A Plasmid-Transposon Hybrid Mutagenesis System Effective in a Broad Range of Enterobacteria

Random transposon mutagenesis is a powerful technique used to generate libraries of genetic insertions in many different bacterial strains. Here we develop a system facilitating random transposon mutagenesis in a range of different Gram-negative bacterial strains, including Pectobacterium atrosepticum, Citrobacter rodentium, Serratia sp. ATCC39006, Serratia plymuthica, Dickeya dadantii, and many more. Transposon mutagenesis was optimized in each of these strains and three studies are presented to show the efficacy of this system. Firstly, the important agricultural pathogen D. dadantii was mutagenized. Two mutants that showed reduced protease production and one mutant producing the previously cryptic pigment, indigoidine, were identified and characterized. Secondly, the enterobacterium, Serratia sp. ATCC39006 was mutagenized and mutants incapable of producing gas vesicles, proteinaceous intracellular organelles, were identified. One of these contained a β-galactosidase transcriptional fusion within the gene gvpA1, essential for gas vesicle production. Finally, the system was used to mutate the biosynthetic gene clusters of the antifungal, anti-oomycete and anticancer polyketide, oocydin A, in the plant-associated enterobacterium, Dickeya solani MK10. The mutagenesis system was developed to allow easy identification of transposon insertion sites by sequencing, after facile generation of a replicon encompassing the transposon and adjacent DNA, post-excision. Furthermore, the system can also create transcriptional fusions with either β-galactosidase or β-glucuronidase as reporters, and exploits a variety of drug resistance markers so that multiple selectable fusions can be generated in a single strain. This system of various transposons has wide utility and can be combined in many different ways.


INTRODUCTION
Since their initial discovery, transposable elements have greatly assisted our understanding of eukaryotic and prokaryotic genetics. Though transposable elements were first identified and studied in maize, they have been found in virtually all organisms, such as Drosophila melanogaster, Escherichia coli, and Homo sapiens (McClintock, 1956;Craig, 1990;Frost et al., 2005;Cordaux and Batzer, 2009). Transposons in bacteria can also move from the chromosome to a plasmid or bacteriophage genome, driving genetic evolution and contributing to the spread of antibiotic resistance gene clusters between different bacteria (Jimenez and Davies, 1980;Berg and Berg, 1983).
Advances in molecular genetics and gene cloning have allowed development of genetic tools derived from mobile elements. For example, engineered mobile elements (transposons) have been used in genetic analysis to generate random insertions within the chromosome of a target organism (Berg and Berg, 1983) and insertion of a transposon near, or within, a gene can alter or destroy its function.
Transposons can jump between any genetic elements in their bacterial hosts: chromosome, plasmids or phage genomes (Berg and Berg, 1983). The insertion or excision of a transposon is facilitated by a transposase protein, which is usually encoded within the transposon itself. However, when expression of the transposase protein is decoupled from the transposon, a system can be engineered to generate immobile single transposon insertions in target DNA of interest.
Here we describe the creation of a type of transposon system that has a broad host range within the Enterobacteria that complements existing systems (Dennis and Zylstra, 1998;Larsen et al., 2002). This plasmid-transposon (or plasposon) hybrid system, has been engineered so that it carries an origin of replication, and so, can form a replicating plasmid if excised and self-ligated. Such a plasmid can be sequenced and the transposon insertion site identified with ease. We also used a second method to identify the transposon insertion sites, which relied upon Random Primed PCR (RP-PCR; Jacobs et al., 2003;Fineran et al., 2005). These two parallel methods facilitate quick and straightforward identification of any transposon insertion within a host's genome.
Three test studies were conducted to demonstrate the efficacy of this system. Firstly, the plasposon system described above was used to mutagenise the plant pathogen Dickeya dadantii, a cause of soft rot in many different plant species and a pathogen of the aphid Acyrthosiphon pisum (Grenier et al., 2006). Secondly, the system was used to create stable transcriptional fusions within gvpA1, the first gene in the gas vesicle operon of Serratia sp. 39006 (S39006). Thirdly, transposon mutants interrupting the oocydin A biosynthetic gene clusters in Dickeya solani MK10 were isolated-one of them generating a transcriptional fusion within the large oocJ-W operon. Together, these three studies demonstrate the wide utility of this system for genetic screens in a broad range of bacterial species.

DNA Manipulations, Oligonucleotides, and Sequencing
Unless otherwise stated, standard molecular biological methods were used for all DNA manipulations. Plasmid DNA was extracted using a Qiagen Miniprep Kit (Qiagen) or an Anachem Keyprep Kit (Anachem) according to the manufacturer's instructions. Where required, DNA was extracted from individual strains using a Qiagen DNeasy Kit according to manufacturer's instructions. All restriction enzymes used were obtained from New England Biolabs and used according to manufacturer's protocols. DNA fragments were ligated using T4 DNA ligase (NEB). Oligonucleotides were obtained from Sigma Aldrich and are listed in Table 2. DNA sequencing was conducted in the Department of Biochemistry Sequencing Facility, University of Cambridge, Cambridge, United Kingdom.

Cloning of pDS1028 and pKRCPN1
The plasmid pBM1001 was created by insertion of the XmnI/BstBI fragment from pACYC184 (containing the aph gene) into pBluescript II KS+. The plasmid pBM1002 was created by insertion of the StyI/XbaI fragment from pACYC184 (containing the tetA gene) into pBluescript II KS+. The plasmid pDS1028 was created in three steps. Firstly, the XbaI/EcoRI fragment containing the tetracycline resistance gene (tetA) from pBM1002 was ligated into pRL27 to create plasmid pDS1017. Secondly, pDS1017 was digested with SacI and pBM1001 was digested with AccI to remove the cat gene. Both fragments of DNA were treated to create blunt ends and were ligated together to create pDS1022. Finally, oligonucleotides 5 ′ PAC and 3 ′ ERV were used to amplify the transcriptional and translational terminator from pHP45 , then this fragment was digested with EcoRV/PacI, and ligated into pDS1022 that had been compatibly digested, to create plasmid pDS1028.
The plasmid pKRCPN1 was created by modifying pDS1028. The promoterless lacZ1 fragment from miniTn5lacZ1 was ligated into the KpnI site of pDS1028, to create pNRW112. Oligonucleotide pair KR19/KR21 were used to amplify the aph gene from pACYC177. This was digested with BstBI and PacI and ligated into compatible sites of digested pNRW112 to create pKRCPN1. Escherichia coli CC118λpir (ara-leu) araD lacX74 galE galK phoA20 thi-1 rpsE rpoB argE(Am) recA1 λpir Herrero et al., 1990 Escherichia coli DH5α

Transposon Mutagenesis
Overnight cultures of each recipient strain and the E. coli donor strain were grown at 30 • C and 37 • C, respectively. The donor and recipient strains were mixed together in the ratio indicated. The most efficient ratio for transposon mutagenesis for most strains was 1:3 (donor: recipient). Thirty microlitres of this mixture was spotted onto LBA + DAPA where required, and allowed to dry. The mixture was incubated overnight at 30 • C and the mixed culture then resuspended in 1 ml of sterile LB. The conjugation mixture was serially diluted and plated onto minimal agar [0.2% glucose, 0.41 mM MgSO 4 , 0.1% (NH 4 ) 2 SO 4 , 0.7% K 2 HPO 4 , 0.2% KH 2 PO 4, 1.5% agar], caseinase agar (nutrient broth agar with 1% Marvel skim milk, 1.5% agar), or LBA, with appropriate antibiotic selection, and incubated for 48 h at 25 • C (D. dadantii and D. solani MK10) or 30 • C (S39006, S. plymuthica A153). Putative mutants were then tested for acquisition of the full plasmid by patching a selection of colonies onto LBA + Tc.

Identification of Transposon Insertion Sites by Replicon Cloning
DNA was extracted from transposon mutants and digested with an enzyme that does not cut within the transposon (See Table 3 for a full list). Digested DNA was purified using an Anachem Spin PCR Clean Up Kit then self-ligated. The ligation mixture was then used to transform E. coli strains CC118λpir or β2163 by heat shock and plated onto LBA (containing DAPA for β2163) and the appropriate antibiotic. Replicon DNA was subsequently isolated by plasmid extraction and the precise transposon insertion site identified by sequencing using either oREM7 or oREM8 primers.

Identification of Transposon Insertion Sites by RP-PCR
RP-PCR was conducted largely as previously described (Jacobs et al., 2003;Fineran et al., 2005). Briefly, DNA from transposon mutants was amplified using a two-step PCR process. In the first round, DNA was amplified using a random oligonucleotide mix (oPF106, oPF107, oPF108) and the transposon specific oligonucleotide (See Table 2 for examples). DNA from this reaction was used in a second amplification using oPF109 and a second transposon specific oligonucleotide. The resulting DNA fragments were amplified with the appropriate transposon specific oligonucleotide.

Indigoidine Liquid Assay
Production of indigoidine was assessed in liquid as described previously (Reverchon et al., 2002). Briefly, cells were grown in PMB, LB, or MM9, pelleted by centrifugation, supernatant samples removed and the cell pellets snap frozen in liquid nitrogen. Cells were thawed on ice, the pellet resuspended in 1 ml dimethyl sulfoxide (Sigma) and vortexed. Cellular debris was pelleted by centrifugation (10 000 g, 10 min) and the A 615 of the supernatant measured. Indigoidine levels were expressed as the A 615 OD 600 −1 . Where appropriate, a student's t-test was used to determine statistical significance of differences between mutants.

Flotation Assay
Flotation assays of S39006 were carried out as described previously (Ramsay et al., 2011). Briefly, strains were grown in a 30 ml sealed universal plastic tube overnight at 30 • C in LB.
The following day, a normalized number of cells were used to inoculate a 5 ml culture which was grown for 24 h on a roller wheel at 30 • C. Cultures were left to settle at room temperature for 24 h.

Microscopy
Samples were taken directly from liquid culture without further preparation, largely as described previously (Ramsay et al., 2011). Phase contrast light microscopy was undertaken using an Olympus BX-51 microscope with a 100x oil immersion lens. Images were captured using a QICAM camera and QCapture Pro software. Images were cropped and the scale bar added using Adobe Photoshop. All images were representative of those observed for any particular strain.
β-galactosidase Assay β-galactosidase activity was determined by monitoring the breakdown of 4 ′ -Methylumbelliferyl-β-D-galactopyranoside (Melford Laboratories). At the indicated time point, samples of liquid culture (100 µl) were taken and frozen at −80 • C until needed. Samples were thawed and 10 µl removed and frozen at −80 • C for 15 min and thawed at 37 • C. Next, 100 µl of Reaction Mix (Phosphate-buffered saline, 400 µg ml −1 lysozyme, 250 µg ml −1 4 ′ -Methylumbelliferyl-β-D-galactopyranoside) was added to the samples and they were monitored in a Gemini XPS plate reader using the following parameters: 360 nm excitation, 450 nm emission, 435 nm cut off, eight reads per well, measured every minute for 30 min. Relative fluorescence units (RFU) per min were calculated during a linear phase of fluorescence increase and were normalized to the OD 600 creating an activity measurement of RFU OD 600 −1 .

Sequence Information
The sequences of pKRCPN1 and pDS1028 were deposited in Genbank and given the accession numbers KT991288 and KT991389, respectively.

RESULTS
The Plasmids pDS1028 and pKRCPN1 are Capable of Mutagenizing Some Gram-Negative Bacteria The pDS1028 and pKRCPN1 plasmids were created as described in Materials and Methods. Both plasmids were sequenced and a schematic of them is shown in Figure 1. Though both pDS1028 and pKRCPN1 contain the transposase gene to facilitate transposition, it is not located within the transposon itself. Thus, when the transposon has "hopped" into a chromosomal location, if the plasmid is unable to replicate in this host (all pir − strains) a stable insertion will be created. These two systems were then used to isolate gene knockouts or transcriptional gene fusions. Furthermore, as the two transposons contain different antibiotic selections, they can also be used in combination. In addition, a version of this transposon, described by Ramsay and colleagues has been engineered with a uidA cassette in place of the lacZ cassette described in this work (Ramsay et al., 2011). The utility of pDS1028 and pKRCPN1 was assessed in 20 different bacterial host strains. Different ratios of E. coli donor and recipient cells were tested and putative transposon insertions identified. A tetracycline resistance cassette was encoded in the plasmid backbone of both pDS1028 and pKRCPN1 but not within the transposon. Therefore, in the case of pDS1028, if the full plasmid transferred between the donor and recipient strain, any resulting mutants would be resistant to both chloramphenicol and tetracyline, and any transposon insertions would be resistant to chloramphenicol but sensitive to tetracycline. Of the 20 strains tested using both plasposon systems, putative transposon insertions were observed in 16 strains ( Table 4). For each strain where putative transposon insertions were observed, 50 colonies were tested for tetracycline sensitivity. No tetracycline resistant colonies were observed for any strain with successful transposon insertions; therefore we concluded that our mutants were transposon insertions and not the result of plasmid transfer.
We were unable to detect putative transposon insertions in Serratia marcescens MSU97 or the Gram-positive, Bacillus subtilis. We tested for the presence of E. coli donor cells after an 8 h incubation with S. marcescens MSU97 and no donor cells were detected (data not shown). MSU97 produces potent bioactive compounds, potentially killing donor cells during conjugal mating (Matilla et al., 2012). Therefore, the lack of transposon mutants from MSU97 may have been due to killing of the E. coli donor strain, making successful conjugation unlikely. We were also unable to detect any transposon insertions in B. subtilis, though unlike S. marcescens MSU97, this was not due to donor cell death (data not shown). B. subtilis is a Gram-positive organism and conjugation with the Gram-negative E. coli donor strain was not successful.
In the remaining 17 strains, putative transposon mutants were detected with high frequency in all strains except Agrobacterium tumefaciens using both pKRCPN1 and pDS1028 (Table 4). For almost all strains, the optimal ratio of donor to recipient cells was 1:3 ( Table 2). To assess randomness and whether the plasposon system could be used to isolate mutants with particular traits, we performed three pilot studies (i) in the plant pathogen D. dadantii strain 3937; (ii) in the enterobacterium S39006; and (iii) in two oocydin A producing strains, the biocontrol rhizobacterium, Serratia plymuthica A153, and the phytopathogen, D. solani MK10. A schematic showing the full model of how transposon mutagenesis was performed, and mutants were identified is shown in Figure 2.
Mutations in the vfm Operon in D. dadantii are Defective for Protease Production D. dadantii 3937 was conjugated with E. coli CC118λpir carrying pDS1028. Following conjugation, transposon insertions were screened on caseinase agar containing chloramphenicol (to select for the presence of the transposon), and casein degradation was assessed by scoring for clearing around colonies. Over 15,000 random transposon insertion mutants were screened for reduced caseinase activity, and 10 mutants were identified. Of these 10 mutants, we chose to continue characterization of two mutants: REM392 and REM394.
The transposon encoded within pDS1028 also contains an origin of replication: oriR6K. As a result, genomic DNA from any transposon mutant can be digested, with a restriction endonuclease that does not cut within the transposon and can then be self-ligated to form a plasmid that will form a replicon in a pir + strain. Genomic DNA from REM392 was digested with NsiI, self-ligated, and used to transform E. coli strain CC118λpir. Using oligonucleotides internal to the transposon, the sequence adjacent to the transposon was identified. By applying this method, we identified an insertion in vfmE (REM392) and vfmA (REM394) that caused reduced caseinase (or protease) production in these strains.
VfmE is a transcriptional regulator of the AraC family and was previously identified as an activator of virulence determinants in D. dadantii (Nasser et al., 2013). We also identified a mutation in vfmA, the first gene in one operon of the vfm region, and disrupts the function of five genes: vfmA, vfmZ, vfmB, vfmC, and vfmD through polarity (Figure 3). VfmA shares high levels of homology with 3-oxoacyl-acyl-carrier-proteins, a subclass of decarboxylating condensing enzymes that include polyketide synthases, though the full role of this protein has not been studied in D. dadantii (Nasser et al., 2013).
To characterize these mutations further, each mutant was tested for protease, cellulase and pectate lyase production (Figure 3). We also examined siderophore production and swimming motility in each of mutant strains. Swimming motility was reduced in both of the mutants tested ( Figure 3G). Protease and cellulase production were also reduced in both mutants. Siderophore production was slightly reduced in both of the mutants tested and we were unable to detect a difference in pectate lyase production between mutant and wild type D. dadantii using plate assays (Figure 3).

A Mutation in pecS Results in Production of the Pigment Indigoidine
During our screen of D. dadantii mutants, we also identified a mutant that appeared darker on caseinase plates. D.
dadantii is known to have the cryptic capacity to produce the blue pigment indigoidine, though not naturally under standard laboratory conditions (Reverchon et al., 2002). On LB agar rather than caseinase agar, REM393 still produced the blue pigment (data not shown) and we concluded that this blue pigment was likely indigoidine and the transposon insertion in this strain led to cryptic activation.
Previous work demonstrated that the transcriptional regulator PecS represses production of indigoidine under standard laboratory growth conditions (Reverchon et al., 2002). Therefore, we examined whether the transposon insertions in this strain was located within pecS. Using RP-PCR, we determined that the transposon insertion within REM393 was located within pecS (Figure 4). Mutations in pecS have been characterized previously (Reverchon et al., 2002), but we wanted to show quantitatively that indigoidine production was activated (or derepressed) in REM393, compared with wild type. Pigment levels were assayed in wild type and REM393 in different growth media. Indigoidine levels in wild type cells were significantly lower than in a pecS mutant. We also tested indigoidine levels in the vfmA and vfmE mutants. In both of these strains, indigoidine levels were significantly less than wild type when grown in PMB media, suggesting a link between the vfm region and indigoidine production.

Identification of a Transcriptional Fusion in gvpA1 in S39006
S39006 produces gas vesicles, proteinaceous intracellular organelles that are permeable to gas but not liquids. These organelles facilitate buoyancy in a static water column, allowing cells to colonize the air-liquid interface (Ramsay et al., 2011). Gas vesicles are visible in cells by phase contrast microscopy, as they conglomerate as a light-refractile gas "vacuole" inside the cell. S39006 colonies appear opaque on plates due to the production of gas vesicles and colonies incapable of producing gas vesicles appear translucent. To identify mutants that lack gas vesicles, S39006 was mutagenized using the donor plasposon pKRCPN1 and translucent (presumptive gas vesicle defective) strains were identified. We were particularly interested in identifying transcriptional fusions of the βgalactosidase gene to gvpA1, the first gene in the primary gas vesicle synthesis operon. Over 5000 colonies were screened and a single translucent colony, containing the transposon 88 bp 3 ′ of the gvpA1 translational start site, in the correct orientation to form a transcriptional fusion, was identified ( Figure 5A). To confirm that the phenotypes were the result of this transposon alone, mutations were moved into a clean genetic background using the generalized transducing phage ϕOT8 followed by phenotypic confirmation in the transductants. β-galactosidase activity (a proxy for gvpA1 transcription) was monitored throughout growth and showed similar activity to β-glucuronidase fusions in the same gene that were reported by Ramsay et al. (2011; Figure 5). Thus, the plasposon system produces stable insertions and gene fusions in S39006.

Mutagenesis of the Oocydin A Gene Clusters in Serratia and Dickeya Strains
The halogenated haterumalide, oocydin A, was initially isolated from a plant epiphytic bacterial strain due to its strong bioactivity against plant-pathogenic oomycetes (Srobel et al., 1999). However, oocydin A has been also shown to possess antifungal (Thaning et al., 2001), anticancer (Takada et al., 1999), and anti-hyperlipidemic (Sato et al., 2005) properties. Previously, a random transposon mutant library in the oocydin A producing strain, S. plymuthica A153 was constructed and mutants defective in oocydin A production were isolated. We found recently that the ooc gene cluster is widely distributed within the Dickeya genus, including the aggressive phytopathogen D. solani MK10 (Matilla et al., 2015). Here, we employed the plasposon pKRCPN1 to isolate two oocydin A-defective mutants in the phytopathogen D. solani MK10, MK10oocG, and MK10oocN1 (Figures 6A,C). The transconjugant library of MK10 was screened for mutants defective in bioactivities against the plant-pathogenic fungi and oomycete, V. dahliae and P. ultimum, respectively (Figures 6B,C). The transducing bacteriophage φXF3 was used to confirm that the observed phenotypes ( Figure 6C) were associated with  single transposon insertions . One of the of the isolated mutants, strain MK10oocG, formed a transcriptional fusion which will allow the study of the expression of the large oocJ-W operon ( Figure 6A).

DISCUSSION
Random transposon mutagenesis has wide ranging usages in many different bacterial systems. This work demonstrates the FIGURE 6 | Isolation of oocydin A-defective mutants using the plasposon pKRCPN1. (A) Genetic organization of the ooc biosynthetic gene cluster in Serratia plymuthica A153 and Dickeya solani MK10. Multidomain polyketide synthase genes are shown in gray. Arrows indicate the location of the Tn-KRCPN1 transposon insertions and red arrows indicate that the transposon has been inserted in the right orientation, therefore generating a transcription fusion. The oocA gene is absent from the ooc gene clusters in all Dickeya strains. (B) Antifungal activity of S. plymuthica A153 strains against Verticillium dahliae. MmnO15 produce none and 3-5% of the oocydin A wild type levels, respectively (Matilla et al., 2012). (C) Anti-oomycete bioactivities of D. solani MK10 strains toward P. ultimum. The bioassays were repeated at least three times, and representative results are shown. P. ultimum and V. dahliae pictures were taken after 48 and 96 h of incubation at 25 • C, respectively. utility of the plasposon system in different hosts, in particular, by creating transcriptional fusions which can be combined together easily. Firstly, we isolated mutants in D. dadantii defective in protease production and with cryptic activation of indigoidine production. We also used this system to create a stable transcriptional fusion in gvpA1, the first gene in the gas vesicle cluster of S39006, allowing transcriptional quantification of the gas vesicle operon throughout growth. Finally, we identified mutants in the ooc gene cluster in the phytopathogenic bacterium D. solani MK10.
The plasposon system described in this work is not the first of its kind (Dennis and Zylstra, 1998), but allows any transposon insertion to be identified in two ways, either through what we term "Random Primed PCR" (RP-PCR) or through the creation of a replicon using the origin of replication found within the transposon itself. The use of two different plasposon selection systems allows increased flexibility, e.g., the construction of double mutants, making this a useful resource when engineering targeted mutants. Often one of the most laborious stages of a transposon mutagenesis is the identification of individual insertion sites. The plasposon system described here allows for identification by two different methods, an important feature should any single technique fail to provide a clean answer.
Our pilot studies also yielded interesting results that demonstrated the efficacy of our system. The vfm region of D. dadantii has been identified as important for virulence (Nasser et al., 2013). This operon is responsible for production of a new, as yet unidentified, intercellular signal. Here we identified transposon insertion in two of the four transcriptional units comprising the vfm region, vfmE, and vfmA-vfmD. Tests by Nasser and colleagues showed that mutants defective in both of these transcriptional units showed reduced virulence in the Saintpaulia ionantha (African violet) plant model (2013). We found similar results in our analysis of pectate lyase and protease production in our transposon mutants (though not for cellulase production, where our vfmE and vfmA mutants did not produce significantly less cellulase activity). This may be due to differences in the assays that were used.
The cryptic pigment indigoidine is not expressed under normal laboratory conditions in D. dadantii (Reverchon et al., 2002). We identified a mutant (REM393), within pecS, that produced indigoidine. The pecS mutant showed increased pectate lyase activity, cellulase activity and siderophore production. In media containing glycerol as a carbon source (MM9), we saw a significant increase in indigoidine (Figure 4). These observations are consistent with the results of Reverchon et al. (2002) who demonstrated that pecS acts as a repressor of indigoidine.
We also examined indigoidine production in our vfmA and vfmE mutants. Though indigoidine is not normally produced in standard laboratory media, we observed small amounts of production in media containing polygalacturonic acid (PMB), a degradation product of plant cell walls-and perhaps a mimic of in planta conditions. However, in a vfmA or vfmE mutant, no such production was observed, suggesting that the metabolite or putative signaling molecule produced from this region may somehow induce indigoidine production in addition to affecting plant virulence (Nasser et al., 2013).
The identification of biosynthetic clusters for different antifungal compounds is of great interest. From previous work, the genetic cluster responsible for production of the halogenated haterumalide, oocydin A, was identified in A153 (Matilla et al., 2012). From bioinformatic analysis, we knew that this cluster was also present in the phytopathogen D. solani MK10. Using the plasposon system, two transposon insertion mutants within the MK10 cluster were identified, one a transcriptional fusion, that could be used to examine transcriptional changes of the cluster under different environmental conditions.
The primary purpose of this study was to construct and determine the general utility of the plasposon system. We believe that these constructs will have wide utility in studies of Gramnegative bacteria. High rates of conjugation were observed in most Gram-negative strains (unless donor killing was a problem). This system will improve functional genomics in Serratia, Pectobacterium, Citrobacter, Dickeya, and Chromobacterium species, but this was a limited repertoire of test hosts. It is very likely that this plasposon system will have far wider utility in taxonomically distant Gram-negative bacteria.

AUTHOR CONTRIBUTIONS
RM, DS, MM, KR, NW, JR, MW, and GS conceived of the study. RM, DS, MM, KR, ER, NW, and JR designed and performed experiments. AD sequenced the plasmid constructs. RM wrote the manuscript. RM, AD, MW, and GS edited the manuscript.