Multicopy Chromosome Integration and Deletion of Negative Global Regulators Significantly Increased the Heterologous Production of Aborycin in Streptomyces coelicolor

1. Introduction

Aborycin is a class I lasso peptide and was first isolated from Streptomyces sp. SP9440 as a novel anti-HIV metabolite with the origin name RP 71955 [1,2]. It was independently rediscovered from soil S. griseoflavus Tü 4072 as an antibiotic [3], then from deep-sea Streptomyces sp. SCSIO ZS0098 as an anti-infective natural product [4] and from marine sponge-associated Streptomyces sp. MG010 as an antibacterial marker for screening gain-of-function mutants [5]. Aborycin has a typical lasso topology in which the N-terminal 9 amino acids form a macrocyclic ring, and the C-terminal 12 amino acids form a tail that folds back and threads through the ring [2]. Two disulfide bonds between the ring and the tail further increase the structural stability and distinguish the class I lasso peptides from others [6,7].

Lasso peptides are a growing class of bioactive bacterial peptides with unique lasso topology, which differentiates them from other members within the much larger ribosomally synthesized and posttranslationally modified peptide (RiPP) superfamily [8]. The compact and constrained topology endows most lasso peptides with remarkable thermal and proteolytic stability and favors peptide-protein interactions, accounting for the diverse biological activities of lasso peptides, mainly as enzyme inhibitors and receptor antagonists [6,9]. The robust scaffolds of lasso peptides have attracted attention for drug development, such as epitope grafting [10], the incorporation of noncanonical amino acids [11] and protein fusion [12]. While these modifications provide the opportunity to develop novel biological activities with therapeutic potential, they always lead to lower production levels [7]. Moreover, genome mining approaches have greatly accelerated lasso peptide discovery in recent years. Since the first lasso peptide was isolated by genome mining in 2008 [13], the number of lasso peptides discovered by such approaches has steadily increased [9]. By applying the RODEO algorithm, > 1,400 lasso peptide biosynthetic gene clusters (BGCs) were identified from DNA sequence databases, a great increase over the previously known numbers [14]. Although lasso peptide BGCs are widely distributed among bacteria, only approximately 80 lasso peptides were previously characterized [15]. Therefore, both drug development and functional characterization demand effective production systems to explore this rich source of lasso peptides and their modifications.

The heterologous production of lasso peptides in Escherichia coli often provides higher yields than the use of native producers. However, this production advantage in E. coli seems to be confined to lasso peptides from proteobacteria and is less viable for clusters from other phyla [7,9]. Recently, Streptomyces hosts, especially S. coelicolor, S. lividans, and S. albus, have shown significant potential for producing lasso peptides from actinobacteria [14,16,17]. Several technological advances were also achieved for Streptomyces systems, for example, CRISPR/Cas9 genome editing technology [18,19] and the multiplexed site-specific genome engineering (MSGE) method [20]. Genome editing offers a rapid way to modify regulatory elements involving secondary metabolism and therefore impact heterologous production. The MSGE method led to the successful development of a panel of S. coelicolor heterologous hosts, in which up to five copies of BGCs could be integrated into the specific sites of the host chromosome in a single step, leading to significant yield improvements [20]. In our previous study, an aborycin BGC was identified from a marine sponge-associated Streptomyces sp. HNS054 [5]. This provided an opportunity to produce this lasso peptide heterologously in Streptomyces systems. Thus, in this study, a Streptomyces system for aborycin production that is compatible with up-to-date technologies was established. By increasing the integrated copy numbers of the aborycin BGC in the host chromosome and by deleting the negative global regulatory genes involved in secondary metabolism by genome editing, the production of aborycin was significantly improved. This study provides a useful reference to improve Streptomyces systems for the production of lasso peptides and their modifications and thus benefits drug development and functional characterization.

2. Results

2.1. General information on the aborycin gene cluster from Streptomyces sp. HNS054

As previously reported [5], the aborycin gene cluster gul (named after Gulei Town, where the host marine sponge was collected from its coastal sea) from marine sponge-associated Streptomyces sp. HNS054 consists of 14 ORFs spanning an approximately 12 kb region from gulR1 to gulR2 (Figure 1). Sequence alignment showed that the DNA sequence of this region shares 98% identity to the aborycin gene cluster abo from Streptomyces sp. SCSIO ZS0098 [4]. The gul cluster also shows high identity to the siamycin-I gene cluster msl from Streptomyces sp. M-271 [21] in gene sequence and gene organization (Figure1). GulA encodes a 42-residue peptide with a leader peptide at its N-terminus and a 21-residue core peptide (CLGIGSCNDFAGCGYAVVCFW) at its C-terminus. The primary and secondary structures of aborycin and siamycin-I are almost the same, except that the residues at the 4th and 17th positions are switched (Figure1). Due to the high similarity between aborycin and siamycin-I, commercial reagents of siamycin-I could be used as a reference to locate the high performance liquid chromatography (HPLC) signal of aborycin.

2.2. Cloning of the gul BGC and construction of strains for heterologous expression of aborycin

The cloning procedure of the gul BGC is shown in Figure S1. The 14 kb fragment containing the gul BGC was amplified from Streptomyces sp. HNS054 by high-fidelity PCR and subsequently cloned and inserted into the integrative vector pSAT209, yielding pSAT-GUL (Figure S2). The plasmid was fully sequenced, and the results confirmed that no mutation occurred within the plasmid frame and the gul ORFs. Then, the plasmid pSAT-GUL was transferred to S. coelicolor strains M1146, M1246, M1346, M1446 and M1546 by conjugation, respectively. As a blank control, the empty plasmid pSAT209 was also transferred to S. coelicolor M1146. The numbers of successful attP-attB recombinations in exconjugants were determined by PCR. The results showed that strains M1146::gul (Figure 2A), M1246::2gul (Figure 2B) and M1346::3gul (Figure 2C) were successfully constructed. However, four-copy and five-copy integrations were unsuccessful after four rounds of conjugation and screening.

2.3. Extraction and detection of aborycin

With siamycin-I as the control, the M1346::3gul strain was used to test whether aborycin was successfully produced. Aborycin was extracted as shown in Figure 3A. The butanone extract had an HPLC signal at 23.2 min, the same as the signal of siamycin-I (Figure 3B). In the following test of the macroporous adsorption resin method, the AB-8 resin showed high enrichment efficiency for aborycin, equivalent to that of the butanone method. Mass spectrum analysis confirmed that a substance with a relative molecular mass of m/z = 1082.43 (z = 2) accounted for the majority of the composition at the 23.2 min peak of the AB-8-enriched samples (Figure 3C). The MS data of this substance were perfectly matched to the aborycin data from a previous report [5]. Thus, two facts were confirmed by these tests. First, the M1346::3gul strain can express aborycin. Second, compared with the butanone method, the AB-8 resin method offered a safer, easier and cheaper method to enrich aborycin from the fermentation supernatant. This method was thus applied in the following quantification studies. After pure aborycin was isolated from 5.6 L culture broth of the M1346::3gul strain, a precise standard curve of concentrations to the HPLC peak areas of aborycin was obtained (Figure 3D). By applying this curve, aborycin production from different strains or conditions could be quickly quantitated by HPLC.

2.4. Yield comparison among different strains

Titers of aborycin from different strains are shown in Figure 4. Adding the supernatant and mycelial products, the native HNS054 strain average yielded 4.9 ± 2.6 mg/L aborycin. No aborycin signal was detected from the blank control M1146::pSAT209 and the one-copy strain M1146::gul. Signals became obvious for the M1246::2gul and M1346::3gul strains, which had average titers of 1.8 ± 1.0 mg/L and 10.4 ± 1.6 mg/L, i.e., 0.4 times and 2.1 times that of the native strain, respectively. Both supernatant and mycelia contributed to the production, accounting for approximately 38 percent and 62 percent, respectively.

Then, the M1346::3gul strain was selected for genetic modification to verify whether aborycin production could be further improved by deleting negative regulatory genes (Table 1), which were reported to act at higher levels of the gene-regulatory networks to control secondary metabolism, whose knockout mutants always resulted in an increase in secondary metabolism. The ΔphoU, ΔwblA, ΔSCO1712, ΔorrA and ΔgntR mutants were successfully constructed (Figure S3, S4). A titer study showed that the aborycin production of the M1346::3gul ΔphoU strain was 12.1 ± 0.7 mg/L, which was slightly higher than that of the M1346::3gul strain but not significantly different (P > 0.05). The ΔwblA strain was significantly (P < 0.01) higher than the M1346::3gul strain, with a titer of 23.6 ± 10.2 mg/L, i.e., 4.8 times that of the native strain or 2.3 times that of the strain before wblA knockout. Greater improvements were displayed in the ΔorrA and ΔgntR mutants, which had titers of 56.3 ± 18.0 and 48.2 ± 15.1 mg/L, respectively (Figure 4).

3. Discussion

In this study, efforts were made to use the Streptomyces system for the heterologous production of aborycin. First, the cloning procedure was simplified. The 14 kb genomic fragment containing the gul gene cluster was amplified by high-fidelity PCR, and sequencing confirmed its accuracy. High-fidelity PCR techniques have been developed to amplify long DNA fragments with lengths as long as 15-20 kb. The cloning of RiPP gene clusters could benefit from these developments because a large portion of RiPP gene clusters have lengths below 15 kb (Table S3). In a recent review on newly discovered RiPPs, out of 33 listed gene clusters, 25 (76%) had lengths below 15 kb ([27], Table S4). Second, to facilitate foreign DNA integration following genome editing, the integrative vector pSET152 was modified to pSAT209 to prevent antibiotic resistance conflict. Third, the extraction of aborycin was optimized. Aborycin is a peptide with amphiphilic characteristics and thus reversible affinity to certain macroporous adsorption resins. In this case, AB-8 resin was found to be the matched resin to adsorb aborycin in water solution and desorb it in methanol. By simply adsorbing and desorbing, approximately 40% of the total production was recovered from the culture supernatant. A similar situation occurred with the heterologous expression of sviceucin, where approximately 1/3 of the product was released in the culture supernatant [16]. Large-scale production would discard the supernatant because of the high cost of solvent extraction. With this optimization, the product in the supernatant could be recovered at a low cost. We speculated that this theory and operation could be applied to other RiPPs due to their peptide nature. Finally, multiple chromosomal integration of foreign gene clusters following CRISPR/Cas9 genome editing was successfully achieved, and the best case resulted in a 25-fold increase in aborycin production compared to the native strain.

Multiple chromosomal integration of foreign BGC was proven to be a mature technique to increase heterologous expression. The approach was successfully demonstrated in S. coelicolor [20], S. albus J1074 [28] and S. lividans [29]. In this study, one-copy integration showed no signal, while three-copy integration showed a 2.1-fold increase compared with the native strain. However, more advanced techniques were needed to obtain 4- or 5-copy integration in one single step. In our attempts to transfer other BGCs into these M1146-M1546 hosts, the number of exconjugants decreased rapidly with increasing integrated copy number (data not shown). It was previously speculated that a high copy number of chromosomal integration caused the accumulation of target products, endangering bacterial growth [20]. However, the titer of aborycin was further improved 5.4-fold by orrA gene knockout, implying that the Streptomycete actually tolerate higher concentrations of aborycin.

CRISPR/Cas-based genome editing tools provide swift, accurate and traceless ways to modify the genomes of Streptomyces [30]. A straightforward way to exploit the CRISPR/Cas9 tools was to delete negative regulator genes. This study proposed to delete genes that were reported to have a negative impact on secondary metabolism at a high level of the regulation networks. Five mutants, ΔphoU, ΔwblA, ΔSCO1712, ΔorrA and ΔgntR, were successfully constructed from the M1346::3gul strain.

Although the detailed function of the phoU gene was unclear, it was speculated to be involved in the pho regulon, which responds to phosphate starvation. PhoR senses such conditions, and then PhoP is phosphorylated following PhoP-P binding to specific sequences named PHO boxes, thus activating or repressing a set of genes [31]. Under phosphate starvation conditions, the ΔphoU mutants showed an approximately 6-fold increase in the production of actinorhodin [22]. Our study showed that aborycin production in the ΔphoU mutants was not significantly improved from that in the strain before mutation when strains were cultured in the R5 medium. Moreover, the growth of the ΔphoU mutants on MS-agar was retarded in a 9-day morphology observation (Figure 5). This type of the mutants requires more optimization before it can be applied in antibiotic production.

The wblA gene was reported as a pleiotropic downregulator of antibiotic biosynthesis in S. coelicolor. ΔwblA mutants exhibited a defect in sporulation, achieved higher biomass than the wild-type, and overproduced secondary metabolites [32]. Overproduction of antibiotics by disruption of the wblA orthologs was also observed in other Streptomyces bacteria [33]. This study confirmed that the production of aborycin in the ΔwblA mutant was significantly improved by 2-fold. Higher growth and defects in sporulation were also observed (Figure 5).

SCO1712 is a member of the TetR family. TetR family transcriptional regulators are among the most common prokaryotic transcriptional regulators. When SCO1712 was overexpressed or disrupted, ACT production decreased or increased compared with that in S. coelicolor M145, respectively, suggesting that SCO1712 is a pleiotropic downregulator of antibiotic biosynthesis in S. coelicolor [24]. It was further speculated that there is a synergistic effect between SCO1712 and precursor flux pathways in antibiotic production [34]. Unfortunately, in this study, after SCO1712 knockout, the production of aborycin decreased significantly. This provides ideas for future optimizations on precursor flux pathways or mediums for this mutant.

OhkA (SCO1596) - OrrA (SCO3008) is a group of prokaryotic two-component regulatory systems with highly similar transcriptomic features. ΔorrA mutants lead to significant overproduction of antibiotics and downregulation of bld, chp, rdl, and wbl genes associated with morphological development [25]. Correspondingly, we found that ΔorrA mutants overproduced approximately 5.4-fold of aborycin than the strains before mutation (Figure 4). Their morphological development also was similar to ΔwblA mutants (Figure 5). These results implied that the OrrA regulatory system likely covered the wbl regulatory system and controls a wider range of resources for antibiotic production.

The bacterial GntR family is one of the most abundant groups of helix-turn-helix transcription factors that respond appropriately to metabolite micro-environments [35]. It was reported that deletion of a GntR-like gene allowed platensimycin and platencin overproduction in S. platensis [36]. GntR (SCO1678) of S. coelicolor encodes a repressor protein to control the gluconate operon, which enable Streptomyces to utilize gluconate in the mediums [26]. No obvious evidences linked GntR (SCO1678) to antibiotic overproduction to date. Interestingly, both overproduction of aborycin (Figure 4) and overgrowth (Figure 5) were observed from the ΔGntR mutants in this study. No additional gluconate were added in the R5 or MS-agar mediums to obtain these results. These observations provide a significant gene that is worthy of further studies to discover underlying metabolic regulation mechanisms.

Although great production was achieved by these genetic modifications, we believe the Streptomyces systems could be further improved. With more global regulators that govern secondary metabolism being characterized [37], it is worth manipulating them one by one to obtain further knowledge.

4. Materials and Methods

4.1. Strains, plasmids and primers

The strains and plasmids used in this study are listed in Table S1, and the primers used in this study are listed in Table S2. The construction of the related vectors is shown in Figure S1 and Figure S2.

4.2. Construction of heterologous expression strains to produce aborycin

Foreign DNA integration could be fulfilled by the integrative plasmid pSET152 [38], and genome editing could be executed by the suicide plasmid pKCcas9 [19]. However, both vectors contain the apramycin resistance gene aac(3)IV. To achieve BGC integration following genome editing, the antibiotic-resistance markers between the two genetic manipulations should be different. Thus, the aac(3)IV gene in pSET152 was replaced by the ampicillin resistance gene bla from the pUC57 plasmid and the thiostrepton resistance gene tsr from the pGM1190 plasmid. The resulting plasmid was named pSAT209 (Figure S1). Genomic DNA of Streptomyces sp. HNS054 was extracted from the later logarithmic phase cells by a bacterial genomic DNA extraction kit (Cat # DP2001, Biotake, Beijing, China). Using primers 054LasF and 054LasR, a 14 kb DNA fragment containing the gul BGC (located at 15932-29864 nt of GenBank: AC003_RS35325) was amplified from the Streptomyces sp. HNS054 genome (Assembly: GCF_001044185.1). After gel purification, the gul fragment was assembled with EcoRV-linearized pSAT209 by a one-step cloning kit (Cat # C113, Vazyme, Nanjing, China). The resulting vector pSAT-GUL (Figure S1) was sequenced to confirm the accuracy. Then, it and the control plasmid pSAT209 were transferred into E. coli ET12567/pUZ8002 competent cells and conjugated into S. coelicolor M1146-M1546 hosts by standard protocols (Kieser et al. 2000). Exconjugants were randomly picked to check attP-attB recombination by PCR with the following primers. The forward primer ID-oriT-fw targets a position approximately 400 bp upstream of the attP locus in the vector pSAT-GUL. The reverse primers, ID-native-attB-Rev, ID-CPK-attB-Rev, ID-RED-attB-Rev, ID-CDA-attB-Rev and ID-ACT-attB-Rev, each target a position downstream of an individual attB locus in the S. coelicolor M1546 genome. The integrated copy number of the gul gene cluster in the host genome was determined by the number of positive attP-attB recombinations [20]

4.3. Gene knockout in S. coelicolor M1346::3gul by the CRISPR/Cas9 method

The phoU (SCO4228), wblA (SCO3579), SCO1712, orrA (SCO3008) and gntR (SCO1678) genes (Table 1) in S. coelicolor M1346::3gul were independently knocked out by the CRISPR/Cas9 genome editing method [19]. To generate the ΔphoU mutant, the sgRNA expression cassette (f1 fragment) was PCR amplified with the plasmid pKCcas9dO as the template and f1J01-fwd/f1gRNA-R as primers. The upstream (f2 fragment) and downstream (f3 fragment) regions of the phoU gene were PCR amplified with the primer pairs f2J01-fwd/rev and f3J01-fwd/rev, respectively. Then, three DNA fragments were assembled by overlapping PCR using primers f1J01-fwd and f3J01-rev. The resulting DNA fragment was cloned and inserted into pKCCas9 using the one-step cloning kit (Cat # C113, Vazyme, Nanjing, China) to yield pKCCas9-dJ01. The obtained plasmid was introduced into the M1346::3gul strain by conjugal transfer. The correct knockouts were verified by PCR using primers ID-dJ01-fwd/rev. By using the J04, J05, J07 and J08 sets of primers and the same procedure, pKCCas9-dJ04, -dJ05, -dJ07 and -dJ08 were generated and yielded the ΔwblA, ΔSCO1712, ΔorrA and ΔgntR mutants after successful conjugation into the M1346::3gul strain, respectively. The positions of the spacers, primers and target genes are shown in Figure S3.

4.4. Metabolite analysis

The wild-type Streptomyces sp. HNS054 and the S. coelicolor strains were grown on MS solid medium to achieve sporulation. Approximately 10⁸ spores (or 40 μL mycelia store) were transferred to 40 mL R5 medium with appropriate antibiotics and cultured at 200 rpm and 28 °C for 3 d as the seed solution. Ten milliliters of the seed was transferred to 500 mL of R5 medium and cultured at 200 rpm and 28 °C for 9 d to complete the fermentation. Aborycin was extracted as shown in Figure 3A. The fermented broth was centrifuged to separate the supernatant from the mycelia and then independently extracted. The supernatant was extracted by two methods. The first method was butanone extraction three times, and then the butanone phase was concentrated by rotary evaporation. Then, four macroporous adsorption resins, AB-8, DC201-C, NK-9 and D101 (Hecheng New Material, Zhengzhou, China), were tested for their aborycin recovery efficiencies. It was found that the AB-8 resin had a high enrichment ratio for aborycin (Figure 3B), and AB-8 was therefore used as the second method to extract aborycin. Because the AB-8 resin method was safer, easier and cheaper, all aborycin titers of the supernatants in this study were obtained by this method. The mycelium was extracted three times with 300 mL methanol and then concentrated by rotary evaporation. Extracts from both the supernatant and mycelium were adjusted to a final concentration of approximately 5 mg/mL in methanol and filtered with 0.22 µm filter membranes before HPLC or LC‒MS analysis. Analytical HPLC was performed with a Shimadzu Prominence LC-20A (Shimadzu, Tokyo, Japan) using a 2.6×250 mm Ultimate XB-C18 column (Welch, Shanghai, China). The elution conditions were 1% solvent B for 0-7 min and then a linear gradient to 95% solvent B (solvent B: 0.1% formic acid + 100% CH₃CN; solvent A: 0.1% formic acid in H₂O) for 7-30 min at a flow rate of 1 mL/min and monitored at 277 nm. To locate the aborycin signal, the commercial siamycin-I reagent (Adipogen Life Sciences, San Diego, USA) was used as the control. Because the amino acid sequences between aborycin and siamycin-I were the same except that the residues at the 4th and 17th positions were switched (Figure 1), it is reasonable to expect the HPLC signal position of aborycin to be slightly different from that of siamycin-I. LC‒MS analysis was performed with a Q Exactive spectrometer (Thermo Scientific, USA) by the same elution program as HPLC. The software MZmine 2 [39] was used to assess the chromatograms.

Purification of aborycin and preparation of the calibration curve

A 5.6 L culture broth of the M1346::3gul strain was prepared as described in the previous section. The crude extract was enriched by the AB-8 resin method from the supernatant and then purified by semipreparative HPLC with a YMC-PACK ODS-A column (10×250 mm, φ 5 μm, YMC, Kyoto, Japan). F1-F4 fractions near the target position were collected, and the F2 fraction was determined by analytical HPLC to be the major fraction containing aborycin. Then, approximately 22 mg of the F2 fraction was obtained after rotary evaporation. Ten milligrams of the F2 fraction was dissolved in 2 mL of methanol, and analytical HPLC was run as described in the previous section. The 23.2 min peaks were collected and concentrated to approximately 4 mg white powder of pure aborycin. A methanol solution of aborycin was prepared with a precise concentration of 1.00 mg/mL and then diluted to a series of concentrations. These dilutions were subjected to analytical HPLC, and the 23.2 min peak areas were recorded. A standard curve of concentrations to HPLC peak areas of aborycin was then constructed (Figure 3D), and the following formula was deduced.

$$\mathbf{y}=\mathbf{4.970}\mathbf{x}+\mathbf{0.046}\hspace{1em}\hspace{1em}({\mathrm{R}}^2=0.999)$$

(1)

where y (10⁶∙mAU∙min) is the peak area of the HPLC signal of aborycin, and x (mg/mL) is the concentration.

5. Conclusions

This study demonstrated improvements in the S. coelicolor system for aborycin overproduction. The maximum shake flask titer was over 50 mg/L, comparable to the best examples of E. coli systems or Streptomyces systems for lasso peptide heterologous expression [7]. We confirmed that increasing the copy number of chromosome integration and disrupting the negative global regulators governing secondary metabolism were effective ways to improve antibiotic production in S. coelicolor systems. GntR (SCO1678) was proposed to be a significant target that is worthy of further studies on metabolic regulation mechanisms. The optimized extraction method, which could efficiently recover aborycin from the supernatant, also contributed to the total titer. Due to the similar chemical properties of RiPPs to aborycin, this tactic is also suitable for the development of other RiPP relatives. By applying these modifications to Streptomyces systems, the drug development and functional characterization of new lasso peptides from genome mining will be greatly improved.