The bifunctional enzyme, GenB4, catalyzes the last step of gentamicin 3′,4′-di-deoxygenation via reduction and transamination activities

Background New semi-synthetic aminoglycoside antibiotics generally use chemical modifications to avoid inactivity from pathogens. One of the most used modifications is 3′,4′-di-deoxygenation, which imitates the structure of gentamicin. However, the mechanism of di-deoxygenation has not been clearly elucidated. Results Here, we report that the bifunctional enzyme, GenB4, catalyzes the last step of gentamicin 3′,4′-di-deoxygenation via reduction and transamination activities. Following disruption of genB4 in wild-type M. echinospora, its products accumulated in 6′-deamino-6′-oxoverdamicin (1), verdamicin C2a (2), and its epimer, verdamicin C2 (3). Following disruption of genB4 in M. echinospora ΔgenK, its products accumulated in sisomicin (4) and 6′-N-methylsisomicin (5, G-52). Following in vitro catalytic reactions, GenB4 transformed sisomicin (4) to gentamicin C1a (9) and transformed verdamicin C2a (2) and its epimer, verdamicin C2 (3), to gentamicin C2a (11) and gentamicin C2 (12), respectively. Conclusion This finding indicated that in addition to its transamination activity, GenB4 exhibits specific 4′,5′ double-bond reducing activity and is responsible for the last step of gentamicin 3′,4′-di-deoxygenation. Taken together, we propose three new intermediates that may refine and supplement the specific biosynthetic pathway of gentamicin C components and lay the foundation for the complete elucidation of di-deoxygenation mechanisms.


Background
Drug-resistant pathogens have spread dramatically across the world and have become a major threat to public health [1]. Studies have found that several aminoglycosides can resist multiple drug-resistant pathogens, especially for Gram-negative bacteria [2,3]. One example is gentamicin, which contains 3′,4′-di-deoxygenation structures and avoids modification of o-phosphotransferases (APHs) and o-adenyltransferases (ANTs) to 3′,4′-hydroxy groups [4]. Since the hydroxyl group of the aminoglycoside antibiotic is the main attack site of resistant-bacteria inactivation enzymes, these hydroxyldeoxygenated atoms cannot be chemically modified. New semi-synthetic antibiotics used in clinical applications also use this modification. For example, arbekacin and dibekacin contain 3′,4′-di-deoxygenation structures obtained by chemical syntheses. Therefore, elucidation of the mechanisms of gentamicin 3′,4′-di-deoxygenation

Open Access
Microbial Cell Factories *Correspondence: nixianpu126@126.com; hzxia@syphu.edu.cn 1 School of Life Science and Biopharmaceutics, Shenyang Pharmaceutical University, No.103 Wenhua Road, Shenyang, Liaoning, China Full list of author information is available at the end of the article has theoretical and practical value for the development of novel aminoglycoside antibiotics.

Inactivation of genB4 in M. echinospora
To investigate the role of genB4, a 939-bp internal DNA fragment was deleted in-frame with the pD2925B4 plasmid in wild-type M. echinospora and M. echinospora ΔgenK (Additional file 1: Fig. S1a). The mutants were confirmed by PCR (Additional file 1: Fig. S1b and c). The disrupting strains were fermented, and their products were analyzed through high-performance liquid chromatography with evaporative light-scattering detection (HPLC-ELSD). Compared to that in the wild-type strain, M. echinospora ΔgenB4 did not produce any of the gentamicin C complex. Rather, its products accumulated as intermediates such as (1), (2), and (3) as well as minor components in the form of (6) and (7) (Fig. 2). Additionally, (1) and (6) had the same retention time in HPLC-ELSD during separation by cation-exchange chromatography (Additional file 2: Fig. S2). The disrupting strain M. echinospora ΔgenKΔgenB4 accumulated in (4) and (5) (Fig. 2). To determine the structures of these new products, the intermediates were separated, purified, and analyzed through mass and nuclear magnetic resonance (NMR) spectroscopic analyses. The exact mass of intermediate (4) was 448.2770 ([M+H] + ) (Additional file 3: Fig. S3a), and its retention time was consistent with that of sisomicin. The exact mass of intermediate (5) was 462.2944 ([M+H] + ) (Additional file 3: Fig. S3b), which was consistent with that of 6′-N-methylsisomicin (5, G-52) [10][11][12]. Hence, we hypothesized that in the other parallel pathway of gentamicin biosynthesis, the genB4 disrupting strain would accumulate as verdamicin and 6′-N-methylverdamicin (VF3-1) [13][14][15][16]. Interestingly, the exact mass of intermediate (2)  and its congener, C2, from sisomicin in vitro in 2008, at which time this synthesis was first mentioned and its chiral isomerization structure was named [17]. Our present in vitro catalytic findings demonstrate that (2) was verdamicin C2a and (3) was verdamicin C2. Based on the retention times of HPLC-ELSD, we hypothesized that (6) was VF3-1a(S) and (7)

Complementation of the genB4 disrupting strain
To further determine the in vivo role of genB4, we constructed the genB4 complement plasmid, pEAP1B4, which contained the entire genB4 under the control of the PhrdB promoter (Additional file 6: Fig. S6a). After sequence verification, pEAP1B4 was introduced into ΔgenB4 and ΔgenKΔgenB4 to construct the complementation strains, ΔgenB4::genB4 and ΔgenKΔgenB4::genB4, respectively. The exconjugants were confirmed by PCR (Additional file 6: Fig. S6b). As shown in Additional file 7: Fig. S7, the complementation strain ΔgenB4::genB4 restored production of the gentamicin C complex such that they were the same as those in the wild-type strain, and ΔgenKΔgenB4::genB4 restored the productions of gentamicin C1a (9) and C2b such that they were the same as those in the original strain of ΔgenK.
Interestingly, the catalytic reaction produced a new minor component when we increased the amount of the substrate sisomicin (4) and supplemented exogenous PLP (Fig. 3a). As we expected, the new minor component was gentamicin C1a (9) (Additional file 10:  (8) to gentamicin C1a (9). The result showed that the yield of gentamicin C1a (9) increased significantly with L-Glu as an amino donor (Additional file 11: Fig. S11).
We also used 6′-deamino-6′-oxoverdamicin (1) as a substrate, but GenB4 was unable to transform it. Interestingly, when we used GenB1 or GenB2 to catalyze (1), no transamination reaction occurred in this situation either. It was only when either GenB1-GenB4 or GenB2-GenB4 was combined to catalyze (1) that products corresponding to the new compounds (10), (11), and (12) were yielded (Fig. 6). These findings indicate that in the biosynthetic pathway of gentamicin, (1) is upstream of (2) and (3). Additionally, only transamination of (1) to (2) and (3) can be recognized by GenB4. Therefore, we conclude that the 6′ amino group is an important recognition site for GenB4 reduction activity.

Conclusions
In the present study, we disrupted genB4 in wildtype M. echinospora, and its products accumulated as 6′-deamino-6′-oxoverdamicin (1), verdamicin C2a (2) and its epimer, verdamicin C2 (3). We also disrupted genB4 in M. echinospora ΔgenK, and its products accumulated as sisomicin (4). Following in vitro catalytic reactions, GenB4 transformed sisomicin (4) to gentamicin C1a (9), and transformed verdamicin C2a (2) and its epimer, verdamicin C2 (3), to gentamicin C2a (11) and gentamicin C2 (12). Taken together, these findings demonstrate that GenB4 is a bifunctional enzyme with both reduction and transamination activities that are responsible for the last step of gentamicin 3′,4′-di-deoxygenation. Additionally, we provide the first proposal of the structures of intermediates (1), (8), and (10). Collectively, our present study refines and supplements the specific biosynthetic pathways of gentamicin C components, which lays the foundation for fully elucidating the mechanisms of di-deoxygenation. With clarifying genes and the unique biosynthetic pathway of gentamicin di-deoxygenation, deoxygenation enzyme complex will also be utilized for constructing strain producing semi-synthetic dibekacin at the same time. This work will lay the foundation for utilizing dideoxygenation enzyme complex for biosynthesis semisynthetic antibiotics like dibekacin or create new drugs with better bioactivity.

Construction of genB4 disruption plasmid
The genB4 gene was disrupted via pD2925-mediated double-crossover recombination. DNA isolation and manipulation were performed as described by Sambrook et al. [23]. Additional file 14: Table S1 lists the primers used in the present study. Primers were designed using the biosynthetic gene-cluster sequence for gentamicin (GenBank accession number: JQ975418.1). Primers B4up 1 and B4up 2 were used to amplify a 1434-bp fragment containing the upstream sequence of genB4. Primers B4dn 1 and B4dn 2 were used to amplify a 1666-bp fragment containing the downstream sequence and the last 375 bp of genB4. The two fragments were cloned separately into pMD 18-T (Takara, Japan) and were then excised from the resulting plasmids using HindIII-XbaI and XbaI-KpnI. The excised products containing the upstream and downstream fragments of genB4 were then ligated with the HindIII-KpnI fragment of pD2925 to yield pD2925B4 for an in-frame deletion of genB4 (Fig. 2).

Construction of genB4 disruption strain
The disruption plasmid, pD2925B4, was introduced into E. coli ET12567/pUZ8002 via the CaCl 2 method and then into wild-type M. echinospora via conjugational transfer [24]. After incubation at 37 °C for 24 h, each dish was overlaid with 1 mL of sterile water containing apramycin at a final concentration of 20 μg/mL. Since pD2925B4 contains an apramycin-resistance gene, the exconjugants were selected as follows: the apramycinresistant (for the first crossover event) phenotype was first selected, and the apramycin-sensitive (the second crossover event) phenotype was then selected to isolate the genB4-disruption strain. The exconjugants were subsequently incubated at 37 °C for 7 days to select for homologous recombinants (for the first crossover event) and were identified by PCR using the primers B4Y1, B4Y2, B4Y3, and B4Y4. Then, the exconjugants were cultured on antibiotic-free medium for sporulation, and the cycle was repeated three times to enhance the probability of recombination. Single clones were replica-plated onto the apramycin-containing plates, as well as on plates without antibiotic for sporulation. The apramycin-sensitive strains were selected based on the growth conditions of the clones on the two different plates (the second crossover event), and the expected disruption genotype was identified by PCR using the primers, B4Y1 and B4Y4. The selected strain was named M. echinospora ΔgenB4. We used the same method to generate M. echinospora ΔgenKΔgenB4.