GATA6 regulates differentiation of distal lung epithelium

Embryonic lung development in the mouse can be divided into four overlapping stages: embryonic (E9.5-E12.5), pseudoglandular (E12.5-E16.0), canalicular (E16.0-E17.5), and saccular (E17.5-birth) (for reviews, see Costa et al., 2001; Warburton et al., 2000). During embryonic development, the epithelium lining the airways differentiates from a uniform cell type to a diverse population of specialized epithelial cells that differentiate along a proximal-distal axis (Warburton et al., 2000; Weaver et al., 1999). Much of this differentiation occurs in the canalicular (E16.0-E17.5) and saccular (E17.5-birth) stages of lung development when normal proximal-distal patterning results in the development of specialized distal airway epithelium, which consists of type 1 and 2 alveolar epithelial cells (AEC-1 and -2) required for gas exchange and expression of surfactant proteins, respectively (for a review, see Warburton et al., 2000). Current paradigms in lung epithelial cell biology suggest that AEC-2 cells can self-renew and trans-differentiate into AEC-1 cells (Evans et al., 1975; Warburton et al., 2000). During later stages of lung development (E16.5 and later), genes encoding proteins such as the Clara cell-specific 10 kDa protein (CC10) and Foxj1 (previously known as human forkhead homologue 4), are expressed exclusively in the proximal epithelium, while genes encoding proteins including surfactant protein C (SP-C), GATA6, and Foxp2 are expressed in the distal but not proximal epithelium (Morrisey et al., 1996; Shu et al., 2001; Tichelaar et al., 1999b; Zhou et al., 1996b).

Several transcription factors have been implicated in regulating lung epithelial development including thyroid transcription factor-1 (TTF-1) (also known as Nkx2.1 and Titf1), several Fox gene products including Foxa2 and Foxj1, and the zinc finger transcription factor GATA6. TTF-1 is a member of the Nkx homeodomain gene family and has been shown to bind and trans-activate sequences present in many lung-specific promoters including the surfactant proteins A (SP-A/Sftpa), B (SP-B/Sftpb), and C (SP-C/Sftpc) and CC10 (Utg) (Bruno et al., 1995; Margana and Boggaram, 1997; Toonen et al., 1996). Mice harboring a null mutation in the TTF-1 gene exhibit severely attenuated lung epithelial development with a dramatic decrease in airway branching (Kimura et al., 1996). In addition, lung epithelial cells present in these mice lack expression of certain putative target genes such as SP-C suggesting that TTF-1 is an important and essential transcription factor for lung epithelial gene expression (Minoo et al., 1999). Members of the winged-helix/forkhead or Fox gene family are expressed in a wide range of cells in the lung including the distal epithelium (Foxp2 and Foxa2), proximal epithelium (Foxj1 and Foxa2), and mesenchyme (Foxf1 and Foxf2) and Fox DNA binding sites are located in the promoters and enhancers of several lung specific genes including CC10, TTF-1 and SP-B (Braun and Suske, 1998; Clevidence et al., 1994; Ikeda et al., 1996; Kaestner et al., 1994; Mahlapuu et al., 1998; Margana and Boggaram, 1997; Miura et al., 1998; Shu et al., 2001; Tichelaar et al., 1999b; Zhou et al., 1996b). Furthermore, gain-of-function and loss-of-function experiments have demonstrated the necessity of wild-type expression levels of some of these Fox genes during lung differentiation and development including Foxf1, Foxa2 and Foxj1 (Kalinichenko et al., 2001; Mahlapuu et al., 2001; Tichelaar et al., 1999a; Zhou et al., 1997).

The zinc finger transcription factor GATA6 has recently been implicated in regulating lung gene expression and development. GATA6 belongs to the GATA family of zinc finger transcriptional regulators, several of which have been shown to play important roles in tissue specification and gene expression (Simon, 1995). GATA6 is the only known GATA factor expressed in the distal epithelium of the developing lung where its expression is observed as early as E10.5 (Bruno et al., 2000; Keijzer et al., 2001; Morrisey et al., 1996; Morrisey et al., 1997). Binding sites for GATA6 are present in several lung-specific promoters and enhancers including SP-A, SP-C, and TTF-1 (Bruno et al., 2000; Shaw-White et al., 1999). In addition, GATA6 is able to trans-activate these promoters in non-lung cells suggesting a role in the transcriptional regulation of these genes (Bruno et al., 2000; Shaw-White et al., 1999). Previous experiments implicated an essential role for GATA6 in early lung epithelial development because Gata6^–/– embryonic stem (ES) cells were unable to contribute to the airway epithelium of Gata6^–/–/C57BL6 chimeric mice at E13.5 (Morrisey et al., 1998). However, a recent report suggests that under certain in vivo experimental conditions, Gata6^–/– ES cells can differentiate down the lung epithelial cell lineage pathway (Keijzer et al., 2001). These Gata6^–/–/C57BL6 chimeric lungs displayed defects in branching morphogenesis and down-regulation of putative target genes such as SP-C (Keijzer et al., 2001). However, the specific defects in epithelial cell differentiation were not reported. Therefore, much is still not known about the cell intrinsic function(s) of GATA6 transcriptional regulation in lung epithelium, which is important because of the expression of GATA6 in the adjacent mesenchyme of the lung (Bruno et al., 2000; Keijzer et al., 2001; Morrisey et al., 1996).

We have examined the role of GATA6 in distal lung epithelial development by expressing a GATA6/Engrailed dominant-negative fusion protein (GATA6/En) in transgenic mice using the well-characterized human 3.7 kb SP-C promoter which drives expression exclusively in distal lung epithelium (Koutsourakis et al., 2001; Lu et al., 2001; Weaver et al., 1999; Wert et al., 1993; Zhou et al., 1997). Several reports, from our laboratory and others, have shown that the GATA6/En fusion protein can repress GATA6 transcriptional activation (Bruno et al., 2000; Liang et al., 2001). In addition, similar Engrailed repressor domain/transcription factor fusion protein strategies have been used to perturb the function of c-myb and Nkx2.5 (Badiani et al., 1994; Fu et al., 1998). Although the GATA6/En dominant-negative may repress the function of most if not all GATA factors, GATA6 is the only known GATA factor expressed in distal lung epithelium (Morrisey et al., 1996; Morrisey et al., 1997). The morphological characteristics of the surfactant protein C/GATA6-Engrailed transgenic (SP-C/G6en) embryos are distinct from other lung transgenic models previously reported. SP-C/G6en transgenic mice exhibited a lack of squamous epithelium in the distal regions of the lung during late alveolar development indicating a defect in AEC differentiation. In particular, AEC-1 cells were absent in the distal airways of SP-C/G6en embryos as demonstrated by electron microscopy and attenuated aquaporin 5 gene expression. AEC-2 cells had been properly specified as shown by normal TTF-1 gene expression in these embryos. Furthermore, the presence of lamellar bodies, secreted surfactant and normal expression of SP-A suggests that AEC-2 differentiation had progressed beyond the pseudoglandular stage (E12.5-E16.0) of lung development in SP-C/G6en embryos. However, decreased endogenous SP-C expression as well as increased Foxp2 gene expression after E17.5 indicates that certain aspects of AEC-2 cell gene expression were perturbed. Proximal airway development was also disrupted as shown by a decrease in CC10- and Foxj1-positive proximal airway tubules. Finally, we show that binding sites for GATA factors are conserved in the aquaporin 5 promoters from mouse and rat and that GATA6 can trans-activate the mouse aquaporin 5 promoter, supporting a role for GATA function in AEC-1 cell gene expression. Taken together, these data implicate a role for GATA6 in the regulation of distal airway epithelial differentiation, leading to a disruption in AEC-1 differentiation and proximal airway development.

The GATA6/Engrailed chimeric protein was generated by cloning together, in frame, cDNAs encoding the DNA binding domain of the mouse GATA6 protein (aa233-532) and aa1-298 of the Drosophila Engrailed (En) transcriptional regulator, which has been shown to confer transcriptional repression when fused to heterologous DNA binding domains (Badiani et al., 1994; Fu et al., 1998). This fusion construct was further cloned into the human 3.7 kb SP-C transgenic construct with a modified multiple cloning site resulting in SP-C/G6en (Wert et al., 1993). The 298 amino acid Engrailed repressor domain containing a nuclear localization signal at the amino terminus (MKRKKK) was also cloned into the same SP-C transgenic vector resulting in SP-C/en. The resulting constructs, SP-C/G6en and SP-C/en, were digested with AscI to release the transgenes away from the backbone plasmid, and purified using Geneclean (Bio 101, Inc.). Purified transgenic DNA was injected into FVB/N fertilized oocytes and transferred into pseudopregnant females. For histological analysis, six independent founders at E17.5, five at E19.5, and two deceased pups just after birth (P0) were examined for the SP-C/G6en construct while five independent founder lines were examined at E19.5 for the SP-C/en construct. All transgenic embryos examined contained at least 15 copies of the transgene and the phenotypes reported were consistent in all of the founders tested.

E17.5 embryos were dissected from the uterus and fixed in 4% paraformaldehyde for 48 hours. The lungs from E19.5 embryos and P0 neonates were dissected and removed from the chest cavity and fixed for 24 hours in 4% paraformaldehyde. All tissue samples were dehydrated in a series of ethanol washes and embedded in paraffin blocks. Hematoxylin and Eosin (H+E) staining was performed as previously described (Soudais et al., 1995). The in situ probes for Gata6, SV40 poly(A)+ intron, engrailed, SP-A, SP-C, Foxp2, CC10 and Foxj1 have been described (Lu et al., 2001; Morrisey et al., 1996; Shu et al., 2001). The in situ probe for mouse aquaporin 5 was generated by RT-PCR from adult mouse lung mRNA using the following primers: sense 5′-GCTGTGGTCAAAGGCACATATGAG-3′, antisense w/T7 site 5′-TGTAATACGACTCACTATAGGGCGACTGAACTGCTGTGAGCTTGCAC-3′. Radioactive in situ hybridization was performed as previously described (Soudais et al., 1995). To quantitate the numbers of CC10-positive airways in wild-type and SP-C/G6en mice, five sections corresponding to the same region of the left lung from three wild-type and three SP-C/G6en transgenic littermates at E17.5 and E19.5 were probed with the CC10 riboprobe and the resulting positive airways were counted and analyzed using the NIH Image 1.62 software.

Non-radioactive in situ hybridization was performed by probing deparaffinized tissue sections with a digoxigenin-labeled riboprobe and developing and immunologically detecting the positive signal with an anti-digoxigenin monoclonal antibody (Roche Biochemicals, Inc.). Non-radioactive in situ hybridization was used to detect SP-C gene expression because the signal from radioactive in situ hybridization was too intense for high resolution analysis of expression. Periodic acid-Schiff staining (PAS) was performed by soaking deparaffinized embryo sections in 0.5% Periodic acid for 5 minutes, rinsing with dH₂O and then staining in Schiff’s reagent for 60 seconds. PAS-stained slides were then counterstained with Hematoxylin. For amylase treatment, slides were predigested in diastase (with α-amylase) solution for 15 minutes prior to PAS staining. Further details of all histological procedures can be obtained at the University of Pennsylvania Molecular Cardiology Research Center Web page-Histology Core Facility Protocols (http://www.med.upenn.edu/mcrc/histology/histologyhome.html).

Electron microscopy procedures were performed as previously described (Soudais et al., 1995). Briefly, lung tissue was dissected from E19.5 SP-C/G6en, SP-C/en and wild-type littermates and fixed in 2% glutaraldehyde with 0.1 M sodium cacodylate pH 7.4 for 48 hours at 4°C. Samples were then incubated with 2% osmium tetroxide and 0.1 M sodium cacodylate pH 7.4 for 1 hour at 4°C. Ultrathin sections were stained with lead citrate and uranyl acetate and viewed on a JEM 1010 microscope. Digital images were captured on a Hamamatsu HamC4742-95-12 CCD camera using AMT Advantage software.

SP-C/G6en transgenic mice were harvested at E19.5 and the whole lung was dissected from the fetus and frozen at –80°C. Two genotype positive and wild-type lung samples from the same litter were homogenized in Trizol and total RNA was extracted according to the manufacturers protocol (Invitrogen, Inc.). 10 μg of total RNA was resolved on a 1.5% formaldehyde agarose gel and blotted to a Hybond-N membrane (Amersham, Inc.). Generation of radiolabeled probes and hybridization was performed as described previously (Morrisey et al., 1996). The DNA fragments used to generate probes for aquaporin 5 (Aqp5), Foxp2 and engrailed were the same as described for the in situ hybridization analysis.

The –1437 bp mouse Aqp5 promoter was amplified from mouse genomic DNA and subcloned into the MluI/XhoI sites of pGL3basic (Promega) to generate pGL3/Aqp5-1.4 using the following primers: sense 5′-CACACGCGTAAACCTAGAAGGTCCTCCCTCC-3′, antisense 5′-CACCTCGAGTTCTGCGAGACGTGCGGTGCCC-3′. A truncated version lacking the four potential GATA6 binding sites (pGL3/Aqp5-0.5) was generated by PCR from pGL3/Aqp5-1.4kb using the same antisense primer described above and the following sense primer: 5′-CACACGCGTGGCGCTGTCCTCAGAAACTCATC-3′. The fidelity of both constructs was verified by DNA sequence analysis. These reporter constructs (0.5 μg) were co-transfected into NIH-3T3 cells along with the pcDNA3G6 construct (Morrisey et al., 1996) (2.5 μg) containing the entire open reading frame of the mouse Gata6 cDNA using Fugene 6 (Roche Biochemicals, Inc.). To test whether the GATA6 activation of the Aqp5 promoter could be repressed with the GATA6/Engrailed fusion protein, pcDNA3 plasmids containing either the GATA6/Engrailed fusion protein (pcDNA3G6/En), the Engrailed repressor domain (pcDNA3En) or a mutant GATA6/Engrailed fusion protein (pcDNA3G6mut/En), which contains mutations changing aa293-294 of GATA6 cysteine-alanine to serine-arginine (eliminating DNA binding), were transfected along with the pGL3/Aqp5-1.4kb reporter plasmid and the pcDNA3G6 expression plasmid (1 μg). All transfections contained 0.5 μg of the pMSVβgal vector to control for transfection efficiency. Forty-eight hours after transfection, cells were harvested and luciferase and β-galactosidase assays were performed using commercially available kits (Promega).

To determine the role of GATA6 during lung epithelial development, we generated transgenic mice using the human 3.7 kb SP-C promoter to express a fusion protein consisting of the DNA binding domain of the mouse GATA6 protein (aa233-352) fused to the well-characterized repression domain of the Drosophila Engrailed protein (aa1-298) followed by the SV40 intron/poly(A)+ sequence (Fig. 1A). In addition, a construct consisting of the human SP-C promoter driving expression of the Drosophila Engrailed repressor domain (aa1-298) with a nuclear localization signal was also injected and used as a control for specificity of the GATA6/Engrailed fusion protein (Fig. 1B). The human SP-C promoter was chosen for its ability to restrict high level expression of transgenes to the distal airway epithelium in mice (Wert et al., 1993). Initially, we attempted to generate adult founder transgenic lines with the SP-C/G6en construct. However, 121 potential F0 founders were genotyped at 2 weeks of age and no SP-C/G6en-positive animals were obtained (data not shown). These data suggest that expression of the GATA6/Engrailed fusion protein in the distal epithelium of the lung is lethal in postnatal mice. Therefore, six independent F0 founder embryos were harvested at E17.5 and five at E19.5. In addition, two transgene-positive P0 neonates that died soon after birth were also analyzed. SP-C/G6en embyros harvested at E17.5, E19.5 and P0 displayed high levels of transgene expression in the distal airway epithelium as determined by in situ hybridization using probes that recognize the SV40 poly(A)+ sequence or the engrailed repressor domain (Fig. 2 and data not shown). Of note, the expression pattern of the SP-C/en transgene was similar to that of the SP-C/G6en transgene (Fig. 2).

At E17.5, wild-type distal airway epithelium differentiates to assume a squamous morphology as a result of the differentiation of AEC-1 cells from AEC-2 cells (Fig. 3A,C). The lungs of SP-C/G6en embryos were approximately equal in size to wild-type littermates at E17.5 but the distal airways were dilated and lacked squamous epithelium (Fig. 3B,D). Instead, the distal airways of E17.5 SP-C/G6en embryos were lined completely with thick, cuboidal AEC-2-like cells (Fig. 3D, yellow arrowhead). At E19.5 and P0, the number of interalveolar septa was decreased with thicker mesenchyme between the distal airways in SP-C/G6en embryos and neonates suggesting a defect in distal airway branching or delayed maturation (Fig. 3F,H,J,L). In contrast, at E19.5, the SP-C/en control transgenic embryos displayed normal morphology in the distal epithelium of the lung (Fig. 3N,P). All other tissues appeared normal histologically, consistent with the tissue-restricted expression pattern of the human SP-C promoter (Koutsourakis et al., 2001; Lu et al., 2001; Weaver et al., 1999; Wert et al., 1993; Zhou et al., 1996a; Zhou et al., 1997) and data not shown). These data suggest that distal airway epithelial cell differentiation and development is specifically disrupted in SP-C/G6en mice.

To examine the ultrastructural morphology of the AECs in SP-C/G6en embryos, transmission electron microscopy was performed on SP-C/G6en transgenic and non-transgenic littermates and SP-C/en transgenic embryos at E19.5. Normal distal epithelium of the lung at this time point is lined with squamous (flattened) AEC-1 cells and cuboidal AEC-2 cells with AEC-1 cells covering a majority of the surface area in the airways (Ten Have-Opbroek, 1991). Electron micrographs show normal morphology in the lungs of wild-type and SP-C/en embryos at E19.5 with squamous AEC-1 cells covering most of the airway tubules (Fig. 4A,B,G,H). However, SP-C/G6en embryos exhibit a deficiency in these squamous AEC-1 cells when compared to non-transgenic littermates and SP-C/en mice (Fig. 4C-F). Instead, the distal airways in SP-C/G6en embryos were populated almost completely with large cuboidal AEC-2-like cells that contained large vacuoles reminiscent of the glycogen storage vesicles seen in immature AEC-2 cells during early lung development (Fig. 4D,E arrowheads and (Ten Have-Opbroek, 1991). To confirm that these large vacuoles represented an increase in glycogen content in these cells, Periodic acid-Schiff (PAS) staining was performed on histological sections from transgenic and wild-type littermates at E19.5. As shown in Fig. 5, SP-C/G6en mice exhibit an increase in PAS staining in the distal airway epithelium indicating an increase in glycogen content (Fig. 5B). Pretreatment with amylase eliminated the staining, indicating the specificity of the staining for glycogen (Fig. 5C).

The presence of lamellar bodies in many of these abnormal distal airway cells in SP-C/G6en embryos confirms their identity as AEC-2 cells and also suggests that AEC-2 cell differentiation had progressed beyond the pseudoglandular stage (Fig. 4F, black arrow). The presence of secreted surfactant in the distal airways of SP-C/G6en mice also supports these findings (Fig. 4F, white arrow). In addition, pulmonary vascular development was normal in SP-C/G6en embryos as demonstrated by a normal capillary plexus closely adjoining the distal airways (Fig. 4C-E). Together, these data indicate that SP-C/G6en mice have a specific disruption in AEC differentiation and development resulting in a lack of squamous AEC-1 cells in the distal airways.

To further investigate pulmonary epithelial cell differentiation in SP-C/G6en embryos, in situ hybridization utilizing gene- specific riboprobes for SP-C, a marker for AEC-2 cells, and TTF-1 an important transcription factor expressed in both AEC-2 cells and in proximal Clara epithelial cells, was performed (Glasser et al., 1991; Wert et al., 1993; Zhou et al., 1996b). Both of these genes have been used previously as specific markers of lung epithelial cell specification (Zhou et al., 1996b). TTF-1 null embryos die during gestation and exhibit severe defects in lung epithelial development and lack SP-C expression (Minoo et al., 1999). In addition, the mouse TTF-1 promoter contains an important GATA factor DNA binding site suggesting that GATA factors may regulate its expression (Shaw-White et al., 1999). TTF-1 expression was observed in both wild-type and SP-C/G6en lung epithelium at E17.5 and E19.5 (Fig. 6A-D). No differences were observed in the level or pattern of TTF-1 expression between wild-type and transgenic littermates (Fig. 6A-D). At E17.5, expression of endogenous SP-C was observed throughout the distal airway epithelium of wild-type embryos (Fig. 6E,G). Interestingly, endogenous SP-C expression was not observed in the thick cuboidal epithelium where the SP-C/G6en transgene was expressed (Fig. 6H, bracket). This observation was consistent in all five SP-C/G6en embryos examined. However, endogenous SP-C was expressed in regions of the distal lung epithelium that lacked expression of the SP-C/G6en transgene (Fig. 6F,H arrowheads). These data suggest that pulmonary epithelium has been specified but expression of endogenous SP-C is attenuated in SP-C/G6en embryos. However, attenuated endogenous SP-C expression in SP-C/G6en embryos is not due to the loss of TTF-1, which is expressed at wild-type levels.

Ultrastructural analysis suggested that distal airway epithelium in SP-C/G6en embryos exhibit some characteristics of late epithelial cell differentiation such as lamellar body formation and secreted surfactant. However, attenuated endogenous SP-C expression also suggests that early AEC-2 marker gene expression had been disrupted in SP-C/G6en embryos. To further delineate how far distal airway epithelial cell differentiation had progressed, in situ hybridization was performed with a gene-specific riboprobe for SP-A. SP-A expression is initiated in the early canalicular stage of lung development (E16.0) and as such is an excellent marker gene to detect whether SP-C/G6en embryonic distal airway epithelium had progressed to the later stages of lung development (Mendelson, 2000). At E17.5, SP-A expression is observed in a punctate pattern in the distal airway epithelium of both wild-type and SP-C/G6en embryos where both the level and pattern of expression were the same (Fig. 7). These data, combined with the ultrastructural analysis showing lamellar body formation, indicate that distal airway epithelium in SP-C/G6en embryos had differentiated to the later stages (E16.0 and later) of lung epithelial development.

Although expression of GATA6 is not observed in proximal airway epithelium during development, previous studies with chimeric mice generated using Gata6^–/– ES cells suggested that proximal cell gene expression is altered upon disruption of GATA6 expression (Keijzer et al., 2001). To determine whether SP-C/G6en mice exhibited defects in proximal epithelial cell development and gene expression, in situ hybridization was performed on sections from E17.5 and E19.5 transgenic embryos using probes specific for the mouse Foxj1 and CC10 mRNAs. Foxj1 is a winged-helix transcription factor expressed specifically by ciliated epithelium of the upper respiratory tract during lung development in mice, whereas CC10 is a secreted glycoprotein expressed specifically in non-ciliated Clara epithelial cells of the upper respiratory tract (Tichelaar et al., 1999b; Zhou et al., 1996b). Thus, their expression can be used to identify two specific proximal airway epithelial cell lineages. As has been previously reported, Foxj1 and CC10 expression is observed in the epithelium of the major proximal airways in the lungs of non-transgenic mice (Fig. 8A,C). Although the levels of Foxj1 and CC10 expression appeared normal in SP-C/G6en mice, the number of individual airways positive for Foxj1 and CC10 expression was reduced in all of the independent SP-C/G6en founders at E17.5 and E19.5 (Fig. 8B,D,F). This defect is most obvious in the significant reduction in the number of smaller CC10- and Foxj1-positive airways whereas the number of the largest most proximal CC10- and Foxj1-positive airways remains relatively the same (Fig. 8G,H). Quantification of the number of CC10-positive airways shows a 55%±7.2% decrease at E17.5 and a 39%±4.1% decrease at E19.5 in SP-C/G6en mice. Examination of H+E stained slides from E17.5 wild-type and SP-C/G6en embryos confirms that the reduced number of airways expressing CC10 and Foxj1 correlates with a decrease in proximal airways containing convoluted and columnar epithelium in SP-C/G6en mice (Fig. 8G,H). Together, these data suggest that expression of the SP-C/G6en transgene results in a reduction in proximal airway development.

Aquaporin 5 (Aqp5) is a water channel gene whose expression is restricted to AEC-1 cells during lung development and is one of the few AEC-1 marker genes known (Krane et al., 1999; Kreda et al., 2001; Lee et al., 1997). Foxp2 is a winged-helix/forkhead transcription factor gene that is expressed in the distal airway epithelium of the developing lung (Shu et al., 2001). To investigate whether expression of either of these genes was disrupted in the lungs of SP-C/G6en embryos, in situ hybridization was performed using gene-specific riboprobes for Aqp5 and Foxp2.

At E17.5, Aqp5 gene expression was observed at low levels in the distal airways of wild-type lungs as has been previously demonstrated (Fig. 9A) (Kreda et al., 2001). However, Aqp5 expression was not observed in the distal airways in any of the E17.5 SP-C/G6en embryos (Fig. 9B). By E19.5, expression was very robust in the distal airways of wild-type embryos but severely attenuated in all E19.5 SP-C/G6en embryos (Fig. 9C,D). SP-C/en embryos showed no difference in Aqp5 gene expression from their wild-type littermates (Fig. 9E,F). These data suggest that aquaporin-5 gene expression is significantly down-regulated in SP-C/G6en embryos.

At E17.5 and E19.5, Foxp2 is expressed in the distal airway epithelium of wild-type embryos in a pattern similar to that observed for SP-C (Fig. 9G,I,K). In addition, Foxp2 expression is down-regulated in wild-type embryos from E17.5 to E19.5 (Fig. 9G,K). However, in SP-C/G6en transgenic mice, Foxp2 expression is not down-regulated at E19.5 but instead remains robust and is greater than in E19.5 wild-type littermates (Fig. 9K,L). This difference was observed in all of the E19.5 SP-C/G6en embryos analyzed. Of note, the pattern of Foxp2 expression in SP-C/G6en embryos closely resembles that of the transgene as determined by the SV40 poly(A)+ in situ probe, clearly showing the dilated nature of the distal airways at E17.5 and E19.5 in SP-C/G6en embryos (compare Fig. 9H,J with Fig. 2B).

To confirm the in situ hybridization results showing a decreased level of Aqp5 expression and an increased level of Foxp2 expression in the airways of SP-C/G6en embryos, equal amounts of total RNA from the lungs of two E19.5 SP-C/G6en transgenic embryos and two wild-type littermates was analyzed by northern blotting to quantitate the changes in Foxp2 and Aqp5 expression. Both of the independent F0 SP-C/G6en founder embryos expressed high levels of the SP-C/G6en transgene (Fig. 10). Phosphoimager analysis shows that Aqp5 gene expression is decreased an average of 7.8-fold while Foxp2 gene expression is increased by 3.5-fold in SP-C/G6en transgenic embryos at E19.5 (Fig. 10). These data support the in situ hybridization analysis showing a significant decrease in Aqp5 and increase in Foxp2 expression in the lungs of SP-C/G6en embryos.

Previous reports have implicated GATA6, as well as other tissue restricted transcription factors such as TTF-1 and Foxa2, in the regulation of AEC-2 cell-specific gene expression (Bohinski et al., 1994; Bruno et al., 1995; Bruno et al., 2000; Margana and Boggaram, 1997; Shaw-White et al., 1999). In contrast, AEC-1 cell-specific gene expression has not been extensively studied and few cell-specific marker genes or transcriptional regulators have been characterized. The disruption of AEC-1 cell differentiation in SP-C/G6en mice, as demonstrated by ultrastructural analysis and reduced Aqp5 gene expression, suggested that GATA6 may play a role not only in the differentiation of AEC-1 cells from AEC-2 cells but also in the regulation of AEC-1 cell-specific gene expression. In addition, expression of Aqp5 is induced in AEC-2 cells in response to lung injury or remodeling (Tichelaar et al., 2000). These data suggest that AEC-2 and AEC-1 cells may share some transcriptional regulators and mechanisms for cell-specific gene expression. Isolation and sequencing of the 1.4 kb promoter for the AEC-1 restricted mouse Aqp5 revealed four putative GATA6 binding sites, two of which are conserved in the rat Aqp5 promoter (Fig. 11A) (Krane et al., 1999). To determine whether GATA6 could activate the mouse Aqp5 promoter, co-transfection studies were performed using two different regions of this promoter and an expression plasmid containing the GATA cDNA. The pGL3/Aqp5/–1.4 plasmid contains all four of the identified GATA6 DNA binding sites while the truncated pGL3/Aqp5/–0.5 plasmid lacks these four sites (Fig. 11A). Co-transfection of the GATA6 expression plasmid dramatically stimulated the pGL3/Aqp5/–1.4 plasmid by more than 25-fold, while deletion of the four GATA6 binding sites in the pGL3/Aqp5/–0.5 plasmid resulted in a 68% decrease in this GATA6-dependent trans-activation (Fig. 11B). The residual trans-activation of the pGL3/Aqp5/–0.5 plasmid is likely due to GATA6 interactions with ubiquitous transcription factors such as Sp-1 or binding of GATA6 to cryptic GATA DNA binding sites within the backbone of the pGL3basic reporter plasmid (Merika and Orkin, 1995). To verify that the GATA6/Engrailed fusion protein could specifically repress the Aqp5 promoter, the pGL3/Aqp5/–1.4 plasmid was transfected into NIH-3T3 cells along with a constant amount of the GATA6 expression plasmid and with increasing amounts of either a GATA6/Engrailed fusion protein expression plasmid, an expression plasmid encoding a mutant GATA6/Engrailed fusion protein with point mutations in the conserved carboxy-terminal zinc-finger that abrogate Gata6 DNA binding, or an expression plasmid encoding the Engrailed repression domain used in the SP-C/en transgenic mice. As shown in Fig. 11C, only the GATA6/Engrailed fusion protein expression plasmid was capable of repressing GATA6-dependent trans-activation of the Aqp5 promoter. These data show that GATA6 can directly activate mouse Aqp5 gene expression in vitro.

From the data presented above, we propose a model wherein GATA6 regulates late distal epithelial differentiation processes in the lung that lead to the generation of AEC-1 cells (Fig. 12). These cells are required to generate a thin, diffusible membrane for efficient gas exchange between the airway lumen and the pulmonary vasculature. Little is known about the differentiation of AEC-2 to AEC-1 cells in the lung. Previous reports have shown that AEC-2 cells have the ability to self-renew in addition to trans-differentiating into AEC-1 cells (Evans et al., 1975). However, most of these data have been derived from in vitro culture systems and therefore may not accurately reflect in vivo developmental processes. Although our data suggest a necessary role for GATA6 in the differentiation of AEC-1 cells, a recent study suggests that over-expression of GATA6 in distal airway epithelium does not result in increased AEC-1 cell differentiation (Koutsourakis et al., 2001). Thus, GATA6 may be necessary, but is not sufficient to promote AEC-1 cell differentiation.

SP-C/G6en embryos are distinct from other lung transgenic phenotypes in that they exhibit a specific loss of AEC-1 cells, as shown by ultrastructural analysis and attenuated Aqp5 gene expression. The study of AEC-1 cell differentiation has been hampered by knowledge of only a few lineage-restricted marker genes. The best analyzed of these are Aqp5, a water channel gene expressed in lung AEC-1 cells as well as in tracheal and nasal epithelium, and a gene of unknown function called T1α (Lee et al., 1997; Nielsen et al., 1997; Ramirez et al., 1999; Ramirez et al., 1997). Both of these genes have been reported to be expressed exclusively in AEC-1 cells in the lung (Lee et al., 1997; Nielsen et al., 1997; Ramirez et al., 1999; Ramirez et al., 1997). However, the expression of T1α in early lung epithelium (E11.5) suggests that it is not specific to AEC-1 cells since these cells do not appear until E17.5 (Ramirez et al., 1999). Also, Aqp5 expression can be induced in AEC-2 cells upon remodeling of the epithelium in adult mice (Tichelaar et al., 2000). In light of these data, Aqp5, which is first expressed in lung development at approximately E17.5 , is an excellent AEC-1 marker gene because its expression pattern and timing are identical to the appearance of AEC-1 cells during lung development (Lee et al., 1997). Thus, attenuated Aqp5 gene expression supports the concept that AEC-1 differentiation is blocked in SP-C/G6en embryos.

Other loss-of-function experiments have revealed phenotypes that are related but distinct to that observed in the lungs of SP-C/G6en embryos. Parathyroid hormone-related peptide (PTHrP) mice also lack Aqp5 expression and die at birth from respiratory failure (Karaplis et al., 1994; Ramirez et al., 2000). However, these mice appear to contain squamous epithelium in their lungs (Karaplis et al., 1994; Ramirez et al., 2000). In addition, since mice homozygous for an Aqp5 null allele do not exhibit a postnatal lung phenotype, the lack of Aqp5 in the PTHrP mice would not account for perinatal lethality (Ma et al., 1999). Mouse lungs containing a null mutation in the tumor necrosis factor-α converting enzyme (TACE) also exhibit attenuated SP-C and Aqp5 gene expression and display dilated airways similar to those observed in SP-C/G6en embryos (Zhao et al., 2001). As with SP-C/G6en embryos, TACE mice die soon after birth as a result of respiratory failure (Zhao et al., 2001). However, Tace^–/– mice have defects in lung vascular development which were not observed in SP-C/G6en embryos (Zhao et al., 2001). Therefore, the SP-C/G6en transgenic embryos have a unique phenotype that correlates well with a specific loss in AEC-1 cell differentiation during late lung development. This may contribute to the lethality observed in the transgenic mice because of the increased diffusion distance for gas exchange caused by the thicker AEC-2 cells lining the airways. Although AEC-2 differentiation was disrupted as shown by reduced SP-C expression, high levels of transgene expression are not observed with the human SP-C promoter until after E15.5 (Wert et al., 1993). Thus, the SP-C/G6en transgene is not likely to have affected GATA6 function until late lung development, further supporting a defect in the AEC-2 to AEC-1 cell differentiation process.

GATA6 is the only GATA factor expressed in distal lung epithelium and its ability to trans-activate the mouse aquaporin-5 promoter through GATA factor DNA binding sites conserved in the mouse and rat promoters, further reinforces a model wherein GATA6 regulates AEC-1 differentiation and gene expression at the transcriptional level. These data suggest that GATA6 can regulate AEC-1 cell-specific gene expression in addition to the differentiation process leading to AEC-1 cell formation. Indeed, one of the hallmark functions of GATA factors is their ability to regulate cell-specific gene expression programs crucial for tissue-specific differentiation processes. For instance, GATA1 is absolutely required for terminal erythrocyte differentiation while GATA2 and GATA3 are essential for hematopoeitic stem cell and T-cell differentiation, respectively (Pevny et al., 1991; Ting et al., 1996; Tsai et al., 1994). Our data suggests that the AEC-2 to AEC-1 cell differentiation process and AEC-1 cell-specific gene expression may be added to the repertoire of the roles performed by GATA6 during development.

Histological and gene expression data using the proximal airway epithelial marker genes Foxj1 and CC10 suggests a defect in late stage proximal airway development in SP-C/G6en mice as exhibited by reduced numbers of proximal airway tubules. A previously reported transgenic model using the SP-C promoter to over-express a constitutively active form of the signaling protein TGFβ1 resulted in decreased expression of the proximal epithelial gene CC10 (Zhou et al., 1996a). This result led the authors to conclude that the lung epithelium of SP-C/Tgfβ1 mice was arrested in the pseudoglandular stage of development. In addition, over-expression of Foxa2 (HNF3β) in the distal airway epithelium led to decreased levels of CC10 expression in lung epithelium (Zhou et al., 1997). Because the SP-C promoter drives expression in the distal and not proximal airway epithelium, these reports suggest that disruption of distal airway development and differentiation results in aberrant proximal airway development. In SP-C/G6en mice, normal expression levels of CC10 and Foxj1 in the upper airways indicates that lung development has not been arrested in the pseudoglandular stage of development. In contrast, the number of proximal airways is decreased in these embryos compared to wild-type littermates. In light of these results, it is interesting to note that a previous model of lung development proposed that proximal airway epithelium develops from more distal epithelial cell types through combined proliferation and differentiation events (Weaver et al., 1999). Because we have not observed any differences in epithelial cell proliferation or programmed cell death in SP-C/G6en mice as compared to wild-type or SP-C/en mice (data not shown), the decreased numbers of CC10-positive airways probably result from a disruption of distal epithelial cell differentiation in SP-C/G6en mice. Low levels of transgene expression in some part of the proximal airways may also confer this phenotype. However, endogenous GATA6 is not expressed in proximal airway epithelium and the SP-C promoter does not drive expression in these cells (Morrisey et al., 1997; Weaver et al., 1999; Wert et al., 1993). Therefore, our results correlate with those observed in a previous study showing that CC10 expression was attenuated in chimeric lung tissue made from Gata6^–/– ES cells (Keijzer et al., 2001). However, in contrast to that report, our data suggests that reduced CC10 expression is due to a reduction in proximal airway development and not to direct control of CC10 expression by GATA6, indicating a role for GATA6 in distal-proximal airway epithelial differentiation.

The relatively normal expression of SP-A suggests that SP-C/G6en embryos are not arrested at the pseudoglandular stage of development as has been observed in other transgenic models of disrupted lung epithelial differentiation (Zhou et al., 1996a; Zhou et al., 1997). In contrast, Foxp2 expression was elevated during late gestation and SP-C expression was attenuated, indicating aberrant AEC-2 development. These data suggest a model in which GATA6 plays an important role in certain aspects of AEC-2 development, in particular the regulation of SP-C at the transcriptional level (Fig. 12). These results also support a hypothesis wherein inhibition of GATA6 function leads to a broader and earlier inhibition of lung epithelial cell differentiation that results in the later observed defects in proximal airway development and AEC-1 cell differentiation. Although the human SP-C promoter does not produce high levels of transgene expression until E15.5 and later (Wert et al., 1993), tissue-restricted inactivation of the mouse Gata6 gene will be required to discern the difference between an early and broad role and/or a later and more specific role for GATA6 in lung epithelial differentiation as is suggested by our results.

SP-C expression is initiated as early as E11.0 in the mouse where it is restricted to the distal tips of developing airway epithelium (Wert et al., 1993; Zhou et al., 1996b). Later in development SP-C expression is restricted to AEC-2 cells in the lung (Zhou et al., 1996b). The transcriptional regulation of SP-C has been previously examined and several lung restricted transcription factors have been implicated in its regulation including TTF-1 and Foxa2 (Glasser et al., 2000; Glasser et al., 1991; Wert et al., 1993). A recent study has shown that SP-C expression was down-regulated in chimeric lung tissue derived from Gata6^–/– ES cells (Keijzer et al., 2001) and our results corroborate this finding. These observations are supported by the presence of conserved GATA factor DNA binding sites in the proximal promoter region of both mouse and human SP-C genes, which can bind GATA6 and mediate GATA6-dependent trans-activation of these promoters (data not shown). Another study showed that over-expression of GATA6 in distal lung epithelium resulted in decreased levels of SP-C expression (Koutsourakis et al., 2001). Together, these data indicate that wild-type GATA6 activity is essential for SP-C expression.

Another transcription factor that has been shown to be essential for SP-C expression in a loss-of-function experiment is TTF-1 (Minoo et al., 1999). TTF-1 null mice exhibit arrested lung development and lack expression of SP-C (Kimura et al., 1996; Minoo et al., 1999). Interestingly, TTF-1 (Nkx2.1) belongs to the same family of homeodomain proteins as Nkx2.5, an important regulator of cardiac gene expression, which has been shown to interact and synergize with GATA4 to activate cardiac-specific gene expression (Durocher et al., 1997; Sepulveda et al., 1998). It will be interesting to determine whether GATA6 and TTF-1 are also capable of physically interacting and synergistically regulating lung epithelial gene expression.

SP-C/G6en embryos also exhibited a reduction in airway branching morphogenesis as shown by the dilated distal airways containing fewer interalveolar septa at E17.5 and E19.5. Branching morphogenesis is a complex process that is regulated by both autocrine/paracrine signaling pathways and regulation of gene expression at the transcriptional level (for reviews, see Hogan et al., 1997; Metzger and Krasnow, 1999). Thus, this process is extremely sensitive to both cell intrinsic and extrinsic perturbations. Many in vitro and in vivo loss-of-function models of lung development involve some level of disrupted branching morphogenesis (Arman et al., 1999; Bellusci et al., 1997; Miettinen et al., 1997; Pepicelli et al., 1998; Serra et al., 1994; Volpe et al., 2000). These findings are not surprising since airway branching and epithelial cell differentiation and development are closely linked (Warburton et al., 2000). Therefore, GATA6 may regulate lung branching morphogenesis directly through pertinent signaling pathways, but it is equally possible that disrupted branching observed in SP-C/G6en embryos is a secondary effect of disrupted distal and proximal airway differentiation.

In summary, SP-C/G6en transgenic embryos exhibit a unique phenotype during late lung epithelial development, leading to a failure of AEC-1 cell differentiation and disruption in proximal airway development. The SP-C/G6en transgenic model provides key information on the role of GATA6 during lung development, both during late processes (AEC-1 differentiation) and earlier epithelial development (SP-C gene regulation). Disrupted proximal airway development in these embryos also suggests that GATA6 plays an important role in the distal-proximal differentiation of lung airway epithelium.

The authors wish to thank Jon Epstein, Phil Ballard and Celeste Simon for their helpful comments on the manuscript. This work was supported, in part, by a NIH-NHLBI grant (HL-64632) to E. E. M.