and suggest a strategy for structure-based inhibitor design to obtain broad-spectrum inhibitors for glutamate racemases.     

Molecular and structural analysis of C4-specific PEPC isoform from Pennisetum glaucum plays a role in stress adaptation

As an activator of adenylate cyclase, the neuropeptide Pituitary Adenylate Cyclase Activating Peptide (PACAP) impacts levels of cyclic AMP, a key second messenger available in brain cells. PACAP is involved in certain adult behaviors. To elucidate PACAP interactions, a compendium of microarrays representing mRNA expression in the adult mouse whole brain was pooled from the Phenogen database for analysis. A regulatory network was computed based on mutual information between gene pairs using gene expression data across the compendium. Clusters among genes directly linked to PACAP, and probable interactions between corresponding proteins were computed. Database “experts” affirmed some of the inferred relationships. The findings suggest ADCY7 is probably the adenylate cyclase isoform most relevant to PACAP's action. They also support intervening roles for kinases including GSK3B, PI 3-kinase, SGK3 and AMPK. Other high-confidence interactions are hypothesized for future testing. This new information has implications for certain behavioral and other disorders.     

Cloning and functional expression in E. coli of a polyphenol oxidase transcript from Coreopsis grandiflora involved in aurone formation

Polyphenol oxidases are involved in aurone biosynthesis but the gene responsible for 4-deoxyaurone formation in Asteraceae was so far unknown. Three novel full-length cDNA sequences were isolated from Coreopsis grandiflora with sizes of 1.80 kb (cgAUS1) and 1.85 kb (cgAUS2a, 2b), encoding for proteins of 68–69 kDa, respectively. cgAUS1 is preferably expressed in young petals indicating a specific role in pigment formation. The 58.9 kDa AUS1 holoproenzyme, was recombinantly expressed in E. coli and purified to homogeneity. The enzyme shows only diphenolase activity, catalyzing the conversion of chalcones to aurones and was characterized by SDS–PAGE and shot-gun type nanoUHPLC–ESI-MS/MS.     

Implications of oligomeric forms of POD-1 and POD-2 proteins isolated from cell walls of the biocontrol agent Pythium oligandrum in relation to their ability to induce defense reactions in tomato

The cell wall protein fraction (CWP) isolated from the biocontrol agent Pythium oligandrum induces defense reactions in tomato. CWP contains two novel elicitin-like proteins, POD-1 and POD-2, both with seven cysteines. To determine the essential structure in the defense-eliciting components of CWP, five fractions (F1, F2, F3, F4 and F5) were fractionated from CWP using cation chromatography and their components and disulfide bond compositions were analyzed. The expression levels of three defense-related genes (PR-6, LeCAS and PR-2b) were determined in tomato roots treated with each of the five fractions. Of the five fractions, F4 containing a heterohexamer of POD-1 and POD-2, and F5 containing a homohexamer of POD-1, both with disulfide bonds formed between all cysteine residues, induced the expression of three genes. F4 treatment also induced the accumulation of ethylene in tomato. The predicted three-dimensional structures of POD-1 and POD-2, and the results of SEC and MALDI-TOF MS analyses suggest that F4 consists of three POD-1 and POD-2 disulfide-bonded heterodimers that interleave into a hexameric ring through noncovalent association. These results suggest that this structure, which F5 also appears to form, is essential for stimulating defense responses in tomato.     

Synthetic biology and metabolic engineering of actinomycetes for natural product discovery

Actinomycetes are one of the most valuable sources of natural products with industrial and medicinal importance. After more than half a century of exploitation, it has become increasingly challenging to find novel natural products with useful properties as the same known compounds are often repeatedly re-discovered when using traditional approaches. Modern genome mining approaches have led to the discovery of new biosynthetic gene clusters, thus indicating that actinomycetes still harbor a huge unexploited potential to produce novel natural products. In recent years, innovative synthetic biology and metabolic engineering tools have greatly accelerated the discovery of new natural products and the engineering of actinomycetes. In the first part of this review, we outline the successful application of metabolic engineering to optimize natural product production, focusing on the use of multi-omics data, genome-scale metabolic models, rational approaches to balance precursor pools, and the engineering of regulatory genes and regulatory elements. In the second part, we summarize the recent advances of synthetic biology for actinomycetal metabolic engineering including cluster assembly, cloning and expression, CRISPR/Cas9 technologies, and chassis strain development for natural product overproduction and discovery. Finally, we describe new advances in reprogramming biosynthetic pathways through polyketide synthase and non-ribosomal peptide synthetase engineering. These new developments are expected to revitalize discovery and development of new natural products with medicinal and other industrial applications.     

Metabolic Enzyme Alterations and Astrocyte Dysfunction in a Murine Model of Alexander Disease With Severe Reactive Gliosis

Alexander disease (AxD) is a rare and fatal neurodegenerative disorder caused by mutations in the gene encoding glial fibrillary acidic protein (GFAP). In this report, a mouse model of AxD (GFAP^Tg;Gfap^+/R236H) was analyzed that contains a heterozygous R236H point mutation in murine Gfap as well as a transgene with a GFAP promoter to overexpress human GFAP. Using label-free quantitative proteomic comparisons of brain tissue from GFAP^Tg;Gfap^+/R236H versus wild-type mice confirmed upregulation of the glutathione metabolism pathway and indicated proteins were elevated in the peroxisome proliferator-activated receptor (PPAR) signaling pathway, which had not been reported previously in AxD. Relative protein-level differences were confirmed by a targeted proteomics assay, including proteins related to astrocytes and oligodendrocytes. Of particular interest was the decreased level of the oligodendrocyte protein, 2-hydroxyacylsphingosine 1-beta-galactosyltransferase (Ugt8), since Ugt8-deficient mice exhibit a phenotype similar to GFAP^Tg;Gfap^+/R236H mice (e.g., tremors, ataxia, hind-limb paralysis). In addition, decreased levels of myelin-associated proteins were found in the GFAP^Tg;Gfap^+/R236H mice, consistent with the role of Ugt8 in myelin synthesis. Fabp7 upregulation in GFAP^Tg;Gfap^+/R236H mice was also selected for further investigation due to its uncharacterized association to AxD, critical function in astrocyte proliferation, and functional ability to inhibit the anti-inflammatory PPAR signaling pathway in models of amyotrophic lateral sclerosis (ALS). Within Gfap⁺ astrocytes, Fabp7 was markedly increased in the hippocampus, a brain region subjected to extensive pathology and chronic reactive gliosis in GFAP^Tg;Gfap^+/R236H mice. Last, to determine whether the findings in GFAP^Tg;Gfap^+/R236H mice are present in the human condition, AxD patient and control samples were analyzed by Western blot, which indicated that Type I AxD patients have a significant fourfold upregulation of FABP7. However, immunohistochemistry analysis showed that UGT8 accumulates in AxD patient subpial brain regions where abundant amounts of Rosenthal fibers are located, which was not observed in the GFAP^Tg;Gfap^+/R236H mice.