Single-Stranded DNA-Binding Protein and Exogenous RecBCD Inhibitors Enhance Phage-Derived Homologous Recombination in Pseudomonas.

The limited efficiency of the available tools for genetic manipulation of Pseudomonas limits fundamental research and utilization of this genus. We explored the properties of a lambda Red-like operon (BAS) from Pseudomonas aeruginosa phage Ab31 and a Rac bacteriophage RecET-like operon (RecTEPsy) from Pseudomonas syringae pv. syringae B728a. Compared with RecTEPsy, the BAS operon was functional at a higher temperature indicating potential to be a generic system for Pseudomonas. Owing to the lack of RecBCD inhibitor in the BAS operon, we added Redγ or Pluγ and found increased recombineering efficiencies in P. aeruginosa and Pseudomonas fluorescens but not in Pseudomonas putida and P. syringae. Overexpression of single-stranded DNA-binding protein enhanced recombineering in several contexts including RecET recombination in E. coli. The utility of these systems was demonstrated by engineering P. aeruginosa genomes to create an attenuated rhamnolipid producer. Our work enhances the potential for functional genomics in Pseudomonas.

Introduction

Recombinant DNA technology in Escherichia coli (E. coli) has been greatly boosted by bacteriophage-encoded recombination systems (Muyrers et al., 1999, Zhang et al., 1998, Zhang et al., 2000). A decisive advantage of this technology, termed recombinogenic engineering or recombineering, is the easy incorporation of the flanking short homology sequences required for homologous recombination into synthetic oligonucleotides. Oligonucleotides and linear dsDNAs with homology arms as short as 35 nucleotides have been exploited in many ways for point mutations, deletions, insertions, and subcloning with base pair precision regardless of the size of the target DNA (Murphy, 1998, Muyrers et al., 1999, Zhang et al., 1998, Zhang et al., 2000).

In E. coli, recombineering utilizes the Redα, Redβ, and Redγ proteins from the lambda phage Red operon or RecE and RecT from the Rac prophage (Fu et al., 2012, Zhang et al., 1998). Redα and RecE are 5′-3′ exonucleases that generate single-stranded DNA (ssDNA) intermediates (Maresca et al., 2010). Redβ and RecT are single-stranded DNA-annealing proteins (SSAPs) that bind to ssDNA to promote the search for the complementary sequence (Erler et al., 2009). The Redα/Redβ and RecE/RecT phage pairs each include a specific protein-protein interaction, which is required for double-stranded DNA (dsDNA) homologous recombination (Muyrers et al., 2000).

Redγ forms a dimer to mimic DNA that binds to and inhibits the exonuclease and helicase activities of the RecBCD complex (Court et al., 2007, Murphy, 1991), which aggressively degrades linear dsDNA (Court et al., 2007, Wang et al., 2006). The RecET operon does not appear to have a Redγ equivalent. However, the inclusion of Redγ with RecE/RecT increases homologous recombination efficiency via increased persistence of the linear dsDNA substrates (Fu et al., 2012, Zhang et al., 1998). The significance of a RecBCD inhibitor for homologous recombination efficiency was also demonstrated in our study describing Photorhabdus luminescens Pluγ for the development of a recombineering system for Photorhabdus and Xenorhabdus. We found that Pluγ could inhibit the RecBCD complex in both Photorhabdus and E. coli (Yin et al., 2015).

The Red system has been applied to precisely and fluently edit the genome of not only E. coli but also closely related bacteria Salmonella enterica (Bunny et al., 2002), Yersinia pseudotuberculosis (Derbise et al., 2003), Shigella (Beloin et al., 2003), Serratia (Rossi et al., 2003), and Escherichia albertii (Egan et al., 2016). However, its wider application has been limited by apparent host specificities (Yin et al., 2015). Consequently endogenous SSAPs alone or together with partner exonucleases have been used for oligo repair or cassette insertion in Mycobacterium tuberculosis (Van Kessel and Hatfull, 2006), Lactococcus lactis (Pijkeren and Britton, 2012), Lactococcus reuteri (Pijkeren et al., 2012), Clostridium acetobutylicum (Dong et al., 2013), Lactobacillus plantarum (Yang et al., 2015), Bacillus subtilis (Sun et al., 2015), P. luminescens, and Xenorhabdus stockiae (Yin et al., 2015).

Pseudomonas is a gram-negative, aerobic rod that belongs to the bacterial family Pseudomonadaceae (EUZéBY, 1997). The best characterized species include the opportunistic human pathogen Pseudomonas aeruginosa (Stover et al., 2000), the plant pathogen Pseudomonas syringae (Xin and He, 2013), the plant growth-promoting Pseudomonas fluorescens (Paulsen et al., 2005), and the soil bacterium Pseudomonas putida, which is used in bioremediation (Gomes et al., 2005) and biocontrol (Validov et al., 2007). The members of this genus demonstrate a great deal of metabolic diversity (Aditi et al., 2017).

Here, we report the development of recombineering systems for Pseudomonas based on two host-specific phage protein-encoding operons from P. aeruginosa phage Ab31 and P. syringae. BLAST analysis suggested that Orf38, Orf37, and Orf36 are analogs of Redβ, Redα, and single-stranded DNA-binding protein (SSB), respectively. Recombineering experiments indicated that Redγ or Pluγ worked in some Pseudomonas species and SSB significantly increased efficiency. SSBs are often found in recombinase-encoding operons from various phages (Szczepańska, 2009). Because proteins in the same operon are usually functionally associated, we were particularly interested in evaluating the contribution of orf36/S to phage recombinase-mediated homologous recombination. Using these systems, we efficiently modified Pseudomonas genomes including gene deletions and insertions. In particular, a highly attenuated rhamnolipid producer was obtained after deleting pathogenic factors and overexpressing rhlAB and estA in P. aeruginosa.

Results

Endogenous Phage exo/SSAP Protein Pairs in Pseudomonas

We looked for candidate DNA recombination proteins in Pseudomonas and Pseudomonas phage genomes with BLAST using the coding sequences of λ Redβ, rac RecT, or Pluβ as queries in a non-redundant protein sequence database. Two operons including an exonuclease (exo) and SSAP were identified. One was RecTEPsy from P. syringae, which was previously developed as a recombineering system for use in P. syringae (Swingle et al., 2010). The second operon, from P. aeruginosa phage vB_PaeP_Tr60_Ab31, encoded three proteins including a candidate SSAP (orf38, here named B) that was 62% identical to Redβ over its 177-amino acid sequence (Figure 1) and was adjacent to a candidate exonuclease (orf37, here named A; Figure 1), which showed significant similarity to Redα (sequence identity of 32% in a 195-amino acid region). The next coding region, orf36 (here named S), showed significant similarity to SSBs (sequence identity of 54% across 148 amino acids of the E. coli SSB; Figure 1). SSBs are often found in recombinase-encoding operons from various phages (Szczepańska, 2009). Because proteins in the same operon are usually functionally associated, we evaluated the contribution of orf36/S to phage recombinase-mediated homologous recombination and we named this candidate recombineering operon, BAS.

Figure 1

Operon Architecture of Lambda Red and Pseudomonas Phage vB_PaeP_Tr60_Ab31 orf36-orf38 Orf38, Orf37, and Orf36 Are Related to Redβ (beta), Redα (alpha), and SSB, Respectively

The amino acid sequences were compared via ClustalW alignment (Figure S1). The percent identities are displayed between the homology regions.

In gram-negative bacteria, RecBCD, which is the major exonuclease in E. coli, is highly conserved (Table S5). Notably neither RecTEPsy nor BAS operons contained a homolog of Redγ or Pluγ, which are inhibitors of RecBCD exonuclease activity. Because co-expression of Redγ with Redβα or RecET significantly increases recombineering efficiency (Fu et al., 2012), the addition of Redγ or Pluγ to the RecTEPsy and BAS operons were tested for their abilities to increase recombineering efficiency.

Critical Time Points for Electroporation of Pseudomonas

In E. coli, transformation efficiency determines optimal recombineering (Sharan et al., 2009). When E. coli DH10B-derived cells are used, the recombinant proteins are usually induced when the cells enter log phase growth (OD600 = 0.30–0.35). After two cell divisions, at OD600 = 0.70–0.80, the electrocompetent cells are prepared (Fu et al., 2010). This principle was also verified by recombineering in Photorhabdus (Yin et al., 2015). To optimize protocols for Pseudomonas, we first plotted the growth of the four Pseudomonas strains at 30°C. Overnight cultures were diluted to OD600 ≈ 0.085 to start the growth-monitoring cultures. After approximately 2 h, P. syringae and P. aeruginosa cultures entered the log phase and the plasmid pBBR1-Rha-GFP-kan was transformed into electrocompetent cells prepared at different time points (Figure 2). We found that cells prepared at 4 and 2.5 h, respectively, yielded the most transformants (Figures 2B and 2D). The same test was performed for P. fluorescens (Figures S2A and S2B) and P. putida (Figures S2C and S2D). Thereby we established the time points for induction and electrocompetent preparation for each strain (Table S6).

Figure 2

Optimization of Transformation in P. syringeae and P. aeruginosa

(A) Growth curve of P. syringeae. The optical density (OD) at 600 nm (OD600) was measured from a starting OD600 of 0.085.

(B) Time when cells were harvested for electroporation.

(C and D) (C) and (D) are the same as (A) and (B), respectively, for P. aeruginosa. After electroporation of pBBR1-rha-GFP-km, colonies were selected on kanamycin plates and counted.

Data represent the mean ± SD from three independent experiments. The statistical analysis used is Student's t test. *p < 0.05, **p < 0.01, ***p < 0.001.

Efficiency of Recombineering Systems in Pseudomonas

We compared the RecTEPsy, BAS, Redγβα, and Pluγβα systems in four Pseudomonas strains. Pluγ or Redγ was added to the RecTEPsy and BAS operons to generate four more candidate systems: PluγTEPsy, RedγTEPsy, PluγBAS, and RedγBAS. The eight expression plasmids were transformed into the four Pseudomonas strains, followed by plasmid DNA extraction and restriction analysis for confirmation. The expression plasmids were based on a broad host range origin (pBBR1) (Antoine and Locht, 1992), and the rhaR-rhaS PRha promoter (Egan and Schleif, 1993, Egan and Schleif, 1994) was used for inducible expression of the recombination operons.

To compare the recombineering efficiency of different systems in one species, all the electroporation was carried out using competent cells prepared from a 1.3 mL fresh culture that are adjusted to the same cell density. After electroporation, 1 mL Luria-Bertani (LB) medium was added for recovery. The data presented are exactly the result from one electroporation, and this methodology was utilized in our previous study (Fu et al., 2012, Yin et al., 2015).

We used an assay based on the integration of a PCR product containing gentamycin resistance gene flanked by 100 bp homology arms into the pBBR1 expression plasmid to obtain the plasmid product, pBBR1-Kan-Genta (Figure 3A). Results are based on counting gentamycin-resistant colonies followed by verification of pBBR1-Kan-Genta by restriction analysis.

Figure 3

Recombineering with Different Protein Combinations in Different Pseudomonas Species

(A) Diagram of the recombineering assay. A PCR product carrying a gentamycin resistance gene (Genta) flanked by 100-bp homology arms (thick lines) was integrated into the expression plasmid in place of the artificial operon of the recombination system.

(B) Results from the recombineering assay in P. fluorescens upon the expression of Pluγβα, Redγβα, TEPsy, PluγTEPsy, RedγTEPsy, BAS, PluγBAS, and RedγBAS at 30°C.

(C) Results from P. syringae, which was set up as described in (B).

(D) Results from P. putida, which was set up as described in (B), with the exception that RedγTEPsy was missing owing to the failure of transformation.

(E) Results from P. aeruginosa at 30°C, which was set up as described in (B), and the correct rates are indicated at the top of each column.

(F) Results from P. aeruginosa at 30°C, which was calculated from the correct rates of (E), and the y axis represents the number of correct colony.

(G and H) (G) and (H) are as in (E) and (F), respectively, except the proteins were expressed at 37°C. Colonies were selected on gentamycin plates and counted.

Data represent the mean ± SD from three independent experiments. The statistical analysis used is Student's t test. *p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001.

P. fluorescens, P. syringae, and P. putida were grown and tested at 30°C (Figures 3B–3D). Eight recombinants of pBBR1-Kan-Genta were verified by restriction analysis and found to be 100% in all three strains. Expression of Pluγ and Redγ enhanced recombination in P. fluorescens but appeared to impair it in P. putida possibly because the expression of Pluγ or Redγ had a toxic effect. Indeed, we could not cultivate P. putida carrying the RedγTEPsy expression plasmid. In all three strains, both TEPsy and BAS systems worked better than Redγβα and Pluγβα.

A different outcome was observed in P. aeruginosa where the acquisition of gentamycin resistance, possibly due to intrinsic multidrug resistance mechanisms (Savoia, 2014), did not faithfully reflect recombination into pBBR1. Consequently, both gentamycin-resistant colonies and the correct recombination rates are presented in Figures 3E–3H. Because P. aeruginosa normally grows at 37°C, we tested recombination at both 30°C and 37°C. At 30°C, Pluγ significantly increased the efficiency of both the TEPsy and BAS systems, but it failed to have an effect at 37°C indicating thermolability. In contrast, Redγ increased the efficiency of BAS at both 30°C and 37°C. As with the other three strains, the PluγTEPsy, PluγBAS, and RedγBAS systems were superior to Redγβα and Pluγβα. Notably, the correct ratio of PluγTEPsy at 30°C was 92%, which was the most efficient of all variations tested in P. aeruginosa. However, PluγTEPsy showed no activity at 37°C and RedγBAS presented the best performance (Figures 3E–3H). We suggest that Pluγ and TEPsy are inactive above 30°C and Redγ and BAS are active at 37°C. These temperature optima concord with the origins of these proteins.

Functional Dissection of BAS

Based on the above evaluation, the BAS system, with or without Redγ, can be used for genome engineering in various Pseudomonas strains. Because the single-stranded DNA-binding protein S is the unusual aspect of this operon, we further investigated its contribution to recombineering first in E. coli where neither BA nor BAS functioned well without Redγ. However, in the presence of Redγ, S promoted recombineering efficiency about 4-fold (Figure 4A). A similar effect was found in P. aeruginosa (Figure 4B), whereas in P. syringae, Redγ stimulated recombination in the absence of S but impaired it in its presence (Figure 4C).

Figure 4

Functional Analysis of SSB

(A–C) Results from the recombineering assay in Figure 3A in E. coli, (A), P. aeruginosa (B), and P. syringae (C) upon expression of BA or RedγBA in the presence and absence of SSB. Colonies were selected on gentamycin plates and counted.

(D) Results from LCHR assay in E. coli upon expression of RedγBA or RecET-Redγ in the presence and absence of SSB.

(E) As in (D) except using LCHR assay.

(F) Efficiency of LLHR mediated by ETg, ETgSSBphage_Ab31, ETgSSB, and ETgSSBF_plasmid. Colonies were selected on chloramphenicol or kanamycin plates and counted.

Data represent the mean ± SD from three independent experiments. The statistical analysis used is Student's t test. *p < 0.05, **p < 0.001, ***p < 0.0001.

To further investigate the effect of S, we evaluated its effect of Red- and RecET-mediated recombination, which differ mechanistically. Redγβα requires a replicating substrate and acts at the replication fork (Maresca et al., 2010), whereas full-length RecE/RecT recombines two linear DNAs before replication (Fu et al., 2012). We term these two mechanisms linear plus circular homologous recombination (LCHR) and linear plus linear homologous recombination (LLHR). Addition of S to Red LCHR has a small positive effect, whereas it substantially enhances RecET LCHR (Figure 4D). Conversely, S does not convey LLHR capacity onto Red but substantially enhances RecET LLHR (Figure 4E). Consequently, we tried two more SSBs, which are from E. coli and F plasmid, and both also enhanced RecET LLHR (Figure 4F).

Optimization of Recombineering in Pseudomonas

After identifying the optimal recombineering configurations for each Pseudomonas strain, i.e., PluγTEPsy for P. aeruginosa at 30°C, RedγBAS for P. aeruginosa at 37°C, BAS for P. putida and P. syringae, and RedγTEPsy or RedγBAS for P. fluorescens, we further optimized the protocols. Apart from electrocompetent cell preparation, the amount of DNA to be transformed and the length of its homology arm to the target play large roles in determining efficiency (Fu et al., 2012, Yin et al., 2015). We titrated the DNA amount from 100 ng to 2 μg. It was observed that 1 μg DNA was sufficient for P. fluorescens, P. syringae, and P. putida (Figures S3A, S3C, and S3E). However, P. aeruginosa required 1.5 μg of DNA to achieve a decent efficiency (Figure S3G). For all three strains, the saturation amount of DNA was 1.5 μg. The length of the homology arm was tested across a range from 30 to 150 bp. One hundred base pairs were sufficient for P. fluorescens (Figure S3B) and P. putida (Figure S3F). P. syringae (Figure S3D) and P. aeruginosa (Figure S3H) required 120 bp for significantly higher efficiency. Although a 150-bp homology arm further boosted the efficiency for P. aeruginosa, we did not employ this advantage owing to cost and concerns over the quality of long oligonucleotide syntheses.

Recombineering Pseudomonas Genomes

To validate the methods, we designed several genome engineering exercises beginning with the construction of an attenuated P. aeruginosa PAO1 mutant. P. aeruginosa produces a number of virulence factors, including exotoxin A (McEwan et al., 2012), pyocyanin (Yang et al., 2016), elastase (McIver et al., 1991), and periplasmic glucans (Mahajan-Miklos et al., 1999), which prevent the safe use of this strain for the production of rhamnolipids. As diagrammed in Figure 5A, a number of gene knockouts were performed by electroporating a PCR product containing a gentamycin resistance gene flanked by lox71/lox66 sites and 75-bp homology arms. The aroA gene (1917 bp) encodes 3-phosphoshikimate 1-carboxyvinyltransferase, which is essential for the synthesis of aromatic amino acids. Deletion of aroA can completely prevent P. aeruginosa infection of mammalian cells (Priebe et al., 2002). The phnAB (2172bp) gene is involved in pyocyanin biosynthesis (Lau et al., 2004). The lasB gene (1497 bp) encodes elastase (McIver et al., 1991). The mdoH gene (2586 bp) is required for the biosynthesis of periplasmic glucans (Mahajan-Miklos et al., 1999). To knock out these genes, each electroporation yielded approximately 100 gentamycin-resistant colonies when PluγTEPsy was expressed, and all the clones were correct. We also used the RedγBAS system and noticed approximately 1000 small and 100 large gentamycin-resistant colonies, but only about half of the large colonies were correct according to colony PCR (Figures 5B, 5D, 5F, and 5H). To generate a selectable marker-free P. aeruginosa PAO1 mutant, Cre catalyzed the excision of the lox-flanked DNA (Figures 5C, 5E, 5G, and 5I).

Figure 5

PCR Verification of the Genome Engineering of P. aeruginosa by Recombineering through RedγBAS System at 37°C

(A) Diagram of the genome engineering of P. aeruginosa and schematic presentation of the PCR setup for verification.

(B) aroA gene knockout via replacement with a gentamycin selection marker. Lane C is the wild-type strain, used as a negative control. Lane M is the Takara DL5000 marker. Lanes 1 and 2 are recombinants. Both clones 1 and 2 are correct.

(C) Gentamycin selection marker deletion via Cre recombination for aroA gene knockout. Both clones 1 and 2 are correct.

(D) phnAB gene knockout via replacement with a gentamycin selection marker. Both clones 1 and 2 are correct.

(E) Gentamycin selection marker deletion via Cre recombination for phnAB gene knockout. Six clones are correct.

(F) lasB gene knockout via replacement with a gentamycin selection marker. Both clones 1 and 2 are correct.

(G) Gentamycin selection marker deletion via Cre recombination after lasB gene knockout. Both clones are correct.

(H) mdoH gene knockout via replacement with a gentamycin selection marker. The tested clone is correct.

(I) Gentamycin selection marker deletion via Cre recombination after mdoH gene knockout. Four clones are correct.

Pseudomonas exotoxin A is a single-chain toxin with three structural domains that inhibit protein synthesis in eukaryotic cells by catalyzing ADP ribosylation of elongation factor 2. The amino-terminal domain I is involved in eukaryotic cell recognition (Jinno et al., 1988), the central domain II may be involved in the translocation function of the protein (Siegall et al., 1989), and the carboxy-terminal domain III has a cleft that is proposed to be the enzyme active site (Carroll and Collier, 1987). Glu-578 in domain III has been identified as an active site residue for nicotinamide adenine dinucleotide binding (Carroll and Collier, 1987). To evaluate the ability of the recombineering system to achieve a seamless mutation, we aimed to mutate this residue. A single-nucleotide substitution of ToxA (GAG Glu-578 codon to GAC Asp) was achieved by two rounds of recombineering (Figure S4A). First, a PCR product with a gentamycin resistance and counter-selection (SacB) genes flanked by 75-bp homology arms was introduced into P. aeruginosa expressing PluγTEPsy. After the gentamycin-resistant clones were verified by colony PCR (Figure S4B), SacB function was tested (Figure S5). In the second round of recombineering, a point mutation (GAG Glu-578 codon to GAC Asp) was introduced using 121-bp oligonucleotide with 60-bp homology arms either side of the point mutation. The reaction was mediated by PluγTEPsy, and the mutant strain was screened on LB plates containing 10% sucrose. Successful site-directed mutagenesis was confirmed by colony PCR (Figure S4C) and sequencing (Figure S4D).

The BAS system was also successfully applied to genome engineering of P. syringae pv. tomato str. DC3000 by placing a promoter (Pgenta) in front of a silent gene cluster for a secondary metabolite pathway. Insertion of the promoter was efficient. Each electroporation yielded about 120 gentamicin-resistant colonies. Three colonies were verified by colony PCR (Figure S6A), and all of them were correct. However, we could not detect any new compound produced by the engineered strain (data not shown). We also engineered the P. fluorescens genome with the RedγTEPsy system. The retS gene of P. fluorescens was knocked out, and the mutant strain was confirmed by colony PCR (Figure S6B). Each electroporation yielded approximately 20 gentamicin-resistant colonies, and all of them were correct. The hybrid sensor kinase (RetS) negatively controls the production of antibiotics, including 2,4-diacetylphloroglucinol, pyrrolnitrin, and pyoluteorin, in P. fluorescens (Zhang et al., 2015). Genetic inactivation of retS results in enhanced biocontrol capacity of the strain (Zhang et al., 2015).

Using the Attenuated P. aeruginosa PAO1 Mutant for Rhamnolipid Production

Although much effort has been invested in heterologous expression of rhlAB, the pathogenic strain P. aeruginosa PAO1 remains the best producer of rhamnolipids (Zhao et al., 2015). To evaluate the safety of an attenuated P. aeruginosa PAO1 strain, we examined their virulence in mice and C. elegans models. The 50% lethal dose for wild-type PAO1 given by intraperitoneal (i.p.) injection was 4 × 108 colony-forming unit (CFU). However, no toxic effects were observed with i.p. doses of the attenuated mutant PAO1aroA up to 1 × 109 CFU (Figure 6A). In contrast, the PABMT mutant (phnAB−, lasB−, mdoH−, and toxA−) was still pathogenic to BALB/c mice (Table S7) and postponed the egg stage in C. elegans (Figure S7).

Figure 6

Overexpression of rhlAB and estA in the PAO1 Mutant for Rhamnolipid Production

(A) The survival rate of BALB/c mice after intraperitoneal injection of wild-type PAO1 and the PAO1aroA mutant strain (1*109 CFU).

(B) Rhamnolipid production in wild-type PAO1 and the PAO1aroA mutant from a 2-day fermentation. The statistical analysis used is Student's t test. *p < 0.05, **p < 0.01.

(C) Emulsification activity of cell culture supernatants. Equal parts of toluene and supernatant were combined in a glass vial, vortexed vigorously, and rested for 2 h.

(D) Plasmid stability in the PAO1aroA mutant. The tested plasmid was pBBRI-estA-genta-rhlAB. This strain was cultured in Murashige and Skoog (MS) medium supplemented with aromatic amino acids without antibiotic, and the number of cells was counted on LB plates with and without gentamycin every day. The ratio of gentamycin-resistant clones to total cells was calculated. Data represent the mean ± SD from three independent experiments.

The PAO1aroA mutant is auxotrophic for aromatic amino acids and must be cultivated with aromatic amino acid supplement. Under these conditions, the wild-type PAO1 strain entered log phase in 2 h, but the PAO1aroA mutant strain grew more slowly, entering log phase in 5 h (Figure S8). Nevertheless, when we compared rhamnolipid production after 48 h of fermentation, there was no significant difference between the wild-type PAO1 and PAO1aroA mutant strains (Figure 6B).

To increase rhamnolipid production in the PAO1aroA mutant, we introduced a plasmid carrying rhlAB, which participates in the biosynthesis of this biosurfactant (Ochsner et al., 1994), and the estA gene, which encodes an autotransporter protein located in the outer membrane (Wilhelm et al., 2007) (Figure S9). Overexpression of rhlAB and estA in wild-type PAO1 and the PAO1aroA mutant increased the yield of rhamnolipids (Figure 6B). Using a toluene emulsion assay (Biggins et al., 2014), we tested the emulsive potential of the PAO1aroA mutant in the presence of rhlAB and estA. Visible emulsions formed in the supernatants of both wild-type PAO1 and the PAO1aroA mutant (Figure 6C). We also confirmed the stability of the expression plasmid pBBR1-estA-genta-rhlAB, which replicated in the PAO1aroA mutant without antibiotic selection (Figure 6D). Thus a safe rhamnolipid producer for industrial fermentation has been established.

Discussion

Although the Red system has been applied to engineer the genome of E. coli and a number of genetically close species, apparent host-specific factors limit its wider application (Yin et al., 2015). Consequently, we and others have searched for other phage exonuclease/SSAP pairs associated with genetically distant bacterium. This includes the prophage RecETPsy system, which has been used in P. syringae so far (Swingle et al., 2010).

Several attempts of recombineering in P. putida have been recently reported. However, the enzymes are lambda Red (Chen et al., 2016, Cook et al., 2018, Luo et al., 2016) and rac RecET (Choi et al., 2018), or an SSAP without an accompanying exonuclease (Aparicio et al., 2016, Aparicio et al., 2018, Ricaurte et al., 2018). It is well known that the use of SSAPs alone is only effective with single-stranded oligonucleotides and will not mediate the dsDNA homologous recombination events.

Pseudomonas is a large genus including P. aeruginosa, P. chlororaphis, P. fluorescens, P. pertucinogena, P. putida, P. stutzeri, P. syringae, and a large number of Incertae sedis. In a search for recombineering solutions to cover this broad range, we worked with representatives from four groups, namely, P. aeruginosa, P. fluorescens, P. putida, and P. syringae. In addition to RecETPsy, we identified another operon in the Pseudomonas phage vB_PaeP_Tr60_Ab31 (Latino et al., 2014) (Figure 1). These operons differ from the E. coli lambda Red operon (Redγβα) or our previously reported P. luminescens operon (Pluγβα) (Yin et al., 2015) because they do not appear to include a Redγ/Pluγ-type inhibitor of the major host exonuclease RecBCD, which protects linear dsDNA ends from degradation and thereby facilitates recombination. Therefore we added Pluγ and Redγ to the TEPsy and BAS systems to evaluate their impact on recombineering. Owing to apparent toxicity and temperature optima, we did not arrive at a single optimal configuration for all four Pseudomonas species. In P. aeruginosa PluγTEPsy delivered the best results at 30°C, whereas RedγBAS was the configuration for 37°C. Furthermore, Redγ and Pluγ increased recombineering efficiency in P. aeruginosa and P. fluorescens, but Redγ was deleterious in P. putida and P. syringae. We suggest that Redγ likely alters the function of the RecBCD complex in E. coli, P. aeruginosa, and P. fluorescens in the same manner by temporarily blocking its exonuclease activity. However, the interaction between Redγ and RecBCD in P. syringae or P. putida might proceed differently, potentially to have an effect on RecBCD-mediated genome repair.

Consequently, after evaluating different configurations in four Pseudomonas species, we found that the BAS system was functional in all four tested Pseudomonas strains and so has the potential to be a generic system for Pseudomonas after testing for Pluγ or Redγ functionality.

We report the usefulness of an SSB in a recombineering system. The three-element BAS and four-element hybrid RedγBAS systems are useful genome engineering tools in certain Pseudomonas strains. Consequently, we were motivated to explore the utility of S, the phage Ab31 SSB, in both established recombineering paradigms, Red and RecET in LCHR and LLHR applications.

In E. coli, S clearly contributed to BA recombination. Addition of S had little impact on Red recombination. However, it substantially enhanced RecET in both LCHR and LLHR assays (Figures 4D and 4E).

SSBs play crucial roles in DNA replication, recombination, DNA damage signaling, and repair in all organisms (Yang et al., 2013). In addition, SSBs can protect ssDNA from further degradation after 5′-3′ exonuclease resection (Meyer and Laine, 1990). SSB also forms a protein interaction platform by recruiting DNA replication and recombination enzymes, such as exonuclease I (Lu et al., 2011) and RecJ (Han et al., 2006). Biochemical studies of different recombination systems have revealed that SSB binds to and removes secondary structures from ssDNA during the presynapsis stage (Kowalczykowski and Krupp, 1987), which facilitates homologous strand pairing. Therefore, an SSB could potentially contribute to recombineering in three ways, namely, protection of ssDNAs involved in recombination from degradation, interaction(s) with the recombination machinery, or removal of secondary structures in ssDNA intermediates. We designed an experiment to test this third possibility to determine if homologous recombination efficiency was affected by secondary structure of the homology arm and whether an SSB could ameliorate the impact. However, on comparison of a homology arm with notable secondary structure to an arm with moderate secondary structure, there was no bias with or without SSB overexpression in the recombineering host (Figures 7A–7C). Although further studies are necessary to elucidate the biological role of SSB in recombineering, we suggest that the variable contributions of a RecBCD inhibitor or an SSB in different contexts reflect different recombination susceptibilities to double- or single-stranded exonucleases.

Figure 7

Effects of Secondary Structure on Recombineering with or without SSB

(A) Secondary structure predictions of the standard homology arm (HA-STD) and tested homology arms (HA-A, HA-B, and HA-D) from the RNAfold Webserver (http://rna.tbi.univie.ac.at/cgi-bin/RNAWebSuite/RNAfold.cgi) (Mathews et al., 2004).

(B) Diagram of the LLHR assay in E. coli (Fu et al., 2012). Each end of the kan-PCR product has a 50-bp homology arm to the p15A-cm PCR product between the chloramphenicol gene (cm) and the p15A origin (indicated by red and green arrows).

(C) Results from the LLHR assay depicted in (B) upon expression of full-length RecE, RecT, and Redγ with or without SSB. Colonies were selected on kanamycin or chloramphenicol plates and counted.

Data represent the mean ± SD from three independent experiments. The statistical analysis used is Student's t test. *p < 0.05, ns: not significant.

Based on the RedγBAS and PluγTEPsy recombineering systems, we constructed two P. aeruginosa mutants, PABMT and PAO1aroA. PABMT could postpone the egg stage for C. elegans but was still pathogenic to BALB/c mice, suggesting the presence of additional pathogenic factors. PAO1aroA is a highly attenuated strain (Figure 6A) with potential for several applications, such as live vaccines (Priebe et al., 2002). In the present study, the highly attenuated mutant was used to produce rhamnolipid biosurfactants. Rhamnolipids are surface-active secondary metabolites produced by P. aeruginosa or related species in the stationary phase (Zhang et al., 2012). In the last decade, these molecules have emerged as a promising class of biosurfactants for several applications, such as moisturizers, lubricants, and shampoos (Dobler et al., 2016). Rhamnolipids are also effective in the bioremediation of organic and heavy-metal-polluted sites. Consequently, various efforts, including heterologous expression of rhlAB, have been made to establish high-yield rhamnolipid-producing strains. However, P. aeruginosa remains the best rhamnolipid producer. Based on the attenuated strain, further modifications to increase secondary metabolite production would involve the overexpression of proteins that participate in biosynthetic and/or regulatory pathways (Dobler et al., 2016). As expected, the overexpression of rhlAB and estA in wild-type PAO1 and the PAO1aroA mutant strain increased the yield of rhamnolipids (Figure 6B). The rhamnosyltransferase 1 complex (RhlAB) is the key enzyme responsible for transferring the rhamnose moiety to the β-hydroxyalkanoic acid moiety during rhamnolipid biosynthesis (Wang et al., 2007). Rhamnolipid production in P. aeruginosa is regulated by the hierarchical quorum-sensing systems (LasI/R and RhlI/R systems) (Soberón-Chávez et al., 2005). EstA indirectly influences the synthesis of quorum-sensing molecules by providing the cells with fatty acids (Riedel et al., 2003).

For P. aeruginosa PAO1, after electroporation the recombinants were selected on LB agar plates supplemented with gentamycin (15 μg/mL). There were false-positive colonies, which were smaller. According to colony PCR using the internal primer pairs binding to the gentamycin resistance gene, this was not random integration of the cassette (Figure S10). As negative controls we observed that there were gentamycin-resistant colonies of P. aeruginosa after electroporation without adding DNA. Most likely the intrinsic multidrug resistance mechanisms of P. aeruginosa were enhanced after electroporation as a stress response. However these false-positive colonies were not problematic for screening, as they did not grow in liquid LB containing gentamycin (15 μg/mL) because of antibiotic selection being more stringent in the liquid medium than in the solid medium (data not shown).

Here, we developed recombineering systems for Pseudomonas for several reasons including circumventing the inconvenience of constructing a suicide plasmid with long homology arms for genome engineering, which then requires tedious PCR screening of single-crossover and double-crossover events (Hmelo et al., 2015). Further work to construct efficient recombineering systems for more distantly related gram-negative bacteria may require not only the identification of new Red-like or RecET-like operons but also testing combinations of SSBs and Redγ-like proteins with the phage exonuclease/SSAP pairs.

Limitations of the Study

We evaluated two Pseudomonas endogenous phage recombinant systems (BAS and RecETpsy) in various strains and use the E. coli Red and Plu Red recombinant proteins (Redγβα and Pluγβα) as references. These proteins were expressed using the same inducible promoter and the same plasmid origin. However, we did not assess the expression levels of such proteins, so future experiments are needed to test whether differences in expression levels of these proteins can explain different recombineering efficiencies among strains.

Methods

All methods can be found in the accompanying Transparent Methods supplemental file.