Genomic-Led Discovery of a Novel Glycopeptide Antibiotic by Nonomuraea coxensis DSM 45129

Glycopeptide antibiotics (GPAs) are last defense line drugs against multidrug-resistant Gram-positive pathogens. Natural GPAs teicoplanin and vancomycin, as well as semisynthetic oritavancin, telavancin, and dalbavancin, are currently approved for clinical use. Although these antibiotics remain efficient, emergence of novel GPA-resistant pathogens is a question of time. Therefore, it is important to investigate the natural variety of GPAs coming from so-called “rare” actinobacteria. Herein we describe a novel GPA producer—Nonomuraea coxensis DSM 45129. Its de novo sequenced and completely assembled genome harbors a biosynthetic gene cluster (BGC) similar to the dbv BGC of A40926, the natural precursor to dalbavancin. The strain produces a novel GPA, which we propose is an A40926 analogue lacking the carboxyl group on the N-acylglucosamine moiety. This structural difference correlates with the absence of dbv29—coding for an enzyme responsible for the oxidation of the N-acylglucosamine moiety. Introduction of dbv29 into N. coxensis led to A40926 production in this strain. Finally, we successfully applied dbv3 and dbv4 heterologous transcriptional regulators to trigger and improve A50926 production in N. coxensis, making them prospective tools for screening other Nonomuraea spp. for GPA production. Our work highlights genus Nonomuraea as a still untapped source of novel GPAs.


INTRODUCTION
Nonomuraea is a genus of so-called "rare" actinomycetes whose potential to produce specialized (secondary) metabolites is still rather poorly explored. 1,2 Recently sequenced genomes of Nonomuraea species appear to be generally larger than the reference Streptomyces ones. The mean genome size of Nonomuraea (based on the three available complete assemblies 2,3 ) is around 12 Mbp, whereas the mean genome size of Streptomyces (calculated on 251 fully assembled genomes available in GenBank) equals 8.6 Mbp. The larger genomes of Nonomuraea spp. encode dozens of putative biosynthetic gene clusters (BGCs). 2−4 Nonomuraea spp. were initially found to be recalcitrant to commonly used genetic engineering manipulations, but new tools are now being developed for this genus. 5−7 This paves the way for unravelling the huge hidden biosynthetic potential of these organisms.
Probably the most important bioactive metabolite produced by a Nonomuraea species is the type IV 8 glycopeptide antibiotic (GPA) A40926 9 (Figure 1) produced by Nonomuraea gerenzanensis ATCC 39727. Like other GPAs, A40926 acts as a selective and potent inhibitor of cell-wall biosynthesis in Gram-positive bacteria. A40926 is structurally related to the clinically relevant GPA teicoplanin (Figure 1), produced by Actinoplanes teichomyceticus ATCC 31121 10,11 and to ristocetin (Figure 1), previously isolated from numerous Amycolatopsis spp. (i.e., A. lurida NRRL 2430, A. japonicum MG417-CF17, and Amycolatopsis sp. MJM2582). 12−14 Like teicoplanin, A40926 is produced as a mixture of related compounds (major components are A40926 B and A40926 A factors), which differ in the length and branching of an aliphatic side chain ( Figure 1). It was recently clarified that N. gerenzanensis produces the GPA in the form of O-acetyl-A40926 (with an Oacetylated mannose residue), but the acetyl group is lost during the alkaline extraction of the antibiotic. 15,16 Since it was this deacetylated GPA that was initially named A40926, we will refer to it as A40926 hereafter.
A40926 is the precursor of the second-generation semisynthetic GPA dalbavancin (Figure 1), which is currently applied in clinics to treat severe infections caused by multidrug-resistant Gram-positive pathogens. 17 Dalbavancin (marketed in Europe and USA under the trade names xydalba and dalvance, respectively) is the first antibiotic designated as a qualified infectious disease product by FDA because of its potency, extended dosing interval, and unique dose regimen (once-a-week), but its cost still largely exceeds that of firstgeneration GPAs. 10 Therefore, improvement of A40926 production by recombinant engineering of N. gerenzanensis has become increasingly relevant. 6,16 Following the sequencing of the A40926 BGC (dbv) almost two decades ago, 18 multiple aspects of A40926 biosynthesis were investigated, including nonribosomal aglycone assembly and tailoring steps, 15,19,20 self-resistance, 21,22 and pathway-specific regulation of its production. 6,23,24 N. gerenzanensis was also engineered to produce A40926 derivatives that are better suited for downstream chemical modification to dalbavancin. 16 Another GPA produced by a Nonomuraea species is the type V GPA kistamicin ( Figure 1) from Nonomuraea sp. ATCC 55076, which was reported to exhibit potent antiviral activity as well as mild antibiosis against Gram-positive bacteria. 2,25 Its structure contains an unusual indole−phenol cross-link which makes this GPA unique among those already known. 5,26 Genome mining has recently shown that other species from the genus Nonomuraea also possess BGCs for GPAs, 27 as in the cases of Nonomuraea sp. WAC 01424 and Nonomuraea coxensis DSM 45129. Notwithstanding the low quality of the available Figure 1. Structures of the GPAs found in genus Nonomuraea: type IV A40926 and type V kistamicin. Clinically used dalbavancin is obtained from A40926 by conversion of the C-terminal carboxyl group into a (3-dimethylamino)-1-propylamide. Type IV teicoplanin and type III ristocetin are shown due to their structural similarities with A40926. For teicoplanin, the main factor (TA 2−2 ) of the complex produced by A. teichomyceticus is shown, the other factors are differing by the length and branching of the lipid chain. Ristocetin is produced by numerous Amycolatopsis species.
draft genomic data, we recently showed that N. coxensis DSM 45129 carries a BGC remarkably similar to dbv. 6 We found that this BGC contains a putative regulatory gene orthologous to dbv3, which encodes the pathway-specific regulator of LuxRtype in N. gerenzanensis. 6 The heterologous expression of this gene from N. coxensis (named nocRI) led to A40926 overproduction in N. gerenzanensis, indicating that it might be functional in N. coxensis as well. Thus, in this paper we present the fully assembled genome of N. coxensis, which has allowed us to properly describe the putative GPA BGC (called noc). Additionally, we report that N. coxensis produces a novel GPA complex, which we named A50926. Structural characterization of A50926 by liquid chromatography−mass spectrometry (LC-MS) and tandem MS (MS/MS) showed it has high similarity to A40926, although A50926 lacks the carboxyl group on the N-acylglucosamine (GlcN-Acyl) moiety. Consistently, the noc BGC lacks an orthologue of dbv29, which in N. gerenzanensis encodes the enzyme oxidizing the GlcN-Acyl moiety to an N-acylaminoglucuronic group. 19 Introduction of dbv29 into N. coxensis changed the GPA production profile of this strain to A40926. Finally, we have introduced dbv3 and dbv4 pathway-specific regulatory genes in N. coxensis to trigger and overproduce A50926 by regulatory gene cross-talking. In conclusion, our results describe the biosynthesis of a novel GPA, which may have superior properties to A40926 28 and thus may contribute to developing a platform for the combinatorial biosynthesis of third generation lipo-GPAs.

RESULTS AND DISCUSSION
2.1. Complete Assembly of N. coxensis Genome Reveals the Presence of a Novel GPA BGC. The presence of a novel GPA BGC in the genome of N. coxensis was recently anticipated. 6,27 However, due to the poor quality of the available draft, fragments of the BGC were found on different contigs and did not cover the full expected sequence of the BGC. Therefore, we sequenced and fully assembled the genome of N. coxensis DSM 45129 using a combination of HiSeq Illumina and GridION ONT technologies. The circular chromosome of N. coxensis was found to have a smaller size in comparison to the other two previously published Nonomuraea genomesonly 9.07 Mbp compared to 11.85 Mbp in N. gerenzanensis 3 and 13.05 Mbp in Nonomuraea sp. ATCC 55076. 2 The average GC-content was 71.8%. Annotation of the N. coxensis genome revealed 8398 predicted protein coding sequences, five operons for 16S-23S-5S rRNA, and 73 tRNA genes. Genome analysis by antiSMASH 5.0, 29 a specialized metabolite BGC identification tool, led to the discovery of 27 putative BGCs when used in the "relaxed" search mode. However, only a few BGCs showed more than 20% similarity to known BGCs (Table S1).
We thus focused our attention on the GPA-like BGC, which we denoted as noc (from Nonomuraeacoxensis). The noc BGC is the fourth GPA BGC described from Nonomuraea genus, following the dbv BGC from N. gerenzanensis, 18 a putative GPA BGC from Nonomuraea sp. WAC 01424 27 and the type V GPA kistamicin (kis) BGC from Nonomuraea sp. ATCC 55076. 2 Overall, noc contains 36 open reading frames (ORFs) with 35 among them homologous to dbv genes (the nonhomologous noc gene encoding for a putative transposase) and 32 being homologous to genes in the Nonomuraea sp. WAC 01424 GPA BGC ( Figure 2, Table 1). The kis BGC differed from noc most significantly (data not shown).
2.2. Comparative Genomics of Nonomuraea GPA Producers. At the time of writing, genomic information for 34 Nonomuraea species was available in GenBank, although there are only three complete assemblies (Table S2). Along with the four reported Nonomuraea GPA BGCs, we found a kis-like BGC in the draft genome of Nonomuraea sp. NN258 ( Figure  S1). We have then reconstructed the multilocus phylogeny (MLP) of all Nonomuraea species with available genomic data using conserved house-keeping proteins (Table S3). It revealed N. coxensis to be most closely related to N. wenchangensis CGMCC 4.5598, N. polychroma DSM 43925, and N. turkmeniaca DSM 43926 ( Figure S2). None of these species have GPA BGCs in their genomes. N. gerenzanensis is most closely related to Nonomuraea sp. FMUSA5−5 and to the kistamicin producer Nonomuraea sp. ATCC 55076, whereas Nonomuraea sp. WAC 01424 is distantly related to both N. coxensis and N. gerenzanensis ( Figure S2). Thus, GPAproducing Nonomuraea species do not form a single phylogenetic group, which is different from what occurs in the majority of Amycolatopsis spp. producing GPAs. 30 Since N. gerenzanensis and Nonomuraea sp. ATCC 55076 are closely related and their genomes had been completely assembled, we compared their sequences using the MAUVE genome alignment tool. 31 We found that the two genomes are very similar, having few rearranged homologous segments ( Figure S3A). Interestingly, the regions flanking the dbv BGC in N. gerenzanensis show synteny in Nonomuraea sp. ATCC 55076, but in this genome, they flank a miscellaneous assemblage of GPA-unrelated genes instead of the dbv genes. No dbv-like BGC is present in Nonomuraea sp. ATCC 55076. Similarly, no kis-like BGC is in the N. gerenzanensis genome, but the regions flanking the kis BGC in Nonomuraea sp. ATCC 55076 have their homologous counterparts in the N.
Homologue is absent.
ACS Chemical Biology pubs.acs.org/acschemicalbiology Articles gerenzanensis genome ( Figure S3A). Dot plots of N. gerenzanensis and Nonomuraea sp. ATCC 55076 confirm the high homology between the two strains ( Figure S3B). A possible explanation is that Nonomuraea sp. ATCC 55076 and N. gerenzanensis genomes might have acquired different GPA BGCs independently through horizontal gene transfer (HGT) events from other Nonomuraea (or not) species. Dot plots of N. coxensis and Nonomuraea sp. ATCC 55076 genomes ( Figure S3C) as well as of N. coxensis and N. gerenzanensis ( Figure S3D) indicate that N. coxensis is more distantly related to the other GPA producing species. Unfortunately, it was impossible to compare the genome of N. coxensis with its closest relatives N. wenchangensis CGMCC 4.5598, N. polychroma DSM 43925, and N. turkmeniaca DSM 43926 ( Figure S2), due to the incompleteness of their genome assemblies. Overall, it seems that the position of GPA BGCs is not conserved within Nonomuraea genomes, which contrasts to what was observed in most Amycolatopsis spp. 30 2.3. Comparing noc and dbv Biosynthetic Pathways: From Genes to Products. The biosynthesis of A40926 is well understood (Figure 3). The heptapeptide core of this antibiotic is synthesized by a nonribosomal peptide synthetase (NRPS) assembly line involving Dbv25, Dbv26, Dbv17, and Dbv16 proteins. The linear peptide is cross-linked by four monooxygenases (Dbv14, Dbv12, Dbv13, and Dbv11) and halogenated by Dbv10, giving the core aglycone. This aglycone is further modified with the glycosyltransferases Dbv9 and Dbv20, which attach N-acetyl glucosamine (GlcNAc) and mannose, respectively. 32 Then, GlcNAc is oxidized by Dbv29, deacetylated by Dbv21, and acylated by Dbv8. Finally, the . Conceptual scheme of the biosynthesis of A40926 and of the GPA (named A50926) from N. coxensis. Please note the dashed acyl group at the mannose residue, which is installed by Dbv23/Noc17 but consequentially lost during antibiotic extraction. For more details and encoded protein names, please refer to the main text and Table 1. mannose moiety is acetylated by Dbv23, giving O-acetyl-A40926.
Considering the A40926 pathway, it was possible to predict the biosynthetic pathway of the putative GPA from N. coxensis ( Figure 3). Sets of genes required for the biosynthesis of the nonproteinogenic precursor amino acids 4-hydroxyphenylglycine (Hpg), 3,5-dihydroxyphenylglycine (Dpg), and βhydroxytyrosine (further used as substrates for NRPS) are the same in noc and dbv BGCs (Table 1, Figures 2 and 3). Next, the NRPS, encoded within noc BGC, was found to have the same organization and A-domain specificities as the dbv NRPS ( Figure S4, Table S4). All other genes, responsible for the cross-linking and tailoring steps, were identical in both the noc and dbv pathways ( Table 1, Figures 2 and 3). However, one notable difference between dbv and noc was the absence of a dbv29 orthologue in the latter. As mentioned above, Dbv29 is a hexose oxidase responsible for the oxidation of the GlcN-Acyl moiety of A40926. 19 On this basis, we predicted that the noc pathway might produce an A40926 analogue lacking the carboxylic group on the GlcN-Acyl residue and therefore resembling teicoplanin in this moiety (Figures 1 and 3).
Beyond the biosynthetic genes, noc and dbv feature homologous regulatory genes. Two master regulators of A40926 biosynthesisLuxR-like Dbv3 and StrR-like Dbv4 have orthologues coded within nocNocRI (94% aa sequence identity) and NocRII (86% aa sequence identity), respectively. 6 In N. gerenzanensis, both Dbv3 and Dbv4 are crucial for biosynthesis activation. 23 Dbv4 was shown to bind the promoter regions of operons dbv30-35 (mainly coding for Dpg biosynthesis enzymes) and dbv14-8 (including the genes coding for cross-linking monooxygenases), and its binding sites were identified. 33 Our in silico analysis indicates that identical binding sites are present in the promoter regions of noc20 and noc8, orthologues of dbv14 and dbv30, respectively ( Figure  S5). DNA-binding sites of Dbv3 remain uncharacterized, but its regulon was defined from gene expression analysis and includes other biosynthetic genes and Dbv4. 23 Given all these similarities, we presume that NocRI/NocRII have functions identical to Dbv3/Dbv4 and both regulatory pairs might crosstalk between these species. Our previous results, 6 where heterologous expression of nocRI in N. gerenzanensis improved A40926 production, support this assumption. The single GPA resistance determinant encoded within noc is Noc27, a close (87%) orthologue of Dbv7 (VanY n ), which is a D,Dcarboxypeptidase involved in A40926 self-resistance. 21,34 Although the biosynthetic, regulatory, and resistance genes are apparently shared by the dbv and noc BGCs, their genetic organization is different. So far, almost all GPA BGCs have NRPS genes located on one strand in an order that is colinear to the order of the modules in the NRPS assembly line. The only exception is the dbv BGC, where the NRPS genes are coded on different strands and are separated by other biosynthetic genes. 18 The noc BGC, although sharing a remarkable similarity with dbv, features an organization of NRPS genes that is typical of all the other GPAs. Interestingly, only two chromosomal inversion events are needed to rearrange noc into dbv ( Figure S6), indicating how a dbv-like gene arrangement might have derived from a noc-like BGC in a common ancestor of N. coxensis and N. gerenzanensis (or in an ancestralprotocluster).
The putative GPA BGC in Nonomuraea sp. WAC 01424 ( Figure 2) differs more substantially from both noc and dbv. It lacks a noc14/dbv20 homologue encoding for a mannosyl-transferase, as well as a noc17/dbv23 homologue encoding for a mannose-O-acetyltransferase (Table 1). Instead, WAC 01424 GPA BGC contains a close homologue of staL ( Figure S7), which encodes for a sulfotransferase involved in the biosynthesis of A47934 from Streptomyces toyocaensis NRRL 15009. 35 Additionally, the WAC 01424 BGC-encoded halogenases seem more related to the ones from the A47934 BGC than to Noc24 and Dbv8 ( Figure S7). Thus, we suggest that WAC 01424 GPA is a nonmannosylated, but sulfated, A40926 analogue, putatively with a halogenation pattern different from A40926 ( Figure S8).
2.4. Optimization of GPA-Producing Conditions for N. coxensis. N. coxensis was first described in 2007, 36 but as far as we know, it was never tested for the production of antimicrobials. Considering the predicted similarity between the putative GPA produced by N. coxensis with A40926, we first applied to N. coxensis the cultivation and A40926 production conditions that we had previously optimized for N. gerenzanensis. 22,37 In these conditions (namely a vegetative preculture in E26 medium and a GPA production step in FM2 medium using baffled flasks), N. coxensis tended to grow poorly, and no antimicrobial activity was detectable throughout the 168h cultivation from inoculum. Thus, we further screened different media and fermentation conditions previously used for growing other GPA producing strains, such as TM1 used for teicoplanin production by A. teichomyceticus 38 and R5 adopted for balhimycin production in Amycolatopsis balhimycina, 39 as well as VM0.1 and ISP2l previously employed for the vegetative cultivation of N. coxensis 6 (media composition detailed in the Supporting Information). The production of antimicrobial activity toward Bacillus subtilis ATCC 6633 was observed only in TM1 and ISP2l media when glass beads were added to favor dispersed growth ( Figure S9). Indeed, adding glass beads to E26 medium cultures allowed us to use it for a successful vegetative preculture step ( Figure S10A). Interestingly, routine analysis of glucose consumption in all media described above indicated that N. coxensis did not visibly consume glucose during growth (data not shown). We thus tested the glucose-lacking E26 (named E27), TM1 (TM1m), and ISP2l (ISP2lm) media variants for N. coxensis growth and putative GPA production. We found that biomass accumulation was similar in E26 and E27 ( Figure S10A) and that biomass and antimicrobial production were equivalent in TM1 and ISP2l as well as in their glucose lacking variants TM1m and ISP2lm ( Figure S10B and C). Currently, it is impossible to say why N. coxensis fails to use glucose throughout cultivation given that all necessary genes are present within its genome ( Figure S10D). Thus, for all the following work with N. coxensis, E27, TM1m, and ISP2lm were used.
2.5. Expression of VanY-like Activity in N. coxensis. As already mentioned, the noc BGC encodes a Dbv7 ortho-logueNoc27. We therefore tested whether D,D-carboxypeptidase activity could be detected in GPA-producing cultures of N. coxensis. This was measured in membrane extracts as previously reported for N. gerenzanensis and its mutant strains. 22 D,D-carboxypeptidase activity was measurable in N. coxensis extracts, although at an inferior level than in N. gerenzanensis ( Figure S11). This indicated that Noc27 is functional and its expression correlates with the antimicrobial producing conditions. These results corroborate the hypothesis that noc genes are expressed and a novel GPA active versus B. subtilis is produced by N. coxensis. As in dbv 40 and WAC 01424 BGCs, a vanY gene seems to be the only cluster-situated determinant of self-resistance in N. coxensis.
2.6. Purification and Identification of the Novel Glycopeptide Complex Produced by N. coxensis. D-Alanine-D-Alanine (D-Ala-D-Ala) affinity resin chromatography was used to capture the putative GPA from cultures of N. coxensis grown in ISP2lm and TM1m media. ISP2lm appeared to be the most suitable medium for GPA purification, since the rich composition and high viscosity of TM1m interfered with affinity chromatography. Analyzed by HPLC, the affinity resin eluates contained two major peaks with the characteristic UV spectra of the commercially available A40926 standard, but with a different retention time ( Figure S12). LC-MS analysis of these peaks revealed they corresponded to ions with m/z 852.3 and 859.3 ([M + 2H] 2+ ), 28 and 14 Da smaller respectively than an A40926 standard ([M + 2H] 2+ = 866.3, corresponding to A40926 B). We therefore tentatively named this new GPA complex A50926 (Figure 4a and c). commercial standard of A40926 and the middle and bottom rows to culture extracts from ISP2lm and TM1m, respectively. For each mass, peak heights are normalized relative to the intensity of the largest peak in the sample set, shown in brackets at the top of each column. (b) MS spectra for A40926 B, A50926 B, and A50926 A. Peak heights are normalized to the intensity of the top peak in each spectrum, shown on the top right corner of each plot. Signature in-source fragments for each of the analyzed molecules are circled in pink, blue, and green, respectively, whereas the fragment corresponding to the mannosylated aglycone common to all of them is highlighted in yellow. (c) Proposed structure for the A50926 molecules. The top schematic represents a generic proposed structure common to A40926 and A50926 while the insets below represent the differential fragments for each of the analyzed molecules, as inferred from MS and MS/MS data. All three molecules showed similar MS spectra with single, double, and triple charge proton adducts as well as in-source fragments corresponding to the aglycone carrying the mannose moiety and the GlcN-Acyl moiety (Figure 4b,c). The mannosylated aglycone fragment (m/z 1374.3) was common to all three peaks (Figures 4b and S13), indicating that they share the same aglycone structure and mannose decoration. In contrast, the in-source fragment corresponding to the acylated sugar carried the signature mass difference for each molecule (Figures 4b, S14, and S15): the main A50926 peak ([M + H] + = 1717.5361) had a fragment with m/z 344.2, whereas the A40926 standard had a fragment with m/z 358.22 (Figures 4b,  S14, and S15). Further MS and MS/MS analyses of these fragments (Figures S14 and S16) allowed us to assign this 14 Da mass difference to the glucosamine moiety. The masses are consistent with this sugar featuring a regular 6-hydroxyl group in A50926 versus being carboxylated in A40926 (Figures 1, 3, 4c, S14, S15, and S16). This correlates with the lack of a homologue of dbv29 in the noc BGC, as it encodes the enzyme responsible for the oxidation of the C-6 hydroxyl group of GlcN-Acyl into a carboxylic acid in A40926. The second A50926 peak ([M + H] + = 1703.5172) had a further 14 Da mass difference in the GlcN-Acyl moiety (Figures 4b and S14), but in this case MS/MS showed this difference to be in the acyl chain ( Figure S16), which is consistent with an A50926 congener with a C11 acyl chain instead of a C12 acyl chain. This is equivalent to the A and B series of congeners in the A40926 complex. 41 Based on this analysis and accurate mass data (Figure 4b), we named the compound with [M + H] + = 1717.54 A50926 B (Figure 4c) and the compound with [M + H] + = 1703.52 A50926 A (Figure 4c).
2.7. Single Gene Expression Leads to A40926 Production in N. coxensis. To support our MS-based characterization of A50926, we hypothesized that we could convert N. coxensis into an A40926 producer by overexpression of the dbv29 gene from N. gerenzanensis, which encodes the hexose oxidase required for oxidation of the C-6 hydroxyl group of GlcN-Acyl into the corresponding carboxylic acid. To achieve this, we used the pSET152A expression platform, which has proven to be very effective for gene overexpression in both N. coxensis and N. gerenzanensis. 6 dbv29 was cloned into pSET152A to generate pSAD29, which was then introduced into N. coxensis by conjugation from Escherichia coli. N. coxensis pSAD29 + was grown in ISP2lm medium for 168 h, and the resulting GPA complex was purified using D-Ala-D-Ala affinity resin. LC-MS analysis determined that N. coxensis pSAD29 + was able to produce a molecule with an identical retention time and MS spectrum to that of A40926 (observed m/z 1731.5181, calculated A40926 [M + H] + 1731.5107, 4.27 ppm difference) (Figure 5a and b).
MS/MS analysis of the molecule showed it also had an identical fragmentation pattern to the A40926 standard, including the in-source fragment with m/z 358.22 characteristic of the carboxylated GlcN-Acyl moiety (Figures 5c and  S17). Traces of A50926 could also be detected in the extract of the complemented strain, indicating that while complementation was very efficient, conversion from A50926 to A40926 was not complete (Figure 5a). Alongside the BGC homology (Figure 2), this provides strong evidence that A50926 is chemically identical to A40926 with the exception of the carboxylated GlcN-Acyl. However, we cannot completely rule out small differences, such as acyl chain branching.
2.8. Heterologous Expression of Transcriptional Regulators dbv3 and dbv4 to Enhance the Production of A50926 in N. coxensis. In previous work, we overexpressed the two dbv BGC situated master regulators in N. gerenzanensis (dbv4 and dbv3) to successfully improve A40926 production. 6 Therefore, hereby we used the previously constructed expression vectors pSAD4 and pSAD3 carrying dbv4 and dbv3, respectively, in N. coxensis to trigger and improve A50926 production. First, we observed that N. coxensis pSAD3 + and pSAD4 + recombinant strains grown in the E27 and VSP vegetative media produced an antimicrobial activity against B. subtilis ( Figure S18A), whereas their parental wild type strain did not exhibit any antimicrobial activity in these media. Overexpression of dbv3 also triggered antimicrobial activity on VM0.1 and ISP2 solid media, whereas the wild type was not active ( Figure S18B). Consistently, in both ISP2lm and TM1m production media N. coxensis pSAD3 + and pSAD4 + produced more antibiotic than the wild type ( Figures  S18C and 6a and b). In ISP2lm (Figure 6a), at 192 h N. coxensis pSAD3 + reached the maximum production of approximately 45 μg mL −1 , exceeding both wild type (approximately 20 μg mL −1 ) and N. coxensis pSAD4 + (approximately 30 μg mL −1 ) productivities. In TM1m medium (Figure 6b), N. coxensis pSAD3 + produced approximately 50 μg mL −1 after 192 h of cultivation. At the same time point in TM1m the wild type and N. coxensis pSAD4 + produced approximately 16 and 22 μg mL −1 of antibiotic, respectively. The control strain carrying the "empty" pSET152A vector performed exactly as the wild type (data not shown). No significant differences between biomass accumulation or pH were observed among the recombinant strains, or in comparison with the parental N. coxensis wild type strain. Thus, overexpression of dbv3 and dbv4 regulatory genes triggered or improved the production of A50926 in N. coxensis under different cultivation conditions.

CONCLUSIONS
A novel GPA, A50926, was identified from N. coxensis DSM 45129. Detailed MS and MS/MS analysis indicates that A50926 differs from the previously characterized A40926 GPA by lacking the carboxyl group on the GlcN-Acyl moiety attached to Hpg4 of the GPA aglycone, resembling teicoplanin in this part of the molecule. A compound with the same chemical structure was described 25 years ago as a chemically prepared derivative of A40926 (named RA 28 ). Extensive study of antibacterial activities of RA in vitro 28 indicated that RA has slightly better antimicrobial activity than A40926: minimal inhibitory concentrations (MICs) of RA were 2−4 times lower against different staphylococcal and enterococcal strains when compared to A40926. The difference of chemical structure between the newly described A50926 and A40926 correlates with the absence of dbv29 orthologue in the A50926 BGC (noc). Consistently, when dbv29 was introduced into N. coxensis, we obtained A40926 production in the recombinant strain. Otherwise, both noc and dbv BGCs share all biosynthetic genes, which are closely related. Heterologous expression of A40926 regulatory genes dbv3 and dbv4 in N. coxensis improved A50926 production.
Although the majority of noc and dbv genes are orthologous, the dbv BGC is significantly rearranged in comparison to the noc BGC, as well as all other characterized GPA BGCs. We have proposed a series of genetic inversions that could have occurred in a common Nonomuraea ancestor to explain these different genetic architectures. Both BGCs are quite similar to the putative GPA BGC from Nonomuraea sp. WAC 01424. The latter lacks genes required for the addition of mannose, but possesses a gene encoding a sulfotransferase and an additional gene encoding a halogenase. Thus, the putative nonmannosylated GPA from Nonomuraea sp. WAC 01424 might be sulfated and have a different chlorination pattern than A40926/A50926. Consequently, Nonomuraea sp. WAC 01424 GPA BGC seems an attractive source for new tailoring genes to obtain A40926 derivatives with altered pharmacological properties. Notwithstanding the GPA BGC similarity, multilocus phylogeny of Nonomuraea spp. shows that GPA producers are not clustered together: GPA producers are found in distinct clades within the genus. Our analysis indicates that type IV and V GPA BGCs are common in Nonomuraea spp., which is in contrast to how rare these BGCs were believed to be. This is comparable to studies that show that BGCs for types I−III−IV GPAs are common in Amycolatopsis, and type V GPAs in Streptomyces. 27,42,43 This highlights how rare actinomycete genera, such as Nonomuraea, may represent a rich untapped source of novel GPAs, as well as GPA tailoring enzymes for the diversification of existing GPA scaffolds.  Table S6. Compositions of all the media used for cultivation and GPA production are also given in the Supporting Information. All media components and antibiotics were supplied by Sigma-Aldrich, unless otherwise stated. For routine maintenance, N. gerenzanensis and N. coxensis strains were cultivated on ISP3 agar medium supplemented with 50 μg mL −1 apramycin-sulfate when appropriate. For genomic DNA isolation, N. gerenzanensis and N. coxensis strains were cultivated in liquid VSP medium on an orbital shaker at 220 rpm and at 30°C. The working cell banks (WCBs) for N. gerenzanensis and N. coxensis strains were prepared as described previously. 22,37 E. coli DH5α was used as a routine cloning host, and E. coli ET12567 pUZ8002 was used as a donor for intergeneric conjugations. E. coli strains were cultivated at 37°C in LB liquid or agar media supplemented with 100 μg mL −1 of apramycin-sulfate, 50 μg mL −1 of kanamycin-sulfate, and 25 μg mL −1 of chloramphenicol when appropriate.
4.2. Plasmid Construction and Generation of Recombinant N. coxensis Strains. To construct the pSAD29 expression vector, the coding sequence of dbv29 (1601 bp) was amplified from the genomic DNA of N. gerenzanensis using dbv29_F/R primer pair (Table S7) and Q5 high-fidelity DNA polymerase (New England Biolabs). The resulting amplicon was digested with EcoRI and EcoRV restriction endonucleases and cloned into pSET152A 44,45 cleaved at the same binding sites. The resulting plasmid was verified by endonuclease restriction mapping and sequencing at BMR Genomics.
pSAD29, as well as pSAD3, 6 pSAD4, 6 and pSET152A, 44 were transferred to N. coxensis conjugatively, as described previously. 6 Transconjugants were selected as resistant to 50 μg mL −1 of apramycin-sulfate. Obtained strains were verified by PCR. To verify the integration of pSAD29, a ∼1.1 kbp fragment of pSAD29 was amplified using the dbv29_seq_int/PAM_seq_R (Table S7) primer pair, in which dbv29_seq_int anneals within dbv29 and PAM_seq_R anneals upstream the EcoRV cleavage site of pSET152A. To verify the integrations of pSAD4 and pSAD3, ∼1 kbp and ∼2 kbp fragments were amplified respectively using PAM_seq_F/dbv4_R and PAM_-seq_F/dbv3_seq_R primer pairs (Table S7). Finally, the integration of pSET152A was verified by amplifying aac(3)IV with the aac(3)IV_F/R primer pair (Table S7). In all cases, genomic DNA was isolated using the Kirby procedure. 46 4.3. N. coxensis Cultivation for A50926 Production. To initiate the cultivation of N. coxensis, one WCB vial was inoculated into a 250 mL Erlenmeyer flask with 50 mL of VSP reactivation medium containing 6 glass beads (ø5 mm). After 72 h of incubation on a rotary shaker at 220 rpm, 30°C the culture was used to inoculate (10% v/v) 500 mL Erlenmeyer flasks containing 100 mL of E27 vegetative medium and 12 glass beads (⌀ 5 mm). Following 72 h of incubation on a rotary shaker at 220 rpm, 30°C this culture was used to inoculate (10% v/v) 500 mL Erlenmeyer flasks with 100 mL of ISP2lm or TM1m production media containing 12 glass beads (⌀ 5 mm). A50926 production cultures were then incubated up to 240 h on a rotary shaker at 220 rpm, 30°C. Samples were collected at regular time points to estimate biomass accumulation (dry weight), pH, and A50926 production.
4.4. VanY-Related Activity Measurement. D,D-carboxypeptidase activity in Nonomuraea spp. was measured in FM2 production medium (N. gerenzanensis) and ISP2lm (N. coxensis) at 24, 48, 72, 96, 120, and 144 h time points. Mycelial lysates were prepared as described previously. 34 The enzyme activity releasing D-Ala from the tripeptide N-Acetyl-L-Lys-D-Ala-D-Ala (10 mM) was followed spectrophotometrically by a D-amino acid oxidase/peroxidase coupled reaction that oxidizes the colorimetric substrate 4-aminoantipyrine to chinonemine. D,D-carboxypeptidase activity was normalized to dry biomass weight, as previously reported. 22 One unit is defined as the amount of enzyme that is able to convert 1 μmol of substrate in 1 min.
4.5. HPLC and LC-MS Analysis of GPAs. For quantitative measurement, A40926 and A50926 were extracted from N. coxensis cultures with equal volumes of borate buffer composed of 100 mM H 3 BO 3 (Sigma-Aldrich) and 100 mM NaOH (Sigma-Aldrich), pH 12. During this extraction the O-acetylated forms were converted in the corresponding deacetylated GPAs A40926 and A50926. A40926 and A50926 were analyzed using HPLC as previously reported. 6,22,37 In all cases the injection volumes of studied samples and standards were the same (50 μL). Concentration of A50926 was estimated as follows: Where, C(A40926 std) is the concentration of the commercial A40926 sample; A(A50926) is the area sum of the peaks corresponding to A50926 B; A(A40926 std) is the area of the peak corresponding to the standard A40926 factor B 0 ; and 2 is the dilution factor. High resolution liquid chromatography−mass spectrometry (LC-MS) and fragmentation (MS/MS) analysis of A40926 and A50926 was carried out on a SYNAPT G2-Si mass spectrometer equipped with an Acquity UPLC (Waters). Samples were injected onto a Waters Acquity UPLC BEH 1.7 μm, 1 × 100 mm C18 column, and eluted with a gradient of (B) acetonitrile/0.1% formic acid in (A) water/0.1% formic acid with a flow rate of 0.08 mL min −1 at 45°C. The concentration of B was kept at 1% for 2 min followed by a gradient up to 40% B over 9 min, ramping to 99% B in 1 min, kept at 99% B for 2 min and re-equilibrated at 1% B for 4 min. MS data were collected in positive mode with the following parameters: resolution mode, scan time 0.5 s, mass range m/z 50−2000 calibrated with sodium iodide, capillary voltage = 2.5 kV; cone voltage = 40 V; source temperature = 120°C; desolvation temperature = 350°C. Leuenkephalin peptide was used to generate a lock-mass calibration with m/z 556.2766 for positive mode, measured every 90 s during the run. For MS/MS fragmentation, a data directed analysis (DDA) method was used with the following parameters: precursor selected from the 4 most intense ions; MS/MS threshold 5000; scan time 2 s; no dynamic exclusion. Collision energy (CE) was ramped between 8 and 35 at low mass (m/z 50) and 10−70 at high mass (m/z 1200).
4.6. Purification of GPAs Using D-Ala-D-Ala Based Affinity Resin. GPAs were purified by affinity chromatography with a D-Alanine-D-Alanine (D-Ala-D-Ala) based resin. Activation of 5 mL HiTrap NHS-activated HP affinity columns (GE Healthcare) and ligand binding was conducted as described before 47 with modifications. Briefly, the resin was activated with 30 mL of 1 mM HCl, followed by injection of 200 mM D-Ala-D-Ala dipeptide, dissolved into 5 mL of coupling buffer (0.2 M NaHCO 3 , pH 7.0). After 30 min incubation, the resin was washed with three cycles of 0.5 M ethanolamine hydrochloride, 0.5 M NaCl (pH 4.0, 30 mL), followed by 0.1 M sodium acetate, 0.5 mM NaCl (pH 4.0, 30 mL), alternately. Finally, the resin was washed with 50 mL coupling buffer and left to equilibrate for at least 1 h before use.
N. coxensis cultures were extracted in borate buffer as reported above, the pH in the obtained extracts was adjusted to 7.5 with HCl, and they were applied to the affinity chromatography system. Thus, extracts in borate buffer, coming from N. coxensis strains cultivated in TM1m or ISP2lm media, were filtered with 0.45 μm cutoff and loaded onto a D-Ala-D-Ala column at a flow rate of 0.5 mL min −1 . After extensive washing with coupling buffer, the bound GPA was eluted with 0.1 M NaOH and the eluate was lyophilized. 4.7. Bioassays for the Detection of A50926. Agar plug or Whatman paper disc (GE Healthcare) antibiotic diffusion assays were used to determine antimicrobial activities. An overnight B. subtilis ATCC 6633 culture in Mueller-Hinton broth II (cation adjusted, Sigma-Aldrich) was used to inoculate (1% v/v) a fresh culture, which was grown to OD 600 = 0.6. A 200 μL portion of this culture was then added to 25 mL of 0.7% (w/v ) Mueller-Hinton agar (Condalab) and plated. After solidification of the media, agar plugs cut from the plates with N. coxensis lawns, or Whatman paper discs containing GPAs, were placed on the agar surface. Bioassay plates were incubated for 16 h at 37°C before examination.
4.8. Sequencing and Annotation of the N. coxensis Genome. The genome of N. coxensis was sequenced using a combination of HiSeq Illumina and GridION ONT technologies. The Illumina data was obtained from SRA (PRJNA165411), while for the ONT data, a sequencing library (SQK-LSK109) was prepared using the Ligation Sequencing Kit (Oxford Nanopore Technologies) according to the manufacturer's instructions and run on a GridION sequencer in an R9.4.1 flowcell (both Oxford Nanopore Technologies). Base-calling of the raw data was performed with GUPPY-FOR-GRIDION v3.0.6. The assembly and polishing were performed as described previously, 48 using canu v.1.8 instead of v.1.6. The ONT data was assembled into 5 contigs, while the Illumina data were assembled into 87 scaffolds containing 310 contigs using NEWBLER v2.8. After manual curation using CONSED, 49 the complete genome of N. coxensis DSM 45129, consisting of one circular chromosome of 9,073,954 bp (72.12% G + C) was obtained. Annotation was performed using PROKKA v1.11 50 resulting in the prediction of 8,398 coding sequences (CDS), 5 rRNA operons, 73 tRNAs, and 5 noncoding RNA elements. The annotated genome and ONT raw data were deposited at DDBJ/ENA/GenBank under the BioProject accession number PRJNA693185.
4.9. In Silico Analysis Tools and Approaches. Routine analysis of nucleotide and amino acid sequences was performed in GENEIOUS v4.8.5. 51 Multiple sequence alignments, selection of the best models for the phylogenetic reconstruction and phylogenetic reconstruction itself were done with the MEGA X package. 52 To reconstruct the multilocus phylogeny of Nonomuraea, orthologues of 30 S. coelicolor house-keeping proteins (Table S3, 53 ) were identified within the genomes of 34 Nonomuraea spp. (Table S2) using reciprocal best hit (RBH) BLAST. Sequences of these proteins from each Nonomuraea spp. were concatenated, and these concatenates were used for the upstream phylogenetic reconstruction.
■ ASSOCIATED CONTENT * sı Supporting Information