Secretory Cells Are the Primary Source of pIgR in Small Airways

Abstract

Loss of secretory IgA (SIgA) is common in chronic obstructive pulmonary disease (COPD) small airways and likely contributes to disease progression. We hypothesized that loss of SIgA results from reduced expression of pIgR (polymeric immunoglobulin receptor), a chaperone protein needed for SIgA transcytosis, in the COPD small airway epithelium. pIgR-expressing cells were defined and quantified at single-cell resolution in human airways using RNA in situ hybridization, immunostaining, and single-cell RNA sequencing. Complementary studies in mice used immunostaining, primary murine tracheal epithelial cell culture, and transgenic mice with secretory or ciliated cell–specific knockout of pIgR. SIgA degradation by human neutrophil elastase or secreted bacterial proteases from nontypeable Haemophilus influenzae was evaluated in vitro. We found that secretory cells are the predominant cell type responsible for pIgR expression in human and murine airways. Loss of SIgA in small airways was not associated with a reduction in secretory cells but rather a reduction in pIgR protein expression despite intact PIGR mRNA expression. Neutrophil elastase and nontypeable H. influenzae–secreted proteases are both capable of degrading SIgA in vitro and may also contribute to a deficient SIgA immunobarrier in COPD. Loss of the SIgA immunobarrier in small airways of patients with severe COPD is complex and likely results from both pIgR-dependent defects in IgA transcytosis and SIgA degradation.

Clinical Relevance

Localized secretory IgA (SIgA) deficiency in small airways is an established driver of chronic obstructive pulmonary disease pathogenesis, but the mechanism of loss remains unclear. We explored the mechanism of SIgA loss in chronic obstructive pulmonary disease, which could inform the development of therapeutics aimed at restoring the SIgA immunobarrier.

Throughout life the lungs are bombarded by inhaled particulates, microorganisms, and endogenous bacteria aspirated from the upper respiratory tract (1). Many such irritants are deposited on the surface of small airways, where bulk airflow ceases and gas transfer proceeds through diffusion (2). To facilitate homeostasis, the small airway epithelium maintains a multifaceted defense system that protects against microbial invasion while limiting potentially injurious inflammatory responses (3, 4). A key component of this system is secretory IgA (SIgA), which lines the lumen of small airways and other mucosal surfaces (5–9). SIgA opsonizes and agglutinates microorganisms and noninfectious irritants to facilitate their clearance through the mucociliary escalator (6). Through a process known as immune exclusion, SIgA limits inflammation by masking pathogen-associated molecular patterns and preventing them from activating inflammatory signaling cascades in the airway epithelium (6).

In 2001, Pilette and colleagues made the seminal discovery of reduced secretory component (SC; a component of SIgA) in the airways of patients with chronic obstructive pulmonary disease (COPD) (10), a common and often fatal lung disease associated with tobacco smoking (11). Furthermore, they reported that loss of SC is associated with increased neutrophilic inflammation (10). We extended these findings to show that loss of SIgA correlates with increased bacterial invasion into small airway walls, COPD-like innate and adaptive immune activation, and small airway wall fibrosis in patients with COPD (12, 13). Furthermore, we reported that SIgA-deficient mice spontaneously develop progressive lung inflammation, small airway wall thickening, and emphysema, similar to patients with COPD (14–16), suggesting that loss of SIgA immunobarrier directly contributes to COPD pathogenesis.

Regulation of SIgA on mucosal surfaces is a complex process involving multiple cell types and processes (5). Dimeric IgA (dIgA) is made by plasma cells that reside beneath the airway epithelium in the lamina propria. To cross the airway epithelium to the luminal surface, dIgA covalently binds to the pIgR (polymeric immunoglobulin receptor) on the basolateral surface of the airway epithelium. pIgR/dIgA complexes are endocytosed and transported to the apical surface through vesicular trafficking (17). At the apical surface, an endoproteolytic cleavage event releases dIgA and the extracellular portion of pIgR (SC) to form SIgA. At the airway surface, SIgA is susceptible to cleavage by host and bacterial proteases (18–24). Thus, the amount of SIgA present on the airway surface depends on rates of production, transport, and degradation.

To better understand regulation of the SIgA barrier in the lungs, we defined cell type–specific expression of pIgR in human and murine small airways using complementary in situ, in vitro, and in vivo approaches and evaluated potential mechanisms accounting for an impaired SIgA barrier in small airways of patients with COPD.

Some of the results of these studies have been previously reported in preprint form (https://doi.org/10.1101/2021.11.10.467794).

Methods

Human Samples

The explanted lungs of patients with COPD were obtained after informed consent according to Institutional Review Board (IRB)–approved protocols from Vanderbilt University Medical Center and Norton Thoracic Institute (Vanderbilt IRB #060165 and #171657, Norton IRB #20181836). Control samples were obtained from deceased organ donors and were exempt from IRB review. Demographic and clinical information for human samples is provided in Table E1 in the data supplement. Demographic and clinical information for control lungs used for single-cell RNA sequencing (scRNA-seq) have been published previously but are included here for reference (25).

Mouse Models

All mouse experiments were performed according to a protocol approved by the Institutional Animal Care and Use Committee at Vanderbilt University Medical Center. Scgb1a1.Cre^ERT2 mice were generated in the laboratory of Dr. Brigid Hogan (26) and were obtained from The Jackson Laboratory (catalog #16225). Transgene activation was induced with tamoxifen citrate–impregnated (400 mg/kg) chow (catalog #130860; Envigo), as described in Results. FoxJ1.Cre mice were originally developed in the laboratory of Dr. Michael Holtzman (27) and were a gift from the developers. The flox-pIgR mouse line was developed with Ingenious Targeting Laboratory. A targeting vector was generated in which loxP (locus of crossover P1) sites were inserted upstream of pigr exon 3 and downstream of exon 11, which includes the binding domains for pIgR (5). The coding regions of the GFP gene were introduced upstream and downstream of the two loxP sites, such that excision of exons 3–11 via Cre recombinase results in GFP expression. Selection was achieved by introduction of a neomycin selection cassette flanked by FRT (flippase recognition target) sites downstream of exon 11. This construct was electroporated into C57BL/6J FLP (flippase) embryonic stem cells, which were screened for neomycin resistance. Positive cells were injected into BALB/c blastocysts and then surgically inserted into pseudopregnant females. Resulting chimeras with a high percentage of black coat color were mated to C56BL/6J mice to generate germline neodeleted mice, which was confirmed by Southern blot. A map of the targeting vector is included as Figure E3. C57BL/6J mice were purchased from The Jackson Laboratory (catalog #000664) or bred in house after originally being purchased from The Jackson Laboratory. Approximately equal numbers of male and female mice 2 months of age or older were used in all experiments.

scRNA-Seq

Tissue processing

Biopsies from each lung sample were digested in an enzymatic cocktail (Miltenyi Multi Tissue Dissociation Kit or collagenase I/dispase II 1 μg/ml tissue) using a gentleMACS Octo Dissociator (Miltenyi, Inc.). Tissue lysates were serially filtered through sterile gauze (100 μm) and 40-μm sterile filters (Thermo Fisher Scientific). Single-cell suspensions then underwent cell sorting using serial columns (Miltenyi Microbeads, CD235a, CD45) or FACS. At Vanderbilt University Medical Center, CD45⁻ and CD45⁺ populations mixed 2–3:1 were used as input for the generation of scRNA-seq libraries. At the Translational Genomics Research Institute, Calcein-AM was used to stain live cells, and 10,000–15,000 total live cells were sorted directly into the 10X reaction buffer and transferred immediately to the 10X 5′ chip A (10X Genomics).

scRNA-seq library preparation and next-generation sequencing

scRNA-seq libraries were generated using the 10X Chromium platform 5′ library preparation kits (10X Genomics) targeting 10,000 cells per sample. Next-generation sequencing was performed on an Illumina NovaSeq 6000 using Illumina S1 flow cell with paired end reads. Reads with Phred scores <30 were filtered out and 10X Genomics Cell Ranger version 3 (for VU_COPD_29 and VU_COPD_34) or version 5.0.0 (all other samples) was used to align reads to the GRCh38 reference genome.

scRNA-seq analysis

Postalignment analysis was performed using the Seurat (https://satijalab.org/seurat/) version 4.0.1 package in R (https://www.r-project.org/). Quality control filtering was performed using detected genes (minimum 750) and mitochondrial reads percentage (minimum 0%, maximum 15%). All data (COPD and control samples) were jointly normalized and scaled using the SCTransform function in Seurat (28). Epithelial (EPCAM⁺ [epithelial cell adhesion molecule]), stromal (EPCAM⁻/PTPRC⁻ [protein tyrosine phosphatase receptor type C]), and immune (PTPRC⁺) cells were then extracted into independent objects which underwent SCT-Integration (28) followed by principal-component analysis, recursive clustering, uniform manifold approximation and projection embedding, and cell-type annotation. Clusters composed of nonphysiologic marker combinations (i.e., PTPRC⁺/EPCAM⁺ cells) were presumed doublets and excluded from further downstream analysis. Cell type–specific differentiation expression analysis was performed as previously described (25). For this data set, only EPCAM⁺ epithelial populations were examined. The code used to generate the data set is available at https://www.github.com/kropskilab/copd/. The data discussed in this publication have been deposited in the National Center for Biotechnology Information’s Gene Expression Omnibus (GSE196341). (https://www.ncbi.nlm.nih.gov/geo/query/acc.cgi?acc=GSE196341).

RNA In Situ Hybridization

RNA in situ hybridization (RNA-ISH) was performed on formalin-fixed, paraffin-embedded tissue using the RNAscope platform (ACDBio; manual multiplex kit V2). Probes used include PIGR-C1, SCGB3A2 (secretoglobin family 3A member 2)–C2, SCGB1A1 (secretoglobin family 1A member 1)–C3, and FOXJ1 (forkhead box J1)–C2. Positive and negative control probes were also used to validate RNA integrity and signal specificity. TSA Plus fluorophores (fluorescein, Cy3, and Cy5) were purchased from Akoya Biosciences. Fluorophores were used at concentrations between 1:750 and 1:1,500. Slides were imaged on a Nikon spinning disk confocal microscope (Yokogawa CSU-X1 spinning disk head, Andor DU-897 electron multiplying charge-coupled device [EMCCD], high-speed piezo [z] stage and a four-line high-power solid-state laser launch) or on a Keyence BZ-710.

HALO Analysis

Quantification of RNAscope images and pIgR immunostaining was performed using HALO image analysis software version 1-3 with the FISH-IF module (Indica Labs). Only cells within airway epithelia were quantified. DAPI was used to identify individual cells, and a cell radius of 3 μm outside of the DAPI-stained area was set to identify probes/signal within each cell while minimizing probe/signal spillover from adjacent cells or signal on the apical surface of the epithelium. Cells were segmented aggressively to avoid cell overlap. The minimum threshold for probe intensity for RNA-ISH or staining intensity for immunofluorescence and contrast threshold was determined for each set of staining by adjusting settings for best mask fitting using real-time tuning to eliminate nonspecific signal using irrelevant areas outside the airway epithelium (predominantly smooth muscle cells) as the baseline for zero expression/staining. Probes were also segmented aggressively.

Bacterial 16S Ribosomal RNA Detection

Slides were deparaffinized then placed in 20 mM Tris buffer until hybridization of the probes. Probes were diluted to 2 μM in prewarmed hybridization buffer (20 mM Tris, 0.9 M NaCl, 0.01% sodium dodecyl) and placed in a humidified hybridization oven at 46°C for 2 hours. After hybridization, slides were washed in prewarmed fluorescence in situ hybridization wash buffer (20 mM Tris, 225 mM NaCl, 5 mM EDTA) three times before DAPI staining and coverslipping. The eubacteria probe sequence was 5′-Cy3-GCTGCCTCCCGTAGGAGT-3′, and the nonspecific probe sequence was 5′-6FAM-CGACGGAGGGCATCCTCA-3′, as previously used by Macedonia and colleagues (29).

Immunostaining

For formalin-fixed, paraffin-embedded tissue, 5-μM sections were deparaffinized, and then antigen retrieval was performed using citrate buffer at pH 6. Samples were then permeabilized with 0.1% Triton (Sigma) before blocking with 1% BSA in PBS. Antibodies were incubated overnight at 4°C. Antibodies used were human pIgR (rabbit, HPA012012 [Sigma-Aldrich], 1:500 amplified), mouse pIgR (goat, AF2800 [R&D Systems], 1:100 amplified), human FoxJ1 (rabbit, HPA005714 [Sigma-Aldrich], 1:500 amplified), mammal acetylated α-tubulin (mouse, sc-23950 [Santa Cruz Biotechnology], 1:100), human Scgb1a1 (rat, MAB4218 [R&D Systems], 1:50 amplified), mouse Scgb1a1 (rabbit, ab40873 [Abcam], 1:1,000), GFP (chicken, GFP-1010 [Aves Labs], 1:250), human IgA (rabbit, A0262 [Dako], 1:75), and human neutrophil elastase (HNE) (rabbit, 1:100, amplified). For amplified targets, the appropriate horseradish peroxidase–conjugated secondary was then incubated at 1:250 in PBS for 45 minutes to 1 hour, followed by TSA fluorophore treatment (Akoya Biosciences). For unamplified targets, appropriate Cy3- or FITC/488-labeled secondary antibodies were used at 1:100. Images were taken on a Nikon spinning disk confocal microscope (CSU-X1 spinning disk head, DU-897 electron-multiplying charge-coupled device, high-speed piezo [z] stage and a four-line high-power solid-state laser launch) or on a BZ-710.

For pIgR antibody validation, our human pIgR antibody (HPA012012; Sigma-Aldrich) was preincubated with an excess of SIgA (that contains the portion of pIgR that the antibody is targeted toward) for 1 hour before staining of control tissue sections and compared with a nonpreincubated sample in serial sections. A secondary only control was also included in an additional serial section. Identical camera settings were used across all samples.

SIgA Assessment

Airways were binned into SIgA⁺ and SIgA⁻ on the basis of fluorescent signal of IgA on the luminal surface, as previously described (13).

Western Blots

Whole-tissue lysates were prepared from flash-frozen lung tissue (for flox-pIgR experiments) or whole-cell lysates (for murine tracheal epithelial cell [MTEC] experiment) using radioimmunoprecipitation assay buffer (RIPA buffer) supplemented with protease inhibitors. Protein concentrations were determined using bicinchoninic acid assay, and 15 μg protein (or 8 μg SIgA for in vitro degradation assays) was separated on a 10% acrylamide gel and transferred to a nitrocellulose membrane. IB was performed using antibodies against murine pIgR (goat antimouse, AF2800 [R&D Systems], 1:1,000), β-actin (mouse antimultispecies, A5316 [Sigma-Aldrich], 1:40,000), α-tubulin (1:1,000), IgA heavy chain (rabbit antihuman, A0262 [Dako], 1:1,000), and SC/pIgR (goat antihuman, AF2717 [R&D Systems], 1:500), followed by detection with fluorescent secondary antibodies (1:10,000) using the Odyssey Imaging System (Li-Cor). Densitometry was performed using Image Studio (Li-Cor) and standardized to β-actin.

Cell Culture

MTEC isolation, culture, and DAPT treatment

MTECs were isolated and cultured as described by You and colleagues (30). Briefly, adult C57BL/6J mice were killed and tracheas isolated and split longitudinally to expose the lumen. Tracheas were placed in 1.5 mg/ml Pronase E (Sigma) in Ham’s/F12 containing antibiotics and incubated overnight at 4°C. Cells were dislodged from the tracheas by gentle agitation in Ham’s/F12 containing 10% FBS, collected by centrifugation at 500 × g, and resuspended in MTEC basic medium (30) containing 10% FBS. Cells were then plated onto a Primaria tissue culture dish (catalog #353801; Corning) for 3–4 hours at 37°C with 5% CO₂ to adhere contaminating fibroblasts. Nonadherent cells were then collected and resuspended in MTEC complete media (30) and seeded at a density of 0.8 × 10⁵ onto Transwell inserts (catalog #3470; Corning) precoated with 0.05 mg/ml rat tail type 1 collagen. After cells reached confluence (approximately Days 7–10), they were lifted to generate air–liquid interface (ALI) cultures with MTEC 2% Nu-Serum (Corning)/Retinoic Acid (RA) (Sigma) (30) added only to the lower chamber. For the DAPT experiment, 10 μM DAPT (catalog #120633; Abcam) was added to media in the basolateral chamber for 14 days of ALI culture. Control wells were treated with diluent (DMSO) only.

In vitro SIgA degradation experiments

For the nontypeable Haemophilus influenzae (NTHi) degradation experiment, NTHi strain 1479 (a gift from Dr. Brahm Segal, University of Buffalo) was grown overnight at 37°C on chocolate agar plates (Hardy Diagnostics). A single colony was used to inoculate 20 ml brain–heart infusion media (Sigma-Aldrich) supplemented with 10 μg/ml nicotinamide adenine dinucleotide and 10 μg/ml Hemin (both from Sigma-Aldrich). The culture was incubated for 4 hours at 36°C with constant shaking and then used to inoculate an additional 200 ml liquid media. After another 4 hours of growth, bacteria were pelleted and the conditioned media filtered using a 0.22-μm polyethersulfone filter. This media (25 μl) was incubated with 25 μg (1 μg/μl, 25 μl) SIgA from human colostrum (16-13-090701; Athens Research and Technology). For the HNE experiment, 25 μg SIgA was incubated with 10 μg sputum-derived HNE (1 μg/μl in 0.05 M NaOAc [pH 5] containing 0.1 M NaCl; SE563GI [Elastin Products Co.]) with or without a protease inhibitor (P8340 [Sigma-Aldrich], 1:100).

Statistics and Graphing

Mice were randomly assigned to study groups, and all animals were included in each analysis. Results are presented as mean ± SEM unless otherwise noted. GraphPad Prism 9 was used for statistical testing and graphing. Type of test and P value are noted in each figure legend.

Results

pIgR Is Expressed Primarily by Secretory Cells in Human Airways

Secretory and ciliated cells constitute the majority of epithelial cells lining human small (<2 mm) airways (31). To determine which of these cell types express pIgR, we performed RNA-ISH for PIGR, SCGB1A1 (secretoglobin family 1A member 1; a marker of secretory cells), and FOXJ1 (forkhead box J1; a marker of ciliated cells) in small airways from deceased lung donors without known respiratory disease whose lungs were rejected for transplantation (e.g., control subjects). Fluorescence microscopy showed areas of colocalization between PIGR and SCGB1A1 but minimal overlap between PIGR and FOXJ1 (Figure 1A). Using HALO image analysis software, we quantified each probe 3 μm around DAPI-stained nuclei in segmented airway epithelial cells (see Figure E1), thus allowing us to define FOXJ1⁺ and SCGB1A1⁺ cells with minimal overlap between markers (Figure 1B, first two panels). Using this technique, we observed higher probe expression for PIGR in SCGB1A1⁺ secretory cells compared with FOXJ1⁺ ciliated cells (Figure 1B, third panel). Furthermore, by using two different markers of secretory cells (SCGB1A1 and SCGB3A2) (32), we were able to account for 83% of PIGR⁺ cells (Figures 1C and 1D), suggesting that these cells are the dominant cell type responsible for PIGR expression in small airways. To confirm these findings at the protein level, we used immunofluorescence microscopy in control lung tissue. After validating the specificity of our pIgR antibody (see Figure E2), we performed a triple stain for pIgR, Scgb1a1 and FoxJ1, which showed that pIgR protein localized in cells that contained Scgb1a1 but not FoxJ1 (Figure 1E).

Figure 1.

PIGR/pIgR (polymeric immunoglobulin receptor) is expressed primarily by secretory cells in human lung sections. (A) RNA in situ hybridization for PIGR (red), the secretory cell marker SCGB1A1 (secretoglobin family 1A member 1) (green, left panel), the ciliated cell marker FOXJ1 (forkhead box J1) (green, right panel), and DAPI (white) in small airways from deceased lung donors without chronic respiratory disease. Scale bars, 50 μm. Insets depict 2.5× magnification. (B) FOXJ1, SCGB1A1, and PIGR expression (probe counts) in cells defined as SCGB1A1⁺ and FOXJ1⁺; anything less than 10 probes per cell was considered background or overlap from adjacent cells. Error bars represent median values and 95% confidence interval. *P < 0.0001 (Mann-Whitney test). (C) RNA in situ hybridization showing heterogeneous expression of PIGR (blue), SCGB1A1 (green), and SCGB3A2 (secretoglobin family 3A member 2) (red) and the combination of each marker in a small airway from a control patient. Scale bar, 50 μm. (D) Percentage of cells expressing PIGR alone or PIGR with secretory cell markers SCGB1A1 and/or SCGB3A2; n = 10,908 cells (4,571 PIGR expressing) from 17 airways in six deceased lung donors without chronic respiratory disease. (E) Immunostaining for pIgR (red), Scgb1a1 (green, left panel), FoxJ1 (green, center panel), and all three markers (pIgR, red; Scgb1a1, green; FoxJ1, blue; DAPI, white; right panel) in small airways from deceased lung donors without chronic respiratory disease. Scale bars, 50 μm. Insets depict 2× magnification.

To gain a more comprehensive understanding of the pattern of PIGR expression across the respiratory epithelium, we performed scRNA-seq of distal portions of explanted lungs from patients with COPD (n = 11) and analyzed these samples alongside our previously published scRNA-seq data from analogous regions of lung from nondiseased control subjects (25). All samples were jointly normalized and scaled before graph-based clustering and manual annotation using canonical markers, as previously described (25) (Figures 2A–2C; see Figure E3). Graph-based clustering yielded three distinct populations of cells bearing markers typical of secretory cells that could be differentiated by expression of SCGB1A1, SCGB3A2, or both markers. Similar to what was observed by RNA-ISH and immunostaining, PIGR expression was highest in these secretory cell populations in both patients with COPD and nondiseased control subjects (Figure 2D). The amount of expression was similar between COPD and control samples except in ciliated cells and SCGB1A1⁺/SCGB3A2⁺ secretory cells, for which expression was higher in COPD samples (Figure 2D). Taken together, these data suggest that secretory cells are the dominant cell type responsible for pIgR expression in human airways and that loss of SIgA in COPD airways is not explained by decreased PIGR expression.

Figure 2.

PIGR is expressed primarily by secretory cells in human lung tissue by single-cell RNA sequencing (scRNA-seq). (A) Tissue processing work flow for scRNA-seq experiments. (B) Uniform manifold approximation and projection (UMAP)–embedded image showing cell clusters derived from nonbiased, graph-based clustering of EPCAM⁺ (epithelial cell adhesion molecule) cells derived from the distal lungs of patients with chronic obstructive pulmonary disease (COPD) and control patients. (C) UMAP-embedded image showing integration of cells according to clinical status. (D) Violin plots show PIGR expression in COPD explants and organ donors without known chronic respiratory disease (e.g., control subjects). The height of each violin corresponds to the degree of expression/cell, and the width of the violin corresponds to the number of expressing cells for a given amount of expression. *P < 0.001 and **P < 0.0001 (negative binomial regression). AT1 = alveolar type 1; AT2 = alveolar type 2; KRT5 = keratin 5; KRT17 = keratin 17; TGen = Translational Genomics Research Institute; VUMC = Vanderbilt University Medical Center.

pIgR Is Expressed Primarily by Secretory Cells in Murine Airways

Similar to human airways, we found that pIgR colocalized with Scgb1a1 by immunostaining but not the ciliated cell marker acetylated α-tubulin in wild-type C57Bl6/J mice (Figure 3A). In addition, we isolated tracheal epithelial cells from wild-type mice and treated them with DAPT, an inhibitor of secretory cell maturation (33), during terminal differentiation in ALI culture (Figure 3B). DAPT markedly reduced pIgR expression in whole-cell lysates (Figures 3C and 3D), consistent with the idea that secretory cells are the dominant cell type responsible for pIgR expression. Concentrations of α-tubulin did not differ between DAPT and control samples, suggesting that loss of pIgR was not due to impaired differentiation of ciliated cells (Figure 3C).

Figure 3.

pIgR is expressed primarily by secretory cells in murine airways. (A) Immunostaining for pIgR (red), ciliated cell marker acetylated-α tubulin (white), and secretory cell marker Scgb1a1 (green) in airways from wild-type C57BL6/J mice. DAPI staining of nuclei (DNA) is shown in blue. Scale bars, 20 μm. (B) Diagram describing methods used for DAPT treatment in primary murine tracheal epithelial cells (MTECs). (C) IB for pIgR in air–liquid interface–differentiated MTECs with and without addition of the γ-secretase inhibitor DAPT, which restricts secretory cell differentiation (n = 3 inserts per group). β-Actin is included as a loading control, and α-tubulin is included to show that DAPT did not affect ciliated cell differentiation. (D) Quantification of (C) by densitometry. All three bands of pIgR are included in the densitometry analysis. pIgR expression was normalized across samples using β-actin. *P < 0.01 (t test). SC = secretory component; WT = wild-type.

As ciliated cells outnumber secretory cells in the distal airways of humans and mice (31), we speculated that even low amounts of pIgR expression in ciliated cells may be relevant to SIgA transport in the distal airways in vivo. To investigate this further, we generated a mouse model for conditional pigr deletion by placing loxP sites upstream of exon 3 and downstream of exon 11 of the pigr gene, which includes the binding sites for dIgA (see Figure E4A). The coding regions of GFP were introduced upstream and downstream of the two loxP sites, such that Cre recombinase–mediated excision of exons 3–11 results in GFP expression. pIgR^fl/fl mice were bred to Scgb1a1.Cre^ERT2 mice (26) to generate tamoxifen-inducible deletion of pIgR in secretory cells (pIgR^Δsecretory mice) and FoxJ1.Cre mice (27) to generate constitutive deletion of pIgR in ciliated cells (pIgR^Δciliated mice) (see Figure E4B). In pIgR^Δsecretory mice, Cre recombinase was activated by placing mice on tamoxifen chow every other week (to minimize toxicity) for 3 months (see Figure E4C). pIgR^Δciliated and pIgR^Δsecretory mice were evaluated using immunofluorescence to assess pIgR expression and GFP to confirm Cre-mediated recombination. Cre⁻ littermate control animals from both lines showed no expression of GFP (Figures 4A and 4B, left panels). We observed weak expression of GFP in pIgR^Δciliated mice (Figure 4A, right panel), suggesting low amounts of pigr expression in FoxJ1⁺ ciliated cells. However, despite pIgR deletion from ciliated cells in these mice, many airway epithelial cells continued to express pIgR (Figure 4A, right panel, white arrows). In contrast, we observed very few pIgR-expressing cells in pIgR^Δsecretory mice (Figure 4B, right panel), and GFP expression was much more robust in pIgR^Δsecretory mice than pIgR^Δciliated mice (see Figures E5A and E5B). We also performed western blotting for pIgR in whole-lung lysates from pIgR^Δciliated and pIgR^Δsecretory mice to measure global pIgR expression. In pIgR^Δciliated mice, there was no difference in pIgR concentrations between Cre⁺ and Cre⁻ littermate control animals (see Figures E5C and E5D). In contrast, in pIgR^Δsecretory mice, there was a marked reduction of full-length pIgR in Cre⁺ mice compared with littermate control and pIgR^Δciliated mice.

Figure 4.

Ciliated cell–derived pIgR is dispensable for protection against emphysema. (A and B) Immunostaining for pIgR and GFP in adult pIgR^Δciliated and pIgR^Δsecretory mice. White arrows in (A) highlight pIgR⁺GFP⁻ cells in pIgR^Δciliated mice. Scale bars, 50 μm. Inset depicts 2× magnification. (C) Low-magnification images (10×) of the lung parenchyma in 6-month-old Cre⁻ and Cre⁺ pIgR^Δciliated mice shows absence of emphysema in both strains (hematoxylin and eosin stain). (D) Quantification of emphysema (by mean linear intercept) in 6-month-old C57Bl6 WT and Cre⁺ and Cre⁻ pIgR^Δciliated mice (n = 4–7 mice per group). Box-and-whisker plots represent median, interquartile range, and range. Results are not significant (ANOVA). ns = not significant.

We previously noted mice with a global loss of pIgR (pigr^−/− mice) lack SIgA in small airways and spontaneously develop emphysema by 6 months of age (14). To evaluate the importance of ciliated cell–derived pigr for prevention of emphysema, we measured mean linear intercept (a morphometric measure of emphysema) in 6-month-old pIgR^Δciliated mice, Cre⁻ littermate control animals, and syngeneic wild-type mice. We found no difference in mean linear intercept between these groups (Figures 4C and 4D), suggesting that SIgA transport by nonciliated cell types is sufficient to maintain homeostasis in the murine lung. Taken together, these in vivo studies support the idea that secretory cells are the dominant cell type responsible for pIgR expression and thus play a unique role in maintaining the SIgA immunobarrier.

Changes in Secretory Cell Numbers Do Not Explain Loss of SIgA in COPD Airways

Having established secretory cells as the dominant cell type responsible for pIgR expression, we questioned whether loss of these cells contributes to reduced SIgA in COPD airways. Using RNA-ISH, we quantified SCGB1A1⁺ and SCGB3A2⁺ cells in small airways and found a similar proportion of secretory cells between COPD and control subjects, although there was a shift toward reduced numbers of SCGB3A2⁺ secretory cells (Figure 5A). To directly test whether loss of the SIgA immunobarrier was associated with a reduction in secretory cells, we performed immunostaining for SIgA as previously described (13) in lung sections from patients with advanced COPD and binned airways as SIgA⁺ or SIgA⁻ on the basis of visual assessment by a pathologist (Figure 5C, left panels). We then quantified secretory cells by RNA-ISH using probes for SCGB1A1 and SCGB3A2 and compared the percentage of these cell types in IgA⁺ and IgA⁻ airways. Consistent with prior data (13), the majority of small airways analyzed from this cohort of patients with severe COPD were SIgA deficient (60%). However, we found no difference in numbers of SCGB1A1⁺ or SCGB3A2⁺ cells between IgA⁺ and IgA⁻ airways, though there was a trend toward reduced SCGB3A2⁺ cells in IgA⁻ airways (Figure 5B). These data indicate loss of secretory cells does not account for SIgA deficiency in small airways of patients with advanced COPD.

Figure 5.

Secretory IgA (SIgA)⁻ airways have reduced pIgR expression compared with SIgA⁺ airways in patients with advanced COPD. (A) Percentage of SCGB1A1⁺ or SCGB3A2⁺ cells per airway among DAPI⁺ cells as assessed using RNA-ISH in control lungs (n = 6; 17 airways) versus COPD (n = 6; 18 airways). *P < 0.05, Mann-Whitney test. (B) Percentage of SCGB1A1⁺ or SCGB3A2⁺ cells per airway among DAPI⁺ cells as assessed using RNA-ISH in SIgA⁺ and SIgA⁻ airways (five patients, 18 positive airways, 28 negative airways) as shown in (C). (C) Representative images of a SIgA⁺ and SIgA⁻ small airways (on the basis of IgA immunostaining) from the explanted lungs of a patient with COPD (left) and staining of pIgR in a serial section of the same airways showing reduced pIgR in SIgA-deficient airways (right). Scale bars, 50 μm. Insets depict 2× magnification. Arrows show that intracellular pIgR is still present. (D) Average percentage of pIgR-expressing cells per patient in SIgA⁺ or SIgA⁻ airways as determined on HALO analysis. *P < 0.05, Mann-Whitney test. (E) Paired analysis of percentage of pIgR-expressing cells in SIgA⁻ versus SIgA⁺ airways per patient. ns = not significant (P > 0.05); RNA-ISH = RNA in situ hybridization.

We next considered the possibility that there was a reduction in pIgR protein expression in IgA⁻ airways despite intact PIGR mRNA expression seen in our scRNA-seq data (Figure 2D). To test this, we performed immunostaining for pIgR in IgA⁺ and IgA⁻ airways. We noted a significant decrease in the percentage of pIgR⁺ cells among total cells in IgA⁻ airways (Figures 5C and 5D), and all subjects had a higher mean percentage of pIgR-expressing cells in IgA⁺ airways relative to IgA⁻ airways (Figure 5E). These data indicate that loss of pIgR protein expression in secretory cells contributes to loss of the SIgA immunobarrier in patients with advanced COPD.

Bacterial and Host Proteases Degrade SIgA In Vitro

Although we found a significant association between loss of SIgA and percentage of pIgR-expressing cells, some IgA⁻ airways had a similar percentage of pIgR-expressing cells to IgA⁺ airways (Figure 6A, first two panels). As previous studies have indicated that SIgA negative airways have an increased presence of bacteria and neutrophils and that SIgA may be degraded by bacterial or host proteases (18–24), we speculated that degradation might account for loss of SIgA in airways with a normal percentage of pIgR-expressing cells. To explore this further, we performed fluorescence in situ hybridization for eubacterial 16S ribosomal RNA in IgA⁻ airways with intact pIgR expression (Figure 6A). We noted sporadic bacteria present in the mucosa of these airways (Figure 6A). We then evaluated whether SIgA is degraded in vitro by secreted proteases from NTHi, the most common bacterium isolated from the lungs of patients with COPD (34). We incubated SIgA derived from human colostrum with conditioned media from NTHi 1479 (a clinical isolate from a patient with COPD) and performed western blots for the IgA heavy chain and the SC. We found that conditioned media from NTHi cultures cleaved the IgA heavy chain but not the SC, consistent with IgA protease activity (Figures 6B–6D). In addition to bacteria, we also found extensive numbers of neutrophils present in and around SIgA⁻ small airways that still contained pIgR (Figure 6E). Incubation of colostrum-derived SIgA with sputum-derived HNE degraded the IgA heavy chain and the SC, forming several lower molecular weight degraded products (Figures 6F and 6G). These in vitro data support the idea that bacterial or host proteases may also contribute to loss of SIgA in COPD.

Figure 6.

Human SIgA is degraded by bacterial and host proteases in vitro. (A) pIgR-positive, SIgA-negative airways were assessed for the presence of bacteria using 16S ribosomal targeted fluorescence in situ hybridization. Scale bars, 20 μm. (B) SIgA was incubated with conditioned media from NTHi strain 1479 for 24 hours and then assessed for degradation/cleavage products in the IgA heavy chain (top) or in the secretory component (bottom). PBS and noninoculated media were used as negative controls. (C) Quantification of full-length IgA heavy chain from the top panel of (B). (D) Quantification of full-length secretory component from the bottom panel of (B). (E) pIgR-positive, SIgA-negative airways were assessed for the presence of neutrophils as detected by the neutrophil protease elastase. Arrow in the center panel indicates that intracellular pIgR is still present. Inset depicts 2× magnification. Neutrophils are present both in (arrow) and surrounding (arrowhead) the airway. Scale bars, 100 μm. (F) SIgA was incubated with human neutrophil elastase (HNE) for 24 hours and then assessed for degradation/cleavage products in the IgA heavy chain (top) or in the secretory component (bottom). The buffer used to dissolve the HNE (vehicle) was used as a negative control. A nonspecific mammalian protease inhibitor was also incubated with SIgA and HNE as a control. (G) Quantification of full-length IgA heavy chain from the top panel of (F). (H) Quantification of full-length secretory component from the bottom panel of (F). *P < 0.01 (t test). NE = neutrophil elastase; NTHi = nontypeable Haemophilus influenzae; PI = protease inhibitor.

Discussion

Loss of SIgA is common in COPD small airways and is an established driver of COPD progression in animal models (10, 12–16). However, the mechanism of loss of SIgA remains poorly understood, particularly in small airways. Here we show that secretory cells are the predominant source of pIgR in human and murine small airways and thus define a new niche for these cells in maintenance of the SIgA immunobarrier. Furthermore, we show that loss of SIgA is associated with reduced numbers of pIgR-expressing cells despite intact PIGR mRNA expression. Finally, we show that bacterial and host proteases known to be present in COPD can degrade SIgA, which may also contribute to SIgA deficiency in COPD small airways.

The finding that secretory cells are the predominant source of pIgR expression is consistent with published human and murine scRNA-seq data sets (25, 35–39) and with the known role of secretory cells in immune defense and airway homeostasis (37). Although ciliated cells also produce low concentrations of pIgR and are more numerous than secretory cells in distal airways, we did not observe lung pathology in mice with ciliated cell–specific deletion of pIgR. This finding suggests that secretory cells play a dominant role in maintaining the SIgA immunobarrier. Interestingly, we found a shift toward reduced SCGB3A2⁺ secretory cells in COPD airways, consistent with other studies showing a reduction in SCGB3A2 gene expression in COPD (40, 41). SCGB3A2⁺ secretory cells are localized predominantly in terminal and respiratory bronchioles (32), where much of the pathology of COPD originates. The mechanism of SCGB3A2⁺ cell loss in COPD airways merits additional investigation.

Here we found that loss of SIgA was associated with a reduction in cells containing pIgR protein in small airways from patients with advanced COPD. Interestingly, this block appears to be at the translational or post-translational level, as we observed similar or increased amounts of PIGR mRNA expression between COPD and control samples across cell types in our scRNA-seq data set. Divergence between PIGR mRNA expression and protein concentrations was also described by Gohy and colleagues in large airway biopsies (42). As PIGR expression is known to be upregulated by inflammatory cytokines including IFN-γ, TNF, IL-1β, and IL-4 (43–47), persistent inflammation in small airways of patients with COPD could explain our observations regarding PIGR mRNA expression in COPD small airways. However, the mechanism responsible for reduced pIgR protein concentrations in COPD remains incompletely understood and is an important area for future study.

We found that the IgA heavy chain was degraded by NTHi 1479, a clinical isolate of the most common bacterium isolated from the lungs of patients with COPD during exacerbations (34). A variety of bacteria have been shown to degrade SIgA (18, 19, 22–24), and it is conceivable that SIgA-degrading bacteria are overrepresented in patients with COPD during stable and/or exacerbated periods (48, 49). In addition, we found that HNE is capable of degrading the IgA heavy chain and SC. Thus, degradation of SIgA by host and/or bacterial proteases could contribute to loss of SIgA in COPD small airways. Furthermore, we previously showed that loss of the SIgA immunobarrier is associated with increased numbers of neutrophils within individual airways (13), and neutrophils are spontaneously recruited to SIgA-deficient airways in mice because of chronic bacterial invasion (14, 15). Thus, it is tempting to speculate that once loss of SIgA is established in an airway because of loss of pIgR, chronic bacterial invasion and neutrophilic inflammation further reduce SIgA concentrations in the airway in a feed-forward mechanism. Future studies examining SIgA fragmentation patterns in BAL fluid samples could prove useful for determining the specific proteases degrading SIgA.

There are several important limitations to our study. Many patients in our scRNA-seq control cohort were light smokers, whereas all patients in the COPD cohort were former smokers, which could have affected pIgR’s protein abundance or mRNA expression. Gohy and colleagues previously noted increased PIGR expression in the large airways of smokers (42). Conversely, McGrath and colleagues found that cigarette smoke attenuated pIgR protein expression in vivo in response to challenge with ovalbumin and LPS (50). Similarly, Rostami and colleagues showed that cigarette smoke extract treatment in vitro reduced multiple secretory cell defense genes, including PIGR (39). Further studies will be required to more adequately define the impact of smoking on PIGR/pIgR expression. In addition, as control samples were taken from nondiseased lung donors whose lungs were declined for transplantation, it is likely that some of these donors had acute lung pathology such as pulmonary hemorrhage or pneumonia, which also could have influenced pIgR expression. Although also subject to biases, future studies involving unaffected tissue from lung nodule/cancer resections would be complementary if matched to COPD samples by age and smoking status. Finally, our study does not exclude the possibility that there are additional pIgR-independent defects in SIgA transcytosis. Future in vitro experiments using COPD-relevant models of abnormal epithelial differentiation such as that recently reported by Rao and colleagues (51) would be helpful in examining this further.

In summary, our study suggests that secretory cells are predominantly responsible for maintaining the SIgA immunobarrier in small airways and that loss of SIgA in COPD small airways is complex and likely due to both reduced expression of pIgR and degradation by host and bacterial proteases. Strategies to restore the SIgA immunobarrier will likely have to account for multiple layers of dysregulation in COPD.