Structural Basis of Amide-Forming Adenylation Enzyme VinM in Vicenistatin Biosynthesis.

Adenylation enzymes activate amino acid substrates to aminoacyl adenylates and generally transfer this moiety onto the thiol group of the phosphopantetheine arm of a carrier protein for the selective incorporation of aminoacyl building blocks in natural product biosynthesis. In contrast to the canonical thioester-forming adenylation enzymes, the amide-forming adenylation enzyme VinM transfers an l-alanyl group onto the amino group of the aminoacyl unit attached to the phosphopantetheine arm of the carrier protein VinL to generate dipeptidyl-VinL in vicenistatin biosynthesis. It is unclear how VinM distinguishes aminoacyl-VinL from VinL for amide bond formation. Herein we describe structural and biochemical analyses of VinM. We determined the crystal structure of VinM in complex with VinL using a designed pantetheine-type cross-linking probe. The VinM-VinL complex structure in combination with site-directed mutagenesis analysis revealed that the interactions with both the phosphopantetheine arm and VinL are critical for the amide-forming activity of VinM.

Cross-Linking of the Nonribosomal Peptide Synthetase Adenylation Domain with a Carrier Protein Using a Pantetheine-Type Probe.

Adenylation domains (A-domains) are responsible for the selective incorporation of carboxylic acid substrates in the biosynthesis of nonribosomal peptides and related natural products. The A-domain transfers an acyl substrate onto its cognate carrier protein (CP). The proper interactions between an A-domain and the cognate CP are important for functional substrate transfer. To stabilize the transient interactions sufficiently for structural analysis of A-domain-CP complex, vinylsulfonamide adenosine inhibitors have been traditionally used as molecular probes. Recently, we have developed an alternative strategy using a synthetic pantetheine-type probe that enables site-specific cross-linking between an A-domain and a CP. In this chapter, we describe the laboratory protocols for this cross-linking reaction.

Structure-Based Analysis of Transient Interactions between Ketosynthase-like Decarboxylase and Acyl Carrier Protein in a Loading Module of Modular Polyketide Synthase.

Ketosynthase-like decarboxylase (KSQ) domains are widely distributed in the loading modules of modular type I polyketide synthases (PKSs) and catalyze the decarboxylation of the (alkyl-)malonyl unit bound to the acyl carrier protein (ACP) in the loading module for the construction of the PKS starter unit. Previously, we performed a structural and functional analysis of the GfsA KSQ domain involved in the biosynthesis of macrolide antibiotic FD-891. We furthermore revealed the recognition mechanism for the malonic acid thioester moiety of the malonyl-GfsA loading module ACP (ACPL) as a substrate. However, the exact recognition mechanism for the GfsA ACPL moiety remains unclear. Here, we present a structural basis for the interactions between the GfsA KSQ domain and GfsA ACPL. We determined the crystal structure of the GfsA KSQ-acyltransferase (AT) didomain in complex with ACPL (ACPL=KSQAT complex) by using a pantetheine crosslinking probe. We identified the key amino acid residues involved in the KSQ domain-ACPL interactions and confirmed the importance of these residues by mutational analysis. The binding mode of ACPL to the GfsA KSQ domain is similar to that of ACP to the ketosynthase domain in modular type I PKSs. Furthermore, comparing the ACPL=KSQAT complex structure with other full-length PKS module structures provides important insights into the overall architectures and conformational dynamics of the type I PKS modules.

Unique Initiation and Termination Mechanisms Involved in the Biosynthesis of a Hybrid Polyketide-Nonribosomal Peptide Lyngbyapeptin B Produced by the Marine Cyanobacterium Moorena bouillonii.

Lyngbyapeptin B is a hybrid polyketide-nonribosomal peptide isolated from particular marine cyanobacteria. In this report, we carried out genome sequence analysis of a producer cyanobacterium Moorena bouillonii to understand the biosynthetic mechanisms that generate the unique structural features of lyngbyapeptin B, including the (E)-3-methoxy-2-butenoyl starter unit and the C-terminal thiazole moiety. We identified a putative lyngbyapeptin B biosynthetic (lynB) gene cluster comprising nine open reading frames that include two polyketide synthases (PKSs: LynB1 and LynB2), four nonribosomal peptide synthetases (NRPSs: LynB3, LynB4, LynB5, and LynB6), a putative nonheme diiron oxygenase (LynB7), a type II thioesterase (LynB8), and a hypothetical protein (LynB9). In vitro enzymatic analysis of LynB2 with methyltransferase (MT) and acyl carrier protein (ACP) domains revealed that the LynB2 MT domain (LynB2-MT) catalyzes O-methylation of the acetoacetyl-LynB2 ACP domain (LynB2-ACP) to yield (E)-3-methoxy-2-butenoyl-LynB2-ACP. In addition, in vitro enzymatic analysis of LynB7 revealed that LynB7 catalyzes the oxidative decarboxylation of (4R)-2-methyl-2-thiazoline-4-carboxylic acid to yield 2-methylthiazole in the presence of Fe2+ and molecular oxygen. This result indicates that LynB7 is responsible for the last post-NRPS modification to give the C-terminal thiazole moiety in lyngbyapeptin B biosynthesis. Overall, we identified and characterized a new marine cyanobacterial hybrid PKS-NRPS biosynthetic gene cluster for lyngbyapeptin B production, revealing two unique enzymatic logics.

Acyltransferase Domain Exchange between Two Independent Type†I Polyketide Synthases in the Same Producer Strain of Macrolide Antibiotics.

Streptomyces graminofaciens A-8890 produces two macrolide antibiotics, FD-891 and virustomycin A, both of which show significant biological activity. In this study, we identified the virustomycin A biosynthetic gene cluster, which encodes type I polyketide synthases (PKSs), ethylmalonyl-CoA biosynthetic enzymes, methoxymalony-acyl carrier protein biosynthetic enzymes, and post-PKS modification enzymes. Next, we demonstrated that the acyltransferase domain can be exchanged between the Vsm PKSs and the PKSs involved in FD-891 biosynthesis (Gfs PKSs), without any supply problems of the unique extender units. We exchanged the malonyltransferase domain in the loading module of Gfs PKS with the ethylmalonyltransferase domain and the methoxymalonyltransferase domain of Vsm PKSs. Consequently, the expected two-carbon-elongated analog 26-ethyl-FD-891 was successfully produced with a titer comparable to FD-891 production by the wild type; however, exchange with the methoxymalonyltransferase domain did not produce any FD-891 analogs. Furthermore, 26-ethyl-FD-891 showed potent cytotoxic activity against HeLa cells, like natural FD-891.

Biochemical and crystallographic assessments of the effect of 4,6-O-disulfated disaccharide moieties in chondroitin sulfate E on chondroitinase ABC I activity.

O-sulfated N-acetyl-d-galactosamine (GalNAc) residues in chondroitin sulfate (CS) play a crucial role in chondroitinase ABC I (cABC-I) activity. CSA containing mainly 4-O-monosulfated GalNAc was a good substrate for the enzyme, but not CSE containing mainly 4,6-O-disulfated GalNAc [GalNAc(4S,6S)]. Each CS isomer exhibits structural heterogeneity; CSE has di-sulfated disaccharide units and mono-sulfated disaccharide units. Disaccharide composition analysis of digested products revealed that mono-sulfated disaccharide units in CSE contributed to the enzyme reactivity. Although enough substrate (CSA) was present in mixtures of CSA and CSE for reaction, the reactivity was reduced depending on the amount of CSE in the mixture. These results suggested that CSE is not only resistant to enzyme digestion but also attenuates enzyme activity. To understand the mechanism of action, crystallography of cABC-I in complex with unsaturated CSE-disaccharide, ΔDi-(4,6)S, was performed. Both 4-O- and 6-O-sulfate groups in ΔDi-(4,6)S interact with Arg500, suggesting that there was a greater interaction between ΔDi-(4,6)S and Arg500 than between mono-sulfated disaccharides and Arg500. Besides, this interaction attenuated enzyme activity by interfering with a function of Arg500, which is the charge neutralization of the carboxy group of D-glucuronic acid (GlcA) residues in CS. When interacting with the CSE-disaccharide unit [GlcAß1-3GalNAc(4S,6S)] in CS, cABC-I cannot interact with other CS-disaccharide units until it has digested the CSE-disaccharide unit. The low reactivity of cABC-I with CSE is attributable to two suggested factors: (a) resistance of E-units in CSE molecules to digestion by cABC-I, and (b) tendency of E-units in CSE molecules to attenuate cABC-I activity.

Structural Basis of Transient Interactions of Acyltransferase VinK with the Loading Acyl Carrier Protein of the Vicenistatin Modular Polyketide Synthase.

Acyltransferase (AT) recognizes its cognate acyl carrier protein (ACP) for functional transfer of an acyl unit in polyketide biosynthesis. However, structural characterization of AT-ACP complexes is limited because of the weak and transient interactions between them. In the biosynthesis of macrolactam polyketide vicenistatin, the trans-acting loading AT VinK transfers a dipeptidyl unit from the stand-alone ACP VinL to the ACP domain (VinP1ACPL) of the loading module of modular polyketide synthase VinP1. Although the previously determined structure of the VinK-VinL complex clearly illustrates the VinL recognition mechanism of VinK, how VinK recognizes VinP1ACPL remains unclear. Here, the crystal structure of a covalent VinK-VinP1ACPL complex formed with a pantetheine-type cross-linking probe is reported at 3.0 Å resolution. The structure of the VinK-VinP1ACPL complex provides detailed insights into the transient interactions between VinK and VinP1ACPL. The importance of residues in the binding interface was confirmed by site-directed mutational analyses. The binding interface between VinK and VinP1ACPL is similar to that between VinK and VinL, although some of the interface residues are different. However, the ACP orientation and interaction mode observed in the VinK-VinP1ACPL complex are different from those observed in other AT-ACP complexes such as the disorazole trans-AT-ACP complex and cis-AT-ACP complexes of modular polyketide synthases. Thus, AT-ACP binding interface interactions are different in each type of AT-ACP pair.

Recent advances in the structural biology of modular polyketide synthases and nonribosomal peptide synthetases.

Polyketides and nonribosomal peptides are an important class of natural products with useful bioactivities. These compounds are similarly biosynthesized using enzymes with modular structures despite having different physicochemical properties. These enzymes are attractive targets for bioengineering to produce "unnatural" natural products owing to their modular structures. Therefore, their structures have been studied for a long time; however, the main focus was on truncated-single domains. Surprisingly, there is an increasing number of the structures of whole modules reported, most of which have been enabled through the recent advances in cryogenic electron microscopy technology. In this review, we have summarized the recent advances in the structural elucidation of whole modules.

Recent advances in the structural analysis of adenylation domains in natural product biosynthesis.

Adenylation (A) domains catalyze the biosynthetic incorporation of acyl building blocks into nonribosomal peptides and related natural products by selectively transferring acyl substrates onto cognate carrier proteins (CP). The use of noncanonical acyl units, such as nonproteinogenic amino acids and keto acids, by A domains expands the structural diversity of natural products. Furthermore, interrupted A domains, which have embedded auxiliary domains, are able to modify the incorporated acyl units. Structural information on A domains is important for rational protein engineering to generate unnatural compounds. In this review, we summarize recent advances in the structural analysis of A domains. First, we discuss the mechanisms by which A domains recognize noncanonical acyl units. We then focus on the interactions of A domains with CP domains and embedded auxiliary domains.

Structural Basis of Stereospecific Vanadium-Dependent Haloperoxidase Family Enzymes in Napyradiomycin Biosynthesis.

Vanadium-dependent haloperoxidases (VHPOs) from Streptomyces bacteria differ from their counterparts in fungi, macroalgae, and other bacteria by catalyzing organohalogenating reactions with strict regiochemical and stereochemical control. While this group of enzymes collectively uses hydrogen peroxide to oxidize halides for incorporation into electron-rich organic molecules, the mechanism for the controlled transfer of highly reactive chloronium ions in the biosynthesis of napyradiomycin and merochlorin antibiotics sets the Streptomyces vanadium-dependent chloroperoxidases apart. Here we report high-resolution crystal structures of two homologous VHPO family members associated with napyradiomycin biosynthesis, NapH1 and NapH3, that catalyze distinctive chemical reactions in the construction of meroterpenoid natural products. The structures, combined with site-directed mutagenesis and intact protein mass spectrometry studies, afforded a mechanistic model for the asymmetric alkene and arene chlorination reactions catalyzed by NapH1 and the isomerase activity catalyzed by NapH3. A key lysine residue in NapH1 situated between the coordinated vanadate and the putative substrate binding pocket was shown to be essential for catalysis. This observation suggested the involvement of the ε-NH2, possibly through formation of a transient chloramine, as the chlorinating species much as proposed in structurally distinct flavin-dependent halogenases. Unexpectedly, NapH3 is modified post-translationally by phosphorylation of an active site His (τ-pHis) consistent with its repurposed halogenation-independent, α-hydroxyketone isomerase activity. These structural studies deepen our understanding of the mechanistic underpinnings of VHPO enzymes and their evolution as enantioselective biocatalysts.

Protein-Protein Recognition Involved in the Intermodular Transacylation Reaction in Modular Polyketide Synthase in the Biosynthesis of Vicenistatin.

The ketosynthase (KS) domain is a core domain found in modular polyketide synthases (PKSs). To maintain the polyketide biosynthetic fidelity, the KS domain must only accept an acyl group from the acyl carrier protein (ACP) domain of the immediate upstream module even when they are separated into different polypeptides. Although it was reported that both the docking domain-based interactions and KS-ACP compatibility are important for the interpolypeptide transacylation reaction in 6-deoxyerythronolide B synthase, it is not clear whether these findings are broadly applied to other modular PKSs. Herein, we describe the importance of protein-protein recognition in the intermodular transacylation between VinP1 module 3 and VinP2 module 4 in vicenistatin biosynthesis. We compared the transacylation activity and crosslinking efficiency of VinP2 KS4 against the cognate VinP1 ACP3 with the noncognate one. As a result, it appeared that VinP2 KS4 distinguishes the cognate ACP3 from other ACPs.

Structural Insight into the Reaction Mechanism of Ketosynthase-Like Decarboxylase in a Loading Module of Modular Polyketide Synthases.

Ketosynthase-like decarboxylase (KSQ) domains are widely distributed in the loading modules of modular polyketide synthases (PKSs) and are proposed to catalyze the decarboxylation of a malonyl or methylmalonyl unit for the construction of the PKS starter unit. KSQ domains have high sequence similarity to ketosynthase (KS) domains, which catalyze transacylation and decarboxylative condensation in polyketide and fatty acid biosynthesis, except that the catalytic Cys residue of KS domains is replaced by Gln in KSQ domains. Here, we present biochemical analyses of GfsA KSQ and CmiP4 KSQ, which are involved in the biosynthesis of FD-891 and cremimycin, respectively. In vitro analysis showed that these KSQ domains catalyze the decarboxylation of malonyl and methylmalonyl units. Furthermore, we determined the crystal structure of GfsA KSQ in complex with a malonyl thioester substrate analogue, which enabled identification of key amino acid residues involved in the decarboxylation reaction. The importance of these residues was confirmed by mutational analysis. On the basis of these findings, we propose a mechanism of the decarboxylation reaction catalyzed by GfsA KSQ. GfsA KSQ initiates decarboxylation by fixing the substrate in a suitable conformation for decarboxylation. The formation of enolate upon decarboxylation is assisted by two conserved threonine residues. Comparison of the structure of GfsA KSQ with those of KS domains suggests that the Gln residue in the active site of the KSQ domain mimics the acylated Cys residue in the active site of KS domains.

Complex structure of the acyltransferase VinK and the carrier protein VinL with a pantetheine cross-linking probe.

Acyltransferases are responsible for the selection and loading of acyl units onto carrier proteins in polyketide and fatty-acid biosynthesis. Despite the importance of protein-protein interactions between the acyltransferase and the carrier protein, structural information on acyltransferase-carrier protein interactions is limited because of the transient interactions between them. In the biosynthesis of the polyketide vicenistatin, the acyltransferase VinK recognizes the carrier protein VinL for the transfer of a dipeptidyl unit. The crystal structure of a VinK-VinL covalent complex formed with a 1,2-bismaleimidoethane cross-linking reagent has been determined previously. Here, the crystal structure of a VinK-VinL covalent complex formed with a pantetheine cross-linking probe is reported at 1.95 Šresolution. In the structure of the VinK-VinL-probe complex, the pantetheine probe that is attached to VinL is covalently connected to the side chain of the mutated Cys106 of VinK. The interaction interface between VinK and VinL is essentially the same in the two VinK-VinL complex structures, although the position of the pantetheine linker slightly differs. This structural observation suggests that interface interactions are not affected by the cross-linking strategy used.

Substrate specificity of Chondroitinase ABC I based on analyses of biochemical reactions and crystal structures in complex with disaccharides.

Chondroitinase ABC I (cABC-I) is the enzyme which cleaves the ß-1,4 glycosidic linkage of chondroitin sulfate (CS) by ß-elimination. To elucidate more accurately the substrate specificity of cABC-I, we evaluated the kinetic parameters of cABC-I and its reactivity with CS isomers displaying less structural heterogeneity as substrates, e.g., approximately 90 percent of disaccharide units in Chondroitin sulfate A (CSA) or Chondroitin sulfate C (CSC) is D-glucuronic acid and 4-O-sulfated N-acetyl galactosamine (GalNAc) (A-unit) or D-glucuronic acid and 6-O-sulfated GalNAc (C-unit), respectively. cABC-I showed the highest reactivity to CSA and CSC among all CS isomers, and the kcat/Km of cABC-I was higher for CSA than for CSC. Next, we determined the crystal structures of cABC-I in complex with CS disaccharides, and analyzed the crystallographic data in combination with molecular docking data. Arg500 interacts with 4-O-sulfated and 6-O-sulfated GalNAc residues. The distance between Arg500 and the 4-O-sulfate group was 0.8 Å shorter than that between Arg500 and the 6-O-sulfated group. Moreover, it is likely that the 6-O-sulfated group is electrostatically repulsed by the nearby Asp490. Thus, we demonstrated that cABC-I has the highest affinity for the CSA richest in 4-O-sulfated GalNAc residues among all CS isomers. Recently, cABC-I was used to treat lumbar disc herniation. The results provide useful information to understand the mechanism of the pharmacological action of cABC-I.

Mutational Biosynthesis of Hitachimycin Analogs Controlled by the ß-Amino Acid-Selective Adenylation Enzyme HitB.

Hitachimycin is a macrolactam antibiotic with an (S)-ß-phenylalanine (ß-Phe) at the starter position of its polyketide skeleton. (S)-ß-Phe is formed from l-α-phenylalanine by the phenylananine-2,3-aminomutase HitA in the hitachimycin biosynthetic pathway. In this study, we produced new hitachimycin analogs via mutasynthesis by feeding various (S)-ß-Phe analogs to a ΔhitA strain. We obtained six hitachimycin analogs with F at the ortho, meta, or para position and Cl, Br, or a CH3 group at the meta position of the phenyl moiety, as well as two hitachimycin analogs with thienyl substitutions. Furthermore, we carried out a biochemical and structural analysis of HitB, a ß-amino acid-selective adenylation enzyme that introduces (S)-ß-Phe into the hitachimycin biosynthetic pathway. The KM values of the incorporated (S)-ß-Phe analogs and natural (S)-ß-Phe were similar. However, the KM values of unincorporated (S)-ß-Phe analogs with Br and a CH3 group at the ortho or para position of the phenyl moiety were high, indicating that HitB functions as a gatekeeper to select macrolactam starter units during mutasynthesis. The crystal structure of HitB in complex with (S)-ß-3-Br-phenylalanine sulfamoyladenosine (ß-m-Br-Phe-SA) revealed that the bulky meta-Br group is accommodated by the conformational flexibility around Phe328, whose side chain is close to the meta position. The aromatic group of ß-m-Br-Phe-SA is surrounded by hydrophobic and aromatic residues, which appears to confer the conformational flexibility that enables HitB to accommodate the meta-substituted (S)-ß-Phe. The new hitachimycin analogs exhibited different levels of biological activity in HeLa cells and multidrug-sensitive budding yeast, suggesting that they may target different molecules.

One-pot enzymatic synthesis of 2-deoxy-scyllo-inosose from d-glucose and polyphosphate.

2-Deoxy-scyllo-inosose (2DOI, [2S,3R,4S,5R]-2,3,4,5-tetrahydroxycyclohexan-1-one) is a biosynthetic intermediate of 2-deoxystreptamine-containing aminoglycoside antibiotics, including butirosin, kanamycin, and neomycin. In producer microorganisms, 2DOI is constructed from d-glucose 6-phosphate (G6P) by 2-deoxy-scyllo-inosose synthase (DOIS) with the oxidized form of nicotinamide adenine dinucleotide (NAD+). 2DOI is also known as a sustainable biomaterial for production of aromatic compounds and a chiral cyclohexane synthon. In this study, a one-pot enzymatic synthesis of 2DOI from d-glucose and polyphosphate was investigated. First, 3 polyphosphate glucokinases (PPGKs) were examined to produce G6P from d-glucose and polyphosphate. A PPGK derived from Corynebacterium glutamicum (cgPPGK) was found to be suitable for G6P production under ordinary enzymatic conditions. Next, 7 DOISs were examined for the one-pot enzymatic reaction. As a result, cgPPGK and BtrC, the latter of which is a DOIS derived from the butirosin producer Bacillus circulans, achieved nearly full conversion of d-glucose to 2DOI in the presence of polyphosphate.

Stepwise Post-glycosylation Modification of Sugar Moieties in Kanamycin Biosynthesis.

Kanamycin A is the major 2-deoxystreptamine (2DOS)-containing aminoglycoside antibiotic produced by Streptomyces kanamyceticus. The 2DOS moiety is linked with 6-amino-6-deoxy-d-glucose (6ADG) at O-4 and 3-amino-3-deoxy-d-glucose at O-6. Because the 6ADG moiety is derived from d-glucosamine (GlcN), deamination at C-2 and introduction of C-6-NH2 are required in the biosynthesis. A dehydrogenase, KanQ, and an aminotransferase, KanB, are presumed to be responsible for the introduction of C-6-NH2 , although the substrates have not been identified. Here, we examined the substrate specificity of KanQ to better understand the biosynthetic pathway. It was found that KanQ oxidized kanamycin C more efficiently than the 3''-deamino derivative. Furthermore, the substrate specificity of an oxygenase, KanJ, that is responsible for deamination at C-2 of the GlcN moiety was examined, and the crystal structure of KanJ was determined. It was found that C-6-NH2 is important for substrate recognition by KanJ. Thus, the modification of the GlcN moiety occurs after pseudo-trisaccharide formation, followed by the introduction of C-6-NH2 by KanQ/KanB and deamination at C-2 by KanJ.

Structural Characterization of Complex of Adenylation Domain and Carrier Protein by Using Pantetheine Cross-Linking Probe.

Adenylation domains (A-domains) are responsible for selective incorporation of carboxylic acid substrates in the biosynthesis of various natural products. Each A-domain must recognize a cognate carrier protein (CP) for functional substrate transfer. The transient interactions between an A-domain and CP have been investigated by using acyl vinylsulfonamide adenosine inhibitors as probes to determine the structures of several A-domain-CP complexes. However, this strategy requires a specific vinylsulfonamide inhibitor that contains an acyl group corresponding to the substrate specificity of a target A-domain in every case. Here, we report an alternative strategy for structural characterization of A-domain-CP complexes. We used a bromoacetamide pantetheine cross-linking probe in combination with a Cys mutation to trap the standalone A-domain-CP complex involved in macrolactam polyketide biosynthesis through a covalent linkage, allowing the determination of the complex structure. This strategy facilitates the structural determination of A-domain-CP complexes.

Generation of incednine derivatives by mutasynthesis.

The macrolactam antibiotic incednine, isolated from Streptomyces sp. ML694-90F3, contains a (S)-3-aminobutyric acid moiety in its polyketide aglycon. In this study, we performed mutasynthesis to generate incednine derivatives. We successfully obtained 28-methylincednine by feeding 3-aminopentanoic acid into culture of a strain in which the glutamate 2,3-aminomutase gene idnL4, whose product is responsible for supplying 3-aminobutyric acid, was disrupted. 28-Methylincednine showed similar suppressive activity of the antiapoptotic function of oncoprotein Bcl-xL to that of incednine. Thus, this study highlights the applicability of the mutasynthesis approach in generation of novel ß-amino acid-containing macrolactam polyketide derivatives.

Biochemical and Structural Analysis of a Dehydrogenase, KanD2, and an Aminotransferase, KanS2, That Are Responsible for the Construction of the Kanosamine Moiety in Kanamycin Biosynthesis.

Kanosamine (3-amino-3-deoxy-d-glucose) is a characteristic sugar unit found in kanamycins, a group of aminoglycoside antibiotics. The kanosamine moiety originates from d-glucose in kanamycin biosynthesis. However, the timing of the replacement of the 3-OH group of the d-glucose-derived biosynthetic intermediate with the amino group is elusive. Comparison of biosynthetic gene clusters for related aminoglycoside antibiotics suggests that the nicotinamide adenine dinucleotide (NAD+)-dependent dehydrogenase KanD2 and the pyridoxal 5'-phosphate (PLP)-dependent aminotransferase KanS2 are responsible for the introduction of the amino group at the C3 position of kanosamine. In this study, we demonstrated that KanD2 and KanS2 convert kanamycin A, B, and C to the corresponding 3″-deamino-3″-hydroxykanamycins (3″-hks) in the presence of PLP, 2-oxoglutarate, and NADH via a reverse reaction in the pathway. Furthermore, we observed that all of the 3″-hks are oxidized by KanD2 with NAD+, but d-glucose, UDP-d-glucose, d-glucose 6-phosphate, and d-glucose 1-phosphate are not. Crystal structure analysis of KanD2 complexed with 3″-hkB and NADH illustrated the selective recognition of pseudotrisaccharides, especially the d-glucose moiety with 2-deoxystreptamine, by a combination of hydrogen bonds and CH-π interactions. Overall, it was clarified that the kanosamine moiety of kanamycins is constructed after the glucosylation of the pseudodisaccharide biosynthetic intermediates in kanamycin biosynthesis.