Expression of Carcinoembryonic Cell Adhesion Molecule 6 and Alveolar Epithelial Cell Markers in Lungs of Human Infants with Chronic Lung Disease

Abstract

The membrane protein carcinoembryonic antigen cell adhesion molecule (CEACAM6) is expressed in the epithelium of various tissues, participating in innate immune defense, cell proliferation and differentiation, with overexpression in gastrointestinal tract, pancreatic and lung tumors. It is developmentally and hormonally regulated in fetal human lung, with an apparent increased production in preterm infants with respiratory failure. To further examine the expression and cell localization of CEACAM6, we performed immunohistochemical and biochemical studies in lung specimens from infants with and without chronic lung disease. CEACAM6 protein and mRNA were increased ~4-fold in lungs from infants with chronic lung disease as compared with controls. By immunostaining, CEACAM6 expression was markedly increased in the lung parenchyma of infants and children with a variety of chronic lung disorders, localizing to hyperplastic epithelial cells with a ~7-fold elevated proliferative rate by PCNA staining. Some of these cells also co-expressed membrane markers of both type I and type II cells, which is not observed in normal postnatal lung, suggesting they are transitional epithelial cells. We suggest that CEACAM6 is both a marker of lung epithelial progenitor cells and a contributor to the proliferative response after injury due to its anti-apoptotic and cell adhesive properties.

Introduction

CEACAM6 (CEA6, NCA 50/90, CD66c), a GPI-anchored membrane protein of the carcinoembryonic antigen (CEA) gene family, is expressed in epithelial cells of many mammalian tissues and is highly enriched in cancer cells of the gastrointestinal tract (Scholzel et al. 2000), pancreas (Blumenthal et al. 2007), and certain pulmonary tumors (Han et al. 2014). The protein is involved in maturation of the human gastrointestinal tract epithelium (Ilantzis et al. 1997) and serves as a component of the intestinal barrier defense. CEACAM6 also functions as an intercellular homophilic and heterophilic adhesion molecule and likely has signaling properties that relate to its anti-apoptotic properties (Blumenthal et al. 2005).

We have previously examined the expression and functions of CEACAM6 in human lung. CEACAM6 protein was identified in second trimester human fetal lung epithelial cells (Kolla et al. 2009) and was induced by treatment with dexamethasone and cAMP, which promotes differentiation of the type II cell phenotype in culture (Wade et al. 2006; Kolla et al. 2009). A closely related family member, CEACAM5 (CEA, CD66e) is also present in the alveolar epithelium and its expression is increased with hormone treatment (Wade et al. 2006; Kolla et al. 2009). In this culture model of type II cell differentiation, CEACAM6 was found to be localized to both plasma membranes and surfactant-containing lamellar bodies, and some protein was secreted and bound to lung surfactant. In functional studies, CEACAM6 was shown to have an anti-apoptotic effect in cultured fetal lung cells (Kolla et al. 2009), consistent with findings in gut epithelium (Ordonez et al. 2000), and protected surfactant function in vitro in the presence of inhibitory plasma proteins (Kolla et al. 2009). We also identified elevated CEACAM6 levels in the tracheal aspirate fluid and enhanced immunostaining signals in the alveolar epithelium lung tissue of premature newborns with respiratory failure (Chapin et al. 2010). These findings led us to further examine the expression and cellular localization of CEACAM6 in infants with lung disease.

In the normal lung, more than 95% of the surface area consists of type I cells with flattened cytoplasmic extensions; the remaining ~5% of surface area is covered by cuboidal type II cells. In normal lung tissue, type II cells are interspersed among type I cells. Injury to the lung from a variety of insults causes damage to type I cells and results in the appearance of contiguous, cuboidal type II-like cells in a process that has been designated as type II cell hyperplasia. Hyperplasia can result from insults/injuries such as environmental toxicants (e.g., silica) (Miller and Hook 1990), hyperoxia (Narasaraju et al. 2003), drugs (e.g., bleomycin) (Degryse et al. 2010), growth factors, such as KGF (Fehrenbach 1999), or chronic lung disease caused by inherited surfactant abnormalities (Hamvas et al. 2004; Stevens 2005; Bullard et al. 2005; Groves et al. 2013), as recently reviewed (Wert et al. 2009). The hyperplasia is hypothesized to reflect proliferation of type II cells and/or progenitor cells, which occurs during the repair of the damaged epithelium (Fehrenbach 2001).

The goal of this study was to further characterize CEACAM6 expression in lung tissue from infants with severe and chronic lung disease. We hypothesized that CEACAM6 expression is up-regulated during prolonged lung injury as a result of various causes (surfactant deficiency, infection, mechanical ventilation, hyperoxia) and contributes to epithelial cell proliferation and hyperplasia as part of the repair process. Our finding of elevated and widespread CEACAM6 expression in the hyperplastic alveolar epithelium in several types of lung injury confirms the alveolar epithelium as the major source of CEACAM6 in the injured lung. Co-expression of both type I and type II cell markers in these cells suggests that hyperplastic epithelial cells are transitional cells in a repairing alveolar epithelium.

Materials & Methods

Lung Tissue

Postmortem and biopsy lung tissues were obtained under IRB-approved protocols of the Children’s Hospital of Philadelphia and included informed consent procedures. Paraffin slides of de-identified cases were obtained from the Pathology Department, Children’s Hospital of Philadelphia. Some donor tissues were obtained from lungs unused (excess tissue) for transplant, under IRB-approved protocols.

Samples of surgical specimens, lung tissue obtained at transplant, and postmortem samples were quick-frozen and stored at -70°C for later extraction of RNA and protein, or fixed and embedded in paraffin, or used for frozen-block preparation. Normal samples included donor lung tissues unsuitable for transplantation, normal tissue resected from lung specimens at the time of surgery for congenital cystic adenomatoid malformation, and normal lungs from infants who died of nonpulmonary causes (n=15). Three of the normal samples were kindly provided as sections from unfixed, frozen blocks by G. Pryhuber (Univ. Rochester). Pathologic lung tissues from infants born at term included surfactant-related mutations mostly obtained at transplant (SP-B deficiency, 10 cases; surfactant protein C (SP-C) mutations, 2 cases; ABCA3 (ATP-binding cassette, sub-family A (ABC1), member 3) mutation, 1 case), alveolar proteinosis (4 cases), and pneumonitis (3 cases); from prematurely born infants, there were 5 cases with a clinical diagnosis of bronchopulmonary dysplasia (BPD). For some samples in each category, frozen tissue for protein or mRNA analysis was not available. A list of samples is included (Supplementary Table S2). The age of normal and injured lung specimens mostly ranged from 1 day to 2 years, with some of unidentified age and two specimens from patients that were older (1 BPD, 7 years; 1 ABCA3 mutation, 12 years).

Human Fetal Lung Explants and Isolated Epithelial Cells

Human fetal lung tissue (18–23 weeks’ gestation) was obtained from Advanced Bioscience Resources, Inc (Alameda, CA) under IRB-approved protocols. Explants were prepared and cultured, and epithelial cells were isolated and cultured as previously described using Waymouth’s media with DCI (10 nM dexamethasone, 0.1 mM 8-Br-cAMP, 0.1 mM isobutylmethylxanthine) to promote differentiation (Gonzales et al. 2002).

Immunofluorescence Staining

Sections (5 µm) of paraffin-embedded tissues (prefixed with 1–4% paraformaldehyde) were cut, mounted, and deparaffinized (including 1 hr in 70% EtOH, 0.25% NH₃ to decrease endogenous fluorescence (Baschong et al. 2001)), treated by antigen retrieval (no. H-3300, Vector Laboratories, Burlingame, CA), permeabilized with 0.1–0.3% Triton-X 100, and then immunostained as described elsewhere (Kolla et al. 2009) with minor modifications.

Primary antibodies used were (see Supplementary Table S3): anti-human CEACAM6 (1:100, 9A6; Santa Cruz Biotechnology, Dallas, TX), rabbit anti-sheep SP-B (1:100; Millipore, Billerica, MA), goat anti-human SP-A (1: 100, #S8400-01; US Biological, Swampscott, MA), rabbit anti-human SP-D (1:100, scH-120; Santa Cruz Biotechnology), anti-human CD45-Alexa 488 tagged (1:50, H130 clone; Biolegend, San Diego, CA), rabbit anti-TTF1 (1:100, H-140; Santa Cruz Biotechnology), synaptophisin (1:50, SY-38; Millipore, Darmstadt, Germany), rabbit anti-human CC-10 (1:100, H-75; Santa Cruz Biotechnology), rabbit anti-pan cytokeratin (1:50, H-240; Santa Cruz Biotechnology), rabbit anti-human aquaporin 5 (AQP5, 1:50, LS-B8978; LifeSpan Biosciences, Seattle, WA), CD68-Alexa Fluor 488 (1:50, KP1; Santa Cruz Biotechnology), goat anti-human CGRP (N-20, sc8856; Santa Cruz Biotechnology), rat anti-mouse CD11b (1:50; Serotec, Raleigh, NC), anti-HTII-280, a type II cell-specific plasma membrane marker in human lung (1:100; Gonzalez et al. 2010), DC-LAMP (1:100, clone 104.G4; Beckman-Coulter, Danvers, MA), and anti-HTI-56, a type I cell-specific plasma membrane marker in human lung (1:10; Dobbs et al. 1999). Secondary antibodies, species appropriate, were conjugated to Alexa Fluor 488 or Alexa Fluor 594 (Invitrogen). Immunostaining with anti-PCNA-biotin tagged (#MA5-11355, PC10; Thermo Scientific, Waltham, MA) was detected with HRP-conjugated streptavidin followed by the tyramide-enhanced method with fluorescein label (Erber et al. 1997). Anti-CEACAM6 antibody was directly labeled with Alexa Fluor 488 using the Zenon labeling kit, as per the manufacturer’s instructions (Life Technologies; Grand Island, NY). Controls were performed in the absence of primary antibody or with nonimmune IgG of appropriate species.

Frozen sections of DCI-treated explants or injured lung tissues were triple-stained first with anti-CEACAM6 (1:100) and anti-HTII-280 (1:50), followed by anti-mouse IgG1-Alexa Fluor 488 (1:1000) and anti-mouse IgM-Alexa Fluor 594 (1:500), then with anti-HTI-56 (1:10) directly conjugated with Alexa Fluor 405, again using the Zenon labeling kit. Incubations were 60 min at room temperature for each antibody.

Most slides, except in the case of the triple staining, were stained with DAPI (1.6 µg/ml, 4’, 6-diamidino-2-phenylindole; Southern Biotech, Birmingham, AL) before mounting with Fluoromount-G (Molecular Probes; Eugene, OR), and imaged with an Olympus 1X81 microscope equipped with epifluorescence and Metamorph imaging system (Universal Imaging; West Chester, PA). Sections of each lung tissue were stained with hematoxylin and eosin to assess morphology (not shown).

Frozen Sections

Most tissue samples were fixed with 4% paraformaldehyde in PBS, then embedded in Tissue Freezing Medium (Triangle Biomedical Sciences; Durham, NC). Some sections were cut from frozen blocks of unfixed lung (3 normal blocks; a gift from Gloria Pryhuber, University of Rochester, NY) and were fixed in 4% paraformaldehyde immediately prior to staining. Frozen sections (5 µm) were permeabilized and stained as above (no antigen retrieval used).

Western Immunoblotting

Frozen tissue was crushed, then extracted in RIPA lysis buffer (50 mM Tris-HCl, 150 mM NaCl, 5 mM EDTA, 5% glycerol, 1% Triton X-100 with protease inhibitor cocktail tablets added (Complete Protease Inhibitor tablets; Roche Applied Science, Indianapolis, IN)). Protein concentration of extracts (centrifuged at 11,500 RPM, 10 min, 4°C) was measured with the Bio-Rad protein assay (Hercules, CA). Extracts (20-50 µg protein) were run on 10% NuPage Bis-Tris polyacrylamide gels with MES running buffer (Thermo Scientific), then transferred to nitrocellulose membranes as per the manufacturer’s instructions, as previously described (Kolla et al. 2009). Immunoblotting was carried out with primary antibodies (1:1000–2000) against CEACAM6, SP-B, TTF1 and GAPDH (loading control; Chemicon, Temecula, CA) and infrared-detectable secondary antibodies conjugated to Alexa Fluor 680 (1:10,000 dilution; Molecular Probes) or to IRDye800 (1:10,000 dilution; Rockland, Gilbertsville, PA). Densitometry of immunoblots was assessed with the Odyssey infrared imaging systems (Licor Biosciences; Lincoln, NE).

qPCR to Assess Gene Expression

Total RNA was extracted from frozen lung tissues with RNeasy Mini Kit (Qiagen; Boston, MA) as per the manufacturer’s instructions, and 2 µg of RNA was reverse transcribed using the Superscript First-Strand RT-PCR kit (Invitrogen), as previously described (Kolla et al. 2009). Complementary DNA products were amplified by PCR using 3’ and 5’ GAPDH primers as a test for adequate product formation, and assessed by agarose gel electrophoresis with ethidium bromide under UV. Quantitative real-time PCR (qPCR) reactions were run with a singleplex format on an ABI SDS-7900HT machine (Applied Biosystems) using the ABI primer/probe sets listed in Supplementary Table S1.

Data were analyzed using a Student’s unpaired t-test and significance is indicated as p<0.05 or p<0.01 for injured versus normal samples.

Results

Cellular Expression of CEACAM6 in Human Fetal Lung

We previously reported the stimulatory effect of dexamethasone plus cAMP/isobutylmethylxanthine (DCI), a hormone mixture that accelerates differentiation of isolated alveolar type II cells (Gonzales et al. 2002), on the expression of CEACAM6 in explant cultures, and isolated epithelial cells of second trimester human fetal lung (Kolla et al. 2009). To further explore the cellular identity of CEACAM6-expressing cells in fetal lung, we performed fluorescence immunostaining in lung explants for CEACAM6 and markers of adult type I (HTI-56) (Dobbs et al. 1999) and type II (HTII-280) (Gonzalez et al. 2010) cells.

Explants of human fetal lung expressed higher levels of CEACAM6 in DCI-treated cultures as compared with non-DCI treated controls (Figure 1A, 1D, 1G), confirming previous mRNA and protein results in isolated cells (Kolla et al. 2009). The CEACAM6 immunoreactivity co-localized with markers for both type I cells (HTI-56, Fig 1C–1E) and type II cells (HTII-280, Fig 1F–1H) in much of the epithelium that lined the airspaces of DCI-treated explants.

Figure 1.

Immunostaining for CEACAM6 and markers of alveolar type I and type II cells in human fetal lung explant cultures (A–N) and isolated epithelial cells (O–T). CEACAM6 expression is low in control (5 days, no hormones) explant cultures (A, 23-weeks’ gestation lung) and was increased by 5 days in culture with DCI (D, G). Explants cultured with DCI (C–H) also showed high expression of HTI-56 (E) and HTII-280 (H), and CEACAM6 colocalized with either HTI-56 (C) or HTII-280 (F) in many of the epithelial cells lining the airspaces. Utilization of anti-CEACAM6 antibody tagged directly with Alexa Fluor 488 allowed co-staining with the HTI-56 monoclonal antibody. In a representative experiment (n=3), explants of uncultured lung (20 weeks; I–K) showed a few airspace-lining epithelial cells staining positively for HTI-56 (J) and/or HTII-280 (I, K), with HTII-280 typically observed in larger airspaces, (I, arrowhead) and HTI-56 staining smaller airspaces (J, arrow) and some co-staining (asterisk) as well; airspace size likely reflects sectioning across various regions along the branching airways in the developing lung, with distal tips appearing smallest (unpublished observation). Expression of both markers increased markedly in DCI-cultured explants (L–N). Many cells showed both HTI-56 and HTII-280 localized to the epithelial cell apical plasma membrane (high power insets, L–N), and luminal staining (L–N) was sometimes present from sloughed cells or debris. Culture of isolated epithelial cells in DCI for 5 days also increased the expression of both HTI-56 and HTII-280, which co-stained the plasma membranes of many cells (O–Q, arrow). HTI-56 is also co-expressed with surfactant protein B (SP-B), a well-studied lamellar body marker, in some of the cells (R–T, arrow). blue, DAPI nuclear staining; a, alveolar space. Scale, 20 µm.

The expression of both type I cell and type II cell markers also increased after culture in DCI (Fig. 1L–1N) as compared with uncultured tissues (Fig. 1I–1K) or explants cultured 5 days in the absence of DCI, which showed a minimal increase in expression (Supplementary Fig. S4). In the DCI-treated explants, many of the contiguous epithelial cells lining the airspaces expressed both HTI-56 and HTII-280 but other cells expressed only one of the two antigens. Much of the co-staining was localized to the plasma membrane, as shown in the higher power views (Fig. 1L–1N insets). In the uncultured fetal lung, relatively few epithelial cells expressed either marker, with HTII-280 preferentially expressed in the epithelium of larger airspaces (Fig. 1I, arrowhead) and HTI-56 in the epithelium lining the smaller airspaces (Fig. 1J). However, some co-expression was observed (see asterisk). This pattern suggests differential spatiotemporal expression of the two markers, but induction of both by DCI. CC10 was not expressed in the parenchymal epithelial cells of explants cultured in the presence or absence (control) of DCI; however, a few cells in the small airways (columnar epithelium) of the explants were CC10 positive (CEACAM6 negative, not shown).

In monolayer cultures (5 days in DCI) of epithelial cells isolated from uncultured lung, co-expression of HTI-56 and HTII-280 was present in many cells (Fig. 1O–1Q), similar to that observed in the explants. Moreover, cells expressing HTI-56 also expressed surfactant protein B (SP-B), a type II cell marker for the induction of the surfactant program by DCI treatment of the cells (Fig. 1R–1T, arrow). Thus, hormone-treated fetal lung epithelium co-expressed markers for type I and type II cells, and in some cells, CEACAM6 was also co-expressed, consistent with the appearance of a transitional cell type during lung maturation.

Co-expression of all three antigens (CEACAM6, HTI-56, HTII-280) by a fraction of the epithelial cells in DCI-treated explants was confirmed by triple staining with indirectly detected CEACAM6 (Alexa 488), and HTII-280 (Alexa 594) and directly tagged HTI-56 (Alexa 405), as shown in Fig. 2. Staining with the individual antibodies is shown in Fig. 2A–2C, and the overlay in Fig. 2D. Approximately 10% of the CEACAM6-positive cells also co-stained for both HTI-56 and HT2-280, with most other cells co-staining for CEACAM6 and either HTI-56 or HTII-280.

Figure 2.

CEACAM6 co-expressed with both HTI-56 and HTII-280 in epithelium of DCI-treated fetal lung explants. Explants of human fetal lung (23-weeks’ gestation) were cultured in DCI for 5 days to promote epithelial differentiation. Frozen sections were stained first with anti-CEACAM6 and anti-HTII-280, followed by secondary antibodies, anti-mouse IgG1-Alexa Fluor 488 and anti-mouse IgM-Alexa Fluor 594, followed by directly conjugated anti-HTI-56-Alexa Fluor 405. HTI-56 (A), HTII-280 (B) and CEACAM6 (C) were each expressed in some cells, with all three proteins co-expressed in an estimated 10% of the epithelial cells (D; arrow/arrowhead). (E) Morphology is shown by phase contrast. Scale, 20 µm.

CEACAM6 Expression in Normal Postnatal Lung

CEACAM6 immunostaining at 1 day of age after term birth was mostly localized to dispersed, infrequent cells that did not co-stain for the type II cell marker HTII-280, which was expressed in many noncontiguous cells (Fig. 3A–3C). The findings at 12 days of age were similar to that at 1 day, except for additional punctate CEACAM6 immunoreactivity in some of the cells lining the airspaces (asterisk, Fig. 3D–3F). By 10 months of age, small punctate CEACAM6 staining was evident over much of the alveolar epithelium, consistent with low but widespread staining of type I cells in adult lung (Chapin et al. 2010), with little localization to cells staining with HTII-280 (Fig. 3G–3I, **) or with TTF1, another type II cell marker (not shown). Although consistent with type I cell localization observed in adult lung, CEACAM6 localization to HTI-56-expressing cells in normal infant lung was not convincingly demonstrated, perhaps due to a low expression and/or low sensitivity of the directly labeled antibody.

Figure 3.

CEACAM6 and HTII-280 expression in normal infant lung (1 day, 12 days, and 10 months of age). Punctate CEACAM6 staining is of low intensity in scattered epithelial cells at 1 day of age (A–C, arrow). Staining intensity is somewhat increased in the 12-day lung (D–F, asterisk), and markedly increased with punctate staining covering most of the alveolar surface at 10 months (G–I, **). HTII-280 stained the apical membrane of discrete cells in most alveoli; CEACAM6 immunostaining did not co-localize with HTII-280 at any age. Sections were from unfixed tissues embedded in frozen blocks, and were fixed with 4% paraformaldehyde immediately prior to staining. Nuclei in all images are stained with DAPI (blue). Scale, 20 µm.

Cell counts in three normal lung specimens (surgical resections, 1–3 months age) showed that only 4.7% ± 1.2% of the CEACAM6-positive cells co-stained with either HTII-280 or SP-B (counted 112–273 cells in 3–5 fields per sample), whereas 90% of the SP-B-positive cells co-stained for HTII-280. These results resemble findings in adult human lung, where CEACAM6 was detected at the plasma membrane of most type I cells and only in a subpopulation of type II cells (Chapin et al. 2010), a pattern reflecting the immunostaining for the type I cell antigen HT1-56 in these specimens (not shown). Thus, the CEACAM6 staining pattern changes during the alveolar phase of lung development.

Effect of Lung Injury on CEACAM6 mRNA and Protein Expression

Based on the observations in fetal lung explants, where hormone treatment induces CEACAM6 and produces a hyperplastic-like type II cell response, we hypothesized that injury resulting in epithelial cell hyperplasia in infant lung would increase CEACAM6 expression. In postmortem lungs of infants with respiratory failure secondary to various disorders, CEACAM6 protein was increased ~4-fold as compared with control lungs (Fig. 4A, 4B). Both the fully glycosylated (~90 kDa) and unglycosylated (~50 kDa) bands were increased in the injured samples (Fig. 4B), as well as partially glycosylated intermediate forms.

Figure 4.

Effect of lung injury on CEACAM6 expression. (A) CEACAM6 protein expression determined by western analysis was increased ~4-fold (*p<0.05 vs normal) in the surfactant-related mutation group (n=10; surfactant protein B (SP-B) deficient (n=8), surfactant protein C (SP-C) mutation (n=1), ATP-binding cassette, sub-family A (ABC1), member 3 (ABCA3) mutation (n=1), light gray bar) and ~5-fold (*p<0.05 vs normal) in the group with other lung diseases (n=9; bronchopulmonary dysplasia (BPD) (n=5), alveolar proteinosis (n=2), pneumonitis (n=2), dark gray bar). Densitometric values were corrected to GAPDH levels for each sample, then normalized to the mean of normal samples on each blot. TTF1 protein expression was not significantly different between groups (not shown). (B) A representative blot showing several normal samples (lanes 1–4) and several samples from injured lungs (lanes 5–11). The anti-CEACAM6 antibody recognizes the unglycosylated CEACAM6 protein (~50 kDa) as well as partially and fully glycosylated forms ranging from~70 to 90 kDa (Chapin et al. 2010). (C) CEACAM6 and CEACAM5 mRNAs expression in lung tissues (same as used for protein expression analysis) showed levels for both genes were significantly increased 4- to 5-fold (*p<0.01) for surfactant-related mutations (n=10, light gray bar) and other lung diseases (n=9, dark gray bars) vs normal lungs (n=8, open bar). All values (corrected for 18S) were normalized to the EPCAM content of each sample (to correct for epithelial cell content) and then this corrected value for each injured or normal lung was normalized to the mean of all the normal lung tissues. (D) mRNA expression for TTF1 and aquaporin5 (AQP5) was not different between normal and injured lungs. SP-B mRNA was decreased (p<0.05 vs normal) and SP-C was increased (p<0.05) in the SP-B mutation group, as expected (Beers et al. 2000). Values of these four mRNAs were normalized to EPCAM for each sample and then to the mean value for all normal tissues, as described above.

The expression of CEACAM6 mRNA was also elevated ~5-fold in lungs of term infants with surfactant-related genetic mutations (surf muts; Fig. 4C) and ~4-fold in preterm infants with BPD (other; Fig. 4C). These data were normalized to EPCAM expression to account for changes in epithelial cell density that may have occurred as a result of injury, age or tissue sampling. In microarray studies of fetal lung induced by DCI-treatment to differentiate in culture, CEACAM6 expression was elevated ~10-fold, whereas EPCAM expression was not detectably elevated (Kolla 2009; Wade 2006). CEACAM5 mRNA content was also elevated ~4-fold in the group with surfactant mutations (Fig. 4C), but the ~4- to 5-fold increase in the group with other diseases was not statistically significant. In the same tissues, levels of TTF1 and AQP5 mRNAs, as markers of type II and type I cells, respectively, were not significantly different between the groups (Fig. 4D). The eight SP-B deficient tissues in the surfactant mutation group had lower SP-B mRNA, as expected (0.54 ± 0.16 vs 1.19 ± 0.17 for the normal, p<0.01; Beers et al. 2000), and there was a compensatory 2-fold increase in SP-C mRNA; this resulted in a lower ratio of SP-B/SP-C for the SP-B deficient group (0.42 ± 0.15 (n=8) vs 2.29 ± 0.61 (n=9) for other tissues; p<0.02). These findings are consistent with transcriptional activation of CEACAM6 expression with lung injury.

CEACAM6 Co-expression with Type II Cell Markers in Injured Neonatal Lung Tissue

By immunofluorescence staining, CEACAM6 expression was shown to be elevated in the injured alveolar-lining epithelium and was often present in contiguous cuboidal cells, as shown in the representative fields from lungs with SP-B deficiency (3 weeks; Fig. 5A–5F), an ABCA3 mutation (12 years old, Fig. 5G–5I), or BPD (2.5 months, Fig. 5J–5L, 5N). These cells were shown to be CC10 negative (Suppl Fig. S2) but TTF1 positive (not shown). Anti-HTII-280 immunostained only the apical plasma membrane of the alveolar type II epithelial cells, clearly overlying the CEACAM6 localized in a punctate pattern beneath the plasma membrane (Fig. 5D–5F). There was a similar immunostaining pattern for CEACAM6 in all lungs with hyperplasia of the alveolar epithelium (6 surfactant mutation, 3 BPD, 2 pneumonitis samples).

Figure 5.

Co-localization of CEACAM6 expression with type II cell markers in hyperplastic alveolar epithelium of diseased lung. (A–C) Surfactant protein B (SP-B)-deficient lung (3 weeks; resected at transplant) showed extensive co-localization of CEACAM6 (green) and HTII-280 (red) immunostaining in type II cells of lung parenchyma. (D–F) Higher-power images of SP-B-deficient lung demonstrated HTII-280 staining of plasma membrane of epithelia lining the airspaces (D, asterisk) with underlying CEACAM6 vesicular cytoplasmic staining. (G–I) Lung parenchyma from a patient with an ATP-binding cassette, sub-family A (ABC1), member 3 (ABCA3) mutation (12 years old at lung transplant) showed co-localization of CEACAM6 (green) with many, but not all, of the HTII-280-positive type II cells (red). (J–L) CEACAM6-positive type II cells were present in the airspaces of a bronchopulmonary dysplasia (BPD) lung tissue (2.5 months; postmortem). The cells appear to have rounded up and sloughed into the airspace (J, arrow). A higher-power view clearly shows that the cell cytosol is positive for CEACAM6 (green), and surrounded by HTII-280 staining of the apical plasma membrane (M, red, arrowhead); this identifies the cells in the airspace as type II cells (L, arrowhead for HT2-280). In this tissue (N), the plasma membrane of the hyperplastic epithelium showed yellow co-staining for CEACAM6 (green) and HTII-280 (red). (O–Q)Anti-CEACAM6 (green) also stained many (asterisk) but not all (arrow) of the SP-B (red)-positive cells in the hyperplastic tissue of the ABCA3 mutant sample. (R–T) Immunostaining of injured lung (surfactant protein C (SP-C) mutation; 13 months) showed extensive content of both CEACAM6 and SP-A in the airspaces (R, arrow) as well as co-staining of the extensive hyperplastic type II cells lining the airspaces (R, arrowhead). SP-D immunostaining also co-localized with CEACAM6 in the airspaces (not shown). Nuclei in all images are stained with DAPI (blue). Scale, 20 µm.

CEACAM6 was also co-expressed with SP-B in some cells, but many SP-B-expressing cells did not express CEACAM6, as shown in a representative section (Fig. 5O–5Q; arrow).

Material in the airspaces of injured lung specimens contained many cells, as well as CEACAM6-positive stained material. Some of these cells appeared to be sloughed type II cells, identified by positive immunostaining for CEACAM6 surrounded by an HTII-280-positive membrane (Fig. 5J–5N) as well as staining for SP-B surrounded by HTII-280 (not shown). Both the lower power (Fig. 5L) and higher power (Fig. 5M) images clearly show a CEACAM6-stained cell cytosol around the nuclei with HTII-280-stained limiting membranes. The epithelial cells lining the airspaces in this particular tissue sample co-express CEACAM6 and HTII-280 (Fig. 5N), in contrast with the normal postnatal lung (Fig. 3). The contents of these alveolar spaces stained minimally with anti-CD-68 Alexa Fluor 488, suggesting they are not primarily of macrophage origin (not shown) (O’Reilly et al. 2003). These findings suggest shedding of hyperplastic type II cells into the alveolar space, which is consistent with the presence of type II cells in bronchoalveolar lavage fluid, as previously described (Stanley et al. 1992).

CEACAM6 co-localized extensively with cytokeratin, but not with CD45 (common leukocyte antigen, which marks most hematopoietic-derived cells except platelets; Trowbridge and Thomas 1994), or with CD68 (a monocyte/macrophage marker; O’Reilly et al. 2003) (Supplementary Fig. S1), but co-staining with anti-CD11b showed some overlap, suggesting CEACAM6 expression in some macrophages (not shown). Synaptophisin and CGRP (markers of neuroendocrine cells; Lyda and Weiss 2000; Watkins et al. 2003) were negative (not shown). No CEACAM6 was detected in CC10-stained small airway cells (a Clara cell protein, Supplementary Fig. S2; Ryerse et al. 2001), confirming parenchymal epithelial cell identity of the hyperplastic cells.

CEACAM6 Co-localization with Other Type II Cell Proteins: SP-A, SP-D

Co-staining for CEACAM6 and SP-A showed little immunoreactivity for either protein in the airspaces of normal lungs (3 lungs; data not shown) but large amounts for both proteins in the airspaces of eight injured lungs (Fig. 5R–5T). Similarly, SP-A co-localized with CEACAM6 in the hyperplastic alveolar epithelium of the injured tissues (Fig. 5R–5T), with little co-staining in the normal epithelium. Immunostaining with anti-SP-D showed a similar pattern to that of SP-A relative to the CEACAM6 staining, but SP-D immunosignals were not as strong as that seen for SP-A (data not shown). These observations are consistent with CEACAM6 release into airspaces of injured epithelium both by secretion, as for SP-A and SP-D (both secreted glycoproteins), and by cell sloughing, as observed for SP-B.

Expression of Type I Cell and Type II Cell Markers in Injured Alveolar Epithelium

In the alveolar epithelium of normal infant lung, anti-HTI-56 and anti-HTII-280 stained discretely different cells with no co-localization (Fig. 6A–6C; 1-month-old surgical specimen), as reported for adult lung (Dobbs et al. 1999, Gonzalez et al. 2010). In contrast, the hyperplastic epithelium of injured lungs showed extensive co-staining for markers of both mature type I cells (HTI-56) and mature type II cells (HTII-280), as illustrated by the representative section from a pneumonitis specimen (10 months; Fig. 6D–6F) and a high-power image from an SP-B-deficient lung (10 months, postmortem; Fig. 6G–6I). In addition, CEACAM6 directly tagged with Alexa Fluor 488 (to allow co-staining with 2 monoclonal antibodies) displayed extensive co-localization with HTI-56 in injured alveolar epithelial cells (Fig. 6J–6L; pneumonitis, 10 months). A few cells were positive for all three markers (CEACAM6, HTI-56 and HT2-280) by triple immunofluorescence staining (Supplementary Fig. S3; pneumonitis, 10 months). Thus, during the response to various types of lung injury, a population of epithelial cells appears that expresses type I and/or type II cell markers as well as CEACAM6.

Figure 6.

Co-expression of type I cell and type II cell markers in injury. Immunostaining of normal lung (A–C; 1 month, surgical resection tissue) showed discrete staining for markers of type I (HTI-56) and type II (HTII-280) cells in different epithelial cells, with the absence of detectable co-staining of these markers, as expected (Dobbs et al. 1999). In contrast, in representative injured lungs (D–F, pneumonitis, 2 months, postmortem; G–I, surfactant protein B (SP-B)-deficient lung, 10 months at transplant) there was extensive co-expression of HTI-56 and HTII-280 in most of the hyperplastic alveolar epithelium. (J–L) Anti-CEACAM6 antibody directly tagged with Alexa Fluor 488 showed extensive co-localization of CEACAM6 with HTI-56 in the hyperplastic alveolar epithelium of the injured lung (pneumonitis; 2 months). Nuclei were stained with DAPI (blue). Scale, 20 µm.

Cell Proliferation in Normal and Injured Lung Tissue

Lung tissue from two patients with surfactant-protein mutations (SP-B-deficient, SP-C mutation) and normal tissue were stained for both CEACAM6 and the nuclear proliferation marker PCNA. Figure 7 show a representative section illustrating both intensely and lightly stained PCNA-positive nuclei in the SP-B deficient lung. Of CEACAM6-positive epithelial cells (103–250 cells in 5–14 fields counted), 34.8% ± 4.5% and 32.7% ± 11.6% per field were PCNA-positive in the two injured lungs, versus 7.3% ± 3.2% of CEACAM6-positive cells for the normal lung (p<0.05); more PCNA-positive nuclei were also present in the airspaces (~30% for injured lungs versus ~0% for normal). CEACAM6-negative interstitial cells (214–325 cells in 5–8 fields counted) were 13.8% ± 4.8% and 12.7% ± 7.4% PCNA-positive in the injured lungs and 36.3% ± 4.3% positive in the normal lung (p<0.05). Thus, cell proliferation likely contributes to the expanded population of CEACAM6-positive epithelial cells associated with lung injury. In the injured tissues, many of these cells also express type I and/or type II cell markers (see Fig. 5A–5N; Fig. 6D–6I), consistent with a transitional cell in a pathway to mature type I and type II cells.

Figure 7.

Increased proliferation of the hyperplastic alveolar type II cells in injured lung. (A–C) Some of the CEACAM6-positive cells (red) in the alveolar epithelium of an surfactant protein B (SP-B)-deficient lung (5 weeks, postmortem) contain PCNA-positive nuclei, with some strongly positive (A, arrows) and some weakly positive (B, arrowheads); some CEACAM6-negative cells in the interstitium are also PCNA-positive (A, B, asterisk). In the normal neonatal lung (transplant donor, age unknown), very few alveolar epithelial cells were PCNA-positive (not shown). Nuclei were stained with DAPI (blue). Scale, 20 µm.

Discussion

Our findings indicate that CEACAM6 expression is increased in the lung parenchyma of infants/children with a variety of lung disorders that result in severe chronic lung disease. This supports the report that CEACAM6 is one of the genes with increased expression, as determined by transcriptomic analysis of postmortem lung tissue from infants with BPD (Bhattacharya et al. 2012). We localized CEACAM6 expression to hyperplastic epithelial cells that represent proliferating cells in response to injury. These cells also co-express markers of type I and/or type II cells, which is not observed in normal adult lung (Dobbs et al. 1999; Gonzalez et al. 2010), suggesting that they are transitional epithelial cells arising from proliferation of progenitor cells. The transitional cell hypothesis is supported by our explant studies in which fetal lung samples, cultured in DCI, showed induced expression of both type I and type II markers, as well as CEACAM6, in the fetal epithelium. The occurrence of transitional cells was also suggested in an earlier study of developmental changes in the expression of a lectin surface marker of type II cells (Joyce-Brady and Brody 1990).

Based on these observations, we propose the following model for the role of CEACAM6-expressing cells in regeneration of the injured epithelium: In the normal infant lung, there is a small population of progenitor cells within the alveolar epithelium that expresses CEACAM6 at a relatively low level, and comprises neither type I nor type II cells. In response to lung injury, these cells increase the expression of CEACAM6 and CEACAM5 (which responds similarly but is in much lower abundance in fetal lung cells (Kolla et al. 2009)), as well as other unidentified proteins, and undergo proliferation and differentiation into a transitional epithelial cell-type that expresses both type I and type II markers, with subsequent transition into type I or type II cells. As such, these cells contribute to the replenishment of the epithelial cell layer. We further propose that CEACAM6 is both a marker of these progenitor cells and a contributor to the proliferative response by virtue of its anti-apoptotic and cell adhesive properties (Ordonez et al. 2000; Cameron et al. 2012; Han et al. 2014).

Possible stimuli for increased CEACAM5/6 gene expression with lung disease are uncertain. Both genes are induced in epithelial cells of human fetal lung by treatment with glucocorticoids and cAMP (Kolla et al. 2009), and the newborn lung disease Respiratory Distress Syndrome is associated with increased plasma cortisol concentration (Ballard et al. 1980); thus, stress-related hormones may contribute to the induction of the proteins. Lung disease and associated oxygen and ventilator treatment in infants produces an inflammatory response that could contribute to increased CEACAM5/6 expression; however, regulation by inflammatory mediators has not been tested. Desai et al. (2014) have proposed that progenitor cell proliferation in injured lung may be mediated by protein(s) released from injured type I cells; this possibility could be tested in future cell culture experiments with human lung tissue.

Recent interest and investigation in the area of lung progenitor cells has elucidated mechanisms of alveolar epithelial cell development and repair in mice; however, experimental adaptation to explore similar processes in human lung is incomplete. Several investigators have identified niche type II cells as possible progenitor cells that refurbish damaged alveolar epithelium in mice after injury (Rock and Hogan 2011; Barkauskas 2013), although others find ‘a previously uncharacterized cell type’ (SPC⁺, CC10⁺ (Kim et al. 2005); EpCAM⁺, CD49⁺, CD104⁺ (McQualter et al. 2010); or α6β4⁺, SPC^- (Kim et al. 2005; McQualter et al. 2010; Chapman et al. 2011) as the likely alveolar progenitor cell population. In a recent study with mice, using a battery of molecular markers, lineage tracing, and clonal analysis, Desai et al. (2014) elegantly elucidated the lineage pathway of alveolar epithelial type I and type II cells arising from bi-potent progenitor cells. During development, type I and type II cells were found to be directly derived from bi-potent progenitor cells expressing markers attributable to both cell types, whereas, following birth, a ‘switch’ occurs after which both type I and type II cells are derived from ‘rare’ self-renewing mature type II cells, presumably ‘niche’ type II cells. The ‘transitional cells’ that simultaneously express markers specific to both the type I and type II cells in developing normal fetal epithelium (Fig. 1) and injured human infant (Fig. 6) and adult lung (Gonzalez et al. 2010), may be homologous to the mouse progenitor cell population described by Desai et al. We found that CEACAM6 expression appears to be congruous with the localization pattern of progenitor cells delineated by Desai and colleagues. Moreover, CEACAM6 expression was identified in cells simultaneously expressing both type I and type II cell antigens, supporting a progenitor role for these cells. Although lineage tracing in humans is not feasible, and some markers used for mouse studies are not useful for humans, analysis of ‘human progenitor cells’ using single cell RNAseq (Treutlein et al. 2014) at various stages of development and during repair after injury should prove useful in characterizing these cells and elucidating the role of CEACAM6.

Our results confirm and extend upon previous observations regarding the expression of epithelial cell proteins detected by antibodies HTI-56 and HTII-280, which are unique markers for human alveolar epithelium. In adult human lung, HTI-56 immunostaining is restricted to type I cells (Dobbs et al. 1999), and HTII-280 immunosignals to type II cells (Gonzalez et al. 2010); in areas of adult lung, with nearly contiguous HTII-280 staining, which presumably represent epithelial cell hyperplasia secondary to localized injury, some co-localization of the two markers is observed (Gonzalez et al. 2010). In our studies of human fetal lung epithelial cells differentiating in culture, we found similar examples of co-expressed HTI-56 and HTII-280 markers. Expression of both antigens increases with advancing gestational age and during culture under conditions that induce differentiation of the type II cell phenotype (Fig. 1). These observations suggest that the two antigens are co-expressed at some time point during differentiation of fetal alveolar epithelial cells, and that subsequently their expression later in differentiation is restricted to either of the two mature cell types. This process appears to be recapitulated during the repair process following lung injury.