The submucosal glands (SMGs) of the upper respiratory tract play an important role in secreting mucus, lysozyme, defensins, and other agents that help protect the lungs from particles and infectious agents. They develop from small buds that arise in the dorsolateral airway epithelium between the cartilage rings. These buds extend into the surrounding mesenchyme, undergo branching morphogenesis, and differentiate into mucous and serous cells that produce distinct secreted products (Finkbeiner, 1999). In the human, SMGs are found along the airways from the larynx down to the distal part of the main bronchi. However, in the mouse, they are restricted to the upper trachea, more specifically to the regions between the first few cartilage rings, with the precise distribution depending on genetic background (Borthwick et al., 1999; Innes and Dorin, 2001). The significance of the SMGs for human health lies in the fact that they undergo enlargement in patients suffering from respiratory disorders. These disorders include severe asthma and bronchitis, both associated with airway inflammation and the secretion of excessive amounts of mucus (Jeffery, 2000). For example, in patients with severe/fatal asthma, the size of SMGs increases more than twofold as judged from histological sections of bronchial biopsy and postmortem material (Benayoun et al., 2003). Hypertrophy and hyperplasia of SMGs are also seen in dogs and rats exposed to environmental agents such as sulfur dioxide that damage the surface epithelium (Lamb and Reid, 1968). Furthermore, recent work suggests that a change in the volume and the viscosity of secretions from the SMGs may be the primary defect in cystic fibrosis (Salinas et al., 2004). However, despite the significance of SMGs for human respiratory disease, little is known about the mechanisms controlling their growth, differentiation, and homeostasis, either in the early postnatal animal or in the adult. It is likely that these mechanisms are activated or subverted in response to conditions that ultimately cause pathological changes in SMGs.
One approach to identifying factors controlling SMG growth and differentiation is to study mouse mutants in which their development is defective. Along these lines, it was shown several years ago that SMGs are absent from Lef1 null mutant mice, suggesting that the canonical Wnt signaling pathway is involved in their development (Duan et al., 1999). The localization of Lef1 transcripts to the early buds of the neonatal ferret tracheal SMGs is consistent with the initiation of the glands being dependent on a Wnt signal from either the underlying mesenchyme or adjacent epithelium (Duan et al., 1998; Driskell et al., 2004). Lef1 is also required for the development of many ectodermal appendages, including hair follicles, whiskers, teeth, and mammary gland (van Genderen et al., 1994; DasGupta and Fuchs, 1999; Sasaki et al., 2005). In all cases, there is a dynamic and reciprocal interaction between the mesenchymal and epithelial components, mediated by signaling factors that include not only the Wnts but also bone morphogenetic proteins (Bmps), fibroblast growth factors (Fgfs), Sonic hedgehog, and their respective antagonists. The potential role of these factors in the induction of SMGs and their subsequent morphogenesis has not been explored.
In this study, we follow-up on the original observation by Gruneberg that tracheal SMGs are absent in Tabby mice. These mutants also have defects in the development of hair, teeth, and sebaceous and salivary glands (Gruneberg, 1971a, b). Tabby is an X-linked gene that encodes Ectodysplasin (Eda), a member of the TNF superfamily of intercellular signaling factors (Ferguson et al., 1997; Srivastava et al., 1997; Mikkola and Thesleff, 2003). A similar phenotype is seen in mice with mutations in autosomal genes encoding components of the Eda signaling pathway. These genes include Edar, which is mutated in downless (Edardl) mice and encodes a TNF receptor of the death domain class, and Edaradd, which is mutated in crinkled (Edaraddcr) mice and encodes an adaptor protein with a death domain (Headon and Overbeek, 1999; Headon et al., 2001; Thesleff and Mikkola, 2002). The Eda/Edar signaling pathway is evolutionarily conserved and mutations in the corresponding human genes, EDA, EDAR, and EDARADD, cause the inherited disorder known as hypohidrotic ectodermal dysplasia (HED). Patients are characterized by abnormalities in multiple organs including hair, teeth, and sweat and sebaceous glands. Significantly for this study, patients with HED lack mucus-producing glands in the nose, larynx, and bronchi and frequently have severe chronic pulmonary complications that may require lung transplantation (Smythe et al., 2000).
To learn more about the mechanisms underlying the normal and abnormal development of SMGs in the mouse, we have followed their development in relation to genes encoding signaling factors that control epithelial budding and branching morphogenesis in other organ systems such as hair, teeth, and the distal lung. These include BMPs, HHs, and FGFs. In addition, we have confirmed that SMGs are indeed absent in adult Tabby and homozygous mutant crinkled mice. Finally, we present evidence that the Eda/Edar and other TNF signaling pathways are expressed in the lung outside of the SMGs and, therefore, may play more extensive roles in maintaining the homeostasis of the lung and its response to pathological conditions.
Early Development of Normal Submucosal Glands
The buds that give rise to the SMGs first appear in the most proximal intercartilage region of the tracheal epithelium at postnatal day 2 (P2; Fig. 1A). By P4, the first buds to form have enlarged and extended and have undergone some branching, but new buds are still forming. There is evidence at this stage, but not earlier, for some production of mucus within the older glands, as judged by Alcian blue staining and transcription of the gene encoding Demilune Cell and Parotid Protein (Dcpp), which is produced by the mature glands (Fig. 1B, and data not shown). In addition, immunohistochemistry with anti–smooth muscle actin reveals myofibroblasts around the developing glands at this time (Fig. 1D). Mucus production is well under way by P7 (Fig. 1C). Even at P7 and P14, new buds are still forming and proliferation of cells within these buds can be detected by a short pulse of labeling with bromodeoxyuridine (BrdU; data not shown). However, no evidence for new bud formation is seen in animals older than P21. These results suggest that the factors inducing the formation of new SMGs are likely to be present over a period of several weeks.
At P2, when buds first appear, basal cells positive for keratin 14 (K14) are present within the tracheal epithelium (Fig. 1F, arrows). The initial buds, and the proximal regions of the early glands that will form the future ducts, appear to be enriched in K14-positive cells (Fig. 1F, arrowheads). K14 basal cells are also present within the ducts and secretory region of P7 and adult glands (Fig. 1G, arrows and data not shown), as noted by others for adult mice (Borthwick et al., 2001).
Gene Expression in the Mesenchyme Associated With Developing Submucosal Glands
The development of hair follicles in the embryonic mouse skin is associated with the dynamic and localized expression of genes encoding extracellular signaling factors within either the mesenchymal dermis or the epidermis. Among these genes are members of the Bmp superfamily, including Bmp4 (Kulessa et al., 2000; Botchkarev and Sharov, 2004). We therefore examined the expression in the developing SMGs of Bmp4lacZ, which faithfully reports gene transcription (Lawson et al., 1999). As shown in Figure 2, at P2 Bmp4lacz is widely expressed in the mesenchyme underneath the tracheal epithelium and around early buds. By P7 expression is more restricted to the mesenchymal cells around the developing glands and by 4 weeks after birth, Bmp4lacZ is only seen in a few mesenchymal cells underneath the tracheal epithelium and SMG acini (Fig. 2C,D). Expression is not detected in the epithelial cells at any stage.
Fgf10 Is Required for Submucosal Gland Morphogenesis
Studies in hair and feather development have implicated Fgfs in the reciprocal signaling between mesenchyme and epithelium at several stages (Ohuchi et al., 2003; Petiot et al., 2003; Mandler and Neubuser, 2004). In addition, recent results have uncovered a role for Fgf10 in the development of the lacrimal and salivary glands of both the mouse and human. Fgf10 is transcribed in the mesenchyme adjacent to several different budding organs, these organs include the lacrimal gland, feather placodes, and distal lung buds (Bellusci et al., 1997; Makarenkova et al., 2000; Mandler and Neubuser, 2004). Fgf10 heterozygous mice do not develop lacrimal glands and have hypoplastic or absent salivary glands (Makarenkova et al., 2000; Entesarian et al., 2005). We therefore examined serial sections through the trachea of Fgf10+/− heterozygotes and their wild-type siblings. At P20, when the SMGs are largely developed, wild-type mice had well-formed, highly branched glands above the first tracheal ring and smaller SMGs between the more distal cartilage rings extending as far as the sixth cartilage element (Fig. 3A,B, and data not shown). The few SMGs present in the Fgf10+/− trachea were in the most proximal position above the first cartilage ring; these were not branched as extensively as the controls. More distally, there were no SMGs between the tracheal rings (Fig. 3C,D).
Submucosal Glands Are Absent in Tabby (EdaTa/Y) Males and crinkled (Edaraddcr) Homozygotes
To confirm the observation of Gruneberg that SMGs are absent in Tabby mice, we examined histologically the tracheas of four adult EdaTa/Y males. All samples showed the complete absence of glands (Fig. 4A,B). The mesenchyme beneath the tracheal epithelium where glands should have been present appeared normal and contained blood vessels and calcitonin gene-related peptide (CGRP) positive nerves (data not shown). This analysis clearly established that SMGs are absent in adult animals defective in signaling through the Ectodysplasin pathway. However, it did not distinguish between a failure in the initial formation of the gland primordia and a situation in which glands first form and then regress. This second possibility is consistent with the observation that submandibular salivary glands do form in mice with mutations in the Eda pathway, although their subsequent morphogenesis is disrupted (Jaskoll et al., 2003). We therefore analyzed the earliest stages of SMG development in postnatal pups homozygous for a null mutation in Edaradd. This gene encodes an essential intracellular adaptor protein that interacts with the death domain of the Eda-1 receptor (Headon et al., 2001). Homozygous crinkled (Edaraddcr/cr) pups at P7 and P14 lacked any sign of newly formed or nascent SMGs (Fig. 4E,F). By contrast, wild-type or heterozygous littermates had both extending and newly formed buds. This result suggests that signaling through the Eda-1 receptor (Edar) is required for the initial formation of buds, rather than for their subsequent maintenance.
Expression of Components of the Eda Signaling Pathway in the Postnatal Trachea and Lungs
The pseudostratified tracheal epithelium of the adult EdaTa/Y mice contained the normal complement of keratin 14-positive basal cells, ciliated cells, and secretory cells. However, individual columnar epithelial cells appeared higher than normal and the epithelium as a whole had a more disorganized appearance (Fig. 4C,D). These changes could be either a direct result of a function for Eda signaling in the tracheal epithelium itself, or secondary to the absence of submucosal glands over a prolonged period. The latter explanation is supported by the observation that the Edarddcr pups do not have an obvious tracheal epithelial phenotype. Nevertheless, to explore possible additional roles for the Eda signaling pathway in the trachea and lung, we asked if pathway components continue to be expressed after submucosal gland formation is complete. The genes examined were Eda, Edar, and Xedar and Tnfsrf19, which encode related members of the tumor necrosis factor receptor superfamily (Hu et al., 1999; Kojima et al., 2000; Yan et al., 2000). At P49, transcripts from all four genes were detected by reverse transcriptase-polymerase chain reaction (RT-PCR) in the upper region of the trachea, where SMGs are present, and also in RNA from more distal tracheal regions and the distal lung (Fig. 5A). Similarly, at P3 and P9, when SMG morphogenesis is occurring, transcripts for all four genes were present throughout the airways (data not shown). In situ hybridization confirmed that Edar mRNA is expressed throughout the entire tracheal epithelium at P4 and was not just restricted to the gland-forming regions (Fig. 5B,C).
In this study, we document the development of mouse tracheal submucosal glands. We show that they bud from the surface of the tracheal epithelium over the first 3 weeks of postnatal development. In addition, although the earliest-formed glands are apparently functional by P7, they continue to branch and grow until adulthood. Furthermore, we show that the Eda signaling pathway is absolutely required for the development of SMGs and that it functions at the earliest stage of SMG formation, namely the initial outgrowth of buds into the mesenchyme. Our analysis raises the possibility that Eda and related signaling pathways have additional function(s) in the normal lung and trachea, for example in innate immunity.
The SMGs develop from small epithelial placodes that bud into the mesenchyme in a similar manner to the primordia of the hair follicle, tooth bud, and other ectodermal organs. Like these organs, the SMGs require Lef1 function for their development, implicating the Wnt intercellular signaling pathway in the formation of the initial buds (Duan et al., 1999). Here, we have shown that the Eda signaling pathway is absolutely required for SMG formation, just as it is for the first wave of hair follicles. We therefore asked whether hair and SMG development share any other early signaling events. In the developing hair follicle, Bmp4 expression is first restricted to the mesenchymal condensation beneath future placodes. Here, it is thought to control follicle spacing by acting as an inhibitor of placode formation in the surrounding regions (Jung et al., 1998; Noramly and Morgan, 1998; Wilson et al., 1999). In contrast, we found that Bmp4 is expressed throughout the tracheal mesenchyme during the development of the SMGs so that it presumably cannot be functioning to control their spacing. However, Bmp4 expression does become restricted to mesenchymal cells around the SMGs at late stages in their development and this expression persists into adulthood. Here, by analogy with its presumed role in the intestinal crypts (He et al., 2004), it may function to regulate SMG homeostasis. NogginlacZ, Patchedlacz, and Fgf8LacZ, which are highly expressed in specific cells of the developing hair follicle and tooth bud, were undetectable in or around the SMGs throughout the P2 to P14 period when morphogenesis is occurring (data not shown).
There is plenty of evidence to support the idea that Fgf10 plays an important role in the initial budding and subsequent branching morphogenesis of many organs that develop from epithelial placodes (Makarenkova et al., 2000; Ohuchi et al., 2003; Mandler and Neubuser, 2004; Steinberg et al., 2005). In particular, it plays a key role in the distal lung, promoting outgrowth of epithelial buds (Bellusci et al., 1997; Min et al., 1998; Sekine et al., 1999; Weaver et al., 2000). We have found that in Fgf10 heterozygotes fewer SMGs are formed, and these do not undergo branching morphogenesis to their full extent. This finding supports the notion of general functions for Fgf10 both in initial budding and branching morphogenesis of these organs. Interestingly, there was a proximal–distal distribution of the severity of the effect on SMGs in the Fgf10 heterozygotes: the most distal SMGs did not bud at all, the SMGs above the first cartilage element budded and branched but did not complete their branching morphogenesis, and the morphogenesis of the more proximal nasal glands was apparently normal (data not shown). This finding suggests that either there is a proximal–distal difference in the requirement of the glands for Fgf10 or, more likely, that other Fgfs are compensating in the more proximal regions. Fgf10 is typically transcribed in the mesenchyme adjacent to budding organs, these include the lacrimal gland, feather placodes, and distal lung buds (Bellusci et al., 1997; Makarenkova et al., 2000; Mandler and Neubuser, 2004).
Both Wnt and Eda signaling are required for the formation of the first wave of hair and feather follicles and in tooth development, but the precise connection between these two signaling pathways, and other intercellular signaling events, is still unclear. In vitro experiments have suggested that Eda is directly regulated by Wnt signaling in the tooth bud (Pengue et al., 1999; Laurikkala et al., 2001; Durmowicz et al., 2002). More recently, it has been suggested that Eda signaling is a competence factor for placode formation in the early stages of ectodermal organ development (Mustonen et al., 2004; Houghton et al., 2005). Both of these hypotheses are consistent with our data for an early role of Eda signaling in SMG development. Interestingly, overexpression of Eda ligand itself in adult mice cannot rescue the Tabby phenotype but can cause sebaceous gland hyperplasia in wild-type animals (Cui et al., 2003; Gaide and Schneider, 2003; Mustonen et al., 2003). Therefore, it is tempting to speculate that an increase in levels of Eda signaling plays a role in the SMG hyperplasia observed in severe asthma and other respiratory diseases.
We have shown that Eda, Edar, and two related receptors, Xedar and Tnfsfr19 are expressed in the lower trachea and lung, both in the early postnatal period and in the adult. In addition, Traf 4, which encodes an adaptor protein for some TNF receptor family members, is expressed in airway epithelial cells, and homozygous mutants have a constriction of the upper trachea, although SMGs have not been studied (Krajewska et al., 1998; Shiels et al., 2000; Regnier et al., 2002). These data suggest that signaling through the TNF pathway has important roles not only in the development of the SMGs but also in the normal function of the rest of the lung. These pathways all converge on the Rel/Nf-kB complex, which is translocated to the nucleus to regulate gene transcription. Recent studies have shown that bronchial epithelial cells express nuclear RelA in response to allergen or lipopolysaccharide-induced inflammation (Poynter et al., 2002, 2003, 2004). This is evidence that Nf-kB can act directly within airway epithelial cells but the downstream targets are not yet known. Although the defects that we observed in the morphology of the airway epithelial cells of Eda mutant mice are probably a long-term consequence of life without submucosal glands, it is still possible that the Eda signaling pathway mediates some responses to injury or inflammation within the epithelium. Tabby and crinkled mutant mice represent useful models with which to explore these possibilities.
Wild-type animals were ICR outbred mice (Harlan Sprague-Dawley, Indianapolis, IN). EdaTa/Y adult males (four males from two litters with wild-type siblings) and Edarddcr/+ breeding pairs were purchased from the Jackson Laboratory. BMP4LacZ and Fgf10+/− heterozygotes as described, were maintained on an outbred genetic background (Lawson et al., 1999; Sekine et al., 1999).
In Situ Hybridization
For section in situ hybridization, wild-type trachea were fixed in 4% paraformaldehyde/phosphate buffered saline, dehydrated, and embedded in paraffin. Seven-micrometer sections were used for in situ hybridization with 35S-labeled sense and antisense probes. These were prepared from a 1.4-kb fragment of Edar (Laurikkala et al., 2001) using the Boehringer–Mannheim RNA labeling kit. Slides were exposed to emulsion for up to 10 days and counterstained with hematoxylin.
Bmp4LacZ heterozygotes were mated with ICR mice. Trachea were excised from the pups from P0 to P28 fixed in 4% paraformaldehyde/phosphate buffered saline (PBS), permeabilized, and stained for LacZ from overnight to 2 days. They were subsequently embedded in paraffin, sectioned at 7 μm, and counterstained with eosin.
Immunohistochemistry and Alcian Blue Histology
Seven-micrometer sections of paraffin-embedded tissue were dewaxed and rehydrated before antibody staining; immunohistochemical reactions were carried out in parallel with control reactions lacking primary antibodies. Primary antibodies used were mouse monoclonal anti–α-smooth muscle actin (Clone 1A4, Sigma, St. Louis, MO) 1:200 dilution after protease antigen retrieval, mouse monoclonal K14 antibody (Clone LL002, Neomarkers, Fremont, CA) 1:200 dilution after antigen retrieval by microwave treatment in citrate buffer, and rabbit anti-CGRP (Peninsula Laboratories, Inc., San Carlos, CA) 1:1,000 dilution. Primary antibodies were incubated at 4°C overnight. Secondary antibodies were goat anti-mouse Cy3 from Jackson ImmunoResearch (West Grove, PA) and biotinylated horse anti-mouse IgG (Vector Laboratories, Burlingame, CA). For the biotinylated antibody, the signal was amplified using the Vectastain Elite ABC kit and visualized with diaminobenzidine (DAB) substrate (Vector Laboratories). Sections were counterstained with 4′,6-diamidine-2-phenylidole-dihydrochloride (DAPI) or eosin.
Alcian blue staining for acid mucins was performed on 7-μm sections of paraffin-embedded tissue, which were dewaxed and rehydrated to water, incubated in Alcian blue pH 2.5 (1% Alcian blue in 3% acetic acid) for 5 min, washed in water, and counterstained with 0.1% Nuclear fast red for 5 min and washed in water. Sections were then dehydrated, cleared, and mounted in Permount (Fisher, Pittsburgh, PA).
To detect cell proliferation, BrdU (Amersham Biosciences, Piscataway, NJ) was diluted in PBS and injected intraperitoneally into pups at a dose of 10 μl per gram body weight. BrdU was detected using mouse monoclonal anti BrdU (Sigma-Aldrich, St. Louis, MO) with heating in dilute HCl, microwave treatment in citrate buffer and protease digestion for antigen retrieval. The Mouse on Mouse Kit and DAB (Vector Laboratories) were used to detect the antigen–antibody complexes.
Total RNA was extracted from the upper and lower trachea and the left lobe of the lung at P3, P9, and P49 using the Rneasy kit (QIAGEN, Inc., Valencia, CA). The cDNA was synthesized from 1 μg of total RNA using the Superscript First-Strand Synthesis Kit (Invitrogen, Carlsbad, CA). Thirty cycles of amplification were performed using the following primer pairs: β-actin, 5′-GTCGTACCACAGGCATTGTGATGG-3′ and 5′-GCAATGCCTGGGTACATGGTGG-3′; EDA, 5′-GAGAAGCAAAAGTGGTGAAG-3′ and 5′-GTTCATAGTGATGCGAGACC-3′; Edar, 5′-TGTGTATGCCAACGTGTGTG-3′ and 5′-TCCCTTCATTTGCCTAGGTG-3′; Xedar, 5′-CCCTCTACTGGACCTGAAAC-3′ and 5′-CAGGCAAACTCCACCTCATT-3′; TNFSRF19, 5′-TGTCACCACCCAGAGGATTC-3′ and 5′-CGTCCTTGTGCTGTGAAGAG-3′.
The authors thank Irma Thesleff for Eda probes; Eric Meyers, Matt Godderis, and Roger Ilgan for providing Fgf8LacZ, PtcLacZ, and Fgf10 heterozygous mice; and Scott Randell for generous discussion and advice.