PPARγ is reduced in the airways of non-CF bronchiectasis subjects and is inversely correlated with the presence of Pseudomonas aeruginosa

Background Chronic airway inflammation in conditions such as cystic fibrosis (CF) and non-CF bronchiectasis is characterised by a predominant neutrophilic inflammatory response, commonly due to the presence of pathogenic bacteria such as Pseudomonas aeruginosa. We hypothesised that down-regulation of the anti-inflammatory nuclear transcription regulator peroxisome proliferator-activated receptor gamma (PPARγ in non-CF bronchiectasis subjects may explain why this exuberant neutrophilic inflammation is able to persist unchecked in the inflamed airway. Methods PPARγ gene expression was assessed in bronchoalveolar lavage fluid (BAL) of 35 macrolide naïve non-CF bronchiectasis subjects and compared with that in 20 healthy controls. Human RNA was extracted from pelleted BAL and PPARγ expression was determined by reverse-transcription quantitative PCR. Bacterial DNA was extracted from paired induced sputum and total bacterial load was determined by 16S rRNA qPCR. Quantification of individual bacterial species was achieved by qPCR. Results PPARγ expression was lower in subjects with non-CF bronchiectasis compared with healthy control subjects (control: 1.00, IQR 0.55–1.44, n = 20 vs. Bronchiectasis: 0.49, IQR 0.12–0.89; n = 35; p<0.001, Mann-Whitney U test). This lower PPARγ expression correlated negatively with Pseudomonas aeruginosa (r = -0.53, n = 31; p = 0.002). No significant association was seen between PPARγ and total bacterial levels or levels Haemophilus influenzae. Conclusion PPARγ is expressed in low levels in the airways of non-CF bronchiectasis subjects, despite an aggressive inflammatory response. This low level PPARγ expression is particularly associated with the presence of high levels of P. aeruginosa, and may represent an intrinsic link with this bacterial pathogen.


Introduction
Non-cystic fibrosis bronchiectasis is a chronic pulmonary condition characterised by exuberant airway inflammation and damage caused by chronic airway infection and exacerbation [1]. Infection with the opportunistic pathogen Pseudomonas aeruginosa is associated with accelerated lung function decline [2] and increased mortality [3,4]. Why P. aeruginosa colonised patients have poorer outcomes is unclear, but is likely to reflect an imbalance between the pro-inflammatory and anti-inflammatory nature of the host-pathogen relationship.
In cystic fibrosis (CF), the expression of anti-inflammatory nuclear transcription regulator peroxisome proliferator-activated receptor gamma (PPARγ is low compared to healthy controls, and appears to be reduced further in the presence of P. aeruginosa [5]. PPARγ binds to exogenous lipid-like ligands and then interacts with specific PPAR response elements on nuclear DNA to modulate gene expression [6]. Inhibition of PPARγ has been shown to result in a pro-inflammatory phenotype by interfering with its inhibitory action on the target genes of NFκB [6], supportive of the overall anti-inflammatory effect of PPARγ. The mechanism by which PPARγ production is suppressed in CF is not known. However, in vitro studies have demonstrated that P. aeruginosa acyl-homoserine lactones (AHLs)-lipid-like exoproducts of P. aeruginosa-suppress expression of PPARγ in various cell types, including epithelial and endothelial cells [5,7,8]. Bacterial AHLs have been shown to enter, and function in, mammalian cells [9][10][11]. One way that P. aeruginosa might interfere with the host immune response is through the suppression of PPARγ function, through inhibition of PPARγ transcription or translation. Despite the availability of commercial PPARγ agonists, which might provide clinical benefit, there have been no reported investigations of the role of PPARγ in non-CF bronchiectasis.
We examined the expression of PPARγ in the airways of macrolide naïve non-CF bronchiectasis patients who were recruited as part of a prospective randomised controlled trial [12]. Further, we explored the relationship between PPARγ expression and the presence of important colonising pathogens, such as P. aeruginosa.

Study population
The study was approved by the Mater Human Research Ethics Committee and all participants provided written informed consent. Subjects in this study were participants in the BLESS trial [12]. A subgroup of 41 participants underwent bronchoscopic sampling at baseline, along with 20 healthy controls. BLESS participants were recruited for this subgroup sequentially at the same time as they were recruited to the parent study and were only excluded if they had a comorbidity that would preclude safe bronchoscopy and biopsy, or after the pre-specified number of participants had been reached. The current analysis is based upon the 35 bronchoscopy subjects and 20 healthy controls from whom sufficient bronchoalveolar lavage (BAL) sample a not-for-profit hospital and health services company. The funder provided support in the form of salaries for the authors listed above, but did not have any additional role in the study design, data collection and analysis, decision to publish, or preparation of the manuscript. The specific roles of these authors are articulated in the 'author contributions' section.
Competing interests: Please note that no authors on this manuscript have any relevant competing interests to declare, however Dr Lucy Burr, Dr Simon Bowler and Rebecca Keating are employed by Mater Misericordiae Health Services Ltd. This is a not-for-profit hospital and health services company. The funder provided support in the form of salaries for the authors listed above, but did not have any additional role in the study design, data collection and analysis, decision to publish, or preparation of the manuscript. The specific roles of these authors are articulated in the 'author contributions' section. This does not alter our adherence to PLOS ONE policies on sharing data and materials.
was obtained for gene expression analysis. Table 1 shows the baseline clinical characteristics of these two groups. Six subjects in the non-CF bronchiectasis group did not have sufficient BAL sample for analysis, however, there was no significant difference between this group and the original 41 patients recruited for bronchoscopy in the trial (data not shown).

BAL and sputum processing
Gauze-filtered BAL was separated into sterile specimen jars and transported to the laboratory on ice. One portion was sent for cell counting. BAL was then further divided into 1 mL aliquots. A minimum of 2 aliquots were centrifuged at 500 x g for 5 min at 4˚C to pellet large cellular debris. The pellet was then frozen rapidly and stored at -80˚C prior to RNA extraction. Matched, induced sputum was frozen rapidly and stored at -80˚C until subsequent molecular analysis.

Cell counts
BAL was processed within 60 minutes of collection. After filtration, the sample was centrifuged, the supernatant aspirated and the cell pellet resuspended in PBS for cell counting in the presence of trypan blue and both Blue and Red Rapi-Diff cell stains.

DNA extraction
DNA extraction for both PCR and microbiome analysis was performed on 200 μL portions of paired induced sputum using a combination of physical disruption and a phenol/chloroformbased methodology as described previously [13].

RNA extraction and cDNA synthesis
RNA extraction from BAL was performed using an on-column method as per the manufacturer's instructions for RNeasy RNA Extraction Kit (Qiagen, Netherlands). cDNA synthesis for the non-amplified RNA samples was performed using the iScript cDNA synthesis kit (Bio-Rad, California USA), as per the manufacturer's instructions.

Quantitative PCR for the determination of bacterial load
Total bacterial load was determined in paired sputum samples using a Taqman assay directed against a conserved region of the 16S rRNA gene, as described previously [14]. P. aeruginosa levels were determined using a Taqman assay, in which a 117 bp region between positions 330 to 447 of the P. aeruginosa oprL gene was amplified, as described previously [13]. H. influenzae density was determined using a Taqman assay, in which a 90-bp region between positions 518 to 608 of the H. influenzae hel gene was amplified [13].

Clinical correlates
Lung function and assessment of clinical measures were performed at enrolment as described previously [12]. Spirometry was performed pre-and post-inhalation of salbutamol bronchodilator. Serum CRP was measured at each visit. A 24-hour sputum sample and a spontaneously expectorated sputum sample were collected at each visit for volume quantification, microbiology, and differential cell count.
Levels of PPARγ gene expression were assessed relative to airway immune cell counts. PPARγ gene expression was correlated to total neutrophil and macrophage levels in BAL fluid. No significant correlation was observed between total neutrophil count and PPARγ gene expression (r = -0.19, p = 0.329) (Fig 5A). However a weak, but significant positive correlation was observed between PPARγ gene expression and total macrophage count (r = 0.38, p = 0.044) (Fig 5B). P. aeruginosa load was then correlated with total macrophage levels, where there was a significant, negative correlation between macrophages and P. aeruginosa (r = -0.408, p = 0.023) (Fig 6). Agonising PPARγ has been implicated in the inhibition of stimulated mucin expression, potentially reducing sputum volume in subjects with excessive inflammation [15]. To investigate whether airway PPARγ gene expression was related to sputum amount, levels were correlated with 24 hour sputum weight. There was a trend to lower PPARγ gene expression with increasing sputum volume, suggesting that the more PPARγ expression, the less sputum produced. However, this relationship did not achieve statistical significance (r = -0.190, p = 0.276) (Fig 7A). This was also reflected with measurements of CRP (r = -0.137, p = 0.449). However, again, the association was weak and non significant (Fig 7B).
To determine whether the alterations in PPARγ were associated with important clinical outcomes, correlations with lung function (as measured by FEV1% predicted) were performed. There was no significant correlation between PPARγ expression and lung function as measured by FEV1 (r = 0.082, p = 0.225).

Discussion
Our aim was to assess the relevance of PPARγ in the context of P. aeruginosa colonisation in non-CF bronchiectasis. PPARγ is an anti-inflammatory product of leukocytes, which has been shown to inhibit, or switch off, NFκB in response to lipids [6]. In doing so, PPARγ might help to limit exuberant inflammation within the context of suppurative lung disease. P. aeruginosa AHLs are lipid-like structures that readily interact with PPARγ. P. aeruginosa AHLs have been shown to reduce PPARγ gene expression in endothelial and other cell lines when exposed in vitro [8].
Low levels of PPARγ expression have been reported in CF airways and that PPARγ levels are lowest in patients colonised by P. aeruginosa [5]. Our data suggests reduced levels of PPARγ are also characteristic of the non-CF bronchiectasis airways, despite an aggressive inflammatory response. In keeping with the observations made in patients with CF, high airway levels of P. aeruginosa in patients with bronchiectasis are associated with low levels of PPARγ expression, suggesting an intrinsic link with this bacterial species. This is supported by data demonstrating that agonising PPARγ inhibits P. aeruginosa biofilm production in vitro [16], and is further supported by evidence suggesting that clearance of macrophages infected with P. aeruginosa is enhanced with activation of PPARγ [17]. Whilst further work would be necessary to confirm these findings, our data supports a specific bacterial interaction between P. aeruginosa and PPARγ.
We assessed PPARγ gene expression in cells derived from the airway fluid of healthy and non-CF bronchiectasis subjects. These samples comprise mainly immune cells, such as neutrophils, eosinophils and macrophages, with minimal contribution of epithelial cells. Airway measures of PPARγ therefore likely originate from luminal immune cells in this context. In order to determine whether there was any correlation with PPARγ expression and airway immune cells, PPARγ expression was correlated with absolute and percentage immune cell counts. Interestingly there was no correlation with airway neutrophils, but there was a weak but significantly positive correlation with airway macrophages. This suggests that macrophages may be the primary source of PPARγin non-CF bronchiectasis airways. Interestingly, the more P. aeruginosa present in the airways, the fewer macrophages were found. There is evidence that the P. aeruginosa AHL molecule 3-oxo-12-HSL accelerates apoptosis in macrophages [18,19], suggesting that, in the presence of P. aeruginosa, there is the potential for fewer, less active macrophages and less PPARγ.
In order to determine whether PPARγ levels could predict important clinical outcomes, PPARγ, measured at baseline, was correlated with clinical outcomes. There were no significant correlations with specific clinical outcomes, however, there was a trend to decreased sputum amount with increasing PPARγ expression. This trend is consistent with previous literature implicating PPARγ agonism with decreased mucus hypersecretion [15,20].
Our study had a number of limitations. Firstly, we focused on levels of PPARγ gene expression, rather than on its functional activity. However, previous studies have shown a strong correlation between PPARγ gene expression and protein activity suggesting gene expression is an accurate reflection of activity [21]. Previous studies assessing airway PPARγ expression in association with disease states have also arrived at similar conclusions [5,22]. Another limitation was that the healthy control group was significantly younger than the diseased group, which can affect PPARγ [23] and could explain the differences between the groups. The healthy group, by its nature, did not have any airway disease, meaning that their lung function was better, they did not have colonising pathogenic airway bacteria, and had no evidence of systemic inflammation, all of which necessarily confounded the analysis. While we can say that PPARγ expression is lower in the diseased group compared with a non-diseased group, the specific reason for that difference cannot be reliably answered from the available data. Within the diseased group there were significant correlations between PPARγ expression and specific disease markers such as presence of P. aeruginosa and abundance of macrophages. Correlation is not causation however. It must be also noted that in the disease cohort used in this study, there was a high prevalence of patients colonised by P. aeruginosa (as detected by qPCR). This rate is higher than previously found in the non-CF bronchiectasis population, [12] therefore further studies characterising P. aeruginosa levels in other non-CF bronchiectasis cohorts are required.
Nonetheless, the data presented here demonstrates for the first time that lower levels of PPARγ are present in the airways of non-CF bronchiectasis subjects, despite an aggressive inflammatory response. This low level PPARγ expression is particularly associated with the presence of P. aeruginosa, hinting at a possible mechanism involving this bacterial species. This is similar to data acquired from patients with CF [5] and supports the notion that low levels of traditional anti-inflammatory molecules may also play a role in disease due to the inability of the host to rein in exuberant inflammation. Notably, PPARγ agonists are commercially available treatments for Type 2 Diabetes Mellitus and may offer a potential therapeutic agent to aid in the treatment of inflammatory airways diseases.