CASCADE-Cas3 enables highly efficient genome engineering in Streptomyces species

Abstract

Type I clustered regularly interspaced short palindromic repeat (CRISPR) systems are widespread in bacteria and archaea. Compared to more widely applied type II systems, type I systems differ in the multi-effector CRISPR-associated complex for antiviral defense needed for crRNA processing and target recognition, as well as the processive nature of the hallmark nuclease Cas3. Given the widespread nature of type I systems, the processive nature of Cas3 and the recombinogenic overhangs created by Cas3, we hypothesized that CASCADE-Cas3 would be uniquely positioned to enable efficient genome engineering in streptomycetes. Here, we report a new type I based CRISPR genome engineering tool for streptomycetes. The plasmid system, called pCRISPR-Cas3, utilizes a compact type I-C CRISPR system and enables highly efficient genome engineering. pCRISPR-Cas3 outperforms pCRISPR-Cas9 and facilitates targeted and random sized deletions. Furthermore, we demonstrate its ability to effectively perform substitutions of large genomic regions such as biosynthetic gene clusters. Without additional modifications, pCRISPR-Cas3 enabled genome engineering in several Streptomyces species at high efficiencies.

Graphical Abstract

Figure 1.

pCRISPR-Cas3 enables streamlined genome engineering of streptomycetes. (A) Distribution of type I and type II CRISPR systems in streptomycetes and the number of PAM sites identified in selected streptomycetes for both systems. Spycas9 and cas3 from S. albidoflavus were used as references. The BLAST search was run on all high quality publicly available Streptomyces genomes (n = 2401). Type I CRISPR systems appear to be much wider distributed than type II CRISPR systems in streptomycetes. However, only 108 hits with a sequence similarity above 50% were identified. An order of magnitude fewer PAM sites were identified in three selected streptomycetes for the CASCADE-Cas3 PAM compared to the NGG Cas9 PAM, highlighting the much lower number of potential off-target sites. (B) Plasmid map of pCRISPR-Cas3. The plasmid is based on the pSG5 replicon and carries the codon optimized type I-C minimal CASCADE under control of the inducible tipA promoter. All elements of the backbone are highlighted in gray. The crRNA is cloned between to repeats in the chromosomal RNA (cRNA) cassette, which is controlled by the constitutive ermE* promoter. Repair templates are cloned into the multiple cloning site (MCS) on the backbone of the plasmid. (C) The second repeat has a modified sequence to prevent recombination between the two repeats. The CASCADE complex comprised of Cas5, Cas8, and Cas7 units binds the target sequence and recruits Cas3. Cas3 has a 3′–5′ helicase nuclease activity, resulting in directionally biased deletions. (D) cRNAs are cloned between two repeats, posing some challenges due to sequence homologies. Since type IIS restriction enzymes cannot be used in high GC Streptomyces plasmids, a PCR and Gibson Assembly based cloning approach was established, allowing cloning of cRNAs with high efficiencies. (E) pCRISPR-Cas3 can be used for targeted deletions of large genomic regions, or for substitutions of such with a specified cargo. It can also be used for random sized deletion experiments.

Figure 2.

pCRISPR-Cas3 introduces genomic deletions with higher efficiencies than pCRISPR-Cas9. (A) Plate pictures of S. coelicolor mutants harboring pCRISPR-Cas9 or pCRISPR-Cas3 with spacers targeting the actinorhodin BGC in three different locations, and with or without repair templates. pCRISPR-Cas3 displayed higher toxicity without repair templates but resulted in more exconjugants overall and more with the desired red phenotype once repair templates were provided. (B) Representation of sequencing results of selected colonies, both for pCRISPR-Cas3 and pCRISPR-Cas9 with and without repair templates. Both pCRISPR-Cas3 and pCRISPR-Cas9 introduced random sized deletions without repair templates. With repair templates, precise deletions were observed for both pCRISPR-Cas3 and pCRISPR-Cas9. The spacers 1 for Cas9 and CASCADE-Cas3 targeted the left flank of the BGC, spacers 2 the middle, and spacers 3 the right flank. (C) Efficiencies for actinorhodin deletions with pCRISPR-Cas9 and pCRISPR-Cas3. For pCRISPR-Cas3, the efficiencies were consistently high (>80%), while with pCRISPR-Cas9 the observed efficiencies were highly sgRNA dependent. Circles correspond to data for spacers 1, triangles for spacers 2, and rectangles for spacers 3. (D) Read alignments for the junction site of the two homologous flanks. A HindIII site was integrated, demonstrating that the DSB was repaired using the repair templates cloned into pCRISPR-Cas3 using HindIII. Shown in panel (C) are the means ± standard deviations of three deletion experiments targeting the actinorhodin region with three different spacers.

Figure 3.

Installation of highly efficient deletions in S. venezuelae (A), S. albidoflavus J1074 (B), and Streptomyces sp. NBC1270 (C). In S. venezuelae, antiSMASH region 22 (122 kb) was deleted, encoding a dense accumulation of BGCs. In S. albidoflavus J1074 and S.sp. NBC1270, the 27 kb fluostatin-like BGC encoded by region 10 was deleted. In all strains, deletions were achieved with high efficiencies, ranging from 75% to 100% of all screened colonies.

Figure 4.

Simultaneous deletions and integrations enable streamlined genome engineering. (A) The PhiC31 Streptomyces integrase integrates cargo DNA into target attB sites. The consensus attB site from S. coelicolor is 51 bp long and features a central TT sequence where the cargo is integrated. (B) Substitution of the entire actinorhodin BGC with an additional attB site. The attB site was cloned between the repair templates. Coverage plots of mappings of ONT data against the wild type and the in silico generated substitution strain reveal precise genome engineering. (C) pCRISPR-Cas3 was used for construction of a S. coelicolor expression host using both targeted and semi-targeted deletions and substitutions. (D) Oxford Nanopore sequencing results for all deletions based on minimap2 mappings to the reference genome. (E) The final strains S. coelicolor CW5 and CW6 both displayed >200% increase in actinorhodin production compared to the base strain S. coelicolor CW1 upon integration of an actinorhodin BGC BAC. (F) Phenotypes of S. coelicolor CW6 C and E2 without and with actinorhodin integrations. Shown in panel (E) are the means ± standard deviations of four biological replicates. Significance was tested using unpaired two tailed t-tests, where **P < .01, ***P < .001, and ****P < .0001.