Direct and indirect roles for β-catenin in facultative basal progenitor cell differentiation

Abstract

The conducting airway epithelium is maintained and repaired by endogenous progenitor cells. Dysregulated progenitor cell proliferation and differentiation is thought to contribute to epithelial dysplasia in chronic lung disease. Thus modification of progenitor cell function is an attractive therapeutic goal and one that would be facilitated by knowledge of the molecular pathways that regulate their behavior. We modeled the human tracheobronchial epithelium using primary mouse tracheal epithelial cell cultures that were differentiated by exposure to the air-liquid-interface (ALI). A basal cell subset, termed facultative basal cell progenitors (FBP), initiate these cultures and are the progenitor for tracheal-specific secretory cells, the Clara-like cell, and ciliated cells. To test the hypothesis that β-catenin is necessary for FBP function, ALI cultures were generated from mice homozygous for the Ctnb^flox(E2–6) allele. In this model, exons 2–6 of the β-catenin gene are flanked by LoxP sites, allowing conditional knockout of β-catenin. The β-catenin locus was modified through transduction with Adenovirus-5-encoding Cre recombinase. This approach generated a mosaic epithelium, comprised of β-catenin wild-type and β-catenin knockout cells. Dual immunostaining and quantitative histomorphometric analyses demonstrated that β-catenin played a direct role in FBP-to-ciliated cell differentiation and that it regulated cell-cell interactions that were necessary for FBP-to-Clara-like cell differentiation. β-catenin was also necessary for FBP proliferation and long-term FBP viability. We conclude that β-catenin is a critical determinant of FBP function and suggest that dysregulation of the β-catenin signaling pathway may contribute to disease pathology.

the human conducting airway epithelium is composed of three main cell types: basal, secretory, and ciliated (26). Each of these cell types is present at approximately equivalent frequency in the healthy airway. Cells that die are replaced through proliferation and differentiation of endogenous epithelial progenitor cells. These progenitor cells are subsets of the basal and secretory cell populations (40).

Chronic lung disease is often characterized by airway epithelial dysplasia (18, 26). Diseased airways exhibit changes in progenitor cell proliferation and are characterized by regions of basal cell hyperplasia, ciliated cell hypoplasia, and secretory cell metaplasia. These changes are indicative of an injury process and of altered progenitor cell function. Our long-term goal is to identify the signals that regulate progenitor cell behavior and to normalize epithelial cell-type frequency through manipulation of these pathways.

Human airway architecture including epithelial structure and cellular composition are recapitulated by the mouse trachea (7). Thus we have used mouse trachea to model the human airway epithelium. Normal tracheal basal cells express keratins (K) 5 (41) and 15 (7). Approximately 20% of normal basal cells coexpress K14 (7). Because K5 and K15 are panbasal cell markers, basal cells were defined in this study by immunostaining for K5 or K15. Tracheal ciliated cells are defined by motile cilia, which are detected by immunostaining for acetylated tubulin (ACT, Ref. 34). Tracheal secretory cells are termed Clara-like cells (48). These cells, like the bronchiolar Clara cell (35), express Clara cell secretory protein (CCSP; also known as Scgb1a1) (45) and Cytochrome P450–2F2 (Cyp2F2) (5). However, tracheal Clara-like cells express less CCSP and Cyp2F2 than bronchiolar Clara cells (7), have a distinct ultrastructure (29), and express high levels of the CCSP-related protein Scgb3A1 (36). In this study, Clara-like cells were detected by immunostaining for CCSP.

Lineage-tracing studies indicated that the basal cell progenitor produces different types of cells under normal and repairing conditions. We showed that the K14+ basal cell subset generates only basal cells in the normal epithelium (13). Others used the panbasal cell promoter, K5, to lineage trace basal cells and showed that the majority of these cells self-renewed and that they generated small numbers of Clara-like and ciliated cells (39). After naphthalene-mediated epithelial injury, all tracheal basal cells express K14 (7). Lineage tracing demonstrated that repairing K14+ cells were the progenitor for Clara-like and ciliated cells (13). Similarly, K5+ basal cells became progenitors for numerous Clara-like and ciliated cells after acid injury (39). Thus basal cells proliferate in response to injury and increase their differentiation potential when repairing the epithelium.

Respiratory progenitor cells have been termed “facultative progenitors” (49). This term was originally coined to explain the two functional states of these cells. In the normal epithelium, respiratory epithelial cells are long-lived, exist in a quiescent state, and perform differentiated functions that are critical to tissue maintenance and function. Following epithelial injury, basal, Clara-like, and Clara cells increase their proliferation rate (7, 9–12, 20, 34). Thus basal, Clara-like and Clara cells have the capacity to proliferate to replace dead cells. Herein, we take note of the fact that basal cell progenitors also change the frequency with which they generate ciliated and Clara-like cell progeny (13, 39). This phenotypic plasticity suggested that the facultative basal progenitor (FBP) cell behavior could be modified, leading to normalization of epithelial cell type frequency.

The molecular mechanisms that influence proliferation and differentiation of FBP were the focus of this study. We examined the role of β-catenin in FBP cell fate determination for several reasons. First, β-catenin-dependent gene expression is activated in basal cells following naphthalene injury (4). Second, potentiation of β-catenin-dependent gene expression in the developing mouse airway epithelium inhibited differentiation and resulted in tracheal polyposis (27). Third, conditional potentiation of β-catenin signaling in Clara cells blocked Clara cell maturation and ciliated cell differentiation in vivo (35). Finally, and perhaps paradoxically, targeted deletion of β-catenin in Clara-cells showed that β-catenin was not necessary for postnatal maturation of bronchiolar Clara cells or repair of naphthalene-injured bronchiolar airways (57). These data suggested that β-catenin regulated gene expression in a context- and cell-type-dependent fashion. The purpose of this study was to determine whether β-catenin was necessary for FBP cell-fate determination.

We and others have shown that air-liquid-interface (ALI) cultures of primary mouse tracheal epithelial cells can be used to evaluate the cellular and molecular mechanisms regulating basal cell function (4, 13, 41, 55, 56). K5-positive basal cells comprise 98% of the ALI culture on culture day 2 (4). These cells proliferate, form a polarized epithelium, and differentiate first to ciliated and then to Clara-like cells (Fig. 1) (4). The advantage of ALI cultures over in vivo analysis is that ALI cultures separate rapid proliferation from differentiation and resolve FBP-to-ciliated and FBP-to-Clara-like cell differentiation into distinct waves. Thus ALI cultures are an optimal model for analysis of signaling pathways that regulate FBP cell-fate determination.

Fig. 1.

Experimental design. A: facultative basal cell progenitor (FBP) proliferation and differentiation are presented as a function of time in air-liquid-interface (ALI) culture. The percentage of cells that participate in each process is presented. The rapid proliferation phase occurs on culture days 0–5 (dotted line). In subsequent text and figures, proliferation phase is referred to as P. ALI cultures polarize to ∼2,000 Ohms/cm² on P5 ± 1 day. Because the timing of polarization can vary, differentiation day 0 is defined as the day the cultures polarize, referred to as D. The differentiation phase occurs on D0.5–D9 (vertical line). Differentiation is divided into 2 subprocesses, generation of ciliated (dot-dash line) or Clara-like (dashed line) cells. Cultures evaluated in Figs. 3 and 5–9 were transduced with adenovirus 5 (Ad5) constructs on P5. Cultures evaluated in Fig. 4 were treated on P4 and were maintained in differentiation medium to P6. B–E: immunofluorescence analysis of ALI cultures that were transduced on P5 and fixed on D2. Scale bars = 50 μM. Shown are split-color images for DAPI, blue (B), Keratin 5 (K5), green (C), and β-catenin (Bcat), red (D); merged image in E. White arrows, β-catenin wild-type (WT) FBP; Yellow arrows: β-catenin knockout (KO) FBP.

This study used polarized ALI cultures to test the necessity of β-catenin for basal cell differentiation, survival, and proliferation. ALI cultures were generated from Ctnb^{floxE2-E6/floxE2-E6} (Ctnb^f/f) mouse trachea (3). To determine whether β-catenin was required for basal cell fate determination, ALI cultures were transduced with adenovirus 5 (Ad5) expressing Cre recombinase. The resultant cultures were a mosaic of β-catenin wild-type (WT) and knockout (KO) cells. Differentiation, apoptosis, and proliferation were compared in WT and KO cells as a function of time. We demonstrate that β-catenin played a direct role in FBP-to-ciliated cell differentiation and an indirect role in FBP-to-Clara-like cell differentiation. β-Catenin KO basal cells became apoptotic and were replaced by β-catenin-expressing cells over time. These findings demonstrated that β-catenin is necessary for FBP cell function.

MATERIALS AND METHODS

Animals.

Mice were housed and tissue recovered according to procedures approved by the National Jewish Health Animal Care and Use Committee. Primary tracheal epithelial cells were recovered from adult (6–8 wk of age) mice. Specific strains were BATGal (28), Ctnb^{floxE2–E6/floxE2–E6} (Ctnb^f/f) (3), ROSA26^floxStopLacZ (RS) (46), and CCSP-Cre/RS (25, 44).

ALI cell cultures.

Cultures of primary tracheal cell isolates were produced according to previously published methods (4, 13, 56). Briefly, whole mouse tracheas from age-matched individuals of both sexes were pooled and digested with 0.15% Pronase (Roche; 11459643001) in Ham's F-12 (Cellgro; 10–080-CV) overnight at 4°C. Tracheal cells were plated on collagen-coated 24-well Transwell inserts. The day that cells were plated was termed Proliferation Day (P) 0 (Fig. 1). Keratin 5 immunopositive cells constituted >98% of cultures on P0–P5 (Fig. 1, B–E) (4). The cultures polarized (e.g., reached a transepithelial resistance of ∼660–2,300 Ohms × cm²) within 4–6 days. The culture period preceding polarization is termed the rapid proliferation phase. Following polarization, growth factors were removed and cells are cultured in 2% NuSerum (BD; 355100). This culture period is termed the differentiation (D) phase (Fig. 1). During this phase the mitotic index was low relative to that observed during the rapid proliferation phase (4).

Viral transduction of tracheal epithelial cell cultures.

ALI cultures were transduced on P5 (Fig. 1), unless otherwise noted. Ad5-green fluorescent protein (GFP) (no. 1060; Vector Biolaboratories) was used as a control and Ad5-Cre recombinase-IRES-GFP (no. 1700; Ad5-Cre-GFP, Vector Biolaboratories) was used to induce recombination. Optimal multiplicity of infection and timing of transduction were determined in RS cultures and confirmed in Ctnb^f/f cultures. The optimal multiplicity of infection was 400.

Transduction efficiency was determined by viral GFP expression in live cells. Recombination efficiency was determined by X-gal staining of Ad5-GFP or Ad5-Cre-GFP transduced RS cell ALI cultures. GFP expression indicated that nearly 100% of cells were transduced, yet X-gal staining and β-catenin depletion were evident in no more than 50% of the cells at any given time point. This discrepancy could be attributable to differences in expression levels or relative stability of GFP and Cre recombinase in transduced cells. Importantly, viral GFP fluorescence was destroyed by fixation and therefore did not interfere with immunofluorescence imagining.

BrdU labeling.

Cultures were pulsed 10 μM bromodeoxyuridine (BrdU) (Sigma), 16 h before harvest, and then fixed in 3.2% paraformaldehyde (Sigma)/3.0% sucrose/PBS for 20 min at 4°C.

Immunostaining and imaging.

Fixed membranes were immunostained for BrdU, K5, K15, ACT CCSP, or β-catenin with validated antibodies using previously published methods (4, 7, 13). Apoptotic cells were detected by immunofluorescence using a rabbit α-activated (cleaved) caspase 3 (CC3) at 1:300 (Abcam no. 2302). Staurosporine-treated ALI cultures served as positive controls for apoptotic cells. The β-galactosidase (β-gal) reporter was detected using a guinea pig α-β-gal at 1:8,000 (MP Biomedical). Images were captured on a Zeiss Axiovert Imager-Z1 or a Zeiss LSM 700 confocal microscope.

Morphometry.

The data were collected according to stereological principles (21, 54). A critical component of the stereological technique is unbiased sample selection, which is typically referred to as “random” field selection. The details of this method as applied to analysis of ALI membranes follows. The membrane was positioned within the illuminated area of the microscope stage. The first field to be photographed was visualized in the DAPI channel, and the first image was taken in this channel. Separate images of the same field were taken for each fluorophore present on the immunostained cells. Most images presented in this manuscript were comprised of three individual photographs: one taken in the DAPI channel, one taken in the green channel, and one taken in the red channel. Following image acquisition, the camera shutter was closed and the membrane was moved to another region. Steps 1–4 were repeated until at least six photographs of each ALI membrane were taken. The ×400 images included 3.73 × 10⁴ μm² per image. Images were randomly divided into six counting areas (2,243 μm² per area) using a preprinted template. Typical cell numbers per counting varied as a function of time and ranged from 600 to 1,200 cells.

For analysis of early ciliated cell differentiation (Fig. 4), images were centered on a ciliated cell and captured. The frequency of β-catenin WT and β-catenin KO cells was determined as indicated above. This method allowed analysis of ciliated cells during the period in which they are rare constituents of the ALI culture.

Fig. 4.

β-Catenin is necessary for initiation of FBP-to-ciliated cell differentiation. ALI cultures were generated from WT or Ctnb^f/f mice. Ctnb^f/f cultures were treated on P4 and maintained in proliferation medium through P6. A–D: histological analysis of β-catenin and ciliated cell differentiation on D1 and D3. A and C: D1. B and D: D3. A and B: dual-color images illustrating the frequency of β-catenin KO cells (yellow arrows). C and D: triple-color images of the same fields presented in A and B. White arrows indicate β-catenin KO ciliated cells. ACT, green; β-catenin, red; DAPI, blue. Scale bars = 50 μm. E–G: quantification of ciliated cell differentiation. Bars indicate the mean and SD for each data set. E: kinetics of ciliated cell differentiation in WT cultures on P4-–P6 and D1–D4. n = 24 per time point. F: kinetics of recombination in Ad5-Cre-green fluorescent protein (GFP)-treated Ctnb^f/f cultures. n = 24 per time point. G: distribution of ciliated cells to the β-catenin WT and KO subsets in Ctnb^f/f cultures on D1–D4. n = 10 per time point and genotype. Plus sign indicates β-catenin WT (black); minus sign indicates β-catenin KO (red).

For cell-type frequency determinations, the number of DAPI-positive nuclei per area was set as the denominator. The number of cells expressing the marker(s) of interest was set as the numerator. For determination of the cellular mitotic index, the number of cells expressing the marker of interest was set as the denominator, and the number of cells that were dual-positive for the marker and BrdU was set as the numerator. The resulting frequencies were presented as a percentage for each counting area. Thus the sample sizes (n) noted in the figure legends refer to the number of areas, not the number of cells, counted for the marker(s) of interest.

Interexperimental consistency.

To determine whether Clara and/or ciliated cell differentiation varied significantly among experiments, we performed statistical analyses in consultation with the National Jewish Health Department of Biostatistics and Bioinformatics. Data for identically treated cultures from three different experiments were used to compare ciliated cell differentiation or Clara-like cell differentiation. Inspection of the raw data revealed that there were zero values in some data sets. Zero values resulted when none of the cells within a counting area expressed the phenotypic marker of interest. Therefore, zero values were legitimate data points. The data sets also had different numbers of observations. Because the data sets contained zero values and had unequal numbers of observations, a nonparametric statistical analysis was required (29a).

The Kruskal-Wallis test, a nonparametric version of ANOVA, is commonly used to compare three or more groups. This test ranks raw data by value and analyzes differences in the rank values, adjusting for ties that occur when the raw data in each group contain identical values. Kruskal-Wallis tests of the data for ciliated and Clara cell differentiation resulted in P values >0.05. Thus we concluded that ciliated and Clara-like cell differentiation was consistent among experiments.

Statistical analysis.

The significance of differences between data sets within experiments was determined by one-way ANOVA (Tukey posttest) or two-tailed Student's t-test where appropriate. Differences were considered to be statistically significant when P < 0.05.

RESULTS

Activation of β-catenin-dependent gene expression in vitro.

To determine whether β-catenin-dependent genes are active in nascent ciliated and Clara cells, ALI cultures were generated from mice harboring the β-catenin-dependent reporter transgene BATGal (28). In this model, β-gal is expressed under regulation of a multimerized T cell factor consensus binding site. β-Gal is directed to the nucleus by a nuclear localization sequence facilitating codetection of the reporter and cell type-specific markers.

BATGal transgene expression was evaluated on D3 and D8, when ciliated and Clara-like cells are detected (Fig. 1) (4, 13). At each time point, β-gal+ cells were a subset of cells within the culture (Fig. 2). On D3, β-gal+ cells were clustered and formed a homogenous group of transgene positive cells (Fig. 2, A and B). This clustering suggested focal activation of β-catenin-dependent gene expression. Cells within the β-gal+ clusters stained for the ciliated cell marker ACT (Fig. 2A). Interestingly, numerous β-gal-ciliated cells were also noted. These data suggested that FBP-to-ciliated cell differentiation involved transient expression of β-catenin-dependent genes in FBP and ciliated cells. On D8, CCSP+ cells were adjacent to β-gal+ cells (Fig. 2, C and D). The CCSP+ cells frequently encircled a β-gal+ cell. However, CCSP+ cells never expressed the β-catenin reporter transgene. These data suggested distinct roles for β-catenin in ciliated cell and Clara-like cell differentiation. K5+ FBP expressed β-gal at both time points (Fig. 2, E and F).

Fig. 2.

β-Catenin-dependent gene expression is activated in vitro. ALI cultures were generated from mice harboring the BATGal transgene. This β-catenin reporter expresses a nuclear-localized β-galactosidase (β-gal). A–F: histological analysis of β-catenin-dependent gene expression. Scale bars = 50 μm. A and B: D3. C and D: D8. β-gal, red; acetylated tubulin (ACT) and Clara cell secretory protein (CCSP), green; DAPI, blue. Arrows indicate the following: β-gal+/ACT+ cell (A); β-gal+ cell (B); β-gal-/CCSP+ cells (C and D). E and F: D1. β-gal, red; K5, green; DAPI, blue. Arrows indicate the following: β-gal+/K5+ cells (E); β-gal+ cells (F).

β-Catenin is required for FBP-to-ciliated cell differentiation.

To test the hypothesis that FBP-to-ciliated cell differentiation is regulated by β-catenin, ALI cultures derived from Ctnb^f/f mice were mock transduced or transduced with Ad5-Cre-GFP (Cre+) on P5. We first determined the kinetics of FBP-to-ciliated cell differentiation in mock-transduced cultures. Differentiation was evaluated by dual-immunofluorescence analysis of ACT and β-catenin on D2, D4, D6, and D9 (Fig. 3, A–D). The frequency of ciliated cells increased from 0.14% on D2, to 26% on D4 (Fig. 3G). On D6 and D9, ciliated cell frequency remained constant at 28% and 25% (Fig. 3G).

Fig. 3.

β-Catenin is not necessary for late FBP-to-ciliated cell differentiation. ALI cultures were generated from WT or Ctnb^f/f mice and treated on P5. A–D: histological analysis of ciliated cell differentiation in WT cultures on D2 (A), D4 (B), D6 (C), and D9 (D). ACT, green; DAPI, blue. Scale bars = 50 μm. E and F: histological analysis of ciliated cell differentiation in Ctnb^f/f cultures that were transduced with Ad5-Cre on P5 and analyzed on D6. ACT, green; β-catenin, red; DAPI, blue. E: wide-field micrograph, scale bar = 50 μm. F: confocal micrograph, scale bar = 10 μm. Arrows in F: yellow, β-catenin WT ciliated cell; white, β-catenin KO cell. G–I: quantification of ciliated cell differentiation. Bars indicate the mean and SD for each data set. G: kinetics of ciliated cell differentiation in WT cultures on D2–D9. n = 24 per time point. H: effect of treatment on ciliated cell differentiation in Ctnb^f/f cultures. Analysis on D6 and D9. Mock-transduced cultures were treated with saline (○); Cre+ cultures were treated with Ad5-Cre- (▼). n = 13 per treatment for D6; n = 24 per treatment for D9. I: effect of genotype on ciliated cell frequency in Ctnb^f/f cultures. Analysis on D6–D9. Plus sign indicates β-catenin WT (●); minus sign indicates β-catenin KO (■). n = 18 per genotype on days 6-8; n = 15 per genotype on day 9.

To determine whether viral transduction influenced FBP-to-ciliated cell differentiation, the frequency of ciliated cells in mock and Cre+ cultures was compared on days 6 and 9 (Fig. 3H). The frequency of ciliated cells did not vary according to treatment on either day.

Ad5-Cre transduction resulted in a mosaic epithelia composed of β-catenin WT and KO cells. To determine whether the frequency of ciliated cells varied according to β-catenin genotype, the ciliated cell differentiation within the β-catenin WT and KO cellular subsets was evaluated on D6–D9 (Fig. 3I). The frequency of ciliated cells did not vary according to β-catenin genotype at any time point.

Careful examination of the data in Fig. 3I revealed two types of β-catenin KO regions. One type was devoid of ciliated cells (zero values in Fig. 3I), whereas the second type contained numerous ciliated cells. Furthermore, Fig. 3G suggested that a rare cohort of FBP initiated the differentiation process before day 2. We reasoned that these cells may have initiated the FBP-to-ciliated cell differentiation process before recombination of the Ctnb^f/f allele and generation of the β-catenin KO phenotype.

To test this hypothesis, the kinetics of FBP-to-ciliated cell differentiation was evaluated on P4 through D4 (Fig. 4, A–E). Small numbers of ciliated cells were first detected on P6 (Fig. 4E). The frequency of ciliated cells increased slowly through D2 and increased dramatically between D2 and D4. These data indicated that ciliated cell differentiation could be divided into two phases: an early phase, during which ciliogenesis proceeds slowly, and a late phase, during which ciliogenesis proceeds rapidly. These data agree with previously published kinetics of ciliated cell differentiation in vitro (22, 55, 56).

To determine whether the early ciliated cell differentiation overlapped with Cre-mediated generation of β-catenin KO cells, cultures were transduced on P4 and maintained in proliferation medium for 2 days after polarization to 660 Ohms/cm². Cultures were fixed daily, beginning 12 h posttransduction. β-Catenin expression was evaluated by immunofluorescence staining. β-Catenin KO cells were first detected 3 days after transduction, on D1 (Fig. 4F). The frequency of β-catenin KO cells then increased linearly between D2 and D4 (Fig. 4F). These data indicate that transduction on P5 (as presented in Fig. 3) resulted in overlap of Cre-mediated recombination and ciliated cell differentiation. Consequently, the analysis of ciliated cell differentiation presented in Fig. 3 captured ciliated cells that differentiated before as well as after establishment of the β-catenin KO phenotype.

To quantify the effect of β-catenin deficiency on early ciliated cell differentiation, the genotype of ciliated cells was determined in cultures that were transduced on P4 and evaluated on D1–D3 (Fig. 4G). During the D1–D2 interval, all ciliated cells were β-catenin WT even though the frequency of β-catenin KO cells increased from 10% to 30% (Fig. 4F). Thus sufficient recombination had occurred to detect β-catenin KO ciliated cells if they were generated. These data indicated that the earliest phase of ciliated cell differentiation required β-catenin.

As previously shown for cultures transduced on P5 (Fig. 3), cultures transduced on P4 and held in proliferation medium to P6 exhibited a rare population of β-catenin KO ciliated cells on D3–D4. On D3, 96 ± 3% of ciliated cells were β-catenin WT. On D4, 76 ± 9% of ciliated cells were β-catenin WT. ANOVA analysis did not detect statistically significant change in the frequency of β-catenin KO ciliated cells over the D1–D4 interval. On the basis of the experiments presented in Figs. 3 and 4, we conclude that β-catenin is required for FBP to initiate ciliated cell differentiation. However, β-catenin is not required to complete the ciliation process or to maintain the ciliated cell phenotype.

FBP-to-Clara-like differentiation in vitro.

The kinetics of Clara-like cell differentiation in mock-transduced cultures derived from Ctnb^f/f mice were determined on D2, D4, D6, and D9 by immunofluorescence staining for CCSP (Fig. 5, A–F). The frequency of Clara-like cells increased from 0% on D2 and D4, to 12% on D6, and 25% on D9 (Fig. 5G). A dense time course analysis, focused on D6–D9, detected a trend toward increased numbers of Clara-like cells between D6 and D7, and constant frequency between D7 and D9 (Fig. 5H). Thus the role of β-catenin in FBP-to-Clara-like differentiation was evaluated on D6–D9.

Fig. 5.

The effect of β-catenin on FBP-to-Clara-like cell differentiation. ALI cultures were generated from WT or Ctnb^f/f mice and treated on P5. A–F: histological analysis of Clara-like cell differentiation. A–D: kinetics of Clara-like cell frequency in WT cultures on D2 (A), D4 (B), D6 (C) and D9 (D). ×400 (E) and ×630 (F) confocal images showing the distribution of CCSP+ cells to the β-catenin WT and KO subsets in Ctnb^f/f cultures on D6. CCSP, green; β-catenin, red; DAPI, blue; scale bars = 50 μm. Arrows: yellow, CCSP+/β-catenin+ cells; white, β-catenin KO cell. Asterisks in F, CCSP+ β-catenin KO edge cell. See K for nomenclature. G–J: quantification of Clara-like cell differentiation. Bars indicate the mean and SD for each data set. G: kinetics of Clara-like cell differentiation in WT cultures on D2–D9. n = 27 areas/time point. H: dense time course analysis of Clara-like cell differentiation on D6–D9. n = 12 areas/time point. I: effect of treatment on Clara-like cell differentiation. Mock-transduced cultures were treated with saline (○); Cre+ cultures were treated with Ad5-Cre-GFP (▼). n = 16 for each treatment group. J: effect of genotype on the frequency of Clara-like cells on D6–D9. Plus sign indicates β-catenin WT (●); minus sign indicates β-catenin KO (■). n = 12 per genotype and day. K: types of cells found in mosaic cultures: β-catenin WT (blue) and β-catenin KO cells (pink) composed the epithelium. Within each genotype a subset of cells contacted another cell of the same genotype (cluster cell) or a cell of the opposite genotype (edge cell). CCSP+ Clara like cells (C) were identified in WT cluster and edge cells. In contrast, Clara-like cells were never identified in KO cluster cells but were identified in KO edge cells. L: frequency of CCSP+ cells within various cell subtypes. Bars indicate the mean and SD for each data set. Horizontal lines indicate a P value of 0.0001.

FBP-to-Clara-like differentiation is influenced by β-catenin.

To test the hypothesis that FBP-to-Clara-like cell differentiation is regulated by β-catenin, ALI cultures derived from Ctnb^f/f mice were transduced on P5. Treatment- and genotype-dependent effects were evaluated on D6–D9. Clara-like cell differentiation was evaluated by dual immunofluorescence staining for CCSP. The frequency of Clara-like cells in mock cultures was greater than in Cre+ cultures on day 6 (P = 0.0023; Fig. 5I). This treatment effect suggested that β-catenin was necessary for FBP-to-Clara-like differentiation.

To evaluate a genotype-dependent difference in FBP-to-Clara-like cell differentiation, the frequency of CCSP+ cells in the β-catenin WT and KO cell subsets was determined. The frequency of β-catenin WT Clara-like cells increased between days 6 and 9 (Fig. 5J). In contrast, β-catenin KO cells that were surrounded by other β-catenin KO cells did not generate Clara-like cells at any time point (Fig. 5J).

Careful inspection of β-catenin KO cells that were adjacent to a β-catenin WT cell showed that a subset of these cells was able to differentiate into a CCSP+ cell (Fig. 5F). Quantification of these “β-catenin KO edge cells” showed that 7.3 ± 2.7% of these cells were CCSP+. The frequency of CCSP+:β-catenin KO edge cells was greater than the frequency of CCSP+ cells in the entire β-catenin KO cell population (P = 0.0001, Fig. 5L). Although β-catenin KO edge cells could generate a Clara-like cell, the frequency of this process was decreased approximately threefold and was significantly different from Clara-like cell differentiation within the β-catenin WT cell population (P = 0.0001 for comparison to WT cluster and WT edge cells).

To determine whether the β-catenin KO by β-catenin WT interaction had an impact on β-catenin WT differentiation to Clara-like cells, we compared the frequency of CCSP+:β-catenin WT cells that were surrounded by β-catenin WT cells (WT cluster cells) with that of WT cells that were adjacent to a KO cell (WT edge cells). The frequency of CCSP+ cells was similar in the two subsets of WT cells (P = 0.086). Collectively these data sets indicated that β-catenin was necessary for FBP-to-Clara-like differentiation, that the β-catenin-dependent mechanism signaled via direct cell contacts, and that the proposed cell-cell interaction required β-catenin on the nondifferentiating cell.

Fate of clustered β-catenin KO cells.

The above data suggested that β-catenin KO cells were divided into two functional subsets, edge cells and cell clusters. Although, a β-catenin WT-to-KO-cell interaction explained detection of CCSP+ within the β-catenin KO edge cell population, we considered two alternative explanations for the lack of Clara-like cell differentiation in KO cell clusters. First, it was possible that a β-catenin KO FBP was able to generate a Clara-like cell but the β-catenin KO Clara-like cell then died. Thus we determined whether β-catenin was necessary for maintenance of a differentiated Clara-like cell. Second, it was possible that KO-FBP underwent apoptosis before initiation of Clara-like cell differentiation. Because this putative apoptotic event could be a consequence of β-catenin deficiency, transduction, or recombination, we tested each of these possibilities separately.

β-Catenin is not necessary for maintenance of Clara-like cells.

To determine whether β-catenin was necessary for persistence of differentiated Clara-like cells, ALI cultures were generated from mice that were homozygous for the Ctnb^f/f allele and harbored the CCSP-Cre transgene (25, 44). This model allows recombination of the β-catenin locus after CCSP is expressed (37). Ad5-Cre transduction was not used in this experiment because the EGTA treatment that facilitates transduction of polarized mouse tracheal ALI cultures (data not shown) (8, 53) was toxic. On days 6 (Fig. 6, A–C) and 9 (Fig. 6D) the frequency of CCSP+ cells was indistinguishable in the β-catenin WT and KO cell subsets. These data indicated that β-catenin was not necessary for maintenance of the Clara-like cell phenotype. Thus the failure of FBP-to-Clara-like cell differentiation in clustered β-catenin KO cells was not due to death of the Clara-like cell.

Fig. 6.

β-Catenin is not necessary for maintenance of Clara-like cells. ALI cultures were generated from mice harboring the CCSP-Cre transgene and the Ctnb^f/f allele. A and B: histological analysis of CCSP (green) and β-catenin (red) on D6. A: wide field, scale bar = 50 μm. B: confocal, scale bar = 10 μm. White arrows denote β-catenin KO cells. C and D: quantification of Clara-like cell frequency in the β-catenin WT (+, ●) and β-catenin KO (−, ■) cellular subsets. Bars indicate the mean and SD for each data set. C: D6. D: D9. n = 18 per genotype and day.

β-Catenin is necessary for FBP cell survival.

To approach the question of apoptosis, we reasoned that attrition of β-catenin KO cells would result in enrichment of β-catenin WT cell population. Thus we first determined whether the frequency of β-catenin WT cells varied as a function of time in mosaic cultures (Fig. 7, A–E). This study demonstrated that β-catenin WT cell frequency was greater on day 9 than on days 6 (P = 0.0046) or 7 (P = 0.0017) and that the frequency of β-catenin WT cells stabilized between days 8 and 9. These data suggested that about 25% of β-catenin KO cells were lost from the cultures on days 6 and 7.

Fig. 7.

β-Catenin KO cells are lost as a function of time. ALI cultures were generated from Ctnb^f/f mice and treated with Ad5-Cre-GFP on P5. A–D: histological analysis of β-catenin WT and KO cells on D6–D9. β-catenin, red; DAPI, blue. Scale bars = 50 μm. Arrows indicate β-catenin KO cells. E: quantification of β-catenin WT cell frequency on D6–D9. Bars indicate the mean and SD for each data set. n = 30 for days 6-8; n = 18 for day 9.

To determine whether the decrease in β-catenin KO cells could be an effect of Ad5-Cre transduction, mock and Cre+ cultures were immunostained with an antibody directed against a marker of apoptosis, CC3 (Fig. 8, A–F) (23, 24). The frequency of CC3+ cells was determined on D6–D9 (Fig. 8G). Apoptotic cells comprised a very small fraction of the cell population (0–3.3%), yet Cre+ cultures had significantly more apoptotic cells than mock transduced cultures on days 6 (P = 0.0067) and 7 (P = 0.0261). Apoptotic cells were not detected on days 8 and 9. These results paralleled enrichment of β-catenin WT cells presented in Fig. 7E.

Fig. 8.

A subset of β-catenin KO FBP cells are apoptotic. ALI cultures were generated from Ctnb^f/f mice and transduced with Ad5-Cre-GFP on P5. A–F: histological analysis of apoptosis as a function of time. Apoptotic cells were detected by staining for cleaved caspase 3 (CC3) (red). Scale bars = 50 μm. A: mock-treated culture on D6. B: Ad5-Cre-GFP-transduced culture on D6. C: analysis of apoptosis in ACT+ cells (green). D: analysis of apoptosis in CCSP+ cells (green). E and F: analysis of apoptosis in K15+ cells using wide-field imaging (E) or confocal imaging (F). Arrows: red, cleaved caspase 3+ cells; yellow, cleaved caspase 3+/K15+ cells. G–J: quantification of apoptosis as a function of time and treatment. Bars indicate the mean and SD for each data set. G: effect of treatment on the frequency of apoptotic cells. Mock-transduced cultures were treated with PBS (○); Cre+ cultures were treated with Ad5-Cre-GFP (▼). n = 30 per treatment and time point. H: effect of viral exposure on the frequency of apoptotic cells on D6. Cultures were treated with PBS, transduced with Ad5-GFP, or transduced with Ad5-Cre-GFP. n = 12 per treatment group. I: effect of genotype on apoptosis on D6–D9. Plus sign indicates β-catenin WT (●); minus sign indicates β-catenin KO (■). n = 30 per genotype. J: phenotype of apoptotic cells. n = 12 per marker.

To determine whether apoptosis on D6 was an unintended effect of viral transduction or Cre-mediated recombination, ALI cultures from RS (46) mice were transduced with Ad5-GFP, which encodes GFP but not Cre recombinase. A second set of RS cultures was transduced with Ad5-Cre-GFP. Cre-mediated recombination in RS cells results in expression of the LacZ gene, which can be detected with anti-β-gal antibody. Ad5-GFP and Ad5-Cre-GFP cultures were immunostained for GFP to detect productive infection, for β-gal to detect recombination, and for CC3 to evaluate apoptosis. On D6, all three test groups contained ∼4% CC3+ cells (Fig. 8H). The frequency of apoptotic cells in the three groups was indistinguishable (Fig. 8H). These data demonstrated that neither transduction nor recombination resulted in apoptosis.

To determine whether β-catenin deficiency was associated with increased apoptosis on days 6 and 7, Cre+ cultures were dual-immunostained for β-catenin and CC3 on D6 (Fig. 8I). Apoptotic cells were significantly more frequent among β-catenin KO cells than in the β-catenin WT cell population. These data demonstrated that a subset of β-catenin KO cells initiated an apoptosis program on D6 and potentially on D7.

To determine which β-catenin KO cell type(s) underwent apoptosis, Cre+ cultures were triple-immunostained for β-catenin, CC3, and differentiated cell markers on days 6–9. In this experiment, K15 (7) was used as a basal cell marker because antibodies against CC3 and the previously used FBP marker (K5) were generated in the same host. No apoptotic Clara-like cells or ciliated cells were observed at any time point in Cre+ cultures (Fig. 8J). The CC3+ cells were distributed equally between K15-positive FBP cells and cells that did not express any of the differentiation markers tested (Fig. 8J). These data suggest that a subset of β-catenin KO FBP were predisposed to apoptosis on D6 and likely on D7.

Because the apoptotic period overlapped with the initiation of Clara-like cell differentiation (Figs. 1 and 5A), we could not determine whether the absence of β-catenin KO Clara-like cells on days 6 and 7 was due to failure of FBP-to-Clara-like cell differentiation or FBP cell death. However, because apoptotic cells were not detected on days 8 and 9, time points when Clara-like cell differentiation still occurred within the WT cell population (Figs. 1 and 5A) and within the β-catenin KO edge cell population, we concluded that that β-catenin was necessary for FBP-to-Clara-like cell differentiation.

β-Catenin is necessary for FBP proliferation.

To determine whether transduction with Ad5-Cre affected proliferation, ALI cultures were pulse-labeled with BrdU and the mitotic index determined by staining for BrdU and differentiation markers (Fig. 9, A–F). Both mock and Cre+ cultures showed low levels of proliferation that decreased between D7 and D9 (Fig. 9G). However, mock cultures had a significantly higher mitotic index than Cre+ cultures on D6 (P = 0.0072; Fig. 9G). These data suggested that β-catenin deficiency decreased the proliferative capacity of Cre+ cultures. To assess the necessity of β-catenin for proliferation, Cre+ cultures were dual immunostained for BrdU and β-catenin (Fig. 9, A–C). Only β-catenin WT cells were BrdU+; no mitotic β-catenin KO cells were detected at any time point (Fig. 9H). As presented above, we were unable to distinguish the proapoptosis and antiproliferation effects of β-catenin on days 6 and 7. However, these data indicated that β-catenin was necessary for proliferation of surviving β-catenin KO cells on D8 and D9.

Fig. 9.

β-catenin is necessary for FBP proliferation. ALI cultures were generated from Ctnb^f/f mice, treated on P5, and bromodeoxyuridine (BrdU)-labeled 16 h before harvest. A–F: histological analysis of proliferation on D7 (A and D), D8 (B and E), and D9 (C and F). Scale bars = 50 μm. A–C: β-catenin (red), BrdU (pink), DAPI (blue). D–F: BrdU (red), ACT, CCSP, K5 (green), DAPI (blue). G and H: quantification of proliferation on D6–D9. Bars indicate the mean and SD for each data set. G: effect of treatment on proliferation. Mock-transduced cultures were treated with saline (○); Cre+ cultures were treated with Ad5-Cre-GFP (▼). n = 30 per treatment for days 6–8; n = 18 per treatment for day 9. H: effect of genotype on proliferation. Plus sign indicates β-catenin WT (●); minus sign indicates β-catenin KO (■). n = 30 per genotype on days 6–8; n = 18 per genotype on day 9. I–K: quantification of cellular mitotic index on D6. The frequency of each cell type, frequency of mitotic cells, and the cellular mitotic index are presented. Bars indicate mean and SD for each data seta. I: ACT+ cells. n = 24 for each data set. J: CCSP+. n = 36 for each data set. K: K5+ cells. n = 36 for each data set.

To determine the phenotype(s) of mitotic cells, cultures were pulse-labeled with BrdU 16 h before harvest, fixed, and dual-immunostained for BrdU and cell type-specific markers (Fig. 9, D–F). The frequency of cells that were dual-positive for a differentiation marker and for BrdU was expressed as a fraction of the total number of cells that were positive for the differentiation marker (Fig. 9, I–K). No BrdU-positive Clara-like or ciliated cells were observed. These data indicated that FBP-derived Clara-like and ciliated cells were postmitotic. The proliferative cell type in all cases was the K5-positive basal cell. These data indicated that β-catenin KO FBP could not divide. Consequently, we suggest that the apoptotic β-catenin KO cells were not replaced, resulting in depletion of the KO cell population on D6 and D7.

DISCUSSION

Summary.

We demonstrate that FBP require β-catenin to 1) differentiate into ciliated and Clara-like cells, 2) survive, and 3) proliferate in vitro.

β-Catenin-dependent gene expression.

We previously evaluated expression of the TOPGal transgene, a β-catenin reporter, in mice treated with NA (4). The transgene was expressed in FBP on recovery days 3–9 at a time when ciliated cells were being replaced through FBP-to-ciliated cell differentiation (7, 13). The timing of TOPGal transgene activity in vivo suggested a role for β-catenin-dependent genes in ciliated cell differentiation. However, variation in the kinetics of injury and repair in vivo prevented definitive analysis of cause and consequence.

Using ALI cultures, we were able to separate FBP proliferation and differentiation and to subdivide FBP differentiation into two waves, generation of ciliated and Clara-like cells (4, 13) (Figs. 1, 3, and 5). We previously showed that the TOPgal transgene was active in ALI cultures but we were unable to correlate activity patterns with specific FBP functions. In contrast with the TOPGal transgene, the BATGal transgene is more sensitive to signals that regulate β-catenin-activity (1). Our analysis of BATGal transgene activity in ALI cultures (Fig. 2) suggested that β-catenin-dependent genes played a direct role in FBP-to-ciliated cell differentiation and an indirect role in FBP-to-Clara-like differentiation.

β-Catenin regulates FBP-to-ciliated cell differentiation.

Traditional and genetic approaches indicate that the ciliated cell is long-lived, nonmitotic, and terminally differentiated (32, 33). As a consequence of this terminal differentiation state, damaged ciliated cells must be replaced by a progenitor cell. These progenitors include tracheal basal cells and their subtypes (13, 19, 20, 31, 39–41). We now show that FBP-to-ciliated cell differentiation occurs in two phases in vitro (Figs. 3–4). The early phase begins between P5 and P6, and extends to D1. During this stage, ciliated cell frequency increases in a linear fashion (Fig. 4E). The late ciliation phase begins on D2 and results in an exponential increase in ciliated cell frequency (Fig. 4E). Once differentiated, ciliated cells are neither apoptotic (Fig. 8J) nor mitotic (Fig. 9I), and ciliated cell frequency is constant. These data suggest that the FBP-to-ciliated cell differentiation process is negatively regulated.

By dissecting the FBP-to-ciliated cell differentiation into early and late phases, we showed that β-catenin is necessary for early ciliated cell differentiation (Fig. 4G) but not late phase differentiation (Fig. 3I). The necessity of β-catenin for early FBP-to-ciliated cell differentiation (Fig. 4) and differences in the pattern of BATGal reporter transgene activity, clustered vs. dispersed (Fig. 2, A and D), indicated that the signals that coordinate initiation of FBP-to-ciliated cell differentiation act through β-catenin. Our previous study showed that genetic stabilization of β-catenin (16) increased ciliated cell differentiation twofold in vitro (4). Together these two studies indicate that loss of the β-catenin-activating signal limits FBP-to-ciliated cell differentiation.

The Brody group demonstrated an association between the frequency of cells containing a primary cilium and those that later formed motile cilia (22). Primary cilia sequester receptors and adaptor proteins that are necessary for sonic hedgehog signaling (15, 47). Thus the sonic hedgehog pathway is an appealing candidate for a transient signal that activates β-catenin-dependent gene expression in FBP and initiates generation of ciliated cells. Investigation of this mechanism is beyond the scope of the present study.

β-Catenin is necessary for FBP survival and proliferation.

In ALI cultures, a subset of FBP survive and proliferate infrequently to maintain the monolayer (Figs. 8 and 9). Analysis of β-catenin status in mosaic cultures showed that β-catenin deficiency compromises the capability of FBP cells to survive within a differentiated epithelium (Fig. 8) and to proliferate for maintenance of this epithelium (Fig. 9). Because clusters of β-catenin KO cells were noted at all time points (Figs. 3–9) and most analysis regions lacked an apoptotic cell (Fig. 8I), it was unlikely that β-catenin KO cells died as a result of defective adherens junctions. Several studies identified plakoglobin as a surrogate for β-catenin in adherens junction formation (2, 6, 30, 43). We suggest that a change in plakoglobin distribution complemented the β-catenin deficiency and allowed formation of functional cell-cell junctions.

β-Catenin is necessary for FBP-to-Clara-like differentiation.

In contrast with the ciliated cell, the tracheal Clara-like cell functions as a self-renewing progenitor in the steady-state epithelium (9, 11, 12). After NA and other types of chemical injury (4, 7, 14, 38, 40) the Clara-like cell is replaced by basal cells and their subtypes. We now show that β-catenin-dependent effects on FBP-to-Clara-like cell differentiation must be defined according to the context: the behavior of a β-catenin KO FBP is dependent on whether its neighbors are β-catenin KO or WT (Fig. 5K).

Clustered β-catenin KO FBP undergo more frequent apoptosis than β-catenin WT FBP on D6 and potentially on D7 (Fig. 8). These data suggested that the earliest cohort of β-catenin KO FBP died before they could differentiate into Clara-like cells and that FBP that that did not enter a differentiation pathway required β-catenin transcriptional activity to survive. However, the β-catenin KO cell population stabilized on D8 and D9, and these β-catenin KO cells were nonapoptotic (Fig. 8G). These data suggested that clustered FBP could survive in the absence of β-catenin. It is unknown whether these clustered FBP are senescent. Because of the complex effect of β-catenin on clustered FBP survival, we were not able to determine whether β-catenin played a role in Clara-like cell differentiation by clustered β-catenin KO FBP.

In contrast, we detected a positive effect of β-catenin on differentiation of β-catenin KO cells that were adjacent to a β-catenin WT cell. These so-called edge cells were able to differentiate into a CCSP+ Clara cell and were the only β-catenin KO cells that underwent this process (Fig. 5L). These data in combination with demonstration that CCSP+ cells did not express the β-catenin reporter transgene BATGal, suggesting that FBP-to-Clara-like cell differentiation required activation of β-catenin-dependent genes in an adjacent cell.

The present study did not determine the mechanisms regulating β-catenin-dependent gene expression. However, it is likely that FBP-to-Clara-like cell differentiation is regulated by an interaction between a cell that is expressing β-catenin-dependent genes (inducer cell) and a cell that is not expressing β-catenin-dependent genes (responder cell). We previously reported that constitutive activation of β-catenin blocked FBP-to-Clara-like differentiation in vitro (4). Thus we propose that the β-catenin-dependent differentiation of FBP into Clara-like cells is regulated by a transient β-catenin-dependent signal that mediates direct cell-cell interactions or short-range paracrine signals.

The Notch pathway was identified as the mediator of cell-cell interactions in differentiating basal cells (38) and in bronchiolar Clara cells that arise during development, mucus metaplasia, and after NA-mediated bronchiolar injury (42, 50–52). Although, Notch signaling was a plausible candidate for regulation of β-catenin-dependent gene expression during FBP-to-Clara-like cell differentiation, a recent study by Rock and colleagues (38) demonstrated that a Notch-intracellular domain reporter transgene was inactive in basal cells in vivo and in vitro. Interestingly, this transgene was active only in cells that expressed an intermediate phenotype that was defined by the absence of basal, Clara-like, and ciliated cell markers. Because the Rock study clearly demonstrated that basal cells did not utilize a Notch-mediated mechanism, other β-catenin-dependent signals involved in FBP-to-Clara-like cell differentiation await discovery.

Clara-like cell differentiation of β-catenin WT FBP edge cells was comparable to that of β-catenin WT cluster cells. These data indicate that the β-catenin KO cell did not have a deleterious affect on its WT neighbor. This observation reinforces the notion that the β-catenin-dependent process signaled from the WT cell to the KO cell. In contrast, Clara-like cell frequency among β-catenin KO edge cells was less than that observed for either β-catenin WT cluster or edge cells. These data suggest that the β-catenin-regulated pathway exerts a dose-dependent effect of on Clara-like cell differentiation. This type of mechanism may help to explain inconsistencies among the previous analyses of β-catenin-dependent effects in the tracheal and bronchiolar epithelium (57).

Clinical relevance.

The number and types of cells that populate the airway epithelium are dysregulated in many chronic lung diseases (17, 26). Thus modification of progenitor cell fate is an attractive therapeutic goal and one that would be facilitated by detailed knowledge of the cellular and molecular pathways that control this process. Our study advances this initiative by providing evidence that β-catenin signaling determines FBP cell fate.

GRANTS

This work was supported by NIH RO1 (HL075585) and Supplement (HL075585-S1) to S. Reynolds.

DISCLOSURES

No conflicts of interest, financial or otherwise, are declared by the authors.

AUTHOR CONTRIBUTIONS

Author contributions: M.K.S. performed experiments; M.K.S. and S.D.R. analyzed data; M.K.S. and S.D.R. interpreted results of experiments; M.K.S. and S.D.R. prepared figures; M.K.S. and S.D.R. drafted manuscript; M.K.S., P.J.K., and S.D.R. edited and revised manuscript; M.K.S., P.J.K., and S.D.R. approved final version of manuscript; P.J.K. and S.D.R. conceived and designed research.