Peptide surfactants with post-translational C-methylations that promote bacterial development

Abstract

Bacteria produce a variety of peptides to mediate nutrient acquisition, microbial interactions and other physiological processes. Of special interest are surface-active peptides that aid in growth and development. Herein we report the structure and characterization of clavusporins, unusual and hydrophobic ribosomal peptides with multiple C-methylations at unactivated carbon centers, which help drastically reduce the surface tension of water and thereby aid in Streptomyces development. The peptides are synthesized by a previously uncharacterized protein superfamily, termed DUF5825, in conjunction with a vitamin B₁₂-dependent radical S-adenosylmethionine metalloenzyme. The operon encoding clavusporins is widespread among actinomycete bacteria, suggesting a prevalent role for clavusporins as morphogens in erecting aerial hyphae and thereby advancing sporulation and proliferation.

Natural products are small organic molecules synthesized by a wide array of organisms for diverse purposes¹. Their functions fall into severalbroad categories, many involving chemical warfare or otherwise competitive or collaborative associations^1–5. An interesting class is those that physically interact with the environment. Siderophores, for example, mine iron from insoluble sources and deliver the metal to the host⁶, while osmolytes enhance the ionic strength of water⁷. Another intriguing class is surfactants, which can alter the surface tension of water and allow microbes to raise aerial structures, among other processes^8,9. Surfactants are especially important for the development of microbes, notably in streptomycetes and filamentous fungi, which produce aerial hyphae^10–15. These structures develop into spores that, upon dispersion and germination, allow the respective life cycles to start anew⁴. Sporulation is a highly complex and coordinated process, and the small molecules underlying it remain underexplored. The recent accumulation of genome sequencing data suggests that many natural product groups remain to be discovered, including those involved in microbial development¹⁶.

SapB is a ribosomally synthesized and post-translationally modified peptide (RiPP) in the lanthipeptide family and the only well-known surface-active natural product in Streptomyces development^17–19. It is absent or highly divergent in many actinomycete genomes, suggesting that alternative surfactants may be involved in this process. We became interested in the involvement of surfactant peptides in Streptomyces clavuligerus, the producer of clavulanic acid, as it codes for a score of uncharacterized biosynthetic gene clusters, notably a RiPP cluster, the expression of which requires a rare leucine tRNA. This tRNA is encoded by bldA and its presence is a marker for genes that have functions related to aerial hyphae formation and sporulation^10–13. The RiPP cluster is found on the megaplasmid pSCL4 in S. clavuligerus and prior studies have shown that removal of this plasmid leads to a sporulation-defective phenotype; however, the underlying molecular basis has remained elusive²⁰. We have herein investigated the product of this RiPP gene cluster and found it to be a new family of peptides that are synthesized by a highly unusual mechanism. They are encoded in diverse actinomycete genomes and effectively reduce the surface tension of water, thus acting as prevalent biosurfactants that advance microbial development.

Results

An unusual RiPP operon in S. clavuligerus

The S. clavuligerus genome contains a RiPP gene cluster that we have termed mpc (methylated peptides in S. clavuligerus; Fig. 1a). It encodes a transporter (mpcT), an S9 peptidase (mpcP), a 39mer precursor peptide (mpcA), a vitamin B₁₂-dependent radical S-adenosylmethionine (rSAM) enzyme (mpcB) and a domain of unknown function (DUF5825, mpcC). The cluster is notable for the presence of a rare TTA leucine codon in mpcT, which suggests involvement in morphological development, as well as the rSAM enzyme MpcB, and the DUF5825 MpcC. The hypothetical protein MpcC has no sequence similarity with known proteins; members of DUF5825 have not yet been characterized. MpcC does show high structural similarity with the C terminus of the rSAM domain and the putative RiPP recognition element (RRE) of MpcB by HHPred analysis, implying functional relevance between these two proteins (Supplementary Fig. 1; ref. 21). Radical SAM enzymes utilize a [4Fe–4S] cluster to generate a highly reactive 5′-deoxyadenosyl radical (5′-dA·), a radical initiation process that underlies many unusual transformations in biology^22–24. A subfamily of rSAM enzymes binds B₁₂. These so-called class B rSAM methyltransferases can methylate unactivated sp³ carbons^25–27. Several RiPPs have been reported with methyl groups installed by B₁₂–rSAM enzymes, including polytheonamides, bottromycins and pheganomycin^28–32. Unlike canonical rSAM enzymes that share a conserved CX₃CX₂C motif for [4Fe–4S] cluster binding, MpcB contains a CX₇CX₂C binding sequence, which was shown to be important for catalysis in the noncanonical rSAM enzyme PoyC³³.

Fig. 1 |. Discovery and structure of clavusporins.

a, The mpc gene cluster as identified by genome mining. The gene locus of mpcA is marked by a black circle, and its protein sequence is shown. Strategies for genetic activation are shown below, including promoter exchange and codon optimization. b, MALDI-TOF MS analysis of SDS extracts of S. clavuligerus strains indicated. The experiment was repeated with at least three biological replicates, and representative results are shown. c, Numbering and parts of the MpcA peptide. d,e, MS/MS profiles of clavusporin A (d) and B (e). Blue and red arrows represent observed b and y ions, respectively. Each asterisk represents methylation on the corresponding residue. f, Structure of clavusporins. Post-translational methylations are marked in red. The L-configured α-carbons are proposed based on the ribosomal origin of the clavusporins.

To circumvent any transcriptional regulation of the mpc operon and identify its product, we engineered an S. clavuligerus strain, wherein the native promoter P_mpc was replaced with a Streptomyces constitutive promoter P_ermE* and the first 33 basepairs of mpcT were codon-optimized, including a switch from the rare Leu codon (TTA) toa common one (CAG; Fig. 1a, Supplementary Fig. 2 and Supplementary Table 1; Methods)³⁴. However, we were unable to identify methylated peptides from culture supernatants or organic extract of cell pellets of the engineered strain via untargeted peptidomic analysis. We reasoned that the potential methylations catalyzed by MpcB could drastically elevate the hydrophobicity of the peptide product(s), which could become recalcitrant to solvent extraction. Inspired by the discovery of SapB, the cell pellet of S. clavuligerus P_mpc::P_ermE* was extracted with a 1% sodium dodecyl sulfate (SDS) solution, and the extracts analyzed by matrix-assisted laser desorption/ionization time-of-flight (MALDI-TOF) mass spectrometry³⁵. Two signals (m/z 1,512.9 and 1,526.9) were observed exclusively in the engineered strain but not WT (Fig. 1b and Supplementary Fig. 3). The 14 Da difference pointed to additional methylation, further confirming the hypothesis that the gene cluster produces methylated peptides. Genetic deletion of either mpcB or mpcC in S. clavuligerus P_mpc::P_ermE* abolished the production of these two signals (Fig. 1b and Supplementary Figs. 2 and 3), thus linking their production to the mpc locus. The two products match 13mer MpcA peptides (KPSVGITITVPFR; Fig. 1c) with seven or eight methylations; we have named these clavusporin A and B, respectively. MALDI–tandem mass spectrometry (MS/MS) analysis revealed the methylation patterns (Fig. 1d,e, Supplementary Fig. 4 and Supplementary Table 2), consisting of bismethylated Pro2 and Pro11, as well as monomethylated Val4, Ile6 and Ile8 on both peptides. Clavusporin B was additionally monomethylated at Thr7.

Extraction of the peptides into acidic ethyl acetate facilitated analysis by ultra-performance liquid chromatography (UPLC)-coupled high-resolution MS (HRMS), the results of which were consistent with the assignment of hypermethylated 13mer peptides above (Extended Data Fig. 1a,b). To unequivocally determine the sites of methylations, we isolated the clavusporins, comprised of a mixture of clavusporin A and B, using a combination of unusual chromatographic methods and subjected them to multidimensional NMR spectral analysis, which revealed β-methylation at each modified residue and completed structural elucidation (Fig. 1f, Extended Data Fig. 1c,d, Supplementary Fig. 5 and Supplementary Table 3). Clavusporins are a new family of β-methylated amphiphilic peptides; the occurrence of a series of β-methylated and β-dimethylated residues is rare in RiPP-natural products^19,29,36.

Surface-active clavusporins promote bacterial development

The structures of clavusporins are unique in that the hydrophobic residues are heavily β-methylated, forming a tertiary alcohol on Thr7 and quaternary carbon centers on Pro, Val and Ile residues, while the side chains of Lys and Arg are positively charged under physiological conditions. We hypothesized that clavusporins can function as cationic surfactants in the course of Streptomyces development. To examine this hypothesis, we first measured the ability of clavusporins to reduce the surface tension of water using a contact angle goniometer. At a concentration of 1 μg μl⁻¹, clavusporins substantially altered the shape of aqueous droplets by reducing the surface tension to 31 mN m⁻¹, similar to those of saturated SDS or SapB solutions (Fig. 2a–c)^14,37. At lower concentrations, water surface tension decreased linearly with clavusporin titers (Fig. 2c, red line). Between 1.0 and 1.5 μg μl⁻¹, no further decreases were observed, indicating a critical micelle concentration of clavusporins at around 1 μg μl⁻¹. To test the effects of post-translational methylations, the unmodified 13mer core peptide, MpcA_1–13, was synthesized (Methods; Supplementary Fig. 6 and Supplementary Table 4) and subjected to surface tension analysis (Fig. 2c, blue line). At 1 μg μl⁻¹, MpcA_1–13 only lowered surface tension to 57 mN m⁻¹, thus acting as a weaker surfactant and indicating that the methylations substantially affect the surfactant properties of clavusporins.

Fig. 2 |. Clavusporins reduce surface tension and promote aerial hyphae growth.

a,b, Images of a water droplet consisting of 10% DMSO (a; vehicle control) and a droplet of clavusporins at 1 μg μl⁻¹ in a 10% DMSO solution (b). c, Quantification of surface-tension reduction by clavusporins (red) and the unmethylated 13mer, MpcA_1–13 (blue). Mean ± s.d., n = 3 per condition, two-way ANOVA with Sidak’s multiple comparison test; *P = 0.0152, **P = 0.0080 and ****P < 0.0001. Surface tension is linearly dependent on compound concentration. The lowest surface tension occurs at ~1 μg μl⁻¹ clavusporins. d, Chemical complementation of S. clavuligerus ΔmpcB with DMSO control, MpcA_1–13 or clavusporins (10 μg dissolved in 2 μl of DMSO). e–g, SEM images of cell surfaces supplemented with DMSO control (e), MpcA_1–13 (f) and clavusporins (g). Each experiment was repeated with at least three biological replicates, and representative results are shown.

We next assessed the effect of clavusporins on sporulation by generating gene inactivation mutants via replacement of mpcB or mpcC with an apramycin resistance cassette (apr^R; Supplementary Fig. 7). The mutants were unable to sporulate, growing only substrate mycelia and limited aerial hyphae on agar, whereas WT cells developed aerial hyphae and spores under the same conditions (Extended Data Fig. 2a–c). These results are in line with prior reports of the developmental deficiency of S. clavuligerus mutants lacking the megaplasmid pSCL4 (ref. 20). The phenotype differences between WT and ΔmpcB and ΔmpcC were verified by scanning electron microscopy (SEM; Extended Data Fig. 2d–f). The WT showed ‘club’-like spore clusters, for which the strain has been named³⁸; these were lacking in the mutants. The inability of the mutants to sporulate precluded genetic complementation. Together, these observations suggest that the mpc operon is involved in S. clavuligerus development through the production of methylated peptides, consistent with the requirement of the rare leucine tRNA, encoded by bldA, for translation of MpcT.

To further test the function of clavusporins, we examined their ability to restore the bald phenotype of the ΔmpcB and ΔmpcC strains. Cells growing on agar surfaces were either chemically supplemented with DMSO (as vehicle control) or with clavusporins or MpcA_1–13. The clavusporin-treated area showed accelerated growth of aerial hyphae, based on visual inspection and SEM imaging, whereas the control did not (Fig. 2d–g). Surprisingly, supplementation with unmodified MpcA_1–13 peptide had a similar effect, which was comparable to that of clavusporin at different doses (Supplementary Fig. 8). However, MpcA_1–13 was not detected in the S. clavuligerus P_mpc::P_ermE* ΔmpcB/ΔmpcC strain (Supplementary Fig. 3), suggesting that mutants lacking MpcB/MpcC are unlikely to accumulate the unmodified core peptide, thus leading to the bald phenotype. We also tested whether other general surfactants could promote the growth of aerial hyphae (Supplementary Fig. 9). Surfactin, a naturally occurring lipopeptide, had no observable effect on cell growth. Cationic (cetrimonium bromide) and anionic (SDS) surfactants only exhibited cytotoxicity at high concentrations, indicating that general surfactants do not complement the bald phenotype and that there is a degree of specificity in the surfactant properties of clavusporin.

We then investigated any regulatory roles for clavusporins by assessing transcript levels of key regulatory genes in response to the absence of the RiPP. To do so, we generated a mutant strain devoid of the entire mpc cluster to eliminate any possible confounding effects of the precursor peptide (Supplementary Table 5). RT-qPCR experiments were then carried out in both WT and the mutant strain monitoring the expression of bldDHN, pivotal in regulating aerial hyphae formation, ramSR, which govern the biosynthesis of the morphogenetic peptide SapB, as well as ssgAB and whiBD, integral to the sporulation process^10–13. No substantial disparity in transcript levels was observed between the WT and mutant strain (Supplementary Fig. 10 and Supplementary Table 6), which suggests that clavusporins do not exert direct regulatory effects but rather serve primarily as surfactants in the development of S. clavuligerus. This hypothesis is also supported by the observation that unmodified core peptide MpcA_1–13, albeit a weaker surfactant, was able to rescue aerial hyphae growth. The post-translational methylations serve to increase the stability of the peptides in vivo and to maximize surfactant activity with limited clavusporin production.

Lastly, we explored the generality of the findings above by exploring the role of the orthologous cluster in Streptomyces ghanaensis (named mpg for methylated peptides in S. ghanaensis). This cluster encodes the same set of genes with MpgA, MpgB, MpgC, MpgP and MpgT proteins that are 77%, 81%, 58%, 63% and 56% homologous to those in S. clavuligerus, respectively (Extended Data Fig. 3a). We generated a mpgB::apr^R insertion in S. ghanaensis (Supplementary Fig. 11 and Supplementary Table 7), which exhibited a white phenotype and altered development by visual inspection and by SEM. The WT, however, developed normally under similar conditions, producing dark gray spores (Extended Data Fig. 3b–e). Together, these results show that the mpc/mpg operons produce peptide surfactants, which lower the surface tension of water and facilitate the erection of aerial hyphae, thus playing a key role in the developmental processes of S. clavuligerus and S. ghanaensis.

Biosynthesis of clavusporins

With the structure and function of clavusporins established, we next explored its biosynthesis, focusing on the introduction of methyl groups at unactivated carbon centers. B₁₂-dependent rSAM enzymes are notoriously difficult to purify, despite advances in cofactor incorporation, protein yields and solubility of this class of enzymes^39,40. We therefore utilized an alternative in vivo approach to coexpress maltose-binding protein (MBP)-tagged MpcA with MpcB in Escherichia coli cells (Fig. 3a and Supplementary Table 8). Plasmids pDB1282 encoding Fe–S cluster assembly proteins and pBAD42-BtuCEDFB to enhance cellular cobalamin availability were also introduced into E. coli^40,41. After expression, the MBP-tagged peptides were enriched through affinity-based chromatography and subsequently trimmed by proteolysis and analyzed by high-performance liquid chromatography (HPLC)-coupled HRMS.

Fig. 3 |. Biosynthesis of clavusporins.

a, Workflow for in vivo reconstitution of peptide methylations by coexpression of MBP-tagged precursor peptide and the modification enzymes. b–g, HPLC-coupled HRMS analysis of MpcA core peptides from MBP-peptide fusions after column-enrichment and trypsin digestion. Full-length MpcA was co-expressed with void control (b), MpcB (c), MpcC (d) and MpcBC (e), and truncated MpcA was co-expressed with with MpcB (f) and MpcBC (g). Shown are extracted ion chromatograms of MpcA_1–13 (core peptide) with or without methylation, as indicated. The traces are offset on the y axis for clarity. Each experiment was repeated with at least three biological replicates, and representative results are shown. h, Proposed biosynthetic pathway of clavusporins.

As expected, only the linear, unmodified MpcA core peptide was observed when MBP-MpcA was expressed in the absence of any modification enzymes (Fig. 3b). Coexpression with MpcB and MpcC, but not with either protein alone, yielded hypermethylated core peptides, indicating that MpcBC acts as a methyltransferase pair, consistent with the results from genetic deletions of mpcB and mpcC above (Fig. 3c–e). We detected not only singly methylated species but also peptide fragments with multiple (up to eight) methylations. The fully methylated peptide contained the same methylation pattern as clavusporin B, as judged by tandem MS analysis (Supplementary Table 9). To probe the sequence of methylation events, the single-methylated species were analyzed by tandem HRMS; the results pointed to a mixture of monomethylated species, suggesting a lack of order of β-methylation by MpcBC.

We next explored the role of the MpcA leader peptide in post-translational modification. The first 17 residues of MpcA were removed from MBP-MpcA, yielding MBP-MpcA_(−1)–21, in which the Arg at the (−1) position was retained for trypsin digestion. The truncated MBP-MpcA_(−1)–21 was coexpressed with MpcB alone or MpcBC following the same procedure. No methylation was detectable in the absence of MpcC (Fig. 3f). Only a small fraction (<10%) of leader-less MpcA was monomethylated in the presence of MpcBC, showing that the reactivity of MpcBC is largely dependent on the MpcA leader peptide (Fig. 3g).

The leader–peptide dependence of MpcBC suggests that methylation on MpcA precedes peptidolysis. Two separate hydrolysis events are required to remove the leader and follower peptides and thus generate the mature product. As both cleavages occur C-terminally to Arg, it is possible that the only peptidase encoded in the BGC, MpcP, performs these reactions. To test this idea, His₆-tagged MpcP and NusA-His₆-tagged MpcA were produced recombinantly in E. coli. Incubation of the purified proteins yielded a single product [M + 3H]³⁺ with m/z of 758.7177, which matched MpcA_1–21 ([M + 3H]³⁺, calc. 758.7137, 5.3 ppm; Extended Data Fig. 4a). MpcP activity is unlikely to be affected by the NusA tag, as trypsin was able to hydrolyze both upstream and downstream of the core peptide (Extended Data Fig. 4b). Moreover, when modified MBP-tagged MpcA obtained from coexpression with MpcBC was treated with MpcP, only removal of the leader peptide, but not the follower, was observed, yielding MpcA_1–21 with 0–8 methylations (Extended Data Fig. 4c). These observations show that MpcP is a site-specific protease that removes the leader peptide from MpcA. We postulate that the acidic follower of MpcA solubilizes the peptide after extensive methylations in the core region. Once the modified MpcA_1–21 peptide(s) are transported outside of the cell by MpcT, a yet-unknown extracellular protease removes the follower. Together, these observations allow us to propose a biosynthetic pathway of clavusporins (Fig. 3h).

Mpc-type BGCs are widespread in actinomycetes

Given the substantial role of the mpc/mpg operon in S. clavuligerus and S. ghanaensis, we explored its prevalence bioinformatically using MpcB (WP_003958123.1) as a query in protein BLAST searches (Fig. 4a). The first 500 nonredundant proteins were subjected to genomic neighborhood analysis using RODEO⁴². MpcB-homologs situated in gene clusters that encode upstream MFS transporters and S9 peptidases as well as downstream DUF5825 proteins were selected to focus on genuine clavusporin-like clusters, leading to the identification of 229 distinct mpc BGCs, all from actinomycetes with identical genomic contexts. It is, therefore, reasonable to assume that these share analogous functions and generate mature products that are similar to clavusporins.

Fig. 4 |. The mpc-type BGCs are widespread in actinomycetes.

a, Workflow for identification of mpc clusters and their putative precursor peptides. b,c, SSN of ‘long’ precursor peptides with acidic follower peptides (b) and ‘short’ precursor peptides without acidic follower peptides (c). Each node represents a unique BGC and the lines connecting them indicate sequence similarity in the precursor peptide (MpcA). The S. clavuligerus MpcA and S. ghanaensis MpgA nodes are labeled in red. d, Logo plot for precursor peptides in each subfamily shown in b and c. The precursor peptides are color-coded to the BGCs in b and c, and the number of representatives for each cluster is shown.

We next identified precursor peptides for these BGCs using three criteria—location of the ORF between MpcP and MpcB, high GC content and a peptide sequence rich in Val, Ile, Thr and/or Pro residues. We were able to find a single precursor peptide for each of the 229 BGCs, except for one, which failed due to poor genome sequencing quality. Most of the peptides resemble MpcA, consisting of leaders, hydrophobic cores and acidic followers. However, we also identified several (22 of 228) short precursor peptides lacking the acidic follower sequence. Coincidently, the MpcA-like ‘long’ peptides are mostly from Streptomyces, with two exceptions out of 206 peptides, while the truncated precursors are mostly found in rare actinomycetes (Supplementary Table 10).

To further group the identified mpc clusters, the 228 precursors were subjected to sequence similarity network (SSN) analysis⁴³. As the algorithm is especially sensitive to sequence length, the long and short precursors were separated into two networks (Fig. 4b,c). This analysis organized the long precursors into four subfamilies with multiple members and four singletons. BGCs containing the short precursor were organized into two subfamilies. For each, a conserved precursor peptide motif was generated using the MEME suite (Fig. 4d). The majority of the long peptides (169 of 206) cluster with MpcA, featuring two conserved Arg residues for peptidase recognition, and two conserved Pro residues for dimethylations. Although the number of Pro residues in the putative core regions varies among the long peptides, all peptides have the first Arg residue for MpcP-type processing. In contrast, the second Arg, Arg13, in MpcA is not conserved, again pointing to a separate enzyme for follower removal. For the short precursors, the Arg residue upstream of the hydrophobic core is conserved, while Pro is no longer present in the core peptides. Instead, the short peptides are rich in Val/Ile residues, with one group having an alternating pattern of Val/Ile, and another group bearing seven consecutive Val/Ile residues. These analyses annotate >200 BGCs for which the underlying chemistry, biosynthesis, enzymology and function can now be explored.

Discussion

It is by now near trivial to find uncharacterized BGCs in microbial genomes^16,44. The much harder proposition is identifying the products of these clusters, and clavusporins demonstrate the challenges associated with this process as we needed to perform a genetic promoter swap, codon optimization and unusual extraction protocols to uncover the product of the mpc cluster. The structure and biosynthesis of clavusporins are highly unusual. The latter not only revealed MpcBC as a new iterative β-methyltransferase pair but also highlighted both the promiscuity and selectivity of the reaction^28–31,33. On one hand, MpcBC can recognize four different residues, but on the other hand, it consistently methylates at the β-carbon. MpcC may serve as a scaffolding domain, perhaps assisting in substrate binding as previously described for RRE proteins⁴⁵. However, MpcB has its own intrinsic RRE domain and MpcC also shares structural similarity with the rSAM domain of MpcB, suggesting an unknown protein–protein interaction that requires further biochemical characterizations (Supplementary Fig. 1). The reconstitution of MpcBC catalysis in E. coli now allows further functional studies with this catalytic pair.

To the best of our knowledge, clavusporins join polytheonamides, the related pythonamides and bottromycins as heavily C-methylated RiPP-natural products^30,36,46,47. A common approach used by nature to reduce the polarity of peptides is attachment of fatty acyl groups at the peptide N terminus⁴⁸. Installation of a bevy of individual methyl groups on amino acid side chains, as seen in clavusporins and polytheonamide, is an alternative strategy to elevate hydrophobicity. Structure and function go hand-in-hand, and the amphiphilic structure of clavusporins is well-suited for its function in lowering the surface tension of water and allowing bacteria to raise aerial structures. It has long been known that removal of the large plasmid in S. clavuligerus leads to a bald phenotype. Our results provide a mechanistic rationale for this observation. Clavusporins, if immobilized with a certain conformation, could be more effective as surfactants compared to free molecules in the droplet tests. General surface-active agents do not complement the absence of clavusporins, suggesting specificity in the function of this hypermethylated surfactant. The exact mechanism of deployment of clavusporins, whether they have specific molecular targets on the cell surface, and whether they assemble on the cell wall and interact with chaplins and rodlins remain to be determined. Sporulation is a complex and coordinated process involving a number of regulators, notably the autoinducer ɣ-butyrolactone⁴. In this regard, how clavusporins connect with other regulatory pathways and known processes in sporulation provides a rich area for future research. It is also highly likely that yet-unknown natural products act as surfactants in sporulation, including those in the family of >200 mpc-like BGCs that we annotate, and these offer intriguing avenues for further study at the intersection of microbial development, natural product chemistry and enzymology.

Online content

Any methods, additional references, Nature Portfolio reporting summaries, source data, extended data, supplementary information, acknowledgements, peer review information; details of author contributions and competing interests; and statements of data and code availability are available at https://doi.org/10.1038/s41589-025-01882-8.

Methods

Bacteria strains and growth

S. clavuligerus ATCC 27064 and S. ghanaensis ATCC 14672 were obtained from the American Type Culture Collection (ATCC). S. clavuligerus and S. ghanaensis strains were cultured in TSB liquid medium (3% (wt/vol) tryptone soy broth) for general purposes. S. clavuligerus cultures were plated on GYM agar for sporulation (0.4% (wt/vol) d-glucose, 0.4% (wt/vol) yeast extract, 1% (wt/vol) malt extract, 0.2% (wt/vol) calcium carbonate, 1.2% (wt/vol) agarose, pH = 7.2). S. ghanaensis strains were cultured on soy flour–mannitol (SFM) agar plates for sporulation. SFM agar medium was prepared by dissolving 2% (w/v) soybean meal (NutriSoy soy flour) in tap water, followed by autoclave sterilization and filtration through cheesecloth. The filtrate was then supplemented with 2% (wt/vol) mannitol and 2% (wt/vol) agar, and autoclaved again. For the production of clavusporins, the engineered S. clavuligerus strain was cultured in MYM medium (0.4% (wt/vol) d-maltose, 0.4% (wt/vol) yeast extract, 1% (wt/vol) malt extract). E. coli strains were cultured in Luria–Bertani (LB) medium or LB agar with the appropriate antibiotics. Antibiotics used were ampicillin (Amp; 100 μg ml⁻¹), apramycin (Apr; 50 μg ml⁻¹), kanamycin (Kan; 50 μg ml⁻¹), chloramphenicol (Cm; 25 μg ml⁻¹), thiostrepton (Tsr; 20 μg ml⁻¹) and spectinomycin (Spec; 100 μg ml⁻¹). All media components, unless otherwise mentioned, were purchased from Beckton Dickinson. All antibiotics were obtained from Sigma-Aldrich.

Genetic manipulation of Streptomyces by conjugation

Plasmid pJTU1289 (Amp^R, Tsr^R) was used for the gene knockouts (aprR insertion) of S. clavuligerus and S. ghanaensis. Plasmid pKC1139 (Apr^R) with a temperature-sensitive origin of replication was used for the marker-less promoter exchange for S. clavuligerus. In brief, ~2 kb fragments upstream and downstream of the targeted region were cloned, and inserted into the knockout plasmid, flanking the desired insertion sequence (for example, antibiotic marker). All restriction enzymes and T4 ligase were purchased from New England Biolabs. Constructions of the knockout plasmids were completed in E. coli DH5α cells (Supplementary Note), and chemically transformed into methylation-deficient E. coli ET12567 (Cm^R) carrying pUZ8002 (Kan^R). The donor E. coli strain was then cultured in 25 ml of LB medium supplemented with antibiotics to an OD₆₀₀ of 0.5–0.7. Cells were collected by centrifugation (4,000g, 10 min), washed with antibiotic-free LB medium (10 ml, twice) and resuspended in LB medium for later use. Streptomyces spores were collected from agar plates by scratching the cell surface with cotton swabs and suspended in water. Excess mycelial cells were removed through crude cotton filtration. Spores remaining in the filtrate were then spun down by centrifugation, washed with 1 ml of N-tris(hydroxymethyl)methyl-2-aminoethanesulfonic acid (TES) buffer (50 mM TES, pH = 8.0) and then resuspended in 500 μl of TES buffer. Spores were heat-shocked for 10 min in a 50 °C water bath, subsequently incubated on ice for 5 min, and then mixed with 500 μl of 2× spore-activation medium (1% (wt/vol) yeast extract and 1% (wt/vol) casein hydrolysate) and calcium chloride (5 mM, final concentration). The spores were then incubated at 37 °C, 200 rpm for germination for 2–3 h, then collected by centrifugation, and resuspended in LB medium. Spore suspensions were mixed thoroughly with E. coli (donor/acceptor 1:1 to 2:1), plated on selected agar plates, and incubated at 30 °C for 16–18 h. Exconjugants were selected by plating Apr and trimethoprim (50 μg ml⁻¹) onto plates, and colonies emerged on plates after incubation at 30 °C for 2–4 days. For gene knockouts, double-crossover colonies that were sensitive to Tsr, but resistant to Apr were selected and verified by PCR. For generating the S. clavuligerus-engineered strain, exconjugants were first cultured in TSB with Apr at 30 °C, then without Apr in TSB at 37 °C for plasmid integration, and then selected with Apr again at 30 °C to obtain single-crossover cells. The single-crossover cells were then plated on antibiotic-free GYM plates for single-colony selection. Colonies sensitive to Apr were picked and tested with PCR for successful promoter replacement.

SEM

Streptomyces cells grown on agar plates were cut into cubes, sliced into a thin layer and then transferred to aluminum SEM stubs (Electron Microscopy Sciences, 75220) covered with double-sided carbon tape. Samples were coated with a 3–5 nm layer of gold on a VCR IBS/TM 200S ion beam sputterer and imaged on a Quanta 200 FEG environmental-SEM in high vacuum mode.

Detection of clavusporins from S. clavuligerus P_mpc::P_ermE*

S. clavuligerus strains were cultured in TSB medium for 2 days at 30 °C per 200 rpm, and then inoculated in a 125 ml Erlenmeyer flask containing 25 ml of MYM medium (0.02% wt/vol), supplemented with CoCl₂·6H₂O (2 μg ml⁻¹, final concentration). Cells were then cultured at 30 °C, 250 rpm for 6 days. After growth, cells were pelleted by centrifugation at 4,000g for 20 min, resuspended in 15 ml of 1% (wt/vol) SDS solution and heated for 1 h in a 55 °C water bath. The supernatants from SDS extraction were analyzed by MALDI-TOF mass spectrometry. Samples were mixed with matrix (α-cyano-4-hydroxycinnamic acid in 50% MeCN in H₂O + 0.1% trifluoroacetic acid (TFA), 10 mg ml⁻¹) in 1:1 ratio and spotted on a ground steel 384-well target plate. MALDI–MS analysis was performed on a Bruker UltrafleXtreme TOF/TOF mass spectrometry in reflector/positive mode. Data were collected using FlexControl software and extracted by FlexAnalysis software. For UPLC-coupled HRMS analysis, supernatants from SDS extractions were acidified with 1% (vol/vol) TFA and then extracted with ethyl acetate (1:1) twice. The organic phase was then combined, dried in vacuo, resuspended in MeOH and subjected to UPLC–MS analysis performed on an Agilent 6546 Accurate Mass Quadrupole Time-of-Flight (Q-TOF) with a 1290 Infinity II Series LC system. LC separation was carried out on an Eclipse Plus C18 column (Agilent, 1.8 μm, 2.1 × 50 mm), with H₂O and MeCN as mobile phases (+0.1% formic acid) at a flow rate of 0.5 ml min⁻¹. The LC method started isocratically at 10% MeCN for 0.5 min, followed by a gradient from 10% to 100% MeCN over 4.5 min, and then finished with 3 min of isocratic 100% MeCN. Data were collected using the Agilent MassHunter Workstation Data Acquisition 10.0 software and extracted ion chromatograms were obtained using Agilent MassHunter Workstation Qualitative Analysis 10.0.

Isolation of clavusporins

S. clavuligerus P_mpc::P_ermE* spore stock was inoculated into a 14-ml culture tube containing 3 ml TSB medium, and cultured at 30 °C, 250 rpm for 2 days. The culture was then diluted 1:100 into two 125-ml Erlenmeyer flasks containing 25 ml of TSB medium and cultured at 30 °C, 250 rpm for 2 days. The resulting cultures were used for large-scale growth. Cells were collected by centrifugation at 4,000g, resuspended in MYM medium and inoculated into 8 × 2.8 l baffled Fernbach flasks, each containing 1 l of MYM medium (0.02% wt/vol), supplemented with CoCl₂·6H₂O (2 μg ml⁻¹, final concentration). After growth at 30 °C, 200 rpm for 6 days, cells were spun down at 9,500g, and transferred to 8 × 1 l Erlenmeyer flasks containing 600 ml of 1% SDS solution. Cell suspensions in SDS solution were warmed for 1 h in a 55 °C water bath with occasional shaking, and then stored at 4 °C overnight for the precipitation of insoluble proteins. The cooled cell suspensions were then centrifuged at 9,500g for 90 min. The subsequent cell-free supernatants were acidified with 1% (vol/vol) TFA and extracted with EtOAc (1:1) twice. Liquids from the organic phase were combined, treated with sodium sulfate and concentrated by rotary evaporation. The concentrated EtOAc extract was an oily dark brown substance; it was lyophilized to remove residual water and TFA. The resulting material was resuspended with a solution of dichloromethane (DCM)/MeOH (50:1) and loaded onto a normal phase silica column (Thermo Fisher Scientific, 230–400 mesh, grade 60). The column was eluted with DCM/MeOH 50:1, DCM/MeOH 20:1, DCM/MeOH 10:1, DCM/MeOH 5:1, DCM/MeOH 1:1, MeOH, MeOH/H₂O 4:1 and MeOH/H₂O 2:1. Clavusporins were detected in the last three fractions, which were then combined, and dried in vacuo. The material was then resuspended in 70% MeCN in H₂O (+0.1% TFA), and further purified on an Agilent analytical HPLC system, equipped with a 1260 Infinity pump, an automatic liquid sampler, a temperature-controlled column compartment, a photodiode array detector and an automated fraction collector. The material was repeatedly injected into an analytical Eclipse XDB-C8 column (Agilent, 5 μm, 4.6 × 150 mm), operating at a flow rate of 0.5 ml min⁻¹ with a mobile phase of H₂O and MeCN (+0.1% TFA), which was controlled by Agilent Openlab software version A.0105. Elution was carried out isocratically at 25% MeCN for 5 min, followed by a gradient of 25–50% MeCN over 25 min, a gradient of 50–100% MeCN in 3 min and lastly an isocratic wash of 100% MeCN for 12 min. Fractions containing clavusporins (20–26 min), as detected by MALDI-TOF MS, were pooled, dried and subjected to NMR analysis.

NMR analysis

Samples were dissolved in 120 μl of DMSO-d₆ and transferred to 2.5-mm diameter tubes. Spectra were collected at the Princeton Chemistry NMR Core Facility on a Bruker Acance III HD 800 MHz spectrometer equipped with a triple resonance probe, which was operated using the Bruker TopSpin software v.3.7.0. NMR data were processed and analyzed in MestReNOVA 12.0 software.

Surface tension measurement

Compounds were dissolved in 10% DMSO (in water) for surface tension analysis on a Rame-Hart contact angle goniometer (Model 90-U3-Pro). Photos of hanging droplets (7–10 μl) were taken and analyzed by the DROPimagePro software using a surface tension tool. For each concentration, three technical replicates were performed. Data were processed in Origin software.

Chemical complementation of clavusporins and other surfactants

S. clavuligerus ΔmpcB and ΔmpcC knockout strains were cultured in TSB medium at 30 °C, 250 rpm for 2 days, and then plated onto GYM agar plates to form an even cell lawn. Cells were first incubated at 30 °C for 2 days to enable the growth of substrate mycelium. Then, 2 μl of compound solutions in DMSO at various concentrations was applied to the surface of growing cells. Aerial hyphae emerged after 1–2 days of chemical complementation. Surfactin (S3523) and SDS (L3771) were obtained from Sigma-Aldrich. Cetrimonium bromide (227160100) was purchased from Thermo Fisher Scientific.

Reconstitution of the MpcBC reaction in E. coli

Plasmid pRSF-Duet (Kan^R) was used for heterologous expression of precursor peptides and the corresponding MpcBC methyltransferase in E. coli. Briefly, MBP-fused precursor peptide was cloned into the first multiple cloning site, and modifying enzymes were cloned into the second multiple cloning site, which were both downstream of T7 promoters (Supplementary Material). Expression plasmids were then cotransformed with pDB1282 (Amp^R) and pBAD42-btuCEDFB (Spec^R) into E. coli BL21(DE3) by electroporation. Colonies from the transformation plates were cultured in LB medium (Kan, Amp and Spec). Overnight cells from LB cultures were then inoculated into a 125 ml Erlenmeyer flask containing 80 ml of M9 salt–ethanolamine medium⁴¹ (Kan, Amp and Spec) to a final OD₆₀₀ of 0.02, and cultured in the dark at 37 °C, 180 rpm. Cells were induced with 0.2% (wt/vol) L-arabinose (final concentration) at an OD₆₀₀ ~0.4 and induced with isopropyl β-D-1-thiogalactopyranoside (IPTG) at an OD₆₀₀ of 0.6–0.7 with a final concentration of 0.4 mM. Iron(III) chloride and L-cysteine hydrochloride were supplemented with both L-arabinose and IPTG induction, with final concentrations of 25 μM and 150 μM, respectively. After the addition of IPTG, the cells were cultured in the dark at 37 °C, 120 rpm for 20 h. The cells were then harvested by centrifugation at 4 °C (4,000g, 10 min), and resuspended in 10 ml of lysis buffer A, which consisted of 25 mM Tris base, 300 mM NaCl, 10 mM imidazole, 10% glycerol, pH = 7.7. The cell suspension was lysed by sonication (Thermo Fisher Scientific, FB505 sonicator) on ice for 50 s in 5-s on/15-s off cycles at 30% power. Two milliliters of the cell lysate was then centrifuged at 4 °C, 17,500g for 10 min, and the supernatant was loaded onto a pre-equilibrated mini-Ni column (Thermo Fisher Scientific, HisPur Ni-NTA Spin Column, 0.2 ml). The Ni column was then washed with 1 ml of buffer A, 1 ml of wash buffer B (25 mM Tris base, 300 mM NaCl, 50 mM imidazole, 10% glycerol, pH = 7.7), and then eluted with 1 ml of Elution Buffer C (25 mM Tris base, 300 mM NaCl, 300 mM imidazole, 10% glycerol, pH = 7.7). Eluates from the Ni column were combined and loaded onto a desalting column (Cytiva, PD MiniTrap G-10) to buffer exchange into storage buffer D, which consisted of 25 mM Tris base, 150 mM NaCl, 10% glycerol, pH = 7.7, following the manufacturer’s instruction. Enriched MBP-tagged peptides (100 μl) were subsequently treated with trypsin (~0.5 μg) in the presence of CaCl₂ (20 mM) at 37 °C overnight. The digested peptide mixture was quenched at 95 °C for 5 min, diluted with 100 μl of MeOH, and subjected to HPLC–MS analysis, which was performed on a 6540 UHD Accurate Mass Q-tof HPLC–MS system (Agilent), consisting of a 1260 Infinity Series HPLC, an automated liquid sampler, a diode array detector, and a JetStream ESI source, using an analytical Jupiter C18 column (Phenomenex, 5 μm, 300 Å, 4.6 × 150 mm) with a gradient of 5% MeCN in H₂O (+0.1% formic acid) to 95% MeCN in H₂O (+0.1% formic acid) over 15 min.

Protein overexpression and purification

Genes mpcA and mpcP were amplified from S. clavuligerus genomic DNA and cloned into the BamHI/XhoI site of plasmid pIJ-NusA (Amp^R) and the NdeI/XhoI site of pET28b (Kan^R) for overexpression of NusA-MpcA and His₆-MpcP, respectively (Supplementary Table 8). The plasmids were then transformed into E. coli BL21 (DE3) via chemical transformation. Colonies from the transformation plates were inoculated into LB medium (with antibiotics), and then transferred to a 125 ml Erlenmeyer flask containing 25 ml of LB as seed culture. After 16–18 h, cells from the 25-ml culture were inoculated into a 4 l Erlenmeyer flask containing 800 ml of LB medium with an initial OD₆₀₀ of 0.02, and cultured at 37 °C, 180 rpm. Cultures were induced with IPTG (0.1 mM, final concentration) at an OD₆₀₀ of 0.4–0.6, and then cultured at 18 °C, 180 rpm for 18 h. The cells were pelleted by centrifugation at 4 °C (4,000g, 20 min) and resuspended in ~60 ml of buffer A. These were lysed by sonication on ice for 5 min in 5-s on/15-s off cycles at 35% power. The cell lysate was then centrifuged at 4 °C, 16,900g for 60 min, and the supernatant was loaded onto a Ni-affinity column (5 ml), which was pre-equilibrated with 50 ml of buffer A. The column was washed with 50 ml of lysis buffer A and subsequently 50 ml of wash buffer B. Overexpressed protein was then eluted with elution buffer C, and fractions containing the target proteins, as analyzed by SDS–PAGE gel, were combined and buffer-exchanged into storage buffer D using ultra-filters (Amicon). Concentrated protein solutions were then aliquoted and stored at −80 °C.

MpcP assay

Typical MpcP reactions were performed in 100 μl of storage buffer D with ~20 μM of purified MpcP and ~100 μM of NusA-MpcA. Control assays were conducted by adding heat-inactivated (95 °C, 10 min) MpcP or no addition of MpcP. Assay mixtures were incubated at 30 °C overnight, quenched with 2 volumes of MeOH and centrifuged at 21,100g for 10 min. The supernatant was then subjected to HPLC–MS analysis as described above.

Purification of the unmodified core peptide

NusA-tagged MpcA and MpcP were overexpressed and purified as described above, and a large-scale reaction (10 ml) was performed with 200 μM of NusA-MpcA and ~50 μM of MpcP at 30 °C overnight. The reaction was quenched with 2 volumes of acetonitrile, centrifuged (4 °C, 4,000g, 20 min) and the supernatant then transferred to a glass vial for rotary evaporation. After the solvent was removed, 100 μg of trypsin and calcium chloride (20 mM final concentration) were added to the aqueous solution for further digestion at 37 °C overnight. Then, 5 g of HP20 resins (Thermo Fisher Scientific) was added to the reaction mixture to absorb the desired product. The HP20 resin was then isolated, washed with 5 volumes of each water, 5% MeCN + 0.1% TFA, 25% MeCN + 0.1% TFA, 50% MeCN + 0.1% TFA, 75% MeCN + 0.1% TFA and 100% MeCN + 0.1% TFA. Fractions containing unmodified MpcA_1–13, as judged by HPLC–MS, were pooled, dried in vacuo and resuspended in 70% MeCN + 0.1% TFA for further purification. The material was then purified on an Agilent analytical HPLC as described above, with an analytical Eclipse XDB-C8 column (Agilent, 5 μm, 4.6 × 150 mm), operating at a flow rate of 0.5 ml min⁻¹ with mobile phase of H₂O and MeCN (+0.1% TFA). The elution was carried out first isocratically at 15% MeCN for 5 min, followed by a gradient of 15–40% MeCN over 25 min, a gradient of 40–100% MeCN over 3 min and lastly isocratic elution with 100% MeCN for 12 min. Fractions containing unmodified MpcA_1–13 (21–24 min), as judged by HPLC–MS, were pooled, dried and subjected to NMR analysis.

Bioinformatics

Protein domain alignment was performed using HHpred online webtool (https://toolkit.tuebingen.mpg.de/tools/hhpred)²¹. Protein BLAST was performed with the MpcB sequence and default blastp parameters on NCBI (https://blast.ncbi.nlm.nih.gov/BlastAlign.cgi). The accession numbers of the first 500 aligned sequences were downloaded and subjected to genome neighborhood analysis using the RODEO webtool⁴². Gene clusters encoding MFS transporters, S9 peptidase or α/β hydrolase fold, B₁₂-rSAM enzymes and hypothetical proteins (DUF5825) were selected manually and further analyzed for potential precursor peptides. DNA and ORF sequences were visualized in SnapGene Viewer 2.7.1. The GC% of codon usage for different open-reading frames was analyzed by online in FramePlot⁴⁹. In total, 228 mpc-like gene clusters were identified with putative precursors, encompassing 104 unique peptide sequences. Precursor peptides are grouped into ‘long’ peptides (with acidic follower, length ≥ 37 aa) and ‘short’ peptides (without acidic follower, length ≤ 33 aa). Two groups of precursor peptides were separately input into EFI-EST for SSN analysis using option C⁴³. An alignment score threshold of 15 was chosen for ‘long’ peptides, and 8 for ‘short’ peptides. The SSNs were visualized in Cytoscape 3.5.1 with prefused force-directed layout. Conserved motif analysis of each cluster was performed on the MEME suite using the GLAM2 algorithm (https://meme-suite.org/meme/tools/glam2; ref. 50).

Extended Data

Extended Data Fig. 1 |. Structural elucidation of clavusporins.

a,b, UPLC-coupled HRMS analysis of extracts of WT S. clavuligerus and the deletion mutants indicated. Shown are extracted ion chromatograms of clavusporin A (a) and clavusporin B (b). The traces are offset on the y axis for clarity. Each experiment was repeated with at least three biological replicates, and representative results are shown. c, Key NMR correlations used for determination of post-translational methylations in clavusporin B. d, Key NMR correlations used for determination of the overall structure of clavusporin B.

Extended Data Fig. 2 |. Phenotypic differences of WT S. clavuligerus and mpc mutants.

a–c, Images of WT S. clavuligerus (a), ΔmpcB (b) and ΔmpcC (c) growing on GYM agar for 10 days. Note the green spores produced by the WT. d–f, SEM images of WT (d), ΔmpcB (e) and ΔmpcC (f), which confirm sporulation in WT, and minimal aerial hyphae growth in mutants. Each experiment was repeated in three biological replicates, and representative results are shown.

Extended Data Fig. 3 |. Phenotypic differences between WT S. ghanaensis and the mpgB mutant.

a, The orthologous mpg gene cluster in S. ghanaensis. The gene locus of mpgA is marked by a black circle, and its protein sequence is shown. b,c, Images of WT S. ghanaensis (b) and ΔmpgB S. ghanaensis (c) growing on SFM agar for 3 days. Note the robust production of dark gray spores by the WT. d,e, SEM images of WT (d) and ΔmpgB S. ghanaensis (e), which show typical formation of Streptomyces spore chains in WT. Each experiment was repeated in three biological replicates, and representative results are shown.

Extended Data Fig. 4 |. MpcP is a site-specific leader peptidase.

a, HPLC–HRMS analysis of the MpcP reaction with NusA-MpcA. Heat-inactivated MpcP and a no-enzyme reaction were used as controls. Shown are extracted ion chromatograms of MpcA_1–21. b, HPLC–HRMS analysis of trypsin-digested NusA-MpcA as a positive control of MpcP assay. Shown are extracted ion chromatograms of two tryptic fragments of MpcA, MpcA_1–13 (blue) and MpcA_{(−15)–(−1)} (red). c, HPLC–HRMS analysis of MpcP-treated MBP-MpcA fusion, which was coexpressed with MpcBC in E. coli. Shown are extracted ion chromatograms of MpcA_1–21 with different methylation states. The traces are offset on the y axis for clarity. Each experiment was repeated in three biological replicates, and representative results are shown.

Supplementary Material

Supplementary information The online version contains supplementary material available at https://doi.org/10.1038/s41589-025-01882-8.

Data availability

Experimental data supporting the conclusions of this study are available within the article and Supplementary Information. Protein sequences were retrieved from the NCBI Nonredundant Protein Database (http://www.ncbi.nlm.nih.gov/protein/) and NCBI accession numbers of proteins in the S. clavuligerus and S. ghanaensis BGCs are as follows: WP_003958126.1(MpcT), WP_003958125.1 (MpcP), WP_003958123.1 (MpcB), WP_003958122.1 (MpcC), WP_004993370.1 (MpgT), WP_004993372.1 (MpgP), WP_004993375.1 (MpgB), WP_004993376.1 (MpgC). Raw NMR data used to elucidate natural product structures as well as the other data in this paper are available from the corresponding author upon request. Source data are provided with this paper.