Discovery at the VICB
Oxygenases on the Pathway to New Antibiotics
By: Carol A. Rouzer, VICB Communications
Published: August 25, 2015
The characterization of novel oxygenases in orthomycin biosynthesis provides new tools for antibiotic drug discovery.
The rapid appearance of pathogenic bacteria displaying resistance to many, or nearly all, currently available antibiotics is a major public health problem that has yet to be adequately addressed. Contributing to this problem is the fact that many antibiotics that are now used in the clinic share the same chemical scaffold. Acquisition of resistance to one antibiotic of a particular scaffold often leads to at least some resistance to other drugs derived from the same scaffold. Thus, the search for structurally novel antibiotics directed against equally novel targets is of great importance. This led Vanderbilt Institute of Chemical Biology members Tina Iverson and Brian Bachmann to investigate enzymes on the biosynthetic pathway to the orthosomycins [K. M. McCulloch, et al., (2015) Proc. Natl. Acad. Sci. U.S.A., published online August 3, DOI: 10.1073/pnas.1500964112].
The orthosomycins are a class of natural products notable for broad spectrum antibiotic activity. For example, some everninomicins (Figure 1A) are effective against highly antibiotic resistant bacteria such as methicillin resistant Staphylococcus aureus and vancomycin resistant enterococci, through their action at a unique ribosomal binding site. A promising congener of everninomicin A reached phase III clinical trials but eventually failed due to pharmacological issues. This suggests that orthosomycins may become an important new class of antibiotic drugs if appropriate analogs can be found or synthesized.
Figure 1. (A) Structures of avilamycin A, everninomicin D, and hygromycin B. Orthoester linkages are shown in red and methylenedioxy bridges are in blue. (B) Phylogenetic analysis of the oxygenase enzymes of orthosomycin gene clusters. A red star indicates enzymes that were characterized structurally in this study. A blue triangle designates enzymes characterized by gene disruption, and a yellow circle indicates those for which binding affinity was measured. Figure reproduced by permission from K. M. McCulloch, et al., (2015) Proc. Natl. Acad. Sci. U.S.A., published online August 3, DOI: 10.1073/pnas.1500964112 . Copyright 2015, K. M. McCulloch, et al.
The orthosomycins are decorated oligosaccharides that are notable for the presence of one or more spirocyclic ortho-δ-lactone (orthoester) linkages joining two monosaccharide units. Many orthosomycins also contain a methyledioxy bridge as part of their structure (shown in red and blue, respectively, in Figure 1A). The orthoester group fixes the geometry of the molecule between the two linked monosaccharides, a structural feature that contributes to the compound’s antibiotic potency. Despite their importance, the biosynthetic enzymes responsible for orthoester linkage and methylenedioxy bridge formation during orthosomycin synthesis have not been identified. This led the Bachmann lab to conduct a detailed evaluation of known orthosomycin gene clusters in a quest to find likely candidates.
Hygromycin B from Streptomyces hygroscopicus was the first orthosomycin described. It contains only one orthoester linkage and no methylenedioxy bridges, whereas everninomicin and avilamycin each contain two orthoester linkages and one methylenedioxy bridge. Gene clusters for the latter two compounds comprise genes for up to 50 enzymes; however, the function of many of these enzymes is not known. The Bachmann investigators were interested to learn that the gene clusters for all orthosomycins encode one or more enzymes of the α-ketoglutarate, iron- (AKG/Fe(II))-dependent oxygenase family. These enzymes catalyze a wide range of oxidation reactions on an equally wide range of substrates. Thus, they seemed likely candidates for orthoester linkage and methylendioxy bridge biosynthesis.
To test this hypothesis, the Bachmann group conducted a more detailed analysis of the five orthosomycin gene clusters that are available in GenBank. These included ava (from Streptomyces mobaraensis) and avi (from Streptomyces virdochromogenes) for avilamycin synthesis, evd (from Micromonospora carboniacia var. aurantiaca) and eve (from Micromonospora carboniacia var. africana) for everninomicin synthesis, and hyg (from Streptomyces hygroscopicus) for hygromycin biosynthesis. All five gene clusters encoded proteins that were recognizable as necessary for oligosaccharide biosynthesis. In addition, hyg encoded one, while ava, avi, evd, and eve each encoded three enzymes of the AKG/Fe(II)-dependent oxygenase family. This finding was notable because the number of oxygenases in each case corresponded to the number of orthoester linkage plus methyledioxy bridge structures in the orthosomycin being produced by the cluster’s encoded enzymes. Phylogenetic analysis of the thirteen AKG/Fe(II)-dependent oxygenases indicated that they form a distinct subfamily containing three subgroups (Figure 1B). Three of the subgroups contain one enzyme from each of the avilamycin and everninomicin gene clusters, while the fourth subgroup contains the one enzyme from the hygromycin gene cluster. The sequence identities between enzymes of different subgroups was 12-40%, suggesting that they act on structurally distinct substrates, whereas the identity between enzymes within a subgroup was 70-94%, suggesting that they act on structurally similar substrates.
The Iverson group obtained X-ray crystal structure data on a representative enzyme from each subgroup, AviO1 (a monomer), EvdO2 (a monomer), EvdO1 (a dimer), and HygX (a tetramer). All four enzymes exhibited the double-stranded β-helix motif, with an active site metallocenter between β-sheets, that is characteristic of an AKG/Fe(II)-dependent oxygenase (Figure 2). All four enzymes also contained loop insertions between the β-strands. These loops, which form a large binding cleft, are thought to control substrate specificity. A high crystallographic temperature factor suggested that the loops are flexible, enabling them to change conformation upon substrate binding.
Figure 2. Ribbon diagram of the HygX structure. Major strands are in yellow, and minor strands are in red. Ni2+, which was substituted for iron to enhance protein stability, is indicated by a green sphere. Figure reproduced by permission from K. M. McCulloch, et al., (2015) Proc. Natl. Acad. Sci. U.S.A., published online August 3, DOI: 10.1073/pnas.1500964112 . Copyright 2015, K. M. McCulloch, et al.
Most AKG/Fe(II)-dependent oxygenases contain an HXD/E…H motif for coordination of the catalytic iron. This motif was present in three out of four of the crystalized enzymes. In HygX, however, the acidic residue (D/E) in this motif was replaced by glycine, and a glutamic acid located four residues proximal to the distal histidine completed the coordination sphere around the metal ion via a long hydrogen bond or one or more water molecules. This displacement of the acidic residue has not been seen in any other AKG/Fe(II)-dependent oxygenases for which structural data are available.
To confirm that the enzymes they were exploring are, indeed, AKG/Fe(II)-dependent oxygenases, the Iverson group obtained structural data on co-crystals of EvdO2 and HygX with AKG. In both cases, AKG was bound near the metal center of the enzyme, directly coordinating to the metal by displacing two water molecules in the coordination sphere. Small movements of the interacting protein side chains and a large movement of a conserved arginine residue accompanied AKG binding (Figure 3).
Figure 3. Structure of HygX showing the active site metal coordination with AKG and key enzyme residues. Figure reproduced by permission from K. M. McCulloch, et al., (2015) Proc. Natl. Acad. Sci. U.S.A., published online August 3, DOI: 10.1073/pnas.1500964112 . Copyright 2015, K. M. McCulloch, et al.
Based on the other enzymes present in the hyg gene cluster, the investigators hypothesized that HygX acts on a substrate containing two glycosidic linkages. Ideally, a search for the substrate would involve synthesis of all the possible intermediate structures that might contain two such linkages, but this proved to be impractical. However, the discovery that HygX retains micromolar affinity for the product, hygromycin B, led the Iverson group to obtain crystal structure data on hygromycin B complexed to the enzyme. An important finding in this structure was that an oxygen atom of hygromycin B’s orthoester linkage is positioned close to the metal center, and this interaction is stabilized by ten direct and five water-mediated interactions (Figure 4). These findings strongly support the hypothesis that the function of HygX is to catalyze the formation of the orthoester linkage of hygromycin B. Also of interest was the observation that this binding interaction could not take place without the modification of HygX’s metal binding motif that the investigators had previously discovered.
Figure 4. Structure of HygX showing a potential binding pose for hygromycin B. The yellow arrow indicates that one possible site of hydrogen atom abstraction is 5.2 Å from the metal center of the enzyme. Figure reproduced by permission from K. M. McCulloch, et al., (2015) Proc. Natl. Acad. Sci. U.S.A., published online August 3, DOI: 10.1073/pnas.1500964112 . Copyright 2015, K. M. McCulloch, et al.
To explore the binding interactions of possible substrates to the other three enzymes, the investigators used in silico approaches based on their crystal structure data. To simplify their calculations, they constructed models of disaccharides that form a portion of hygromycin B, everninomicin, and avilamycin containing an orthoester linkage. They first showed that the hygromycin B disaccharide model docked into the active site of the HygX structure in a pose that strongly resembled that of hygromycin B in its co-crystal structure with the enzyme. Similar efforts to dock the avilomycin and everninomicin disaccharide models into the structures of their respective enzymes yielded multiple structures, the most favored of which placed an oxygen atom of the orthoester linkage in close proximity to the metal center. These results strongly support the hypothesis that each of the four crystalized enzymes is capable of orthoester linkage biosynthesis.
Finally, the investigators attempted to confirm the role of the AKG/Fe(II)-dependent oxygenases EvdO1 and EvdMO1, proteins encoded by the evd gene cluster, in orthoester linkage biosynthesis. The Bachmann lab first characterized the everninomicins produced by M. carbonacea var. aurantiaca, demonstrating that the wild-type organisms produced everninomicins D, E, F, and G. They then created strains of the bacteria lacking each of the genes of interest. They were not surprised to discover that neither of the mutant strains produced any everninomicins. However, they also failed to find evidence of the intermediates that would be expected to build up as a result of removing a single enzyme from the biosynthetic pathway. These results confirmed the importance of both oxygenases to everninomicin biosynthesis but failed to pinpoint their exact role in the pathway.
Together the data strongly support the conclusion that AKG/Fe(II)-dependent oxygenases are responsible for orthoester linkage and methylenedioxy bridge formation in orthosomycin biosynthesis. These findings add to our knowledge of the biosynthetic pathways of these novel antibiotics and provide an important biochemical tool that may be used in the future to create new orthosomycin congeners. Such congeners are a key step forward in the search for the next drug in the armamentarium to fight resistant bacteria.