Background
Respiratory infections are one of the leading causes of mortality worldwide and smoking is considered a risk factor for developing upper and lower respiratory infections [
1‐
3]. In addition, exposure to cigarette smoke is associated with increased risk of airway bacterial colonization compared to non-smokers [
4,
5].
The pathogenesis of infectious airway disease caused by cigarette smoke is complex. Smoking damages airway epithelia, increases mucus production, decreases mucociliary clearance, impairs cell immunity and the production of antimicrobial peptides and proteins (AMPs) in the airway [
6‐
11]. Besides impairing the host response to infectious challenges, cigarette smoke can also affect bacterial virulence [
12].
The airway surface liquid (ASL) is a layer of fluid covering the airways that is a first line of defense responsible for antimicrobial activity against airborne pathogens. One of the most abundant AMPs present in the ASL is lactoferrin, a bacteriostatic protein that chelates iron, which is required for bacteria to grow and form biofilms [
13]. The impairment of AMP activity plays a fundamental role in the origin of infectious lung diseases. Several factors can alter the activity of AMPs such as decreased pH, increased ionic strength especially due to divalent cations such as iron, magnesium, and calcium [
14‐
18].
Tobacco contains particulate matter and multiple chemicals that can potentially alter the iron homeostasis in the lungs [
19,
20]. Since iron promotes bacterial growth and biofilm formation and inhibits AMP activity against pathogens, we hypothesized that the ASL from smokers would grow more bacteria and develop more biofilm compared to non-smokers [
21‐
23]. Repeated respiratory infections in smokers influences the development of chronic inflammation and lung function decline leading to chronic obstructive pulmonary disease (COPD) [
24,
25].
We have chosen
Staphylococcus aureus and
Pseudomonas aeruginosa as models to study relevant culturable airway pathogens. They are both representative airway Gram positive and negative pathogens.
S. aureus colonizes the nostrils of smokers in a higher prevalence than non-smoking population [
26]. This carrier state has been associated with an increased risk of lethal infections by endogenous strains [
27].
Pseudomonas aeruginosa is a pathogen present in the airways of patients with COPD at both baseline and during exacerbations [
28]. This organism is more prevalent in severe COPD and is associated with poor clinical outcomes in hospitalized patients [
29,
30]. Therefore, it is relevant to determine mechanisms implicated in increased risk of respiratory infections in smokers.
Sampling ASL is extremely challenging as it is present at a very small volume in the lungs [
31]. Therefore, we have used bronchoalveolar lavage as a surrogate of ASL. Bartlett et al. examined the protein composition of bronchoalveolar lavage (BAL) and ASL from new born pigs and found they had 514 protein in common, including AMPs such as lactoferrin, lysozyme and cathelicidins [
32]. We challenged the BAL from smokers and non-smokers with bacteria and assessed growth and biofilm formation. We also investigated the effect of supplementing iron, cigarette smoke extract, and lactoferrin to explore the role of iron in the bactericidal and anti-biofilm properties of the airway.
Methods
Human BAL collection and processing
We used biobank-stored BAL samples from non-smokers (
n = 11) and smokers (n = 11) from the study Human Lung Responses to Respiratory Pathogens that aimed to study the relationship of vitamin D levels and the innate defense of the lung against inhaled bacteria. Participants were selected if they were between the age 18–60, if smoker, FEV1 had to be more than 60% of predicted. Participants characteristics were similar between smokers and non-smokers (Table
1). Participants were excluded if they had history of positive tuberculin test or tuberculosis, pneumonia, recent airway infections, antibiotic use or vaccination, were taking vitamins or medications with selected exceptions, were pregnant, breast-feeding, had asthma, diabetes, heart disease or allergy to lidocaine. More detailed inclusion and exclusion criteria for this study were previously published [
33]. The collection was approved by the Institutional Review Board at the University of Iowa (IRB# 200607708). BAL collection was performed as previously described [
33]. Briefly, after subjects signed an informed consent, and pregnancy was ruled out using a urine test, participants were premedicated with atropine (0.6 mg intramuscularly (IM)), and either morphine (10 mg IM) or meperidine (12.5–25 mg IM). The airways were locally anesthetized using 2–4% lidocaine. A pulmonary physician performed the bronchoscopies by a standard procedure using a flexible bronchoscope (model P160 or P180; Olympus) at the University of Iowa Hospitals and Clinics (Iowa City, Iowa, USA). Under clinical monitoring, five BAL samples (20 mL) were suctioned into a trap container from three segments of the lungs. The liquid in the collection traps was transferred into conical 50 mL centrifuge tubes. Tubes were centrifuged to separate cells from the rest of the BAL. The supernatant of the tubes was pooled into one Falcon® Cell Culture Flasks and stored at − 80 °C. The selection of flasks that was retrieved from the biobank were thawed at once on ice, aliquoted into working samples, and stored again at − 80 °C. Working samples were thawed and used once for every experiment to avoid multiple freeze-thaw cycles.
Table 1
Comparison of participant characteristics by smoking status
Age mean | 32.8 (11) | 37.7 (13) | 0.3881 |
Male gender (%) | 54 | 81 | NA |
Recovery rate (%) | 77.3 (10.5) | 70.7 (16.4) | 0.2542 |
Assessment of BAL bacteriostatic effect on bacteria
To assess bacterial growth, we used bioluminescent
Pseudomonas aeruginosa Xen 05 and
Staphylococcus aureus Xen 29 (Caliper Biosciences, USA). It has been reported that Relative Light Units (RLU) correlate closely with Colony Forming Units (CFU) [
16]. Briefly,
P. aeruginosa Xen 05 was cultured overnight in tryptic soy broth (TSB) and then subculture in iron-free media M9 (BD Difco™, USA) overnight at 37 °C, and then washed twice with Phosphate Buffered Saline without calcium and magnesium (PBS−/−). Thereafter, we combined 100 μL of BAL samples with 10 μL of bacteria (~ 5 × 10
5 CFU) in a 96-well plate (Optiplate-96, Perkin-Elmer, USA). We measured RLU (527 nm) 6 h after bacterial challenge as a surrogate of live bacteria.
S. aureus Xen 29 was cultured overnight in TSB. The next day we washed a subculture of mid-log phase bacteria, twice with PBS−/−. Because S. aureus does not grow in M9 iron-free media, we resuspended in minimal essential media (10 mM Sodium phosphate buffer, pH 7.4, 100 mM NaCl, 1% TSB). Thereafter, we combined 10 μL of BAL samples with 10 μL of bacteria (~ 5 × 105 CFU) in a Optiplate-96 and measured RLU 30 min after challenge as a surrogate for live bacteria.
To test the effect of lactoferrin supplementation on bacterial growth, we combined 10 μL of increasing concentrations of recombinant lactoferrin from human milk (10, 30, 100, or 300 μg/mL, final concentration) (Sigma-Aldrich) or phosphate buffer control with 90 μL of the BAL for 30 min at 37 °C. Subsequently, we added 10 μL of bacteria (~ 5 × 105 CFU) and measured RLU at 30 min for S. aureus and 6 h for P. aeruginosa after bacterial challenge.
We investigated the formation of biofilms in the BAL from smokers and non-smokers using two methods. For
S. aureus we used a microtiter dish biofilm formation assay as previously described [
34]. Briefly, we combined 10 μL of BAL with 190 μL of bacteria Xen 29 (~ 5 × 10
5 CFU) suspended in minimal media in a 96 well plate. After 48 h, we extracted the liquid, washed the wells with distilled water, stained the biofilm using crystal violet, removed the excess crystal violet with distilled water, dissolved the stain with 30% acetic acid and read the OD
600 of every well.
To test the effect of lactoferrin supplementation on the smokers BAL biofilm formation, we combined 40 μL of PBS containing varying concentrations of lactoferrin (30, 300, and 1000 μg/mL) or PBS control with 150 μL of BAL samples from smokers and non-smokers and assessed biofilm formation. This mixture was incubated for 30 min at 37 °C. Thereafter, we added 10 μL of Xen 29 S. aureus (~ 1.5 × 104 CFUs) to the 96-well plate and incubated for 48 h at 37 °C. Biofilm formation was assessed at 48 h as previously described.
For
P. aeruginosa we grew PAO1 overnight in TSB. The overnight culture was washed twice with M9 and cultured for 4 h in M9. Subsequently, 50 μL of the bacterial solution (~ 1.5 × 10
7 CFUs) was added to a 96 well plate containing 100 μL of BAL with a coverslip mounted perpendicular to the bottom of each well. After 48 h, the coverslips were stained using concanavalin-A conjugated with Texas-red. We used confocal microscopy to measure the depth of biofilm formation by quantifying fluorescence. The Z-stacks were used to create surface plots of biofilm growth based on concanavalin A–Texas red intensity. Intensities are expressed as relative fluorescent units (RFUs) as previously described [
21]. To test the effect of lactoferrin supplementation on the smokers BAL biofilm formation, we used a final concentration of 300 μg/mL lactoferrin or phosphate buffer control in the samples and assessed biofilm as previously described.
We produced the CSE as previously described with modifications [
35]. We filled a 50 mL syringe with 10 mL of media and inserted the filter end of a research cigarette from the University of Kentucky (Code 3R4F) into the wide end of a 1000 μL pipette tip. We lit the cigarette and the narrow end of the pipette was placed into the tip of the 60 mL syringe. The syringe plunger was pulled smoothly to aspirate smoke from the cigarette, through the pipette tip, and into the syringe and mixed with the media by vigorous shaking. We repeated this process of smoke aspiration and media/smoke mixing until one cigarette was consumed. The resulting mixture was filtered through a 0.22 μm filter unit (EMD Millipore). CSE was made fresh for each experiment using the OD
320 to standardize the solution. An OD
320 equal to 1.00 corresponded to 100% CSE.
Bacterial growth in the presence of CSE alone
For S. aureus, we combined 50 μL of increasing concentrations of CSE solutions in PBS −/− (0.1, 0.3, 1, or 3%) with 150 μL of ~ 500 CFUs of log phase Xen 29 S. aureus bacteria in sodium phosphate buffer (10 mM, 100 mM NaCl, 1% TSB) in a round bottomed 96-well plate for 30 min at 37 °C. After 30 min, each condition was plated onto tryptic soy agar (TSA) plates, incubated overnight at 37 °C, and CFUs were counted.
For P. aeruginosa, Xen 05 bacteria were cultured overnight in TSB media and subculture in M9 media for 12 h at 37 °C. We combined 100 μL of bacterial solution in M9 (2.25 × 105 CFUs) with 100 μL of increasing concentration of CSE in M9 (0.01, 0.1, 1, or 3%) in a 96 well-plate and incubated for 18 h at 37 °C. The next day the samples were plated on TSA plates, incubated overnight at 37 °C to count colonies the following day.
Bacterial growth in BAL supplemented with CSE and ferric chloride
To supplement BAL with CSE we coincubated 50 μL of increasing concentrations of CSE (0.1, 0.3, 1, or 3%) with 140 μL of non-smokers BAL for 30 min. Thereafter, we added 10 μL of S. aureus Xen 29 (~ 1.5 × 104 CFU) and measured RLU at 30 min.
For P. aeruginosa, we coincubated 100 μL of a combination of non-smokers BAL with either CSE (1%), FeCl3 (1 μg/mL) or control with 100 μL of Xen 05 in M9 (~ 1.5 × 107 CFU) in a 96 well-plate. The sealed plate was incubated overnight and RLU were measured at 18 h.
Measurement of lactoferrin in the BAL
We diluted BAL 1:5 using water. Thereafter, we measured lactoferrin using a human lactoferrin (HLF2) ELISA kit (Abcam, USA) interpolating the unknowns to a standard curve to calculate the lactoferrin concentration in the BAL samples. The data was corrected for the dilution factor.
Trace metals in BAL were analyzed using a Thermo X series II inductively coupled plasma mass spectrometer ICP-MS with a collision cell (ThermoFisher Scientific). Samples were spiked with 0.5 mL of a 1 part per million indium solution (Inorganic ventures CGIN1–1) to serve as an internal standard. A 1:10 dilution was performed with 2% nitric acid (Fisher Chemical, Trace metal grade) and final sample volume was 10 mL. Calibration curves were prepared for each analyte of interest from a multi-element standard (Inorganic Ventures QCP-QCS-3 s source). Standards were prepared in concentrations from 1 to 200 parts per billion with the same internal standard. Standards were diluted with 2% nitric acid. Calibration curves were plotted using the response ratio of analyte/internal standard on the Y axis and concentration of analyte on the X axis. The slope was then used to obtain concentration of analyte in samples after a blank subtraction (2% nitric acid). The data was corrected for the dilution factor and samples that were below blank level concentration were considered as below the limit of detection and plotted as 0.
Data analysis
Data are expressed as mean ± SEM. For BAL growth, we used raw RLU and for the CSE experiments we normalized data as percent of control using this formula:
$$ \mathrm{Percent}\ \mathrm{of}\ \mathrm{live}\ \mathrm{bacteria}=\kern0.5em \frac{\mathrm{RLU}\ \mathrm{from}\ \mathrm{sample}}{\mathrm{RLU}\ \mathrm{from}\ \mathrm{control}\ \mathrm{vehicle}}\ \mathrm{x}\ 100 $$
All experiments had n = 11, were done in replicates in at least two independent experiments. The exception was the metal measurements in the BAL that was done once with all samples using the same standard to ensure that they were comparable. We determined the statistical significance between two related groups using paired t-tests. We used multiple comparison ANOVA and Kruskal-Wallis to compare three or more concentrations to their respective control. We also used the Pearson test to calculate correlation coefficients. Data analysis was performed using Graph Pad Prism 6.00 (GraphPad Software, California, USA).
Discussion
We investigated the bacteriostatic properties of human BAL collected in vivo from smokers and non-smokers using a standard bacterial inoculum. We found that S. aureus and P. aeruginosa grow more and develop more biofilm in the BAL samples taken from the lungs of smokers compared to non-smokers.
It has been proposed that the surface of the airways is iron-depleted to limit bacterial growth and virulence [
38,
39]. Although the study was underpowered to detect significant differences in iron concentrations, we found that smokers had four times higher mean iron concentration in the BAL than non-smokers. This finding is consistent with several reports that have found that smokers have increased iron concentration in their lungs [
20,
40‐
43]. This result and the report of other investigators is in part explained by the disruption of iron homeostasis mechanisms in the lungs of smokers.
Cigarette smoke particles are rich in iron and can directly increase iron concentration in the airways [
20] where particulate matter is deposited. Some of the particles are endocytosed and metal oxides on their surface can adsorb intracellular free iron to form ferruginous bodies decreasing its concentration available for the cell. In turn, low iron can be sensed by iron-regulatory proteins that activate iron-responsive elements to post-transcriptionally increase transferrin receptors in the cell basolateral membranes, upregulating iron uptake and further increase the iron content in the lungs [
44‐
46]. In addition, cigarette smoke contains polyhydroxybenzenes that can react with ferritin to release iron [
47]. Furthermore, iron in the airways might come from damage in the airway epithelial cells, which results in serum leakage [
40].
Increased iron concentration in the airways correlates with the severity of lung disease in cystic fibrosis and chronic bronchitis [
38,
48]. It has been proposed that changes in iron homeostasis can affect the susceptibility of the airway to develop infections [
46]. Most bacteria rely on a continuous supply of host iron to proliferate [
49] and high levels of serum iron can increase the risk of developing active infections such as tuberculosis [
50,
51]. In addition, iron nanoparticles can directly impair airway innate mechanism such as AMP activity [
21].
We found that other metals such as aluminum, lead, and vanadium that were not present in the airways of non-smokers. Accumulation of metals other than iron has also been observed in the lung and serum of patients with COPD [
52,
53]. Some these metal have been showed to impair mechanism involved in airway immunity associated with the pathogenesis of COPD such as decreased release of the AMP and decreased cystic fibrosis transmembrane conductance regulator (CFTR) function [
52,
54‐
56].
When we exposed
S. aureus and
P. aeruginosa to only CSE, there was no increased growth compared to a solution control. It has been reported that CSE has variable effects on bacterial growth. In general, it has an inhibitory effect that is greater in Gram positive than Gram negative bacteria [
12,
57]. However, these experiments with only CSE were done using doses that would also cause cell death in airway epithelial cells and do not necessarily recapitulate what occurs in the airways [
35].
When we supplemented BAL from non-smoking subjects with CSE, as a way to recreate in part the airway microenvironment, we found that CSE impaired BAL bacteriostatic properties. We found a similar impairment of the BAL when we supplemented with iron chloride. These results might suggest that iron bioavailability could be a mechanism for regulating airway antimicrobial activity. One key AMP in the airways is lactoferrin. One of its major function is iron chelation, which reduces the amount of bioavailable iron in biological fluids, including ASL [
13]. The function of lactoferrin is affected when saturated with iron [
58]. Therefore, we speculate that cigarette smoke contains iron nanoparticles that might not only be a source of iron for bacterial growth but could also inhibit the bacteriostatic properties of the ASL. Conversely, longer times were needed to notice a difference between control BAL and CSE or iron supplemented BAL in
P. aeruginosa (6 vs 18 h). These results suggest that other factors present in the BAL from smokers such as heme-iron, inflammatory mediators, different composition of AMP might also contribute to the increased bacterial growth.
Both active and passive smoking has been associated with an increase in lactoferrin concentrations in human secretions [
36,
37]. In our samples, we found no significant differences in the concentration of lactoferrin between smokers and non-smokers. However, we found an increased ratio of iron to lactoferrin. We speculate that the relative variability in iron/lactoferrin ratio of our samples is responsible to the heterogeneity of some of the results. Especially those experiments using
Pseudomonas (Figs.
1b,
5b & d). (Fig.
3c).
Pseudomonas has evolved to efficiently uptake iron by robust and redundant mechanisms [
59,
60]. This feature has allowed them to survive in a wide array of environments, such as water currents, plants, nematodes, insects, and in mammals, including humans [
61]. Previous studies have suggested that iron in the lungs might be important for
Pseudomonas airway colonization [
62]. As airway disease progresses in COPD, iron deposits in the airways also increases [
63]. It is also known that lung diseases such as severe COPD has higher rates of
Pseudomonas airway colonization. We also speculate that iron/lactoferrin imbalance will also increase the probabilities of being colonized by this pathogen associated with poor clinical outcomes in hospitalized patients [
29,
30].
When we supplemented BAL samples with excess lactoferrin we reverted the impaired bacteriostatic activity and biofilm formation observed in the BAL of smokers. One likely mechanism for excess lactoferrin reverting bacterial growth and biofilm formation in smokers is by decreasing bioavailability of iron. Despite that lactoferrin has other antimicrobial mechanisms such as direct binding to LPS, osmotic effect, and bacterial membrane permeabilization, these are also impaired when iron is bound to lactoferrin [
64‐
66]. We acknowledge that the concentration of lactoferrin added to the samples could be considered supraphysiologic. However, commercially available lactoferrin is partially saturated with iron. For this reason, higher doses of AMP were used to observe an effect, particularly in the biofilm experiments.
As a limitation of our study, we acknowledge that Haemophylus influenzae, an important organism in COPD exacerbations was not considered for this study. However, this organism needs hemin, an iron containing protoporphyrin and nicotinamide adenine dinucleotide to grow in vitro, the addition of these elements could confound the hypothesis of smoking as a source of iron for bacteria.
Several investigators have recently reported that current and former smokers with preserved lung function by spirometry have increased respiratory symptoms and evidence of airway disease by imaging [
67,
68]. This study demonstrates that the imbalance between iron content and lactoferrin abundance in the airways can result in conditions that will impair airway innate immune mechanisms, resulting in an increased risk of respiratory infections. Since the development of airway infections has been proposed as an important mechanism for lung function decline and development of chronic bronchitis, our results provide a potential mechanism for some of the recent reports of respiratory symptoms in smokers without spirometry evidence of COPD [
24,
25].
The implications of iron/lactoferrin imbalance in the development of COPD might go beyond increasing bacterial growth. A recent discovery demonstrated that a gene that encodes for an iron receptor protein was associated with dysfunctional mitochondrial iron loading affecting mucociliary clearance and contributing to the development of COPD [
69,
70]. In the same study, the use of deferiprone, an USDA approved drug that functions as an iron chelator in a mouse model of COPD, improved features of airway disease progression and acute lung injury such as weight loss, pulmonary inflammation, and decreased mucociliary clearance despite continuous cigarette smoke exposure. Further studies will have to determine the feasibility of this intervention but suggest a promising avenue to prevent the progression from smoking to COPD by iron chelation.