Experiments using human subjects and experimental animals were performed in strict accordance with guidelines provided by Yonsei University. Protocols were reviewed and approved by Institutional Review Board of Yonsei University College of Medicine. Permit numbers for primary culture of human tissues and mouse infection experiment were 2014-1842-001 and 2013-0369-5, respectively.

A prototype strain of P. aeruginosa called PAO1 was used in this study (Yoon et al., 2006). The bacterial strains of the other species, as listed in Figure 1, were obtained from our laboratory stock and were previously reported (Gi et al., 2015). Bacterial cultures were grown in Luria-Bertani medium (LB; 10 g tryptone, 5 g yeast extract, and 10 g NaCl/l), unless otherwise stated. Bacterial cells were cultured at 37°C for 16 h and shaken at 200 rpm. Bacterial growth in AMS and survival following the treatments with diverse antibiotics were assessed by quantifying colony forming units (CFUs).

A transposon (Tn) insertion mutant library of PAO1 was constructed following the procedures described elsewhere (Lee et al., 2016). In brief, PAO1 was conjugated with Escherichia coli SM10/λpir harboring the pBTK30 plasmid (Table 1). Gentamicin (Gm)-resistant transconjugants were grown on LB agar plates containing 50 μg/ml Gm and 50 μg/ml Irgasan (Sigma). Irgasan was added to eliminate E. coli donor cells. To isolate lysozyme-sensitive mutants, each individual mutant was grown in 0.5 × LB media (50% LB + 50% distilled water) containing 1 mg/mL lysozyme (Sigma) in 96 well plates. Bacterial growth was monitored by measuring the OD₆₀₀. The Tn insertion site of each defective mutant was determined by arbitrary PCR, followed by DNA sequencing (Lee et al., 2016). Preparation of AMS samples from Normal Healthy Trachea Epithelial (NHTE) cells and bacterial growth in AMS were conducted as previously described (Gi et al., 2015).

The bioA, bamB, and fabY deletion mutants were created by allelic replacement as previously described (Lee et al., 2012). Briefly, flanking sequences (~600 bp) at both ends of each gene were PCR amplified with primers listed in Table 2. Two inner primers (upstream reverse primer and downstream forward primer) are complementary to each other. In this strategy, the 3′-end of the upstream sequence and the 5′-end of the downstream sequence are annealed during PCR amplification without further treatment. The deletion of the PA0420, PA3800, and PA5174 genes was confirmed by PCR and DNA sequencing.

To complement the ΔPA3800 and ΔPA5174, DNA fragments containing the entire PA3800 and PA5174 genes were amplified from the PAO1 genome and ligated into EcoRI/XbaI-treated pJN105. The resultant plasmid and the control empty plasmid (i.e., pJN105) were transferred into the ΔPA3800 and ΔPA5174 mutant by electroporation, respectively. Expression of PA3800 or PA5174 gene is induced by 0.01% L-arabinose (Sigma).

Bacterial cultures grown in LB for 8 h were diluted in PBS to get bacterial suspensions with the 10⁵ CFU/mL. Ten microliters of each diluent was inoculated into 100 μL AMS to achieve the initial inoculum size of ~10³ CFU in AMS. Bacterial cells were then grown for 16 h in a humidified 37°C incubator. Bacterial growth was assessed by measuring the growth index. The values for [CFU after 16 h in AMS / CFU after 0 h in AMS] were calculated and plotted as the growth index.

Animal experiments were approved by the Committee on the Ethics of Animal Experiments of Yonsei University College of Medicine (IACUC permit number: 2013-0369-5). Animal experiments were conducted following national guidelines provided by the Korean government (Ministry for Food, Agriculture, Forestry and Fisheries) and in strict accordance with the institutional guidelines for animal care and use of laboratory animals. To test the virulence of the ΔbamB and ΔfabY mutants in vivo, 8-week-old C57BL/6N inbred female mice (Orient, Korea) were intranasally infected with 5 × 10⁶ cells of each strain (n = 4). The lungs of the infected mice were removed at 16 h post-infection. The harvested lungs were homogenized, and the number of bacteria in each organ was measured by quantifying CFUs. Caenorhabditis elegans survival tests were performed as described previously (Go et al., 2014).

Lysozyme-treated bacterial cells were visualized using a scanning electron microscope (SEM), as described previously (Yoon et al., 2011). Briefly, fixed bacterial suspensions were stained with 1% OsO4 (Sigma) and then coated with gold via an ion sputter (IB-3 Eiko, Japan). SEM (FE SEM S-800, Hitachi, Japan) was used at an acceleration voltage of 20 kV. All images were processed using ESCAN 4000 software (Bummi Universe Co., Ltd., Seoul, Korea).

Bacterial resistance against hypo-osmotic stress was measured by monitoring the decrease of OD₆₀₀ values. Each bacterial strain, grown to mid-exponential phase was harvested and washed with 1 × PBS. The bacterial cells were resuspended in sterilized deionized water. The OD₆₀₀ of each bacterial suspension was measured every 10 min for 60 min. The data were normalized with their initial OD₆₀₀-values. The PAO1, ΔbamB, and ΔfabY mutants were also observed under confocal laser scanning microscope (CLSM) with LIVE/DEAD® BacLight™ Bacterial Viability Kit to detect the difference in membrane integrity among these bacterial strains.

The experiments were repeated at least in triplicate, and data are expressed as mean ± standard deviation (SD). Data were analyzed using unpaired Student's t-test, unless otherwise stated and P < 0.05 were considered to be statistically significant. Log-rank test was used to provide statistical significance in the C. elegans lifespan experiments.

Prior literature demonstrated that P. aeruginosa can become lysozyme-sensitive only when co-treated with EDTA at pH 8.0 (Voss, 1964) or pretreated with acetone or heat (Warren et al., 1955). As expected, PAO1 grew unencumbered when grown in LB media supplemented with 1 mg/mL lysozyme (Figure 1). Likewise, two other Gram-negative species, Escherichia coli (Ec) and Salmonella enterica serovar Typhimurium (St), also grew normally in these conditions (Figure 1). These results further support our notion that Gram-negative bacterial cells are intrinsically resistant to lysozyme because their outer membrane prevents lysozyme from accessing the peptidoglycan in the periplasm. On the other hand, two Gram-positive species, Bacillus subtilis (Bs) and Listeria monocytogenes (Lm), lost their viability in response to the same treatment (Figure 1). Consistent with previous reports (Bera et al., 2005, 2006), Staphylococcus aureus (Sa) was completely resistant to the lysozyme treatment (Figure 1). PAO1 growth was not affected, even in the presence of 8 mg/mL lysozyme (data not shown), demonstrating an exceptionally high resistance to lysozyme.

In order to devise a better strategy for P. aeruginosa infection control, it would be desirable to identify lysozyme-sensitive mutants. We therefore constructed a Tn insertion mutant library and screened it for mutants that had lost their resistance to lysozyme treatment. Since lysozyme is more active in environments with reduced ionic strength (Sorrentino et al., 1982; Verhamme et al., 1988), bacterial growth was tested in two-fold diluted LB media (termed 0.5 × LB). Approximately 5,500 mutants were screened, three of which were determined to lack lysozyme resistance. In each mutant, the transposon insertion occurred in PA0420, PA3800, or PA5174 gene. To further verify the effects of these gene disruptions, we generated an in-frame deletion of each gene and examined cell growth in the presence of lysozyme. In 0.5 × LB, ΔPA3800 and ΔPA5174 mutants exhibited significant growth inhibition in the presence of 1.0 mg/mL lysozyme (Figure 2A). After 16 h of growth at 37°C with lysozyme, the CFUs of these two mutants remained similar to those at the time of inoculum (indicated by an arrow on the y-axis), indicating that bacterial growth was prevented by lysozyme at a physiological concentration. PA3800 gene encode β-barrel assembly machinery protein B (BamB) involved in outer membrane protein assembly (Jansen et al., 2012). PA5174 gene encodes probable beta-ketoacyl synthase, which plays a role in fatty acid biosynthesis (Yuan et al., 2012; Figure 2B). The deletion of PA3800 gene did not result in any growth inhibition, whereas the ΔfabY mutant exhibited a slightly affected growth, yielding ~10-fold less viable cell counts after 16 h growth in 0.5 × LB media (Figure 2A). When these mutants were grown with lysozyme, bacterial cells with crumbled rod-shaped morphology were observed (Figure S1), further demonstrating that lysozyme, an enzyme that degrades peptidoglycan polymers, damages the bacterial cell wall structure.

PA0420, bioA gene, produces adenosylmethionine-8-amino-7-oxononanoate aminotransferase, which is responsible for biotin biosynthesis (Beaume et al., 2015). Of note, the ΔbioA mutant exhibited inactive growth, even in the absence of lysozyme treatment, indicating that an interruption in biotin synthesis leads to defective growth in P. aeruginosa. When the mutant was grown in media supplemented with extraneous biotin, its growth and lysozyme susceptibility were completely restored (Figure S2). This result suggests that lysozyme sensitivity of the ΔbioA mutant is likely associated with its faulty growth; therefore, we did not pursue any further investigation on this particular mutant.

When wild type copy of bamB or fabY gene was expressed by the arabinose-inducible promoter, each mutant became fully resistant to lysozyme treatment (Figure 3). These results further confirm that the lysozyme sensitivities observed in the mutants are indeed caused by the deletion of bamB or fabY gene.

Airway mucus secretion (AMS) plays a role in the host innate defense and contains various antimicrobial components including lysozyme (Gi et al., 2015). To examine bacterial response to AMS, we treated the bacterial strains with AMS collected from primary cultures of three different human tracheal tissues. In each treatment, 10³ bacterial cells were inoculated and incubated for 16 h. After the treatments, the number of PAO1 cells increased to ~5 × 10⁶ cells (growth index of ~5,000), demonstrating the capability of wild type P. aeruginosa to propagate in AMS (Figure 4). The growth of the ΔbamB mutant was not as robust as that of PAO1. The average growth index of the mutant in AMS was ~68 (Figure 4). Of note, the growth of the ΔfabY mutant was more severely affected, yielding the growth index of ~1.0. Together, these results suggest that two lysozyme-sensitive mutants are less capable of proliferating in AMS, a frontline substance that P. aeruginosa encounters during the early stage of airway infection.

We next examined whether these mutants were also defective in establishing infections in vivo. To address this question, we utilized two different animal infection models: mouse and nematode. First, mice were intranasally infected with bacterial cells. At 16 h post-infection, mouse lung homogenates were prepared, and the bacterial cells were enumerated. Increased numbers of PAO1 cells were recovered following the 16 h infection (Figure 5A), demonstrating that wild type P. aeruginosa can replicate inside the mouse airway. By comparison, decreased numbers of ΔbamB and ΔfabY mutant cells were recovered, indicating a compromise in their ability to multiply inside a host airway (Figure 5A).

C. elegans is a nematode that has been widely used to study host-microbe interactions (Go et al., 2014). Furthermore, C. elegans is known to express 15 homologs of lysozyme that act as digestive enzymes for bacterial prey and, therefore, as key players for innate immunity (Mallo et al., 2002; Boehnisch et al., 2011). Given these characteristics, we postulated that C. elegans could serve as an appropriate model to study in vivo infectivity of lysozyme-sensitive mutants. As shown in Figure 5B, C. elegans lived significantly longer when fed ΔbamB and, to a greater extent, ΔfabY mutants; the average lifespans were 9.2 ± 0.5 (days) and 11.6 ± 0.6 (days), respectively. As expected, a considerably shorter lifespan (6.6 ± 0.3 days) was observed in worms grown by feeding on PAO1 cells (Figure 5B). Together, these results demonstrate that lysozyme-sensitive mutants are also less capable of establishing in vivo virulence.

Because lysozyme targets bacterial peptidoglycan, we postulated that the increased lysozyme sensitivity of the mutants may be associated with altered bacterial cell membrane integrity. To address this issue, we analyzed how the lysozyme-sensitive mutants responded to hypo-osmotic stress. We also assessed the penetration of propidium iodide, a fluorescent dye that normally impermeable of healthy bacterial membrane, into bacterial cells of each strain. When PBS-washed bacterial cells were resuspended in distilled water, a marked and persistent drop in OD₆₀₀-values were observed in the ΔfabY mutant over 1 h time period (Figure 6A). In contrast, bacterial cell density monitored by measuring OD₆₀₀-values was decreased only in the later stage of the experiment for the case of the ΔbamB mutant (Figure 6A). Consistent with the hypo-osmotic resistance result, ΔfabY mutant cells were more permeable to propidium iodide, a red fluorescent dye that only stains bacterial nucleic acids when the membrane integrity was compromised. Together, these results demonstrate that bacterial membrane integrity is affected, especially in the ΔfabY mutant and such a defect likely accounts for the enhanced sensitivity (i) to the AMS treatment (Figure 4) and (ii) inside the C. elegans (Figure 5B).

We then explored whether lysozyme-sensitive mutants also exhibited elevated sensitivity to antibiotic treatments. To examine bacterial responsiveness to antibiotics of diverse classes, we tested vancomycin (Day et al., 1993), ceftazidime (O'Callaghan, 1986), ampicillin (Ghobashy and Chiori, 1984), tobramycin (Hoff et al., 1974), and ciprofloxacin (Roy et al., 1983), the first three of which target cell wall synthesis. Tobramycin and ciprofloxacin inhibit protein synthesis and DNA replication, respectively. At the concentrations tested, PAO1 was either unaffected or only mildly affected (Figures 7A–E). Importantly, the ΔbamB mutant completely lost its viability in response to treatment with vancomycin, ceftazidime, and ampicillin (Figures 7A–C). The viability of the ΔfabY mutant was also decreased about 100~1,000 fold in response to the same treatment (Figures 7A–C). The degree of growth inhibition in these two mutants was not as great when treated with tobramycin and ciprofloxacin (Figures 7D,E). Collectively, these results suggest that (i) the ΔbamB mutant is exceptionally sensitive to cell wall-targeting antibiotics and, therefore, (ii) BamB may be an effective target in the elimination of P. aeruginosa.

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.