Plasmid-curing investigation
Rapid plasmid-curing is important for genome editing, especially for iterative genome engineering. Due to the high copy number and intrinsic stability of modern cloning vectors, plasmid-curing can often be challenging. The broad-host-range plasmid pBBR1MCS2 features a multiple cloning site, mobilizability (RK2), and compatibility with diverse plasmid groups (IncP and IncQ), making it popular for gene expression and metabolic engineering in various bacteria, including P. putida [22]. However, plasmid stability can vary depending on hosts, growth conditions, selection pressure, and genetic content. For instance, Liu et al. observed the instability of pBBR-mpd in Sphingomonas sp. CDS-1 [23], while Cook et al. reported the instability of pBBR-UP, a pBBR1 variant, in P. putida [12]. The instability of the plasmid provides the possibility for rapid plasmid-curing, so we constructed a series of pBBR1MCS2 derivatives and investigated the influence of temperature changes and various CRISPR/Cas9 cargoes on pBBR1MCS2 stability in P. putida without antibiotics. Figure 1A depicts the loading of different cargoes onto pBBR1MCS2 before transfer into P. putida, resulting in strains PP01, PP02, and PP03. Subsequently, we evaluated plasmid curing efficiencies in these strains.
As shown in Fig. 1B, during 34 hours of continuous culture of PP01, we found pBBR1MCS2 to remain stable in P. putida at 30 oC, while it was eliminated within 14 hours at 35°C or 37°C. Each experiment was performed in triplicate, ensuring that the loss of plasmids was not due to random chance. Previous studies have linked plasmid instability to specific mutations. Jiang et al. reported that 3nt mutations (T-43A, A73G and A436G) in the replication origin of pBBR1MCS2 caused its instability in E. coli at 42 oC [24], Similarly, Cook et al. observed poor stability with the pBBR-UP mutation (C299T) [12]. Upon comparing these sequences, we confirmed that replication origin of pBBR1MCS2 in our study was identical to that of the wild type, ruling out mutations as a contributing factor to plasmid instability in our experiments.
The Streptococcus pyogenes (sp) CRISPR-Cas9 system, while adapted for genome editing across multiple bacteria, exhibits toxicity due to the double-strand breaks caused by SpCas9 cleavage or its tight binding to PAM sequences [25–27]. In our study, using the Plac promoter with a strong RBS sequence to express SpCas9 in E. coli DH5a consistently resulted in truncated mutations or stop codons within the Cas9 gene, confirming its toxicity. We mitigated this by modifying the sequence upstream of the Cas9 gene, removing spacers and intervals and replacing the RBS (Fig. S2). This enabled successful construction of pBBR-Cas9, which was then transferred into P. putida to create strain PP02. Figure 1C shows that in PP02, the Cas9 gene caused plasmid instability, with complete loss at both 35 oC and 37 oC within 6 hours. We also assessed the plasmid stability of strains PP01, PP02, and PP03 at 30 oC, where the presence of Cas9, sgRNA, and homology arms led to complete plasmid elimination within 8–10 hours. The plasmid curing was also confirmed by point-to-point streaking on both LB Km and LB agar plates (Fig. 1E). These observations suggest that rapid plasmid curing in P. putida using a pBBR1MCS2-based CRISPR/Cas9 system is highly feasible, providing a practical tool for genome editing applications.
Various strategies have been reported for plasmid curing in P. putida. For instance, over 10 cycles of overnight culture in plain LB media have been used for plasmid curing in a three-plasmid genome editing system [14]. A self-curing pQURE vector has been shown to facilitate suicide plasmid-based genome eiditing and rapid plasmid curing in P. putida [28]. Additionally, Lauritsen et al. developed the pFREE tool for rapid removal of various plasmids in E. coli and P. putida by using CRISPR-Cas9 targeting of plasmid replicons [29]. Compared with these existing methods, our findings suggest a promising approach for more rapid and facile plasmid curing after genome editing, requiring only an 8 h culture without the need for additional plasmids or inducers.
CRISPR/Cas9 system construction and optimization
The “all-in-one” CRISPR/Cas9 system (Fig. 2A) was constructed based on pBBR1MCS2, containing spCa9, sgRNA and homology arms, to facilitate streamlined genome editing and plasmid curing. Initially, we targeted a non-essential gene yqhD (PP_RS13015) with 5’-TCGGTAACCAAACCATTGCC-3’ as the target. The sgRNA, along with 1000 bp of the homology arms (500 bp each), was incorporated into pBBR-Cas9 to generate pBBR-Cas9ΔyqhD-HA500. As illustrated in Fig. 2B, the deletion of yqhD was confirmed via colony PCR and sequencing, achieving a deletion efficiency of 59% (10/17). We also attempted a large size deletion (ech-vdh-fcs, 4324 bp), which resulted in a low editing efficiency of 10% (2/20). Therefore, we further investigated the effects of different homology arm lengths and sgRNA targets on editing efficiency.
Increasing the length of homology arms can enhance recombination efficiency by facilitating homology-dependent recombination [18, 30, 31]. The effectiveness of the CRISPR/Cas9 system for precise genome editing is influenced by factors such as Cas9 expression, sgRNA transcript levels, sgRNA target sites, and homology arm lengths. Although increasing sgRNA or Cas9 levels may enhance editing efficiency in systems using two or three plasmids [32], in our single-plasmid system, such strategies may lead to cellular toxicity or plasmid mutations, ultimately resulting in editing failure. Consequently, we opted to increase the length of the homology arms to optimize our editing outcomes. As illustrated in Fig. 2C, we engineered a series of plasmids with varying homology arm lengths, ranging from 150 bp to 1200 bp per arm, targeting deletions of yqhD (500 bp) and ech-vdh-fcs (4324 bp). As shown in Fig. 2D, we observed a significant improvement in yqhD deletion efficiency when the homology arm length increased from 150 bp to 500 bp, but there was no statistically significant difference when the length increased further to 1200 bp. For the ech-vdh-fcs deletion, increasing homology arm length from 150 bp to 1200 bp did improve the editing efficiency (from 3–20%) when using sgRNA target 1 (sgRNA-T1), but the difference was not statistically significant. With sgRNA-T2, editing efficiency improved from 500 bp (18%) to 1200 bp (55%), with 1200 bp showing a statistically significant difference. Based on these results and the current 1800 bp limit for DNA fragment synthesis, we suggest using 500 bp homology arm length to ensure editing efficiency while facilitating synthesis and plasmid construction for future genome editing experiments. We further used our “all-in-one” CRISPR/Cas9 system to perform sequential genome editing for vdh and vanAB deletions, and the efficiency of vdh (1440 bp) was 38% and vanAB (2025 bp) was 17%. (Fig. 2F-G). It took less than 1.5 days to complete one cycle of genome editing in P. putida. Due to the rapid and simple plasmid-curing procedure, it is easy to carry out the second round of genome editing with a new pBBR-Cas9HA plasmid.
Several CRISPR systems have been reported for genome editing in P. putida. For instance, Cook et al. [12] developed a three-plasmid system consisting of pCas9 (a low-copy pRK2 plasmid with constitutive Cas9 expression and inducible λRed expression), pJOE (a suicide vector carrying the repair template), and pgRNA (pBBR1-UP carrying the sgRNA). Their two-step protocol achieved a high gene deletion efficiency (85–100%). However, the protocol has several drawbacks: the chromosomal integration of pJOE is very inefficient, and the process requires two rounds of transformation and selection, which are time-consuming. Additionally, although pgRNA was easily cured while pCas9 was maintained, the co-transformation of pJOE and pgRNA for subsequent editing resulted in low editing efficiency. Aparicio et al. [14] merged ssDNA recombineering with CRISPR/Cas9 in a three-plasmid system. They achieved a large cluster (~ 69 kb) deletion, however, the complex system composition and time-consuming plasmid curing make it difficult to apply for iterative genome editing. Sun et al. [13] introduced a two-plasmid system, with the pCas9 plasmid carrying both Cas9 and λRed genes, and pSEVA-gRNA carrying the repair template and sgRNA. They also incorporated PrhaB-dgRNA and SacB into the pCas9 plasmid, responsible for rhamnose-induced curing of pSEVA-gRNA and sucrose-assisted curing of pCas9, respectively. This system achieved 70–100% editing efficiencies and improved the plasmid curing. However, the large size of the pCas9 plasmid reduced the electroporation efficiency. Additionally, the use of L-arabinose for λRed induction, rhamnose for PrhaB-dgRNA induction, and sucrose for SacB selection increased the operational complexity, requiring proficient skills to successfully carry out genome editing using this system. Mougiakos et al. [16] introduced a one-plasmid system using a suicide plasmid pEMG carrying ThermoCas9/sgRNA/repair template, however, the low efficiency of chromosomal integration of pEMG is very inefficient and it requires two rounds of transformation and selection. Zhou et al. [15] reported a one-plasmid system that integrated Cas9n/λRed cassettes into P. putida genome, which requires inducing and curing processes and extends the time needed to achieve a single gene deletion to at least four days. Even though these methods reported high editing efficiency, the scientific community favored conventional methods over these CRISPR systems for P. putida, possibly due to multiple functional parts and complicated procedures, which hindered its adoption [5].
Table 1
Summary of gene deletion, substitution, and insertion in P. putida KT2440.
Editing Type | Gene location and function | Target size | Efficiency |
Deletion | | | |
catBC | PP_RS19345, PP_RS19340, Muconate cycloisomerase | 1434 bp | 13% (2/16) |
Substitution | | | |
yqhD::gfp | PP_RS13015, Iron-containing alcohol dehydrogenase | 500 bp::910 bp | 14% (2/14) |
yqhD::rfp | PP_RS13015, Iron-containing alcohol dehydrogenase | 500 bp::717 bp | 60% (12/20) |
PP_RS14690::MluI | PP_RS14690, Zinc-dependent alcohol dehydrogenase | 1080 bp | 13% (2/15) |
Insertion | | | |
::RBS FabH | Before PP_RS22730, Ketoacyl-ACP synthase III | 39 bp | 45% (9/20) |
To assess the adaptability and user-friendliness of our single-plasmid CRISPR/Cas9 system, one experienced PhD student and two new Master’s students new to this technology from different labs successfully performed genome editing in P. putida KT2440, as summarized in Table 1. The summary of gene editing experiments on P. putida KT2440, illustrated various genomic alterations: deletions, substitutions and insertion. For instance, the catBC genes were deleted with a 13% efficiency, removing 1434 bp. Substitutions of the yqhD gene with gfp or rfp reporter genes achieved 14% and 60% efficiencies, respectively. Additionally, a segment of the PP_RS14690 gene was substituted with the MluI restriction site, resulting in an 13% efficiency. An 45% efficiency was observed for inserting a ribosome binding site (RBS) upstream of the fabH gene.
These results effectively demonstrate the varying efficiencies of different gene editing using our CRISPR/Cas9 system in P. putida KT2440. Compared with existing CRISPR genome editing methods, our editing efficiency is not exceptionally high and the deletion size is limited. However, the system offers significant advantages such as simple single-plasmid construction and a rapid gene editing and plasmid curing process. The successful experiments conducted by graduate students highlight the accessibility and efficacy of our CRISPR/Cas9 system, emphasizing its user-friendliness, even for novice users. These achievements illustrate the potential for widespread adoption of this single-plasmid-based, easily curable CRISPR/Cas9 system and underscore its practical applicability across various genetic research settings.
Biosynthesis of valencene
As a demonstration of practical application, we engineered P. putida using the developed CRISPR/Cas9 system to achieve the biosynthesis of valencene. Valencene, prized for its citrus aroma has been pursued for de novo synthesis in recombinant strains such as S. cerevisiae and E. coli via different pathways (Fig. S3). P. putida KT2440 has emerged as a promising alternative due to its high tolerance to toxic byproducts, facilitating robust production processes for valencene and other terpenes.
As shown in Fig. 3A, the endA and endX genes were deleted sequentially to create a host strain PpΔend, enhancing exogenous gene expression stability [33]. Then, the plasmid pV01, carrying the valencene synthase (vs) gene from Callitropsis nootkatensis [34], was introduced into PPpΔend, resulting in the strain PpV01. This strain generated 2.11 mg/L of valencene. DXP synthase (dxs) is crucial in isoprenoid biosynthesis, playing a central role in the production of isopentenyl pyrophosphate (IPP) (Fig. S3 and Fig. 3B). The ispA (GPP/FPP synthase) catalyzes the condensation of IPP and DMAPP to form farnesyl pyrophosphate (FPP), the valencene precursor. Overexpression dxs and ispA in PpV01 generated strains PpV02 and PpV03, respectively. Notably, PpV02 with dxs overexpression with VS showed a significant 4.5-fold improvement in valencene production, but noverexpressing ispA with vs in PpV03 did not increase production significantly. These results indicated that enhancement of key enzymes of the upstream pathway of MEP is favorable to increasing the yield of the target product mainly due to the insufficient supply of precursors of the upstream pathway at this time.
In P. putida KT2440, Ac-CoA can naturally be converted to polyhydroxyalkanoate (PHA) by phaG and phaC1ZC2 [35]. Deleting phaG and phaC1ZC2 genes may reduce precursor depletion by the bypass pathway thus improve valencene production. As shown in Fig. 3C, deleting phaG in PpV04 did not improve valencene production, indicating the presence of other isoenzymes. However, PpV05, with phaG and phaC1ZC2 knocked out, exhibited an 11.8% increase in yield compared to PpV02. Further overexpression of ispA in strain PpV06 significantly increased valencene yield and cell density compared to PpV04.
Finally, adding another copy of the vs gene linked with ispA in genome using a designed α-helical rigid linker (EAAAK)2 [36] resulted in strain PpV07, which showed a significant improvement in valencene yield. Remarkably, the engineered PpV07 achieved a 10-fold increase in valencene production compared to the starting strain PpV01. Overexpressing dxs, ispA, and vs genes mainly contributed to valencene yield, while deleting phaG and phaC1ZC2 genes contributed indirectly by enhancing cell growth.
Heterologous biosynthesis of valencene has been reported in different microorganisms [37]. The highest valencene yield (12.4 g/L) was reported in Saccharomyces cerevisiae through gene screening, protein engineering, and pathway optimization [38]. Bacteria as hosts showed relatively lower valencene production, ranging from 2.41 mg/L in Corynebacterium glutamicum to 352 mg/L in Rhodobacter sphaeroides. This study is the first report of valencene biosynthesis using P. putida as a host. Although our current yield is not as high as in engineered yeast, the fast growth rate of P. putida and our advantageous CRISPR/Cas9 editing technology allow for faster metabolic engineering, thus having great potential in increasing valencene production.