INTRODUCTION
Actinomycetes, particularly
Streptomyces species, are among the most productive sources of bioactive natural products in medicine and agriculture
1. Their large, guanine–cytosine (GC)-rich genomes encode extensive repertoires of biosynthetic gene clusters (BGCs), many of which remain silent or poorly expressed under routine laboratory conditions, leaving substantial chemical diversity inaccessible to conventional cultivation-based discovery. This untapped biosynthetic capacity has made actinomycetes a central chassis for natural-product mining and pathway engineering, but it has also exposed a persistent technical barrier: many strains of practical interest remain genetically difficult to manipulate because of low DNA delivery efficiency, slow growth, complex developmental programs, and the limited portability of existing gene editing systems across non-model hosts
2-
6.
Site-specific recombination (SSR) systems have therefore become foundational tools in actinomycete genetics. Among them, the phage φC31 integrase has been especially influential because it catalyzes efficient and stable recombination between
attP and
attB sites, and can also exploit pseudo-
attB sites distributed across
Streptomyces genomes, making integrative vectors highly valuable for stable gene insertion and heterologous BGC expression
7-
9. Cre-
loxP-based strategies further extended this framework by enabling removal of selectable markers or dispensable vector sequences, thereby supporting the generation of marker-free mutants after integration of genes or gene clusters
10-
12. In existing implementations, however, Cre is still delivered in a separate genetic step, for example by an additional plasmid or transient phage-mediated introduction, thereby increasing both time and labor costs
10,
12,
13. Conversely, when Cre is placed on the same construct and controlled by an inducible promoter, leaky expression creates a practical risk of early recombination, leading to unwanted plasmid rearrangement or excision before the intended engineering step has been completed. These limitations are particularly consequential when the goal is not only stable integration but also post-integration removal of unnecessary vector sequences. Although recent reported recombinase-mediated cassette exchange (RMCE) and Micro-host engineering platform (HEP) systems can realize backbone-free integration of BGCs, they both rely on prior installation of
loxP sites in the recipient genome. Such a prerequisite is inherently unfriendly to genetically recalcitrant strains, because the host must first undergo an additional round of engineering before the desired integration event can even be performed
14,
15.
The emergence of CRISPR-based gene editing provides an unprecedented opportunity to accelerate genetic manipulation in actinomycetes. Cas9-based systems provide powerful targeted mutagenesis in actinomycetes, yet their application to genetically recalcitrant actinomycetes is often constrained by delivery barriers and the burden of double-strand-break repair
16,
17. Base editors (BE) partially solve this problem by enabling programmable single-base mutations without generating double-strand cleavage, and establish that both cytosine and adenine base editing can be deployed efficiently in actinomycetes
18-
22. For strains that are difficult to transform or poorly amenable to conventional CRISPR workflows, integrative vectors offer an attractive route to deliver editing machinery into the genome; however, once integrated, the gene editing cassette cannot be removed readily, leaving residual foreign DNA in the genome and raising downstream safety, regulatory, and strain-stability concerns
23. There is therefore a clear need for a streamlined workflow that combines efficient plasmid delivery, high-efficiency gene editing, and subsequent removal of unnecessary vector sequences in a single controllable platform. Recent recombinase-based engineering frameworks further illustrate both the power of site-specific integration and the demand for cleaner, more modular actinomycete engineering systems
12,
13,
24.
Recent advances in genome engineering have further expanded the strategies available for activating silent BGCs in
Streptomyces. Programmable transcriptional regulation using CRISPR interference and activation (CRISPRi/CRISPRa) has enabled direct rewiring of endogenous regulatory networks to stimulate natural-product biosynthesis
25. In parallel, CRISPR-assisted BGC capture and heterologous expression have facilitated the activation and functional characterization of large cryptic gene clusters
26,
27. More recently, multiplexed promoter refactoring and repurposing of endogenous CRISPR-Cas systems have provided additional routes for coordinated activation and reconstruction of complex BGCs in
Streptomyces28,
29. Together, these approaches substantially broaden the capacity for natural-product discovery and pathway engineering. Nevertheless, most such strategies primarily focus on activating or reconstructing biosynthetic pathways, whereas the efficient removal of editing components after genome modification remains an important consideration for iterative strain engineering. In this context, a genome-engineering platform that couples efficient genetic modification with subsequent elimination of the editing-vector backbone would provide a complementary strategy for constructing genetically cleaner and more readily re-engineered
Streptomyces strains.
Here, we address this need by developing a comprehensive platform, named erasable base editor (EraBE), that combines a site-specific integrating vector, a base editing system, and Cre-loxP recombination system within one design. The key innovation of EraBE is the introduction of a function-disrupting amino-acid mutation into Cre recombinase to generate an inactive dead Cre (dCre) state. This mutation is deliberately designed to be correctable by cytosine or adenine base editing. As a result, regardless of how strongly dCre is expressed, it cannot trigger early recombination. Only upon induction of an sgRNA that restores the mutated codon does the base editor convert dCre back to functional Cre, enabling precise excision of the dispensable plasmid backbone from the chromosomal integration locus. This design compresses efficient plasmid delivery, precise gene editing, and controlled plasmid backbone erasure into a single genetic manipulation. Leveraging the widespread distribution of SSR systems in actinomycetes, EraBE should in principle be deployable across a broad range of actinomycete hosts. By integrating the base editor or cargo-delivery module through an SSR-based vector, EraBE enables genome editing or target gene introduction to be completed within a single round of genetic manipulation, followed by the safe and controllable excision of the plasmid backbone from the chromosomal integration locus. This design minimizes permanent retention of unnecessary vector sequences and is therefore expected to substantially improve the genomic stability of the resulting engineered strains.
We validated EraBE in both model and non-model actino-mycetes, in which it supported highly efficient gene editing or BGC introduction, while the rate of inducible plasmid-backbone erasure reached up to 100%. Corresponding EraBE-engineered mutants achieved high-production of multiple natural products. Our findings indicate that EraBE can provide a versatile route for “hit-and-run” genome engineering, and be compatible with heterologous expression or enhancement of natural product biosynthesis.
RESULTS
Site-specific recombination system has high programmability in actinomycetes
SSR systems have been widely exploited in actinomycetes, and the availability of multiple integrative platforms, including those derived from φC31, φBT1, φJoe, VWB, SV1, μ1/6, TG1, R4, and pSAM2, highlights the breadth and maturity of this genetic toolkit (Supplementary Fig. S1)
24,
30,
31. To evaluate the potential host coverage of SSR-based engineering, we collected all genome assemblies available in the National Center for Biotechnology Information (NCBI) RefSeq database for the phylum Actinomycetota as of March 14, 2025, comprising 42,942 assemblies in total. As genome fragmentation may increase ambiguity in the identification and enumeration of short
attB-like sequences, we retained only assemblies containing fewer than five contigs, resulting in 7,735 genomes. The corresponding RefSeq accession numbers are provided in Supplementary Information (Supplementary Data 1). We then examined the distribution of
attB sites corresponding to nine representative SSR systems and found that 98.24% of these genomes contained at least one cognate
attB site (Fig. 1a). Among the systems analyzed, the pSAM2
attB site was the most prevalent, occurring in 96.28% of genomes, whereas φBT1 showed the lowest representation at 13.32%; the widely used φC31
attB site was present in 28.18% of genomes (Fig. 1b). In addition, 76.7% of the genomes contained at least two
attB sites, and 27.03% contained five or more, indicating substantial combinatorial integration potential across actinomycetes (Fig. 1c).
Together, these results indicate that SSR systems are both broadly distributed and highly deployable, supporting the development of SSR-based genome engineering tools with broad host applicability, including genetically recalcitrant strains. At the same time, the unavoidable retention of vector backbone sequences remains a fundamental limitation of conventional integrative plasmids. This combination of exceptional applicability and persistent structural burden provided the rationale for developing EraBE, an SSR-enabled actinomycete engineering platform designed to preserve the delivery efficiency and generality of integrative vectors while supporting genome editors or functional cargos without permanent plasmid-backbone retention. Based on these findings, we also established a web-based tool for attB-site prediction, available at website (attb-find.com/), to assist users in selecting appropriate SSR systems for genetic manipulation.
Design and construction of EraBE
EraBE was designed as a three-module platform comprising plasmid integration, base editing, and plasmid erasure. For proof-of-concept validation, the integration module was built on the widely used φC31-based integrative shuttle vector pSET152
7.
The editing module was configured either as a C-to-T cytosine base editor or an A-to-G adenine base editor. The cytosine base editor (CBE) version, designated EraCBE, incorporated eSCBE3-NG-Hypa, an NG-PAM-recognizing editor optimized for both efficiency and fidelity in
Streptomyces22, whereas the adenine base editor (ABE) version, designated EraABE, was generated by replacing the cytidine deaminase in eSCBE3-NG-Hypa with the hyperactive adenine deaminase TadA8e and removing uracil DNA glycosylase inhibitor (UGI), yielding a derivative termed SABE8e-NG-Hypa (Fig. 2a)
32. For the plasmid-erasure module, two directly repeated
loxP sites were positioned immediately upstream and downstream of the
attP locus, so that, following
attP-
attB-mediated DNA recombination, the entire plasmid backbone became flanked by
loxP sites and could in principle be excised by Cre recombinase (Fig. 2b).
In fact, design of plasmid-erasure module was motivated by our experimental observation that direct placement of wild-type Cre under an inducible promoter was insufficient to ensure plasmid stability: even a promoter with low basal leakage readily caused premature recombination during plasmid maintenance in
Escherichia coli (
E. coli)(Supplementary Fig. S2)
33,
34. To overcome this limitation, we reasoned that, if leaky Cre expression could not be completely eliminated, Cre could instead be rendered catalytically inactive and subsequently reactivated through a controllable, orthogonal gene-editing event. This strategy shifts the critical control point from Cre expression to Cre reactivation, thereby enabling the safe and tightly controlled induction of Cre-mediated plasmid erasure. Previous structural and
in vitro biochemical studies have demonstrated that several amino acid residues in Cre recombinase are critical for its catalytic activity
35. Among these, Arg173 and His289 are located within the catalytic region of Cre and have been identified as essential residues for efficient phosphodiester-bond cleavage and strand-exchange reactions
35. Accordingly, designed H289R and R173Q mutant codons could be restored to the wild-type codons by the CBE and ABE modules, respectively, with compatible protospacer adjacent to NG PAMs. These combined catalytic and sequence-design constraints determined the two dCre variants, H289R and R173Q, used in EraCBE and EraABE (Fig. 2c). Corresponding H289R and R173Q substitutions were introduced in synthesized Cre DNA fragments by polymerase chain reaction (PCR) and Gibson assembly. This design makes sequence explicit: oxytetracycline (OTC) induces erasing-sgRNA expression; the erasing sgRNA directs base editing of the mutant dCre codon; corrected dCre regains Cre activity; and active Cre excises the
loxP-flanked backbone.
Moreover, an orthogonal erasing sgRNA cassette driven by the inducible promoter Potr* was then incorporated to target the mutated codon in dCre; upon OTC induction, the sgRNA, together with the cognate CBE or ABE, restored the original amino acid residue and thereby reactivated Cre, coupling plasmid-backbone excision to base-editing-dependent rescue of recombinase activity (Fig. 2c). Consistent with this design, EraBE plasmids retained the correct architecture after 12 serial passages at 12-h intervals in E. coli, indicating high structural stability (Supplementary Fig. S3). Although leaky expression of the orthogonal erasing sgRNA cannot be excluded, unintended activation of dCre would additionally require concurrent leakage of the base editor and productive correction of the inactivating mutation; our results indicate that this compound event occurs at very low frequency, supporting the robustness of the EraBE design.
To increase the efficiency of recovering plasmid-erased mutants, EraBE additionally carries the counter-selectable marker
pheSA339G/T278A. In the presence of 4-chloro-phenylalanine, strains harboring
pheSA339G/T278A misincorporate this analogue into cellular protein synthesis, resulting in cell death and thereby enriching for clones that have lost the plasmid backbone (Fig. 2b)
36.
EraBE enables efficient gene editing and the acquisition of plasmid-free mutants
To evaluate the performance of EraBE in a model actinomycete host, we examined both the editing and plasmid-erasure modules in Streptomyces coelicolor (S. coelicolor) M145. For assessment of base-editing efficiency, two NGN PAM-containing protospacers were selected in the S. coelicolor genome for targeting by EraCBE and EraABE, respectively. After cloning the corresponding target sgRNAs (Supplementary Fig. S4), the plasmids were introduced into the host by E. coli–Streptomyces intergeneric conjugation (Fig. 3a). Exconjugants were obtained by antibiotic selection, and several colonies were subsequently expanded and subjected to resistance verification to ensure successful chromosomal integration of the plasmid (Fig. 3a). Genomic DNA was then extracted and the editing outcomes at the target loci were analyzed by Sanger sequencing (Fig. 3a). Both EraCBE and EraABE produced clear single-base conversions at the intended sites, with maximal editing efficiencies reaching 100% for individual bases (Fig. 3b). At the EraCBE target PSP_CGG-3, mixed sequencing peaks were detected, indicating coexistence of edited and unedited genomes within a single exconjugant. Following streak purification, a clonal population carrying the desired edit with a uniform genotype was successfully isolated (Fig. 3b). To remove the integrated plasmid backbone, the edited strains were streaked on plates containing 2 μg/mL OTC and 10 μg/mL 4-chloro-phenylalanine, thereby inducing orthogonal erasing sgRNA expression, activating dCre rescue and simultaneously applying counter-selection. To distinguish quantitative backbone erasure from the subsequent strain-purification procedure, we defined backbone-erasure efficiency based exclusively on random colony PCR screening after the first round of OTC induction and counter-selection. Single colonies recovered from these plates were expanded, and plasmid-erasure events were examined by PCR using primers flanking the chromosomal attB integration site. Successful backbone excision was expected to yield a 639-bp amplicon (Supplementary Fig. S5a). Consistent with this prediction, the 639-bp band was detected in both the EraCBE- and EraABE-treated groups: 67.78% ± 22.69% (mean ± SD) clones were positive in the EraCBE group (Fig. 3c; Supplementary Fig. S5c), whereas 93.33% ± 6.67% (mean ± SD) clones were positive in the EraABE group (Fig. 3c; Supplementary Fig. S5d).
Notably, a 505-bp band was also observed. Sequencing showed that this product was identical to the wild-type genomic configuration, despite prior antibiotic verification of plasmid integration (Supplementary Figs. S5c, d). The appearance of the 505-bp PCR product may result from low-frequency reverse excision of the integrated vector. Although φC31 integrase is generally considered to favor unidirectional
attP ×
attB integration, recent studies have shown that continued integrase expression can mediate low-frequency
attL ×
attR excision in
Streptomyces37. RT-PCR analysis of strains carrying integrated pSET152, EraCBE, or EraABE detected the expected 1,531-bp
int transcript, whereas no product was detected in the corresponding no-RT controls, confirming that
int remained transcriptionally active after chromosomal integration (Supplementary Fig. S6). These observations suggest that continued integrase expression may contribute to the formation of the 505-bp product by promoting partial re-excision of the integrated vector.
This finding suggests that continued expression of the integrase after chromosomal integration may have promoted excision events that restored the wild-type genotype independently of the intended plasmid-erasure route
37. The subsequent counter-selection step was used primarily to facilitate purification of backbone-erased mutants rather than to quantify erasure efficiency. Specifically, colonies showing a high proportion of the excised genotype in the first-round PCR screen were preferentially selected for an additional round of streak purification on 4-chloro-phenylalanine-containing medium. Because this purification step involved intentional selection based on PCR profiles rather than random sampling, it was not included in the calculation of backbone-erasure efficiency. After yielding mutants with the correct PCR pattern and a uniform genotype (Fig. 3d), Sanger sequencing further confirmed that the final mutants had undergone plasmid-backbone excision of the
loxP-flanked region, including the bacterial replication/transfer elements, antibiotic-selection cassette, integrase cassette, base-editor cassette, dCre cassette, erasing and targeting sgRNA modules, and counter-selectable marker (Figs. 2b, 3e). The final locus retains a single
loxP scar and an
att-derived recombination junction (Fig. 3e).
To assess off-target effects associated with EraCBE and EraABE, we performed whole-genome resequencing on two plasmid-free S. coelicolor mutants, M145-CGG-3 and M145-CGA-2, generated following EraCBE- and EraABE-mediated editing, respectively. The results indicated that EraCBE caused evident genome-wide off-target mutagenesis, whereas EraABE did not produce substantial off-target effects (Supplementary Data 2; Fig. 3f). Meanwhile, neither system led to appreciable indel formation (Supplementary Data 2; Fig. 3f). The higher off-target burden observed with EraCBE may reflect the stronger activity of its deaminase component together with a greater susceptibility of the corresponding sgRNA target to mismatch-tolerant binding. These findings suggest that, from the standpoint of editing safety, EraABE provides a more suitable framework for future expansion into additional actinomycete engineering applications.
To further evaluate the portability of EraBE beyond
Streptomyces, we tested the system in
Micromonospora echinospora (
M. echinospora) ATCC15835, a gentamicin-producing rare actinomycete that is relatively difficult to genetically manipulate and has been reported to exhibit very low transformation efficiency
38. EraCBE was successfully integrated into the chromosome of
M. echinospora ATCC15835, and antibiotic-resistant integrants were subsequently transferred to medium containing 2 μg/mL OTC and 10 μg/mL 4-chloro-phenylalanine to induce backbone erasure and counter-selection. PCR screening confirmed successful removal of the integrated vector backbone, with an erasure efficiency of 13.33% ± 6.67% (mean ± SD) (Fig. 3g, h; Supplementary Fig. S5e). Although this efficiency was substantially lower than that observed in
S. coelicolor M145 (Supplementary Fig. S5f), the successful operation of EraCBE in
M. echinospora demonstrates that the core EraBE workflow can function in a phylogenetically distinct rare actinomycete. The reduced erasure efficiency observed here may therefore reflect host-dependent differences in the activity of the regulatory elements driving the EraCBE components. Further optimization using host-adapted promoters or other regulatory elements may improve the performance of EraBE in genetically distinct actinomycete hosts. Together, these results extend the experimental validation of EraBE beyond
Streptomyces and support its potential application to a broader range of actinomycetes.
EraABE enables plasmid-free heterologous expression of pqq gene cluster
In addition to supporting plasmid-free genome editing, EraBE was further evaluated for plasmid-free heterologous gene-cluster expression. Given the high editing safety observed for EraABE, this system was selected as the carrier for proof-of-concept validation.
Moreover, because some actinomycete hosts may be insensitive to a given counter-selectable marker
pheSA339G/T278A, thereby reducing the recovery efficiency of mutants that have undergone plasmid-backbone excision, we generated an alternative pEraABE derivative, pEraABE-CodA(sm), by replacing
pheSA339G/T278A with the reported counter-selectable marker
codA39. This modification was intended to improve the versatility and practical applicability of the EraBE system across different host backgrounds.
The previously reported
pqq gene cluster encoding the cofactor pyrroloquinoline quinone, which has been shown to activate or enhance natural-product biosynthesis in
Streptomyces, was cloned into the EraABE plasmid (Fig. 4a)
40. The 3.4 kb expression cassette of
pqq cluster was inserted between the
attP site and the
loxP site in pEraABE-CodA(sm) to ensure that, following chromosomal integration, Cre-mediated recombination would remove only the dispensable plasmid backbone while retaining the heterologous cargo in the genome (Fig. 4b). The resulting recombinant plasmid was successfully integrated into the
S. coelicolor M145 chromosome (Fig. 4b). After inoculation of the corresponding exconjugants to plates containing 2 μg/mL OTC and 10 μg/mL 4-chloro-phenylalanine, PCR analysis confirmed successful plasmid-backbone erasure, with all five randomly selected single colonies displaying the expected plasmid-free genotype (Fig. 4c). Sanger sequencing further verified that the
kasOp*-driven
pqq cluster had been integrated at the intended chromosomal locus and that the predicted plasmid-erasure event had occurred (Fig. 4c).
We used the edited strains after EraABE removal rather than strains retaining the integrated editing cassette for the next step of fermentation evaluation. Prolonged retention of the active base-editing machinery could lead to sustained editor expression during extended cultivation, potentially increasing cumulative off-target editing and introducing additional physiological effects that confound evaluation of the intended mutation. To examine whether removal of the integrated EraABE backbone itself affected strain physiology, we compared the growth profiles of the edited strain retaining pSET152 backbone with those of the corresponding vector-erased strain. The corresponding strains showed comparable growth profiles under the tested conditions (Supplementary Fig. S7), indicating that the vector-erasure process did not cause an evident growth defect.
For fermentation evaluation, although
pqq has previously been associated with enhanced biosynthesis of multiple natural products in
S. coelicolor40, LC-MS analysis further revealed an effect on the production of the antibiotic-pigment compound undecylprodigiosin (Supplementary Table S1; Fig. 4d). Quantitative mass spectrometric analysis showed that the plasmid-free mutant carrying the integrated
pqq cluster produced more than twice as much undecylprodigiosin as the control strain (Fig. 4e, f). Together, these results demonstrate that EraABE can efficiently mediate plasmid-free chromosomal integration of heterologous functional genes and provide a useful additional platform for actinomycete genome engineering.
EraCBE supports gene inactivation and plasmid-backbone erasure in a non-model Streptomyces strain
To further assess the generality of EraBE beyond model hosts, we established an EraCBE-based premature-stop strategy for gene inactivation in strain
Streptomyces sp. 211726, a non-model
Streptomyces isolate from a mangrove environment in Hainan, China, that contains abundant biosynthetic gene clusters but exhibits relatively low conjugation efficiency (Supplementary Table S2)
41-
43. As a target, we selected
wblA211726, a conserved WhiB-like regulatory gene whose homologue in
S. coelicolor has been implicated in morphological differentiation and secondary metabolism (Fig. 5a; Supplementary Table S3)
44,
45.
An sgRNA was designed to convert a tryptophan codon within
wblA into a premature stop codon and cloned into pEraCBE. Following successful chromosomal integration of the recombinant plasmid in strain
Streptomyces sp. 211726, targeted sequencing of exconjugants confirmed the expected cytosine conversion at the intended site (Fig. 5b). After streak purification, we obtained a homogeneous mutant in which the premature stop codon had been stably introduced into
wblA211726 (Fig. 5b). This mutant displayed a striking phenotypic shift relative to the wild-type strain, including a bald morphology, loss of sporulation and a colony color change from white to bright yellow, consistent with disruption of developmental regulation, and altered secondary metabolism (Fig. 5c)
45.
To remove the integrated plasmid backbone, the mutant was subsequently streaked on plates containing 2 μg/mL OTC and 10 μg/mL 4-chloro-phenylalanine to induce plasmid-backbone excision (Fig. 5d). PCR screening identified 12 positive clones among 15 randomly selected single colonies, corresponding to a recovery rate of 80% (Fig. 5e). After an additional round of streak purification followed by sequencing validation, we successfully obtained mutants with the expected genotype and complete plasmid-backbone erasure (Fig. 5f).
Together, these results show that EraCBE can mediate efficient gene inactivation and controlled plasmid erasure in a genetically less tractable, non-model Streptomyces strain, underscoring the portability of the EraBE system across distinct actinomycete hosts.
EraCBE enhances natural products biosynthesis in non-model Streptomyces
After obtaining the plasmid-free
wblA-inactivated mutant Δ
wblA211726, we evaluated its secondary-metabolite production by fermentation followed by HPLC and LC–MS analysis. Relative to the parental strain, Δ
wblA211726 exhibited a broad and pronounced increase in metabolite output. Production of the azalomycin congeners azalomycin F3a, F4a, and F5a, which are marginolactone natural products with reported antimicrobial and antitumor activities, increased by 11.42-, 20.89-, and 18.59-fold, respectively (Fig. 6a–c)
42. Elaiophylin, a bioactive macrodiolide with documented antibacterial and antitumour properties, increased by 1.89-fold (Fig. 6a–c)
46. Nigericin, a polyether ionophore with established antibacterial, antifungal, antimalarial, and anticancer activities, showed a 3.99-fold increase (Fig. 6d–f)
47.
To assess whether EraCBE induced off-target mutagenesis during genome editing in Streptomyces sp. 211726, genomic DNA from the mutant strain ΔwblA211726 was subjected to whole-genome resequencing. Comparative analysis revealed only 23 single-nucleotide variants (SNVs) and 49 indels in ΔwblA211726, whereas the reference wild-type genome contained 15 SNVs and 48 indels (Supplementary Data 2; Supplementary Fig. S8). The minimal difference in variant counts between the mutant and wild-type strains indicates that EraCBE did not introduce obvious off-target effects at an appreciable level in this host, supporting an acceptable safety profile for genome editing in Streptomyces sp. 211726.
These results demonstrate, at a technical level, that EraCBE can be used to generate plasmid-free regulatory mutants in a genetically less tractable host and thereby efficiently reprogram endogenous secondary metabolism. At the application level, the simultaneous enhancement of multiple bioactive metabolites highlights the value of this strategy for improving production titers of pharmaceutically relevant natural products and for accelerating strain development in non-model actinomycetes.
DISCUSSION
This study establishes EraBE as an erasable genome engineering system for actinomycetes. The main conceptual advance lies in linking site-specific integration, base editing, and inducible plasmid-backbone erasure within a single vector architecture. This design addresses a practical limitation of current actinomycete engineering workflows: integrative plasmids are often the most reliable route for DNA delivery, especially in genetically recalcitrant hosts, yet they leave permanent non-essential sequences in the chromosome after editing or cargo introduction. By coupling SSR-mediated integration with base-editing-dependent rescue of an inactive Cre variant, EraBE converts a normally irreversible delivery event into a controllable and largely self-contained engineering cycle. In this respect, EraBE is important not only because it functions, but also because it provides a general solution to the long-standing conflict between delivery efficiency and genomic cleanliness in actinomycetes.
The feasibility of this integrated strategy is supported by several aspects of the present study. First, the broad genomic distribution of
attB sites across actinomycetes indicates that SSR-based delivery should, in principle, be applicable to a wide range of hosts
7-
9. Second, the dCre design provides a mechanistically meaningful solution to early recombination, a problem that is difficult to avoid when wild-type Cre is placed directly under inducible transcriptional control. Recent studies have addressed Cre leakage by splitting the recombinase into two inactive fragments and restoring its activity through chemical or light induction, thereby improving controllability
48,
49. However, split Cre fragments can still undergo spontaneous association in the absence of induction, creating a residual risk of unintended recombinase activity. By contrast, the dCre design used in EraBE remains functionally inactive unless the mutated codon is precisely corrected by base editing. Cre activity is therefore restored only after the intended editing event has occurred, providing a more stringent and genetically defined layer of control over recombinase activation. Third, EraBE supported both plasmid-free base editing and plasmid-free genetic retention of a heterologous functional cargo, demonstrating that the same framework can serve two common engineering objectives: construction of clean edited mutants and stable introduction of useful exogenous genes or gene clusters.
Compared with existing actinomycete genome-editing systems that rely on autonomously replicating plasmids
20-
22, EraBE uses site-specific integrative vectors and is therefore less constrained by host-specific replicon compatibility. Consistent with this design, our
attB-find analysis showed that 98.24% of the analyzed actinomycete genomes contain at least one potentially applicable site-specific recombination system, supporting the potential portability of EraBE across diverse hosts (Fig. 1a). In addition, EraBE couples vector erasure directly to the genome-editing event: Cre remains catalytically inactive before editing and is restored only by the programmed base-editing event, thereby minimizing leakage-associated premature recombination and triggering backbone excision only after editing. Importantly, EraBE can also retain a heterologous cargo in the chromosome while removing the functional plasmid backbone, enabling backbone-free heterologous gene expression. Together, these features integrate broad integrative delivery, editing-responsive vector clean-up, and stable cargo expression within a single framework.
Despite these strengths, the current study also exposes several limitations that define the next stage of development. First, EraBE is not fully scarless: following excision, a
loxP site and a residual
attB-
attL-derived recombination sequence remained at the integration locus (Fig. 3e). This residual sequence burden is substantially smaller than that imposed by the full plasmid backbone, but it may still be undesirable in applications requiring repeated rounds of integration at the same site or ultra-clean industrial genotypes. Future versions could alleviate these limitations by employing heterospecific lox sites or orthogonal recombinase–target pairs, which can enable independent site-specific recombination events during successive engineering rounds
50. More prospectively, RNA-guided bridge recombinases derived from IS110 elements allow programmable recognition of both target and donor DNA and support sequence-specific DNA insertion and excision
51. Adaptation of such programmable recombination systems to
Streptomyces may ultimately reduce reliance on fixed
loxP/
attB recognition sites and facilitate cleaner iterative genome engineering.
Second, the appearance of the 505-bp PCR product in
S. coelicolor indicates that continued integrase activity after integration can generate unintended restoration of the wild-type chromosomal configuration, bypassing the intended erasure route
37. This observation is mechanistically informative because it identifies integrase persistence, rather than Cre leakage, as a remaining source of genotype heterogeneity after integration. To improve the overall dynamic range and erasure efficiency of the system, the dCre activation site and base editor could be co-optimized according to editing window, target sequence context, and the efficiency of the desired amino-acid restoration.
Third, although the genome survey indicates broad theoretical host coverage, the experimental platform was validated here primarily with a φC31-based backbone, one heterologous cargo example and two actinomycete hosts. Additional validation with other SSR systems, larger BGCs and more phylogenetically diverse strains will be needed to determine how broadly the current design principles transfer in practice. These considerations suggest several rational improvements, including deployment of alternative integrases selected through the
attB-prediction tool, minimization or self-limitation of post-integration integrase expression, reduction of recombination scars
52.
In future implementations, EraBE-like architectures could be adapted for multiplex regulatory rewiring, iterative pathway optimization, backbone-free installation of large biosynthetic cassettes, or integration of accessory functions that activate silent BGCs and are then selectively removed once host-state remodeling has been achieved
53. The
attB-find web tool reported here should further facilitate this expansion by allowing rational matching of hosts and SSR systems before experimental design.
The present work shows that EraBE already functions as a portable and operationally streamlined platform for “hit-and-run” genome engineering in actinomycetes. Its significance therefore lies less in any single editing event than in providing a versatile systems-level framework for future strain development, pathway engineering and natural-product discovery.
MATERIAL AND METHODS
Bacterial strains, culture conditions, and reagents
Information on bacterial strains used in this study is presented in Supplementary Table S4
. E. coli DH10B was used for plasmid construction and routine subcloning. For intergeneric conjugation,
E. coli ET12567 harboring pUZ8002 served as the donor strain for DNA transfer to
Streptomyces.
E. coli strains were cultivated in 2×TY medium (1% yeast extract, 1.6% tryptone, and 0.5% NaCl) at 37 °C with appropriate antibiotic selection, whereas
S. coelicolor and
Streptomyces sp. 211726 were cultured at 28 °C on mannitol SFM agar medium (2% mannitol, 3% soybean powder, and 2% agar) or in TSB liquid medium (2% tryptone soya broth) unless otherwise specified
54.
M. echinospora was cultured at 28 °C on ABB13 agar medium (0.5% soluble starch, 0.5% tryptone soya broth, 0.21% MOPS, 0.0012% FeSO
4·7H
2O, 0.001% thiamine hydrochloride, 0.3% CaCO
3, and 2% agar). For sporulation,
S. coelicolor M145 and
Streptomyces sp. 211726 were grown on SFM medium, while
M. echinospora was grown on ABB13 agar medium. Antibiotics were supplemented as required at the following concentrations: ampicillin (100 μg/mL), apramycin (50 μg/mL), kanamycin (50 μg/mL), chloramphenicol (25 μg/mL), and nalidixic acid (25 μg/mL) for
E. coli strains. Apramycin (25 μg/mL) was added to SFM medium for
S. coelicolor M145 and
Streptomyces sp. 211726. Thiostrepton (8 μg/mL) in ABB13 medium for
M. echinospora. All chemical reagents and antibiotics were obtained from commercial suppliers, including Sigma-Aldrich and Sangon Biotech, while molecular cloning reagents were purchased from New England BioLabs and QIAGEN. Oligonucleotide synthesis and Sanger sequencing were performed by Wuhan Tsingke.
Plasmid construction
Information on plasmids and primers used in this study is listed in Supplementary Tables S5, S6. pEraCBE and pEraABE were constructed using pSET152, an integrative
E. coli–
Streptomyces shuttle vector, as the parent plasmid
55. For pEraCBE construction, five initial fragments were prepared. Fragment 1, containing the eSCBE3-NG-Hypa operon, was amplified from peSCBE3-NG-Hypa using the primer pair Fwd-BE-operon-152 and Rev-BE-operon-152
22. Fragment 2, containing the target sgRNA cloning cassette, was amplified from peSCBE3-NG-Hypa using Fwd-target sgcc and Rev-target sgcc
22. Fragment 3, comprising
pheSA339G/T278A operon
36, the orthogonal erasing sgRNA cassettes driven by the inducible promoter P
otr*, and
otrR driven by the constitutive promoter
hrdBp
33, was synthesized by GenScript and amplified using Fwd-
pheS-erasing sgRNA and Rev-
pheS-erasing sgRNA. Fragment 4, containing dCre driven by
hrdBp, was synthesized by GenScript and amplified using primer pair Fwd-dCre and Rev-dCre. Fragment 5, corresponding to the pSET152 backbone, was amplified from pSET152 using primer pair Fwd-vec-152 and Rev-vec-152. These five fragments were assembled by Gibson assembly to generate an intermediate plasmid. Fragment 6, containing paired
loxP sites, the φC31
attP site and the
ermEp* promoter, was synthesized by GenScript and amplified using primer pair Fwd-
loxP-
attP and Rev-
loxP-
attP, and was subsequently inserted into the intermediate plasmid by Gibson assembly to replace the original
attP site and the original promoter driven φC31 integrase (the recipient backbone was amplified using Fwd-vec-
loxP and Rev-vec-
loxP), yielding pEraCBE. pEraABE was generated from pEraCBE by replacing the APOBEC3A(Y130F) deaminase with TadA8e
32. To construct pEraABE-CodA(sm), a synthesized fragment containing the
codA(sm) operon was introduced in place of
pheSA339G/T278A in pEraABE by Gibson assembly
39. For pEraABE-CodA(sm)-
pqq construction, the
pqq cluster driven by
kasOp* was amplified from pLQ646 and inserted into pEraABE-CodA(sm) by Gibson assembly
40. The detailed DNA sequence information of the genetic modules used in this study is provided in Supplementary Table S7.
sgRNA design and cloning
Information on the sgRNAs used in this study is listed in Supplementary Table S8. The target sgRNAs for EraCBE and EraABE were designed using the Benchling web tool (benchling.com/) and cloned by Gibson assembly after plasmid linearization with
Xba I and
Age I (Supplementary Fig. S4)
22.
Interspecies conjugation
Conjugation experiments were performed according to the standard protocol described previously
54. Briefly,
E. coli ET12567/pUZ8002 harboring pEraBE were grown to an OD
600 nm of 0.4–0.6, collected by centrifugation at 2,400×
g for 5 min, washed twice with 2×TY broth and resuspended in 100 μL 2×TY.
Streptomyces spores were washed twice with TES buffer (0.05 mol/L, pH 8.0), resuspended in 5 mL TES buffer, and incubated at 50 °C for 10 min to induce germina-tion. An equal volume of 2×TY broth and 10 μL 5 mol/L CaCl
2 were then added, and the mixture was incubated at 37 °C for 2–3 h with shaking at 220 rpm. Germinated spores were collected by centrifugation under the same conditions and resuspended in 100 μL 2×TY. Donor
E. coli cells and prepared spores were mixed and plated on SFM agar supplemented with 10 mmol/L MgCl
2 for
S. coelicolor M145 or 10 mmol/L CaCl
2 for
Streptomyces sp. 211726. After incubation for 12 h at 28 °C, plates were overlaid with 1 mL sterile water containing nalidixic acid and apramycin to final concentrations of 25 μg/mL each for both
S. coelicolor M145 and
Streptomyces sp. 211726. For conjugation delivery of plasmids into
M. echinospora, the conjugation plates were incubated for 16 h at 28 °C, then the plates were overlaid with 1 mL of sterile water containing nalidixic acid to final concentration of 25 μg/mL and the thiostrepton to a final concentration of 8 μg/mL.
Base editing efficiency evaluation
To assess base editing events in individual exconjugants, randomly selected colonies were cultured on SFM agar supplemented with apramycin for an additional 3 days to obtain sufficient genomic DNA and confirm plasmid integration. Target loci were amplified from genomic DNA using Phanta Max Super-Fidelity DNA Polymerase (Vazyme) with primers listed in Supplementary Table S6. Amplicons were verified by agarose gel electrophoresis, purified using a Universal DNA Purification Kit (Tiangen Biotech) and subjected to Sanger sequencing (QIAGEN). Base conversion efficiencies were quantified using EditR, which analyzes editing frequencies from Sanger sequencing chromatograms
56.
Off-target effects evaluation through whole genome resequencing
The corresponding S. coelicolor M145 and Streptomyces sp. 211726 mutants were inoculated into 30 mL fresh TSB broth and cultivated at 28 °C and 220 rpm for 2–3 days. Mycelial fragments were harvested by centrifugation at 6,200× g for 10 min, washed twice with sterile water and collected for genomic DNA extraction. Sequencing libraries were prepared according to the manufacturer’s instructions (GENEWIZ), and qualified libraries were subjected to paired-end 150-bp sequencing on the Illumina HiSeq X Ten, NovaSeq or MGI2000 platform. Clean reads were mapped to the S. coelicolor A3(2) genome (NC_003888.3) or the genome of Streptomyces sp. 211726, and SNVs were called. Raw sequencing reads were processed using fastp (v0.23.0) to remove adapters, PCR primers, reads containing more than 14 ambiguous bases, and reads with Q20 lower than 40%. Clean reads were mapped to the reference genome using the Sentieon pipeline (v202112.02), which was also used for duplicate removal and SNV/InDel calling. SNVs and InDels were annotated using ANNOVAR (v21 Apr 2018), while BreakDancer and CNVnator were used for structural variation analysis.
RT-PCR analysis of integrase transcription
S. coelicolor M145 strains carrying chromosomally integrated pSET152, EraCBE, or EraABE were first activated on agar plates. An approximately 1-cm2 agar plug containing mycelia from each strain was inoculated into TSB medium and cultured for 24 h. The mycelia were then harvested for total RNA extraction using the RNAprep Pure Cell/Bacteria Kit (TIANGEN, China) according to the manufacturer’s instructions. The resulting RNA samples were further purified using the RNAclean Kit (TIANGEN, China) to minimize residual genomic DNA contamination. RT-PCR was performed using the purified RNA samples with the integrase-specific primers Seq-fwd-RT-int and Seq-rev-RT-int (Supplementary Table S6). The resulting products were analyzed by agarose gel electrophoresis.
Fermentation conditions
SFM agar medium was used for strain rejuvenation. Mycelia from SFM agar plates were inoculated into 250 mL flasks containing 30 mL TSB seed medium and cultured at 28 °C and 220 rpm for 2 days. For fermentation, S. coelicolor was grown in SFM liquid medium, whereas Streptomyces sp. 211726 was cultivated in YMG liquid medium containing 4 g/L yeast extract, 10 g/L malt extract, and 10 g/L glucose. 50 mL of fermentation medium was inoculated with 5% seed culture and incubated at 28 °C and 220 rpm for 5–7 days.
Isolation and detection of secondary metabolites of fermentation broth
For extraction of fermentation products from S. coelicolor and Streptomyces sp. 211726, 20 mL broth from three biological replicates was mixed with 30 mL methanol and sonicated for 30 min. After centrifugation at 13,800× g for 10 min, the supernatants were concentrated under reduced pressure. The resulting residues were dissolved in 1 mL methanol, filtered through a 0.22 μm Nylon66 membrane and analyzed by HPLC (SHIMADZU) and LC-ESI-HRMS (Thermo). HPLC analysis was performed on a Phenomenex Luna C18 column (5 μm, 250 mm × 4.6 mm) at a flow rate of 1.0 mL/min using solvent A (0.1% formic acid in water) and solvent B (0.1% formic acid in acetonitrile). For compounds from S. coelicolor, the gradient was 0–2 min, 65% B; 2–17 min, 5%–100% B; 17–22 min, 100% B; 22–23 min, 100%–5% B; and 23–27 min, 5% B. For compounds from Streptomyces sp. 211726, the gradient was 0–20 min, 35%–65% B; 20–28 min, 65%–95% B; 28–30 min, 95% B; 30–33 min, 95%–35% B; and 33–38 min, 35% B. LC-ESI-HRMS analysis was conducted in positive-ion mode on a Thermo Electron system at a flow rate of 0.4 mL/min using the same mobile phases and gradients as in the HPLC analysis. Data were collected and processed using Thermo Xcalibur software v3.0.63 for peak alignment and formula assignment.
Prediction of attB sites in actinomycetes
To construct the web for prediction of attB sites in actinomycetes, all genome assemblies assigned to the phylum Actinomycetota and available in the NCBI RefSeq database as of March 14, 2025 were collected, comprising 42,942 assemblies in total. Collecting 7,735 genomes by filtering assemblies containing fewer than five contigs (Supplementary Data 1). Candidate attB sites were identified using blastn-short. Searches were performed with an E-value threshold of ≤ 0.01 and a word size of 7, and hits with ≥ 75% alignment coverage and a bit score ≥ 30 were retained as putative attB sites. These criteria were used to survey attB sites corresponding to the representative site-specific recombination systems included in attB-find (attb-find.com/). In the web application, users can submit target genome sequences for attB-site identification, and the E-value and bit-score thresholds can be adjusted to modify the search stringency according to specific requirements.
Statistics analysis
Statistical analyses were performed using GraphPad Prism 10.1.2 with two-tailed t-test and analysis of variance hypothesis. Significance was denoted as *P < 0.05, **P < 0.01, ***P < 0.001, ****P < 0.0001. All data are presented as mean ± SD unless otherwise indicated, and the number of biologically independent samples was three for each panel unless stated otherwise in the figure legends.
The Author(s) 2026. Published by Higher Education Press.
