Elucidation of unusual biosynthesis and DnaN-targeting mode of action of potent anti-tuberculosis antibiotics Mycoplanecins

Article https://doi.org/10.1038/s41467-024-44953-5 Elucidation of unusual biosynthesis and DnaN-targeting mode of action of potent anti-tuberculosis antibiotics Mycoplanecins Chengzhang Fu 1,2,7 ,YunkunLiu 1,7 , Christine Walt 1,3,7 , Sari Rasheed 1,3,7 , Chantal D. Bader 1,3 , Peer Lukat 4 ,MarkusNeuber 1,3 ,F.P.JakeHaeckl 1,3 , Wulf Blankenfeldt 4 ,OlgaV.Kalinina 5,6 &RolfMüller 1,2,3 DNA polymerase III sliding clamp (DnaN) was recently validated as a new antituberculosis target employing griselimycins. Three (2 S,4 R)−4-methylproline moieties of methylgriselimycin play significant roles in target binding and metabolic stability. Here, we identify the mycoplanecin biosynthetic gene cluster by genome mining using bait genes from the 4-methylproline pathway. We isolate and structurally elucidate four mycoplanecins comprising scarce homo-amino acids and 4-alkylprolines. Evaluating mycoplanecin E against Mycobacterium tuberculosis surprisingly reveals an excitingly low minimum inhibition concentration at 83 ng/mL, thus outcompeting griselimycin by approximately 24-fold. We show that mycoplanecins bind DnaN with nanomolar affinity and provide a co-crystal structure of mycoplanecin A-bound DnaN. Additionally, we reconstitute the biosyntheses of the unusual L-homoleucine, L-homonorleucine, and (2 S,4 R)−4-ethylproline building blocks by characterizing in vitro the full set of eight enzymes involved. The biosynthetic study, bioactivity evaluation, and drug target validation of mycoplanecins pave the way for their further development to tackle multidrug-resistant mycobacterial infections. Tuberculosis (TB) is a severe infectious disease caused by the pathogenic bacterium Mycobacterium tuberculosis (Mtb). TB remains an enormous global health burden, causing an estimated 1.5 million deaths and 10 million new cases in 2020. TB was the leading cause of death in all infectious diseases until the coronavirus (COVID-19) pandemic. Additionally, the growing percentage of multidrug-resistant (MDR) TBisa dauntingobstacleto globalTB treatment and prevention efforts. Since MDR TB is resistant to at least the two most potent TB drugs, isoniazid and rifampin, patients need new treatment options1,2. Consequently, new drugs addressing novel targets in Mtb are increasingly desired. Griselimycins (GMs) are such promising anti-TB antibiotics that bind and inhibit the mycobacterial DNA polymerase III sliding clamp (DnaN).This unique target makes GM distinctfrom other antibioticsto evadecommonTBdrugresistance 3. The unusual nonproteogenic amino acid (2S,4 R)−4-methylproline ((2S,4 R)−4-MePro) acts as an essential building block occurring twice in GM and thrice in methylgriselimycin (MGM). Oxidation of the metabolically unstable Pro8 Received: 22 June 2023 Accepted: 8 January 2024 Check for updates 1 Helmholtz Institute for Pharmaceutical Research Saarland (HIPS), Helmholtz Centre for Infection Research (HZI), and Department of Pharmacy, Saarland University, 66123 Saarbrücken, Germany. 2 Helmholtz International Lab for Anti-Infectives, Helmholtz Center for Infection Research, 38124 Braunschweig, Germany. 3 German Centre for Infection Research (DZIF), 38124 Braunschweig, Germany. 4 Structure and Function of Proteins, Helmholtz Centre for Infection Research, Inhoffenstr. 7, 38124 Braunschweig, Germany. 5 Medical Faculty, Saarland University, 66421 Homburg, Germany. 6 Helmholtz Institute for Pharmaceutical Research Saarland (HIPS), Helmholtz Centre for Infection Research (HZI), and Center for Bioinformatics, Saarland Informatics Campus, 66123 Saarbrücken, Germany. 7 These authors contributed equally: Chengzhang Fu, Yunkun Liu, Christine Walt, Sari Rasheed. e-mail: [email protected] Nature Communications | (2024) 15:791 1 1234567890():,; 1234567890():,; residue initiates GM degradation. The significantly improved metabolic stability of GM derivatives with substituent groups at Pro8, such as the natural MGM with (2 S,4 R)−4-MePro, has corroborated this finding3. More importantly, GM congeners with different 4-alkyl groups show improved activity against Mtb which might be attributed to their enhanced lipophilicity3. In our previous study, the Fe(II)/ α-ketoglutarate (α-KG)-dependent dioxygenase GriE was proven to initiate the biosynthesis of (2 S,4 R)−4-MePro by hydroxylating L-leucine to form (2 S,4 R)−5-hydroxyleucine. The zinc-dependent dehydrogenase GriF then converts (2 S,4 R)−5-hydroxyleucine to (2S,4 R) −4-methylglutamate-5-semialdehyde. The resulting semialdehyde will be turned to (2 S,4 R)−4-MePro by a spontaneous cyclization and a final reduction by the pyrroline-5-carboxylate reductase ProC or GriH (Fig. 1a)4. We previously reasoned that the genes required for synthesizing the (2 S,4 R)−4-MePro building block could be utilized to explore compounds containing such an unusual residue4. An earlier study discovered new natural products containing the diastereoisomer (2 S,4 S)−4-MePro from cyanobacteria employing different biosynthetic genes from the Nostopeptolide pathway as genome mining bait5. In this work, we address the question of whether novel (2S,4 R)-4MePro-containing compounds from Actinobacteria with improved bioactivity and physicochemical properties could be discovered via genome mining using the gene pair griE and griF as the probe, considering that the GM producer Streptomyces sp. DSM408356,7belongs to the order Actinomycetales, the most significant industrial source of natural product drugs8,9. We discover the previously unknown mycoplanecin (MP) biosynthetic gene cluster (BGC) and isolate several MP derivatives that indeed exhibited significantly improved pharmaceutical properties compared to GM. Analysis of a co-crystal structure and biophysical assays prove that MPs bind to DnaN in the nanomolar range. In addition, we comprehensively characterize in vitro three interwoven and uncommon biosynthetic pathways to the non-natural amino acids building blocks of MPs: L-homoleucine, Lhomonorleucine, and (2 S,4 R)-4-ethylproline ((2 S,4 R)-4-EtPro). Results Genome mining employing griEF as bait identifies the mycoplanecin biosynthetic pathway Our search for BGCs including griE and griF homologs in all actinobacterial genomic sequences in the NCBI nuccore and assembly revealed dozens of hits (Fig. 1b and Supplementary Table 1). This mining strategy was effective as it found the BGCs known to produce compounds containing 4-MePro, such as GM4and acyldepsipeptide (ADEP) antibiotics10 (Fig. 1b, c). Among the unknown hits which are mainly nonribosomal peptide synthetase (NRPS) or NRPS hybrid BGCs, one hit similar to the GM pathway drew our attention because it exhibited significant differences. However, only a partial BGC could be recovered due to the low genome sequence quality. This short contig containing griEF homologous genes belongs to the genome of Actinoplanes awajinensis subsp. mycoplanecinus NRRL B-1671211. Additional genes that are absent in the GM pathway were identified, including such encoding a radical S-adenosylmethionine (SAM) protein, an acyl carrier protein (ACP), and a 3-ketoacyl-ACP synthase(KAS) III homolog (Fig. 1b, d). Fortuitously, we found that this NRRL strain is equivalent to A. awajinensis subsp. mycoplanecinus subsp. nov. ATCC 33919 (Strain No. 41042) producing anti-mycobacterial MP peptides discovered in the 1980s12–16. Notwithstanding the structural similarity Fig. 1 | Discovery of mycoplanecin biosynthetic gene cluster (BGC) by sequence similarity network (SSN) analysis. a GriE and GriF are essential for biosynthesis of (2S,4 R)-4-MePro in the griselimycin (GM) pathway.bSequencesimilarity networks for GriE (left) and GriF (right). All edges corresponding to sequence identity below 80 % were removed. The color of the edge represents the sequence similarity, with darker color corresponding to higher similarity. Node colors correspond to the type of the biosynthetic gene cluster (Supplementary Table 1). cChemical structures of GM and ADEP1 which contain (2 S,4R)-4-MePro. dTheincompleteNRPS BGC from Actinoplanes awajinensis NRRL B-16712 shows similarity to GM BGC but also contains several different genes. Article https://doi.org/10.1038/s41467-024-44953-5 Nature Communications | (2024) 15:791 2 to GM, MPs possess unique characteristics, including the exceptional N-terminal α-keto butyric moiety, as well as the other two 4-alkylprolines: 4-EtPro and 4-propylproline (4-PrPro), homoleucine (5-Methylnorleucine), and N-methyl-homonorleucine (N-methylheptanoic acid) (Fig. 2). Characterization of the MP BGC possessing a selfresistance gene Because of the highly fragmented genome of B-16712 we PacBio sequenced ATCC33919 and obtained the intact 59 kb BGC of MP by searching for the griEF hit sequence in its completed genome (Fig. 2a). The putative MP BGC comprises 15 genes divided into three hypothetical operons that encode NRPSs, precursor biosynthetic enzymes, and a self-resistance protein (Fig. 2a, Supplementary Table 2). We found a DnaN-encoding gene mypR nearby transposase genes, which was also observed for griR in the GM pathway4.Furthermore,mypR is found to be the second DnaNgene in the genome of strain ATCC33919, implying that MPs are potential DnaN inhibitors, and mypR confers self-resistance against MPs3. The majority ofgenes are clustered in one operon, which encodes NRPSs MypABC, the MbtH protein MypD, redox enzymes MypP (P450), MypE (hydroxylase), and MypF (dehydrogenase), the thioesterase MypG, the discrete ACP MypI, and the ketosynthase MypJ, as well as a radical SAM enzyme MypH. Inconsistencies in the assembly of the NRPS gene mypA caused by repetitive sequences were solved by comprehensive PCR amplification and restriction digestion analysis (Supplementary Figs. 1–3). The three gene operon comprising mypKLM is located upstream adjacent to the primary operon and oriented in opposite direction (Fig. 2a). MypK and MypL show sequence similarity to enzymes involved in the branchedchain amino acid (BCAA) biosynthesis. MypM is a SAM synthetase probably involved in regenerating the cofactor required by MypH. MypA, MypB, and MypC comprise six, three, and one single NRPS module, respectively. The substrate specificity prediction of adenylation (A) domains correctly suggested the incorporation of Thr3, Leu4, Fig. 2 | MP biosynthetic gene cluster and proposed biosynthetic pathway. a The gene organization and comparison of GM and MP biosynthetic gene clusters. bThe MP biosynthetic pathway proposal. The A domains and corresponding atypical building blocks are shown in orange, red, or olive green, respectively. The proteins involved in the hypothetic biosynthetic pathway of the α-ketobutyrate precursor are in purple. For eachmodule, the domains are:C = condensation, A = adenylation, MT = methyltransferase, T = thiolation domain, also known as peptidyl carrier protein domain in NRPS, E = epimerase, TE = thioesterase. Article https://doi.org/10.1038/s41467-024-44953-5 Nature Communications | (2024) 15:791 3 Pro8, Leu9, and Gly10 (Fig. 2b). The four N-methyltransferase (MT) domains align with the N-methylation of Val1, Thr3, Val7, and Leu9 in MPs. However, modules 1 and 7 were inappropriately predicted to activate Thr instead of Val. Intriguingly, several nonproteogenic amino acid building blocks were found in the biosynthesis of MPs. A6 is predicted to activate ornithine, but L-homoleucine was found incorporated in MPs in this position. The A9 domain has a promiscuous substrate tolerance since L-N-methyl-homonorleucine and L-N-MethylLeu are present in different MPs at this position. The substrate specificity of A2, A5, and A8 was predicted as proline, indicating that the current bioinformatic analysis could not discriminate proline from different alkyl-prolines4.Thesubstratespecificity of A2 is flexible as three different 4-alkylproline residues areincorporated in different MP derivatives. However, it is also possible that different 4-alkylproline residues are formed by post-modifications of 4-MePro in MP B (Fig. 2b). It is noteworthy that uncommon genes responsible for lipopeptide chain initiation are found in the MP pathway, partially similar to the proposed pathway for calcium-dependent antibiotics (CDAs)17. Here, it is hypothesized that the holo-MypI (ACP) is either malonylated by the malonyl-CoA-acyl carrier protein transacylase (MCAT) FabD or functions as a self-priming ACP that widely exists in fatty acid biosynthesis18,19. Subsequently, the KAS III-like enzyme MypJ extends the acyl chain by a molecule of acetyl-CoA to form the acetoacetyl-SACP species, which is probably reduced to butyryl-S-MypI by reductases from the type II fatty acid biosynthesis. A P450-dependent oxidase encoding gene mypP is assumed to be responsible for the cryptic α-keto group formation, albeit the mechanism and oxidation timing are still unexplored (Fig. 2b, Supplementary discussion). Isolation and structure elucidation of MPs with unusual building blocks Mycoplanecins were first investigated four decades ago due to their activity against Mycobacteria while possessing low cytotoxicity, but no subsequent studies are known from the literature. The absolute configurations of MPs are not completely clear12–16, nor is the biosynthesis or the molecular target. Given the presence of the putative selfresistance gene mypR and the finding of novel biosynthetic genes, we re-isolated these potent anti-tubercular compounds aiming to identify unprecedented derivatives for subsequent characterization. The three most abundant derivatives were purified and structurally elucidated as the previously described MP A, B, and D (Fig. 3a)12,13,15. A complete set of 1D and 2D nuclear magnetic resonance (NMR) experiments was acquired for each derivative to underpin the structures and to supplement previous MP spectral data (Supplementary NMR Data)12,13,15. The last derivative isolated, MP E, is structurally novel with a 4-PrPro and a N-methyl-homonorleucine residue incorporated, the largest of the MPs (Fig. 3a). Comprehensive studies ofCOSY,HSQCandHMBCspectraindicatethepresenceofα-ketobutyrate, N-methyl-valine, leucine, 4-methylproline, 5-methylnorleucine, proline, glycine and N-methyl-threonine analog to the known derivatives MP A, B and D. Furthermore, the spectral data reveal that MP E contains the same 4-propylproline moiety as described in MP D, which is substituted by 4-methylproline in MP A and 4-ethylproline in MP B. N-methylleucine found in MP A, B and D is replaced by the rare N-methylheptanoic acid building block in MP E. This moiety is characterized by the three methylene signals at δ(1H) = 1.16 (2H, m) δ(13C) = 38.9, δ(1H) = 1.26 (2H, m) δ(13C) = 27.3 and δ(1H) = 1.64 (2H, m) δ(13C) = 24.9 as well as the more shielded methyl signal at δ(1H) = 0.90 (3H, m) δ(13C) = 14.1 ppm. Besides this, the chemical shifts of the remaining structure of MP D and MP E exhibit a deviation of less than 0.12 ppm for δ(1H) and 2.5 ppm for δ(13C) (Supplementary NMR Data). We proposed the stereochemistry of all derivatives through Marfey’s analysis and biosynthesis analysis, confirming that MP A comprises Nmethyl-valine, 4-EtPro, leucine, 4-MePro, proline, N-methyl-threonine in L-configuration while N-methyl-leucine is the only D-configured amino acid underlined by the only epimerization (E) domain (Fig. 2b and Supplementary Figs 21-28, 35−41, 48-55, 60-67). MPs show low nanomolar anti-Mtb potency Importantly, bioactivity testing of the MPs against the attenuated M. tuberculosis strain H37Ra uncovered superior activity in the nanomolar range (minimum inhibitory concentration (MIC) from 30 to 100 ng/mL (Fig. 3b)). Additional MPs tested against M. smegmatis revealed potent activity from 0.0625 to 0.5 µg/mL. Although MPs remain relatively effective against a GM-resistant M. smegmatis strain3with MIC values ranging between 4 to 8 µg/mL, there was a clear MIC shift (ranging between 16 to 64-fold), indicating that mycoplanecins have the same target as GM. It should be noted that resistance against griselimycins is Fig. 3 | Chemical structures and bioactivity of MPs. a The chemical structures of MPs isolated in this study. bThe comparison of bioactivity of MPs and GM against Mycobacteria and dissociation constants against DnaN from M. smegmatis mc2155. The K D values with standard deviation were calculated from three independent microscalethermophoresismeasurements.cThestronginductionofSOSresponse through RecA in M. smegmatis treated with 2x MIC MP A. dThe strong induction of DNA repair through RadA in M. smegmatis treated with 2x MIC MP A. Article https://doi.org/10.1038/s41467-024-44953-5 Nature Communications | (2024) 15:791 4 associated with severe fitnesslossinthemutants 3. While MPs were less potent against the virulent M. tuberculosis strain H37Rv than the attenuated H37Ra, they still exhibit an impressive ~24-fold more potent bioactivity than GM (Fig. 3b). Surprisingly, the best MIC was 0.083 µg/ mL for the new congener MP E, while the MIC of GM is 2 µg/mL. MPs were additionally tested against a panel of Gram-positive (S. aureus Newman and B. subtilis DSM 10) and Gram-negative (E. coli BW25113, C. freundii DSM 30039 and A. baumannii DSM 30008), but exhibited no activity up to 64 ug/mL (Supplementary Table 6). The mycobacteriaspecific inhibitory bioactivity of MPs highlighted the high selectivity of this compound class, also concerning members of the human microbiota that are likely spared in the potential use of MP in TB therapy. In addition, MPs showed no cytotoxicity against CHO-K1 cells up to a concentration of 37 µg/mL, which underpins the high potential of MPs in the application to combat mycobacterial infections. DNA damage response after exposure to MP A was evaluated in M. smegmatis reporter strains harboring mycobacterial promoters that drive the expression of DNA damage-inducible genes, recA and radA. Strong induction of luminescence was observed for the reporter strains upon treatment with 2x MIC MP A, indicating that the target of MPs is involved in DNA metabolism that results in genotoxic stress. (Fig. 3c, d). MPs bind to DnaN in nanomolar affinity Thehomologyoftheself-resistancegenegriR from the GM BGC with mypR in the MP BGC and MPs’structural analogy to GM implied that DnaN is the molecular target of MPs. We determined the binding affinity of all isolated MPs as well as GM to DnaN by microscale thermophoresis (MST). The highest affinity to M. smegmatis DnaN (msDnaN) was observed for GM with K D =6.5±5.9nM. Lower DnaN binding affinities were determined for MPs as the K D -value for MP A was 95.4 ± 58.0 nM while MP B showed strongest binding to msDnaN among the congeners with a K D = 24.4 ± 11.9 nM. The other two MPs showed a similar binding range to msDnaN (Fig. 3b, Supplementary Fig. 4). Interestingly, GM manifests a slightly stronger DnaN binding affinity than the MPs but is less active than MPs against mycobacteria. This intriguing phenomenon is unresolved but is likely a result of MPs’ improved physicochemical properties including increased lipophilicity, which is important for TB activity. Crystal structure of DnaN in complex with MP A Although the co-crystal of msDnaN and MP A could not be obtained after intensive attempts,DnaN of E. coli (ecDnaN) was overproduced in E. coli and co-crystallized with MP A, GM and cyclohexyl-GM (CGM). Despite sharing only 29 % sequence identity with msDnaN, the binding pocket of ecDNA shows a sufficient degree of conservation (Supplementary Fig. 5). The crystal structures were solved at resolutions of 2.12 Å (MP A), 1.65 Å (GM) and 1.50 Å (CGM) (Supplementary Fig. 6 and Supplementary Tables 8, 9). MP A was foundto bind in the same mode to ecDnaN as GM or CGM. The binding pocket can be divided into a larger (subsite 1) and a smaller cavity (subsite 2) and as hasbeen shown previously for GM and mycobacterial DnaN, all compounds occupy both subsites of the binding pocket. All three substances form hydrogen bonds to only 2 –3 protein residues, but the additional oxygen in the N-terminal α-keto butyrate residue of MP A provides a second hydrogen bond acceptor for Arg365 (Supplementary Fig. 7), which is also conserved in mycobacterial DnaN. The compounds bind mainly via van der Waals interactions with interfaces ranging from 550 Å2for GM to 590 Å2for CGM to 600 Å2for MPA. Compared to available structures of uncomplexed DnaN, only a few residues undergo significant conformational changes upon binding of the compounds - most notably Met362, which has been described to function as a gate between the subsites (Supplementary Fig. 8)20. The linear, N-terminal part of the ligands in subsite 2 superimposes almost perfectly, while MP A seems to be tilted slightly (approximately 7-10 ° compared to GM/CGM)away from the protein in theregion around Pro8. This tilting can be attributed to the longer 5-methyl-l-norleucine residue of MP A atposition 6,whereasleucineispresentatthe samepositioninGMand CGM (Fig. 4and Supplementary Fig. 9). The comparison with the previously published mycobacterial GM/DnaN complexes shows that whereas the N-terminal parts of the ligands in subsite 2 superimpose very well, there is higher variance on the opposite side of the molecules around Pro8 (Supplementary Fig. 9). In the complex with msDnaN, GM Fig. 4 | Binding of MP A to ecDnaN. a Superimposition of the ecDnaN co-crystal structures containing MP A (8CIZ, dark/light green for the two entities in the asymmetric unit),GM (8CIX, orange),and CGM(8CIY,violet).The proteinsurfaceis shown in gray. The two subsites of the binding pocket are separated by a dashed white line. The structures were superimposed on the level of single amino acid chains of DnaN. bBinding comparison of GM (orange) and MP A (green) to ecDnaN (gray cartoon). The 5-Methyl-norleucine residue at position 6, that pushes MPA away from the protein in comparison with leucin 6 in GM is highlighted in light green. cView of the GM/MP A binding comparison turned 180° around the Y axis. Article https://doi.org/10.1038/s41467-024-44953-5 Nature Communications | (2024) 15:791 5 adopts a conformation very similar to that of MP A in the ecDnaN complex, although GM lacks the 5-methyl-l-norleucine at position 6 and the residues lining the protein pocket around this position are conserved between mycobacteria and E. coli.IncomplexwithM. tuberculosis DnaN, the tilt of GM away from the protein is less pronounced, instead the macrocycle seems to be tilted approximately 7° towards positon 7 in comparison to the msDnaN complex. Although differences in crystal packing make a direct comparison of the binding strength difficult, the linear part of the ligands (1-3) and the residues adjacent to the bifurcation (4-5, 10) seem to be bound tighter/show less flexibility than the opposing residues (6-9) in all structures of DnaN/GM complexes (Supplementary Fig. 10). In vitro reconstitution of the α-keto acid elongation subpathway NRPS megasynthetases exhibit the unique feature of incorporating unusual and nonproteinogenic amino acids into bioactive peptidic molecules. For instance, 4-ethyl-L-proline/4-propyl-L-proline, L-homoleucine, and L-homonorleucine as present in MPs, often contribute to the bioactivities whereas their biogenesis remains mysterious. We proposed the biosynthesis of the homo-amino acid residues in MPs as biosynthetically sharing an analogous process to leucine biosynthesis, as previously also suggested for a few other homo-amino acids. Two homologous genes of leucine biosynthesis were identified within the MP BGC. mypK encodes a 2-isopropylmalate synthase-like enzyme that likely catalyzes the condensation of different six-carbon α-keto acids, including 4-methyl-2-oxopentanoate (1), 2-oxohexanoate (2), and 3methyl-2-oxopentanoate (3) with acetyl-CoA. The other gene encodes a didomain enzyme MypL, which contains a putative N-terminal isopropylmalate isomerase subunit (LeuD) homolog and a C-terminal pyridoxal phosphate (PLP)-dependent transaminase domain. Although the dedicated dehydrogenase (leuB) and the other isopropylmalate isomerase subunit (leuC) genes were not found in the MP pathway, the broad substrate tolerance of leucine biosynthesis enzymes was nevertheless considered to be important for homo-amino acids biosynthesis as well21. Thus, MypK, MypL, together with aaLeuB and aaLeuC from primary metabolism of A. awajinensis ATCC 33919, were proposed to biosynthesize the homo-amino acid residues (10, 11,and 12)(Fig.5a). We set out to study these intriguing steps in vitro because Actinoplanes species are notoriously recalcitrant to genetic manipulation, includingstrainATCC33919.Theacceptanceof1, 2,and3assubstrates by the recombinant His-tagged MypK was demonstrated by the newly detected mass peaks showing 60 Da mass increase, corresponding to the compounds 2-isobutylmalate (4), 2-n-butylmalate (5), and 2-secbutylmalate (6), respectively (Fig. 5b). Moreover, MypK exhibits a preference for 1compared to 2and 3, which was verified by measuring enzyme kinetics (Supplementary Table 10). A new peak with m/z shift of 1 Da corresponds to the expected mass of L-homoisoleucine (12)and was detected by LC-MS in the reaction of the recombinant aminotransferase MypL C in the presence of L-glutamate as the amino group donor and 4-methyl-2-oxohexanoate (9)asthesubstrate.MypL C was tested under the same condition revealing an extensive range of proteinogenic amino acids as amino donors (Supplementary Fig. 11)22. The substrate acceptance of 4-methyl-2-oxohexanoate for MypL C indicated that MypL C was likely involved in forming 10, 11,and12.The successful purification of the remaining enzymes enabled the total invitroreconstitutionofhomo-aminoacid biosynthesisinonepot(see below and Supplementary Fig. 12). MypL N was co-purified with aaLeuC in an intensive brown color, indicating the binding of an iron-sulfur cluster. The one-pot reaction was executed in three stages initiated by MypK, followed by the reaction of aaLeuC and MypL N co-purified protein, and aaLeuB together with β-Nicotinamide adenine dinucleotide in its oxidized form (NAD+). After the last stage reaction with MypL C ,D-FDLA was used to derivatize potential amino acid products. The reaction product using 1as the substrate had the same retention time compared with a peak with the same mass from the hydrolyzed and D-FDLA derivatized MP D, suggesting the formation of 10.Similarly, the mass corresponding to the derivatized 11 or 12 was detected from the reaction using 2or 3as substrate, respectively. (Fig. 5band see below). In vitro reconstitution of the 4-alkylproline subpathway Previously, 4-EtPro and 4-PrPro were characterized to be derived from L-tyrosine in the biosynthesis of lincomycins (Supplementary Figs. 13 and 14a)23, however, no homologous genes were found in the MP pathway. As reported in GM biosynthesis, (2 S,4 R)−4-MePro was the Fig. 5 | The homo-amino acid building blocks of MPs are built by a pathway mimicking BCAA biosynthesis. a Proposed subpathway and intermediates of Lhomoleucine (10),L-homonorleucine (11)andL-homoisoleucine (12)formation.The names of proteins from the MP pathway involved in the homo-amino acid subpathway are highlighted in orange. bIn vitro assays of the purified protein MypK. Extracted ion chromatograms (EICs) corresponding to seven-carbon α-keto acids detected in negative ion mode ([M-H]− ,m/z 189.07). No formation of 2-isobutylmalate (4) could be detected with boiled MypK and substrate 1as control (I). The formation of 4,2-n-butylmalate (5), and small amount 2-sec-butylmalate (6) were observed upon incubation of MypK and 1(II), 2(III), and 3(IV), respectively. Article https://doi.org/10.1038/s41467-024-44953-5 Nature Communications | (2024) 15:791 6 product of a series of enzymatic modifications of leucine (Fig. 1a)4.A pair of genes mypE and mypF from the MP pathway shows homology with the genes involved in the 4-MePro biosynthesis, including ldoA and nosE from Cyanobacteria, and griE and griF in GM biosynthesis4,24. However, the biosynthesis pathways of 4-EtPro and 4-PrPro in MPs remained elusive. By supplementing the culture of strain ATCC33919 with LMethionine-(methyl-13C) and subsequent MS analysis of the resulting mycoplanecins, a 1 Da difference between control and isotope-labeled MPB ( + 4 Da),A ( + 5 Da) and D( + 6 Da) suggested the extracarbons of 4-EtPro and 4-PrPro are derived from methionine that is used for radical-SAM based methylation of 4-MePro (Fig. 6a). We speculated that MypE/F form 4-MePro as described in the GM pathway (Fig. 6b). However, the timing of subsequent methylationby the radical SAMMT MypH is still elusive (Fig. 6c). MypH might iteratively methylate the unreactive methyl-carbon of nascent substrates at different stages, including the α-keto acids before chain elongation, the homo-amino acids before hydroxylation by MypE, or 4-MePro (Fig. 6c). In addition, we can not exclude that 4-EtPro and 4-PrPro-containing MPs could be produced by modifying MP B by MypH as a post-assembly tailoring enzyme. The formation of 4-hydroxymethyl-L-homoisoleucine (13)was observed asa new mass with a 16 Da increase in the D-FDLA derivatized reaction product of recombinant MypE incubated with homoisoleucine (stereoisomer mixture) since the pure (2S,4 R)-homoisoleucine is not commercially available. Surprisingly, three individual new peaks with the same m/z were observed after derivatization by DFDLA, suggesting loose stereospecificity of MypE (Fig. 7b). A series of structural analogs were investigated for substrate tolerance of MypE, but no conversion was detected with L-valine, D-isoleucine or D-alloisoleucine. In contrast, a trace amount of products of D-leucine and Lisoleucine, and nearly complete conversion of L-leucine and L-alloisoleucine was observed by LC-MS (Fig. 7b). These results indicated MypE is a promiscuous enzyme recognizing substrate stereoisomers, however, preferring L-forms of amino acids, especially the L-allo form amino acid in the case of two stereocenters. The reaction of MypF proceeded well with the substrate 4-hydroxymethylhomoisoleucine (13)fromE. coli overexpressing recombinant MypE and incubated with homoisoleucine, as determined by a significant rise of absorbance at 340 nm, indicating NADH formation (Fig. 6d). A new signal corresponding to the D-FDLA-derivatized (2 S,4 R)−4-EtProwasobserved inthecoupledassaywith MypE, MypF, and aaProC (pyrroline-5-carboxylate reductase from the MP producer), which is consistent with the mass and retention time of the derivatized hydrolysis product of MP D (Fig. 7a). Discussion Giventhe promising activity of griselimycinas a potential anti-TB drug, its innovative target DnaN, and the involvement of nonproteogenic amino acids in target binding, we became interested in identifying additional natural products exhibiting similar chemical moieties. Investigation of all publically available actinobacterial genomes using (2 S,4 R)−4-MePro biosynthesis genes4unveiled a previously unknown BGC with additional interesting precursor biosynthetic genes compared to the GM pathway. This finding inadvertently disclosed the mycoplanecin BGC and correlated it to these compounds initially discovered in the 1980s with potent and specific anti-tuberculosis bioactivities12–16. Fig. 6 | The biosynthesis of alkylproline residues in MPs. a MS analysis of production of MPs by fermentation without (I: MP B, III: MP A, V: MP D) or with (II: MP B, IV: MP A, VI: MP D) L-Methionine-(methyl-13C). bProposed subpathway of 4-EtPro and 4-PrPro formation in MP biosynthesis. cPotential methylation timepoints by MypH. dUV/Vis spectra of MypF and heat-inactivated MypF in the assay. Article https://doi.org/10.1038/s41467-024-44953-5 Nature Communications | (2024) 15:791 7 Here, four MPs, including one new congener with the most potent anti-Mtb bioactivity, were isolated and found to contain several unusual nonproteogenic amino acid residues, including three different 4-alkylprolines. Marfey’s analysis plus the NRPS E domain for the first time proved the absolute stereochemistry of the MPs. These amino acid residues, especially (2 S,4 R)−4-EtPro and (2 S,4 R)−4-PrPro, are scarce in natural products. Interestingly, the 4-PrPro and 4-EtPro in lincomycins instead derive from L-tyrosine degradation as catalyzedby a subcluster of six genes25 which are not present in the MP BGC. A step-by-step in vitro total reconstitution of the BCAA-like pathway of MP biosynthesis revealed a set of promiscuous enzymes which build unusual homo-amino acids from different 2-oxo acids. Subsequently, we in vitro reconstituted 4-EtPro using a pair of GriEFlike enzymes from the MP pathway. A feeding study with L-Methionine- (methyl-13C) suggested the extra carbons of 4-EtPro and 4-PrPro over 4-MeProin MPslikelyto comefrom radical SAM methylation byMypH. The successful total in vitro reconstitution of 4-EtPro formation starting from 3implied a potential 4-alkylproline pathway with a yet uncertain methylation timepoint (Figs. 6cand7a). Considering the substrate tolerance of all related Myp enzymes as shown for homo-amino acids and 4-EtPro reconstitution, one reasonable pathway is MypE hydroxylation of 12 to form 13 and subsequent oxidation and spontaneous cyclization yielding the pyrroline intermediate that is further reduced to (2S,4 R)−4-EtPro. The 4-PrPro residue canbeproduced followingthe samelogic.However,itrequiresthe previously unknown 3-methyl-2-oxohexanoic acid as the starting substrate, which needs experimental proof in the future (Fig. 6c). It is still mysterious when exactly MypH performs its function since potentially the radical methylation can either take place on the 4-alkylproline precursor or as a tailoring step to modify MP B directly. Intriguingly, MPs showed significantly more potent anti-Mtb bioactivity than GMs. We demonstrated that the new congener MP E shows the most potent MIC (0.083 µg/mL), exhibiting 24-fold higher potency against Mtb H37Rv than GM. We showed MP A in our hands has a four-fold lower MIC (0.102 µg/mL) against Mtb H37Rv than initiallyreported(0.39 µg/mL)14.The analogouschemicalstructuresofMPs and GMsalign with their identical molecular target site withinDnaN. In comparison to GM and CGM, the higher potency of MP A might be attributed to the increased hydrophobicity of the substance and the larger van der Waals interface with DnaN. The 5-methyl-L-norleucine at position 6 might also play a role by pushing the generally less-tight bound region around Pro8 out of the binding pocket but stabilizing the tighter interaction around the ligand’s bifurcation. The co-crystal structure of ecDnaN bound with MP A and the nanomolar binding affinity of MPs to msDnaN validates DnaN as the molecular target. The binding mode is similar to GM and CGM, albeit MPs showed significantly stronger anti-Mtb bioactivity. This finding is of great importance as GM development towards a TB drug was stopped dueto unwanted but target-independent side effects (unpublished results). As mycoplanecins show significantly lower MICs they might be useful to overcome this limitation as the potential therapeutic window over the side effect is significantly improved. TB is still one of the leading public health threats, even during the global COVID-19 pandemic. Drug-resistant strains of Mtb further deepen the crisis because we are running out of our antibiotic arsenal. The low cytotoxicity and specific anti-mycobacterial activity Fig. 7 | In vitro reconstitution of the biosynthesis of homo-amino acid and alkylproline residues. a In vitro analysis of the extender unit reconstitution. EICs of both D-FDLA derivatized reaction samples and hydrolyzed MP D and A ascontrol were analyzed. Derivatized 10 (m/z 440.213) was detected in one pot incubation (I) with 1, MypK, aaLeuC+MypL N , aaLeuB and MypL C , and from hydrolyzed and derivatized MP D standard (II). The peak at ~19.5 min (retention time) in II corresponds to Marfey’s derivatization product of N-methyl-D-leucine. Derivatized 11 and 12 products (m/z 440.213) were found using 2and 3as substrate in one pot assay, respectively (III, IV). The mass (m/z 438.198) corresponding to 4-EtPro formation was observed in reaction of 12 with MypE, MypF and aaProC (V), and from hydrolyzed MP A standard (VI). bIn vitro analysis of MypE (EICs). The 4-hydroxymethylhomoisoleucine (13)(m/z 456.208) was detected after incubation of MypE and homoisoleucine (I), but not with boiled MypE (II). The formation of 4hydroxy-L-Leu (m/z 442.193) after incubation of MypE and L-Leu (III). Trace amount of 4-hydroxy-D-Leu (m/z 442.193) after incubation of MypE and D-Leu (IV). Trace amount of 3-hydroxy-L-allo-Ile (m/z 442.193) after incubation of MypE and L-Ile (V). The formation of 3-hydroxy-L-allo-Ile (m/z 442.193) after incubation of MypE and Lall-Ile (VI). No product was detected with using D-Ile, D-allo-Ile and L-Val as substrate (VII-IX). Article https://doi.org/10.1038/s41467-024-44953-5 Nature Communications | (2024) 15:791 8 thus make MPs promising candidates to be developed as potent and safe anti-TB drugs that are likely not detrimental to human microbiota. The discovery of highly potent antibiotics exhibiting the rare target DnaN, and the unusual biosynthesis of the molecular building blocks sets the stage for continuous efforts into future mycoplanecin development. Methods Sequence similarity network analysis Sequence similarity networks were constructed for all homologs of GriE and GriF found in biosynthetic gene clusters (BGCs) in all species of the phylum Actinobacteria. To this end, we conducted a BLAST (v. 2.9.0)26 search with a threshold e-value of 10–5and filtered for hits where the similarity is found for at least 75% of the sequence of the correspondingprotein.Wehaveretrieved thecorrespondinggenomes or genome fragments and predicted BGCs with antiSMASH (v. 5.1.2)27. Only GriE/GriF homologs within BGCs were retained. The protein sequences werealigned with MAFFT (v. 7.450)28 and pairwise sequence similarities were calculated. The sequence similarity networks were visualized with Cytoscape (v. 3.9.1)29. Strains, plasmids and reagents The mycoplanecins producer strain Actinoplanes awajinensis ATCC 33919 was purchased from the American Type Culture Collection. The plasmid pET28b was purchased from Novagen. The plasmid pColdI was purchased from TAKARA. CoA esters and amino acids were purchased from Sigma-Aldrich and Enamine. The 4-ethylproline was synthesized and supplied by Sanofi. Reagents used for cultivation, expression, catalyzation, derivatization and all enzymes required for molecular cloning were purchased from ThermoFisher Scientific, NEB, Sigma-Aldrich or Chemspace. Strain cultivation, compound extraction and purification Cultivation of Actinoplanes awajinensis was carried out in 5 ×1-L flasks using 300mL culture volume in APL medium (1% soluble starch, 0.2% yeast extract, 1% glucose (dextrose), 1% glycerol, 0.25% corn steep liquor, 0.2% peptone (meat), 0.1% NaCl, 0.3% CaCO 3 ,adjustedtopH 7.2). Subsequent supercritical fluid extraction (SFE) was carried out followingtheprotocoldescribedbyBaderetal.30 usinga WatersMV-10 ASFE system and yielding the mycoplanecins in the ethyl acetate phase. The resulting SF extract was used to purify MP A, B, D, and E using a Dionex Ultimate 3000 SL system comprising a SWPS 3000 SL autosampler, P680pump, TCC100 columnoven, PDA100 UVdetector andAFL3000 fractioncollector incombinationwith a XBridgePeptide BEH C18 OBD prep column (130 Å, 5 µm, 10 mm × 250mm). The flowrate was set to 5 mL/min with a mobile phase containing (A) 0.1% formic acid in ddH 2 O and (B) 0.1% formic acid in acetonitrile heated to 45°C. After an equilibration phase for 2min at 5% B in A, a 30 min gradient from 50 to 80% B was adapted for separation of the MP derivatives. The column was flushed with 95% B for 2 min before subsequent re-equilibration at 5% B for 5 min. NMR analysis for planar structure elucidation NMRmeasurementsofMPA,B,D,andEwererecordedusingaBruker UltraShield 500 or a Bruker Ascend 700 spectrometer equipped with a 5mm TCI cryoprobe ( 1Hat500or700MHz,13Cat125or175MHz, respectively). 1H, 13C, COSY, TOCSY, HSQC and HMBC experiments were carried out in deuterated chloroform. The NMR Spectra were recorded by Bruker TopSpin 4 and processed by ACD/Labs 2021.2.0. Marfey’s analysis for stereochemical assignment Marfey’s derivatization of MP A, B, D, and E was performed to assign the amino acids’stereochemistry by comparing the retention time of the hydrolyzed and derivatized amino acids with respective standards. Hydrolysis of the amide bonds and subsequent derivatization of the fragments was carried out as follows: 0.1 mg of the respective MP derivative was hydrolyzed with 100 µL of 6 N HCl in a closed glass vial for 45 min at 110 °C under nitrogen. The solvent was dried at 110 °C and the residue was dissolved in 110 µLH 2 O. The solution was split into two 50 µL aliquots in PP tubes and both were treated with 20 µL of 1 N NaHCO 3 followed by 20 µLof1%1-fluroro-2,4-dinitrophenyl-5-leucine-amide (D-FDLA and L-FDLA in acetone). After incubation for 2 h at 40 °C and 700 rpm, the reaction was stopped by addition of 10 µL of 2 N HCl and 300 µL of acetonitrile. Both samples were centrifuged for 10 min at 15 °C and 21,500 × g, and the supernatant was measured by UHPLC-HRMS. Analysis was carried out on a Dionex Ultimate 3000 SL system coupled to a Bruker Daltonics maXis 4 G UHR-TOF applying electrospray ionization (ESI). Separation was performed using a Waters Acquity BEH C18 column (100 × 2.1 mm, 1.7 μm) with a mobile phase consisting of (A) 0.1 % formic acid in ddH 2 O and (B) 0.1 % formic acid in acetonitrile with a flow rate of 0.6 mL/min at 45 °C. A multi-step gradient was applied to separate the derivatized amino acids: 5-10 % B in 1 min, 10-35 % B in 14 min, 35–55% in 7 min, 55–80% in 3 min, hold at 80% for 1 min and re-equilibration at 5 % B for 5 min. Before entering the mass spectrometer, the flow was split to 75 µL/min. UV data were obtained at 340 nm simultaneously with the MS detection in centroid mode ranging from 250 to 3000 m/z. Standard LC-MS methods UHPLC-HRMS measurements for determination of exact mass, sum formula and retention time were carried out with a Dionex Ultimate 3000 SL system consisting of SWPS 3000 SL autosampler, P680 pump module, TCC100 column oven, and PDA100 UV-detector coupled to a maXis 4 G UHR-TOF platform. The ionization mode was ESI with the following MS settings: capillary voltage 4000 V, end plate off-set -500 V, nebulizer gas pressure 1 bar, dry gas flow rate 5 L/min, dry gas temperature 200°C, mass scan range 150–2500 m/z. The mass spectrometer was externally calibrated to sodium formate cluster using subsequent lock mass calibration. Metabolite profiling was performed using Waters Acquity BEH C18 columns (50 × 2.1 mm, 1.7 µmand 100 × 2.1 mm, 1.7 µm) connected to a Waters VanGuard BEH C18 precolumn. The standard methods for both columns are linear gradients with 5–95% ACN+ 0.1% FA over 6, 9, or 18 min. The LC-MS results were analyzed by Bruker Compass DataAnalysis V4.4. MIC determination To determine the inhibitory activity of the mycoplanecins, Minimum Inhibitory concentration (MIC) was carried out. M. tuberculosis (Mtb) strain H37Ra and H37Rv, M. marinum (Mm)strainM (ATCC BAA-535) and M. smegmatis (Msmeg) strain mc2155 were cultured in 7H9 complete medium (BD Difco; Becton Dickinson) supplemented with oleic acid-albumin dextrose-catalase (OADC, 10%; BD), 0.4% glycerol, and 0.05% Tween80 at 37 °C for Mtb and Msmeg and at 30 °C for Mm. At mid-log phase (OD 600 between 0.4 and 0.8), cultures were harvested and centrifuged (3700 × g, 10 min). For Mtb and Mm, bacterial cells were then thoroughly resuspended in 7H9 medium (10% OADC) in the absence of glycerol and Tween80 by use of a syringe and a 26-gauge syringe needle. Mtb and Mm single-cell PBS stocks were thawed and used to prepare a final suspension of 105CFU/mL. Compounds were tested in twofold serial dilutions in final 1% DMSO. Mtb MIC values were determined by addition of 50 µL of Alamar Blue (Promega) with 5% v/v Tween80 to each well on the sixth day of incubation followed by 16–24 h of incubation at 37 °C. Fluorescence was measured (excitation at 570 nm and emission at 590 nm). Mtb MIC was defined as the lowest concentration affecting an 80% reduction in fluorescence relative to the signal for the control without antibiotic. For Mm and Msmeg, MIC was determined visually as the lowest concentration where no visible growth was observed. Article https://doi.org/10.1038/s41467-024-44953-5 Nature Communications | (2024) 15:791 9