Pseudomonas aeruginosa Induced Host Epithelial Cell Mitochondrial Dysfunction

The pathogenicity of P. aeruginosa is dependent on quorum sensing (QS), an inter-bacterial communication system that can also modulate host biology. The innate immune function of the lung mucosal barrier is dependent on proper mitochondrial function. The purpose of this study was to define the mechanism by which bacterial factors modulate host lung epithelial cell mitochondrial function and to investigate novel therapies that ameliorate this effect. 3-oxo-C12-HSL disrupts mitochondrial morphology, attenuates mitochondrial bioenergetics, and induces mitochondrial DNA oxidative injury. Mechanistically, we show that 3-oxo-C12-HSL attenuates expression of peroxisome proliferator-activated receptor-γ coactivator-1α (PGC-1α), a master regulator of mitochondrial biogenesis, antioxidant defense, and cellular respiration, and its downstream effectors in both BEAS-2B and primary lung epithelial cells. Overexpression of PGC-1α attenuates the inhibition in cellular respiration caused by 3-oxo-C12-HSL. Pharmacologic activation of PGC-1α restores barrier integrity in cells treated with 3-oxo-C12-HSL. These data demonstrate that the P. aeruginosa QS molecule, 3-oxo-C12-HSL, alters mitochondrial pathways critical for lung mucosal immunity. Genetic and pharmacologic strategies that activate the PGC-1α pathway enhance host epithelial cell mitochondrial function and improve the epithelial innate response to P. aeruginosa. Therapies that rescue PGC-1α function may provide a complementary approach in the treatment of P. aeruginosa infection.


P. aeruginosa QS molecules disrupt mitochondrial morphology in bronchial epithelial cells.
Mitochondria, which are central to cell metabolism and also important mediators of other cellular pathways including those that regulate cell death, are frequent targets of microbial products 18 . Mitochondrial morphology is tightly regulated by dynamic mitochondrial quality control mechanisms including mitochondrial biogenesis, mitophagy, fusion, and fission. Changes in mitochondrial morphology can be associated with perturbations in mitochondrial function 19 . Previous research has demonstrated that the QS molecule, 3-oxo-C12-HSL, activates the apoptotic pathway in a variety of cell types [20][21][22] and causes mitochondrial membrane permeabilization in fibroblasts 20 , but little is known about their specific effects on lung epithelial cells. We hypothesized that QS molecules would disrupt mitochondrial morphology in bronchial epithelial cells. Indeed, a 6-hour treatment with 100 μM 3-oxo-C12-HSL had a profound qualitative effect on mitochondrial morphology and cristae structure (Fig. 1A). Quantitatively, it caused significant decrease in number of mitochondrial per high power field (Fig. 1B), mitochondrial area (Fig. 1C), and length-to-width ratio (Fig. 1D) as detected by electron microscopic morphometric analysis. Infection with P. aeruginosa strain, PAO1 (MOI 20) for 6 hours had a similar effect on mitochondrial length to width ratio (Fig. 1D). Collectively, these data demonstrate that P. aeruginosa and its QS molecules may evade host defenses by disrupting host epithelial mitochondria.

3-oxo-C12-HSL attenuates metabolic potential and mitochondrial respiration in bronchial epithelial cells.
Since at least 90% of cellular energy is produced by mitochondria 23 , we wanted to investigate whether the changes in mitochondrial structure reflect a disruption in cellular bioenergetics. We utilized the Seahorse XF Cell Energy Phenotype test, which allows for the real time measurement of both extracellular acidification rate (ECAR), a measure of glycolysis, and oxygen consumption rate (OCR), a measure of oxidative phosphorylation. Because the assay did not provide reliable data in cells treated with live bacteria because of microbial metabolism that interfered with the assay, we focused on the specific effect of QS molecules on cellular bioenergetics ( Fig. 2A). A 6-hour treatment with 100 μM 3-oxo-C12-HSL resulted in a significant decrease in baseline OCR (Fig. 2B) and ECAR (Fig. 2C) levels. In addition, cells treated with the QS molecule, had less metabolic potential to increase the rates of glycolysis and OXPHOS in response to the stress caused by the addition of an uncoupling agent, FCCP, and the ATP synthase inhibitor, oligomycin, (Fig. 2B,C).
Next, we wanted to specifically interrogate the effect of the QS molecule on various aspects of the OXPHOS pathway. First, we analyzed whether 3-oxo-C12-HSL reduced steady-state levels of ATP, since OXPHOS is the primary cellular process by which ATP is generated. In cells treated with 100 μM 3-oxo-C12-HSL for 6-hours, there was a significant decrease in ATP concentration normalized to intracellular protein (Fig. 3A). Next, we employed the Seahorse XF Mito Stress Test to interrogate various aspects of the mitochondrial respiration pathway (Fig. 3B). We found that a 6-hour treatment with 50-200 μM 3-oxo-C12-HSL attenuated basal respiration (Fig. 3C), maximal respiration (Fig. 3D), spare respiratory capacity (Fig. 3E), and ATP production (Fig. 3F) in a dose-dependent manner. Together, these data suggest that QS molecules compromise mitochondrial respiration in epithelial cells, and this effect may result in reduced immune function. 3-oxo-C12-HSL and PAO1 disrupt mitochondrial morphologic parameters. Results are mean ± SEM. *P < 0.05, **p < 0.01, ****p < 0.0001 all vs. control. One-way ANOVA with Tukey's multiple comparisons test used for statistical analysis.

3-oxo-C12-HSL decreases mitochondrial content in bronchial epithelial cells. Others have
demonstrated that cellular bioenergetic crises are generally a potent stimulus for activation of the mitochondrial biogenesis pathway in order to create new mitochondria to replace those that have been damaged 24 . We hypothesized that the QS molecule would similarly cause an increase in mitochondrial biogenesis in host cells. Mitochondrial biogenesis is principally regulated by PGC-1α, a transcriptional co-activator that promotes the transcription of nuclear and mitochondrial genes crucial for mitochondrial function 25,26 . Interestingly, 16-hour treatment with 100 μM 3-oxo-C12-HSL or PAO1 (MOI 1) resulted in a significant decrease in the mRNA and protein levels of PGC-1α (  Supplementary Fig. 2) in BEAS-2B cells. In primary human bronchial epithelial cells, 100 μM 3-oxo-C12-HSL and infection with PAO1 similarly reduced expression of PGC-1α and TFAM (Fig. 4D,H). 3-oxo-C12-HSL also caused a significant decrease in expression of the mitochondrial specific protein, voltage-dependent anion-selective channel (VDAC) 1 (Fig. 4I). Furthermore, it significantly decreased relative mtDNA content normalized for nuclear DNA in BEAS-2B cells (Fig. 4J) and in human primary bronchial epithelial cells (NhBEs) (Fig. 4K). These findings together confirm that 3-oxo-C12-HSL not only attenuates mitochondrial bioenergetics but also suppresses the ability of the cell to regenerate new mitochondria through biogenesis to respond to this energetic stress.
3-oxo-C12-HSL induces ROS generation, oxidative mitochondrial DNA damage, and the apoptotic pathway in bronchial epithelial cells. Electron transport chain complexes regularly leak a small percentage of electrons to partially reduce oxygen to produce superoxide anion, which can go on to produce additional ROS. Because of the potential damage that can be caused by these ROS, there are several antioxidant defenses within mitochondria. The balance between ROS and antioxidant defenses is tightly regulated under physiologic conditions, but under pathologic conditions, this balance can be disrupted leading to excessive ROS generation, oxidative injury, and apoptosis 27 . We utilized MitoSOX ™ Red, a fluorogenic dye, which specifically detects superoxide anion, to determine the effect of the QS molecule, 3-oxo-C12-HSL on ROS generation. Flow cytometry analysis demonstrated a noticeable increase in fluorescence intensity in cells treated with 3-oxo-C12-HSL (Fig. 5A). 3-oxo-C12-HSL caused a significant increase in cells staining positive for MitoSOX ™ as compared to control and on the order with the positive control, the respiratory chain toxins, rotenone and antimycin A (Fig. 5B).
Since mtDNA is particularly susceptible to ROS generation, we next investigated whether the QS molecule enhances mtDNA damage. Indeed, 3-oxo-C12-HSL caused a significant increase in mtDNA damage as measured by an increase in the mt79/mt230 DNA fragment ratio (Fig. 5C) in BEAS-2B cells. This effect was also seen in human primary lung epithelial cells treated with 3-oxo-C12-HSL (Fig. 5D). www.nature.com/scientificreports www.nature.com/scientificreports/ Finally, because excessive ROS are a known stimulus for induction of cell death, we interrogated whether the apoptosis pathway was activated in response to the mitochondrial damage and oxidative stress caused by 3-oxo-C12-HSL. The QS molecule caused a significant increase in cytochrome C release (Fig. 6A), an early event in the mitochondrial-dependent intrinsic apoptosis pathway, but activation of the intrinsic pathway initiator caspase, caspase-9, was not observed (Supplementary Fig. 3). Instead, cleavage of the extrinsic pathway initiator caspase, caspase-8, was induced by 3-oxo-C12-HSL at 3 and 6 hours (Fig. 6B, Supplementary Fig. 3). The QS molecule also activated the executioner caspase, caspase-3 ( Fig. 6C) and downstream cleavage of Poly (ADP-ribose) Polymerase (PARP) (Fig. 6B, Supplementary Fig. 3). Collectively, these data show that P. aeruginosa QS molecules induce production of mitochondrial ROS with oxidative mtDNA damage and triggers the apoptotic pathway. Since epithelial cells are critical for host defenses for P. aeruginosa, these damaging effects likely contribute to impaired epithelial immune response.
Overexpression of PGC-1α partially rescues 3-oxo-C12-HSL-induced impairment in mitochondrial biogenesis and respiration. PGC-1α is a master regulator of cellular respiration, mitochondrial biogenesis, and anti-oxidant defenses. Given its central role in the pathways that are perturbed by the QS molecule, 3-oxo-C12-HSL, we hypothesized that overexpression of PGC-1α would ameliorate some of its deleterious effects.   Supplementary Fig. 4). In addition, TFAM levels were restored with AdPGC-1α transduction ( Fig. 7B,C, Supplementary Fig. 4).
Using the SeahorseXF Mito Stress Test assay, we also found that overexpression of PGC-1α partially restored parameters of mitochondrial respiration in cells treated with the QS molecule ( Fig. 7D-H). In particular, there was a significant increase in basal respiration (Fig. 7E), maximum respiration ( Fig. 7F), spare respiratory capacity (Fig. 7G), and ATP production (Fig. 7H), in 3-oxo-C12-HSL-treated cells that were overexpressing PGC-1α as compared with AdGFP controls. Together, these data show that overexpression of PGC-1α can at least partially rescue mitochondrial biogenesis pathways and cellular respiration that is impaired by the QS molecules.
Metformin and resveratrol restore epithelial barrier integrity following 3-oxo-C12-HSL treatment. Given the beneficial effects of a genetic strategy of PGC-1α overexpression on cellular bioenergetics, we next wanted to investigate whether a pharmacologic strategy targeting enhancement of PGC-1α activity would rescue functional improvements in lung epithelial cells. PGC-1α activity is induced by post-translational modifications including phosphorylation by AMPK and deacetylation by SIRT1 25 . For these studies we utilized resveratrol, a potent activator of SIRT1, and metformin, a canonical activator of AMPK, to determine whether PGC-1α could rescue the functional effects of QS molecules on host epithelial cells. We found that 24-hour pre-treatment with resveratrol (20 μM) or metformin (1 mM) prevented attenuation of TFAM expression by QS molecules in BEAS-2B cells (Fig. 8A,B). In our experimental model, BEAS-2B cells could not form a tight epithelial barrier (baseline trans-epithelial electrical resistance [TEER] measurements did not exceed 200 Ω x cm 2 ). Therefore, we utilized Calu-3 lung epithelial cells that form tight monolayers on transwell supports. In Calu-3 cells inoculated with PAO1 (MOI 1) on the apical chamber for 6 hours, pre-treatment with resveratrol (20 μM) or metformin (1 mM) significantly reduced the transmigration of bacteria across the epithelial barrier (Fig. 8C). Further, 3-oxo-C12-HSL disrupted the distribution of the tight junction-associated protein, Zonula Occludens-1 (ZO-1), as detected by immunofluorescence microscopy. Pretreatment with resveratrol or metformin resulted in preserved tight junction architecture (Fig. 8D). In additional experiments in Calu-3 cells, treatment with the QS molecule, 3-oxo-C12-HSL, caused a significant decrease in TEER that persisted over several days (Fig. 8E,H). However, a 24-hour pre-treatment with resveratrol (20 μM) or metformin (1 mM) restored barrier function 24-and 48-hours following treatment with the QS molecule 3-oxo-C12-HSL (Fig. 8F,G,I,J). In primary human airway cells, pre-treatment with metformin or resveratrol prevented the early (3-12 hour) decline in TEER caused by 3-oxo-C12-HSL and protected barrier integrity up to 24 hours following treatment ( Fig. 8K-M,N-P). Collectively, these data for the first time show that pharmacologic approaches to induce PGC-1α may help restore barrier integrity thus improving immune function preventing transmigration of bacteria.

Discussion
In this study, we have identified a key pathway that is involved in the pathogenesis of lung epithelial injury caused by a virulence factor produced by P. aeruginosa. 3-oxo-C12-HSL, the principal quorum sensing molecule used by P. aeruginosa, disrupts mitochondrial morphology, decreases bioenergetic potential, attenuates cellular respiration, enhances ROS generation and subsequent oxidative mtDNA damage, and induces apoptosis in. It also decreases expression of mitochondrial proteins, PGC-1α and TFAM, and decreases mtDNA content, indicative of a reduction in mitochondrial biogenesis (Fig. 9A) in both cell line and primary human lung epithelial cells. Given the central role of PGC-1α as a master regulator of mitochondrial biogenesis, mitochondrial respiration, and anti-oxidant defenses, we hypothesized that activation of the PGC-1α pathway would ameliorate these deleterious effects. Genetic overexpression of PGC-1α does partially rescue markers of mitochondrial biogenesis and cellular respiration. Furthermore, pharmacologic activation of PGC-1α with the agents, metformin or resveratrol, restores epithelial barrier integrity and mitochondrial biogenesis following exposure to 3-oxo-C12-HSL and www.nature.com/scientificreports www.nature.com/scientificreports/ prevents bacterial transmigration across the epithelial monolayer (Fig. 9B). These data provide a rationale for a novel therapeutic strategy for P. aeruginosa lung infections.
Mitochondria play a central role in the generation of energy necessary to supply cellular functions. In addition, mitochondria regulate important second messenger cellular signaling pathways including calcium signaling and the induction of the intrinsic apoptosis pathway through cytochrome c release. Given these important functions and perhaps owing to the ancestral origin of mitochondria as a product of bacterial endosymbiosis, it is not surprising that these organelles are frequent targets of bacterial virulence factors 28 . In the lung, mitochondrial function is critical for the proper function of the epithelium including the maintenance of barrier integrity, ciliary function, and the regeneration of epithelial cells following injury 12 . Therefore, mitochondrial dysfunction can result in cellular and tissue dysfunction and promote perpetuation of infection. In addition, damaged mtDNA can act as a danger-associated molecular pattern (DAMP) that can activate TLR9 on nearby immune cells. Further, mitochondrial ROS and mtDNA can activate the NLRP3 inflammasome thereby resulting in increased tissue inflammation and organ injury 29 .
P. aeruginosa QS molecules, specifically acyl-homoserine lactones, have been shown to affect host epithelial cells in several ways including loss of barrier integrity 30,31 , induction of apoptosis 22,31,32 , and cytosolic calcium release 33,34 . These molecules have also been shown to induce ROS generation in intestinal epithelial cells 32 . We demonstrate for the first time that QS molecules also disrupt mitochondrial morphology, attenuate metabolic potential, induce ROS production and oxidative injury to mtDNA in lung epithelial cells. In addition, 3-oxo-C12-HSL markedly reduces oxidative phosphorylation and ATP production in lung epithelial cells. www.nature.com/scientificreports www.nature.com/scientificreports/ Interestingly, Tao et al. found that 3-oxo-C12-HSL enhanced activity of the respiratory complexes IV and V in intestinal goblet cells 35 . This suggests that there may be cell type-and tissue-specific effects of QS molecules on cellular bioenergetics, which may contribute to the predilection of P. aeruginosa to cause respiratory infections.
In the face of mitochondrial injury, cells have evolved finely tuned processes designed to maintain tight regulation of mitochondrial quality control. The elimination and replacement of damaged mitochondria are important for the promotion of cell survival. Mitochondrial biogenesis, the process of creating new mitochondria, results from growth and division of existing mitochondria and requires the coordinated activity of PGC-1α and TFAM. We demonstrate for the first time that P. aeruginosa QS molecules reduce expression of these key regulators in immortalized and primary human lung epithelial cells further supporting the relevance of this finding to clinical infections in humans. In addition, decreased levels of mitochondrial DNA and expression of the mitochondrial marker, VDAC1, demonstrates a reduction in mitochondrial biogenesis. This is a novel finding not previously reported in any cell types. There is scant research highlighting the effect of bacterial pathogens on host mitochondrial biogenesis. One previous study in a murine model of Staphylococcus aureus pneumonia found that mitochondrial biogenesis was upregulated during lung repair in the distal lung epithelium following antibiotic www.nature.com/scientificreports www.nature.com/scientificreports/ treatment 14 . Others have found that mitochondrial biogenesis is attenuated early in other sepsis models with recovery over time 36,37 . In addition, dysregulation of mitochondrial biogenesis potentiated the degree of liver injury in an experimental model of sepsis 36 . Finally, a recent study by Cui et al. demonstrated that PGC-1α is downregulated in mouse alveolar epithelial cells exposed to LPS and conditional deletion of PGC-1α in these cells aggravated a mouse model of acute lung injury 38 . These studies support the idea that PGC-1α, through its promotion of cellular bioenergetics and fatty acid oxidation serves a protective role in the homeostatic host response to inflammatory stimuli. Given the importance of proper mitochondrial function in maintaining epithelial barrier integrity and host response [11][12][13] , disruption of mitochondrial function and the mitochondrial biogenesis repair process may reflect an additional virulence mechanism employed by P. aeruginosa.
We hypothesized that a strategy targeting activation of PGC-1α to enhance cellular respiration, mitochondrial biogenesis, and anti-oxidant defenses could protect the epithelial host response to P. aeruginosa infection. Such a strategy has had efficacy in models of other tissue injury including pulmonary fibrosis 16 , hepatic ischemia-reperfusion injury 39 , kidney injury 40 , and aging-related degeneration in the mouse lung 41 . Using an adenovirus overexpression strategy, we found that upregulation of PGC-1α attenuated the reduction in mitochondrial respiration induced by 3-oxo-C12-HSL. In addition, we employed two widely available chemicals, resveratrol and metformin, that act through complementary mechanisms to activate PGC-1α via post-translational modifications. These molecules resulted in more rapid recovery of barrier integrity following QS molecule treatment. This provides preliminary evidence that enhancement of PGC-1α activity represents a novel therapeutic strategy for P. aeruginosa infection.
A limitation of this paper is that we did not investigate other homeostatic mitochondrial quality control processes that work alongside mitochondrial biogenesis, specifically mitophagy. Ongoing studies are necessary to delineate whether mitophagy is disrupted in host lung epithelial cells by QS molecules. Another limitation derives from the fact that resveratrol and metformin exert many pleotropic effects through their effects on AMPK or SIRT1 in addition to activation of PGC-1α. Further investigations are needed to clarify the exact mechanisms by which metformin and resveratrol restore barrier integrity. In addition, a limitation of this manuscript is the lack of animal studies examining the effect of bacterial infection on in vivo epithelial mitochondrial function. These and other ongoing studies in our laboratory will further define the precise role of PGC-1α in mitochondrial host response to invading pathogens.
In conclusion, these studies provide a new paradigm and implications for the role of mitochondria in the epithelial cell host response to P. aeruginosa infections. Our data suggest that novel molecular approaches to rescue mitochondrial function may enhance immune responses to pathogens such as P. aeruginosa. Since drugs such as metformin and resveratrol that can activate PGC-1α are already in clinical use, these results are translatable to patients with P. aeruginosa infections and other pathogens that evade immune defenses by similar mechanisms. www.nature.com/scientificreports www.nature.com/scientificreports/ Methods Cell line models. BEAS-2B bronchial epithelial cells (ATCC, Rockville, MD) were maintained with complete BEGM media (Lonza, Basel, Switzerland) supplemented with 10% FBS and plated on dishes coated with collagen-fibronectin-BSA mixture. Calu-3 cells (ATCC) were maintained in EMEM media supplemented with 10% FBS plated on 3 μM polyester membrane transwell supports (Corning, Corning, NY). After plating cells, epithelial tight junctions were allowed to mature over 7-10 days until the TER was over 1000 Ω x cm 2 . Cells were treated with 3-oxo-C12-HSL (Sigma-Aldrich, St. Louis, MO), trans-resveratrol (Cayman Chemical, Ann Arbor, MI), or metformin (Cayman Chemical), or co-cultured with the P. aeruginosa strain, PAO1, (MOI of 1 or 30), for 6 or 16 hours. Primary NhBE cells were grown in conditional reprogrammed cell (CRC) culture medium as described in Liu et al. 42 , on human type IV placental collagen-coated 12-mm Costar Transwell supports (Corning) with slight modifications, by the CF@LANTA RDP Experimental Models Core at Emory University as described previously 43 . Bacterial stocks. Stock of P. aeruginosa strain PAO1 was prepared as previously described 44  ATP concentration. The ATP Determination Kit (Molecular Probes, Eugene, OR), a bioluminescence assay for quantitative determination of ATP, was used to assess steady state levels of ATP in cell lysates. 3 μg cell lysate samples were loaded in triplicate into a 96 well plate luminometer and the assay was conducted per manufacturer instructions. An ATP standard curve was generated. Luminescence of samples and standards was measured in triplicate per the manufacturer instructions and the quantity of ATP was calculated based on the standard curve.
RoS detection. Cells were detached with warm accutase, then collected by centrifugation, and re-suspended in media. MitoSOX ™ Red (Invitrogen) was added at a final concentration of 5 mM for 20 mins at 37 °C. The cells were washed once in warm PBS and immediately analyzed by FACS analysis. Cells were acquired using BD FACSAria II and analyzed using FACSDiva software. The level of intracellular ROS corresponded to an increase in fluorescence. Cells expressing fluorescence above a defined threshold were considered positive and were evaluated based on the population defined by double discrimination gating and the unstained and untreated controls.
Mitochondrial DNA content. Relative mitochondrial DNA content was determined as described previously 47,48 .
Briefly, DNA was isolated from samples using DNeasy kit (Qiagen) per manufacturer instructions. DNA was quantified using NanoDrop spectrophotometry (Thermo Fisher Scientific, Washington, DE). 40 ng DNA of each sample was amplified with mtDNA tRNA Leu(UUR) and nuclear β-2-microglobulin primer pairs (tRNA Leu(UUR) Forward: 5′-CACCCAAGAACAGGGTTTGT-3′, Reverse: 5′-TGGCCATGGGTATGTTGTTA-3′; β-2-microglobulin Forward: 5′-TGCTGTCTCCATGTTTGATGTTGTATCT-3′; Reverse: 5′-TCTCTGCTCCCCACCTCTAAGT-3′). The mtDNA tRNA Leu(UUR) gene region was selected because it is rarely deleted or duplicated and contains only a few rare SNPs. Relative mtDNA content was normalized to nuclear DNA using the ΔΔC T method. caspase activity assay. Caspase-3/CPP32 colorimetric assay kit (BioVision, Mountain View, CA) was used to determine caspase activity per manufacturer's instructions as previously described 49 . Cells were treated and collected with cell lysis buffer. An equal amount of protein was loaded per assay. Absorbance was measured at 405 nm on a microplate reader. Caspase activity is expressed as fold change in comparison to untreated control cells.
Cytochrome C releasing assay. The water soluble mitochondrial inner membrane protein, cytochrome c was detected by solid phase sandwich enzyme-linked immune-sorbent assay (ELISA, Invitrogen, Camarillo, CA) to detect an early component required for intrinsic apoptosis initiation per manufacturer instructions as previously described 50 . Samples (3 μg protein per well) and standard containing cytochrome c were loaded into wells coated with a monoclonal antibody specific for cytochrome c. Next, a biotinylated secondary antibody specific for cytochrome c is added. Following addition of streptavidin peroxidase enzyme, a substrate solution was added to the wells to produce color detected at 450 nm on a microplate reader. A standard curve was generated and the concentration of each sample was then calculated.
Adenovirus overexpression of PGC-1α. An adenovirus expressing PGC-1α, AdPGC-1α (a gift from Dr. Russ Price, East Carolina University) was used as described previously 46 . Briefly, BEAS-2B cells were transduced with AdPGC-1α (MOI 25) in BEGM media containing 2% FBS. Media was changed to full 10% FBS media 6 hours following transduction. RNA was isolated 48 hours following transduction. Protein was isolated and Seahorse bioenergetics assays were performed 72 hours following transduction. teeR. Calu-3 cells were grown on 3 μM polyester membrane Costar Transwell ® permeable supports (Corning) and tight junctions were matured after a period of 7-10 days. Alternatively, differentiated primary NHBes were grown on type IV placental collagen-coated 6.5-mm Costar Transwell supports. TER was measured using an Ohmmeter (World Precision Instruments, Sarasota, FL) with chopstick electrodes as previously described 43 . Cells were pretreated with resveratrol or metformin for 24 hours prior to treatment with 100 μM 3-oxo-C12-HSL. TER was measured at baseline and at serial intervals following treatment.
Bacterial transmigration assay. Calu-3 grown on Transwell ® permeable supports as above were pretreated with resveratrol or metformin for 18 h. Bacterial suspension (MOI of 1) was added apically. After 6 hours, the apical and basolateral media were collected separately. A sample of the basolateral medium was serially diluted, cultured on LB agar, and incubated at 37 °C overnight. Colony-forming units were recorded (accurate range, ≥30 ≤ 300), as described previously 43 . Immunofluorescence. BEAS-2B cells were plated on 8-well chamber slides (Thermo Fisher Scientific) coated with collagen, fibronectin, and BSA, as described above. Immunofluorescence was performed on cells as follows: post-treatment chamber slides were washed in PBS twice and fixed with 2% paraformaldehyde for 15 minutes. Cells were then permeabilized with 0.1% Triton-X-100 and blocked with PBS (with calcium and magnesium) that contained 0.1% Triton-X-100 and 2% goat serum (Sigma-Aldrich). Cells were incubated with rabbit polyclonal anti-ZO-1 antibody (Thermo Fisher Scientific, #617300) overnight at 4 °C and then incubated with fluorescent-tagged (Alexa Fluor ® 488) secondary goat anti-rabbit IgG (Thermo Fisher Scientific) for 1 hour at room temperature. Slides were mounted with ProLong Gold mounting media (Thermo Fisher Scientific). Images were obtained using Olympus Fluoview FV1000 confocal microscope.

Statistical analysis.
All experiments were repeated at least 3 times. Data are presented as mean ± SEM.
Detailed information regarding statistical tests used is included in figure legends. All statistical analyses were performed by using GraphPad Prism (GraphPad Software, La Jolla, CA, USA).