Improved mini-Tn7 Delivery Plasmids for Fluorescent Labeling of Stenotrophomonas maltophilia

ABSTRACT Fluorescently labeled bacterial cells have become indispensable for many aspects of microbiological research, including studies on biofilm formation as an important virulence factor of various opportunistic bacteria of environmental origin such as Stenotrophomonas maltophilia. Using a Tn7-based genomic integration system, we report the construction of improved mini-Tn7 delivery plasmids for labeling of S. maltophilia with sfGFP, mCherry, tdTomato and mKate2 by expressing their codon-optimized genes from a strong, constitutive promoter and an optimized ribosomal binding site. Transposition of the mini-Tn7 transposons into single neutral sites located on average 25 nucleotides downstream of the 3′-end of the conserved glmS gene of different S. maltophilia wild-type strains did not have any adverse effects on the fitness of their fluorescently labeled derivatives. This was demonstrated by comparative analyses of growth, resistance profiles against 18 antibiotics of different classes, the ability to form biofilms on abiotic and biotic surfaces, also independent of the fluorescent protein expressed, and virulence in Galleria mellonella. It is also shown that the mini-Tn7 elements remained stably integrated in the genome of S. maltophilia over a prolonged period of time in the absence of antibiotic selection pressure. Overall, we provide evidence that the new improved mini-Tn7 delivery plasmids are valuable tools for generating fluorescently labeled S. maltophilia strains that are indistinguishable in their properties from their parental wild-type strains. IMPORTANCE The bacterium S. maltophilia is an important opportunistic nosocomial pathogen that can cause bacteremia and pneumonia in immunocompromised patients with a high rate of mortality. It is now considered as a clinically relevant and notorious pathogen in cystic fibrosis patients but has also been isolated from lung specimen of healthy donors. The high intrinsic resistance to a wide range of antibiotics complicates treatment and most likely contributes to the increasing incidence of S. maltophilia infections worldwide. One important virulence-related trait of S. maltophilia is the ability to form biofilms on any surface, which may result in the development of increased transient phenotypic resistance to antimicrobials. The significance of our work is to provide a mini-Tn7-based labeling system for S. maltophilia to study the mechanisms of biofilm formation or host-pathogen interactions with live bacteria under non-destructive conditions.

neither affect bacterial growth, the antibiotic resistance pattern, biofilm formation, nor virulence of the fluorescently tagged S. maltophilia strains.

RESULTS AND DISCUSSION
The 39-end of the glmS gene in S. maltophilia. Mini-Tn7 transposons have become valuable tools for inserting heterologous genes into bacterial genomes in a site-and orientation-specific manner (44)(45)(46)(47)(48)(49). In the presence of the Tn7-encoded TnsABCD(E) transposase complex, the right (Tn7R) and left (Tn7L) ends of the transposon are sufficient for transposition of mini-Tn7 elements at a high frequency into a single Tn7 insertion site that is located about 25 nucleotides downstream of the conserved glmS gene for the essential glutamine-fructose-6-phosphate aminotransferase involved in cell wall metabolism of many bacteria (44,50,51). The transposase complex additionally requires an attTn7-associated sequence motif at the immediate 39-end of glmS to direct transposition into the single insertion site (50). With very few exceptions that contain multiple glmS genes and attTn7 sites, such as members of the genus Burkholderia (52), most bacterial species carry only a single glmS gene and thus only one attTn7 site, which allows the insertion of a single-copy gene at a unique neutral intergenic site of the bacterial genome.
To the best of our knowledge, there is a very limited number of previous studies using S. maltophilia strains that have been fluorescently labeled with a mini-Tn7 delivery system (53)(54)(55). In fact, the plasmid pURR25 carrying the mini-Tn7KSGFP transposon and engineered for use in other g -Proteobacteria, such as Photorhabdus luminescens, Pseudomonas aeruginosa or Serratia spp. (56), was applied in all these studies for labeling of S. maltophilia with GFP, but without providing further details or background information.
To address the question of whether S. maltophilia is indeed a suitable host for the development of a site-specific chromosome integration system based on Tn7, we searched the NCBI genome database and found only one glmS gene in most S. maltophilia strains, which suggests a single attTn7 site in this bacterium. A total of 75 S. maltophilia strains, belonging to 15 out of 23 different monophyletic lineages (13), was then randomly selected to compare the 39-ends and 40 nucleotides of the intergenic regions downstream of the glmS gene (Fig. S1). Noteworthy, the sequence variations downstream of the glmS gene allowed classification of the intergenic regions into groups that were largely consistent with the monophyletic lineages of the strains, suggesting, like in many other bacteria (57,58), a naturally evolved region as an insertion site of mini-Tn7 transposons in S. maltophilia. This assumption was additionally supported by the fact that the attTn7-associated sequence motif at the immediate 39-end of glmS was as highly conserved in S. maltophilia as in other Gram-negative bacteria. Taken together, the data described above strongly suggested that S. maltophilia contains the genetic background required for single-copy insertion of Tn7-based genetic elements.
Construction of mini-Tn7 delivery plasmids for S. maltophilia. It has long been known that transposable genetic elements based on Tn7 have an extraordinarily broad host range, with at least 20 different bacterial species in which Tn7 is able to transpose (44). The original classification of S. maltophilia as a member of the genus Pseudomonas (59), followed by assignment of the bacterium to the genus Xanthomonas (60) and reclassification as Stenotrophomonas (61) has prompted us to first test the utility of pUC18T-mini-Tn7T-Gm-ecfp, pUC18T-mini-Tn7T-Gm-eyfp and pUC18T-mini-Tn7T-Gm-dsRedExpress for fluorescent labeling of S. maltophilia. The latter mobilizable mini-Tn7 delivery vectors, which have been successfully developed to fluorescently label P. aeruginosa (45), were transferred from E. coli DH5a donor strains to different S. maltophilia recipients by four-parental mating experiments, using E. coli DH5a with pRK2013 for mobilization of mini-Tn7 plasmids and E. coli SM10 (lpir) carrying the plasmid pUX-BF13 for TnsABCDE transposase expression as helper strains. However, although colony PCRs using the primer pairs P Tn7L /PSmlt glmS-up and PSmlt glmS-down /P Tn7R indicated that insertion of the mini-Tn7 elements occured at attTn7 in S. maltophilia, the signals from all fluorescent proteins proved to be quite weak (data not shown).
We therefore decided to develop improved mini-Tn7 delivery plasmids derived from the base vectors pUC18R6K-mini-Tn7T-Gm and pUC18T-mini-Tn7T (44,45). Using mini-Tn7 Delivery Plasmids for Stenotrophomonas Applied and Environmental Microbiology a multiple-step cloning strategy, the fluorescence intensity of S. maltophilia cells as examined by confocal laser-scanning microscopy (CLSM) was successively increased in the course of plasmid construction. As described in detail in Materials and Methods and depicted in Fig. S2, amplification of the fluorescence signal was mainly achieved by optimization of three functional mini-Tn7 elements, i.e., i) adaptation of codon usage of the genes for the fluorescent proteins sfGFP, mCherry, tdTomato and mKate2 to their expression in bacteria of the family Xanthomonadaceae, ii) change to the strong, constitutive promoter P c from class III integron of Delftia acidovorans C17 (62,63) to drive expression of the fluorescent proteins, and iii) replacement of the putative ribosomal site sequence GCGAGC of the rpoD gene of S. maltophilia K279a with the six-base consensus sequence AGGAGG seven bases upstream of the AUG start codons. Interestingly, the P c promoter was also recently used to achieve high expression levels of fluorescent proteins on mini-Tn7 elements in P. fluorescens (64) and could therefore serve as an alternative to other constitutive promoters used for protein expression in different Gram-negative bacteria. If required, genes for other fluorescent proteins can be placed under the control of the P c promoter by replacement of the existing genes using BamHI/Sacl cloning (see Fig. S2). Gentamycin is one of the few antibiotics to which multidrug-resistant S. maltophilia strains are sensitive to varying degree. Therefore, we did not replace the original aacC1 gene for aminoglycoside-(3)-N-acetyltransferase in the improved mini-Tn7 elements, but usually checked the intrinsic resistance to gentamicin of the strains to be labeled before each experiment. For selection, we routinely used a concentration of 60 mg/mL gentamicin, but this could be adjusted if intrinsic resistance to this antibiotic was elevated in certain strains. As part of the initial mini-Tn7T-Gm transposon encoded by pUC18R6K-mini-Tn7T-Gm (Fig. S2), the aacC1 gene was flanked by FLP recombinase target (FRT) sites and could therefore offer the opportunity of being removed from the genome by Flp-mediated excision to yield strains with unmarked mini-Tn7 insertions (44). For expression of the yeast FLP recombinase, the helper plasmid pFLP2 has been widely used (45,65). However, due to the highly intrinsic resistance of S. maltophilia to ß-lactams, the plasmid was unsuitable for removal of the FRT cassette from these bacteria. Following the approach to construct an alternative for pFLP2 in phytopathogenic Xanthomonas spp. (47), we cloned a 5.1-kb Pstl fragment containing the cI 857 -FLP-sacB fragment from pFLP2 into the Pstl site of pBBR1MCS-1, a cloning vector with moderate copy number coding for chloramphenicol resistance (66). However, various attempts to excise the aacC1 gene in the presence of the pBBR1MCS-1-FLP2 helper plasmid failed (data not shown) and were therefore not pursued further.
Chromosomal insertion of mini-Tn7 elements and fluorescent labeling of S. maltophilia. The improved mini-Tn7 delivery plasmids could be transferred with high efficiency into various S. maltophilia isolates by conjugative transfer. For fluorescent labeling, we aimed not only to use a selection of strains from different monophyletic lineages but also to place focus on clinical isolates from different sources such as blood, lung puncture, vascular ulcer, perineum, oropharynx, hematologic neoplasia, or cystic fibrosis patients. As shown in Fig. 1 and Fig. S3, the four-parental mating experiments yielded fluorescently labeled strains with a high signal-to-noise ratio at single-cell resolution. While the site-and orientation-specific insertion of the mini-Tn7 elements downstream of glmS were routinely verified by PCR with primer pairs P Tn7L /PSmlt glmS-up and PSmlt glmS-down /P Tn7R , the actual mini-Tn7 insertion sites in S. maltophilia strains K279a, PC313, H5726, PC240, UV74, OC194 and LMG11112 were determined by sequencing of the flanking mini-Tn7 regions using DNA fragments amplified from fluorescently labeled derivatives of the strains. Because the Tn7 transposition event generates characteristic 5bp duplications of the insertion site (67), the host target sequence of the mini-Tn7 elements could be readily identified in the different S. maltophilia strains (Fig. 2, Fig. S1 and S4B). According to the nomenclature of Craig and colleagues (58,68), in which the base pair located in the middle of the 5-bp sequence is referred to as nucleotide position 0, with sequences toward glmS possessing a positive numbering, the average distance between the mini-Tn7 insertion site and the 39-end of the glmS gene was 25 nucleotides in S. maltophilia ( Fig. 2 and Fig. S1), consistent with the distance determined in other bacteria, such as Xanthomonas spp. (47), P. aeruginosa (44,45), E. coli (68), Acinetobacter baumannii (48), P. putida, Yersinia pestis, or Burkholderia thailandensis (44). Given the degree of sequence conservation of the intergenic regions downstream of the glmS gene within individual monophyletic lineages, it is reasonable to assume that S. maltophilia strains of the same monophyletic lineage have identical target sequences that serve as insertion sites for mini-Tn7 transposons (Fig. S1).
Stability of mini-Tn7 insertions. To answer the question of whether integration of the mini-Tn7 transposons into the chromosome of S. maltophilia affects strain fitness,

mini-Tn7 Delivery Plasmids for Stenotrophomonas
Applied and Environmental Microbiology we used fluorescent derivatives of strains K279a and UV74 throughout this study. The blood isolate K279a is considered the clinical reference strain for which the first wholegenome sequence of an S. maltophilia bacterium was available (69), whereas strain UV74 isolated from a vascular ulcer was found to be genetically very similar to strain D457 (70), a bronchial-aspiration isolate that is used as a model strain for studying the expression and regulation of resistance determinants in S. maltophilia (71). As a first step, we aimed to investigate the chromosomal stability of improved mini-Tn7 elements in strains K279a::sfGFP, K279a::tdTomato, UV74::sfGFP and UV74:: tdTomato. Since the aacC1 gene was genetically linked to the corresponding gene for a fluorescent protein on the inserted mini-Tn7 transposon, we considered a stable inheritance of the gentamicin resistance marker as a measure of the stability of the entire mini-Tn7 element, including a stable replication of the gene for the fluorescence protein in the absence of continued antibiotic selection. We therefore grew all strains in media without gentamicin for five consecutive days, followed by the testing of 100 individual clones of each strain for their ability to grow in media containing the antibiotic. As expected from chromosomally inserted markers, 100% of the clones retained their mini-Tn7-encoded gentamicin resistance in the absence of sustained selection

mini-Tn7 Delivery Plasmids for Stenotrophomonas
Applied and Environmental Microbiology pressure. This result was confirmed by the detection of mini-Tn7 elements in the correct orientation as 502-bp PCR products in 10 randomly selected clones of each strain (Fig. S4). Taken together, these data indicated that the mini-Tn7 transposons are stable in S. maltophilia over a longer time period without the need for continuous selection pressure.
Growth properties of fluorescently labeled S. maltophilia strains. Acquisition of antibiotic resistance determinants, e.g., through mobile genetic elements, may be associated with fitness deficits due to an overall metabolic burden and, as a result, may lead to lower bacterial growth rates (72)(73)(74). While integration of unmarked mini-Tn7 transposons at a unique neutral intergenic site downstream of the glmS gene does not appear to have adverse effects on bacterial fitness (44), we wondered whether the acquisition of the aacC1 gene together with the mini-Tn7 element and the development of resistance to gentamicin might affect the fitness of fluorescent S. maltophilia strains. As shown herein for S. maltophilia strains K279a::sfGFP, K279a::tdTomato, UV74::sfGFP and UV74::tdTomato, there were no growth differences between the parental strains and their fluorescent derivatives (Fig. 3). The mean generation times (mean 6 SD) of K279a, K279a::sfGFP, K279a::tdTomato, UV74, UV74::sfGFP and UV74::tdTomato were similar and amounted to 106. 90  To address the question of whether integrated mini-Tn7 elements have an impact on the antibiograms of fluorescently labeled S. maltophilia in comparison to their parental strains, we examined the resistance profiles of K279a, K279a::sfGFP, K279a:: tdTomato, UV74, UV74::sfGFP and UV74::tdTomato with 18 antibiotics of different classes. As shown in Table 1, the MIC values of almost all tested antibiotics against the parental strains and their fluorescently labeled derivatives were identical. Only the MIC of the combination of the antibiotic ticarcillin with the ß-lactamase inhibitor clavulanic acid against the tdTomato-labeled derivative of strain K279a was reduced by half and, conversely, the MIC values of ticarcillin-clavulanic acid combination and another combinatorial antibiotic, piperacillin-tazobactam, were doubled in UV74::tdTomato compared to UV74 and UV74::sfGFP. It must be emphasized, however, that a 2-fold dilution difference is within the intrinsic error of the broth microdilution method and is not usually considered significant in susceptibility tests (75). Overall, the results allowed us to draw the conclusion that integration of the improved mini-Tn7 transposons does neither alter the growth characteristics nor the resistance patterns of the fluorescently labeled S. maltophilia strains.
Biofilm formation of fluorescently labeled S. maltophilia strains. The ability to form biofilms as an important virulence factor is a characteristic feature of S. maltophilia (15, 41, 76-78). Because biofilm growth is another measure of bacterial fitness, changes in the capacity to form biofilms can have far-reaching consequences, such as

mini-Tn7 Delivery Plasmids for Stenotrophomonas
Applied and Environmental Microbiology altering the pathogenic and resistance potential of bacteria (74). To determine whether fluorescently labeled S. maltophilia strains are unchanged in their ability to produce biofilm, we first used crystal violet (CV) staining to quantify the amount of biofilm relative to bacterial growth to deduce the values of relative biofilm formation for strains K279a, K279a::sfGFP, K279a::tdTomato, UV74, UV74::sfGFP and UV74::tdTomato (Fig. 4). Based on the mean ratio (OD 550 of CV/OD 620 of cell growth), we were able to confirm that strains with a diffusible signal factor-based quorum sensing system of the rpf-2 type (UV74) are stronger biofilm formers than strains of the rpf-1 type (K279a) (78). Our results further showed only marginal, statistically non-significant differences in biofilm formation between the parent strains and their fluorescently labeled derivatives under the experimental conditions used, suggesting once again that the chromosomally integrated mini-Tn7 elements did not negatively affect the fitness of the bacterial cells. Imaging techniques are increasingly being used in biofilm research, with bacteria expressing fluorescent proteins finding wide application, for example, in studies of spatial biofilm structure or the dynamics of biofilm maturation as a function of time (79,80). Here we used confocal laser-scanning microscopy (CLSM) to visualize the threedimensional architecture of biofilms formed by fluorescently labeled S. maltophilia strains. For this purpose, the biofilms were grown on a polymer surface under static conditions for either 48 h (bacteria with tdTomato labeling) or 72 h (bacteria with sfGFP or mKate2 labeling), followed by reconstruction of representative biofilm images from CLSM Z-axis serial sections ( Fig. 5 and Fig. S5). As expected, the CLSM images showed biofilm-forming ability of all strains on a solid polymeric surface and strain-dependent variations in biofilm architecture, characterized by loosely to densely packed populations of bacterial cells. All strains tested appeared to maintain the biofilm mass formed after 48 h throughout the 72-h experimental period, with strain S. maltophilia LMG11112, a clinical isolate from a lung puncture, proving to be the most potent biofilm producer in this study after 48 and 72 h of growth (Fig. 5). However, we attached even more importance to the observation that the various fluorescent proteins encoded on the corresponding mini-Tn7 transposons have no effect on biofilm architecture. This was particularly evident in the Sm454::sfGFP and Sm454::mKate2 strains, which had formed biofilms of microcolonies of varying sizes after 72 h of incubation ( Fig. 5 and Fig.  S6). Interestingly, and as the example of strain Sm454::sfGFP showed (Fig. S7), the microcolonies possessed a characteristic ultrastructure consisting of densely packed cell layers adhering to the polymer surface and layers with loosely packed cell aggregates. The latter were reminiscent of rosettes, which appeared to be formed by adhesion between cells at their poles. Of note, a similar rosette-like biofilm architecture of S. maltophilia Sm454 was recently observed, but visualized using a live/dead staining protocol (41). However, the mechanisms responsible for intercellular cohesion still needs to be elucidated, e.g., whether the aggregates are held together by a unipolar adhesion polysaccharide, similar to that described for rosette formation of cells in biofilms of other bacteria, such as Rhodopseudomonas palustris (81) or Agrobacterium tumefaciens (82). Taken together, the results indicated that the presence of the mini-Tn7 elements does not affect the biofilm formation of the fluorescently labeled strains compared with their parental strains, nor does it lead to differences between biofilms of differently labeled cells of the same strain. The differently labeled S. maltophilia strains would thus be interchangeable if necessary.
Using CLSM, we were also able to monitor the succession of biofilm formation of S. maltophilia Sm314::sfGFP on confluent Calu-3 cell monolayers over a 48-h period (Fig. 6). The gradual damage of Calu-3 cells by the growing biofilm, starting with the formation of aggregates of bacterial cells on the Calu-3 cells after a period of 16 h,  shows, all strains were moderately virulent in the larval killing assays, with no significant difference between wild-type and fluorescently labeled strains. The inoculum used, approximately 5 Â 10 5 CFU, resulted in 5-10% mortality for all strains 24 h postinfection, while 35-45% of larvae were killed after 120 h. However, this assay was only useful for demonstrating that fluorescent labeling has no effect on the virulence of the strains. As shown previously, fluorescently labeled strains are not suitable for monitoring the course of infection within the larvae due to their autofluorescence and the fact that the cuticle does not allow detection of fluorescence inside the larvae (83).

MATERIALS AND METHODS
Bacterial strains, plasmids and growth conditions. All strains and plasmids used in the present study are described in Table 2. Unless otherwise stated, the E. coli and S. maltophilia strains were grown aerobically with shaking (220 rpm) at 37°C in LB-Miller medium (10 g of tryptone, 5 g of yeast extract, 10 g of NaCl per L), except E. coli SM10 (lpir)/pUX-BF13 was routinely cultured at 30°C. For plasmids containing the R6Kg origin of replication, E. coli SY327 expressing the l PI protein served as a cloning host. All other plasmids were constructed and propagated in E. coli DH5a. Gentamycin at concentrations of 15 mg/mL, 30 mg/mL and 60 mg/mL, kanamycin (30 mg/mL), ampicillin (100 mg/mL), or norfloxacin (5 mg/mL) were added to the media as required.  Germany). The cells were grown in cell culture flasks at 37°C in a humidified atmosphere of 5% CO 2 , using RPMI 1640 culture medium with stable L-glutamine, 2.0 g/L NaHCO 3 and 10% of heat-inactivated fetal bovine serum (FBS). Adherent Calu-3 cells were dissociated the day before the infection experiments by trypsinization with a trypsin 0.05%/EDTA 0.02% solution at 37°C for 3 min, collected and sedimented by centrifugation at 280 Â g for 3 min. The cell pellet was then resuspended in culture medium, followed by determination of the cell number with a hemocytometer and evaluation of cell viability with Trypan blue. Finally, the cells were transferred in a volume of 250 mL and a density of 6.25 Â 10 4 cells/cm 2 into each well of a m-Slide 8-Well ibiTreat slide (ibidi GmbH, Gräfelfing, Germany).
Construction of mini-Tn7 delivery plasmids for fluorescent labeling of S. maltophilia. Standard recombinant DNA methods were used for nucleic acid preparation and analysis (84). The sequences of primers with integrated recognition sites for restriction enzymes and control primers are listed in Table 3. Using the codon usage adaptation tool Jcat (85), codon optimization of the genes for fluorescent proteins was performed for protein expression in Xanthomonadaceae. The successful construction of all plasmids was verified by DNA sequence analysis performed by LGC Biosearch Technologies (Berlin, Germany).
As a first step, the gene for the red-fluorescence mCherry protein was placed under the control of the rpoD promoter of S. maltophilia K279a. The rpoD promoter region was obtained by PCR from the genomic DNA of S. maltophilia K279a with primers KpnI-rpoDprom and XhoI-rpoDprom, followed by digestion of the PCR product of 232 bp with Kpnl/Xhol and cloning into the Kpnl/Xhol sites of pUC18R6K-mini-Tn7T-Gm to yield pUC18R6K-mini-Tn7T-Gm-rpoD. Then the primer pair 5XhoI-Fluores/3BamHI-Fluores and pGEM-T Easy-mCherry as a template were used to amplify a 731-bp fragment containing the mCherry gene. The PCR product was digested with Xhol/BamHl and ligated into the Xhol/BamHl sites of pUC18R6K-mini-Tn7T-Gm-rpoD, yielding plasmid pUC18R6K-mini-Tn7T-Gm-rpoD-mCherry. To allow for conjugative transfer of the mini-Tn7 delivery plasmids to S. maltophilia, the pUC18T-mini-Tn7T-Gm-rpoD-mCherry plasmid carrying the oriT transfer origin was constructed by ligation of the internal mini-Tn7 Kpnl/SalI fragment of pUC18R6K-mini-Tn7T-Gm-rpoD-mCherry of 2370 bp with the 3104-bp Kpnl/SalI fragment of pUC18T-mini-Tn7T.
Chromosomal insertion of the mini-Tn7 elements in gentamicin-resistant transconjugants was verified by colony PCRs, using the primer pairs P Tn7L /PSmlt glmS-up and PSmlt glmS-down /P Tn7R (Table 3) to amplify the flanking mini-Tn7 regions of 671 bp and 502 bp with transposon-and bacterium-specific primers, respectively. For determination of the mini-Tn7 insertion site, the PCR products were ligated into the pCR2.1 cloning vector of the Original T/A Cloning Kit (ThermoFisher Scientific) and sequenced.
Confocal laser-scanning microscopy of bacterial cells and biofilms. Confocal laser-scanning microscopy (CLSM) was used for imaging of bacterial cells and biofilms. To prevent movement during imaging, the bacterial cells were immobilized using the agarose pad method essentially as described previously (92). Briefly, the cells of a 1-mL overnight culture were sedimented by centrifugation, washed once with PBS and resuspended in 250 mL of PBS. The bacterial suspension was then mixed with 1 volume of ProLong Live Antifade Reagent in PBS (ThermoFisher Scientific) and placed between a coverslip and a thin pad of 0.5% agarose in PBS.
To prepare the inoculum for biofilm formation on an abiotic polymer surface, exponentially growing cultures of fluorescently labeled strains were adjusted to a cell number of about 1.43 Â 10 4 CFU/mL in 0.5 Â Brain Heart Infusion (BHI) broth. Each well of a m-Slide 8-Well ibiTreat slide was then seeded with a total volume of 350 mL, containing about 5 Â 10 3 CFU of each bacterium, followed by incubation of the m-Slides at 30°C without shaking in a humidified chamber. Exhausted medium was replaced by 350 mL of fresh, prewarmed 0.5 Â BHI broth every 8 to 16 h. After an incubation for 48 or 72 h, the culture supernatants were aspirated and the wells were washed with 350 mL of PBS. The biofilms were then fixed in the dark with 350 mL of a 2% paraformaldehyde (PFA) solution in PBS at room temperature for 20 min. Finally, the biofilms were washed with 350 mL of PBS and overlaid with 200 mL of ProLong Live Antifade Reagent in PBS as recommended by the supplier (ThermoFisher Scientific).
For biofilm experiments on a biotic surface, only Calu-3 cells that reached a confluence of at least 90% were used for infection with S. maltophilia Sm314::sfGFP cells of the early exponential growth phase. The multiplicity of infection was adjusted to 10. Following incubation of the m-Slides at 37°C in a humidified atmosphere of 5% CO 2 for 16, 24 and 48 h, samples were fixed in the dark with 300 mL of a 1% PFA solution in PBS at room temperature for 20 min. The samples were then washed twice with Imaging was performed with a TCS SP5 inverse confocal laser-scanning microscope (Leica Microsystems, Mannheim, Germany) and analyzed with LAS AF software (version 2.73). The microscope was equipped with a Leica 63x/NA 1.40 HCX Plan Apochromat CS oil immersion objective. The sfGFP protein was excited with 488 nm laser light, and fluorescence emission was detected between 495 and 530 nm, whereas mKate2 was excited with laser light at 594 nm and fluorescence emission was detected between 605 and 670 nm. The excitation wavelength of laser light was 561 nm for both mCherry and tdTomato, but emission of their fluorescence was detected between 573 and 629 nm and 580 and 648, respectively. The excitation wavelength of laser light was 405 nm for DAPI, and its fluorescence emission was detected between 415 and 470 nm. Alexa Fluor 647 was excited with laser light at 633 nm, and fluorescence emission was detected between 660 and 685 nm. Three-dimensional biofilm images were routinely generated from confocal image stacks using the daime computer program (93). However, in some specific cases, the confocal image stacks were further processed with Leica's Lightning Deconvolution Tool (Leica Application Suite X, version 3.0.0_15697) and the Imaris Viewer 10.0.0 (Oxford Instruments).
Stability testing. The genome stability of mini-Tn7 insertions was investigated after growth of the bacteria in the absence of antibiotic selection pressure for 5 days essentially as described previously (44,94). To ensure that the initial cultures have retained their selection marker at the beginning of the experiment, fluorescently labeled strains were grown aerobically (220 rpm) at 37°C overnight in LB medium containing 60 mg/mL gentamicin. From the overnight cultures, each strain was streaked onto LB agar with 60 mg/mL gentamicin to grow single colonies at 37°C as controls for later colony PCRs. The initial overnight cultures were then grown for five consecutive days at 37°C on a rotary shaker at 200 rpm, with daily dilutions of the cultures by a thousandfold into fresh LB medium. Finally, serial dilutions of the cultures were plated on LB agar without gentamicin to subsequently test 100 single colonies of each strain for their growth on LB agar containing 60 mg/mL gentamicin. The presence and the correct orientation of the mini-Tn7 elements was examined in 10 randomly selected colonies of each strain by colony PCRs with primer pair PSmlt glmS-down /P Tn7R ( Table 3).
Growth of strains with mini-Tn7 insertions. To compare the growth of fluorescently labeled strains with that of their parental strains, fresh overnight cultures were adjusted to an OD 600 of 0.1 with LB-Miller medium, followed by transfer of 100 mL of each diluted culture to 100 mL of LB-Miller medium in each well of a Honeycomb X100 Bioscreen C sterile plate (ThermoFisher Scientific). The growth at 37°C and 220 rpm was recorded by measurement of the OD 600 values at 15-minutes intervals for a total of 24 h using a Labsystems Bioscreen C automated microbiology growth curve analysis system (Labsystems, Helsinki, Finland). All growth curves were plotted in triplicate. Based on graphs of log OD 600 versus time and estimation of the bacterial cell number from optical density, the generation time (G) of a strain was calculated using the formula G = t/(3.3 log N t /N 0 ), where t was the time interval, N t the bacterial cell number at the end of the time interval and N 0 the bacterial cell number at the beginning of the time interval.
Antibiotic susceptibility testing. MICs were determined by the broth microdilution method in cationadjusted Mueller-Hinton broth (CAMHB) following the recommendations of the Clinical and Laboratory Standards Institute (CLSI) (75). The antibiotic stock solutions were prepared according to the CLSI guidelines and the test panel included 18 antibiotics of different classes. Fluorescently labeled and wild-type parental strains were first grown overnight in CAMHB under the conditions recommended by the CLSI. The MICs were determined in sterile 96-well microplates using 2-fold serial dilutions of each antibiotic. After overnight growth of the strains, 100 mL of diluted bacterial cultures with a cell number of about 5 Â 10 5 CFU/mL each were added to the wells containing the serial antibiotic dilutions. The microtiter plates were analyzed after 20 h of incubation at 37°C. The MIC was defined as the lowest concentration of the antibiotic (in mg/mL) that prevented visual growth. MIC determinations were done in duplicate.
Biofilm formation assay. To evaluate biofilm formation, overnight cultures of the strains were grown aerobically (200 rpm) in 0.5 Â BHI broth at 37°C, followed by dilution of each culture with fresh medium to an optical density at 620 nm (OD 620 ) of 0.05. Sterile, untreated 96-well microtiter plates (BrandTech 781662) were then inoculated with the bacterial suspensions (200 mL per well) and incubated at 37°C for 24 h. Prior to biofilm quantification, cell growth was estimated in each well by measuring the OD 620 value using a microplate reader (Multilabel Plater Reader VICTOR 3 , PerkinElmer). Quantification of the biofilm biomass was performed by crystal violet (CV) staining as described previously (95). The amount of biofilm was quantified by measuring the OD 550 of dissolved CV using the microplate reader. Biofilm formation (OD 550 of CV) was normalized by cell growth (OD 620 ) and reported as relative biofilm formation. For each strain, four biological replicates with six technical replicates each were prepared. Statistical significance was determined by the non-parametric Kruskal-Wallis one-way Analysis of Variance (ANOVA) test corrected for multiple comparisons using Dunn's test.
Virulence in Galleria mellonella. Larvae of Galleria mellonella were reared in-house. A total of 30 larvae with a weight between 300 and 400 mg and no signs of melanization were infected with each S. maltophilia strain. For preparation of bacterial inoculums, fluorescently tagged and corresponding parental wild-type strains were grown overnight in 10 mL of BHI medium at 37°C on a rotary shaker at 200 rpm. The bacterial cells were then sedimented by centrifugation, washed in PBS and adjusted to contain about 5 Â 10 5 cells per larva, which has been shown previously as an optimal dose of S. maltophilia required to kill G. mellonella over a period of 24 to 96 h (38,96). The bacterial burden of the doses was confirmed by plating the cells onto BHI agar. The larvae were infected with 10 mL of the inoculum through the left proleg using a 50 mL-Hamilton Microliter syringe and incubated in the dark at 30°C in empty Petri dishes. Survival of infected larvae was scored at 24-h intervals for 5 days. The larvae were considered dead when they failed to respond to touch, which was equivalent to complete melanization (blackening of the larvae). At least two replicates were performed with different batches of larvae. Kaplan-Meier survival curves were plotted using GraphPad Prism 5.0a, and survival analysis and statistical significance was determined using the log-rank test (GraphPad Software, San Diego, CA).
The plasmids are available through the Addgene repository upon request.

SUPPLEMENTAL MATERIAL
Supplemental material is available online only. SUPPLEMENTAL FILE 1, DOCX file, 3.3 MB.