Optimized Production of Concanamycins Using a Rational Metabolic Engineering Strategy

Abstract

Plecomacrolides, such as concanamycin and bafilomycin, are potent and specific inhibitors of vacuolar-type ATPase. Concanamycins are 18-membered macrolides with promising therapeutic potential against multiple diseases, including viral infection, osteoporosis, and cancer. Due to the complexity of their total synthesis, the production of concanamycins is only achieved through microbial fermentation. However, the low titers of concanamycin A and its analogs in the native producing strains are a significant bottleneck for scale-up, robust structure-activity relationship studies, and drug development. To address this challenge, we designed a library of engineered Streptomyces strains for the overproduction of concanamycins by combining the overexpression of target regulatory genes with the optimization of fermentation media. Integration of two endogenous regulators from the concanamycin biosynthetic gene cluster (cms) and one heterologous regulatory gene from the bafilomycin biosynthetic gene cluster into the attB site significantly increased production of concanamycin A and its low abundant analog concanamycin B in Streptomyces eitanensis. The highest titers reported to date were observed in the engineered S. eitanensis DHS10676, which produced over 900 mg/L of concanamycin A and 300 mg/L of concanamycin B. Heterologous overexpression of the identified target regulatory genes across a panel of Streptomyces spp., harboring a putative concanamycin biosynthetic gene cluster confirmed its identity, and significantly improved concanamycin A production in all tested strains. Strain engineering, optimization of fermentation, and extraction purification protocols enabled swift access to these structurally complex plecomacrolides for semi-synthetic medicinal chemistry-based approaches. Together, this work established a platform for robust overproduction of concanamycin analogs across species.

1. Introduction

Natural products, also referred to as secondary or specialized metabolites, are a valuable source of biologically active molecules and act as important scaffolds to develop new drug leads (Atanasov et al., 2021). Natural products are not essential for the growth, development, or reproduction of an organism, but they can often play a vital role in their ecological interactions (Abrudan et al., 2015; Chevrette et al., 2022; Khalil et al., 2019; Westhoff et al., 2021). Microorganisms are a significant source of natural products with diverse therapeutic applications, such as antibiotics, anticancer agents, antivirals and immunomodulators (Newman and Cragg, 2020). Bacterial survival, resilience, and proliferation across diverse environmental conditions rely on a vast transcriptional regulatory network that triggers metabolic adjustments to adapt rapidly to new environmental conditions. In Streptomyces spp., these processes are coupled with the production of a wide range of pharmaceutically important secondary metabolites (Qiu et al., 2024; Traxler et al., 2013). Strains of the genus Streptomyces are among the most prolific microbial sources of bioactive natural products with complex chemical signatures. Even though these organisms are often challenging to culture and engineer, their biochemical potential remains of significant academic and industrial interest (Ko et al., 2020; Pham et al., 2019).

Concanamycin A (1), B (2) and C (3) (Fig. 1) were first isolated from Streptomyces diastatochromogenes S-45 (Kinashi et al., 1984). These molecules are comprised of an 18-membered macrolactone and include a 6-membered hemiketal ring that belong to the polyketide-derived plecomacrolide family of bioactive natural products. Concanamycins are potent Vacuolar-type adenosine triphosphatase (V-ATPase) inhibitors (Huss et al., 2002; Keon et al., 2022) with diverse biological activities, including antifungal (Okoli et al., 2009), antitumor (Michel et al., 2013), anti-viral (Guinea and Carrasco, 1994), and immunomodulator (Togashi et al., 1997). Concanamycin A has also exhibited sub-nanomolar potency against Nef-dependent MHC-I downmodulation activity in HIV-infected T cells (Painter et al., 2020). Efforts to maximize potency while minimizing mammalian cell toxicity have been attempted (Dröse et al., 2001; McCauley et al., 2024). Due to its complex chemical structure, the total synthesis of concanamycins remains challenging and is not a sustainable source of the compounds (Toshima et al., 2001). Commercially available concanamycin A and B are produced by large-scale fermentation of Streptomyces strains at low titers and are consequently costly. Despite the four decades since the isolation and characterization of concanamycins was first reported, the highest titers of both concanamycin A and B (1 and 2, Fig. 1) in batch fermentations are 60 and 20 mg/L, respectively (Bindseil and Zeeck, 1993; Schuhmann and Grond, 2004). Recently, heterologous expression of concanamycin biosynthetic gene cluster (BGC) from Streptomyces neyagawaensis in a Streptomyces avermitilis SUKA32 chassis enabled the production of concanamycin A (24 mg/L, 2-fold improvement compared with parental S. neyagawaensis) (Kudo et al., 2024). In an effort to develop a high-titer overproducing strain, we pursued a targeted metabolic engineering strategy focused on regulation of secondary metabolism (Mingyar et al., 2021; Robertsen et al., 2018).

Fig. 1. Structure of concanamycins reported in this work.

Concanamycin A (1, X= ethyl, Y=CONH₂), Concanamycin B (2, X= methyl, Y=CONH₂), and Concanamycin C (3, X= methyl, Y=H).

The genetic information required for the biosynthesis of secondary metabolites is spatially organized as biosynthetic gene clusters (BGCs) that often include cluster-situated regulatory genes. The concanamycin BGC (cms BGC) was first characterized in S. neyagawaensis, containing 28 ORFs in a total of ~100 Kb (Haydock et al., 2005). The aglycone core of concanamycins is derived from the condensation of acyl-CoA building blocks by Type I Polyketide Synthases (PKSs). Biosynthetic genes for methoxymalonyl-ACP and ethylmalonyl-CoA production were also identified in this BGC. Concanamycin A differs from concanamycin B (Fig. 1) by the condensation of ethylmalonyl-CoA over methylmalonyl-CoA leading to an ethyl or methyl branch on the macrolactone C8 position. The concanamycin aglycones are modified by the addition of the 4’-O-carbamoyl-2’-deoxyrhamnose (encoded by a six gene deoxysugar subcluster, a carbamoyltransferase, and a glycosyltransferase) at the hemiketal ring C23 hydroxyl group (Haydock et al., 2005). Within the gene cluster, two putative pathway specific regulators show high homology to members of the LuxR and SARP protein families (Haydock et al., 2005).

Pleiotropic and cluster-situated regulatory mechanisms tightly regulate secondary metabolite production (Gehrke et al., 2019; Świątek-Połatyńska et al., 2015). Therefore, engineering new regulatory states that modulate natural product biosynthesis can enable controlled production of low-abundant metabolites (Jung et al., 2008; Malla et al., 2010). Here, we present a rational and systematic approach to increase production of concanamycin A and B. First, optimization of Streptomyces eitanensis cultivation conditions led to a >270-fold increased production of concanamycin A. Further, engineering of native and heterologous regulatory gene overexpression, we achieved production of 909.8±64.7 mg/L of concanamycin A, a 10-fold improvement compared to wild-type. Sodium propionate supplementation boosted production of concanamycin B (CMB) to 306.5±42.1 mg/L, without significantly decreasing concanamycin A (CMA). Moreover, efforts to optimize extraction and isolation enabled the purification of 483.1±25.8 mg/L and 159.4±41.8 mg/L of concanamycin A and B, respectively, under shake-flask cultivation conditions. Proteomics and metabolomics analysis uncovered a metabolic shift for CMA in the engineered versus wild-type strains. Recently, we used concanamycin A and B sustainably produced by engineered S. eitanensis as structural scaffolds to generate unnatural semi-synthetic derivatives for a detailed structure-activity relationship study against HIV-Nef (McCauley et al., 2024). This work highlights that rational metabolic engineering design principles are broadly applicable for the rapid development of Streptomyces sp. for sustainable production of concanamycin analogs.

2. Material and methods

2.1. Strains and culture conditions

Strains used and designed in this study are listed in Table S1. Escherichia coli DH5α (NEB) was used as host for plasmid assembly, replication, and preservation. E. coli S171 strain was used for interspecies conjugation. All E. coli strains were cultivated in LB medium (10g Tryptone, 10 g NaCl and 5 g of Yeast Extract per liter) at 37 °C. For plasmid maintenance and selection, LB media was supplemented with appropriate antibiotics. Concanamycin producing strain S. eitanensis (wild-type) was a gift from Fermentek. All Streptromyces strains were cultivated in 2xYT (16 g tryptone, 10 g yeast extract, and 5 g NaCl per liter) at 28 °C for seed culture and for genomic DNA extraction. Strains were sporulated in OPAH (1 g oatmeal, 1 g pharmamedia, 1 g arabinose, 0.5 g humic acid, 0.5 mM KH₂PO₄, 0.5 mM CaCl₂, 0.5 mM MgSO₄, 1.9 mg Na₂-EDTA·2H₂O, 1.4 mg FeSO₄·7H₂O, 0.2 mg H₃BO₃, 0.05 mg MnSO₄·H₂O, 0.01 mg ZnSO₄·7H₂O, 0.01 mg Na₂MoO₄·2H₂O, 0.01 mg CuSO₄, 0.01 mg CoCl₂, per liter). When appropriate, media was supplemented with antibiotics: apramycin (50 μg/mL) (Goldbio, A-600-10), kanamycin (50 μg/mL) (Goldbio, K-120-25) and nalidixic acid (25 μg/mL) (Cayman, 19807).

2.2. Genome extraction, sequencing, and assembly

High-quality genomic DNA was prepared using Masterpure DNA extraction kit (Lucigen, MC85200) with few modifications. Wild-type and engineered Streptomyces strains were cultivated in 2xYT at 28 °C for 3 days. Cells were pelleted by centrifugation and washed twice with PBS (8 g NaCl, 0.1 g KCl, 1.44 g Na₂PO₄, 0.22 g KH₂PO₄ per liter, pH=7.4). Biomass was resuspended in 480 μL of EDTA and 120 μL of Lysozyme (DotScientific, DSL38100-10) (10 mg/ml) and incubated at 37°C for 45 min. Cells were centrifuged for 1 min at ~5,000×g, supernatant was discarded, and pellet was resuspended in 200 μL of ‘Tissue and Cell Lysis solution’ supplemented with 1 μL of Proteinase K (Qiagen, RP107B-5). Samples were incubated at 65 °C for 15 min, and allowed to cool down to RT for 5 min. Next, samples were incubated at 95 °C for 10 min, cooled to RT before 30 μL of RNAse A (Sigma-aldrich, R5503-1G) (10 mg/ml) was added, and incubated at 37 °C for 1 hr. Genomic DNA was further extracted following kit protocol. Quality of genomic DNA was evaluated by gel electrophoresis. S. eitanensis (wild-type) was sequenced using an hybrid approach combining Nanopore technology and Illumina reads (Plasmidsaurus). Genome assembly, annotation and phylogenetic analysis were performed using KBase (Arkin et al., 2018) platform. Briefly, read quality was checked with FastQC v0.12.1 (Andrews, 2010), and long-reads were cleaned with Filtlong v.0.2.1 (Wick, 2017) before assembly. Genome was assembled with ‘Unicycler - v0.4.8’ (Wick et al., 2017), ‘MaSuRCA Assembler - v3.2.9’ (Zimin et al., 2013) and ‘HybridSPAdes - v3.15.3’ (Antipov et al., 2016). The quality of genome assemblies was assessed with ‘QUAST v4.4’ (Gurevich et al., 2013) (Table S2). Genome completeness and contamination was assessed with CheckM (Parks et al., 2015). Proseek was used for genome visualization (Grant et al., 2023) (Fig. 2A). Genome annotation was performed with RASTtk – v1.073 (Brettin et al., 2015), and eggnog-mapper (Cantalapiedra et al., 2021).

Fig. 2. Genome assembly, annotation and analysis of S. eitanensis.

(A) Genome assembled with Unicycler, visualized with Proseek (Grant et al., 2023) (B) Phylogenetic analysis of 49 core gene (Clusters of Orthologous Groups) families of 31 public RefSeq related genomes and S. eitanensis using SpeciesTree v2.2.0 (KBase) and iTOL v6 (Letunic and Bork, 2021). Phylogenetic tree with bootstrap values (out of 100). (C) Predicted biosynthetic gene cluster (BGC) content in S. eitanensis strain predicted by AntiSMASH and Gecco. RiPP - ribosomally synthesized and post-translationally modified peptides; NRP - Nonribosomal peptides; NRP/PKS - hybrid peptide-polyketides. (D) Comparative analysis of Concanamycin A BGC from S. eitanensis with characterized BGCs from S. neyagawensis and S. scabiei using clinker (Gilchrist and Chooi, 2021).

2.3. Genome-wide phylogenetic analysis

Followed by genome assembly and annotation, phylogenomic analysis was performed using SpeciesTree in KBase suite. A set of closely related genomes (n=33) from public databases was constructed using 49 core, universal genes defined by Clusters of Orthologous Groups gene families. Tree was performed with SpeciesTreeBuilder v2.2.0 (Arkin et al., 2018) that uses FastTree2 (Price et al., 2010) to infer maximum-likelihood (ML) phylogenetic analysis. Alignments used 49 highly conserved core universal genes defined by Cluster of Orthologs Groups (COG) gene families for a set of 31 closely related genomes from public database (RefSeq) to estimate phylogenetic tree, as defined by SpeciesTreeBuilder tool. FastANI (Jain et al., 2017) was also used to estimate whole-genome Average Nucleotide Identity (ANI) between S. eitanensis assembly and the closest genomes in public database. ANI estimations between S. eitanensis and S. scabiei 87.22 (GCF_000091305.1), S. stelliscabiei (GCA_001008135.1), S. griseiscabiei (GCA_020010925.1) and S. neyagawaensis (GCF_028863365.1) were 89.8%, 89.9%, 90.5% and 87.9%, respectively. Phylogenetic tree generated by SpeciesTreeBuilder was then exported to iTOL v6 (Letunic and Bork, 2021) for visualization and figure design (Fig. 2B).

2.4. Plasmid design and genome editing

Strains and plasmids generated in this work are listed In Table S1 and S3. The integrative system, mediated by the attP site of the Streptomyces phage ΦC31 (Bierman et al., 1992), was used for the conjugal transfer of DNA from E. coli S171 (Simon et al., 1983) to S. eitanensis. To design overexpression plasmids, the strong synthetic constitutive promoter kasO*p (Wang et al., 2013) and fd terminator (Ward et al., 1986) were synthetized as gene fragment (Table S4) and assembled into the integrative plasmid pSET152 (Bierman et al., 1992) and pSET152-kan (Lian et al., 2008) via Gibson Assembly, originating plasmid pSET152-kasO*p and pSET152k-kasO*p, respectively. The CmsG and cmsR genes were amplified from gDNA of S. eitanensis strain, while bafR was amplified from gDNA of Streptomyces lohii (Li et al., 2017). Selected regulator genes were amplified and assembled, via Gibson Assembly, in pSET152-kasO*p and pSET152k-kasO*p between BamHI and EcoRI (R0136S and R0101S, New England Biolabs) restriction sites. Synthetic ribosomal binding sites (Bai et al., 2015) were used as overlapping regions for multi-gene vector design. Primers were designed using pyDNA (Pereira et al., 2015) and listed in Table S4. The integrative plasmids were introduced into wild-type S. eitanensis (this work), S. stelliscabiei (NRRL B-24447), S. neyagawaensis (NRRL ISP-5588), S. scabiei (NRRL B-24449), and S. griseiscabiei (NRRL B-2795) via interspecies conjugation from E. coli S171. Following incubation at 28 °C for 12 h, each plate was overlaid with 1 mL sterilized water containing 1.25 mg apramycin or kanamycin and 0.5 mg nalidixic acid. After additional 3–5 days, the recombinants were transferred to OPAH plates with 25 μg/mL nalidixic acid and 50 μg/mL apramycin or kanamycin. The resultant antibiotic resistant strains were PCR confirmed and whole-genome sequenced (Nanopore) (Fig. S3B). Cluster comparison and visualization were performed with clinker (Gilchrist and Chooi, 2021) and DNAviewer (Zulkower and Rosser, 2020). Knockdown of cmsG and cmsR in S. eitanensis was performed using the CUmate-Based Inducible CRISPR interference (CRISPRi) system (Bai and van Wezel, 2023). Protospacers were designed (~65 – 70 bp after ATG) for each gene and assembled in pCB-2 using Golden Gate (Bai and van Wezel, 2023). Plasmids, pCB-2_gRNA-cmsR and pCB-2_gRNA-cmsR, were introduced in S. eitanensis via interspecies conjugation as described above.

2.5. Concanamycin A production conditions

Spores from fresh OPAH plates were inoculated into 50 mL of peanut meal & starch media (glucose 10 g, starch 30 g, bacto peptone 5 g, peanut meal 10 g, yeast extract 5 g, CaCO₃ 2 g per liter, pH=7.0) and incubated for 3 days at 28 °C. Subsequently, 10 mL of pre-culture was used to inoculate 1 L of producing media (GICYE: glucose 10 g, inulin 30 g, bacto peptone 5 g, corn gluten meal 10 g, yeast extract 5 g, CaCO₃ 2 g per liter, pH=7.0). Cultures were incubated 7 days at 22 °C and 170 rpm. When appropriate, 20 mL of a 29.9% (w/v) solution of sodium propionate was added at 48 h. Activation of dCas9 and sgRNA, in strains DHS10683 and DHS10683, was performed by supplementing GICYE media with 10 μM of cumate (Sigma-Aldrich, 268402) and incubated for 7 days at 22 °C and 170 rpm. Detailed composition of all media tested in this work can be found in supplementary information (S2.1).

2.6. Sample preparation and quantification of concanamycin A and B

Quantification of concanamycin A and B, reported in the main manuscript, was performed using 1 mL samples from biomass cultivated in 1 L shake-flasks, from three independent replicates. To estimate growth and production over time, samples were collected every day for 10 days (Fig. 5B,C). Compound quantification presented throughout the manuscript was collected on day 7, unless stated otherwise. From each 1 L culture, 1 mL was collected, and biomass was pelleted by centrifugation for 5 min at 20000 rpm. Cell pellet was resuspended in 1 mL of methanol and 100 μL of glass beads. Tubes were vortexed for 3 h at 4 °C, followed by centrifugation for 5 min at 20000 rpm. Supernatant was mixed with methanol 50/50 and centrifuged for 5 min at 20000 rpm. Routine quantification of concanamycin A and B production was performed by HPLC (Shimadzu) equipped with a PDA detector analyzed with a C18 column (Luna 5 μm C18(2) - Phenomenex, kept at 40 °C), Water(A)/Acetonitrile(B) (10% to 100% B) was used as mobile-phase at 2 ml/min. Calibration curves were performed with known concentrations of pure concanamycin A and B (Cayman Chemical: Item No. 11050 and 15502, respectively), production titers were quantified by comparing the standard curve of known concentrations and sample peak areas (AUC) at 280 nm. Concanamycin A and B were detected in culture supernatants (using LC-MS/MS, see 2.11) at trace levels for all tested strains. Therefore, quantifications reported throughout the manuscript correspond to the intracellular concentration of concanamycin A and B. Statistical analysis and visualization were performed with GraphPad Prism v.10.2.2.

Fig. 5. Contribution of BafR and cms cluster-situated regulators from S. eitanensis in concanamycin A production.

(A) Concanamycin A production, at day 7, in engineered S. eitanensis strains overexpressing bafR alone or in combination with native CMA cluster-situated regulators, cmsG and cmsR, cultivated with diverse organic nitrogen sources and with Zinc supplementation (ZnSO₄). (C) Concanamycin A production over time (days, X axis) in DHS10676 and wild-type. (B) Cell dry weight (DCW) of DHS101676 and wild-type cultivated in GICYE at day 4, 7 and 10, at 22°C. (D) Concanamycin A production in engineered S. stelliscabiei (NRRL B-2795), S. neyagawensis (NRRL ISP-5588), S. scabiei (NRRL B-24449), and S. griseiscabiei (NRRL B-24447), and wild-type cultivated in GICYE at 22°C, 7 days. Symbols represent three biological independent experiments. Error bars indicate standard deviations (SD).

2.7. Biomass quantification

Throughout cultivation, 1 ml samples were taken and filtered through a pre-dried membrane filter. Cells were washed twice with water and dried in a microwave oven for 1 min (Borodina et al., 2008). Sampling was performed in triplicate. Due to small pellets morphology, sampling was performed using sterile wide-bore pipette tips.

2.8. Isolation and purification of concanamycin A and B

Total biomass (from 1 L culture) was separated by vacuum filtration. Cell pellets were broken by coating the cells with 500 mL of a mixture of 20% methanol (MeOH) in dichloromethane (DCM) and shaking overnight. The organics were obtained by further vacuum filtration and concentrated. The dried extract was resuspended in 40 mL of DCM, and a 1 mL aliquot was set aside for analysis. The remaining DCM mixture was concentrated with silica to be dry loaded for the first round of normal phase purification. Both normal phase purifications were performed on a Biotage flash column system with a 40 g column. For the first round of purification an ethyl acetate/hexane gradient (15–100%) was utilized where concanamycins A-C eluted at 100%. The second round of normal phase was run with a step wise gradient of isopropanol and hexanes/chloroform (1:3 parts). The steps were 0%, 4%, 8%, 25%, and 40% isopropanol. Concanamycins A-C start to elute around 25%. The crude concanamycin mixture is further processed through a reverse phase purification on a Biotage flash column system with 55% acetonitrile in water using a C18 40 g column (Phenomenex, SO240040-0). Two main peaks were obtained: 1^st major peak comprised mainly of CMB mixed with minor amounts of CMA and CMC and, 2^nd peak was pure CMA. The first peak was further purified on a preparative HPLC as previously reported (McCauley et al., 2024). Purity assessment of isolated compounds was performed using HPLC (as described in 2.7), NMR and LC-MS/MS (see Supplementary Information, Fig. S8–14). Quantification of concanamycin A and B in crude extracts was also performed using HPLC (following protocol described in 2.7) (Table S7). Extractions and purifications were performed independently for each biological triplicate.

2.9. Protein extraction and proteomics analysis

Proteomics analysis was performed at two time-points, day 4 and 7. Three independent biological replicates were collected for each strain, S. eitanensis wild-type and DHS10676, cultivated under concanamycin producing conditions for the two time points. 10 mL of culture were centrifuged at 3000 rpm for 5 min and cell pellet was washed twice with PBS solution and stored at −80 °C. Frozen cell pellets were lysed using 800 μl of lysis buffer (6 M urea, 2 M thiourea, 5 mM DTT, 0.1 M TRIS-HCl and 40 μl of proteases inhibitor cocktail), followed by three cycles of sonication. Preparation of cell lysate for proteomic analysis was executed by following the protocol for the S-trap mini kit (Protifi). Briefly, 23 μl of the cell lysate was diluted in 23 μl of buffer 1 (10% SDS, 100mM tetraethylammonium bromide TEAB, pH 8.5) and then vortexed for 20 secs. For reduction 2 μl of 120mM tris(2-carboxyethyl)phosphine was added and incubated at 55°C for 15 min. 2 μl of alkylator (500 mM methyl methanethiosulfanote in isopropanol) was added and incubated at RT for 10 min. Lastly, the sample was acidified with 5 μl of 12% phosphoric acid (aq) prior to loading onto the S-trap micro column for binding and washing of proteins as stated in the protocol. Proteins were digested on the S-trap column with trypsin (Promega, VA9000) in 50 mM TEAB with an enzyme to protein ration 1:10 at 37 °C overnight. Peptides were eluted with 80 μl each of three elution buffers in series: (1) 50 mM TEAB, (2) 0.2% formic acid (aq), and (3) 50% acetonitrile (aq). Eluted peptides were pooled and dried down under N₂. Samples were store dried at −20 °C until further analysis. Peptide samples were resuspended in 0.1% formic acid in water and analyzed with a reversed-phase nanoflow UPLC (Ultimate 3000, ThermoFisher) coupled to an Orbitrap Fusion Lumos MS via nano-ESI in positive ion mode. Samples were injected to a C18 trap column (Acclaim PepMap 100, 75 μm × 2 cm, nanoviper) by a loading pump with 2% MeCN with 0.1% formic acid at 5 μL/min, and further separated on a C18 separation column (Acclaim Pepmap RSLC, 75 μm × 50 cm, nanoviper) over a 90 min gradient with a flow rate of 300 nL/min. LC separation and elution was performed by following a gradient method composed of solvent A (2% MeCN with 0.1% formic acid) and Solvent B (80% MeCN with 0.1% formic acid): 0–5 min, 5% B; 5–75 min, 5 to 40% B; 75–82 min, 40 to 95% B; 82–84 min, 95% B; 84–90 min, 5% B. LC-MS/MS analysis was performed in data-dependent mode by applying dynamic exclusion after one scan for 30 seconds. Ions at >5e⁴ signal abundance in the survey MS scan were selected and subjected to HCD MS/MS at 35% normalized activation energy for 3 seconds before acquiring another MS scan and subsequent MS/MS scans. The HCD spectral resolution was set at 15,000.

2.9.1. Data processing and Statistical analysis

The acquired LC-HCD MS/MS data were searched using Sequest HT search algorithm on Proteome Discoverer 2.2 (Thermo Fisher) against S. eitanensis reference proteome (annotated with eggnog-mapper, see 2.2) with reversed peptide decoys and common contaminants. A maximum of three missed tryptic cleavages were allowed with fixed C-carbamidomethylation, variable M-oxidation, variable N-deamidation, and variable N-terminal acetylation. Precursor mass tolerance was set at 10 ppm, MS/MS mass tolerance was 0.02 Da., and peptide length were limited to minimum 5 amino acids. False discovery rate (FDR) of 1% was applied for peptide and protein identification. Raw data from Proteome Discoverer 2.2 were loaded into R (2623 proteins). Only proteins that were quantified with two unique peptides and were identified in, at least, two biological replicates were used in downstream analysis. A total of 1891 proteins were then used for differential protein abundance. The output data was cleaned for potential batch effect using limma (Ritchie et al., 2015) and normalized with vsn (Huber et al., 2002). Missing values were imputed with the impute function ‘QRILC’ with package ‘imputeLCMD’. Differential protein abundance was calculated with limma between wild-type and engineered DHS10676 strains at 4 and 7 days. Proteins were classified as ‘hit’ if fold-change was >1 and with a FDR ≤ 5%. Enrichment analysis was performed with ClusterProfiler v4.0 (Yu et al., 2012).

2.10. Untargeted metabolomics

2.10.1. Sample preparation

Samples of intracellular and extracellular metabolites, prepared in 2.7, were used for untargeted metabolomics. Culture medium was also extracted, followed the protocol in 2.7, as controls and run alongside experimental samples to account for sampling and laboratory contamination. As well, reagent blanks (MeOH) were prepared and measured to account for extraneous signals occurring from chemicals and analytical systems. All samples were prepared in three biological replicates to demonstrate reproducibility.

2.10.2. UHPLC-QTOF-MS/MS profiling of crude extracts

Metabolite extracts were analyzed via ultra-high-performance liquid chromatography-quadrupole time-of-flight mass spectrometry (UHPLC-LCMS) using an Agilent 1290 Infinity II UHPLC coupled to an Agilent 6545 ESI-Q-TOF-MS. The method for data collection is as follows: samples were injected and data acquisition occurred via auto-MS/MS in positive mode. Chromatography was performed using a Phenomenex Kinetex phenyl-hexyl (1.7 μm, 2.1 × 50 mm) column and compounds were eluted with an isocratic elution of 90% solvent A (100% H₂O + 0.1% formic acid) for 1 minute followed by a 9 minute linear gradient elution to 100% solvent B (95% MeCN + 5% H₂O + 0.1% formic acid). The capillary temperature was set to 320°C with a source voltage of 3.5kV and a sheath gas flow rate of 11 L/min for electrospray ionization (ESI). Ion fragmentation occurred using ramped collision energy (5 × m/z/100 + 10 eV) of a maximum of 9 selected precursors per cycle. The positive mode internal lock standards used were Purine C₅H₄N₄ [M + H]⁺ (m/z 121.0508) and hexakis (1H,1H,3H-tetrafluoropropoxy)-phosphazene C₁₈H₁₈F₂₄N₃O₆P₃ [M + H]⁺ (m/z 922.0098) and ions were included in the static exclusion list. A full static exclusion list was optimized based on high-intensity contaminant ions (>1000 counts) found in blank solvent samples.

2.10.3. Data processing

Raw mass spectrometry data was converted from vendor-specific instrument files to mzML format using standard peak detection parameters in the msConvert software from ProteoWizard (Adusumilli and Mallick, 2017). Following this, data processing was performed in MZmine3 following an untargeted LC-MS data pre-processing workflow using the Processing Wizard (Schmid et al., 2023). For HPLC, retention time was selected between 0.4–9 minutes, with 15 maximum peaks in the chromatogram and 4 minimum consecutive scans. The approximate feature full width at half maximum (FWHM) was set to 0.045 min. Retention time (RT) tolerance was set to 0.08 min for intra-sample and 0.4 min for sample-to-sample. The MS¹ noise level was set to 5.0×10² and the MS² noise level was set to 2.0×10². Minimum feature height was set at 1.0×10³ and the scan-to-scan m/z tolerance was set to 0.005 m/z (20 ppm), intra-sample m/z tolerance was set to 0.0015 m/z (3 ppm), and sample-to-sample m/z tolerance was set to 0.004 m/z (8 ppm). Output feature quantification table, peak list, and metadata table was run via feature-based molecular networking on the Global Natural Products Social Molecular Networking (GNPS) platform (Wang et al., 2016). Molecular features as MS/MS spectra were putatively matched to the GNPS MS/MS Spectral Libraries based on a cosine similarity score greater than 0.7 and a minimum of 4 shared peaks. The resultant molecular network was visualized using Cytoscape3.9.1 (Shannon et al., 2003).

2.10.4. Statistical analysis of mass spectrometry data

Statistical analysis of features was performed in R after running in GNPS to pair metadata to features and obtain preliminary annotations (Shah et al., 2023). First, features were filtered to keep only those that were present in two out of the three biological replicates. Next, features that were present in the process and reagent blank samples at greater than 10% intensity were removed from sample feature lists. Subsequently, features were normalized using vsn and missing values were imputed using QRILC imputation.

3. Results and discussion

3.1. Genome analysis of S. eitanensis identifies the concanamycin biosynthetic gene cluster

Concanamycin A production was first identified in S. eitanensis by Fermentek, albeit at low concentrations (personal communication). S. eitanensis genome was sequenced using Illumina short-read and Oxford Nanopore long-reads, and assembled with Unicycler (Wick et al., 2017). Complete genome assembly consisting of three contigs comprising a total of 9,794,182 bp (71.09% GC content) was selected for downstream analysis (Fig. 2A). Assembled genome was annotated with RAST (Brettin et al., 2015) and 8844 coding genes were identified. Maximum-likelihood phylogenetic analysis was performed with FastTree2 (Price et al., 2010) using 31 closely related genomes to estimate the evolutionary relationships between S. eitanensis and reference Streptomyces spp. (Fig. 2B). To estimate species demarcation, whole-genome Average Nucleotide Identity (ANI) was also performed between S. eitanensis genome and the three closest reference genomes, including S. scabiei 87.22 a known CMA-producing strain (Natsume et al., 2017). The estimated ANI between these genomes ranged from 89.8% to 90.5%, suggesting that S. eitanensis belongs to a new Streptomyces species. Biosynthetic potential for secondary metabolite production was also analyzed with ‘antiSMASH v5.1.2’ (Blin et al., 2023) and ‘Gecco’(Carroll et al., 2021). Overall, 38 and 44 biosynthetic gene clusters distributed among two contigs were predicted by antiSMASH and Gecco, respectively (Fig. 2C, Table S5). One of these clusters was identified as a concanamycin A BGC (~102Kb). The cms BGC in S. eitanensis shows a similar organization and strong amino acid sequence similarity when compared to the known concanamycin BGCs from S. neyagawaensis (MIBiG accession: BGC0000040) (Fig. 2D, Fig. 3A) with a gene identity varying between 85 and 97% (Table S6).

Fig. 3. Production of concanamycin A by S. eitanensis.

(A) Architecture and composition of concanamycin A biosynthetic gene cluster in S. eitanensis. Cluster representation to scale (bp) with DNAViewer python. Purple arrows indicate the two genes encoding transcription regulators. Genes encoding the biosynthesis of methoxymalonyl-ACP and ethylmalonyl-CoA are represented by golden arrows. Pink arrows show the sugar biosynthetic pathway, glycosyltransferase and carbamoyltransferase. Modular type-I PKS genes, a Type II thioesterase are indicated by grey arrows. Architecture of modular type-I PKS (dark green AT). Acyl transferase domain AT11 (dark green) accepts ethylmalonyl-CoA and methylmalonyl-CoA, thereby yielding concanamycin A and B, respectively. (B) Effect of media composition and fermentation temperature in concanamycin A content (mg/gDCW), after 7days. Agitation, inoculation and concentration of carbon and nitrogen sources were maintained for all tested conditions. Concanamycin A was quantified by AUC (HPLC) with a standard curve of known concentrations. ND – concanamycin A production not detected. Symbols represent three independent experiments. Error bars indicate standard deviations (SD).

3.2. Optimization of fermentation conditions for concanamycin A production

Concanamycin A (1; CMA) production was detected (< 1 mg/L) in wild-type S. eitanensis strain cultivated in GMSYE at 28°C. Microbial secondary metabolite biosynthesis is regulated by environmental cues such as light, pH, phosphate concentration, oxygenation, temperature, signaling molecules, and the identity and quantity of carbon and nitrogen sources (Basak and Majumdar, 1973; Sánchez et al., 2010; van der Meij et al., 2017). Therefore, we initially investigated the contribution of carbon sources, crude protein (N) sources, and fermentation temperature on CMA production. Polyketides and plecomacrolide production in Streptomyces were shown to be enhanced by soybean oil supplementation (Li et al., 2021; Wang et al., 2017). To investigate if CMA yields could also be improved, 6% soybean oil was added to the fermentation medium composition, which completely abrogated production (not detected) (Fig. 3B). GMSYE media uses soybean meal as the crude nitrogen and minerals source. As soybean oil supplementation completely repressed CMA production (Fig. 3B), we next investigated peanut meal as an alternative crude protein source. Carbon catabolite repression tightly controls secondary metabolite production, and can be alleviated by cultivation in a media containing a mixture of simple (D-glucose) and complex carbon sources (such as oils and polysaccharides) (Sánchez et al., 2010; Sanchez and Demain, 2002). Maltodextrin, initially used as a complex carbon source, consists of 2 to 20 monomers of glucose. Thus, we reasoned that replacing maltodextrin with a larger polysaccharide, such as soluble starch, could relieve glucose repression increasing secondary metabolite production. Indeed, the new fermentation medium, containing soluble starch and peanut meal, enhanced CMA production to 6.3 mg/L (Fig. 3B). Additionally, decreasing the cultivation temperature to 22°C led to increased CMA titer (4.3-fold) and biomass (3-fold) (Fig. 3B). Finally, replacing starch by inulin, a polysaccharide consisting of D-fructose units, and using corn gluten as the crude protein source further increased concanamycin A titers and biomass (Fig. 3B). Apart from concanamycin A, we could also detect the production of natural analogs concanamycin B (2) and C (3) by LC-MS/MS in the optimized culture conditions, at titers below the quantification limit. Optimization of medium composition and fermentation temperature boosted concanamycin A titers to 95.9±17.2 mg/L in the wild-type S. eitanensis strain that, to the best of our knowledge, is the highest reported production in shake-flask cultivation.

3.3. Overexpression of cluster-situated regulators improves concanamycin A production in S. eitanensis

Structure and gene composition of the concanamycin A biosynthetic gene cluster in S. eitanensis is highly conserved between known CMA-producing strains (Fig. 2D, Fig. 3A). The concanamycin BGC has two cluster-situated proteins with a putative regulatory role, Orf3 (923 amino acids) and Orf17* (701 amino acids) (Haydock et al., 2005). The protein encoded by Orf3 shows high sequence similarity to larger ATP-binding regulators of the LuxR family, commonly situated in biosynthetic gene clusters for type I polyketides (Li et al., 2022) (Fig. S1A). This class of regulators is characterized by a polypeptide length of >900 amino acids, an ATP-binding motif at the N-terminus, and a helix-turn-helix domain at the C-terminus of the protein (Fig. S1C). By contrast, Orf17* resembles proteins from the Streptomyces antibiotic regulatory protein (SARP) family, characterized by the presence of OmpR-like DNA-binding and the bacterial transcriptional activator (BTA) domains (Wang et al., 2024) (Fig. S1B, D). The concanamycin cluster-situated regulators in S. eitanensis, show a 85.81% and 95.15% protein identity when compared to Orf3 and Orf17* from the known concanamycin BGC in S. neyagawaensis, respectively (Table S6, Fig. S1A, S1B). Hereafter, the S. eitanensis proteins homologous to Orf3 and Orf17* will be referred to as CmsR and CmsG, respectively). CmsR is an homologue of PikD (37% identity), an ATP-binding protein of the LuxR family that has been characterized as a pathway-specific positive regulator for pikromycin biosynthesis in Streptomyces venezualae (Jung et al., 2008; Wilson et al., 2001) (Fig. S1A, C). CmsG has a 52.3% protein identity to BafG, a positive AfsR family regulator that activates production of the plecomacrolide bafilomycin A₁ in Streptomyces lohii (Li et al., 2021) (Fig. S1B,D). To investigate the role of pathway regulators in concanamycin production, we overexpressed cmsR and cmsG in S. eitanensis following their amplification from genomic DNA and assembly into pSET152 (Bierman et al., 1992) under the constitutive synthetic promoter kasO*p (Wang et al., 2013). The resulting plasmids were integrated into S. eitanensis genome at attB site (ΦC31-based), generating DHS10671 (pSET152- cmsR), DHS10672 (pSET152k- cmsG) and DHS10673 (pSET152k- cmsG-cmsR) strains (Fig. 4A). Disruption of pirA, due to pSET152 integration in the Streptomyces genome, has been reported to modulate primary and secondary metabolism (Talà et al., 2018). Therefore, as control, an empty vector was also integrated in S. eitanensis genome, generating DHS10677 (Fig. 4A). The production of concanamycin A in each engineered strain was assessed under optimized fermentation conditions (GICYE at 22°C, see 3.2). No significant changes in concanamycin A production were observed between DHS10677 and the wild-type strain (Fig. 4B). Overexpression of cmsR, improved CMA production by 2.3-fold (236.2±17.4 mg/L) compared to the wild-type (Fig. 4C). Maximum concanamycin A production was obtained in the strain overexpressing cmsG (DHS10672) reaching 308.5±32.6 mg/L (3.8-fold improvement) compared to the wild-type strain (Fig. 4C). Interestingly, concanamycin A production was not enhanced in strain DHS10673 (integrating pSET152k-cmsG-cmsR) compared to engineered strain DHS10672, overexpressing cmsG alone. Although att-site integration was achieved in S. eitanensis, disruption of cmsG and cmsR by homologous recombination has not been feasible. Thus, the role of CmsR and CmsG in concanamycin A production was accessed by gene knockdown using the cumate inducible CRISPRi system, CUBIC (Bai and van Wezel, 2023). Downregulation of cmsR (DHS10683) or cmsG (DHS10683) inhibited concanamycin A production to levels below quantification limits (Fig. 4C). Together, these results confirm that cmsR and cmsG positively regulate concanamycin A biosynthesis in S. eitanensis.

Fig. 4. Contribution of cluster-situated regulators on concanamycin A production.

(A) Integration of pSET152 vector for overexpression of target regulators under the strong constitutive synthetic promoter kasO*p. Engineering S. eitanensis for overexpression of cmsG (DHS10671), cmsR (DHS10672) and cmsG-cmsR (DHS10673). Empty vector (DHS10677) was also integrated as control. (B) Concanamycin A production (mg/L), quantified by AUC (HPLC) with a standard curve of known concentrations. Symbols represent three biological independent experiments. Error bars indicate standard deviations (SD). (C) Inducible knockdown of native cmsR (DHS10683) or cmsG (DHS10684) using CRISPRi in S. eitanensis. HPLC chromatogram of concanamycin A standard, DHS10683, DHS10684, and wild-type cultivated in GICYE with 10 μM cumate, at 7days.

3.4. Heterologous expression of the bafilomycin cluster-situated regulator bafR in combination with cmsR and cmsG improves concanamycin A production across species

Bafilomycin A₁ is a 16-membered ring plecomacrolide produced by S. lohii that is structurally related to concanamycin A, but bears a smaller macrolactone ring, and lacks the carbamoylated deoxyrhamnose sugar. Bafilomycin A₁ and concanamycin A directly bind and inhibit the eukaryotic V-ATPase (Keon et al., 2022; Schuhmann and Grond, 2004; Bowman et al., 2004). Two cluster-situated regulators, bafG and bafR, are encoded within baf BGC in S. lohii (Li et al., 2021). BafG (609 amino acids) is a close homologue of CmsG from S. eitanensis (Fig. S1B). By contrast, BafR encodes a small 117 amino acid protein, characterized as a LuxR family regulator containing only the conserved helix-turn-helix (HTH) motif with an inducer-independent type activation (Li et al., 2021). A phylogenetic analysis of BafR revealed homologues containing the LuxR-type HTH-domain in multiple actinomycetes species (Fig. S2A and S2B), but not in S. eitanensis. As overexpression of targeted heterologous regulatory proteins has been shown to complement activity of native activators (Garg and Parry, 2010), we decided to investigate if concanamycin A production could be improved by heterologous expression of bafR in S. eitanensis. The bafR gene from S. lohii was amplified using genomic DNA and assembled in the integrative expression plasmid pSET152 under the constitutive synthetic promoter kasO*p. The resulting plasmid (pSET152k-bafR) was then integrated into S. eitanensis genome at attB site, generating DHS10674. Heterologous expression of bafR slightly decreased CMA production compared to the wild-type S. eitanensis strain (Fig. 5A). Constitutive expression of bafR in combination with the native con cluster-situated regulator cmsG was also investigated by integrating pSET152k-bafR-cmsG plasmid into S. eitanensis genome, generating DHS10675 strain. Overexpression of native cmsG and the heterologous bafR, slightly improved concanamycin A production compared to overexpression of cmsG alone (Fig. 5A). An integrative plasmid (pSET152-bafR-cmsG-cmsR) was also designed for overexpression of bafR and both CMA cluster-situated regulators cmsG and cmsR (Fig. S3A). Integration of bafR, cmsG, and cmsR at attP site was confirmed by whole-genome sequencing (Fig. S3B), yielding the engineered DHS10676 strain. Concanamycin A titers in DHS10676 increase to 909.8±64.7 mg/L, a 10-fold improvement compared to wild-type under the same cultivation conditions (Fig. 5A, Table S7). Concanamycin A production was determined over a time course and maximum titers were observed at day 7 for DHS10676 (Fig. 5B). Production of concanamycin B was also improved in the engineered strains DHS10675 and DHS10676, 16.7±12.5 and 38.8±7.4 mg/L, respectively (Fig. 7A, Table S7). Overexpression of three regulators and overproduction of concanamycins, did not significantly change the amount of biomass (DCW) compared to wild-type levels (Fig. 5C).

Fig. 7. Impact of sodium propionate supplementation on concanamycin A and B titers.

Quantified production of concanamycin A (A) and concanamycin B (B) in wild-type, and engineered S. eitanensis strains DHS10675 (bafR, cmsG) and DHS10676 (bafR, cmsG, cmsR) with and without 0.6% (w/v) sodium propionate added at 48 h. Concanamycin A and B were quantified by AUC (HPLC) using standard curves with known concentrations of pure compound. Symbols represent three independent biological replicate. Error bars indicate standard deviations (SD). Significant differences between engineered strains and wild-type under the same culture conditions were calculated by Welch’s t-test. (C) Principal component analysis of untargeted metabolome of wild-type (green) and engineered DHS10676 (pink) cultivated with sodium propionate (circles) and without (squares). Each symbol represents a biological replicate. (D) Isolated titers of pure concanamycin A and B across wild-type, DHS10675 and DHS10676 quantified in A and B. Symbols represent three independent experiments. Error bars indicate standard deviations (SD).

Next, we determined the impact of organic nitrogen sources on concanamycin A production in the engineered strain DHS10676. In contrast to wild-type (Fig. 3B), overexpression of the three regulators enable the production of concanamycin A across all organic nitrogen sources tested at comparable titers (Fig. 5A). Analysis of the cms BGC with antiSMASH (v 7.1.0.) (Blin et al., 2023) revealed a putative transcription-factor binding site for a zinc-responsive repressor (weak score: 17.79 out of 28.19). As zinc supplementation has been shown to affect secondary metabolite production in multiple Streptomyces species (Lyu et al., 2022), we assessed whether concanamycin A production could be altered by this nutrient in the engineered DHS10676 strain. Concanamycin A production in the engineered DHS10676 was not affected by increasing exogenous Zn²⁺, under tested cultivation conditions (Fig. 5A).

Concanamycin A production has been reported at low titers in a handful of Streptomyces species (Haydock et al., 2005; Natsume et al., 2017; Woo et al., 1992). To assess whether the designed plasmid for constitutive expression of bafR, cmsG, and cmsR (Fig. S3A) could activate cms BGC expression across species, we engineered two known concanamycin producers S. scabiei and S. neyagawensis, as well as S. stelliscabiei and S. griseiscabiei that contain a highly homologous cms BGC (Fig. 1D, Fig. S3C). Heterologous expression of bafR, cmsG, and cmsR significantly improved concanamycin A production in all engineered strains when compared to respective wild-type (Fig. 5D). Not only did we show production of concanamycin A in S. stelliscabiei and S. griseiscabiei for the first time, but also a maximum titer of 442 mg/L was observed in the engineered S. stelliscabiei strain (Fig. 5D). In this work, engineered S. neyagawaensis produced 107 mg/L of concanamycin A (Fig. 5D), a 5-fold increase compared to the recently reported expression of S. neyagawaensis concanamycin cluster in a chassis strain (Kudo et al., 2024). Overexpression of S. eitanensis cluster-situated regulators, cmsG and cmsR, and the bafilomycin inducer-independent regulator bafR, improved concanamycin A production across species, providing an important tool for accelerating natural product discovery.

3.5. Proteomics and metabolomics identifies a metabolic switch to CMA/B overproduction

We next characterized the engineered strain DHS10676 and wild-type S. eitanensis using untargeted proteomics and metabolomics under concanamycin production conditions (Fig. S4). Relative protein abundances between DHS10676 and wild-type strains at day 4 and 7 were determined using label-free quantification analysis. Overall, protein abundances between samples cluster based on growth/production stage (day 4 vs 7) (Fig. S4A, B). Functional enrichment analysis was performed in proteins with significantly (log₂ ≤−1 or ≥ 1, p-adj ≤ 0.05) changed abundances between engineered DHS10676 and wild-type strains (Fig. 5B). Proteins associated with ‘Biosynthesis of secondary metabolism’ are overrepresented within proteins with increased abundances in engineered DHS10676 strain. On the other hand, proteins with decreased abundances in engineered DHS1676 strain are enriched for targets involved in the ‘glycine, serine and threonine metabolism’ (Fig. S5B). Variance between wild-type and engineered strain (DHS10676) was mostly driven by the differential abundances of proteins from the cms BGC (Fig. S4B, Fig. S5A). Most proteins from the cms BGC were identified in DHS10676 and in wild-type strains (Fig. 6A). Interestingly, the overexpressed CmsR was not detected in either DHS10676 or the wild-type strain at day 4 or 7. BafR was only identified in the engineered DHS10676 strain on day 4 of cultivation and not in wild-type (Fig. S4C). CmsG was identified at both sampling times (4 and 7 days) but only in the engineered DHS10676 strain (Fig. S4C). Overall, DHS10676 showed a significantly increased abundance of proteins from the cms BGC compared to wild-type cultivated under the same conditions (Fig. 6A, Fig. S5A). Overexpression of the three target regulators (bafR, cmsG and cmsR) led to increased abundances of biosynthetic proteins involved in production of concanamycins.

Fig. 6. Proteomics and untargeted metabolomics analysis of DHS10676 and wild-type S. eitanensis.

(A) Significant changes in protein abundances from the cms BGC, between the engineered DHS10676 strain and wild-type at day 4 (light pink) and 7 (dark pink). Proteins were considered a hit if log₂fold-change > 1 (X axis) and with adjusted P-value moderated t-test with Benjamini-Hochberg false discovery rate (FDR) adjustment ≤0.05 (dot size). (B) Feature based analysis of untargeted metabolomics of DHS10676 (pink) and wild-type (green) at day 7 of strain cultivation under CMA producing conditions.

As a next step, we analyzed the non-targeted metabolome of S. eitanensis wild-type and engineered DHS10676, under concanamycin producing conditions. Metabolites were visualized and annotated by feature-based molecular networking (Nothias et al., 2020), which revealed a total of 142 nodes, from which 69 are singletons (Fig. 6B). Only 22 features were annotated by GNPS (Nothias et al., 2020), using a level 2 identification confidence. All features from the putatively annotated concanamycins node (Fig. 6B, Fig. S4D) show increased abundance in engineered DHS10676 compared to wild-type, with the exception of the parent ion 697.4266 m/z, which corresponds to concanamycin H, the ‘open hemiketal’ analog of CMA (Fig. S8) (Li et al., 2022; Lv et al., 2022). Decreased production of desferrioxamine E, phenazinomycin, 1-hydroxyphenazine, and actinomycin Z₃ was observed in engineered DHS10676 when compared to wild-type strain (Fig. S5C, S6). Simultaneously, proteins located in the predicted BGCs of desferrioxamine E, endophenazine and actinomycin D (antiSMASH, Table S5) were found significantly downregulated in DHS10676 strain when compared to wild-type (Fig. S5A). Increased production of 23-hydroxyundecylprodiginine was observed in the engineered DHS10676 strain, concomitantly with the significant increased abundance of several proteins from the predicted marineosin BGC (Fig. S5A, C). In addition to concanamycins, molecular network (Fig. 6B), clustering (Fig. S4A) and quantification (Fig. S5C) analysis revealed a distinct shift in the identity and abundance of intracellular secondary metabolites from the two strains.

3.6. Sodium propionate supplementation improves production of the low abundant natural analog concanamycin B

Concanamycin B (2) is a less potent inhibitor of V-ATPase compared to its natural analog concanamyin A (1), however it has shown promise for osteoclastic V-ATPase inhibiting bone resorption and as a suppressor of antigen presentation by MHC-class II molecules (Ito et al., 1995; Woo et al., 1996). The S. eitanensis wild-type strain only produces concanamycin B with titers below quantification limit. The engineered S. eitanensis strain, DHS10676, produces 38.8±7.5 mg/L of concanamycin B, which is well below levels achieved for concanamycin A (Fig. 7B). In concanamycin B the C8 ethyl group derived from ethylmalonyl-CoA is replaced by a methyl group that is derived from methylmalonyl-CoA (Kinashi et al., 1984) (Fig. 3A). Elongation of the concanamycin backbone was shown to incorporate up to seven propionate units in 1, and eight in 2 (Bindseil and Zeeck, 1994). Previous studies reported that addition of low concentrations of sodium propionate can favor incorporation of methylmalonyl-CoA extender units in promiscuous type-I PKS AT domains (Matějů et al., 1988; Pospiil et al., 1985). Additionally, supplementation of sodium propionate at later stages of fermentation was shown to further improve secondary metabolite production in select Streptomyces strains (Li et al., 2007). Therefore, we investigated the impact of sodium propionate in concanamycin A and B titers in the engineered DHS10676 strain. We first cultivated DHS10676 in small-scale cultures (50 mL) with GICYE supplemented with sodium propionate (0.1, 0.6, 0.74 and 1% w/v) at three time-points (0, 48, and 72 h). The highest production of concanamycin B in DHS10676 was obtained by supplementing culture media with 0.6% (w/v) of sodium propionate at 48 h (Fig. S7A). Next, we cultivated wild-type, DHS10673, DHS10674, DHS10675 and DH510676 with or without supplementation of sodium propionate (0.6% (w/v) at 48h). Addition of sodium propionate improved concanamycin B production in engineered strains DHS10675 and DHS10676 (Fig. 7), while very low improvement was observed in wild-type and the remaining engineered strains cultivated under the same conditions (Fig. 7, Fig. S7B and D). Sodium propionate supplementation (0.6%) enhanced concanamycin A and B titers in DHS10675 to 643.1±30.0 mg/L and 208.5±31.7 mg/L, respectively (Fig. 7A, B). In the engineered strain DHS10676 concanamycin A titers was 763.6±83.3 mg/L, while the maximum titer of concanamycin B was 306.5±42.05 mg/L (Fig. 7A, B). Principal component analysis of untargeted metabolomics showed that cultivation with sodium propionate changed the metabolic profile of both engineered and wild-type, reducing the variance between the wild-type and DHS10676 engineered strain to the first principal component (PC1, 55% of total variance) (Fig. 7C), with desferrioxamine E, 23-hydroxyundecylprodiginine, and concanamycin B driving this variance. Moreover, untargeted metabolomics was in concordance with targeted analysis, capturing the increased relative abundance of concanamycin B in the engineered strain DHS10676 and wild-type cultivated with sodium propionate (Fig. S7C).

In the current work, we combined strain optimization with improved isolation and purification methods for concanamycin A, B and C (Table S7, Fig. S8–S14). After cultivation, concanamycins were obtained by solvent extraction followed by multiple rounds of liquid-chromatography (see methods for detailed protocol). Previously methods to extract concanamycin from biomass have used either methanol (Ishii et al 1995; Kinashi et al., 1984) or acetone (Bindseil and Zeeck, 1993; Hayakawa et al., 1991; Kundo et al., 2023; Woo et al. 1992). Though, both have been found to have adverse reactions with concanamycins A-C. Methanol, when used on its own, readily methylates the C-21 hydroxy position of concanamycins A-C (Ingenhorst et al., 2001; Kinashi et al., 1984). On the other hand, acetone causes concanamycins A-C to degrade over extended periods of time (McCauley et al. 2024). To minimize loss of material from undesired solvent effects, a new extraction solvent system was developed. It was found that a solution of dichloromethane (DCM) with 20% methanol was the most efficient method to extract the concanamycins as no unwanted side-reactions were observed after overnight incubation (data not shown). Previous methods also necessitated a liquid-liquid extraction following biomass extraction, but as DCM is not soluble with water like acetone or methanol, no additional extractions were necessary. Yet, isolated titers were lower than quantified titers (using AUC) across all tested strains and conditions (Fig. 7D). We surmise that loss of material occurred at various stages: due to incomplete cell disruption, organic extraction of concanamycins and by the multiple rounds of liquid-chromatography required to separate concanamycin A, B, and C from the other compounds produced by this strain. Nevertheless, purified concanamycin A from DHS10676, cultivated with sodium propionate, reached 483.1±25.8 mg/L, an 86% recovery of the total quantified compound from crude extract (Fig. 7D, Table S7). Supplementation of sodium propionate also boosted isolation of concanamycin B to 159.4±41.8 mg/L in the DHS10676 engineered strain, a 79% recovery of the total quantified compound from crude extract (Fig. 7D, Table S7). Addition of sodium propionate not only increased concanamycins titers, but also decreased the production of several other compounds, including 23-hydroxyundecylprodiginine, 1-hydroxyphenazine, and phenazinomycin (Fig. S7E), thereby facilitating isolation and purification.

4. Conclusion

Here, we report the design of concanamycin A and B high producing Streptomyces strains through rational metabolic engineering approaches. Strain optimization included overexpression of two cms pathway specific regulators, one heterologous regulator, and media optimization, which significantly enhanced the production of concanamycin A and B (>1000-fold, compared to wild-type under non-optimized cultivation conditions). The engineered strain DHS10676 cultivated in shake-flasks, produced the highest concanamycin A titer reported to date (909.8±64.7 mg/L). Supplementation with 0.6% (w/v) of sodium propionate at 48h boosted concanamycin B production to 306.5±42.05 mg/L, a 7.8-fold improvement compared to non-supplemented cultivation conditions. Subsequent efforts to improve the extraction protocols of concanamycin A and B from crude extracts enabled the isolation of 483.1±25.8 mg/L and 159.4±41.8 mg/L (>90% purity), respectively. Beyond S. eitanensis, improved concanamycin A production was also obtained in four different species, S. griseiscabiei, S. scabiei, S. stelliscabiei, and S. neyagawaensis, by heterologous expression of S. lohii bafR regulator and the S. eitanensis cms BGC situated-regulators, cmsR and cmsG (Fig. 5D). Conditional knockdown of cmsR or cmsG, using CRISPRi, completely abolished concanamycin A production, confirming the positive regulation of cms BGC in S. eitanensis. Despite showing that heterologous expression of bafR significantly improves concanamycin production, biochemical characterization of bafR is necessary to uncover the molecular mechanisms of heterologous interaction with the cms BGC. Proteomics analysis showed a significant increase in abundance of biosynthetic proteins from the cms BGC in the engineered mutant (DHS10676) compared to wild-type (Fig. 6A).

Untargeted metabolomics analysis confirmed the metabolic shift between the wild-type and engineered DHS10676 strains, leading to increased production of concanamycin and 23-hydroxyundecylprodiginine, and decreased production of desferrioxamine E, phenazinomycin, 1-hydroxyphenazine, actinomycin Z₃ and several other non-identified metabolites (Fig. 6B, S5C). The molecular mechanisms that regulated this secondary metabolism shift are still unknown.

Constitutive heterologous expression of regulators presents a great strategy to screen for novel producing strains, thus accelerating natural product discovery and the design of resilient natural overproducing strains. Here, we demonstrated that this metabolic engineering approach effectively achieved scale-up production of concanamycin A and B, molecules that hold significant promise as leads for medicinal chemistry studies against HIV-Nef (McCauley et al., 2024; Painter et al., 2020) and other disease targets (Chen et al., 2022).

Highlights.

• Overexpression of cluster-situated regulators enabled ready access to structurally complex plecomacrolides concanamycin A and B.

• Engineered Streptomyces eitanensis strain produced the highest titer of concanamycin A (0.9 g/L) reported to date.

• Overexpression of heterologous cluster-situated regulators is an effective strategy to activate silent and low yield BGC across species.