Characterization of Pseudomonas aeruginosa from subjects with diffuse panbronchiolitis

ABSTRACT

Diffuse panbronchiolitis (DPB) is a rare, idiopathic inflammatory disease primarily diagnosed in East Asian populations. DPB is characterized by diffuse pulmonary lesions, inflammation of the respiratory bronchioles, and bacterial infections of the airway. Historically, sputum cultures reveal Pseudomonas aeruginosa in 22% of DPB patients, increasing to 60% after 4 years from disease onset. Although DPB patients have a known susceptibility to respiratory P. aeruginosa infections, as is observed in other chronic lung diseases such as cystic fibrosis (CF), the characterization of DPB P. aeruginosa strains is limited. In this study, we characterized 24 strains obtained from a cohort of DPB patients for traits previously associated with virulence, including growth, motility, antibiotic susceptibility, lipopolysaccharide structure, and genomic diversity. Our cohort of DPB P. aeruginosa strains exhibits considerable genomic variability when compared with isolates from people with cystic fibrosis chronically colonized with P. aeruginosa and acute P. aeruginosa infection isolates. Similar to CF, DPB P. aeruginosa strains produce a diverse array of modified lipid A structures. Antibiotic susceptibility testing revealed increased resistance to erythromycin, a representative agent of the macrolide antibiotics used to manage DPB patients. Differences in the O-antigen type among P. aeruginosa strains collected from these different backgrounds were also observed. Ultimately, the characterization of DPB P. aeruginosa strains highlights several unique qualities of P. aeruginosa strains collected from chronically diseased airways, underscoring the challenges in treating DPB, CF, and other obstructive respiratory disease patients with P. aeruginosa infections.

IMPORTANCE

Diffuse panbronchiolitis (DPB), a chronic lung disease characterized by persistent P. aeruginosa infection, serves as an informative comparator to more common chronic lung diseases, such as cystic fibrosis (CF). This study aimed to better address the interplay between P. aeruginosa and chronically compromised airway environments through the examination of DPB P. aeruginosa strains, as existing literature regarding DPB is limited to case reports, case series, and clinical treatment guidelines. The evaluation of these features in the context of DPB, in tandem with prevailing knowledge of P. aeruginosa strains collected from more common chronic lung diseases (e.g., CF), can aid in the development of more effective strategies to combat respiratory P. aeruginosa infections in patients with chronic lung diseases.

INTRODUCTION

Diffuse panbronchiolitis (DPB) is a rare, idiopathic, and severe obstructive pulmonary disease first recognized in Japan in the 1960s (1–3). Almost exclusively diagnosed in East Asian populations, DPB is characterized by bilateral lung disease with diffuse yellow nodules and inflammation of the respiratory bronchioles. Common symptoms include chronic sinusitis, cough, purulent sputum, breathlessness, wheezing, and weight loss (1–3). Although DPB is likely underdiagnosed worldwide due to its non-specific presentation, the mean age of diagnosis is 40 years old, corresponding with symptom onset (2). The cause of DPB is unclear; however, correlations between specific human leukocyte antigen (HLA) types and the development of DPB have been identified (1–10). HLA-B54, in Japanese populations, and HLA-A11, in Korean populations, are regarded as strong predictors of developing DPB (1–5). Additionally, polymorphisms in two novel mucin-like genes, termed panbronchiolitis-related mucin-like 1 and 2, are also associated with the onset of DPB (6, 7). Another study identified DPB-associated polymorphisms in MUC5B, a mucin gene, suggesting that alterations in mucins may contribute to pathogenesis (8). Other factors have been evaluated as well, such as lymphocyte activity (e.g., elevated CD8+ cell activity in the airway lumen of DPB patients), human beta-defensin regulation (e.g., antimicrobial peptides involved in innate immunity), and environmental cofactors (9, 10).

Following disease onset, symptom manifestations leave DPB patients vulnerable to bacterial infection within the lungs (1–3, 11–14). As a result of excess mucus production in the airway, DPB patients experience bronchiectasis and obstructive lung pathology (1–3, 8). This provides opportunistic bacteria, such as Pseudomonas aeruginosa, an ideal environment for colonization (1–3). P. aeruginosa is a Gram-negative bacterium known to infect individuals with immune-dysregulated conditions by utilizing a wide array of virulence mechanisms to evade immune recognition (15–20). Sputum cultures reveal the presence of P. aeruginosa in 22% of DPB patients, increasing in frequency to 60% after 4 years from disease onset. Furthermore, the 10-year survival rate of P. aeruginosa-infected patients is 12%, compared with 73% for uninfected patients (2, 3). Importantly, macrolide antibiotic therapies were used to treat chronic DPB lung infections at sub-bactericidal concentrations, which resulted in a nearly 30% improvement in DPB survival rates 5 years post-prognosis (21, 22). The success of macrolide therapy in DPB inspired the usage of these antibiotics to treat chronic lung infections in other conditions, such as cystic fibrosis (CF) and bronchiectasis (23, 24). DPB is a rare disease—11 cases per 100,000 persons were reported in Japan in the 1980s—but current disease rates are unknown due to limited epidemiologic surveillance (2). Access to P. aeruginosa strains from DPB patients is limited, and studies characterizing these P. aeruginosa strains in DPB have not yet been performed. Acknowledging both the susceptibility of DPB patients to P. aeruginosa infections and the correlation between P. aeruginosa infection and accelerated pulmonary decline in DPB, this study aimed to characterize 24 DPB P. aeruginosa strains for features known to enable chronic infection and impact the severity of other obstructive respiratory diseases, such as CF. Consequently, features such as swim/swarm motility, in vitro growth, lipid A structure, antibiotic susceptibility, genomic diversity, and O-antigen distribution were evaluated for each DPB P. aeruginosa strain.

RESULTS

Genetic distinctions exist between DPB, CF, and environmental P. aeruginosa strains

All DPB P. aeruginosa strains were initially subject to microbial identification via MALDI Biotyper analysis and confirmed as P. aeruginosa (Table S2). Following P. aeruginosa identification, phenotypic attributes of the DPB P. aeruginosa strains were observed. Differences in gross phenotypes (e.g., color, transparency, colony size, colony morphology, and mucoidy) were identified (Fig. S1).

To evaluate DPB P. aeruginosa genomic variation, whole genome sequencing of each strain was conducted. Phylogenetic relationships between DPB, CF, non-CF bronchiectasis, and environmental P. aeruginosa strains were assessed via comparative genomics, as previously described (25). Genome assembly and comparative analysis of the CF, non-CF bronchiectasis, and regional and environmental P. aeruginosa strains were previously reported and are included as comparators (25–28). DPB P. aeruginosa strains do not form a single defined genomic clade when compared with strains obtained from chronic CF lung infections, acute infections, non-CF bronchiectasis, and environmental strains (Fig. 1A). There are some P. aeruginosa strains from DPB that do appear to exhibit clonal relationships with the following isolates being adjacent to each other in the phylogeny: BE110, BE113, and BE117; BE120, BE127, and BE133; BE125 and BE131; BE108, BE109, and BE130; BE121 and BE134; and BE119 and BE132 (Fig. 1A). These closely related isolates differ by on average 334 snps (SD 437 snp, range: 32–1111 snps), whereas the non-closely related strains differ on average by 24899 snps (SD 9427 snp, range: 6472–42862) snps). When the gene content of the P. aeruginosa strains from DPB was compared with that of P. aeruginosa strains from pulmonary isolations in Asia (Table S1), there were only 10 genes that were overrepresented among the DPB strains compared with other clinical isolates. The majority of these isolates are currently annotated as hypothetical, with the remainder being potentially surface-exposed (Table S3). Considering the genomic diversity of the examined strains, this lack of conserved DPB genes is not surprising. Although limited clinical information exists regarding P. aeruginosa collection from this DPB patient cohort, it is possible that these clonal DPB P. aeruginosa strains were isolated from the same patient, a pattern also observed in CF P. aeruginosa strains (29).

Fig 1.

DPB P. aeruginosa strains are genomically distinct and have variable O-specific antigens. (A) Phylogenetic tree depicting the genetic relationships between the DPB P. aeruginosa strains and P. aeruginosa strains from diverse clinical and geographic origins. DPB strains are in bold and highlighted in yellow. Outside of the inferred phylogeny, each strain is shown in the inner ring as the clinical sample of the disease of origin, and the outer ring indicates the predicted O-antigen of each strain. Legends for each of these criteria are included. The phylogenies were inferred using the Northern Arizona SNP Pipeline(NASP) (30) and were visualized using the iTOL (v6.7.5) program (https://itol.embl.de/) (31). (B) O-antigen distribution of the DPB P. aeruginosa strains relative to reference, acute infection isolates, non-CF bronchiectasis, early CF, and late CF P. aeruginosa strains. Genomes of comparator P. aeruginosa strains were reported previously (25–28).

Heterogeneity of DPB P. aeruginosa strain phenotype, motility, and in vitro growth

As motility plays an important role in P. aeruginosa virulence, flagellar swimming and type IV pili swarming motility were assessed in the DPB P. aeruginosa isolates. Variable motility was observed among the DPB P. aeruginosa samples, where some strains displayed relatively high motility function with average motility diameters over 15 mm, whereas others were non-motile (Table S4).

In vitro growth of the DPB P. aeruginosa strains was similarly diverse, ranging from fast to slow-growing strains. Bacterial growth curves were conducted to quantify differences in growth rate for each DPB P. aeruginosa strain grown in lysogeny broth (LB). The doubling times of the DPB P. aeruginosa samples spanned from 2.0 hours to 22.7 hours, as each strain displayed a unique growth ability (Fig. S2). The doubling time of PAO1 was 3.0 hours. No apparent relationship was observed between bacterial growth rate and other tested phenotypes (i.e., motility).

Antibiotic susceptibility profiles of DPB P. aeruginosa strains are non-uniform

The antibiotic susceptibility profile of each DPB P. aeruginosa strain was identified using Kirby-Bauer disk diffusion assay (32). Twelve antibiotics were tested for each DPB P. aeruginosa strain, including first-line beta-lactams, aminoglycosides, macrolides, and polymyxins (Table 1). Similar to CF P. aeruginosa strains, the DPB P. aeruginosa strains varied in antibiotic susceptibility (Table 1) (17, 18, 33, 34). Uniform antibiotic susceptibility was only observed in response to polymyxin B and colistin (Table 1). Conversely, uniform antibiotic resistance was only observed in response to rifampicin and meropenem-vaborbactam (Table 1). Over 90% of the DPB P. aeruginosa strains were also resistant to ampicillin-sulbactam, trimethoprim, and erythromycin (Table 1). Susceptibility of the DPB P. aeruginosa strains ranged from as low as 25% (3/12) to as high as 58.3% (7/12) of all antibiotics tested.

TABLE 1.

Antibiotic susceptibility data of P. aeruginosa DPB strains^a,b

Antibiotic	BE107	BE108	BE109	BE110	BE111	BE112	BE113	BE114	BE115	BE117	BE118	BE119
Rifampicin (5 µg)	0	0	0	0	0	10	0	0	0	0	0	0
	R	R	R	R	R	R	R	R	R	R	R	R
Tetracycline (30 µg)	25	25	13	25	16	15	13	46	14	24	20	13
	S	S	I	S	S	S	I	S	I	S	S	I
Ampicillin/Sulbactam (20 µg)	15	0	0	0	0	12	0	0	0	0	0	0
	I	R	R	R	R	I	R	R	R	R	R	R
Polymyxin B (300 µg)	20	20	17	18	22	18	15	18	15	15	17	23
	S	S	S	S	S	S	S	S	S	S	S	S
Aztreonam (30 µg)	50	20	0	16	46	30	13	23	30	35	37	0
	S	I	R	I	S	S	R	S	S	S	S	R
Trimethoprim (5 µg)	0	0	0	0	0	12	0	0	0	0	0	0
	R	R	R	R	R	I	R	R	R	R	R	R
Tobramycin (10 µg)	26	26	20	0	33	27	0	35	15	0	23	34
	S	S	S	R	S	S	R	S	S	R	S	S
Erythromycin (15 µg)	0	0	0	0	0	28	0	25	0	0	0	0
	R	R	R	R	R	S	R	S	R	R	R	R
Meropenem (10 µg)	50	37	32	40	0	44	33	15	0	25	20	0
	S	S	S	S	R	S	S	I	R	S	S	R
Ceftazidime (30 µg)	20	13	0	13	0	10	0	0	0	0	0	0
	S	R	R	R	R	R	R	R	R	R	R	R
Meropenem/Vaborbactam (30 µg)	0	0	0	0	0	0	0	0	0	0	0	0
	R	R	R	R	R	R	R	R	R	R	R	R
Colistin (10 µg)	19	20	18	19	23	19	17	21	15	22	21	24
	S	S	S	S	S	S	S	S	S	S	S	S

^a Tabulated summary of the Kirby-Bauer disk diffusion assay results for the DPB P. aeruginosa strains. Each numerical value represents the diameter of the zone of inhibition in millimeters. Diameter measures were utilized to define the DPB P. aeruginosa resistance profiles.

^b R = Resistant, I = Intermediate, S = Susceptible.

O-antigen type and expression are variable in DPB P. aeruginosa strains

Differences in O-antigen type among DPB, CF, and environmental P. aeruginosa strains were additionally assessed via comparative genomics (using PAst) and serotyping (Fig. 1B) (35). The bioinformatic approach (i.e., PAst) assessed genetic components that encode O- specific antigen type, whereas the serologic approach allowed us to determine if O-antigen was expressed. This comprehensive approach was needed as many CF strains do not express full-length O-antigen (i.e., rough lipopolysaccharide; LPS), compared with acutely infectious strains that do express full-length O-antigen (i.e., smooth LPS) (36).

Although multiple O-antigen types were observed in our DPB cohort, the majority of DPB P. aeruginosa strains (6/24) were found to genetically encode the O6 type, a trend also observed among the 21 early CF P. aeruginosa strains (12/21) (Fig. 1B). The O6 serotype was also the most prevalent among the collections of clinical P. aeruginosa strains from respiratory sources in Japan (8/36) and China (9/28).(26, 27) The O1 type was the most prevalent among the late CF cohort (3/8) (Fig. 1B). DPB P. aeruginosa strains encode O2 or O16 and O7 or O8 O-antigen, which are not observed in our cohort of CF P. aeruginosa strains (Fig. 1B). In total, O1, O6, and O11 O-antigen types were the only types present in all of P. aeruginosa collection backgrounds examined (Fig. 1B).

For the DPB P. aeruginosa strains, LPS was isolated and O-antigen expression was determined using western immunoblot with three pools of O-antigen sera. Similar to P. aeruginosa obtained from CF chronic lung infection, a minority of DPB strains express O-antigen (5/24), suggesting the pressures that underlie a loss of O-antigen expression are similar between CF and DPB strains (Table S5).

Gene-level bioinformatic approach to assess for genomic basis of strain variability

We evaluated sequence divergence in other genes important for pathogenic success to better understand how DPB P. aeruginosa strains adapt to the human airway during chronic infection. We took a bioinformatic approach using a large-scale BLAST score ratio (LS-BSR) analysis. Notably, there are eight P. aeruginosa strains with sequence divergence in genes in the arn locus involved in 4-amino-L-arabinose addition to a terminal phosphate of lipid A of the glucosamine backbone (Fig. 2). These differences may explain the variability in polymyxin susceptibility, as 4-amino-L-arabinose modification of lipid A confers polymyxin resistance (37–39). Furthermore, sequence divergence in mucA, a negative regulator of alginate biosynthesis, was also observed, corroborating a mucoid phenotype (Fig. S1 and 2C).

Fig 2.

LS-BSR analysis of virulence and outer membrane protein genes. LS-BSR analysis was performed, where an LS-BSR score of 1 indicates 100% gene homology with reference sequence, and a score of 0 indicates no homology. P. aeruginosa reference gene sequences were obtained from a PAO1 consensus sequence (assembly accession GCF_000006765.1). Laboratory-adapted strains (PA14, PAK, PAO1, and PA7) and one Staphylococcus aureus strain was included in this analysis to serve as positive and negative controls, respectively. (A) Virulence genes were selected based on their role in P. aeruginosa pathogenesis (i.e., type-3 and type-6 secretion systems). (B) Outer membrane protein genes were selected based on their proposed role in bacterial survival and membrane homeostasis during infection. (C) Lipid A biosynthesis genes were selected based on their role in both lipid A synthesis and modification pathways.

When virulence genes were interrogated for sequence divergence, increased variability was observed, further confirming intra-strain heterogeneity. Interestingly, sequence diversity in hcp genes (hcpA, hcpB, and hcpC), involved in type-6 secretion system (T6SS) formation, was present in seven DPB strains (Fig. 2A). Sequence divergence in PA2840 (also known as deaD) was also observed for nine DPB strains. DeaD has RNA helicase activity that is important for the expression of type III secretion system (T3SS) genes involved in injecting bacterial effectors directly into host cells.

We also investigated outer membrane protein genes, which are crucial in understanding how P. aeruginosa interacts with its environment. Sequence divergence in oprB was observed in 11 strains (Fig. 2B). OprB is an outer membrane porin involved in glucose uptake, and its loss of function may suggest altered metabolism during chronic lung infection, a phenomenon proposed in P. aeruginosa chronic infection in CF (40)

DPB P. aeruginosa strains demonstrate lipid A structural diversity

P. aeruginosa obtained from the airways of people with CF demonstrates an altered lipid A structure, which can induce variable signaling through innate immune pattern recognition receptors (25, 41–43). Modification of lipid A structure likely contributes to the ability of P. aeruginosa to chronically infect CF patients; therefore, we determined if lipid A structural diversity is also present in this cohort of DPB P. aeruginosa strains (41–43). Fast lipid analysis technique (FLAT) and matrix-assisted laser desorption/ionization time-of-flight (MALDI-TOF) mass spectrometry (MS) were used to evaluate lipid A structural modifications. LS-BSR was used to identify mutations in lipid A biosynthesis genes (Fig. 2C).

Using this MALDI-TOF MS approach, DPB P. aeruginosa strains demonstrated lipid A structural heterogeneity similar to that observed in P. aeruginosa strains from people with CF (Fig. 3A and C) (41–43). The associated lipid A mass spectra are provided in Fig. S3. Peaks at m/z 1446 and m/z 1616 represent penta-acylated and hexa-acylated lipid A structures, respectively, and are boxed in purple (m/z 1446) and orange (m/z 1616) (Fig. 3A and B) (41, 42). Peaks at m/z 1684 are a result of PagP acyltransferase activity, a lipid A phenotype often seen in CF P. aeruginosa strains, as the increase in mass reflects the addition of palmitate (C16, m/z ∆238) (Fig. 3A) (44). Also observed in CF P. aeruginosa strains, the absence of peaks at m/z 1430, m/z 1446, and m/z 1462 reflect a lack of PagL (lipid A 3-O-deacylase) activity (Fig. 3A and C) (45).

Fig 3.

Lipid A structural variation present in P. aeruginosa strains obtained from subjects with diffuse panbronchiolitis. (A) Representative lipid A mass spectra of the DPB P. aeruginosa strain sample set. Canonical penta-acylated (M/z 1446) and hexa-acylated (m/z 1616) lipid A peaks are boxed in purple and orange, respectively. (B) Penta-acylated (m/z 1446, boxed in purple) and hexa-acylated (m/z 1616, boxed in orange) P. aeruginosa lipid A structures. (C) Tabulated summary of the DPB P. aeruginosa lipid A mass spectra documenting commonly observed P. aeruginosa peaks at m/z 1430, m/z 1446, m/z 1462, m/z 1600, m/z 1616, m/z 1632, m/z 1684, m/z 1700, and m/z 1854.

One of the DPB P. aeruginosa strains, BE107, exhibited a novel lipid A mass spectra peak configuration, with peaks observed at m/z 1656, m/z 1672, and m/z 1692 (Fig. 3A). Tandem mass spectrometry (MS/MS) and gas chromatography with flame-ionization detection (GC-FID) were used to further characterize this novel lipid A structure (Fig. S4). Possessing six acyl chains, the BE107 lipid A structure is distinct in its integration of a C16:1 cis-double bond about carbon 9 (Fig. S4C). To date, this lipid A structure has not been observed in any P. aeruginosa strain and highlights the structural plasticity of P. aeruginosa lipid A.

DISCUSSION

Diffuse panbronchiolitis (DPB) is a severe chronic obstructive lung disease that can result in respiratory failure and fatality when untreated (2, 3). Patients with DPB present clinically with several non-specific respiratory symptoms and bronchiolectasis (1–3). Recognizing both the susceptibility of DPB patients to P. aeruginosa infections and the correlation between P. aeruginosa infection and advanced pulmonary decline, this study aimed to better understand features of DPB P. aeruginosa strains that both enable chronic infection and confer virulence impacting patient outcomes in both DPB and other obstructive respiratory diseases, such as CF (2, 3). Although many gaps exist in the knowledge of DPB pathophysiology, phenotypic and genotypic assessment of DPB P. aeruginosa strains can be a novel comparator to P. aeruginosa strains collected from other chronic lung diseases, such as CF.

Comparative genomics confirmed variation among DPB P. aeruginosa strains relative to each other and P. aeruginosa from other collection backgrounds (Fig. 1). The DPB P. aeruginosa strains appeared to group into several small regions on the phylogeny, indicating pockets of genetic similarity among strains in the DPB patient cohort (Fig. 1A). One of these clusters defines a clonal relationship between BE109 and BE130 (Fig. 1A), suggesting that the alternate strain identifications for BE109 and BE130 (KUD 003–1 and KUD 003–2, respectively; see Table S2) may indicate that these strains were collected from the same patient at different times. The O-antigen distribution of DPB P. aeruginosa strains was also variable (Fig. 1A). Furthermore, DPB P. aeruginosa strains displayed O-antigen diversity greater than the environmental and CF P. aeruginosa strains, supporting the hypothesis that DPB P. aeruginosa strains are phenotypically and genetically unique. Importantly, the majority of DPB P. aeruginosa strains express rough LPS (lacking full-length O-specific antigen), a characteristic once thought to be unique to CF P. aeruginosa strains (36, 46). This suggests that selective pressures on P. aeruginosa lipid A structure and O-antigen expression, once thought to be specific to the CF airway, are present in other conditions of chronic airway infection. One role of full-length O-antigen at the host-pathogen interface is to protect the bacterium from killing by complement (47); therefore, we speculate that the lack of complement-mediated killing in the small airways allows for selection of rough LPS P. aeruginosa mutants in both CF and DPB.

Upon initial characterization, colony morphology revealed several of these DPB strains to be mucoid (Fig. S1), and this mucoid phenotype is corroborated by sequence divergence in mucA (Fig. 2). Mucoidy, defined by constitutive alginate or other extracellular polysaccharide production, is associated with CF P. aeruginosa strains and not often seen in other disease contexts (48–50); therefore, these data suggest that mucoid may arise secondary to chronic lung infection, independent of cystic fibrosis transmembrane conductance regulator (CFTR) dysfunction. Swim and swarm motility assays were performed to assess the retention of motility machinery in the DPB P. aeruginosa strains, as CF P. aeruginosa strains can have altered motility during chronic infection (51). Haze and swarm diameters were variable among the DPB P. aeruginosa strains, indicating a non-uniform motility function (Table S4). Perhaps these differences are a result of the varying DPB patient airway immune responses, as the expressions of flagella, type IV pili, and other motility machinery are associated with enhanced virulence and immune recognition via pattern recognition receptors (15).

The lipid A structural diversity of DPB P. aeruginosa strains not only emphasizes their heterogeneity but also reveals that DPB P. aeruginosa strains modify lipid A similarly to CF P. aeruginosa strains. Lipid A structural heterogeneity is observed in CF P. aeruginosa strains obtained during chronic lung infection; however, here, we also reveal lipid A structural variation is present in DPB P. aeruginosa strains (41–43). Genetic variation in key lipid A biosynthesis genes, such as htrB1, htrB2, pagL, pagP, lpxO1, and lpxO2, support these findings (Fig. 2) (41–43, 45). These data suggest that each DPB P. aeruginosa strain retains a specific lipid A structure, which may improve bacterial barrier function and immune evasion specific to its DPB patient of origin, ultimately enabling sustained chronic infection as observed in CF P. aeruginosa strains (41–43, 45).

Additional experimentation (MS/MS, GC-FID) elucidated the novel lipid A peak configuration displayed for BE107 (Fig. 3A; Fig. S4). The BE107 lipid A structure is unique, with its integration of a C16:1 cis-double bond about carbon 9 (Fig. S4C). This cis-double bond creates a kink in the C16 acyl chain, as the adjacent single bonds point in the same direction. Regarding three-dimensional orientation, we hypothesize that this kink induces spatial separation of the surrounding LPS molecules (41, 43, 52). Acknowledging that BE107 was one of the least resistant strains to the antibiotic panel tested, the supposed spacing of the outer membrane induced by the C16:1 cis-double bond kink may correlate with these findings, as antibiotics would theoretically confront less physical opposition when entering the bacteria (Table 1). Further experimentation identifying which lipid A modifying enzymes are involved in adding a C16:1 cis-double bond may be informative for future experimentation, as this acyl chain structure is not seen in canonical P. aeruginosa lipid A.

Variable antibiotic susceptibility profiles were revealed for the DPB P. aeruginosa strains (Table 1). Across the antibiotic panel tested, no DPB P. aeruginosa strains displayed identical susceptibility. Resistance to erythromycin, the main therapeutic agent used to treat DPB patients, was observed in roughly 91% of the DPB P. aeruginosa strains, suggesting that treatment of chronic P. aeruginosa infections can lead to antibiotic resistance, especially after months or years of infection and subsequent antibiotic exposure (Table 1). Noting that inhaled colistin has been used as a therapeutic approach to combat bacterial infection in CF patients, a similar approach may be promising to treat DPB patients with refractory P. aeruginosa infections as the DPB P. aeruginosa strains display uniform susceptibility to colistin (53).

The results of this study show that our cohort of 24 DPB P. aeruginosa strains exhibit considerable phenotypic and genotypic diversity, with similarities to both acute and chronic P. aeruginosa infections. This diversity of colonization sources underscores the challenges in treating P. aeruginosa infections in DPB, CF, and other obstructive respiratory diseases. Similarities observed between DPB and CF P. aeruginosa strains prove informative for future studies, as this work heightens the current understanding of the interplay between P. aeruginosa and chronic lung disease.

MATERIALS AND METHODS

Bacterial strains

Archived P. aeruginosa strains were obtained from the airways of subjects with diffuse panbronchiolitis (before the year 2000)—strain details are outlined in Table S1 (Dr. Samuel M. Moskowitz, University of Washington; currently Design Therapeutics). Strains were stored at −80°C in 25% glycerol. Strains were streaked on lysogenic broth (LB) agar utilizing a tertiary streak pattern and incubated at 37°C for 18 hours with inversion. Liquid cultures were inoculated utilizing single colonies and incubated at 37°C for 18 hours, shaking at 180 rpm. Due to the archived nature of these historic samples, no associated clinical data exist for this cohort of P. aeruginosa strains. Genomic sequences of other P. aeruginosa strains used in this study (from environmental, regional, and CF sources) are described previously (25–28, 54). Metadata for strains are included in Tables S1a and b, including GenBank accession numbers.

Fast lipid analysis technique

Lipid A structural analysis was performed, as previously described (55). A single colony was scraped onto a steel target plate in duplicate with a toothpick. FLAT extraction buffer (1 µL; 0.2 M anhydrous citric acid, 0.1 M trisodium citrate dihydrate) was pipetted over bacterial spots. The target plate was incubated at 100°C for 30 minutes, washed with ddH₂O and air-dried. Mass spectra were collected in negative-ion mode using a Bruker MicroFlex®. Mass spectra were analyzed using FlexAnalysis v3.4 software.

Whole-genome sequencing

The genomes of all strains analyzed in this study were sequenced, as previously described (56). Sample libraries were prepared with the Illumina DNA prep kit (cat. num. 20018705) and sequenced on the Illumina NextSeq 2500, producing 2  ×  151 bp paired-end reads. All software was used with default values. Raw sequencing reads were filtered to remove contaminating phiX reads using BBDuk, one of the BBTools software suites (sourceforge.net/projects/bbmap/). The raw reads were also filtered to remove contaminating Illumina adaptor sequences and quality-trimmed using Trimmomatic v. 0.36 (57). The resulting filtered reads were assembled using SPAdes v. 3.13.0 (58).

Phylogenomic analysis

Comparison was conducted using an in silico genotype (ISG) (59). Single nucleotide polymorphisms (SNPs) were detected relative to the reference P. aeruginosa strain PAO1 (GenBank accession #NC_002516.2) using the ISG that uses the NUCmer (v.3.22) program (60) for SNP detection. SNP sites that were identified in all analyzed genomes were concatenated and used to construct a maximum likelihood phylogeny using RAxML (v7.2.8) (61). All phylogenies were visualized using the iTOL (v6.7.5) program (https://itol.embl.de/) (61).

SNP distances were determined using snp-dists with default values (https://github.com/tseemann/snp-dists; Data set 1).

Genome-based comparisons

Roary/Scoary: All genomes used in subsequent analyses (Table S1a/S1b) were annotated via Prodigal (62) to generate de novo CDS and then PROKKA v1.14.6 (63). The resulting general feature formats (GFF) were analyzed to identify core and accessory genes using Roary v3.13.0 (64) and Scoary version 1.6.16 (65).

P. aeruginosa O-antigen sequence type determination

O-antigen types were identified using P. aeruginosa serotyper (PAst) program, as previously described (35).

LPS purification and visualization

LPS was purified as previously described (66) with slight modifications. Briefly, P. aeruginosa overnight cultures grown on LB agar and were resuspended in PBS and normalized to OD₆₀₀ = 0.5 (~1e9 CFU / mL). Washed bacteria were then pelleted by centrifugation and resuspended in 200 µL sodium dodecyl sulfate (SDS) sample buffer (BioRad) and briefly boiled. Samples were incubated with 2.5 mL DNaseI (10 mg/mL), and 2.5 mL RNase (10 µg/mL) for 30 minutes at 37°C, followed by 3 hours of incubation with 10 mL Proteinase K (10 mg/mL) at 59°C. Ten microliters of purified LPS were loaded and separated on a 12% Tris-glycine gel (BioRad) using SDS-PAGE and then visualized by western blotting. Samples were probed with P. aeruginosa serogroup rabbit polyvalent antibodies (Denka Seiken, Tokyo Japan) followed by incubation with anti-rabbit secondary antibodies conjugated to horseradish peroxidase (HRP) (Sigma). Reactions were visualized by the addition of Clarity Max Western ECL Substrate (BioRad).

Large-scale BLAST score ratio analysis

Genomes of DPB P. aeruginosa strains were compared using LS-BSR analysis, as previously described (67). Gene sets were chosen to investigate the genetic variation of DPB P. aeruginosa strains and differences in bacterial functions, such as virulence or membrane maintenance. Gene sequences were obtained using the reference P. aeruginosa strain PAO1 (GenBank accession no. NC_002516.2). Heat maps of the LS-BSR values were generated in RStudio using ggplot2.

SUPPLEMENTAL MATERIAL

The following material is available online at https://doi.org/10.1128/spectrum.00530-24.

Figure S1. spectrum.00530-24-s0001.pdf.

Photographs of the DPB P. aeruginosa strains utilized in this study.

Figure S2. spectrum.00530-24-s0002.docx.

In vitro growth curves.

Figure S3. spectrum.00530-24-s0003.docx.

Lipid A mass spectra of each DPB P. aeruginosa strain in the sample set (BE107-BE134).

Figure S4. spectrum.00530-24-s0004.docx.

Structural analysis of BE107.

Table S1. spectrum.00530-24-s0005.xlsx.

Strains used in this study.

Supplemental tables. spectrum.00530-24-s0006.docx.

Tables S2 to S5.

Supplemental materials. spectrum.00530-24-s0007.docx.

Supplemental methods.

ASM does not own the copyrights to Supplemental Material that may be linked to, or accessed through, an article. The authors have granted ASM a non-exclusive, world-wide license to publish the Supplemental Material files. Please contact the corresponding author directly for reuse.