Transforming the untransformable with knockout minicircles

Abstract Gene inactivation studies are critical in pathogenic bacteria, where insights into species biology can guide the development of vaccines and treatments. Allelic exchange via homologous recombination is a generic method of targeted gene editing in bacteria. However, generally applicable protocols are lacking, and suboptimal approaches are often used for nonstandard but epidemiologically important species. Photobacterium damselae subsp. piscicida (Pdp) is a primary pathogen of fish in aquaculture and has been considered hard to transform since the mid‐1990s. Consequently, conjugative transfer of RK2/RP4 suicide vectors from Escherichia coli S17‐1/SM10 donor strains, a system prone to off‐target mutagenesis, was used to deliver the allelic exchange DNA in previous studies. Here we have achieved efficient electrotransformation in Pdp using a salt‐free highly concentrated sucrose solution, which performs as a hypertonic wash buffer, cryoprotectant, and electroporation buffer. High‐efficiency transformation has enabled vector‐free mutagenesis for which we have employed circular minimalistic constructs (knockout minicircles) containing only allelic exchange essentials that were generated by Gibson assembly. Preparation of competent cells using sucrose and electroporation/integration of minicircles had virtually no detectable off‐target promutagenic effect. In contrast, a downstream sacB selection apparently induced several large deletions via mobilization of transposable elements. Electroporation of minicircles into sucrose‐treated cells is a versatile broadly applicable approach that may facilitate allelic exchange in a wide range of microbial species. The method permitted inactivation of a primary virulence factor unique to Pdp, apoptogenic toxin AIP56, demonstrating the efficacy of minicircles for difficult KO targets located on the high copy number of small plasmids.

including a secreted toxin AIP56 that cleaves nuclear factor-κB transcription factor, stopping the upregulation of the inflammatory genes and the downregulation of antiapoptotic genes in macrophages and neutrophils (Nunez-Diaz et al., 2018;Pereira et al., 2014;Valderrama et al., 2019;do Vale et al., 2005). This critical virulence factor lacks homologs among known proteins, and the Pdp genome encodes a plethora of other not-yet-characterized in silico predicted proteins whose function is completely unknown (Baseggio et al., 2021). Gene inactivation is a robust way to determine or confirm a protein function (Nakashima & Miyazaki, 2014;Xu & Zhang, 2016), which, in pathogenic bacteria, can promote development of disease control measures such as protein subunit vaccines or nutritional therapy (Lu et al., 2021). Consequently, established or putative virulence factors are common gene knockout (KO) targets.
The accuracy of the protein characterization hinges on the isogenicity of the KO mutant, because the phenotype may be affected by undesired (secondary, off-target) mutations elsewhere in the genome (Johnson et al., 2003). Yet, reliable mutagenesis protocols are scarce for nonmodel organisms, and approaches demonstrated to have high rate of off-target mutations are widely used (Babic et al., 2008;Ferrières et al., 2010;Strand et al., 2014).
Targeted gene KO in a wide range of bacteria can be achieved via allelic exchange by homologous recombination (Nakashima & Miyazaki, 2014;Xu & Zhang, 2016). In contrast to Red/ET recombination and CRISPR-Cas9 methods that require introduction of plasmids for expression of gene-editing enzymes (Nakashima & Miyazaki, 2014;Xu & Zhang, 2016), this approach relies on endogenous enzymatic machinery conserved across bacterial taxa (Michel & Leach, 2012). To generate the allelic exchange construct (AEC), the desired mutation (deletion, insertion, point mutation) is spliced between 0.5 and 1.5 kb regions homologous to sequences flanking the targeted locus (Nakashima & Miyazaki, 2014;Xu & Zhang, 2016). Next, the AEC is conventionally cloned into a plasmid vector containing marker genes for mutant selection, which is nonreplicating (suicide vector) or conditionally replicating in the recipient bacterial strain. Gibson assembly (GA) appears to be the most rapid and efficient approach to date, allowing simultaneous generation of the AEC and cloning into a plasmid vector in a single reaction (Huang & Wilks, 2017;Rudenko & Barnes, 2018). The next step is delivery of the allelic exchange plasmid, which ideally should be conducted via transformation of chemically competent or electrocompetent cells (Drury, 1994;Wirth et al., 1989). However, universal techniques to render cells competent are lacking and require optimization for many bacterial species and strains (Monk, 2012;Wang et al., 2020;Yildirim et al., 2016). Among epidemiologically important taxa, Pdp has been evidently poorly transformable since multiple conventional protocols were tested and shown inefficient (Cutrín et al., 1995(Cutrín et al., , 2000. When a high-efficiency transformation method is not available, an allelic exchange vector may be delivered by conjugation (Ferrières et al., 2010;Strand et al., 2014). To date, conjugative transfer of pir/RP4 suicide vectors from SM10/S17-1 λ pir donor strains of Escherichia coli was the only method used to deliver the allelic exchange DNA in Pdp KO mutagenesis (Abushattal et al., 2022;Naka et al. 2005Naka et al. , 2007Osorio et al., 2015). These donors contain the chromosomally integrated RP4::Mu conjugative transfer cluster (Simon et al., 1983), which was repeatedly shown to introduce off-target mutations via Mu phage integration into the allelic-exchange vector and/or the recipient's DNA, as well as Mu phage-mediated and RP4-mediated transfer and illegitimate integration of E. coli genes into the target host (Ferrières et al., 2010;Strand et al., 2014). Although alternative vectors and donors were designed (Babic et al., 2008;Ferrières et al., 2010;Strand et al., 2014), conjugation as a biological process has a mutagenic effect as the entrance of single-strand DNA (ssDNA) via conjugation activates the stress response SOS that involves errorprone replication and increased translocation/recombination of mobile elements (Baharoglu et al., 2010;Virolle et al., 2020). Further, whether DNA is delivered via conjugation or transformation, suicide vectors often integrate via illegitimate recombination (Johnson et al., 2003). In some cases, frequency of the off-target mutations generated by suicide vectors can be sufficient for the latter to be used in random mutagenesis (Desomer et al., 1991).
In the present study, we have achieved high-efficiency transformation in Pdp using a very simple protocol that utilizes only nonionic osmotic pressure (sucrose) to render the cells competent. Originally developed for streptococci (Framson et al., 1997), this salt-free approach also performs well in Vibrio (Wang & Griffiths, 2009), and may potentially be a method of choice for a wide range of bacteria.
High-efficiency transformation has allowed us to mutagenize Pdp with nonreplicating DNA without conjugation, or even propagation of mutagenic DNA in E. coli. To further decrease the probability of secondary mutations, instead of conventional cloning of an AEC into a suicide vector, we have employed a vector-free "minicircle" approach where homology arms were spliced with marker genes and circularized by GA. Minicircles were constructed and successfully integrated (single-crossover mutants obtained) for two KO targets, a ubiquitous ssrA/smpB ribosome rescue system (Karzai et al., 2000) and AIP56 toxin, which is unique to Pdp (do Vale et al., 2005). In the latter case, double-crossover mutants were also obtained with ease despite the expectation that aip56 may be very hard to knock out due to its location on high copy number of small plasmids (Freitas et al., 2022). Our donor-and vector-free allelic exchange based on transformation of sucrose-treated competent cells with KO minicircles is a major improvement over previously used mutagenesis techniques in Pdp. The approach offers lower probability of secondary mutations and is potentially applicable to a broad range of bacterial species, including allegedly hard-to-transform species. Australia, identified and deposited at the Department of Primary Industries and Regional Development (DPIRD), Western Australia as Pdp AS-16-0540-1 and Pdp AS-16-0555-7, and subsequently deposited as QMA0505 and QMA0506 at The University of Queensland (UQ), where isolates were sequenced (Baseggio et al., 2021). Bacteria were cultured at 25°C on Tryptic soy agar (TSA) or in Tryptic soy broth (TSB), unless otherwise specified.

| Preparation of electrocompetent cells
Initially, we tested best-performing conventional protocol in Pdp according to benchmarking performed by Cutrin et al. (1995) as described, except using TSB to grow cells instead of Brain heart infusion broth. Subsequently, cells were rendered competent using a salt-free protocol developed for preparation of electrocompetent streptococci (Framson et al., 1997) with minor modifications. The overnight TSB cultures were diluted to 10 3 colony-forming unit (CFU)/mL (OD 600 = 0.025 in Eppendorf BioPhotometer 6131) in 150-300 mL TSB, and grown to early exponential phase (10 4 -10 5 CFU/mL; OD 600 = 0.25-0.4). Harvested cultures were transferred to 50 mL tubes and chilled on ice; pelleted for 5 min at 2000g, 4°C; resuspended in 50-100 mL of ice-cold 0.625 M sucrose solution in water (pellets combined) and pelleted for 10 min at 18 500g, 4°C; resuspended in 10 mL of ice-cold 0.625 M sucrose (pellets combined), transferred into 15 mL tube, pelleted for 15 min at 18 500g, 4°C, and resuspended in 0.5-1 mL of the same buffer (or remaining supernatant if pellet was loose). This preparation was also attempted at room temperature using TSB supplemented with 1% salt to grow the culture. Competent cells were aliquoted by 50-90 μL into chilled 1.5 mL tubes and stored at -80°C.

| Transformation efficiency estimation
Competent cells were defrosted on ice, mixed with 100 ng of a plasmid carrying antibiotic resistance marker, pET-28a(+), pLZ12spec, or pUC19, in 10 μL H 2 O, transferred to chilled 0.1 cm gap electro cuvettes (Sigma), and electroporated at the default P1 in Eppendorf Eporator (1250 V, 5 ms, 600 Ω). Cells were recovered in 0.5-1 mL of 0.25 M sucrose TSB for 2.5 h, 50 μL aliquots were spread on 75 μg/mL kanamycin (Km), 200 μg/mL spectinomycin, or 1 μg/mL ampicillin plates, and transformants were counted after 48 h. Viable cell counts were performed by Miles and Misra method (Miles et al., 1938) immediately after preparation, after defrosting, after electroporation, and after the recovery period. sacB gene with promoter and terminator (counterselection marker conferring sucrose sensitivity) were amplified using Km_F/R and sacB_F/R primer sets (Tables 1 and 2), from 0.5 ng of pET-28a(+) (Addgene; #2526) and pRE107 (Addgene; #43829) plasmids, respectively. The pET-28a reaction was treated postamplification with DpnI (NEB) to eliminate transformation background produced by pET-28a(+) plasmid used as a PCR template. The latter was extracted immediately before amplification to ensure that methylation required for DpnI restriction was retained. Homology regions ssDNA was also attempted. For this, half of the purified assembly reactions (10 μL) were denatured at 95°C for 2 min and snap cooled.

|
Single-crossover mutants (meroploids containing integrated minicircles) were selected on 75 μg/mL Km and confirmed by colony PCR screening for the presence of recombination (crossover) in upstream or/and downstream homology regions-amplification across the homology regions where one primer binds to kanR2 sequence and the second primer binds to a sequence outside the homology (out_aip56_F/Km_R, Km_F/out_aip56_R, out_ssrA_F/Km_R, and Km_F/out_ssrA_R reactions; Tables 1 and 2).

| Selection of aip56-and ssrA-smpB KO mutants
To select double-crossover clones (KO strains) of aip56, singlecrossover mutants were grown to saturation in nonsupplemented TSB, subcultured once, and plated onto agar containing 75 μg/mL Km lethal to wild type (WT) and 15% w/v sucrose (lethal to meroploids). Emergent clones were picked, streaked on the fresh plates, subcultured one more time, and screened for the loss of target genes by colony PCR using an aip56-specific primer set (aip56_F/R; Table 1) (Abushattal et al., 2020). aip56 deletion was further confirmed by amplification across the allelic exchange region yielding a shorter product compared to the WT strain (out_aip56_F/R reaction; Table 1). Single-crossover mutants of ssrA-smpB were grown in multiple liquid culture conditions: TSB, TSB supplemented with 7.5% or 15% sucrose, TSB supplemented with 7.5% or 15% sucrose and 75 μg/mL Km, with decreased agitation or stationary. Most attempts included multiple cultures ranging from 10 to 96. These were plated on 75 μg/mL Km/15% sucrose plates or 50 μg/mL Km/7.5% sucrose plates, and emergent colonies were screened using ssrA-and smpB-specific primer sets (ssrA_F/R, smpB_F/R; Table 2).
2.4.5 | Simultaneous selection of aip56-KO clones and "revertant" clones from multiple independent cultures Multiple aip56 KO clones and clones that reverted to WT genotype ("revertants") were selected in parallel from independent broth T A B L E 1 Primers used in aip56 mutagenesis.  et al., 2021). For the purpose of obtaining isogenic KO mutants, we wanted to avoid mating with SM10/S17-1 λ pir donor strains of E. coli due to high off-target mutation rate produced by RP4::Mu conjugative transfer cluster (Ferrières et al., 2010;Strand et al., 2014). Transformation is a more straightforward and rapid approach compared to conjugation since allelic exchange DNA is delivered directly to the target organism. allows storage (Wirth et al., 1989). However, many bacterial species and strains are not readily transformable by standard methods (Aune & Aachmann, 2010), and high transformation efficiency is required for the delivery of nonreplicating DNA in allelic exchange mutagenesis. Case in point, previous attempts to render Pdp cells electrocompetent using the outlined basic approach were not particularly successful. Multiple conditions were evaluated by Cutrin et al. in 1995, and according to the authors, excessive cell lysis occurred when cells were grown in a salty medium (standard for marine species) and prepared using low ionic strength buffers, and arcing occurred when cells were prepared using "isotonic for the bacteria" (i.e., high ionic strength) buffers (Cutrín et al., 2000). Of all the evaluated conventional methods, the highest efficiency of 9.8 × 10 2 CFU/μg DNA was achieved with a laborious multiple-buffer multiple-reagent protocol immediately after preparation without defrosting/prior storage (Cutrín, 1995). We attempted the same procedure in our laboratory and obtained around 10 2 CFU/μg transformation efficiencies with pET-28a(+) and pUC19 plasmids.
Since higher efficiencies are required for transformation with nonreplicating DNA, we attempted a rapid and straightforward single-buffer protocol originally developed by Dunny et al. (1991) for gram-positive bacteria. The method was subsequently applied by Framson et al. (1997) for group B streptococci and optimized to contain no salt in the buffer, which rendered this single-buffer protocol to be also a single reagent. The sole ingredient in the wash/ electroporation buffer is 0.625 M sucrose, which works as a hypertonic stress agent (nonionic osmotic pressure), electroporation medium, and cryoprotectant. Following the above protocol, we have repeatedly achieved transformation efficiencies of 10 6 -10 7 CFU/μg DNA in Pdp using pET-28a(+) and pUC19 vectors, and somewhat lower efficiencies of 10 5 -10 6 CFU/μg DNA were obtained with streptococcal pLZ12spec vector. Notably, these efficiencies were estimated with defrosted cells previously stored at −80°C. We  We have also tried room-temperature preparation and transformation, which may be used as a more convenient alternative to traditional ice-cold technique (Tu et al., 2016). This approach was functional but 10-100 times less efficient.
We have efficiently transformed sucrose-treated electrocom- 3.2 | "KO minicircles": circular AECs generated by GA Nonreplicating plasmids (suicide vectors) are most often used to deliver mutagenic DNA constructs in allelic exchange mutagenesis.
Yet, suicide vectors are notorious for illegitimate integration into host chromosomes and plasmids and other kinds of off-target mutagenesis (Johnson et al., 2003). Illegitimate recombination is very common in bacteria, and it is facilitated in plasmid sequences by the abundance of repeats and common/universal motifs (Desomer et al., 1991;Li et al., 2018;Oliveira et al., 2010;de Vries & Wackernagel, 2002). Remarkably, some suicide plasmids illegitimately integrate at frequencies sufficient for random mutagenesis experiments (Desomer et al., 1991). In some circumstances, plasmid backbones may be necessary, such as when transformation is not available, so DNA is delivered by conjugation, or when transformation is low in efficiency, so very large quantities of DNA are required the plasmid is replicated in and extracted from E. coli for electroporation. However, whenever high-efficiency transformation and other factors permit, vector-free KO approach should be considered as an alternative offering lower off-target mutagenesis probability.
Vector-free AECs in previous studies were Fusion PCR products circularized by ligation (Gomaa et al., 2017) and linear GA products (Wu et al., 2019). Here we combined these two approaches: we used GA as a more convenient alternative to Fusion PCR technique (Huang & Wilks, 2017;Rudenko & Barnes, 2018), but employed it to generate circular constructs/products, as they are more stable compared to linear DNA. The constructs, here dubbed "KO minicircles," were generated to knock out aip56 and ssrA-smpB ribosomal rescue loci (Figures 1 and 2). Minicircles were designed for marked nonpolar deletions and contained four fragments assembled in the following order: -> upstream homology arm -> selectable marker gene (kanR2, Km resistance) -> downstream homology arm -> counter-selectable marker transcript (sacB, sucrose sensitivity) ->.
In-frame insertion of the selectable marker gene (without transcription elements) in the place of the KO target does not cause polar effects, facilitates mutant selection, and helps to maintain purity of the culture. Alternatively, if unmarked deletion is desired, a selectable marker gene (with transcription elements) can be placed outside the homology regions along with a counter-selectable marker. Overall, the minicircle approach is exceptionally versatile and may be used with any selectable and counter-selectable marker/s, and for other genetic modifications including site-directed mutagenesis and gene knock-in, e.g., insertion of the WT gene for the KO rescue.
T A B L E 3 Selection of aip56 and ssrA KO mutants after transformations with ds and denatured (ss) minicircles.   (Table 3). Randomly chosen selected colonies were PCRscreened for the presence of crossover using primer sets where one primer binds to the Km sequence and the second one binds to the genomic sequence outside the homology region. Over 70% of PCRscreened colonies yielded products of expected size ( Figure 3 and Table 3), which unambiguously showed successful integration of minicircles. In the case of aip56, both homology arms recombined equally well (Figure 3a,b), while, in the case of ssrA, the upstream region was apparently less recombinable (Figure 3d,c).

| aip56 and ssrA-smpB mutagenesis: Selection of double-crossover mutants
To select double-crossover mutants, meroploids obtained from dsDNA-minicircle transformations were used, as more likely to be isogenic (Baharoglu et al., 2010;Virolle et al., 2020). Multiple aip56 KO clones were obtained with ease after two subcultures of singlecrossover mutants in nonselective broth (TSB) and plating on agar supplemented with sucrose (lethal to single crossovers) and Km (lethal to WT). Randomly chosen clones were PCR-screened for the loss of aip56 gene (Table 3). Several aip56-negative clones were further confirmed to be KO mutants by amplification across the allelic exchange region yielding a shorter product compared to WT strain ( Figure 4). PCR screening for aip56 deletion showed that many meroploids were escaping sucrose counterselection, which is often weak and inefficient (Cianfanelli et al., 2020;Lazarus et al., 2019), despite omission of salt supplementation in selective plates. Since salt inhibits sacB expression (Blomfield et al., 1991), the latter marker may be suboptimal for slight halophiles such as Pdp and completely inappropriate for obligate halophiles. Nonetheless, double-crossover mutants were selected from 75% of the broth cultures and accounted for 16.65%-83% of the screened colonies (Table 3). Such a success rate was unexpected since aip56 is encoded on a highly abundant, small (<10 kB) plasmid that was anticipated to be hard to eliminate (Freitas et al., 2022). Potentially, aip56 KO would have been harder, perhaps impossible, to achieve using a suicide vector, since integration of a long vector backbone sequence with plasmid replication elements is likely to compromise the aip56 plasmid stability and/or replication . AIP56 is a known toxin characterized at the protein level, which is a primary virulence factor of Pdp causing apoptosis in professional phagocytic cells (Pereira et al., 2014;do Vale et al., 2005). However, isogenic KO mutants lacking the gene have not been characterized yet, which hampers the research aiming at a complete understanding of Pdp pathogenicity (Freitas et al., 2022).
Although both ssrA-smpB homology arms were efficiently recombining ( Figure 3 and F I G U R E 5 Colony PCR screening of clones selected from (kanamycin-resistant/sucrose-sensitive) aip56-minicircle meroploid cultures on sucrose agar. Amplicons in the corresponding top and bottom parts of the gel are derived from the same clone; the top part of the gel contains aip56-specific products (aip56_F/R) absent in knockout clones and the bottom part contains kanR2-specific products absent in clones that reverted to WT genotype. The marker in the first lane is Fast DNA Ladder (NEB) with 10, 5, 3, 2, 1.5, 1 (reference), 0.766, 0.5, 0.3, 0.15, and 0.05 kb bands. may be essential in Pdp. Indeed, ssrA-smpB is essential in obligatory intracellular pathogens (Karzai et al., 2000), and Pdp causes primarily intracellular infections and shows relatively slow growth on laboratory culture media.
3.5 | Further aip56 mutagenesis and detection of secondary mutations via whole-genome sequencing 3.5.1 | Mutagenesis clones used for sequencing We aimed to confirm that electroporation of KO minicircles into sucrose-treated competent cells can indeed generate highly isogenic KO mutants. However, sequencing of a single KO clone is not sufficient for this purpose as secondary mutations can occur during culturing either completely at random or can be associated with a particular selective pressures or/and inactivation of a particular gene (a target mutation). Thus, we obtained more aip56 KO mutants from several independent broth cultures of a single meroploid clone (clone used to select the initial KO strain). Also, from the same individual broth cultures we have obtained clones that regained the WT genotype ("revertants"), which originate in single-crossover cultures in the same manner as KO clones except for the second crossover happening on the same rather than the opposite homology arm. Since our meroploid contained sacB counterselectable marker conferring sucrose sensitivity, it was possible to select both kinds of clones in parallel on the sucrose agar, which was followed by plating on Km agar to differentiate between revertant/WT and KO genotypes (kanamycin-sensitive and kanamycin-resistant phenotype respectively). Plenty of clones were no longer sensitive to sucrose in three out of four nonselective meroploid broth cultures, and 30%-53% of the clones selected on sucrose were resistant to Km ( three independent cultures, the initial aip56 KO clone selected on sucrose/kanamycin agar, and four meroploid clones selected on Km: two from ds aip56-minicircle transformation and two from ss aip56minicircle transformation (Table 5).  (Tables 5 and 6). Small mutations appeared to be due to random replication errors, unlikely to have functional significance, and were mostly copy-number variations within repetitive sequence regions. In contrast, several de novo large deletions were found in KO and revertant clones selected on sucrose plates (Tables 5 and 7). They were evident as areas of no read coverage such as in the case of intended aip56 deletion (Figure 6), and were subsequently confirmed via long-read Nanopore sequencing (SRR253160-66; JAUY(VA-D, UZ)000000000). Five out of six sucrose-selected clones contained one large deletion, which was unique and not linked to broth culture history, that is, different variants present in KO and revertant clones selected in parallel from the same culture (Tables 5 and 7). These five deletions were flanked by transposable elements (TE), which are recognized as primary mediators of evolution in Pdp. Indeed, TE movement is promoted by stress (Twiss et al., 2005) and 15% sucrose in the medium creates high-osmolarity environment stressful to bacteria (Cesar et al., 2020).

| Secondary mutations
On the other hand, the KO clone originally selected on Km/sucrose agar did not have any large deletion, which may be incidental or T A B L E 6 Small secondary mutations identified in aip56 mutagenesis clones.
Pdp is not readily transformable by standard techniques that involve competent cell preparations using hypertonic saline buffers (Cutrín et al., 1995). As a result, delivery of the mutagenic DNA via conjugation of pir/RP4 suicide vectors from SM10/S17-1 λ pir donor strains of E. coli has been used in previous genetic modifications of Pdp, a method notorious for the high rate of the off-target mutation (Babic et al., 2008;Ferrières et al., 2010;Strand et al., 2014). Here we have achieved high transformation efficiency comparable to commercial E. coli preparations via a simple single-buffer single-ingredient protocol. This method avoids salt and uses concentrated sucrose solution as an all-in-one hypertonic wash buffer, cryoprotecting medium, and electroporation buffer. Nonionic osmotic pressure stabilizes membranes for electroporation and, unlike ions from salts, does not interfere with electric discharge and does not contribute to DNase activity (Wang & Griffiths, 2009). Potentially, this approach may be generally efficient in bacteria as it was successfully used in other Vibrionaceae (Wang & Griffiths, 2009) and phylogenetically distant Streptococcaceae (Framson et al., 1997).
High-efficiency transformation is required for allelic exchange mutagenesis using nonreplicating DNA. The latter is conventionally an allelic exchange construct (AEC) cloned in a suicide vector.
F I G U R E 6 No read coverage area in reference-guided assemblies of the knockout clones confirming aip56 deletion. CDS, protein coding sequence.
F I G U R E 7 Schematic representation of aip56 knockout. GA, Gibson assembly.
However, plasmid backbones increase the chances of secondary mutations (Desomer et al., 1991;Li et al., 2018;Oliveira et al., 2010;de Vries & Wackernagel, 2002). Here we have mutated Pdp via vector-free allelic exchange employing KO minicircles, nonreplicating circular minimalistic AEC containing only mutagenesis essentials that were generated by GA (Gibson et al., 2009). Electroporated KO minicircles targeting aip56 and ssrA-smpB loci readily integrated into plasmid and chromosomal DNA, respectively, evident by highly efficient selection of single-crossover mutants. Furthermore, KO mutants of the aip56 gene encoded on highly abundant small (<10 kb) plasmid were selected with ease. This was potentially facilitated by the minicircle approach, as integration of a longer sequence containing a suicide vector backbone is likely to affect host plasmid stability and/or replication. We summarize the employed method using aip56 KO as an example below, which is schematically represented in Figure 7: aip56-KO minicircle of 4.7 kb (0.9 kb sequence upstream from aip56 -> selectable marker gene (kanR2, Km resistance, 0.8 kb) -> 0.9 kb sequence downstream from aip56 -> counterselectable marker transcript (sacB, sucrose sensitivity, 2.1 kb) was generated by GA ( Figure 1) and electroporated into Pdp cells rendered competent using highly concentrated sucrose solution (as described in subsection 3.1). Single-crossover mutants were selected by the gain of Km resistance, and confirmed by PCR from Km gene to genomic sequence outside the homology arms ( Figure 3a,b), along with the gain of sucrose sensitivity. Doublecrossover mutants were selected by loss of sucrose sensitivity, then confirmed by the lack of aip56-specific PCR amplification and by PCR across allelic exchange locus yielding shorter product than in the WT parent, accounting for the size difference of aip56 and kanR2 ( Figure 4).
Whole-genome sequencing of multiple clones generated during aip56 mutagenesis did not reveal any secondary mutations that could be attributed to sucrose treatment used to render the cells competent or electroporation/integration of minicircles (Tables 5-7). However, it identified several large deletions associated with the mobilization of TEs by prolonged exposure to high sucrose concentrations used for sacB counterselection (Tables 6 and 7). In this respect, our vector-free method is highly flexible as minicircles can be designed/assembled to contain alternative selection markers or/and to generate other kinds of mutations. To conclude, electroporation of minicircles into sucrose-treated cells is an efficient and versatile approach for allelic exchange allowing to generate highly isogenic mutants in Pdp and potentially most of the bacterial species.