An SFTPC BRICHOS mutant links epithelial ER stress and spontaneous lung fibrosis

Adult interstitial lung disease (ILD) and pediatric childhood ILD (ChILD) are defined pathologically by interstitial fibrosis, inflammation, or a combination of fibrosis and inflammation (1, 2). With thickening of the alveolar interstitium causing impairment in effective gas exchange, the clinical course is often progression to chronic respiratory failure (3, 4). Over the past decade, the alveolar type 2 (AT2) cell has come to the forefront as a key driver of lung fibrosis and ILD (5, 6). Focus on the AT2 cell is in part due to the identification of disease-causing mutations in the AT2 cell–restricted surfactant protein C gene (SFTPC) in both pediatric patients with ChILD and in a subset of adult patients with familial ILD (7–9). The recent demonstration by our group of the occurrence of spontaneous lung fibrosis with expression of the Sftpc^I73T trafficking mutation in a preclinical mouse model has provided further proof of concept for the AT2 cell as a proximal driver of ILD (10). While the SFTPC^I73T mutation, which causes a block in AT2 cell macroautophagy, is the most common clinical SFTPC mutation identified, a majority of the 60 described mutations in the SFTPC gene associated with human disease occur in the distal C-terminal (residues 94–197) BRICHOS domain of the surfactant protein C (SP-C) pro-protein (proSP-C). Their expression results in pro-protein aggregation and induction of ER stress (11–16). These cases further highlight the proximal position of the AT2 cell in the pathogenesis of pulmonary fibrosis and suggest that AT2 cell ER stress may provide a broader mechanistic link between SFTPC BRICHOS mutation–associated pediatric ChILD and adult ILD.

Markers of ER stress are also a prominent feature in lung epithelium from patients with sporadic idiopathic pulmonary fibrosis (IPF), the most common adult ILD (17); however, the role of epithelial ER stress in disease pathogenesis is unclear. In vivo proof-of-concept studies linking SFTPC BRICHOS mutations, ER stress, and pulmonary fibrosis have been plagued by over- and underexpression of the mutant pro-protein isoforms, and have failed to establish a mechanistic connection. While constitutive transgenic expression of a highly aggregation prone SFTPC mutation (SFTPC^Δexon4) resulted in disrupted lung morphogenesis with prenatal lethality (18), inducible expression of a less-severe mutation (SFTPC^L188Q) at lower levels produced increased susceptibility to bleomycin fibrosis, but not spontaneous fibrotic remodeling (19). This finding suggested that epithelial ER stress may not be a driver of fibrosis per se, but a marker of a dysfunctional and vulnerable epithelium requiring a “second hit” to cause disease.

Given the low level of SFTPC expression induced in SFTPC^L188Q mice, we sought to challenge this notion with the hypothesis that a more-severe SFTPC BRICHOS mutation expressed at a more biologically relevant level would produce ER stress–based toxic gain of function resulting in a dysfunctional epithelial cell capable of driving a spontaneous lung phenotype. Based upon a pediatric case of ChILD with a mutation in the cysteine at codon 121 (C121), and previous ultrastructural analysis demonstrating that the disulfide bond formed between C121 and cysteine 189 of the BRICHOS domain mediates the primary folding of the proSP-C primary translation product (20), we generated a murine model (Sftpc^C121G). Consistent with a misfolded isoform, the Sftpc^C121G pro-protein is proximally retained in the ER and induces substantial epithelial ER stress both in vitro and in vivo. The resultant lung phenotype of AT2 cell–driven spontaneous inflammation and subsequent fibrotic remodeling provides proof of concept linking SFTPC BRICHOS mutations and pulmonary fibrosis. The translational relevance of this model was further corroborated by a high-dimensional proteomic analysis of bronchoalveolar lavage fluid (BALF) from pediatric patients with SFTPC BRICHOS mutation–associated ChILD showing overlapping cytokine expression. The Sftpc^C121G model, with relevant biomarker elaboration, thus establishes that AT2 cell ER stress is a sufficient driver for spontaneous lung inflammation and aberrant fibrotic lung remodeling.

In vitro expression of a clinical SFTPC cysteine 121 mutation induces ER stress by a proximally retained SP-C pro-protein. A male child of an uncomplicated pregnancy was born at 38 weeks gestation by spontaneous vaginal delivery to a 27-year-old mother. He subsequently developed respiratory failure in the first day of life, requiring endotracheal intubation and mechanical ventilation. He received exogenous surfactant, and was extubated and transitioned to nasal cannula supplemental oxygen. At 2 months of age, a second episode of respiratory failure occurred, requiring tracheostomy for chronic mechanical ventilation. Chest imaging at 3 months demonstrated diffuse ground glass opacities (Figure 1A). Subsequently, a clinical pathologic analysis of an open lung biopsy (Figure 1B) showed ChILD histology with enlargement of airspaces, interstitial widening, inflammatory infiltrate, and prominent AT2 cell hyperplasia. Bronchoalveolar lavage demonstrated increased cellularity (755 cells/ml) with elevated neutrophils (20%). The patient was ultimately weaned from the ventilator and decannulated 2 years later. Genetic testing demonstrated a heterozygous SFTPC mutation in the BRICHOS domain, c.362 G>A (p.C121Y), without SFTPB or ATP binding cassette A3 (ABCA3) mutations.

Figure 1

In vitro modeling of a clinical SFTPC^C121Y mutation identifies mistrafficked pro-protein and epithelial ER stress with mutagenesis of codon 121 in SFTPC. (A) Chest imaging of a 3-month-old patient with heterozygous SFTPC^C121Y mutation shows bilateral diffuse hazy opacity on chest radiograph (left) and diffuse ground glass opacities on chest CT (right). (B) H&E staining of patient lung biopsy demonstrates diffuse abnormality in lobular architecture, with mild enlargement of airspaces and mild to moderate interstitial widening, airspaces filled by foamy macrophages, scattered neutrophils, and eosinophilic material suggestive of proteinosis (original magnification, ×4 left and ×20 right). (C) HEK293 cells were transiently transfected with either EGFP/SP-C^WT or EGFP/SP-C^C121 mutants as indicated, and cell lysates harvested at 48 hours were subjected to Western blotting for EGFP. An SP-C primary translation product (black arrow) was detected in all transfections; however, no processing intermediates (black bracket) were observed with either EGFP/SP-C^C121 mutation. Representative of 3 independent experiments. (D) HEK293 cells transiently cotransfected with either EGFP/SP-C^WT or EGFP/SP-C^C121G and DsRed/ABCA3 and subjected to immunfluorescence microscopy, demonstrating colocalization of EGFP/SP-C^WT with DsRed/ABCA3, but a reticular expression pattern of EGFP/SP-C^C121G without DsRed/ABCA3 colocalization (original magnification, ×60). (E) Cell lysates of HEK293 cells 48 hours after transfection with either EGFP/SP-C^WT or EGFP/SP-C^C121 mutants as indicated were subjected to Western analysis for BiP (top). Representative Western blot of n = 3. Quantitative densitometry (bottom) normalized to β-actin showing increases in BiP with the cysteine mutants compared to EGFP/SP-C^WT. n = 3. *P < 0.05 vs. EGFP/SP-C^WT by 1-way ANOVA with Dunnett’s multiple comparison test.

In vitro modeling was performed to define the functional consequences of C121 mutations on proSP-C biosynthesis. Plasmids containing EGFP-tagged WT SFTPC (EGFP/SP-C^WT) and 2 SFTPC mutants, EGFP/SP-C^C121Y and EGFP/SP-C^C121G, were transfected into HEK293 cell lines, which were interrogated for cellular expression patterns and posttranslational processing. The SFTPC^C121G mutation, which has also been described clinically (21), was selected in addition to the clinical SFTPC^C121Y mutation to generalize the importance of disulfide bonding in the pro-protein via the C121 thiol (20). In contrast to the WT isoforms, Western blot analysis of cell lysates showed a primary proSP-C translation product containing either cysteine mutant failed to undergo normal processing to a cleaved intermediate pro-protein (Figure 1C). Subsequent cotransfection of HEK293 cells with EGFP/SP-C^WT or EGFP/SP-C^C121G and DsRed-tagged ABCA3 (DsRed/ABCA3) to mark lysosome-related organelles (LROs) demonstrated colocalization between EGFP/SP-C^WT and DsRed/ABCA3, but reticular patterning without colocalization of the EGFP/SP-C^C121G isoform in LROs (Figure 1D). Expression of the proximally retained mutant pro-proteins induced an ER stress response in vitro, as transfected HEK293 cell lysates demonstrated increases in BiP (GRP78) protein in cells expressing either SFTPC mutation compared with WT SFTPC (Figure 1E). This was consistent with previously published in vitro studies of SFTPC BRICHOS mutants, which have demonstrated BiP to be a reliable marker of the ER stress response (13, 15). These data collectively define a critical role of the C121 residue in the biosynthetic routing and processing of proSP-C, and reveal an ER stress toxic gain of function with SFTPC^C121 mutations.

Constitutive expression of the Sftpc^C121G mutation in vivo phenocopies the human pediatric mutation. Given the importance of C121 in the processing of proSP-C in vitro, we utilized an embryonic stem (ES) cell recombineering strategy to selectively knock in a homologous cysteine-to-glycine substitution at codon 121 (C121G) to the endogenous mouse Sftpc locus. Given that the functional consequences of each cysteine mutant isoform were equivalent in vitro, the C121G missense substitution was chosen for the in vivo knockin model to limit confounding variables related to residue side chain size, hydrophobicity, or potential posttranslational modification of a tyrosine residue. The resultant founder line (Sftpc^{C121Gneo/C121Gneo}) retained a PGK-neomycin (PGK-neo) cassette in intron 4 flanked by locus of X-over P1 (loxP) sites (Supplemental Figure 1A; supplemental material available online with this article; https://doi.org/10.1172/jci.insight.126125DS1). Retention of the PGK-neo cassette rendered the Sftpc^C121Gneo gene hypomorphic, a strategy employed for knockin models of toxic gain-of-function mutations (10, 22, 23), with mRNA levels in homozygous Sftpc^C121Gneo animals limited to ~6% that of littermates expressing 2 Sftpc^WT alleles (Supplemental Figure 1B). Despite AT2 cells lacking expression of proSP-C (Supplemental Figure 1C) and consistent with the phenotype of the Sftpc-knockout mouse (24), the founder line was devoid of a spontaneous inflammatory or fibrotic phenotype when aged up to 52 weeks (Supplemental Figure 1, D–F).

To evaluate the effect of increasing Sftpc^C121G expression in the developing lung, we constitutively removed the loxP-flanked PGK-neo cassette from Sftpc^C121Gneo alleles by crossing the Sftpc^{C121Gneo/C121Gneo} founder line to mice heterozygous for the CMV-Cre recombinase (a “deleter” line) (25). The resultant progeny were either Sftpc^WT/C121Gneo in mice not receiving the CMV-Cre allele or heterozygous for the activated Sftpc^C121G allele (Sftpc^WT/C121G) in mice receiving the CMV-Cre allele (Figure 2A). Consistent with a dominant negative effect, heterozygous expression of Sftpc^C121G caused abnormal processing of the WT SP-C pro-protein, with reduced processing intermediates on Western blotting for proSP-C (Figure 2B) and an 84.7% ± 3.2% (mean ± SD, n = 4) reduction in mature SP-C (mSP-C) protein in P2.5 lung homogenates (Figure 2C). This finding recapitulates data from SFTPC BRICHOS mutation patients whereby heterozygous expression of the SFTPC mutation dramatically blocks pro-protein processing, reducing the amount of mSP-C in patients’ BALF (26–28). Previous in vitro studies have shown mechanistically that heteromeric association between mutant and WT proSP-C proteins is responsible for disruption of the intracellular trafficking of the WT proSP-C isoform (16, 29, 30).

Figure 2

Constitutive expression of the Sftpc^C121G mutation in vivo causes a lethal toxic gain of function. (A) Schematic for generation of mice constitutively expressing Sftpc^C121G alleles. Breeding of the Sftpc^{C121Gneo/C121Gneo} line to CMV-Cre mice yields litters with heterozygous constitutive expression of the Sftpc^C121G mutation (Sftpc^WT/C121G). (B) Western blotting of lung homogenates from day P2.5 for proSP-C expression shows aberrant banding pattern, with heterozygous Sftpc^C121G expression with decreased processing intermediates (arrow, Sftpc^WT/C121Gneo proSP-C highest-molecular-weight translational product; arrowhead, Sftpc^WT/C121G highest-molecular-weight translational product; star, nonspecific bands; brackets, processing intermediates). (C) Top: Representative Western blot of lung homogenates from P2.5 for mature SP-C (mSP-C). Bottom: Dot plots of mean and SEM of mSP-C densitometry expressed as fold change to Sftpc^WT/C121Gneo. *P < 0.05 using unpaired 2-tailed t test with normalization to β-actin loading control. (D) Kaplan-Meir survival analysis showing high lethality with constitutive Sftpc^C121Gexpression (n = 41 and 37 for Sftpc^WT/C121G and Sftpc^WT/C121Gneo, respectively). P < 0.001 by log-rank (Mantel-Cox) test. (E) Representative photographs of P.2.0 pups show development of cyanosis in the Sftpc^WT/C121G animal. (F) Representative lung histology (×4 magnification) of P2.0 Sftpc^WT/C121G pups with heterogeneous areas (×10 magnification inset) of thickened septae with proteinaceous fluid in the alveolar space. Scale bars: 500 μm, 200 μm (insets).

The constitutive Sftpc^C121G expression model phenocopied many of the features of the pediatric disease (27, 31). Similar to some cases of ChILD, including the one presented here (Figure 1), a toxic gain of function was observed, with Sftpc^WT/C121G mice developing lethal postnatal respiratory failure with a median survival of 2.5 days (Figure 2D). Sftpc^C121G expression manifested as respiratory distress and cyanosis between P1.5 and P3.5 (Figure 2E), with lung histology of distorted architecture, enlargement of airspaces, interstitial widening, inflammatory infiltrate, and proteinacious fluid in the airspaces (Figure 2F) — features resembling the histology in the clinical SFTPC^C121Y case (Figure 1B).

Expression of mutant Sftpc^C121G in adult mice results in ER-retained SP-C pro-protein and ER stress. To evaluate the cellular effect of Sftpc^C121G expression in the adult mouse, the Sftpc^C121Gneo founder line was crossed to the Rosa26-ERT2-Cre line (R26^Cre) to generate a line (Sftpc^C121G R26^Cre) capable of tamoxifen-mediated Cre recombinase removal of the inhibitory PGK-neo cassette from Sftpc^C121Gneo alleles (Figure 3A). In the absence of tamoxifen, adult homozygous Sftpc^C121G/C121G R26^Cre animals injected with vehicle (i.p. oil) retained a phenotype identical to that of the Sftpc^{C121Gneo/C121Gneo} founder line without substantial leakiness of the R26^Cre locus, as shown by the absence of detectable proSP-C on immunohistochemistry (Supplemental Figure 2). Given the absence of a lung phenotype in Sftpc^C121G/C121G R26^Cre mice with i.p. oil and the extreme hypomorphic Sftpc expression in the founder line, we selected tamoxifen-injected Sftpc^WT R26^Cre mice as controls, given previously described cre lung toxicity (32). At 7 days after administration of tamoxifen (500 mg/kg), Sftpc^C121G mRNA increased to 26.2% ± 4.9% (mean ± SD, n = 7) versus Sftpc^WT controls (Figure 3B), a 4-fold increase from the hypomorphic Sftpc^{C121Gneo/C121Gneo} founder line. Using quantitative real-time PCR (qRT-PCR) to estimate PGK-neo cassette numbers, the efficiency of loxP-mediated recombination in purified AT2 cells isolated 1 week after tamoxifen treatment was 92.4% ± 3.1% (mean ± SD, n = 4) referenced against Sftpc^{C121Gneo/C121Gneo} mice. Despite a substantial increase in Sftpc^C121G mRNA, Western blot analysis of the phospholipid-enriched large aggregate surfactant fraction of BALF from tamoxifen-treated Sftpc^C121G/C121G R26^Cre mice showed no evidence of extracellular mSP-C protein (Figure 3C). The absence of the mSP-C protein was the result of a failure to process translated proSP-C, as AT2 cell lysates from these mice showed a single 21-kDa pro-protein band, corresponding to the unprocessed primary translational product. In contrast, 23- and 21-kDa pro-protein bands (23 kDa corresponding to the palmitoylated post-ER pro-protein) with processing intermediates from 8 to 15 kDa were observed in Sftpc^WT R26^Cre controls (Figure 3D) (33, 34). Furthermore, lung sections of tamoxifen-treated Sftpc^C121G/C121G R26^Cre mice demonstrated a reticular pattern of the mutant proSP-C with significant colocalization with the ER marker KDEL, signifying substantial ER retention (Figure 3E). Thus, interrogation of the adult Sftpc^C121G/C121G R26^Cre mice revealed ER retention of the mutant proSP-C in vivo without evidence for posttranslational palmitoylation from post-Golgi or proteolytic processing, consistent with the in vitro data (Figure 1) and what has been reported in SFTPC BRICHOS patients (27, 35).

Figure 3

In vivo inducible expression of the Sftpc^C121G mutation in adult mice causes an ER-retained SP-C pro-protein. (A) Strategy for generation of tamoxifen-inducible mice in which tamoxifen treatment of the Sftpc^C121G/C121G R26^Cre line results in removal of an inhibitory intronic PGK-neo cassette. (B) qRT-PCR analysis for Sftpc expression in purified AT2 cells from homozygous Sftpc^{C121Gneo/C121Gneo} and Sftpc^C121G/C121G R26^Cre mice at 7 days after treatment with tamoxifen. Data normalized to 18S RNA are expressed as Sftpc mRNA as a fraction of Sftpc^WT R26^Cre mice. (C) Western blotting of BALF large aggregate fraction from Sftpc^C121G/C121G R26^Cre and Sftpc^WT R26^Cre mice on day 7 after tamoxifen showing the absence of mature SP-C (mSP-C) in the Sftpc^C121G/C121G R26^Cre mice. (D) Western blotting of AT2 cell lysates from Sftpc^WT R26^Cre or Sftpc^C121G/C121G R26^Cre mice 7 days after tamoxifen shows Sftpc^C121G/C121G R26^Cre AT2 cells with an ER retained pro-protein (arrowhead) without posttranslational palmitoylation (arrow) or processing intermediates (brackets) observed in Sftpc^WT R26^Cre AT2 cells. (E) Double-label immunofluorescence staining of whole lung sections for proSP-C (red) and the ER marker KDEL demonstrates reticular proSP-C staining with significant colocalization with KDEL in Sftpc^C121G/C121G R26^Cre mice, compared with the punctate pattern of proSP-C staining distinct from KDEL observed in the Sftpc^WT R26^Cre mice (original magnification, ×60).

ER retention of mutant proSP-C C121G induced activation of the unfolded protein response (UPR) in the pulmonary epithelium, as reflected by a substantial increase in BiP mRNA in AT2 cells isolated from mouse lungs as early as 7 days after tamoxifen treatment (Figure 4A). In a corresponding fashion, BiP protein was also detected in AT2 cells by both immunohistochemical staining of cells localized to corners of alveoli (Figure 4B) and Western blotting of AT2 cell lysates (Figure 4C). Consistent with our prior in vitro modeling (36), the Sftpc^C121G AT2 cell phenotype was distinct from the Sftpc^I73T model, where expression of that mutant isoform caused a block in macroautophagy and failed to generate substantial AT2 cell BiP expression (Supplemental Figure 3)(10).

Figure 4

Expression of the SP-C C121G pro-protein causes activation of multiple epithelial ER stress pathways and induces AT2 cell apoptosis. (A) qRT-PCR for BiP expression in purified AT2 cells from Sftpc^WT R26^Cre controls and Sftpc^C121G/C121G R26^Cre mice at 7 days after tamoxifen treatment. Data normalized to 18S RNA are expressed as fold change in BiP normalized to control mice. *P < 0.05 vs. control using unpaired 2-tailed t test. (B) Representative immunohistochemical staining for BiP in lung sections (magnification, ×20) from control and Sftpc^C121G/C121G R26^Cre mice at 7 days after tamoxifen shows increased BiP staining in AT2 cells (inset magnification, ×60). (C) Western blotting of AT2 cell lysates at 7 days after tamoxifen for BiP, ATF6 (P90), ATF4, CHOP, and β-actin. (D) qRT-PCR for XBP1 splicing ratio in AT2 cells at 7 days after tamoxifen shows an increase in the spliced fraction in Sftpc^C121G/C121G R26^Cre AT2 cells compared with controls. *P < 0.05 vs. control using unpaired 2-tailed t test. (E) Western blot (left) of AT2 cell lysates at 7 days after tamoxifen for phosphorylated JNK (P-JNK), total JNK (T-JNK), and β-actin. Densitometry ratio (right) of P-JNK/T-JNK, reveals increase in P-JNK in Sftpc^C121G/C121G R26^Cre mice compared with control. *P < 0.05 vs. control using unpaired 2-tailed t test. (F) Representative ×20 fluorescence micrograph from Sftpc^C121G/C121G R26^Cre lung 7 days after tamoxifen stained with proSP-C (red) and cleaved caspase-3 (green) (left). Dot plots of double-positive (cleaved caspase-3⁺/proSP-C⁺) cells expressed as a percentage of total proSP-C⁺ AT2 cells are shown with means and SEM (right). *P < 0.05 vs. control using unpaired 2-tailed t test.

The UPR utilizes 3 signaling pathways: IRE1/XBP1, PERK/ATF4, and ATF6. Each pathway was interrogated in the AT2 cells from Sftpc^C121G/C121G R26^Cre and control mice after tamoxifen treatment. The IRE1 pathway was activated, as demonstrated by an increase in the ratio of active spliced Xbp1(S) to unspliced Xbp1(U) mRNA (Figure 4D), and downstream activation of JNK signaling, as shown by increased phosphorylated JNK (Figure 4E). The PERK/ATF4 pathway was found to be activated by an increase in both ATF4 and its downstream target CCAAT/enhancer-binding protein homologous protein (CHOP) (Figure 4C). Interrogation of the ATF6 pathway revealed a marked decrease in the ATF6 P90 precursor band in Sftpc^C121G AT2 cells (Figure 4C).

While initiation of UPR signaling is viewed as an adaptive cellular response, sustained overactivation can cause deleterious responses such as apoptosis, a feature of the epithelium in both sporadic and SFTPC mutation–related ILD (26, 37, 38). Previously, we have shown that in vitro expression of SFTPC BRICHOS mutations induces apoptosis via multiple UPR pathways, including ATF4/CHOP and IRE1/JNK (14). Double-label fluorescence immunohistochemistry of lung sections for proSP-C and cleaved caspase-3 at 7 days after tamoxifen treatment revealed a significant increase in the apoptotic index, with 9.4% ± 1.7% (mean ± SD, n = 4) AT2 cells undergoing apoptosis in the Sftpc^C121G/C121G R26^Cre lungs compared with 1.6% ± 0.5% (mean ± SD, n = 4) in controls (P = 0.001) (Figure 4F). Thus, Sftpc^C121G expression induced the UPR and produced an ER stress, proapoptotic AT2 cell phenotype.

Expression of the Sftpc^C121G mutation causes early morbidity and mortality from lung injury and AT2 cell–driven polycellular alveolitis. Sftpc^C121G/C121G R26^Cre mice demonstrated tamoxifen dose–dependent morbidity and mortality. At a dose of 350 mg/kg tamoxifen, the mice developed weight loss starting 7 days after tamoxifen treatment, with a nadir of 84.8% ± 9.6% (mean ± SD, n = 13) from baseline at 13 days (Figure 5A) in association with 40% early mortality (Figure 5B). Lower doses were nonfatal, and a higher dose (600 mg/kg) was 100% fatal (Supplemental Figure 4). The weight loss and death were commensurate with hypoxemia (Figure 5C) and increased BALF total protein (Figure 5D). Histologically, the Sftpc^C121G/C121G R26^Cre mice developed acute, diffuse lung injury by day 14 (Figure 5E).

Figure 5

Expression of mutant Sftpc^C121G causes lung injury with polycellular alveolitis. (A) Weight loss curve in surviving Sftpc^C121G/C121G R26^Cre mice and Sftpc^WT R26^Cre controls treated with tamoxifen. *P < 0.05 vs. control group using unpaired 2-tailed t test. (B) Kaplan-Meier survival curve of Sftpc^C121G/C121G R26^Cre mice treated with tamoxifen (n = 43) or vehicle (oil) (n = 16) and Sftpc^WT R26^Cre mice treated with tamoxifen (n = 32). End points were defined as death or body weight <75% on 2 consecutive days. P < 0.001 by log-rank (Mantel-Cox) test. (C) Pulse oximetry of Sftpc^C121G/C121G R26^Cre mice and controls 7 days after tamoxifen. *P < 0.05 vs. control using unpaired 2-tailed t test. (D) BALF protein content following tamoxifen. Controls represent pooled samples from all 4 time points. *P < 0.05 by 1-way ANOVA with post hoc Tukey’s test. (E) Representative ×10 magnification H&E histology at 14 days after tamoxifen. Scale bars: 500 μm. (F) Dot plots with mean and SEM of BALF cell count following tamoxifen. Controls represent pooled samples from all 4 time points. *P < 0.05 by 1-way ANOVA with post hoc Tukey’s test. (G) Representative Giemsa-stained BALF cytospins from control and Sftpc^C121G/C121G R26^Cre mice after tamoxifen (magnification, ×20). (H) Dot plots with mean and SEM of percentage of total lung digest immune cells (CD45⁺) that were alveolar macrophages (SiglecF⁺CD11b^–) at 3 days after tamoxifen. *P < 0.05 vs. control using unpaired 2-tailed t test. (I) Representative control and Sftpc^C121G/C121G R26^Cre flow cytometry gating for CD11b⁺Ly6C^hi monocytes at day 3 after tamoxifen (left). Dot plots with mean and SEM of percentage of total lung digest immune cells (CD45⁺) that were CD11b⁺Ly6C^hi monocytes. *P < 0.05 vs. control using unpaired 2-tailed t test.

The parenchymal lung injury was accompanied by polycellular alveolitis, with a substantial increase in total BALF cell counts beginning 7 days after tamoxifen treatment and peaking at 2 weeks (Figure 5F). BALF cytospins demonstrated multiphasic inflammatory cell alveolitis, with an early and sustained increase in macrophage/monocyte lineages beginning 7 days after tamoxifen, a rise in neutrophils starting 7 days after tamoxifen and peaking by 2 weeks, and an increase in eosinophils at 2 weeks (Figure 5G and Supplemental Table 1). This pattern of granulocyte alveolitis mirrored that found in the Sftpc^I73T model (10). Total lymphocytes also increased by day 14 and were sustained through 28 days, albeit at numbers 1 log fold less than the macrophage/monocyte and granulocyte cell populations. The inflammatory phenotype was dependent on Sftpc^C121G allele status, with mice heterozygous for the Sftpc^C121G allele (Sftpc^C121G/WT R26^Cre) developing less-severe alveolitis 2 weeks after tamoxifen that was sustained to 4 weeks, but in the absence of weight loss or early mortality (Supplemental Figure 5).

To define the effector cell populations that precede the overt alveolitis, flow cytometry analysis of whole lung tissue was performed 3 days after tamoxifen treatment using the gating strategy shown in Supplemental Figure 6. By this algorithm, an early decrease in the relative percent of SigF⁺CD11b^– resident alveolar macrophages was found (Figure 5H), commensurate with an influx of inflammatory CD11b⁺Ly6C^hi monocytes (Figure 5I), a population recently implicated in fibrotic lung remodeling (39, 40).

In order to define the potential mediators of effector cell recruitment, BALF samples at early- and late-stage alveolitis were analyzed by multiplex assay and ELISA (Supplemental Table 2). CCL2 (MCP1), CCL17 (TARC), and CCL7 (MCP-3) were found to be increased in the BALF of Sftpc^C121G/C121G R26^Cre mice compared with Sftpc^WT R26^Cre controls at the early time point (Figure 6A). Importantly, increases in AT2 cell mRNA expression of Ccl2, Ccl17, and Ccl7 occurred by 7 days, suggesting that this cellular compartment is a potential source of these cytokines (Figure 6B). Notably, although elevations in cytokine levels for CCL2, CCL17, and CCL7 were not statistically significant 3 days after tamoxifen treatment, there was a 2.6 ± 1.2– (mean ± SD, n = 4, P = 0.01), 3.1 ± 1.0– (mean ± SD, n = 4, P = 0.01), and 3.0 ± 0.5– fold (mean ± SD, n = 4, P = 0.005) increase in AT2 cell mRNA expression, respectively. These data coupled with the early influx of inflammatory monocytes suggest that local elaboration of these cytokines may contribute to the recruitment.

Figure 6

Sftpc^C121G AT2 cells are a source of cytokines associated with macrophage/monocyte recruitment. (A) Dot plots with mean and SEM of BALF CCL2 (top), CCL17 (middle), and CCL7 (bottom) protein in Sftpc^C121G/C121G R26^Cre mice and Sftpc^WT R26^Cre controls 7 days after tamoxifen treatment determined by Luminex assay (CCL2 and CCL17) and ELISA (CCL7). *P < 0.05 versus controls by 1-way ANOVA (see Supplemental Table 2), followed by post hoc Tukey’s test. (B) qRT-PCR determination of Ccl2 (top), Ccl17 (middle), and Ccl7 (bottom) mRNA expression in AT2 cells 7 days following tamoxifen. Dot plots with mean and SEM. *P < 0.05 vs. control using unpaired 2-tailed t test.

In addition to the monocyte/macrophage lineages, Sftpc^C121G AT2 cells also contributed to granulocyte recruitment. Sftpc^C121G AT2 cells had increased mRNA expression of Ccl11 and Il5 compared with control AT2 cells, and BALF showed an increase in the corresponding eosinophil chemokines CCL11 (eotaxin) and IL-5 at 1 week after tamoxifen treatment (Supplemental Figure 7). This occurred in the absence of detectable increases in the canonical Th2 cytokines IL-4 and IL-13. Additionally, a 10-fold increase in Sftpc^C121G AT2 cell expression of the murine homolog to IL8 (neutrophil chemoattractant KC [Cxcl1/GroA]) occurred and was also reflected by elevated BALF KC/CXCL1 content at 1 week after tamoxifen (Supplemental Figure 7). There was also a significant increase in BALF IL-6, which along with IL-8 has been posited as a biomarker for acute exacerbations of ILD (AE-ILD) (41, 42) (Supplemental Table 2). In contrast to IL-8, there was no increase in Il6 mRNA expression in Sftpc^C121G AT2 cells, suggesting a non–AT2 cell source of IL-6 such as the distal lung mesenchyme (43). In total, Sftpc^C121G AT2 cells were an early source of cytokines that mediate the recruitment of multiple immune effector cell populations and an important contributor to the early lung inflammatory phenotype.

High-dimensional proteomic screen of a pediatric SFTPC BRICHOS mutation cohort translationally overlaps with the Sftpc^C121G cytokine profile. The translational relevance of the Sftpc^C121G model was assessed in a human pediatric cohort of patients with ChILD. Five patients with ChILD carrying SFTPC BRICHOS mutations and 9 control subjects underwent BALF analysis. Subjects with SFTPC BRICHOS mutations had an increase in total BALF cell count and an increase in the percentage of BALF eosinophils compared with controls (Supplemental Table 3). BALF supernatants were analyzed using a SOMAscan aptamer proteomics platform for 1129 proteins, which identified 7 proteins increased in the SFTPC BRICHOS subjects’ BALF that met conservative significance criteria (P < 0.001; difference greater than 1 on a log₂ scale). Five of these proteins were associated with the recruitment of immune cells (Figure 7A). Three were also increased in the Sftpc^C121G model (Figure 6A) and have been well described in adult ILD cohorts (44–49): CCL2 (MCP-1), CCL17 (TARC), and CCL7 (MCP-3) (Figure 7, B–D). All 3 of these cytokines are associated with macrophage/monocyte recruitment, the initial effector population identified in the Sftpc^C121G/C121G R26^Cre lungs.

Figure 7

Pediatric SFTPC BRICHOS mutation patients elaborate multiple cytokines associated with macrophage/monocyte recruitment found in Sftpc^C121G mice. (A) Volcano plot of SOMAscan proteomics platform analysis of BALF from SFTPC BRICHOS mutations cases (n = 5) and disease control (n = 9). Minus log₁₀–transformed P value on y axis, and log₂ difference on x axis. Conservative selection of cytokines associated with immune cell recruitment with relative florescence units (RFU) difference greater than 1 on a log₂ scale and P < 0.001 (shown in red). (B–D) Individual dot plots of mean ± SEM log₂ RFU for (B) CCL2, (C) CCL17, and (D) CCL7. *P < 0.05 for SFTPC BRICHOS mutation cases versus disease control using unpaired 2-tailed t test.

Expression of Sftpc^C121Gresults in the development of spontaneous fibrotic lung remodeling. The lung injury and morbidity observed 2 weeks after tamoxifen treatment in the Sftpc^C121G/C121G R26^Cre mice was followed by partial recovery of weight loss and a decrease in BALF cell counts. However, 4 weeks after tamoxifen treatment, lung sections from Sftpc^C121G/C121G R26^Cre mice demonstrated spatially heterogeneous areas of dense trichrome-positive parenchymal remodeling (Figure 8, A and B) in association with the accumulation of smooth muscle actin–positive (SMA-positive) cells adjacent to hyperplastic AT2 cells (Figure 8C). There was also a demonstrable quantitative increase in lung collagen deposition, measured as both soluble collagen content of the lungs by Sircol assay (Figure 8D) and as fibrillar collagen content assessed by picrosirius red staining (Figure 8E). Additionally, total lung Col1a1 and Col3a1 gene expression was increased (Figure 8F), demonstrating active fibrotic remodeling 4 weeks after tamoxifen. The observed structural and biochemical changes correlated with restrictive impairment on lung mechanics, with flow volume curves exhibiting a 30% decline in static lung compliance compared with control (Figure 8G). Quantification of AT2 cells in serial lung sections identified by either proSP-B staining or proSP-C staining at 4 weeks after tamoxifen demonstrated a 23.3% ± 4.0% (mean ± SD, n = 3) and 27.5% ± 3.8% (mean ± SD, n = 3) decline in AT2 cell number from 1 week after tamoxifen, respectively, with heterogeneous regions of higher and lower AT2 cell density (Supplemental Figure 8). Given the concordance in cell number quantified by each AT2 cell–specific marker, it appeared that at 4 weeks after tamoxifen, the majority of AT2 cells continued to express Sftpc^C121G.

Figure 8

Mice expressing the Sftpc^C121G mutation develop spontaneous fibrotic lung remodeling. Representative trichrome staining of Sftpc^WT R26^Cre (A) and Sftpc^C121G/C121G R26^Cre (B) lung section 28 days after tamoxifen treatment showing patchy areas of fibrotic remodeling in Sftpc^C121G/C121G R26^Cre mice. Scale bars in ×0.5-magnification upper panels: 5 mm; scale bars in ×10-magnification lower panels: 100 μm. (C) Double-label immunohistochemistry of Sftpc^C121G/C121G R26^Cre lungs showing proSP-B⁺ AT2 cells (red) with adjacent smooth muscle actin (SMA; green) myofibroblasts (magnification, ×20; scale bar: 100 μm). (D) Soluble collagen in right lung homogenates measured by Sircol assay. Shown are dot-plots with mean and SEM. *P < 0.05 vs. control using unpaired 2-tailed t test. (E) Left: Representative picrosirius red–stained ×20-magnification fields. Right: Quantification performed using ImageJ expressed as picrosirius staining as percentage of section area with dot plots, and mean and SEM shown. *P < 0.05 vs. control using unpaired 2-tailed t test. (F) Whole lung expression of Col1a1 (top) and Col3a1 (bottom) mRNA assayed by qRT-PCR and expressed as fold change from controls. Dot plots and mean and SEM are shown. *P < 0.05 vs. control using unpaired 2-tailed t test. (G) Pooled flow volume loops (top) (n = 10) and calculated static compliance (Cst; bottom) from pulmonary function testing at 28 days after tamoxifen. *P < 0.05 vs. control using unpaired 2-tailed t test. (H) Top: Active TGF-β1 levels in BALF at indicated times after tamoxifen were measured using Luminex; shown are dot plots with mean and SEM. *P < 0.05 vs. control using unpaired 2-tailed t test. Bottom: qRT-PCR for Tgfb1 mRNA expression in AT2 cells isolated at 3 and 14 days after tamoxifen. Data expressed as fold change from control group are presented as dot plots with mean and SEM shown. *P < 0.05 versus controls by 1-way ANOVA with post hoc Tukey’s test.

An increase in the potent antiinflammatory and profibrotic mediator TGF-β1 preceded the appearance of the fibrotic remodeling and was commensurate with the resolution of inflammation at 2 weeks after tamoxifen treatment (Figure 8H). Sftpc^C121G AT2 cells were a contributing source of TGF-β1, with increases in both gene expression of Tgfb1 (Figure 8G) and AT2 cell TGF-β1 staining (Supplemental Figure 9). Concurrently, MMP-7, a component of the profibrotic milieu that was elevated in SFTPC BRICHOS patients’ BALF (Supplemental Figure 10A) and has been described in adults with IPF (50), was also increased in the Sftpc^C121G/C121G R26^Cre mouse BALF during this transitional period to fibrosis (Supplemental Figure 10B). Thus, following the resolution of the alveolitis, the Sftpc^C121G/C121G R26^Cre mice express fibrotic mediators and biomarkers found in ChILD and ILD, and develop spontaneous fibrotic lung remodeling with a substantial restrictive impairment in respiratory mechanics.

ER stress is a recognized pathway in the pathogenesis of fibrosis in multiple organs, including heart, kidney, liver, and lung (51). AT2 cell ER stress is also an appreciated feature of the epithelium in sporadic ILD and IPF (17, 52). There are multiple postulated etiologies for this, including exogenous factors such as air pollution, cigarette smoke exposure (52), and chronic viral infections (53, 54). Additionally, it is hypothesized that epithelial ER stress can arise from dysfunctional epithelial proteostasis in an aging or senescent epithelium (55–57). However, in vivo modeling to derive proof of concept linking AT2 cell ER stress and fibrotic lung remodeling has been elusive. Here we present a model that provides evidence that AT2 expression of a disease-associated SFTPC BRICHOS mutation is sufficient to activate the UPR and produce spontaneous lung inflammation and fibrotic remodeling. Thus, AT2 cell ER stress and our Sftpc^C121G model fit within the growing paradigm that disturbance in epithelial cell quality control drives fibrotic lung disease (7, 58). In support of this, we recently reported on the non-BRICHOS Sftpc^I73T mouse model, where expression of this trafficking mutation causes defective AT2 cell macroautophagy in the absence of ER stress, with resultant spontaneous inflammation and fibrotic lung remodeling (10). These two models taken together strongly support the notion that distinct endophenotypes of AT2 cell dysfunction may each contribute to ChILD and ILD through both overlapping and distinct pathways.

Previous in vivo SFTPC BRICHOS mutation models have been limited by technical difficulties with the control of relevant gene dosages. Constitutive SFTPC^Δexon4 mutation expression in mice resulted in dose-dependent disrupted lung morphogenesis with prenatal lethality (18). While that model recapitulated the cell biology of the SFTPC mutation and established a toxic gain of function for SFTPC BRICHOS mutations, the transgenic expression of this highly aggregation-prone mutation using a human SFTPC promoter fragment is the likely cause of the severe developmental phenotype, limiting its translational relevance. Whereas the SFTPC^Δexon4 model demonstrated a severe developmental phenotype, the inducible transgenic SFTPC^L188Q model lacked a spontaneous phenotype, but had increased susceptibility to bleomycin-induced fibrotic remodeling (19). This model was limited by both low (10% of WT) SFTPC^L188Q mRNA expression and the observation in vitro that this mutated pro-protein was less aggregation prone than other BRICHOS mutations, suggesting a less-severe cellular response to this isoform (59).

Based on the importance of the disulfide cysteine bonds in the BRICHOS domain for primary folding and processing (20) and our in vitro modeling of C121 mutant proSP-C isoforms (Figure 1), we selected an Sftpc mutation that we anticipated would cause significant ER stress. We developed a hypomorphic founder line that expressed low Sftpc^C121G mRNA and no detectable mutant proSP-C, and lacked a spontaneous phenotype. This was in contrast to the hypomorphic Sftpc^I73T line, where an age-dependent lung phenotype developed in parallel with the steady accumulation of mutant proSP-C isoforms due to their mistrafficking and evasion of proteasomal quality control (10). In contrast, SP-C BRICHOS mutations are substrates for proteasomal degradation, limiting their expression until transcription/translation exceeds proteasomal capacity (15), which likely accounts for the difference in protein expression in the two hypomorphic models. Since the founder Sftpc^C121G line failed to generate a spontaneous phenotype, we developed complimentary models with sufficient gene expression to recapitulate various aspects of the human lung phenotype. Constitutive heterozygous Sftpc^C121G expression allowed both proof of concept of the dominant negative effect of the mutant pro-protein on WT proSP-C processing (Figure 2, B and C), and a neonatal lung phenotype that mimicked clinical cases of ChILD with SFTPC BRICHOS mutations (Figure 2, D–F). However, in the adult inducible model we discovered that heterozygous Sftpc^C121G expression was insufficient to develop a robust phenotype. This was similar to what was described in the Sftpc^I73T model and likely reflects that either the developing epithelium is more sensitive to the mutated isoforms or that low Sftpc ^C121G transcription in the heterozygous adult model was inadequate to produce sufficient mutant pro-protein capable of generating a toxic gain of function.

To overcome the limitation of low Sftpc mRNA in the adult inducible model, we selected homozygous Sftpc^C121G expression with inducible Cre-driven expression under a strong Rosa26 promoter. While using this Cre-ERT2 system limited the ability to perform AT2 cell lineage tracing experiments (as could be performed with Sftpc^ERT2Cre), it did produce a gene dose level closer to predicted levels for heterozygous SFTPC BRICHOS mutation patients, and was capable of inducing a stronger ER stress cellular phenotype. Thus, by modulating gene dose in both the constitutive and inducible in vivo models, we were able to produce lung phenotypes that recapitulated the spectrum of pediatric and adult SFTPC BRICHOS–associated ILD, providing proof of concept linking AT2 cell ER stress and spontaneous fibrotic lung remodeling.

The translational relevance of this model was supported by the discovery of multiple biomarkers associated with human ILD in the Sftpc^C121G BALF. These included MMP-7, which is a well-described marker of fibrotic tissue remodeling in IPF (60) and was identified both in our pediatric proteomic analysis (Supplemental Figure 10) and in the Sftpc^I73T model. Additionally, we identified IL-6 and IL-8 in Sftpc^C121G BALF during the inflammatory phase of the model (Supplemental Table 2), consistent with analyses of patient samples during AE-ILD that have identified increased plasma levels of IL-8 and IL-6 as biomarkers associated with worse prognosis (41, 42).

The role of inflammation and immune effector cell populations in ILD is an area of ongoing research, but the importance of this is highlighted clinically by AE-ILDs. Key cellular features of acute exacerbations include epithelial injury and apoptosis, and the recruitment of a mixed granulocytic cell population (61, 62). In this regard, the AT2 cell apoptosis (Figure 4F), alveolitis (Figure 5, F and G), and subsequent fibrotic remodeling (Figure 8) in our model recapitulate the human disease. We found that the Sftpc^C121G AT2 cell was responsible for early cytokine production associated with immune cell recruitment (Supplemental Figure 7). This corroborated previous in vitro studies where expression of SFTPC BRICHOS mutations caused the secretion of cytokines associated with immune cell recruitment (13, 14). The relationship between ER stress and inflammation has been an active area of study in multiple organ systems, including in the pathogenesis of diabetes in the pancreatic B cell (63), hepatic steatosis liver disease (64, 65), and vascular atherosclerosis (66). However, that the Sftpc^I73T model had both a similar immune cell ontogeny and overlapping AT2 cell–produced cytokines suggests that AT2 cell ER stress per se may not underlie the effector cell recruitment (10), but rather that distinct AT2 cell dysfunction endophenotypes converge to produce a common cytokine profile. The regulation of both this immune cell recruitment and then the transition to resolution of alveolitis represent two important areas of future research requiring additional understanding of epithelial-immune cell crosstalk.

Preceding the overt alveolitis, we identified the recruitment of monocyte/macrophage effector cells (Figure 5I and Figure 6). CD11b⁺Ly6C^hi monocytes have been experimentally shown to participate in the development of fibrotic lung disease in exogenous models of injury/repair such as bleomycin and were found to be an early immune population recruited in the Sftpc^I73T model (10, 39, 67). Our models of spontaneous injury suggest that CD11b⁺Ly6C^hi monocytes are recruited to replenish the alveolar macrophage niche, which is consistent with recent evidence showing monocyte-derived alveolar macrophages as a key fibrogenic subset (39). We found the monocyte/macrophage cytokine CCL2/MCP-1 in Sftpc^C121G BALF, which has previously been identified in fibrotic lung disease and preclinical models (10, 48, 49, 68–70). However, we also uncovered elevated levels of CCL17/TARC and CCL7/MCP-3, cytokines with overlapping functional redundancies with CCL2/MCP-1. All 3 of these cytokines were elaborated in the pediatric proteomic analysis (Figure 7), proving a translational context for this discovered redundancy, which may account for the phase II clinical trial failure of pharmacologic MCP-1 neutralization as a therapy in IPF (71).

There are multiple possible mechanisms by which fibrosis develops in this model. During the inflammatory phase of the model, there is AT2 cell apoptosis, which has been associated with the development of fibrosis in AT2 cell ablation models (72). However the 20%–30% decline in AT2 cell number we observed in this model at 4 weeks after tamoxifen treatment was a degree of apoptosis found to be insufficient to induce spontaneous fibrosis in multiple AT2 cell ablation models (73, 74). Alternatively, our data show that the Sftpc^C121G AT2 cell is a source of the potent fibrotic mediator TGF-β1 (Figure 8G), although the true contribution of TGF-β1 to the development of fibrosis in this model would require TGF-β1 depletion. In the lung there has been limited mechanistic work on the link between epithelial ER stress and TGF-β1 (57, 75). However, UPR pathways such as ATF4/CHOP have been implicated in TGF-β1 production in epithelial injury models of liver and kidney fibrosis (76, 77). That AT2 cell Tgfb1 expression in the Sftpc^C121G model increased 2 weeks after tamoxifen treatment demonstrates a temporal sequence of events, with early AT2 cell proinflammatory cytokine production followed by a transition to the release of antiinflammatory and profibrotic TGF-β1. The regulation of this phenotypic transition may provide additional insights into the accelerated fibrosis that occurs during recovery from an AE-ILD.

In conclusion, the Sftpc^C121G model provides proof of principle for the role of AT2 cell ER stress in pediatric ChILD and adult ILD. We have demonstrated that Sftpc^C121G AT2 cells are a source of multiple relevant chemokines/cytokines associated with the recruitment of significant effector cell populations supporting a proximal position for the AT2 cell in the pathway to fibrosis. While the precise UPR pathways responsible for the expression of these cytokines and the fibrotic mediator TGF-β1 would require genetic or pharmacologic inhibition, this model represents a new tool for therapeutic intervention studies such as these. The translational relevance of the model is further supported by the identification of these cytokines and other biomarkers in the high-dimensional BALF proteomic analysis of patients with ChILD from SFTPC BRICHOS mutations. The Sftpc^C121G model thus connects AT2 cell dysfunction manifesting as ER stress with lung inflammation and aberrant lung remodeling, supporting its role as a novel and relevant preclinical model.

Supplemental Methods are available online with this article.

Mouse models. The Sftpc^C121Gneo founder line was commercially produced (Genoway Inc.) using targeting vector–based embryonic stem cell electroporation transfection strategies to knock in coding sequences for the cysteine-to-glycine substitution at codon 121 into the endogenous mouse Sftpc locus (depicted in Supplemental Figure 1A). For constitutive removal of the lox-P flanked PGK-neo cassette from Sftpc^C121Gneo allele in utero, homozygous Sftpc^{C121Gneo/C121Gneo} female mice were bred to the heterozygous X-linked CMV-Cre recombinase line [strain B6.C-Tg(CMV-cre)1Cgn/J; stock 006054; The Jackson Laboratory] (25). For tamoxifen-inducible removal of PGK-neo from Sftpc^C121Gneo alleles, the Sftpc^C121Gneo founder line was crossed to the Rosa26-ERT2-Cre line [strain B6.129-Gt(ROSA)26Sortm1(cre/ERT2)Tyj/J; stock 008463; The Jackson Laboratory]. The progeny was backcrossed to homozygosity of the Sftpc^C121Gneo allele and the R26^Cre allele. All mouse strains and genotypes generated for these studies were congenic with C57BL/6J. Both male and female animals (aged 8–12 weeks) were utilized in tamoxifen induction protocols. Mice were housed under pathogen-free conditions in a barrier facility, and all experimental and breeding protocols were approved by the IACUC of the Perelman School of Medicine at the University of Pennsylvania.

Materials and reagents. Tamoxifen (non-pharmaceutical grade) was purchased from Sigma-Aldrich. Cytological slides were stained with Giemsa (GS500; Sigma-Aldrich). All other reagents were electrophoretic or immunological grade and purchased from either Sigma-Aldrich or Thermo Fisher Scientific.

Lung histology. Whole lungs were fixed by tracheal instillation of a 1.5% glutaraldehyde/1.5% paraformaldehyde mixture in 0.15 M HEPES buffer at a constant pressure of 25 cm H₂O and removed en bloc. Paraffin-embedded 6-μm lung sections were stained with H&E or Masson’s Trichrome stain by the Pathology Core Laboratory of Children’s Hospital of Philadelphia.

Immunohistochemistry and immunofluorescence. Immunohistochemical staining of paraffin-embedded lung sections was performed with primary antibodies at indicated dilutions as described in Supplemental Table 4. A commercial HRP-based method (Vectastain Elite Rabbit IgG ABC Kit PK6101, Vectastain DAB Substrate Kit SK410; Vector Laboratories) was used to visualize immunoreactivity. Following immunoreactivity staining, sections were counterstained with Mayer’s hematoxylin.

Immunofluorescence staining of lung sections was performed with combinations of primary antibodies and secondary Alexa Fluor–conjugated anti-IgG (Supplemental Table 4). Images were visualized on an Olympus I-70 inverted fluorescence microscope, captured with a Hamamatsu 12-bit coupled-charged camera, and processed using Metamorph, version 7.8.4.0 (Universal Imaging).

Quantitative determination of proSP-C– and ProSP-C–positive cells was performed by fluorescence immunostaining of serial lung sections with anti–proSP-C and anti–proSP-B. Five consecutive ×20 images were obtained from 2 different lobes (10 peripheral lung fields per section), with manual counting followed by quantification of positive cells per field. Apoptotic AT2 cell index was determined using double-label fluorescence immunostaining of lung sections with anti–proSP-C and anti–cleaved caspase-3. Five consecutive ×20 images were obtained from 2 different lobes (10 peripheral lung fields per section). Apoptotic (percentage cleaved caspase-3/proSP-C double-positive cells per total proSP-C–positive cells) indices were determined by manual counting.

Picrosirius red staining. Staining for collagen was performed using the Picrosirius Red Stain Kit (Polysciences Inc.) according to the manufacturer’s instructions. Following staining of lung sections, 5 consecutive ×20 images were obtained from 2 different lobes (10 peripheral lung fields per section without large airways) for analysis. Digital morphometric measurements were performed using ImageJ (NIH). The mean area of positive picrosirius staining of each lung field was expressed as a percentage of total section area (10, 78).

Quantification of lung collagen content. Total right lung acid soluble collagen content was determined using the Sircol assay (Biocolor Ltd.) according to the manufacturer’s instructions and as previously described (69).

BALF collection and processing. BALF was collected from mice using 5 sequential 1-ml sterile saline lung lavages. 300-μl aliquots of each cell-free, first 1 ml of BALF return was removed and stored at –80°C for cytokine analysis. Cell pellets recovered from centrifugation of complete return were resuspended in 1 ml sterile saline, and total cell counts determined by a Z1 Coulter Counter (Beckman Coulter). Differential cell counts were obtained by manually quantifying cytospins of BALF cell pellets stained with modified Giemsa (Sigma-Aldrich, GS500). Large-aggregate surfactant fractions were prepared from cell-free BALF by centrifugation at 20,000 g for 60 minutes at 4°C as described previously (79). Total protein content of BALF, surfactant, cell lysate, and whole lung homogenate were assayed by the Bradford method (79).

Cytokine analyses. Aliquots of first-return, cell-free BALF were analyzed for levels of a panel of cytokines [CCL2, CCL11, IL-4, IL-13, IL-5, IL-6, GM-CSF, IL-1β, KC(GroA); panel MCYTOMAG-70K-9] using the Luminex (Luminex Corp.) multiplex platform analyzed by the Human Immunology Core at the Perelman School of Medicine. Active TGF-β1 and CCL17 (TARC) were measured separately, also using Luminex kits (panels TGFBMAG-64K and MECY2MAG-73K). For select cohorts, CCL7 was measured using an MCP-3 ELISA kit (BM6006INST, Invitrogen).

Mouse AT2 cell isolation. Mouse AT2 cells were isolated as previously reported (69). Briefly, a single-cell suspension was obtained by instilling Dispase (BD Biosciences) into perfused lungs, followed by mechanical dissociation with a McIlwain tissue chopper (Metrohm USA) and treatment with 20 μg/ml DNAase I (Sigma-Aldrich). Differential adherence on plastic culture dishes negatively selected mesenchymal cells. CD45⁺ cells were depleted by negative selection using Dynabeads Untouched mouse T cells kit (11413D) and Dynabeads mouse DC enrichment kit (11429D, Thermo Fisher Scientific). Recovered cells were collected and flash frozen at –80°C. Purity was determined by immunostaining preparations adhered overnight to 10% Matrigel-coated coverslips using DAPI in combination with primary antisera for proSP-C. Manual counts in five 20x fields per sample were done, and purity was defined as number of pro-SP-C⁺ cells (AT2 cells) divided by total nuclei showing >95% purity (n = 4).

SDS-PAGE and immunoblotting. SDS-PAGE using Novex Bis-Tris gels (NP0301, Thermo Fisher Scientific) and immunoblotting of PVDF membranes with primary antibody (Supplemental Table 4) was performed. This was followed by species-specific HRP-conjugated secondary antibody and band detection by enhanced chemiluminescence (ECL2 80196 Thermo Fisher Scientific or WesternSure 926-95000, LI-COR Biotechnology) using a LiCor Odyssey Fc Imaging Station and quantified using the manufacturers’ software.

Antibodies. A polyclonal proSP-C antiserum (“NPRO-SP-C”) raised against rat proSP-C and polyclonal anti-proSP-B (“PT3”) raised against bovine proSP-B were each produced in rabbits in-house and previously validated as published (80, 81). The other antibodies used for these studies were obtained from commercial sources (Supplemental Table 4).

RNA isolation and qRT-PCR. cDNA was prepared from RNA template purified from AT2 cell or whole lung RNA using the RETROscript Kit (Thermo Fisher Scientific). qRT-PCR was performed using Taq polymerase and TaqMan RT-PCR kits (Applied Biosystems/Thermo Fisher Scientific) with primer sets for mouse genes (listed in Supplemental Table 5) on an ABI Prism 7900 system and a QuantStudio 7 Flex Real-Time PCR System.

Multichannel flow cytometry. Tissue immune cell characterization was performed as previously described (10). Briefly, en bloc lungs were digested in DMEM + 5% FBS + 2 mg/ml Collagenase D (catalog 11088866001, Roche), passed through 70-μm nylon mesh to obtain single-cell suspensions, and then mixed with Gibco RBC Lysis Buffer (A10492-01, Thermo Fisher Scientific). Cell pellets were resuspended in PBS + 0.1% sodium azide and blocked with anti–mouse CD16/32 antibody (Fc block, eBiosciences), followed by incubation with antibody mixtures (or isotype controls) and conjugated viability dye (see Supplemental Table 6). Single-cell suspensions were then analyzed on an LSR Fortessa (BD Biosciences). They were then gated on viability and singlets and analyzed with FlowJo software (FlowJo LLC) based on a modification of the strategy of Misharin et al. (82) as previously published (10) (Supplemental Figure 6).

Pediatric SFTPC mutation cohort for comparative BALF proteomics. Clinically indicated flexible bronchoscopy was performed, and BALF was collected, processed, and stored in a standard fashion. Disease control subjects were defined as patients with pulmonary symptoms necessitating flexible bronchoscopy, including unexplained wheeze, cough, congestion, and recurrent pneumonia, but normal-appearing bronchoscopy. BALF cytology and microbiologic testing were performed as part of routine clinical care (83). Briefly, the area of lavage was selected by the pulmonologist performing the procedure, and in general was performed in areas where mucus had collected or where infiltrate was seen on chest imaging. Typically, 3 lavages were performed using sterile nonbacteriostatic saline at room temperature. Each lavage aliquot consisted of 1 ml per kilogram of body weight, with a maximum of 30 ml normal saline per lavage. After BALF collection, any excess fluid not required for clinical laboratory testing was processed for research study. Collected BALF was processed in a standard manner. It was initially centrifuged for 10 minutes at 250 x g at 4°C. The pellet was saved, and the supernatant was spun again for 20 minutes at 4000 x g at 4°C. The second pellet was saved. This supernatant was split in half, with half stored 0.5 ml aliquots at –70°C. The other half of the supernatant was treated with the protease inhibitors PMSF and EDTA to inhibit protease activity. The samples were frozen at –70°C for further study.

SOMAmer proteomic analysis. SOMAscan technology (84, 85) was used to simultaneously measure 1129 proteins from each sample at SomaLogic Inc. BALF was diluted with buffer to a standard protein concentration of 20 μg/ml, and 100 μl diluted BALF was equilibrated with a SOMAmers mix (Gold 2010). The SOMAmer mixture allowed for the detection and quantitation of 1129 protein aptamers simultaneously in each sample recorded as relative florescence units (RFU). Raw RFU were adjusted based on the individual dilution factor for each sample to reflect the initial protein amounts.

Statistics. For in vitro and in vivo studies, all data are presented as dot plots and group mean ± SEM unless otherwise indicated. Statistical analyses were performed with GraphPad Prism. Student’s t test (1- or 2-tailed as appropriate) were used for 2 groups and multiple comparisons were done with ANOVA with post hoc testing as indicated; survival analyses were performed using log-rank test. In all cases, statistical significance was considered at P values less than 0.05.

For the pediatric BALF analysis, adjusted RFU were log₂ transformed, and quantile normalization was used to obtain final values for analysis. RFU for each protein aptamer were compared univariately between groups using difference in log₂ means and a 2-sample t test assuming unequal variances. Conservative selection of aptamers included the intersection set with a difference greater than 1 on log₂ scale and a P value less than 0.001. This approach provides a balance between selecting aptamers with a large absolute change and significant P value.

Study approval. Mice were housed in pathogen-free facilities according to protocols approved by the IACUC of the Perelman School of Medicine at the University of Pennsylvania. For the pediatric proteomics data, the study was approved by the Colorado Multiple Institutional Review Board (nos. 99-113 and 10-0472). Informed consent was obtained from all subjects for sample collection and study enrollment. If the patient was a minor, informed consent was obtained from the subject’s legal guardian. In subjects aged 12–17 years, informed assent was obtained.