A list of the bacterial strains used in this study is shown in Table 1. All species were grown in LB medium (Sambrook and Russell, 2001), supplemented when necessary with 0.3 mM thymidine, 0.3 mM 2,6-diaminopimelic acid (DAP), chloramphenicol (25 μg/ml) or kanamycin (50 μg/ml for all; except for Pseudomonas aeruginosa 300 μg/ml). Solid medium was obtained by adding 15 g/l of agar. Bacteria were incubated at 37°C (Escherichia coli, Acinetobacter baumannii, Acinetobacter baylyi, and Pseudomonas aeruginosa) or 30°C (Shewanella oneidensis, Pantoea ananatis and Pseudomonas fluorescens).

The pISY100mini suicide plasmid was created by inserting a synthetic DNA sequence into the R6K γ origin plasmid pSW23T (Demarre et al., 2005) between the SalI and the SacI restriction sites. The inserted DNA sequence includes ISY100 inverted repeats IRL and IRR flanking the aminoglycoside phosphotransferase gene (kanamycin resistance, Kan^R) from Tn5 (Beck et al., 1982), forming the mini-ISY100 transposon, followed by the T7 g10-L ribosome binding site (Olins and Rangwala, 1989) and the ISY100 transposase gene (Feng and Colloms, 2007). The pISY100mini-LP plasmid contained a ΦC31 integrase-based “landing pad” in pISY100mini, and was constructed by inserting synthetic DNA sequences containing a ΦC31 attB^TT site (GTGCCAGGGCGTGCCCTTGGGCTCCCCGGGCGCG) and the counter-selectable E. coli rpsL gene between the AatII and BsaI sites of pISY100mini, and a fragment containing an attB^TC site (GTGCCAGGGCGTGCCCTCGGGCTCCCCGGGCGCG), in the opposite orientation, between the NheI and the SpeI sites. Replication of these ISY100 donor plasmids requires the pir gene, therefore both plasmids were maintained in E. coli π3 (Demarre et al., 2005). All plasmids were verified by DNA sequencing.

Suicide plasmids were introduced into bacteria using chemical transformation or electroporation. Chemical transformation of E. coli, P. ananatis, P. fluorescens, S. oneidensis used a standard heat-shock protocol with calcium chloride treated cells (Sambrook and Russell, 2001). To prepare electro-competent cells, 200 ml of exponentially growing culture (OD600 = 0.4) was centrifuged for 7 min at 8000 g then resuspended in 100 ml pre-chilled 10% glycerol. Centrifugation was repeated three more times with cells resuspended first in 50 ml and then in 25 ml of 10% glycerol, before final resuspension in 500 μl of 10% glycerol. Electroporation was performed in 0.2 cm cuvettes at 2.5 kV (0.55 kV for S. oneidensis) using 40 μl of cells mixed with 165 ng of plasmid DNA. After electroporation, 1 ml of LB (with thymidine for π3) was added and cells were incubated with shaking for 1.5 h at 37°C or 30°C as appropriate. After this, an aliquot of cells was spread on selective plates. All the experiments were repeated in triplicate, and results reported as mean ± standard deviation.

The donor strain for the conjugation experiments was obtained by introducing pISY100mini or pISY100mini-LP into E. coli MFDpir (Ferrières et al., 2010) by chemical transformation. Overnight cultures of donor and recipient strains (1 ml each) were pelleted by centrifugation (3000 g – 5 min) and washed twice in 500 μl of dilution buffer (10 mM Tris pH 7.5, 10 mM MgSO₄, 68 mM NaCl₂) before final resuspension in 500 μl of dilution buffer. The donor strain (100 μl) was mixed gently with 50 μl of the recipient cells (100 μl for E. coli) by pipetting up and down. The mixture was pipetted without spreading at the center of a LB/DAP agar plate and incubated at 30°C for 5 h (A. baylyi, A. baumannii, and P. aeruginosa), 8 h (E. coli and S. oneidensis) or overnight (16 h; P. ananatis, P. fluorescens). Cells were washed off the plate with 3 ml of LB (700 μl for the experiments conducted with pISY100-LP-mini) and serial dilutions were spread on plates containing kanamycin and no DAP, and incubated at 30 or 37°C for 48 h to select for recipient cells containing a mini-ISY100 transposon inserted into the chromosome. Each experiment was repeated independently three times to determine the mean and standard deviation.

The protocol for colony arbitrary PCR was adapted from Das et al. (2005) using the primers KanR_3 GACCGCTTCCTCGTGCTTTAC, KanR_4 TCTATCGCCTTCTTGACGAGTTC, Arb1 GGCCACGCGTCGACTAGTCANNNNNNNNNNGATAT, and Arb2 GGCCACGCGTCGACTAGTCA. A single colony was picked from a plate, resuspended in 50 μl of distilled water and heated at 95°C for 15 min. An aliquot of this suspension (5 μl) was added to a PCR mixture containing 25 μl Taq 2X Master Mix (New England BioLabs), 2.5 μl each of primer Kanr-3 (10 μM) and ARB1-1 (50 μM) and 15 μl dH₂O. This was subjected to an initial denaturation at 95°C for 5 min, followed by 6 cycles of 95°C – 30 s, 30°C – 30 s, 72°C – 1.5 min, and then 30 cycles of 95°C – 30 s, 45°C – 30 s, 72°C – 2 min, with a final elongation at 72°C for 4 min. Five microliter of this PCR product was used as DNA template in a second PCR mixture containing 25 μl Taq 2X Master Mix (New England BioLabs), 2.5 μl each of primer Kanr-4 and Arb2 (10 μM each) and 15 μl dH₂O and subjected to the following program: initial denaturation 95°C – 5 min, then 30 cylcles of 95°C – 30 s, 52°C – 30 s, 72°C – 2 min, followed by 72°C – 4 min. The PCR products were purified using a Qiagen PCR cleanup kit and 15 μl of each reaction was sequenced with primer Kanr-4.

The completeness of a transposon insertion mutant library can be predicted by assuming a Poisson distribution for the frequency of the transposon insertions in the genome (Baym et al., 2016). Essential genes will not be represented in the library, and are excluded from the calculations. Assuming a genome with N non-essential genes, all with an equal probability of insertion, any single gene will be targeted by the transposase with a probability of:

Where x represents the number of insertions in the gene and k the number of mutants in the collection. From this, the probability of no insertion in any given gene is:

Meaning that the total number of genes that will not have any insertions is:

By complementarity, the total number of genes that will have at least one transposon insertion is:

To estimate the number of essential genes, we referred to the empirically determined relationship of Gao et al. (2015). According to this, the percentage of essential genes y in a bacterium decays exponentially in relation to the length x of the genome following the equation:

The information on the size of the genome and the number of genes for the analyzed bacteria was derived from the KEGG database¹.

The aim of this work was to design and test a suicide plasmid for the delivery of ISY100 mini-transposon insertions to the genomes of a variety of bacterial species. The suicide plasmid (pISY100mini; Figure 1A) contains a mini transposon (mini-ISY100), consisting of the Tn5 aminoglycoside phosphotransferase gene (Kan^R) between ISY100 inverted repeat sequences (IRL and IRR), and has an R6K γ origin of replication. Replication of pISY100mini is strictly dependent on the R6K pir-encoded Π protein, which is not encoded on the plasmid but can be provided in trans allowing replication in pir⁺ host strains. ISY100 transposase is encoded on pISY100mini outside the mini-ISY100 transposon. After introducing pISY100mini into a pir^- recipient strain, the plasmid cannot replicate, and kanamycin resistant colonies will arise only if the mini-ISY100 transposes to the chromosome. Insertions should be stable as the transposase gene is left behind on the non-replicating donor plasmid. pISY100mini carries a gene conferring resistance to chloramphenicol (Cm^R) outside the mini-ISY100 so that colonies arising from transposition should be kanamycin resistant but chloramphenicol sensitive.

The suicide plasmid could be delivered to recipient cells by transformation, however, conjugation might be a more efficient delivery system. An IncP origin of transfer (oriT) was therefore included on the pISY100mini backbone to allow mobilization of pISY100mini using the RP4 conjugative machinery.

To test the function and the efficiency of our mini-ISY100 delivery system, transposition was assayed in two commonly used (pir^-) strains of E. coli: DH5α and TOP10. The pISY100mini suicide plasmid was introduced into these strains using a simple chemical transformation protocol. No colonies were obtained when samples were spread on plates containing chloramphenicol, confirming that the suicide plasmid cannot replicate in these host strains. The number of cells in which transposition had taken place was determined by counting the number of colonies on LB plates containing kanamycin. This number ranged from an average of 18 per transformation for DH5α, to about 120 for TOP10 (Table 2). The total number of cells surviving the transposition procedure was determined by plating on non-selective plates, and the transposition frequency was then calculated by dividing the number of kanamycin resistant colonies by the total number of colony forming units (Table 2). The results show that approximately 1 in 800,000 (DH5α) to 1 in 280,000 (TOP10) cells in the transformation undergo transposition onto their chromosome.

Production of a transposon insertion by transformation with pISY100mini requires two successful steps: delivery of DNA into the cell, followed by transposition from the suicide plasmid to the host genome. To separate these two different events, we defined the transposition “activity” (Table 2) as the ratio between the transposition frequency of pISY100mini, depending on both transposition and DNA uptake, and the transformation frequency of the cells which depends just on delivery of the DNA. The transformation efficiency cannot be measured in DH5α and TOP10 using pISY100mini, as this plasmid cannot replicate in these cells. Instead, the efficiency of DNA delivery by chemical transformation was measured for each batch of competent cells, using the plasmid pUC71K (Taylor and Rose, 1988). This plasmid is comparable in size to pISY100mini and carries the same Kan^R gene, but can be maintained in a pir^- host due to its pMB1/ColE1 origin of replication. The results indicate that after chemical transformation, in the order of 1 in 1000 cells takes up plasmid DNA, and of these about 1 in 1000 has a transposition event to the chromosome (Table 2).

To increase the transformation frequency, electroporation was then tested to deliver pISY100mini for transposition. In these experiments, the transformation frequency increased, but the transposition activity decreased, so that the overall transposition frequency was similar to that obtained with chemical transformation (Table 2). Because of the large number of cells used in the electroporation reactions, significantly more transposition events were obtained by electroporation than by chemical transformation; ∼2000 colonies were obtained by electroporation with both DH5α and TOP10 (Table 2). Thus, a library large enough to have nearly full coverage of all non-essential genes in E. coli (requiring ∼12,000 insertions mutants) could be produced by combining colonies from several electroporation reactions.

Conjugation is a highly efficient method of delivering plasmid DNA to bacterial cells, so we tested it as an alternative delivery method for pISY100mini. The suicide plasmid pISY100mini was introduced into E. coli MFDpir (Ferrières et al., 2010) by chemical transformation. This strain expresses the pir gene for replication of pISY100mini and the RP4 conjugative proteins, allowing mobilization of pISY100mini. MFDpir also contains a mutation in the dapA gene, allowing easy counterselection after conjugation using media lacking diaminopimelic acid (DAP).

A plate-based mating reaction was set up between the MFDpir pISY100mini donor strain and the two different recipient strains, DH5α and TOP10, as described in the Materials and Methods section. After 8 h of co-incubation of donor and recipient strains on LB DAP plates, cells were washed off and dilutions were plated on different media. No colonies were obtained on LB Cm plates, confirming that pISY100mini cannot replicate in the recipients, and that there is no reversion of the dapA mutation in the donor. LB plates lacking DAP were used to count the total number of recipient cells, while LB lacking DAP but containing kanamycin was used to select for transposition events in DH5α or TOP10. Libraries of 10⁵ transposon insertion mutants were routinely obtained from each conjugation reaction (Table 3), although it should be noted that due to the extended incubation times during the conjugation, these colonies probably do not all represent independent transposition events. The transposition frequency after conjugation, ∼10^-2–10^-3, calculated as the proportion of recipient cells that became kanamycin resistant, was significantly higher than obtained by transformation, ∼10^-5–10^-6 (compare Tables 2, 3).

The conjugation frequency of pISY100mini from MFDpir was measured by mating MFDpir/pISY100mini with the pir⁺ E. coli strain π3, allowing replication of pISY100mini in the recipient. Cells were plated on LB, to count the total number of recipient cells, and LB/Kan+Cm, to determine the number of transconjugants, giving a conjugation frequency of ∼4 × 10^-2. The transposition activity of the mini-ISY100 transposon after conjugation in the E. coli pir^- strains was then estimated by dividing the transposition frequency by the conjugation frequency of pISY100mini from MFDpir to π3 (Table 3). These calculations indicate that a substantial proportion of cells that receive pISY100mini by conjugation (as high as 40% for DH5α) undergo transposition from pISY100mini to the host chromosome.

To show that pISY100mini can be used to create transposon insertion libraries in other bacterial species, we tested transposition in S. oneidensis, P. ananatis, and P. fluorescens. These species have all shown biotechnological potential (Fredrickson et al., 2008; Silby et al., 2009; Hara et al., 2012), and are representatives of three different orders of γ proteobacteria. Our first attempts to deliver pISY100mini in these species using chemical transformation and electroporation were unsuccessful, producing no Kan^R colonies. Therefore, we attempted to deliver pISY100mini by inter-species conjugation, using E. coli MFDpir/pISY100mini as donor. Conjugation was conducted on LB DAP plates by mixing donor and recipient strains as described in the Materials and Methods section, and leaving them to incubate at 30°C for 8–16 h. No colonies were obtained when the conjugations were plated on LB Cm plates, demonstrating that pISY100mini cannot replicate in any of these species, and controls with individual parental strains produced no colonies on LB Kan plates. Conjugation between MFDpir/pISY100mini and the different recipient species yielded from 10³ to 10⁵ Kan^R colonies, depending on the recipient species. Transposition frequencies (the fraction of recipient cells that had become Kan^R) were in the order of 10^-6 for P. fluorescens, 10^-5 for P. ananatis and 10^-4 for S. oneidensis (Table 4).

To confirm that the observed Kan^R colonies were due to mini-ISY100 transposon insertions in the recipient host genome, arbitrary PCR was used to amplify the DNA flanking transposon insertion sites (Das et al., 2005). Five Kan^R colonies produced by conjugation of MFDpir/pISY100mini with each species were randomly selected and subjected to colony PCR using a forward primer specific for the kanamycin resistance gene and a reverse “arbitrary” primer containing a 10 bp random sequence at its 3′ end. The products of this reaction were further amplified by PCR using a nested primer from the kanamycin resistance gene and a reverse primer matching a fixed region from the 5′end of the degenerate primer (Figure 1B). PCR products were sequenced by dideoxy sequencing using a primer from the kanamycin resistance gene, allowing unique determination of the mini-ISY100 transposon insertion site (Supplementary Table S1). All the transposon insertions could be identified in the genomic sequence of the target organism and were at TA dinucleotides, as expected for ISY100.

To create a useful insertional knock-out library in the species studied, the library should be large enough for nearly full coverage of the genome. A practical target to aim for is for >95% of non-essential genes to have at least one insertion. From the Poisson distribution (see the Materials and Methods section), this requires approximately 3N insertions, where N is the number of non-essential genes in the genome. The number of non-essential genes has been determined for S. oneidensis as N_oneidensis≅3700 genes (Baym et al., 2016), and can be estimated from the annotated genome sequences of P. ananatis (Hara et al., 2012) and P. fluorescens (Silby et al., 2009). P. ananatis has a genome of 4.88 Mbp, with 4067 protein-coding genes, while P. fluorescens has a genome of 6.72 Mbp, with 5921 protein-coding genes. Based on their genome sizes, about 12% of these genes are expected to be essential (Gao et al., 2015) resulting in N_ananatis≅3600 and N_fluorescens≅5200. Thus, libraries ranging from 12,000 to 16,000 mutants are required for 95% genomic coverage of S. oneidensis, P. ananatis and P. fluorescens. Therefore, the numbers of mini-ISY100 transposition events obtained with our delivery system are sufficient to create knock-out libraries with 95% coverage in S. oneidensis and P. ananatis in a single experiment, while two independent conjugation reactions of this scale would be required for P. fluorescens (Table 4).

Having proved the efficiency of transposition using pISY100mini, we wished to demonstrate the utility of ISY100 for genetic screens. S. oneidensis produces colonies with a pale brown coloration on LB agar plates and we decided to screen for mutations in genes required for this colony color. A library of S. oneidensis mutants was created by mating MFDpir/pISY100mini with S. oneidensis and selecting on LB Kan plates. From a library of approximately 12,000 mutants, four white colonies were identified (Figure 2) and streaked to single colonies. Arbitrary PCR was used to amplify the DNA flanking the mini-ISY100 insertions in these mutants, and the insertion site was determined by DNA sequencing. Three of the mutants had insertions in the ccmF gene of S. oneidensis, at three different locations within this gene, while the fourth colony had an insertion in the dsbD gene (Table 5). The product of ccmF, a cytochrome c maturation protein, has a role in the reduction of heme for its transfer to the apocytochrome c (Verissimo and Daldal, 2014). The dsbD gene encodes a membrane protein belonging to the thioredoxin family, whose function in bacteria is to maintain apocytochrome c in a reduced form by acting on the assembly protein CcmG (Verissimo and Daldal, 2014). Both of these proteins are therefore involved in the biogenesis and maturation of cytochrome c, indicating that a defect in this pathway might underlie the observed pale colony color phenotype. Indeed, a pale colony phenotype has previously been observed for mutants in ccmF in S. oneidensis (Fu et al., 2015). A causal relationship between the mini-ISY100 insertion in dsbD and pale colony color should be confirmed by complementation and/or gene knockout experiments.

Next, we investigated the use of mini-ISY100 to deliver a genetic cargo to the genomes of different bacterial species. The sequence of pISY100mini was modified so that it could be used to deliver a chromosomal “landing pad” for ΦC31 integrase-mediated cassette exchange (Fogg et al., 2014). ΦC31 integrase catalyzes efficient unidirectional recombination between short DNA sites knows as attP and attB (Thorpe and Smith, 1998). A ΦC31 attB site with a TT central dinucleotide (attB^TT) together with the rpsL streptomycin sensitivity gene from E. coli was placed at one end of the mini-ISY100 just inside the left inverted repeat, and attB with a TC central dinucleotide (attB^TC) was placed at the other end, just inside IRR, creating pISY100mini-LP (Figure 3A). These changes add 684 bp to mini-ISY100, bringing the total size of the genetic cargo carried by mini-ISY100-LP to 1,679 bp. pISY100mini-LP can be delivered by conjugation as before, and transposon insertions can be selected using the kanamycin resistance gene. Once on the chromosome of a bacterial host, a genetic cassette flanked by attP^TT and attP^TC recombination sites could be integrated into the landing pad using ΦC31 integrase-mediated recombination (Figure 3B; S. D. Colloms, unpublished results). This cassette exchange could be selected for by loss of the rpsL (Reyrat et al., 1998) or gain of a marker carried on the cassette.

The ability of this larger landing pad ISY100 to transpose was tested in the following bacteria: E. coli MG1655, A. baumannii ATCC19606, A. baylyi ADP1 and P. aeruginosa PAO1. MG1655 was used as a wild-type E. coli control. P. aeruginosa and A. baumannii are human pathogens of high clinical importance – both are listed within the top three species where urgent need for additional research is required (World Health Organization, 2017). A. baylyi is a genetically tractable relative of A. baumannii. New genetic tools that work in these species could help identify genes required for their virulence and/or new drug targets.

MFDpir/pISY100mini-LP was used as a donor for conjugation. Conjugations were plated on LB to count recipient cells, LB Cm to verify that the donor plasmid cannot replicate in the recipient strains, and LB Kan to determine the number of transposition events. No colonies were obtained on LB Cm plates, whereas hundreds (A. baumannii) to over a million (P. aeruginosa) kanamycin resistant colonies were obtained from the conjugations (Table 6). Using arbitrary PCR and DNA sequencing, the locations of transposon insertions were determined in three random kanamycin resistant colonies for each species. Insertions were in apparently random TA dinucleotides in the genomes of all species (Supplementary Table S2).