Aquaporin 5 regulates cigarette smoke induced emphysema by modulating barrier and immune properties of the epithelium

The absence of AQP5 mitigates cigarette smoke-induced pulmonary emphysema, epithelial permeability and alveolar inflammatory cell accumulation

We exposed AQP5^−/− mice and WT littermates to chronic (6 mo) cigarette smoke to assess for emphysema as well as the alveolar inflammatory milieu. WT mice exposed to CS for 6 mo demonstrated a significant, 8% increase in the mean linear intercept (MLI) (Fig. 1A) and a significant decrease in surface area of airspace wall per unit of lung volume (S/V) (Fig. 1B) compared with age-matched WT mice exposed to air. MLI measures interalveolar septal distance and is an index of alveolar size.^29,30 S/V measures septal loss and is an index for alveolar destruction.^31,32 The change in MLI and S/V were typical of previous 6 mo exposures in this facility and in other published reports.^24,33 In contrast to WT mice, AQP5^−/− mice exposed to CS for 6 mo did not have significant changes in MLI (Fig. 1A) or S/V (Fig. 1B) compared with age-matched air-exposed AQP5^−/− mice. Air-exposed controls from WT or AQP5^−/− had similar MLI and S/V values. The only group with a significant change in MLI and S/V were the WT CS-exposed mice, suggesting protection from emphysema in the absence of AQP5. Representative histologic lung sections stained with H&E demonstrate a qualitative increase in airspace only with CS-exposed WT mice in comparison to air exposed controls; significant airspace enlargement does not appear to occur in AQP5^−/− mice irrespective of CS exposure (Fig. 1C). Saving for some variability, 6 mo of cigarette smoke exposure has been shown to cause an 8–12% increase in MLI due to emphysema formation.²⁹

In addition to its water transport properties, our group has demonstrated that the c-terminus of AQP5 alters human airway epithelial permeability in vitro by altering microtubule assembly.^28,34 We were interested in determining whether AQP5 also affected murine epithelial paracellular permeability. Using a previously developed assay of ex vivo tracheal permeability,³⁴ we detected less extraluminal Evans blue dye (EBD) in trachea from air-exposed AQP5^−/− mice compared with trachea from air-exposed WT mice (Fig. 2A), but suspect this difference falls within a physiologically normal range given the absence of apparent baseline functional differences in AQP5^−/− mice compared with WT mice . With the addition of only 1 d of cigarette smoke exposure, trachea from WT mice had significantly more extraluminal EBD efflux compared with trachea from air-exposed WT mice. In contrast, CS-exposed AQP5^−/− mice did not demonstrate a significant increase in EBD tracheal efflux when compared with air-exposed AQP5^−/− mice. Next, we assessed for CS-altered permeability across airway and alveolar epithelial layers using a ratio of EBD concentration in the bronchoalveolar fluid (BAL) to EBD concentration in the lung in WT and AQP5^−/− mice after 4 weeks of air or CS exposure (Fig. 2B). Despite similar ratios in air-exposed WT and AQP5^−/− mice, CS-exposed WT mice had a significant increase in the BAL to lung EBD ratio compared with air-exposed mice, consistent with increased permeability across the airway and alveolar epithelial layers; AQP5^−/− mice were protected from this phenomenon. EBD lung concentration (~8–10 μg/mL), and EBD lung to EBD blood ratios (~0.05) were similar between WT and AQP5^−/− mice suggesting that differences in BAL EBD efflux were not due to altered endothelial permeability (not shown). In addition to solute efflux, we assessed airway expression of an adherens junction protein, E-cadherin, in air or CS-exposed mice by lung immunofluorescence (Fig. 2C). Air-exposed AQP5^−/− mice had increased airway epithelial E-cadherin expression compared with WT mice. CS exposure significantly reduced E-cadherin airway epithelial expression only in WT mice. Similar to what was observed with EBD efflux, AQP5^−/− mice resisted CS-induced changes, in this case to a key protein involved in paracellular permeability.

The absence of AQP5 alters alveolar epithelial cell turnover after cigarette smoke

We used flow cytometry to identify and enumerate AQP5-induced differences in lung epithelial cells after chronic smoke exposure. Using total lung cells as the denominator, we found the % of CD326+ cells, a pan-epithelial cell marker, to be similar between CS-exposed strains (Fig. 3A). When we assessed the percentage of type I alveolar epithelial cells (AECs) identified by T1α,³⁵ there were significantly fewer in CS-exposed AQP5^−/− mice compared with CS-exposed WT mice; a representative flow plot is also shown (Fig. 3B). Type II AECs are known to proliferate and transition into type I AECs. Using SPC to identify type II AECs by immunofluorescence, we observed an increase in CS-exposed AQP5^−/− mice compared with CS-exposed WT mice (Fig. 3C). We further quantified SPC using immunblots of lung homogenate (Fig. 3D). AQP5^−/− mice exposed to 6 mo of CS express higher levels of lung SPC compared with WT mice. Densitometry of SPC expressed as a ratio of SPC to actin was significantly increased in lungs from CS-exposed AQP5^−/− mice. Collectively, this data suggests an increased ratio of type II to type I AEC presence in AQP5^−/− mice compared with WT mice exposed to chronic cigarette smoke. We also quantified proliferation (Ki-67⁺) among lung epithelial cells (CD326⁺) after 6 mo of CS exposure in WT and AQP5^−/− mice; a representative flow diagram demonstrating how CD326 positive and Ki-67 positive lung cells were selected from CS-exposed and air-exposed mice is shown (Fig. S1A). Both WT and AQP5^−/− CS-exposed mice had a 5-fold increase in % of Ki-67⁺ cells, suggesting CS induction of epithelial cell turnover. However, % Ki-67 positivity was similar between CS-exposed WT and AQP5^−/− mice among CD326⁺ cells (Fig. 3E). Of note, air and CS-exposed mice had a similar number of total lung cells and a similar percentage of CD326⁺ cells.

The absence of AQP5 blunts alveolar neutrophil accumulation after sub-acute and chronic CS exposure

In order to identify potential mechanisms by which the absence of AQP5 was protective from CS-induced damage and to provide a potential link between physical and immune barrier properties of the epithelium, we assessed whether AQP5 could influence the inflammatory cell abundance in the alveolar space with CS exposure (4 weeks vs. 6 mo) despite being expressed only on epithelial cells.³⁶ WT and AQP5^−/− mice exposed to CS for 6 mo had significant but similar increases in total BAL cells compared with their respective 6 mo air-exposed controls; there was no difference between CS-exposed groups (Fig. 4A). However, in assessing the alveolar inflammatory cellular milieu by examining BAL differentials we found that CS-exposed WT mice had significantly more BAL neutrophils compared with air-exposed WT mice or CS-exposed AQP5^−/− mice (Fig. 4B). CS-exposed WT mice had a significant increase in BAL macrophages compared with air-exposed WT mice (Fig. 4C), but were similar compared with CS-exposed AQP5^−/− mice. In contrast, CS-exposed AQP5^−/− mice had a similar number of BAL macrophages compared with air-exposed AQP5^−/− mice. BAL lymphocytes were increased similarly in both CS-exposed mouse strains compared with air-exposed mice (Fig. 4C).

Given the altered inflammatory milieu after chronic CS exposure, we were interested in knowing whether strain differences in BAL inflammatory cell profiles were present after a sub-acute CS exposure (4 weeks) thus preceding AQP5-dependent lung structural changes occurring after chronic CS exposure. Both groups exposed to CS had significant but similar increases in total bronchoalveolar (BAL) cells compared with their respective 4-week air-exposed controls (not shown). Compared with WT air-exposed controls, WT mice exposed to cigarette smoke for 4 weeks had a significant increase in BAL neutrophils (Fig. 4D). CS-exposed AQP5^−/− mice had similar numbers of BAL neutrophils compared with air-exposed AQP5^−/− mice, and significantly less than CS-exposed WT mice, a similar pattern to what we observed after 6 mo of CS. In contrast to BAL neutrophils, 4 weeks of CS exposure did not increase the number of BAL macrophages in WT mice. CS-exposed AQP5^−/− mice had a significant, albeit mild, increase in BAL macrophages only when compared with WT air-exposed mice, but not AQP5^−/− air-exposed mice (Fig. 4E). Therefore, alveolar macrophage differences in WT and AQP5^−/− mice were dependent on the length of CS exposure. After 4 weeks of CS exposure, the number of BAL lymphocytes was not different in any of the four groups (Fig. 4E).

With persistent reduction in alveolar neutrophil abundance after sub-acute and chronic CS exposure as well as conferred protection from CS-induced structural changes in AQP5^−/− mice, we assessed whether the presence of epithelial AQP5 directly impacts neutrophil migration using A549 cells, a human alveolar epithelial cell line (Fig. 4F). In the presence of an adenocontrol virus, AQP5 in A549 alveolar epithelial cells is undetectable irrespective of CS exposure (control). Using an adenoviral AQP5 overexpression system, we successfully increased AQP5 expression by immunoblot in A549 cells (+AQP5), which does not change with subsequent CS exposure. There were no differences in neutrophil migration with air exposure. After CS exposure, we observed a significant increase in PMN migration across A549 cells expressing AQP5 (+AQP5) compared with A549 without AQP5 expression (control).

The absence of AQP5 decreases signaling for neutrophil migration to the alveolar space

We have shown that the presence of AQP5 in lung epithelial cells alters barrier function and neutrophil transmigration. However, it is not clear whether the changes in migration are entirely due to AQP5 effects on the paracellular space, or if in addition to this, there are changes in epithelial signaling of neutrophils. To better study this, we investigated additional aspects of epithelial recruitment of neutrophils, including expression of key adhesion molecules and signaling proteins, as well as secretion of chemokines, in response to six months of cigarette smoke exposure.

CD11b expression is critical for migration of neutrophils and macrophages to sites of inflammation especially within the lung, where it interacts with ICAM-1 and other immune modulating receptors.^37,38 CD11b expression is sensitive to changes in inflammatory conditions and immune modulators,^39,40 the latter of which are known to be different between the apical and basolateral surfaces of the airway/alveolar epithelium,^37,41 necessitating measurement in each lung compartment. We utilized established macrophage (F4–80⁺) and neutrophil (Gr-1⁺) antibodies to identify each cell population in BAL and lung, and then determined cell-specific CD11b expression by flow cytometry (Fig. S1B).We observed a significant reduction in CD11b expression on BAL neutrophils from 6 mo CS-exposed AQP5^−/− mice compared with BAL neutrophils from 6 mo CS-exposed WT mice and to those from AQP5^−/− and WT air-exposed mice (Fig. 5A). In contrast, CD11b expression was increased to similar levels on BAL macrophages from CS-exposed AQP5^−/− and WT mice compared with those from air-exposed AQP5^−/− and WT mice (Fig. 5B), suggesting neutrophil-specific differences in signaling in AQP5^−/− mice exposed to CS.

Neutrophil alveolar abundance is determined by changes in chemokine gradient, altered migration across pulmonary endothelial and epithelial layers, or altered life span and removal from the alveolar space. To assess for differences in migration across the endothelium, we measured inflammatory cells in the lung interstitial compartment.⁴² Following bronchoalveolar lavage, we flushed the vasculature via the right ventricular outflow tract, extracted the lung, isolated cells into single cell suspension, and assessed them by flow cytometry. The total number of lung interstitial cells were similar between CS-exposed AQP5^−/− and WT mice (not shown). Chronic CS-exposure did not induce differences in the percentage of lung interstitial neutrophils in either strain (Fig. 5B). CD11b expression on lung interstitial neutrophils was significantly higher after CS exposure in both strains, but not different between CS-exposed WT and AQP5^−/− mice (Fig. 5B). This data demonstrates that significantly fewer neutrophils are present in the alveolar space of CS-exposed AQP5^−/− mice despite a similar number of neutrophils with similar expression of CD11b in the lung interstitium compared with CS-exposed WT mice.

To determine if differences in neutrophil migration across the alveolar space of AQP5^−/− or WT mice were influenced by differences in ICAM-1, a known receptor of CD11b, we measured it after 6 mo of CS exposure by immunofluorescence (Fig. 6A). ICAM-1 is expressed by several different lung structural cells, so changes in abundance in the whole lung do not necessarily reflect differences in specific cell-types. WT mice exposed to CS had increased alveolar and airway epithelial cell ICAM-1 (labeled red) expression compared with WT mice exposed to air. When enumerated as a ratio of total cells (defined by labeled nuclei), ICAM-1 expression on airway and alveolar epithelial cells from WT mice was significantly increased compared with CS-exposed AQP5^−/− mice, and compared with both air-exposed groups. CS-exposed AQP5^−/− mice had similar ICAM-1 expression compared with air-exposed AQP5^−/− mice. Collectively, this data demonstrates an AQP5-specific modulation of adhesion molecules in the context of cigarette smoke exposure that may contribute to differences in neutrophil migration.

As an additional determinant of alveolar neutrophil abundance, we measured chemokine secretion into the alveolar space focusing on cytokines secreted by epithelial cells. LPS-induced chemokine (LIX) levels were altered with CS exposure (Fig. 7A). In WT mice, there was a significant increase in LIX with CS exposure; however in AQP5^−/− mice, there was a significant decrease in LIX with CS exposure, and CS-exposed AQP5^−/− mice secreted significantly less LIX than their WT counterparts. Keratinocyte chemokine (KC) is secreted by epithelial cells, but also inflammatory cells including macrophages. KC did not significantly change with chronic cigarette smoke and was not different between WT and AQP5^−/− at baseline for the mice sampled (Fig. 7B).

Animal use and care

Male AQP5^−/− mice (on a C57BL/6 background) and wild type (WT) littermates were bred and housed at the Johns Hopkins University Asthma and Allergy Center. Experiments were conducted under a protocol approved by the Johns Hopkins Animal Care and Use Committee.

Murine smoke exposure

WT (littermates) and AQP5^−/− mice were exposed in the Johns Hopkins smoke exposure core.^24,33,64 Mice were exposed to cigarette smoke 5 h/day, 5 d/week for 4 weeks or 6 mo by burning 3R4F reference cigarettes (2.45 mg nicotine/cigarette; Tobacco Research Institute, University of Kentucky) using a smoking machine (Model TE-10, Teague Enterprises).^24,64

Animal harvesting

Mice were anesthetized with intraperitoneal ketamine/acetylpromazine (150/13.5 mg/kg) prior to harvest. At specified time points after cigarette smoke or air exposure, 7–10 animals from various groups were anesthetized and killed by exsanguination from the inferior vena cava. The lungs were perfused free of blood with 1 ml of phosphate-buffered saline (PBS) unless otherwise specified.

Analysis of bronchoalveolar lavage (BAL)

BAL was obtained, total and differential cells counted, and BAL protein assessed as previously described.⁶⁵ Keratinocyte-derived chemokine (KC) and LPS-induced chemokine (LIX) were measured in BAL and culture medium by ELISA.

Evans blue dye (EBD) assay

Lung histology and morphometry

Left lungs were collected and processed for histology^29,66 and morphometry³³ by inflating to 25 cm H₂O with 0.6% agarose, isolated, sectioned (5 μm), and stained with hematoxylin and formalin (H&E). Fifteen images per mouse were captured at 100X (Nikon Eclipse 50i), and mean linear intercept (MLI) and surface area of airspace wall per unit of lung volume (S/V) ratio were determined by computer-assisted morphometry using a macro designed (courtesy of Rubin Tuder) with MetaMorph software (Molecular Devices).²⁹ MLI was determined by dividing the length of a line drawn across the lung section by the total number of intercepts encountered, serving as a measure of interalveolar septal distance and as an index of alveolar size. S/V was determined by examining 10 random fields per lung using Weibel’s test grid and the formula described by Chalkley,^31,32 serving as measure of septal loss and as an index for alveolar destruction.^29,30

Lung immunofluorescence

Right lungs were collected, sectioned onto slides, and submerged into 4% formaldehyde/10% formalin in PBS (< 48 h, room temp), and then heated (15 min, 60°C). Slides were deparafinated and hydrated with xylene (2×, 1 min), 100% EtOH (2×, 1 min), 95% EtOH (2×, 1 min), 80% EtOH (1x, 1 min), and then water (1x, 1 min). Citrate buffer (fresh, salts) was added to the slides (20 min, 95$˚$C), followed by passive cooling to room temperature. Samples were lined with hydrophobic pen, then blocked (20% serum, 1% BSA, 0.5% Tween-20 in PBS), rinsed (0.3% Triton X-100, 1% BSA), and covered with primary antibody for E-cadherin (1:200, Cell Signaling, #3195S) or ICAM-1 (1:200, Biolegend, #116110) in PBS with 0.3% Triton X-100, 1% BSA for 18 h (4°C). We then washed (2×, 10mins, 0.3% Triton X-100, 1% BSA), and added a secondary antibody (1:400, for E-cadherin-Ax488 Donkey Anti-rabbit IgG, for ICAM-1 Ax549 Donkey anti-mouse IgG, Invitrogen) diluted in 0.3% Triton X-100, 1% BSA (2 h, 25°C) in the dark. SPC primary antibody was diluted in 4% serum/DPBS (1:100, Santa Cruz, Cat # 13979), washed (3×,10 min, DPBS), biotinylated (1:1000 Biotin goat anti-rabbit, Invitrogen, Cat # B-2770), washed again (2×, 5 min, DPBS), and then we added a secondary antibody (1:400, Alexa Fluor 488 streptavidin antibody, Invitrogen, Cat # S-11223) diluted in 4% serum/DPBS (2 h, 25°C) in the dark. After subsequent rinse (2×, 10 min in PBS, then 1×, 10 min in TBS (pH 8.6)), slides were covered with Fluoromount, sealed, and stored at 4°C. Quantification: ImageJ software by quantifying the ratio of mean fluorescence intensity between the protein of interest and the intensity of the nuclei in the same region of interest. This was done from a minimum of 3 different animals per condition.

In vitro smoke exposure and neutrophil migration

A549 cells (ATCC) were grown on collagen-coated inserts at 37°C with 5% CO₂ in media (450 mL F12K, 50 mL FBS). We used a Vitrocell smoke chamber to deliver tobacco to A549 cells to mimic tobacco smoke exposure of human epithelium in smokers. The cells were cultured on a 6-well insert as described above allowing medium on the bottom of the porous insert, while the top was exposed to either air (control) or cigarette smoke. We exposed cells to 2 cigarettes (7 min/cigarette). After exposure, the inserts were inverted into a larger dish. Freshly isolated peripheral blood PMNs (courtesy of Bruce Bochner) were labeled with a commercially available fluorescent probe (carboxyfluorescein diacetate (CFDA); excitation/emission spectra similar to FITC to allow PMN tracking.⁶⁷ Media containing the labeled PMNs were applied to the basolateral membrane of the inverted insert. After 4 h, we collected the basolateral media; inserts were collected and lysed to determine sample fluorescence to objectively quantify PMN transmigration.

Immunoblotting

Lungs or A549 cells were harvested and lysed in RIPA buffer (50 mM TRIS-HCl pH8, 150 mM NaCl, 1% Triton X-100 + Roche Protease Inhibitor, 15 min),²⁸ followed by spin, and Bicinchoninic acid assay ((BCA) Pierce) for total protein in the whole cell lysate. Normalized proteins were loaded onto 10% gel. We then blocked with 4% BSA for 30 min, and incubated with SPC antibody (Santa Cruz Antibody, Cat # 13979), washed (3×) with TBST, then incubated with secondary in 4% BSA (Bio-Rad, Cat # 170–6515) for an hour. Followed by wash and ECL with HyGLO (Denville Scientific, Cat #E2400). Antibodies to the carboxyl-terminus of human AQP5 were generated and probed for as before.²⁸

Flow cytometry

Cells were processed for flow and analyzed as previously described,⁶⁸ using the following antibodies (or relevant isotypes): anti-LY6G-FITC, anti-CD11b-APCe780, anti-F4/80-APC, anti-CD86-Pacblue, anti-CD326-APC Cy7, anti-T1α-APC, and anti-K_i-67-PE. Mean fluorescence intensity (MFI) was based on the geometric mean expression for a particular marker among cells gated by positive expression.

Statistical analysis

All values are reported as mean ± SEM except Figure 3E, where individual values and the median for each group are shown. Multiple groups were compared using one-way ANOVA with Bonferroni t-test for multiple pairwise comparisons when data are normally distributed or with Kruskal-Wallis assessment on ranks when data are not normally distributed. Two groups were compared using the student’s t-test or Mann-Whitney rank sum test when sample size or variance was not equal. Statistical analysis was performed using Sigmaplot 11.0 (Systat Software). A p < 0.05 was used for significance.