Neuroepithelial body microenvironment is a niche for a distinct subset of Clara-like precursors in the developing airways

Abstract

Clara cells of mammalian airways have multiple functions and are morphologically heterogeneous. Although Notch signaling is essential for the development of these cells, it is unclear how Notch influences Clara cell specification and if diversity is established among Clara cell precursors. Here we identify expression of the secretoglobin Scgb3a2 and Notch activation as early events in a program of secretory cell fate determination in developing murine airways. We show that Scgb3a2 expression in vivo is Notch-dependent at early stages and ectopically induced by constitutive Notch1 activation, and also that in vitro Notch signaling together with the pan-airway transcription factor Ttf1 (Nkx2.1) synergistically regulate secretoglobin gene transcription. Furthermore, we identified a subpopulation of secretory precursors juxtaposed to presumptive neuroepithelial bodies (NEBs), distinguished by their strong Scgb3a2 and uroplakin 3a (Upk3a) signals and reduced Ccsp (Scgb1a1) expression. Genetic ablation of Ascl1 prevented NEB formation and selectively interfered with the formation of this subpopulation of cells. Lineage labeling of Upk3a-expressing cells during development showed that these cells remain largely uncommitted during embryonic development and contribute to Clara and ciliated cells in the adult lung. Together, our findings suggest a role for Notch in the induction of a Clara cell-specific program of gene expression, and reveals that the NEB microenvironment in the developing airways is a niche for a distinct subset of Clara-like precursors.

The airways of the mammalian lung are populated by distinct epithelial cell types distinguished by their morphology and expression of molecular markers. Clara and ciliated cells are the most abundant, whereas basal and neuroendocrine (NE) cells are restricted in number and distribution (1). Clara cells synthesize components of the airway lining fluid, mediate mucociliary clearance in concert with ciliated cells, metabolize environmental toxins, and act as facultative progenitor cells during lung injury repair (1–3).

Clara cells in the mouse lung have been distinguished in molecular terms by the expression of the secretoglobin Scgb1a1/Clara cell secretory protein (Ccsp or CC10). Subpopulations of Clara cells with distinctive morphology and susceptibility to environmental exposures have been reported in the mouse lung and in other species (4). For example, ultrastructural analysis of conducting airways shows that Clara cells surrounding clusters of neuroepithelial cells [neuroepithelial bodies (NEBs)] have a distinct flattened morphology (5). These cells are also functionally distinct from other Clara cells because they are deficient in the cytochrome P450 enzyme Cyp2f2 (6). It is thought that the absence of Cyp2f2 renders resistance to naphthalene injury (6).

Genetic studies in mice demonstrate an essential role for the Notch signaling in the specification of Clara cells during lung development (7, 8). The Notch pathway is a juxtacrine signaling system that regulates cell fate choices in multiple organisms (9–11). The pathway is activated by the binding of transmembrane Notch receptors (Notch1–4) and transmembrane ligands Delta-like (Dll1, Dll3, Dll4) or Jagged (Jag1, Jag2). Ligand–receptor interaction results in γ-secretase–dependent cleavage of the cytoplasmic domain of the Notch receptor. Nuclear translocation and interaction of this cleaved intracellular domain (NICD) with the Rbpjk transcriptional complex then leads to activation of a canonical pathway and the expression of target genes. Developing airways that are deficient in Rbpjk or protein O-fucosyltransferase-1 (Pofut1), an enzyme required for the efficient binding of Notch receptor to ligand, lack Clara cells and have supernumerary ciliated cells (7, 8). Conversely, constitutive activation in the embryonic lung epithelium results in mucous metaplasia and in a decrease in number of ciliated cells (12).

Studies of the mechanisms of Clara cell specification and the role of Notch signaling therein have been limited by the paucity of early markers of differentiation. Transcription factors that label ciliated (Foxj1), neuroendocrine (Ascl1), and basal (Trp63) progenitors are known. However, with the exception of NICD, no nuclear factor specific to Clara cell progenitors has been identified. Hes1, a Notch transcriptional target, is expressed broadly in developing airways, but is not required for the specification of Clara cells (13). Secretoglobins are a family of small, secreted, structurally similar disulfide-linked dimers of which three members, Ccsp (Scgb1a1), Scgb3a1, and Scgb3a2, are expressed in Clara cells in the lung (14). At present, the secretoglobin Ccsp is the only definitive marker for Clara cells. Markers, such as Trp63, Ascl1, and Foxj1 have been localized to the developing airways well before the onset of, Ccsp expression at embryonic day 15.5 (E15.5) (14–16).

Here we examined the early events associated with the induction of Clara cell fate and the contribution of developmental programming in establishing differences among Clara cells in the mouse airway epithelium. We identify activated Notch1 and Scgb3a2 as markers of an early program of Clara cell fate. Moreover, we implicate Notch as a mediator of a Clara cell-specific program of gene expression in airway epithelial precursors. This program is modified in the developing NEB microenvironment to generate a distinct subpopulation of Clara cell precursors.

Results

Notch Activation in the Developing Airways Precedes Expression of Known Markers of Airway Differentiation.

Lineage analysis of cells that experienced Notch1 activation in the developing airways and loss-of-function genetic studies have shown that Notch signaling correlates with and is essential for Clara cell fate (7, 8). To characterize the spatiotemporal regulation of Notch signaling in the developing airways, we examined the distribution of activated Notch1 by immunohistochemistry (IHC) using an antibody against cleaved Notch1 [referred to hereafter as NICD (7)]. NICD labeling was first detected in the respiratory epithelium at E12.5, at low levels, and restricted to the trachea and main bronchi (Fig. 1 A and B). At E14.5, NICD-positive cells were more abundant in the proximal airways but absent in the distal airways (Fig. 1 C–E). At this stage, strong NICD signals were detected in clusters occupying a relatively apical position within the pseudostratified epithelium in proximal intrapulmonary airways (Fig. 1 E and F); this contrasted with lower intensity and more homogenous pattern of NICD labeling outside the clusters. The expression in clusters was reminiscent of the distribution of Ascl1 [mouse achaete–scute complex homolog 1 (Mash1)], a bHLH transcription factor essential for NE cell fate and a marker for presumptive neuroepithelial bodies (pNEBs) (15). Double IHC confirmed the presence of the NICD clusters juxtaposed to and distinct from Ascl1-expressing cells (Fig. 1F). At this stage most, if not all, clusters of NICD were juxtaposed to pNEBs. By E18.5 NICD was detected throughout the airway epithelium and in clusters associated with pNEBs. (Fig. 1G). Previous studies have suggested that Ascl1 (E13.5) is the earliest marker for airway differentiation (15). We conclude that Notch signaling is activated in the developing airway epithelium before any of the known markers of differentiation, and is activated around pNEBs from early stages.

Fig. 1.

Spatial pattern of Notch activation in the developing airways. (A–G) Labeling for cleaved Notch1 (NICD, red) and Ascl1 (green) indicates that Notch signaling is activated in the airways a few days earlier than Ccsp expression is detected (E15.5, 14), and that it is active in the NEB microenvironment from early stages. (A and B) Low levels of NICD were detected at E12.5 in trachea (Tr) and main bronchi (arrow), but not in distal (dl) airways (shown at higher resolution in B). (C–F) At E14.5, NICD labeling was more abundant in both the extrapulmonary airways (Tr in C, higher magnification in D) and proximal intrapulmonary airways (C, higher magnification in E), but no labeling was detected in distal airways (C Inset). (F) Clusters of strong NICD labeling in the proximal airways at E14.5 are associated with clusters of Ascl1-expressing cells (arrows, airway outlined in white). Note the solitary Ascl1-expressing cells (arrowhead) are not associated with strong NICD labeling (F). (G) At E18.5, widespread labeling of NICD was detected in the intrapulmonary airways, including the NEB microenvironment (arrow). L, lumen; Lu, lung; V, vasculature. (Scale bar: 10 μm.)

NEB-Associated Cells Show Distinct Phenotypic Features During Development Suggestive of a Subpopulation of Clara Cell Precursors.

The early onset of Notch activation in the developing lung suggested that the determination of the Clara cell fate initiates earlier than the expression of Ccsp, and led us to search for early markers. Previous studies report that Scgb3a2, another member of the secretoglobin gene family, is also expressed in secretory Clara cells of developing airways (8, 14, 16, 17). We investigated whether Scgb3a2 correlates with Notch activation at early stages. In situ hybridization (ISH) revealed Scgb3a2 expression from E12.5 onward; expression was localized to the trachea and extrapulmonary bronchi (Fig. 2A). Scgb3a2 expression became stronger and expanded to lobar bronchi at E13.5–E14.5 and was widespread by E18.5 (Fig. 2 B–D). At E14.5, Scgb3a2 was strongly expressed in clusters of cells in proximal airways (Fig. 2C). Double-ISH/IHC and confocal analysis of E14.5 lungs confirmed colocalization of Scgb3a2 and NICD (Fig. 2E). Colocalization was not restricted to these cell clusters but was harder to detect elsewhere in the epithelium due to the weaker NICD signals.

Fig. 2.

Scgb3a2 expression correlates with NICD and identifies a distinct subpopulation of Clara-like progenitors associated with pNEBs. (A–D) Time course of Scgb3a2 expression revealed a pattern similar to that of NICD (Fig. 1). Scgb3a2 was first detected in the trachea (Tr) at E12.5 (A) and in the intrapulmonary airways from E13.5–E14.5 (B–D). At E18.5 (D), Scgb3a2 expression was widespread and detected in both trachea and terminal bronchiole (TB). Clusters of cells with strong signal could be discerned at E13.5–E14.5 (B and C Inset, arrows). (E) Colocalization of Scgb3a2 and NICD was readily observed in these cell clusters (arrows; L marks the airway lumen). (F) Double-Scgb3a2 ISH/AsclI IHC revealed Scgb3a2-labeled clusters (F Right) juxtaposed to Ascl1-expressing pNEBs. (F Right Inset) Scgb3a2-expressing cells (blue) have a clear nucleus not labeled by anti-Ascl1 (brown). No Ascl1-expressing cells were detected in the trachea (F Left). (G) High-resolution optical section showing that luminal Scgb3a2-expressing cells (ISH, red) could be distinguished from basal Ascl1-expressing cells (IHC, green). (H) Scgb3a2 and Ascl1 double-labeling at E13.5 suggests that the formation of Ascl1 clusters (distal, arrows) precedes formation of Scgb3a2-Ascl1 dual clusters. (I–O) Analysis of Ccsp and Scgb3a2 expression at E18.5 suggested that the cells associated with pNEBs may be a distinct subpopulation of Clara precursors. (I–K) Labeling of NICD (red, shown separately in J), Ccsp (white, shown separately in K), and Ascl1 (green) showed that the cells in the NEB microenvironment are NICD positive, but express low (arrow)-to-negligible levels of Ccsp at this stage. Cells with both NICD and Ccsp were abundant away from the NEB microenvironment (I). (L–O) Triple labeling for Scgb3a2 (ISH, red, shown separately in M), Ccsp (ISH, green, shown separately in N), and Cgrp (IHC, blue, shown separately in O) at E18.5 showed that cells apposed to pNEBs express Scgb3a2 and low Ccsp (arrowhead), and some have Scgb3a2 but negligible Ccsp (arrow). Elsewhere, Scgb3a2 and Ccsp signals are strong and colocalized (L).

We tested whether clusters of Scgb3a2-positive cells were arising at sites of Ascl1-expressing cells, as seen for NICD. Sections from E13.5–E18.5 lungs were labeled with a Scgb3a2 riboprobe and an anti-Ascl1 antibody. At E14.5, Scgb3a2 signals were present in the expected uniform and clustered patterns (Fig. 2F); 70–80% of the Scgb3a2-expressing clusters were associated with the Ascl1-expressing cell clusters. Confocal microscopy confirmed that cells expressing Scgb3a2 or Ascl1 were distinct and in direct contact (Fig. 2G). Interestingly, analysis of E13.5–E14.5 lungs showed a P–D pattern of pNEB formation in intrapulmonary airways, preceding the local appearance of Scgb3a2 (Fig. 2H).

These data showed that the pNEB microenvironment at E14.5 consists of clusters of Ascl1-expressing cells and another cell population expressing high levels of NICD and Scgb3a2. No Foxj1 labeling was detected in these cells at any of the stages studied. Therefore, we hypothesized that the NEB microenvironment is a niche for Clara cell precursors and examined the distribution of Ccsp in the NEB microenvironment at E18.5. Simultaneous labeling of NICD/Ccsp/Ascl1 at E18.5 demonstrated that cells clustered around NEBs had high levels of NICD but low-to-negligible levels of Ccsp (Fig. 2 I–O). Three-color labeling of Scgb3a2 (ISH)/Ccsp (ISH)/calcitonin gene-related peptide (Cgrp), another NE cell marker (IHC), confirmed that cells in the pNEB microenvironment continue to express Scgb3a2 at this stage and have low-to-negligible levels of Ccsp. Our data suggested that cells associated with NEBs are a distinct set of Clara cell progenitors.

High Levels of Uroplakin 3A Expression Distinguishes Secretory Precursors in the NEB Microenvironment.

The absence of secretory cells in the airways of Notch signaling-deficient mice presented a unique opportunity to screen for additional markers enriched in the developing Clara cells. Thus, we compared the global gene expression profile of E18.5 lungs from control and mutant mice in which the Rbpjk gene was disrupted in the airway epithelium (Shh-Cre;Rbpjk^Flox/Flox or Rbpjk^cnull) (8, 18). As expected, genes associated with the Clara cell phenotype, such as Ccsp, Scgb3a2, and Cyp2f2, were expressed at significantly higher levels in control lungs (Fig. 3A). A comprehensive description of this screen will be reported elsewhere. Importantly, this screen identified Uroplakin 3a (Upk3a; Fig. 3A) as a putative marker of Clara precursors. Upk3a encodes a single-pass transmembrane domain-containing protein of the Uroplakin family that is expressed in the urinary bladder epithelium (19, 20). Mice deficient in Upk3a have compromised urothelial permeability, but no phenotype has been described in the lung (19).

Fig. 3.

High levels of Upk3a expression distinguishes Clara-like precursors in the pNEB microenvironment. (A) Profiling of E18.5 control and Notch signaling-deficient (Rbpj^cnull) lungs identified known markers for Clara cells and implicated Upk3a as a candidate marker. (B and C) ISH at E18.5 in control lungs revealed that Upk3a was highly enriched in clusters of cells in the proximal airways (C, arrow) and expressed at low levels in scattered cells in distal airways (C, bracket). No signal was detected along the airway axis in Rbpj^cnull airways (Tr, trachea). (D) qPCR analysis showed that Upk3a levels can be detected at E12.5 and increase throughout development. (E and F) Upk3a transcripts were detected by ISH from E14.5 onward in cell clusters juxtaposed to Ascl1-expressing cells (E and F; Insets show higher magnification). Some expression away from clusters was detected from E16.5 onward (F, bracket). Upk3a expression was also detected in a few cells in the adult airways (G, arrow) frequently juxtaposed to Cgrp-expressing NEBs (G Inset). (H–K) Upk3a expression was perturbed in Ascl1-null (Ascl1^−/−) lungs. Clusters of high Upk3a expression were detected in control at both E14.5 (H) and E18.5 (J), but not in mutant lungs at either time point [I and K, circled regions; note expression of Upk3A is detected in the esophagus in the mutant at E14.5 (I, eso)]. Rare Upk3a-expressing cells (nonclustered) could still be seen in the E18.5 Ascl1-null mutant (K Inset). (L) qPCR analysis of Upk3a expression in Rbpjk-deficient and Ascl1 mutants at E14.5 and E18.5 showing that Upk3a expression is dependent on both Notch signaling and Ascl1.

We examined Upk3a expression in control and Rbpjk^cnull mice by ISH (Fig. 3 B and C). In E18.5 controls, Upk3a was highly enriched in cell clusters in the proximal airways, although signals could be also detected at lower levels in scattered cells in the distal airway epithelium (Fig. 3B, bracket); this differed from the widespread expression patterns of Scgb3a2 and Ccsp at E18.5 (Fig. 2). Upk3a signals were undetectable in airways of Rbpjk^cnull mice (Fig. 3C).

Developmental analysis of Upk3a expression using ISH and quantitative real-time PCR (qPCR) detected signals at E12.5 (qPCR) and 14.5 (ISH), and showed increasing levels thereafter (Fig. 3 D–F). ISH revealed clustered distribution from early stages and prompted us to investigate the relationship of Upk3a-expressing cells with pNEBs. Double ISH (Upk3a)/IHC (Ascl1) showed that Upk3a-expressing cells were juxtaposed to the Ascl1-labeled clusters from the earliest stages (Fig. 3 E and F). Quantitative analysis showed that 70–80% of pNEBs at E14.5 were associated with Upk3a clusters. Next we investigated whether Upk3a-expressing cells were present in the NEB microenvironment in the adult lung. Double ISH/IHC using a Upk3a riboprobe and an anti-Cgrp antibody showed at least 1–5 Upk3a-expressing cells in association with 50% of all NEBs examined (Fig. 3G).

Because Cyp2f2 has been reported as a marker for Clara cells (17) and was also identified by our expression profiling as down-regulated in Notch-deficient airways, we characterized its developmental pattern of expression. ISH showed Cyp2f2 expression throughout the epithelium of the trachea and extrapulmonary airways from E13.5 and expanded to distal airways at later times (Fig. S1 A and B). To determine the spatial relationships among Scgb3a2, Upk3a, Cyp2f2, and pNEBs, serial sections were labeled with these riboprobes and subsequently stained with an anti-Ascl1 antiserum. Analysis of E14.5 lungs showed that the pNEB-associated cell clusters that typically express strong Scgb3a2 and Upk3a signals had little, if any, Cyp2f2 (Fig. S1 C–E); this contrasted with the strong Cyp2f2 signals in neighboring Scgb3a2-positive cells. At E18.5, this subpopulation could be distinguished by strong Upk3a but not by low Cyp2f2 expression as seen at earlier times (Fig. S1F).

NEB Microenvironment Harbors a Subset of Clara Cell Precursors in Developing Airways.

Next we asked whether the features identified in the pNEB-associated cells were dependent on the pNEBs. Previous studies show that neither solitary NE cells nor NEBs form in Ascl1-null mice (13, 15). Given the small numbers of cells in the pNEB microenvironment and paucity of specific markers for these cells, we reasoned that the ablation of these cells would have escaped detection, and reexamined Ascl1-null mice. Analysis of these mutants at E14.5 showed that, in the absence of pNEBs, the clustered expression of NICD and Scgb3a2 was abolished, although the weak, nonclustered signals remained in proximal airways (Fig. S2 A–D). At E18.5, the strong expression of NICD and Scgb3a2 throughout the airway epithelium did not allow distinguishing differences between control and Ascl1 mutants.

We then investigated the impact of Ascl1 deletion on Upk3a. No Upk3a-labeled cell clusters were detected in Ascl1-null mutant lungs at E14.5 or E18.5 (Fig. 3 H–K). The selective loss of the Upk3a-labeled cell clusters was further supported by the presence of Upk3a signals in other structures (esophagus; Fig. 3I) or in the scattered cells outside the NEB microenvironment (Fig. 3K, Inset). The Ascl1 phenotype contrasts with the generalized disruption of Upk3a that we found in the airways of Rbpjk^cnull mice (Fig. 3C). qPCR showed that Upk3a mRNA is abolished in Rbpjk^cnull lungs both at E14.5 and E18.5 (Fig. 3L). The observation is consistent with the inability to form Clara cells upon disruption of Notch signaling. In Ascl1-null mice, Upk3a expression was also markedly down-regulated at E14.5, but only marginally decreased at 18.5 (Fig. 3L). The more dramatic reduction in Upk3a transcripts seen at E14.5 mutants was expected because at this stage, Upk3a-expressing cells formed essentially around pNEBs. Later, the contribution of non–pNEB-associated cells minimized the differences in Upk3a expression between control and Ascl1 mutants (see Fig. S2 E–G for Scgb3a2).

These results suggest that pNEBs are required to induce if not maintain a population of cells expressing high levels of Upk3a in its microenvironment.

The identification of Upk3a as a marker enriched in pNEB microenvironment provided us with an opportunity to examine the fate of these cells. We investigated the fate of Upk3a-expressing cells in the airways during prenatal and adult life using an Upk3aCreER^T2 transgenic mouse. A single dose of tamoxifen (Materials and Methods and Fig. 4A) was administered to Upk3aCre ER^T2;Rosa26lacZ dams 15.5 d postcoitum. Offspring were analyzed at E18.5 and at postnatal day 60. As putative Clara cell precursors, we investigated the ability of Upk3aCre-labeled cells to give rise to Clara and ciliated cells (2). Thus, we performed double-IHC/X-gal staining using antibodies against Foxj1, β-Tubulin IV (ciliated), and Ccsp, followed by quantitative analysis (Fig. 4). At E18.5, although lacZ signal was detected in a small number of cells, few were double-labeled with Ccsp, and only a single cell was double-labeled with Foxj1; this showed that the Upk3a lineage was largely uncommitted to either Clara or ciliated fates at this stage and was consistent with our IHC/ISH studies. In contrast, analysis of adult lungs showed labeling in both ciliated and Clara cells at nearly similar proportion (Ccsp: 47.2 ± 3.4%; β-Tubulin: 45.3 ± 4.0%; Fig. 4 J and K). The pool of cells that were lacZ positive; Ccsp negative and lacZ positive; β-Tubulin negative may include cells that remained undifferentiated even in adults. Lineage-labeled cells were distributed throughout the airway epithelium, and only a few of these cells were apposed to pNEBs (Fig. 4 F and G). Analysis using the NE marker PGP9.5 showed no colocalization with lacZ (n = 69 lacZ-positive cells). Together our data suggest that during development the pNEB microenvironment harbors a population of Clara-like precursors that generate both Clara and ciliated cells in adult life.

Fig. 4.

Lineage analysis of Upk3a-expressing cells at E15.5 reveals that these cells are precursors of Clara and ciliated cells. (A) Experimental protocol for the induction and harvest of tissue from the Upk3acreER^T2 × Rosa26lacz. (B–E) Upk3a lineage-derived cells (blue) are distributed along the proximal/distal axis (counterstained with Fast Red) in clusters and as solitary cells (B, arrows). Regions 1, 2, and 3 in B are shown at higher magnification in C–E. (F and G) Double labeling of X-gal–stained preparations for PGP9.5 shows that Upk3a lineage-derived cells (arrows) are not in close association with NEBs (arrowheads), although rare examples of apposition are observed (G). Note that LacZ (arrow)- and PGP9.5 (arrowhead)- expressing cells are distinct. (H and I) Upk3a-expressing cells labeled at E15.5 contribute to Clara and ciliated lineages in adults. Double labeling of X-gal–stained preparations for Ccsp (H) and β-tubulin (I) demonstrate that these cells contribute to both Clara (H) and ciliated (I) lineages. (J and K) Quantitation of the numbers of X-gal–stained cells that co-label for Clara and ciliated markers at E18.5 (Ccsp, Foxj1, n = 156 airways) and in adults (Ccsp, β-Tubulin, n = 501 airways). The Upk3a lineage-derived cells are mostly uncommitted to either Clara or ciliated fates at E18.5 but differentiate into these lineages thereafter. K, mean ± SEM.

Notch Signaling Regulates Expression of Scgb3a2, Upk3a, and Ccsp by a Transcriptional Mechanism.

To gain further insights into the mechanisms by which Notch signaling regulates gene expression in Clara progenitors, we examined the effect of loss or gain of Notch function on expression of genes at early stages, before the onset of Ccsp expression. We generated mice in which Notch signaling was disrupted (Rbpjk^cnull) (18) or constitutively activated in the lung epithelium using ShhCre and Rosa-NICD transgenes (12). Analysis of the Scgb3a2 pattern in E14.5 Rbpjk^cnull lungs showed marked down-regulation in extrapulmonary airways and negligible signals in intrapulmonary airways, including the cells in the pNEB microenvironment, which normally express strong Scgb3a2 (Fig. 5 A and B and Fig. 2). In contrast, E14.5 ShhCreNICD mice showed widespread expression of Scgb3a2 in the airway progenitors (Fig. 5E). Upk3a expression was similarly modulated by Notch signaling (Fig. 5 B and F). We did not detect expression of Ccsp in E14.5 lungs overexpressing NICD, but strong signals were present at E18.5 (Fig. S3 A and B). As evidenced by the morphology, NICD overexpression led to aberrantly enlarged airways. Nevertheless, these airways retained respiratory identity as they expressed Ttf1 (Fig. 5 C and D).

Fig. 5.

Notch-dependent regulation of secretory cell-associated genes. (A–F) Analysis of Notch gain and loss of function in E14.5 airways in vivo. (A and B) Disruption of Notch signaling in Rbpj^cnull mutants disrupts expression of Scgb3a2 and Upk3a. Comparable segments of the intrapulmonary airways in A and B (encircled by gray lines) evidence negligible expression of both markers in Rbpj^cnull lungs. Residual low-level expression of Scgb3a2 was detected in the extrapulmonary airways of mutants only after prolonged staining (A, arrow). Double-Scgb3a2 ISH/AsclI IHC shows that specification of pNEBs was unaffected in Rbpj^cnull mutants, but no Scgb3a2 was detected in surrounding cells (A Inset, arrowhead). (C–F) Shh cre-driven NICD perturbed lung architecture (compare distribution of Ttf1 in C and D) and resulted in widespread expression of Scgb3a2 and Upk3a. (G and H) In vitro analysis of the role of Notch signaling in the regulation of the Scgb3a2 promoter. (G) Schematic showing presumptive Rbpjk binding sites (high-affinity site, red arrow; low-affinity site, gray arrow) in the Scgb3a2 promoter 1 kb upstream to the transcription start site. (G Lower) Sequence of a −273-bp fragment of the Scgb3a2 promoter showing putative Rbpjk (red) and TTF1 (green) binding sites. (H) Validation of the predicted high-affinity Rbpjk binding site in the Scgb3a2 promoter by EMSA. An anti-Rbpjk antibody supershifted the labeled oligonucleotide containing the putative high-affinity binding site (CTRL, lane +ab). Note that the band that is supershifted by anti-Rbpjk was competed by coincubation of the labeled oligonucleotide with an unlabeled CTRL oligonucleotide (cognate) but not competed by coincubation with an unlabeled mutant oligonucleotide lacking the high-affinity binding site (MUT, sequences shown below). (I) Luciferase assay examining the sufficiency of NICD and Ttf1 in the transactivation of the Scgb3a2 promoter. NICD expression alone did not significantly transactivate the Scgb3a2 promoter, but coexpression of NICD and Ttf1 synergistically up-regulated reporter expression. This synergistic up-regulation was abolished when the high-affinity Rbpjk binding site was mutated (−273mut; G).

Based on the genetic data above and our evidence establishing Scgb3a2 as an early Clara marker, we investigated whether Notch could directly regulate Scgb3a2 transcription. A screen for Rbpjk binding sites 1-kb region upstream to the transcription start site (TSS) of the murine Scgb3a2 promoter revealed high (YGTGRGAA)- and low (RTGRGAR)-affinity binding sites (Fig. 5G) (21–23). For subsequent analysis we focused on the putative high-affinity binding site. We performed EMSA using COS-1 cell nuclear extracts and oligonucleotides containing the Rbpjk high-affinity binding sequence (intact, mutated); COS-1 cells are known to endogenously express Rbpjk (24). Detection of a DNA–protein band that was supershifted with an anti-Rbpj antibody confirmed that endogenous Rbpjk can bind to this site (Fig. 5H). Next, we performed luciferase reporter assays using Scgb3a2 promoter sequences −273 bp to the TSS containing either a control (WT) or a mutated Rbpjk high-affinity site. Cotransfection with an NICD-containing plasmid resulted in minimal induction of reporter activity (<twofold). This response was similar to that elicited by cotransfection of the same construct with an expression plasmid-containing Ttf1, a known transcriptional regulator of Scgb3a2. We conclude that despite the presence of a high-affinity binding site in the promoter fragment, NICD expression is insufficient to induce reporter expression (Fig. 5I).

Interactions between Notch and other transcription factors have been implicated in the control of gene expression during development (21, 25). Moreover, synergistic regulation of the mouse Scgb3a2 gene by Ttf1 and other transcription factors, such as C/EBP, has been previously reported (26). We tested the possibility that NICD could act in concert with Ttf1 to transactivate the Scgb3a2 promoter (Fig. 5I). Coexpression of Ttf1 and NICD dramatically increased reporter activity (over 40-fold). Importantly, introducing a mutation in the Rbpjk high-affinity site (TTCTCACG to TTGTCTCG) nearly abolished this synergistic interaction (Fig. 5I).

We extended our promoter analysis to other Clara cell-associated genes such as Upk3a and Ccsp. Both genes have multiple Rbpjk and Ttf1 binding sites in their promoters (Figs. S4 and S5). Reporter assays showed that both Upk3a and Ccsp were transactivated by NICD and Ttf1 in a synergistic manner, as observed for Scgb3a2 (Figs. S4 and S5). Together, our data suggest a model in which Notch, in conjunction with locally expressed transcription factors such as Ttf1, regulates gene transcription in Clara cell progenitors.

Discussion

In this study we examined how Notch influences Clara cell fate specification in airway progenitors, and the influence of local microenvironment in generating diversity among Clara cell precursors during development. We identify expression of Scgb3a2 and Notch activation as early events in a program of Clara cell differentiation. Our in vitro assays show that Notch activation alone has a minimal effect in the transactivation of genes in Clara cell progenitors but is highly effective when combined with Ttf1. This context dependency is of biological significance because genetic studies have shown that Ttf1 is critical for the overall induction of respiratory cell fate, including secretory Clara cells (27, 28).

Ultrastructural studies of the developing airways show the pNEB microenvironment contains a distinct population of nonneuroendocrine, nonciliated cells (5). Here we report that this population of pNEB-associated epithelial cells can be distinguished in molecular terms by the expression of markers associated with Clara cells such as NICD, Scgb3a2, Upk3a, Cyp2f2, and Ccsp. Based on the marker that is most enriched and best distinguishes this population of cells, Upk3a, we undertook a lineage study to trace their fate in Upk3aCreER^T2;Rosa26 lacZ mice. We find that labeling at E15.5 results in lacZ expression in both Clara and ciliated cells of the adult airways at nearly the same proportion. Moreover, this lineage did not contribute to NE cells. Thus, our studies demonstrate that the Upk3a-expressing cells, despite expressing little Ccsp, exhibit the properties of a Clara progenitor and implicate the pNEB microenvironment as a niche for Clara cell progenitors. It is likely that pNEBs are critical to specify or maintain the features of this subpopulation; however, the identification of the pNEB signals that mediate this process is beyond the scope of this work. Candidate signals present in NE cells potentially involved in these interactions include the Delta-like family of Notch ligands, or other currently uncharacterized Ascl1 targets (15). Interestingly, lineage analysis of Scgb1a1CreER;R26YFP mice demonstrates that Clara cells labeled in adulthood are more able to generate ciliated cells than Clara cells labeled at E18.5 (2). The identification of the Upk3a lineage-derived Clara cells could account for this difference. Whether the Upk3a-derived Clara cells that remain uncommitted during the embryonic period have a higher propensity to generate ciliated cells in adulthood remains to be determined.

The spatiotemporal pattern of Scgb3a2 we describe in murine airways, including its enrichment in clusters in the pNEB microenvironment, is highly reminiscent of the CCSP pattern reported in developing airways of humans (29). Although SCGB3A2 has been reported in human neonatal lungs, no information is available on its pattern at early developmental stages in humans (14). A number of human conditions are associated with abnormal increase in NE bodies, including neuroendocrine hyperplasia of infancy (NEHI), diffuse idiopathic pulmonary neuroendocrine cell hyperplasia, and bronchopulmonary dysplasia (BPD). The aberrant expansion of neuroepithelial cells in these conditions is likely to have a profound impact in the local microenvironment, as shown by the increase in the NEB-associated clusters of CCSP-expressing cells in BPD lungs (29). Thus, further studies to better understand the mechanisms and consequences of the cellular interactions in this microenvironment have potentially high clinical significance.

Materials and Methods

A detailed description of the reagents and methodologies (mouse models, ISH, IHC, real-time PCR, EMSA, site-directed mutagenesis, DNA transfections, and luciferase reporter assays) used in this work can be found in SI Materials and Methods.