The strains and plasmids used in this study are listed in Supplementary Table S1. The cells of E. coli DH5α and E. coli JM110 were grown in Luria−Bertani medium at 37°C for 12–16 h. All A. keratiniphila strains were grown at 28°C. For sporulation, the cells were grown on Gauze’s synthetic agar medium (GM). For preparing the competent cells, A. keratiniphila HCCB10007 was cultured in complete pre-cultivation medium (CRM) for 48 h. For the selection of transformants, the cells of A. keratiniphila were grown on Bennet’s medium for 4 days, and then were cultivated in tryptic soy broth (TSB) liquid medium for 2 days. All of the above strains were cultured with agitation at 220 rpm in liquid medium. When necessary, 50 μg/mL or 100 μg/mL of apramycin (Apr) was added in the liquid medium or solid medium. For vancomycin and ECO-0501 production, A. keratiniphila HCCB10007 and the mutants were cultivated in seed medium for about 60 h with shaking at 250 rpm, and were then incubated at 200 rpm for 4 days (Shen et al., 2014; Xu et al., 2014).

The primers used in this study are listed in Supplementary Table S2. Restriction enzymes, Taq enzymes and ligases, and other common molecular biology reagents were purchased from TaKaRa. The ClonExpress MultiS One Step Cloning Kit purchased from Vazyme Biotech Co., Ltd. was used for ligation of fragments and vectors by homologous recombination. High fidelity polymerase KOD FX and PrimeSTAR (Toyobo) were used to amplify target gene for cloning purposes and to perform PCR screening of mutant strains according to the manufacturer’s protocol. PCR reactions were carried out in a PCR instrument (Eppendorf). The sequencing of DNA and the synthesis of all primers were conducted by GENEWIZ. DNA recovery kits and plasmid extraction kits were purchased from Toyobo, and a DNA Marker (GenerulerTM 1 kb DNA ladder) was purchased from Fermentas. All chemicals used were analytical grade and commercially available.

Isolation of genomic DNA from Amycolatopsis strains and plasmid DNA from E. coli were carried out using standard protocols (Kieser et al., 2000). Restriction enzymes and molecular biology reagents were used according to recommendation of suppliers (Takara, Vazyme, Toyobo, Eppendorf).

The CRISPR/Cas9 target online predictor CCTop (http://crispr.cos.uni-heidelberg.de), which present the rapid selection of high quality target sites for NHEJ as well as HDR, was selected for the guide sequences and protospacer adjacent motif (PAM) of sgRNA design (Stemmer et al., 2015; Labuhn et al., 2018).

On the basis of pKCcas9dO (Huang et al., 2015), the plasmids containing different promoters to drive the expression of cas9 and sgRNA were constructed as follows. 1) Using the plasmid pKCcas9dO as the template, the sgRNA fragment containing the promoter J23119, the crRNA scaffold and 19-nt direct repeat was amplified with the primers gRNADNrecom/gtfDgRNAspc2. The resulting sgRNA fragment was cloned into the SpeI/HindIII-digested pKCcas9dO by the Solution-I ligation and the plasmid pKCcas9dgtfD-NA was generated. 2) The sgRNA fragment containing the crRNA scaffold and 19-nt direct repeat was amplified from pKCcas9dO with the primers gRNADNrecom/gtfDgRNArecom. The ermE* promoter was amplified from pLYZWG using the primers ermE-F/ermE-R (Xu et al., 2015). With the Gibson method, the resulting sgRNA fragment and the ermE* promoter was recombined into the plasmid pKCcas9dO which was doubly digested by XbaI/HindIII, and the plasmid pKCcas9EgdgtfD-NA was generated. 3) The endogenous gapdh promoter was amplified from the genome of A. keratiniphila HCCB10007 using the primers gapdh-F/gapdh-R, and it was cloned into the XbaI/NdeI-digested pKCcas9EgdgtfD-NA by Gibson ligation, thus generating the plasmid pKCpGcas9EgdgtfD-NA.

Using the genomic DNA of A. keratiniphila HCCB10007 as the template, the upstream and downstream homologous arm fragments of the gene gtfD were obtained by PCR amplification using the primers Vcm-8F/Vcm-8R and Vcm-10F/Vcm-10R. The two fragments were recovered with a DNA recovery kit and then fused into the pMD19-Tsimple vector by overlapping extension PCR. The resulting plasmid was doubly digested by KpnI/PstI and ligated with eGFP fragment, which was originated from pLYZWG by double digestion with KpnI/PstI (Xu et al., 2015). The fragment of eGFP was inserted between the upstream and downstream homologous arms of gtfD by the Solution-I ligation. The homologous arms inserted by eGFP were integrated into pKCcas9dgtfD-NA, pKCcas9EgdgtfD-NA, and pKCpGCas9EgdgtfD-NA by using a homologous recombination kit to generate the CRISPR/Cas9 editing plasmids pLYNY02, pLYNY03, and pLYNY04. The cas9 was driven by tipA or gadph promoter and the target-specific sgRNA was driven by J23119 or ermE* promoter, respectively. Figure 1A showed the detailed CRISPR/Cas9 editing plasmids pLYNY02, pLYNY03, and pLYNY04.

Based on pLYNY04, a series of CRISPR/Cas9-mediated editing plasmids containing different target-specific sgRNAs and homologous DNA arms for deleting the fragments of cds13-17 (7.7 kb), cds23 (12.7 kb), cds22-23 (21.2 kb), and cds4-27 (87.5 kb) in ECO-0501 BGC were constructed, respectively. Taking the editing plasmid for knocking out cds13-17 as an example, the upstream and downstream homologous regions of cds13-17 were amplified from the genome of A. keratiniphila HCCB10007 using the primers 13-17-arm-AF/13-17-arm-AR and 13-17-arm-ZF/13-17-arm-ZR, respectively. These two fragments were recombined with the linear pLYNY04 vector generated by HindIII digestion. This resulting plasmid was linearized by SpeI digestion and ligated with annealed sgRNA oligonucleotide by overlapping recombination to obtain the final cds13-17-specific editing plasmid.

The dual sgRNA-guided plasmid pLYHMY87-5-I was constructed as follows. The plasmid pLYHMY7-5 was amplified with the primers of sgRNA5-F/sgRNA5-R, and the fragment containing sgRNA 5 and ermE ^* promoter was purled and ligated into pLYHMY21-I digested by XbaI.

The plasmids were transferred into E. coli DH5α for cloning and E. coli JM110 for demethylation by heat shock following the manufacturer’s suggested protocol. The competent cells of A. keratiniphila HCCB10007 for electroporation were prepared as previously described (Xu et al., 2015). The CRISPR/Cas9 editing plasmids were transformed into the competent cells of A. keratiniphila HCCB10007 by a Gene Pulser Xcell (Bio-Rad, Inc.) and electroporation was performed at a field strength of 7.5 kV/cm (25 μF, 600 Ω) with a pulse of about 13 ms. After electroporation, 1 mL of liquid TSB was added to the cell suspension, followed by 7 h incubation. The transformants were grown on Bennet’s medium containing Apr for 5 days, and the incubation was continued in 3 mL of liquid TSB medium containing Apr at 200 rpm for 2 days. The resulting colonies were subsequently checked by PCR with a set of primers outside or inside the region of recombination and subsequent Sanger sequencing. These verified colonies were cultivated on Bennet’s medium at 37°C overnight for one or two rounds to eliminate the plasmid, and subsequently incubated on Bennet’s medium with or without Apr for 3 days at 28°C to select the correct mutants.

The analyses of vancomycin and ECO-0501 produced by the strains of A. keratiniphila HCCB10007 and the mutants were conducted as previously described (Shen et al., 2014; Xu et al., 2014).

The strains of A. keratiniphila HCCB10007 and A. keratiniphila HCCB10007 eGFPΔgtfD were cultured in TSB medium for 5 days. The mycelia were observed under a Leica TCS-SP8 confocal laser-scanning microscope with a ×40 objective lens using excitation wavelength (485 nm) and emitting wavelength (535 nm) (Santos-Beneit and Errington, 2017).

The CRISPR/Cas9-mediated editing plasmid in this study was derived from the high-efficiency plasmid pKCcas9dO, which originated from pKC1139 and successfully applied in Streptomyces (Huang et al., 2015). However, the transformation efficiency was much lower than that of the control plasmid pKC1139 in A. keratiniphila HCCB10007 (Figure 1B). To drive the expression of the cas9 gene and the target-specific sgRNA in A. keratiniphila, the three editing plasmids, which cas9 and sgRNA were expressed under control of different promoters, were transformed into A. keratiniphila HCCB10007 by electroporation. Compared with the empty vector pKC1139, the introduction of pKCcas9dO and pLYNY02 caused about 88% decrease in the transformation efficiency, and no transformants were observed in three independent experiments with pLYNY03. It was indicated that Cas9 toxicity was apparent for A. keratiniphila. When cas9 was expressed under control of endogenous gapdh promoter from A. keratiniphila HCCB10007, the transformation efficiency was improved. About 46.5% decrease of transformation efficiency was caused when pLYNY04 was introduced into A. keratiniphila HCCB10007. The number of transformants was acceptable for the genetic manipulation. The deletion of glycosyltransferase gene gtfD and insertion of eGFP with the editing plasmid pLYNY04 was illustrated in Figure 2A.

After eliminating the plasmid of pLYNY04 by high-temperature culturing, the colonies were confirmed by PCR with three pairs of primers (Figure 2B). Because of the replacement of gtfD by eGFP, PCR amplification from the genomic DNA of the original strain produced no specific products, and PCR of the mutants had the amplicons of 1795, 3340, and 3357 bp, respectively. It was confirmed that the introduction of pLYNY04 into A. keratiniphila HCCB10007 caused a 100% editing efficiency of gtfD gene deletion and eGFP gene insertion from three independent replicates. The mutant strain was named A. keratiniphila HCCB1007 eGFP ΔgtfD.

As expected, the HPLC assay for the fermentation products showed that unlike the A. keratiniphila HCCB10007, no vancomycin peak was observed in the mutant of A. keratiniphila HCCB1007 eGFP ΔgtfD (Figure 2C). The absence of vancomycin in the metabolites of the mutant strain proved that the pLYNY04 plasmid successfully deleted the gtfD and further led to the block of vancomycin biosynthesis. The green fluorescence was detected clearly in the mycelia of A. keratiniphila HCCB1007 eGFP ΔgtfD by confocal laser microscopy, which further confirmed the completion of expected insertion of eGFP and the successful expression of green fluorescent protein (Figure 2D). It could be therefore concluded that the CRISPR/Cas9-mediated editing plasmid pLYNY04, harboring cas9 driven by gadph promoter and the target-specific sgRNA driven by ermE* promoter, was capable of HDR-dependent gene editing in A. keratiniphila HCCB1007 and the system could achieve high efficiency for the deletion and insertion of single gene.

The editing ability of CRISPR/Cas9 is mainly determined by the capacity of sgRNA to recognize and cleave at specific sites, and a great quantity of online sgRNA design websites are available to predict and improve the efficiency of gene editing (Cui et al., 2018; Liu et al., 2020). The sgRNA score of CCTop is based on the off-target quality and the distribution of mismatches, and guides the user towards selecting the optimal target site (Stemmer et al., 2015). To achieve the deletion of large chromosomal fragments in A. keratiniphila HCCB1007, cds13-17 (7.7 kb) and cds22-23 (21.2 kb) in the ECO-0501 BGC were selected as targeting clusters, and nine sgRNAs targeting specific sites of cds13-17 and cds22-23, which scores were over 0.75 and there were no predicted off-target effects were designed. The detailed information of location, sequence composition, PAM sequence, scores in CCTop and GC contents were shown in Figure 3A. On the basis of pLYNY04, the plasmids harboring the corresponding sgRNAs and the homologous arms of cds13-17 or cds22-23 were constructed, respectively. No successful deletion of cds13-17 was observed using the plasmids of pLYHMY7-2 and pLYHMY7-4. With the plasmids of pLYHMY7-1, pLYHMY7-3, pLYHMY21-II, and pLYHMY21-III, the editing efficiencies of 5% to 29% were relatively low. The plasmids of pLYHMY7-5, pLYHMY7-6, and pLYHMY21-I exhibited the more efficient editing ability, and the editing efficiencies of 87% ± 12%, 74% ± 20%, and 97% ± 3% were accomplished, respectively for the deletion of cds13-17 and cds22-23 clusters (Figure 3A). The correct deletions of the clusters of cds13-17 or cds22-23 were verified by PCR with two pairs of primers located inside and outside of the targeting deletion clusters (Figure 3B). PCR of the cds13-17 deleted colonies could not produce a 1491bp DNA fragment with the primers of Verify-7in-F/Verify-7in-R, while amplified a 6092bp DNA fragment with the primers of Verify-7out-F/Verify-7out-R. Similarly, the results showed that the cds22-23 deleted colonies gave a specific 6602bp amplicon with the primers Verify-21out-F/Verify-21out-R, and did not have 1403bp amplicon with the primers Verify-21in-F/Verify-21in-R. The sequencing results of the DNA fragments deleted colonies also confirmed that the deletion of the DNA fragments of cds13-17 or cds22-23 and homologous recombination repair were achieved (Figure 3C). The HPLC analyses of the metabolites fermented by the two clusters deleted mutants A. keratiniphila HCCB10007 Δeco-cds13-17 and A. keratiniphila HCCB10007 Δeco-cds22-23 demonstrated that the mutant strains were unable to biosynthesize ECO-0501 and the genomic targets had been deleted correctly (Figure 3D).

To detect the editing efficiency of the same sgRNA expression cassette with the corresponding homologous arms, the plasmids of pLYHMY12-I targeting to cds23 (12.7 kb) and pLYHMY87-I targeting to cds4-27 (87.5 kb) were transformed into A. keratiniphila HCCB10007. The deletion efficiency of cds23 was 81% ± 12%, which was comparable to the editing efficiency of 97% ± 3% which achieved by the plasmid of pLYHMY21-I for deleting cds22-23 (21.2 kb) (Figure 4A). The correct deletion of cds23 was validated by PCR with the corresponding verification primers. The cds23 deleted colonies could amplify a 5169 bp fragment with the primer of Verify-12out-F/Verify-12out-R and did not obtain the amplicon of 1481 bp with primer Verify-12in-F/Verify-12in-R (Figure 4B). No mutant with deletion of cds4-27 was obtained (Figure 4A). The plasmid of pLYHMY87-I failed to delete the fragment of 87.5 kb, which was likely due to the inefficient of single sgRNA targeting to much larger gene cluster. So, the plasmid pLYNY04 derivatives with single sgRNA and two homologous arms flanking the targeted clusters could delete large DNA fragment, which covers a size of 21.2 kb region in ECO-0501 BGC with a higher efficiency.

The dual-sgRNA strategy which introduced DSBs at both ends and bridged the gap with homologous arms clearly demonstrated that it could significantly improve the deletion efficiency of CRISPR/Cas9 system in editing larger fragments (Cobb et al., 2015; Huang et al., 2015). ECO-0501 BGC spans approximately 100 kb of DNA in A. keratiniphila HCCB10007 (Xu et al., 2014), The deletion of larger DNA fragments of ECO-0501 BGC is particularly valuable for increasing the availability of the precursors beneficial for vancomycin biosynthesis. To further delete the cluster of cds4-27, which covers 87.5 kb of DNA fragments in ECO-0501 BGC, the tandem cds4-27-specific sgRNA expression cassettes driven by two copies of the promoters ermE* to increase the cleavage was considered. The plasmid pLYHMY87-5-I, containing the efficient sgRNAs of both sgRNA 5 and sgRNA I with corresponding homologous arms for cds4-27 (Figure 5A), was constructed and transformed into the strain of A. keratiniphila HCCB10007. The average deletion efficiency of cds4-27 cluster was 13% ± 11%, and the cluster deleted colonies were validated by PCR with the corresponding verification primers. There was no amplicons of 1403 bp with the primer of Verify-21in-F/Verify-21in-R, and the amplicon of 6777 bp was amplified with the primer of Verify-87out-F/Verify-87out-R (Figure 5B). The DNA sequencing result of the strain A. keratiniphila HCCB10007 Δeco-cds4-27 demonstrated that the mutant deleted the cluster of cds4-27 completely (Figure 5C), and could not biosynthesize the metabolite of ECO-0501 (Figure 5D). So, the pLYHMY87-5-I with dual sgRNAs and two homologous arms flanking the targeted cluster could improve the deletion efficiency in deleting larger fragment, which covers a size of 87.5 kb region in ECO-0501 BGC.

A. keratiniphila HCCB10007 is well known for producing vancomycin and ECO-0501, and the both biosynthesis pathways share some common precursor (Shen et al., 2014; Xu et al., 2014). Because of the deletions of gene clusters cds13-17, cds22-23 and cds4-27 in the BGC of ECO-0501, the ability of vancomycin biosynthesis in the three mutants were all significantly improved, and the production of vancomycin increased by 30.54% ± 6.2%, 33.99% ± 8.6%, and 40.58% ± 7.5%, respectively (Figure 6).