Proteome of airway surface liquid and mucus in newborn wildtype and cystic fibrosis piglets

Newborn piglets of both sexes were used. Breeding male and female heterozygous CF carrier animals from a previously described CF pig model [10] generated CF and WT littermate piglets. Induction of birth, genotyping, euthanasia and dissection of the airway tree were carried out as described [3]. Within 24 h of birth, airways were placed in a 50 ml tube with Perfadex® solution, pH 7.2 (XVIVO Perfusion, Gothenburg, Sweden) and shipped at 4 °C overnight to Gothenburg.

The experiments were performed in explanted airway tissue from newborn CF piglets [10] and WT littermates removed and dissected directly after euthanasia of the piglets. The complete airway tree from larynx to distal bronchioles was transported overnight in Perfadex buffer. The airway tree was intact as the airway lumen was filled with air at mounting despite transport in liquid. The distal trachea and primary bronchi were excised, opened along the dorsal smooth muscle and mounted in a Sylgard-coated Petri dish, using insect needles, as described previously [5]. The mounted trachea was 30 mm in length and 10–12 mm across, with the bifurcation at the very distal end. For unstimulated samples, 500 µl aerated Krebs-Glucose buffer (116 mM NaCl, 1.3 mM CaCl₂, 3.6 mM KCl, 1.4 mM KH₂PO₄, 23 mM NaHCO₃, 1.2 mM MgSO₄, 10 mM D-glucose, 5.1 mM Na-glutamate, and 5.7 mM Na-pyruvate, pH 7.4, gassed with 95% O₂/5% CO₂) was added and for stimulation with carbachol, 100 µl aerated Krebs-Glucose buffer (pH 7.4) containing 10 µM carbachol was added to the luminal surface. The tissue was incubated at 37ºC on a heating block for 10 min. Then, 400 µl aerated Krebs-Glucose buffer (pH 7.4) was added to the proximal end of the carbachol-stimulated tissue, the Petri dish placed on a platform heated to 37 °C and tilted 20º. Importantly, the cephalad end of the trachea was placed at the highest point of the tilted platform, to allow the buffer to drain from proximal trachea to bronchi (Fig. 1A). Fractions were collected from the lowest point, between the primary bronchi (Fig. 1A, B, Fraction 1). After collection, 500 µl aerated Krebs-Glucose buffer (pH 7.4) was added to the topmost part of the trachea (Fig. 1A) and fraction 2–4 were collected with 10 min intervals (Fig. 1B). After collection of fraction 4, 500 µl aerated Krebs-Glucose buffer (pH 7.4) containing 0.4 mM Alcian blue 8GX was added to the topmost part of the trachea on the tilted platform to stain the mucus, resulting in staining of bundled mucus strands and 10 min later, fraction 5 was collected by pipetting on the airway surface to aspirate the mucus.

The collected fractions were processed for mass spectrometry as described previously [11]. A total number of 80 samples were prepared, comprising five fractions collected out of 10 WT piglet tracheas (4 non-stimulated and 6 stimulated) and 6 CF piglet tracheas (3 non-stimulated and 3 stimulated). In short, samples were mixed overnight at 37ºC with 250 µl of reducing buffer (6 M guanidinium hydrochloride [GuHCl], 5 mM EDTA, 0.1 M DTT in 0.1 M Tris–HCl pH 8.5) and 100 µl of the reduced sample were processed by a modified filter-aided sample preparation (FASP) protocol using 6 M GuHCl [12]. The enzymes LysC (10 ng/µl, Wako Chemicals, Neuss, Germany) and trypsin (10 ng/µl, Promega, Madison, WI) were employed to digest the proteins into peptides. These were purified through C18 StageTips [13] and submitted to nano-liquid chromatography tandem mass spectrometry (nanoLC-MS/MS) on a Q-Exactive HF mass spectrometer (Thermo Fisher Scientific, Waltham, MA) under the conditions previously described [11].

The resulting raw spectra files were analyzed with MaxQuant (version 1.5.7.4, RRID:SCR_014485) against the UniProt pig proteome (version of 20180301). LysC and trypsin were set for the enzyme specificity, Cys carbamidomethylation was selected as fixed modification, and Met oxidation and protein N-terminal acetylation were ascribed to the variable modifications. The false discovery rate (FDR) was set at 1%. A total of 2328 proteins were identified and filtered, eliminating those identified only by site, reverse identifications and potential contaminants. Protein levels were calculated by normalizing individual peptide intensities to the total intensity of all identified proteins. These are presented in Additional file 2: Table S1. Proteins were filtered for presence of predicted signal sequence and lack of transmembrane domain based on information available in UniProt to identify secreted proteins and these are presented in Additional file 3: Table S2.

Tissue pieces from piglet tracheas were fixed in 4% formalin or Carnoy (60% methanol, 30% chloroform and 10% glacial acetic acid; Cat# 322415, 372978 and 695092 respectively, Sigma-Aldrich, St. Louis, MO), embedded in paraffin and cut in 4 μm thick sections. Sections were dewaxed with Xylene (Cat# 534056 Sigma-Aldrich, St. Louis, MO) and hydrated in decreasing concentrations of ethanol. Antigen retrieval was performed by heating to 100ºC for 20 min and cooling for 20 min in room temperature in 0.01 M citric buffer pH 6.0. Unspecific binding was blocked for 60 min with 3% donkey serum in Tris-buffered saline (TBS) and permeabilized with 0.1% Triton X-l00 at ambient temperature. Primary antibodies used were mouse monoclonal anti-MUC5AC clone 45M1 at 1:1000 (Cat# MA1-38224, RRID:AB_2146842, ThermoFisher Scientific, Waltham, MA), goat polyclonal anti-AGR2 (T-17) at 1:100 (Cat# sc-54561, RRID:AB_2225107, Santa Cruz Biotechnology, Dallas, TX), mouse monoclonal anti-ERp57 at 1:2000 (PDIA3, Cat# ab13506, RRID:AB_1140700, Abcam, Cambridge, UK), mouse monoclonal anti-AGP at 1:200 (ORM1, Cat# ab239473, Abcam, Cambridge, UK), were incubated sequentially overnight with custom made rabbit anti-human anti-MUC5B directed against the D3 domain at 1:60,000 or rabbit monoclonal anti-CD31 at 1:1000 (Cat# ab182981, Abcam, Cambridge, UK) in blocking solution at 4 °C. The custom-made anti-MUC5B antibody was characterized previously [14]. Secondary antibodies according to species origin of the primary antibodies were donkey anti-rabbit Alexa Fluor 488 (Cat# A32790, RRID:AB_2762833), donkey anti-goat Alexa Fluor 555 (Cat# A32816, RRID:AB_2762839, Thermo Fisher Scientific, Waltham, MA) and donkey anti-mouse Alexa Fluor 647 (Cat# A-31571, AB_162542, Thermo Fisher Scientific, Waltham, MA). Secondary antibodies were incubated in blocking solution for two hours at ambient temperature in the dark. Nuclei were counterstained with Hoechst 34580 (Cat# 565877, RRID:AB_2869723, BD Biosciences, San Jose, CA) and sections were mounted with Prolong gold mounting medium (Cat# P36934, RRID:SCR_015961, Thermo Fisher Scientific, Waltham, MA). Images were acquired with an upright LSM 900 Axio Examiner Z1 confocal imaging system, Plan-Apochromat 63x/1.4 Oil DIC M27 (Cat# 420782-9900-799), Airyscan 2 detector and Zen blue software (RRID:SCR_013672, Carl Zeiss, Oberkochen, Germany). Images were processed with Zen blue (airyscan processing) and Imaris software, version 9 (Oxford Instruments, Abingdon, U.K.).

The Perseus software (version 1.5.6.0, RRID:SCR_015753) was employed for data filtration and clustering. Replacement of missing values was done by simulation of the normal distribution. Hierarchical clustering was performed by the Euclidean distance, with average linkage. The results were graphically plotted as a heat map. Clustered proteins are presented in Additional file 4: Table S3.

Data are presented as median with interquartile range and all data points are shown. Each piglet represents one data point, n = number of piglets, unless otherwise described in the figure legends. Pooling was performed to compare washable fractions to mucus fractions, compare WT to CF piglets or to demonstrate differences between stimulated and unstimulated fractions. Number of piglets in each group and analysis method are presented in each figure legend. Statistical analyses were performed using GraphPad Prism version 9 (RRID:SCR_002798, GraphPad, San Diego, CA). Differences were assessed with two-tailed non-parametric Mann–Whitney test for comparison of two groups or Kruskal–Wallis and Dunn’s multiple comparisons test for comparisons between multiple groups. Significance was defined as P ≤ 0.05. No methods of randomization or determination of sample size were used due to limited access to piglet tissue. The experiments were not performed blinded due to the unmistakable morphology of newborn CF piglet tracheas.

As expected, the Alcian blue stained mucus collected in fraction 5 contained the two secreted gel-forming mucins MUC5AC and MUC5B. Levels of MUC5AC in both unstimulated and stimulated WT piglets were high in fraction 5 and low in fraction 1–4 (Fig. 2A). The same pattern was observed for MUC5B (Fig. 2B). Immunostainings in paraffin sections confirmed that goblet cells in the airway surface epithelium secrete MUC5AC (red) and submucosal glands secrete MUC5B (green) (Fig. 2C). Both unstimulated and stimulated piglets had higher amounts of MUC5AC in fraction 5 (mucus fraction) than fraction 4 (the last washable fraction) (Fig. 2D). Similarly, MUC5B was lower in fraction 4 than fraction 5 (Fig. 2E). However, only the difference between the unstimulated groups reached significance due to outlier values for both MUC5AC and MUC5B in the stimulated samples (dark green and red, Fig. 2A, B, F, G) originating from the same piglet, likely due to detached mucus slipping into fraction 4. Previous experiments utilizing internal standard heavy peptides for pig MUC5AC and 5B have shown that the relative intensities of the peptides for each mucin roughly reflect the relative amounts of each molecule [5]. The relative intensities of peptides for MUC5AC and MUC5B were different, indicating that the MUC5AC levels in fraction 5 was higher for both unstimulated and stimulated WT and CF piglets (Fig. 2F, G). Pooling normalized intensities in fractions 1–4 for unstimulated and stimulated WT piglets suggested more MUC5AC than MUC5B under both conditions (Fig. 2G). Together these results imply that surface goblet cell derived MUC5AC was prominent in both washable and mucus fractions and that carbachol stimulation induced secretion of predominantly MUC5AC.

In addition to the two mucins, the fraction 5 cluster contained other proteins, among others the protein disulfide isomerases (PDIs), known to be important for protein assembly in the endoplasmatic reticulum. Their presence in fraction 5 suggests that they follow the mucins during secretion. One of these, anterior gradient protein 2 (AGR2; Fig. 3A), is a small protein of 175 amino acids known to be associated with intestinal [15, 16] and airway mucus [17]. In stimulated WT piglets, there was more AGR2 in fraction 5 compared to fraction 4. Tissue staining for AGR2 suggested the highest prevalence in submucosal glands (Fig. 2B, negative controls in Additional file 1: Fig. S1A and B), as has been previously demonstrated [17]. Staining in the surface epithelium was lower than in submucosal glands (Additional file 1: Fig. S2A). The protein PDIA1, the beta subunit of prolyl 4-hydroxylase, was also higher in fraction 5 than fraction 4 from stimulated WT piglets (Fig. 3C). Another PDI, PDIA3, had a slightly different pattern, with increasing levels from fraction 1 to 5 (Fig. 3D). Immunostaining demonstrated that PDIA3 was mainly present in the submucosal glands and interestingly only produced by a subset of MUC5B-positive cells (Fig. 3E, negative controls in Additional file 1: Fig. S1C–F). Surface staining was negative (Additional file 1: Fig. S2B). These PDIs were found on the airway surface, but whether they have any extracellular function is unknown. The fraction 5 cluster also contained the antimicrobial glycoprotein lactoferrin (LTF), known to be produced by submucosal glands [18]. Lactoferrin displayed similar patterns in unstimulated and stimulated WT and CF piglets (Fig. 3F). The levels of lactoferrin were higher in fraction 5 than fraction 1, when analyzing pooled unstimulated and stimulated samples from both WT and CF piglets (Fig. 3G).

The tracheobronchial surface was sequentially washed and the collected material called washable fractions 1–4. The material was secreted from the epithelial cells or submucosal glands and these proteins were lower in the mucus fraction. The first wash differed from the rest as this likely contained proteins from more distal airways, secreted when the airway tree was still intact and transported with mucociliary transport to the distal trachea. We thus focused on the fraction 2–4 cluster (Fig. 1C), which contained the ubiquitously present extracellular molecular chaperone clusterin (CLU) with numerous suggested functions. In line with our observation, it has been found in the epithelium of human airways [19]. In both WT and CF piglets we detected an increasing trend from fraction 1 to 4, but relatively lower levels of clusterin in fraction 5 (Fig. 4A). More clusterin was found in fraction 4 compared to fraction 1 in pooled values from WT piglets (Fig. 4B), suggesting that clusterin was newly secreted by the cells in the explant or alternatively that clusterin was more tightly associated with structures on the airway surface or the mucus than proteins with higher normalized intensity in fraction 1. Slightly higher levels of clusterin were found in CF compared to WT piglets (Fig. 4C).

Alpha-1 antitrypsin (SERPINA1), an endogenous inhibitor of neutrophil elastase is produced in the lung where it controls inflammatory responses, evident from the hereditary disease alpha-1 antitrypsin deficiency [20]. The inhibitor was previously demonstrated not only in humans [21] but also in bronchoalveolar lavage fluid from newborn piglets [8]. We detected alpha-1 antitrypsin in fraction 1–4 from WT and CF piglets (Fig. 4D), but the normalized intensity was low in the mucus fraction (Fig. 4D). The pooled normalized intensity from fraction 1–4 was higher in CF than WT piglets (Fig. 4E).

The acute-phase protein orosomucoid 1 (ORM1), also called alpha-1-acid glycoprotein, clustered to fraction 2–4 and was detected in fraction 1–4 in unstimulated and stimulated WT and CF piglets. There were relatively low levels of orosomucoid 1 in the mucus fraction in both WT and CF piglets (Fig. 4F), but the normalized intensities were higher in CF than WT piglets (Fig. 4G). Immunolocalization of orosomucoid 1 identified the protein on the airway surface, but not inside the epithelial cells (Fig. 4H). The protein was also detected inside small capillaries in the airway submucosa, positive for the endothelial cell marker CD31 (Fig. 4I, negative controls in Additional file 1: Fig. S1C–F).

Further investigation of proteins clustering to fraction 1–4 revealed low levels of intracellular proteins localized to the endoplasmatic reticulum and Golgi (Additional file 2: Tables S1 and Additional file 3: S2). Examples of such proteins were calumenin, involved in CFTR maturation [22], reticulocalbin-3 from the same family, a molecular chaperone important for biosynthesis of pulmonary surfactant-associated proteins [23] and a molecular chaperone involved in collagen production called SERPINH1 or heat shock protein 47 [24]. Haptoglobin, which binds free hemoglobin to inhibit its oxidative activity [25] and porcine SERPINA3-2 or alpha-1-antichymotrypsin 2 also clustered to fraction 1–4. The closest human ortholog is alpha-1-antochymotrypsin, an acute phase protein induced during inflammation encoded by the gene SERPINA3 [26].

The washable fractions from CF piglets contained proteins forming an additional cluster not present in fraction 1–4 (Fig. 1C). This cluster included fetuin-A (alpha-2-Heremans Schmid-glycoprotein, AHSG), fetuin-B (FETUB) and apolipoprotein A1 (APOA1). Fetuin-A decreased from fraction 1 to 4 (Fig. 5A). The normalized intensities of pooled fraction 1–4 were higher in unstimulated CF than WT piglets (Fig. 5B), also observed in pooled fraction 1 (Fig. 5C). Furthermore, the levels were lower in fraction 5 compared to fraction 1 (Fig. 5D). The normalized intensity for fetuin-B followed a similar pattern to fetuin-A (Fig. 5E) with higher levels in CF than WT piglets (Fig. 5F) and also more fetuin-B in fractions 1–4 than fraction 5 of both WT and CF piglets (Fig. 5G, H). Apolipoprotein A1 also had a similar distribution pattern as fetuin-A (Fig. 5I) and with relatively more apolipoprotein A1 in CF than WT piglets (Fig. 5J) and more in fractions 1–4 than fraction 5 (Fig. 5K, L).