TTF-1 Phosphorylation Is Required for Peripheral Lung Morphogenesis, Perinatal Survival, and Tissue-specific Gene Expression*

Thyroid transcription factor-1 (TTF-1) is a 43-kDa, phosphorylated member of the Nkx2 family of homeodo-main-containing proteins expressed selectively in lung, thyroid, and the central nervous system. To assess the role of TTF-1 and its phosphorylation during lung morphogenesis, mice bearing a mutant allele, in which seven serine phosphorylation sites were mutated, Titf1 PM/PM , were generated by homologous recombination. Although heterozygous Titf1 PM/ (cid:1) mice were unaf-fected, homozygous Titf1 PM/PM mice died immediately following birth. In contrast to Titf1 null mutant mice, which lack peripheral lung tissues, bronchiolar and peripheral acinar components of the lung were present in the Titf1 PM/PM mice. Although lobulation and early branching morphogenesis were maintained in the mutant mice, abnormalities in acinar tubules and pulmonary hypoplasia indicated defects in lung morphogenesis later in development. Although TTF-1 PM protein was readily detected within the nuclei of pulmonary epithelial cells at sites and abundance consistent with that of endogenous TTF-1, analyzed using GeneSpring 5.0 (Silicon Genetics, Inc.), JMP4 (SAS Institute, Inc.), and Spotfire 7.12 (Spotfire, Inc.) software. A total of 18 chips were used in this experiment. Hybridization data ( (cid:3) 216,000 data points) were sequentially subjected to normalization, transformation, filtering, clustering, and function classification as pre- viously described (25). Data were normalized to enable the direct com-parisons across chips and across genes. Statistical differences between Titf1 PM/PM and control littermates were identified by distribution analysis and Welch’s t test at p value (cid:3) 0.05. Variations related to process- ing and biological replicates were calculated and separated from the candidate genes to identify primary genotype response. Fold changes were calculated for each gene against its specific control to determine relative gene expression. Additional filters included minimal absolute intensity (cid:4) 30, a minimum of 4 detectable judgments for A-set (12 chips) and 2 detectable judgments for B and C-sets (6 chips), and coefficient of variation among replicates (cid:3) 50%. Genes with average fold changes (cid:4) 2 and genes that were cross-validated via different probes on the same chip, or the same gene on different chips (A, B, and C sets), were prioritized. Differentially expressed genes were classified into functional categories based on gene ontology definitions. To determine representation of functional categories in the selected gene list, the binomial probability was calculated for each category using corresponding U74Av2 genome as the reference dataset. Hierarchal cluster- ing was applied to visualize and further group the selected genes based on their expression similarity. Pearson correlation was used for simi- larity measure. Clusters were constructed by the unweighted pair group method with arithmetic mean. and human promoter sequences downloaded from Harvard-Lipper Center for Genetics Upstream sequences (1 kb) of selected genes were retrieved from above promoter sequence data base and searched for potential TTF-1 regulatory sites. A 15-bp sequence NNWCTCAAGTRY-WNN from Genomatix

Lung formation begins with the outpouching of endodermal tissues from the laryngeal-tracheal-esophageal groove at embryonic day (E) 1 9 -9.5 in the mouse embryo. Epithelial lined tubules invade the splanchnic mesenchyme and undergo branching morphogenesis to form bronchi, bronchioles, and alveolar regions of the adult lung. Thyroid transcription factor-1 (genomic designation Titf1; also termed TEBP, or Nkx2.1) is a phosphorylated, homeodomain-containing, nuclear transcription factor expressed in respiratory epithelial cells of the developing lung, thyroid, and central nervous system (1). Although the trachea and main stem bronchi were formed in Titf1 null mutant mice, peripheral components of the lung, including bronchioles, acinar ducts, and respiratory saccules were lacking in these mice, causing death at the time of birth (2). Likewise, expression of surfactant proteins was lacking in the Titf1 null mice (3,4). TTF-1 is critical for formation of the lung and thyroid, regulating distinct subsets of genes expressed in both organs (2,5,6). TTF-1 binds to regulatory elements located in the promoters of a number of transcriptional targets in the lung (e.g. secretoglobin 1A, and the surfactant proteins Sftpa, Sftpb, and Sftpc) (5).
In thyroid and pulmonary epithelial cells, TTF-1 is phosphorylated at serine and/or threonine residues (7)(8)(9). Although the protein kinases and sites of phosphorylation mediating the interactions of TTF-1 with its various protein partners or DNA at cis-acting sites are not known with certainty, activation of protein kinase A enhanced transcriptional activation of Sftpb (8). However, direct effects of phosphorylation of TTF-1 on activity or DNA binding to thyroid-specific transcriptional target genes were not found in FRTL5 or HeLa cells in vitro (7,9,10). Hypophosphorylation of TTF-1 was observed in transformed thyroid cells in which TTF-1 target genes were not expressed; however, TTF-1 phosphorylation did not alter its binding to the thyroglobulin promoter (11). cAMP-dependent protein kinase stimulated phosphorylation of TTF-1 in several cell types; however, the effects of cAMP-dependent protein kinase on TTF-1-dependent transcription were not directly mediated by its phosphorylation (12). Taken together, TTF-1 is highly phosphorylated in many cell types, but the role of phosphorylation on transcriptional activation of target genes or on cell differentiation in target tissues remains unclear.
TTF-1 interacts directly or indirectly with other transcription factors and co-factors, including Foxa2, NF-1, GATA-6, AP-1, retinoic acid receptors, and associated co-factors, at or near TTF-1 binding, cis-acting elements located in regulatory regions of its target genes (5,(13)(14)(15)(16). Furthermore, TTF-1 expression is spatially regulated during lung morphogenesis, being more highly expressed in peripheral regions of the growing lung buds with advancing development (1,17). In the postnatal lung, TTF-1 is most abundant in type II epithelial cells in the alveolus, where it regulates surfactant protein synthesis. Thus, the activity of TTF-1 may be regulated by stochastic mechanisms, by interactions of TTF-1 with various protein partners, and by phosphorylation, oxidation-reduction, and cytoplasmic-nuclear trafficking (7,8,(17)(18)(19).
Because TTF-1 phosphorylation mutants retained transcriptional activities in vitro, the ability of TTF-1 PM to replace TTF-1 during lung morphogenesis was assessed in vivo. TTF-1 PM substantially, but not completely, corrected the defects in lung morphogenesis characteristic of Titf1 null mice. Microarray analysis was used to identify genes for which expression was influenced by the Titf1 PM/PM gene.

EXPERIMENTAL PROCEDURES
Generation of Titf1 PM/PM Mice-Mouse Titf1 gene was isolated from a strain 129/SV mouse genomic library (Stratagene) using a probe corresponding to the 3Ј-untranslated region of rat Titf1 (Fig. 1). To prepare the targeting vector, a fragment extending from bp 4656 to bp 10443 of the reported mouse genomic sequence (GenBank TM accession no. U19755), containing the entire coding sequence for Titf1, was cloned in pBlueScript. A fragment, spanning from the translation start site of Titf1 (bp 7957) to the end of homeobox (bp 9480) was removed and replaced by the sequence encoding S80, a phosphorylation mutant allele of rat Titf1 in which seven serine phosphorylation sites were replaced by alanine codons as described (10). The SV40 poly(A) sequence was inserted downstream of the S80 stop codon. The construct includes HSV-tk and PGK-neo cassette for selection of transfected ES cells. The target construct was introduced by electroporation in MPI1-ES cells and selected as described (20). Genomic DNA from neomycin resistance clone was digested with BamHI and analyzed by Southern blotting using as a 500-bp probe from nucleotide 10512 to nucleotide 11042 of the 3Ј-untranslated region of the mouse Titf1 gene (GenBank TM accession no. U19755). Chimeric mice were generated by aggregation of ES cells to CD1. Chimeras were mated to CD1 mice.
Animals-The colony of Titf1 PM/ϩ mice was maintained by crossing heterozygous mice with CD1 wild type animals. Embryonic day was estimated considering noon of the day of a vaginal plug as E0.5. Fetuses were collected at E18.5 by Cesarean section.
Genotyping-To genotype Titf1 PM/PM mice, DNA was obtained from a piece of tail from the mouse fetuses. The tissue was incubated overnight at 60°C with lysis buffer (50 mM Tris-HCl, 100 mM EDTA, 100 mM NaCl, 1% SDS, 0.5 mg/ml proteinase K), and genomic DNA was extracted by adding 0.3 volumes of 6 M NaCl and then precipitated with isopropyl alcohol. The genomic DNA was digested with BamHI and analyzed by Southern blotting.
Lung Histology, Immunohistochemistry, and in Situ Hybridization-Lungs were obtained from fetuses at E18 and were fixed with 4% paraformaldehyde. Lung tissue was processed according to standard methods and embedded in paraffin. Paraffin sections of lung tissue were cut at 5 m for histochemical analysis. Staining for the surfactant proteins SP-B and proSP-C, TTF-1, the Clara cell secretory protein (CCSP), PECAM (CD31), and ␣-smooth muscle actin (␣SMA) was performed as described previously (17). In situ hybridization for SP-A, SP-B, SP-C, and VEGF-A mRNAs were performed using 35 S-labeled riboprobes as previously described (21). Slides were coated in NTB-2, emulsion exposed for 2-5 days, and developed with Kodak D19.
RNase Protection and Western Blot Analysis-RNase protection assays for SP-A, SP-B, SP-C, and CCSP mRNAs were performed on lung RNA using 32 P-end labeled DNA probes as previously described (22); L32 mRNA was used to normalize loading. Blots were scanned and differences compared by Student's t test. Proteins from lung homogenates from wild type and Titf1 PM/PM mice (E18) were separated by SDS-PAGE and blotted using antisera against proSP-B, SP-B, proSP-C, (Chemicon AB3430, AB3436, and AB3428, respectively), and napsin (kidney-derived aspartyl proteinase). To generate napsin antibody, the mouse napsin A cDNA was amplified from type II epithelial cell cDNA, sequenced, and the region encoding pronapsin cloned into the bacterial expression vector pET-21 (Novagen). Recombinant napsin protein was purified from bacterial lysates by chromatography on nickel-nitrilotriacetic acid resin and injected into rabbits. The napsin antibody detected a single protein band (M r of 38,000) in immunoblots of mouse kidney.
RNA Microarray and Promoter Analysis-Total RNA from lungs at E18 Titf1 PM/PM and wild type littermates was subjected to reverse transcription using oligo(dT) with T7 promoter sequences attached, followed by second strand cDNA synthesis. Antisense cRNA was amplified and biotinylated using T7 RNA polymerase, prior to hybridization to the version 2 of murine genome U74 set, which consists of three GeneChips and ϳ36,000 full-length mouse genes/ESTs (Affymetrix Inc.), using the Affymetrix recommended protocol (23,24). Affymetrix MicroArray Suite version 5.0 was used to scan and quantitate the GeneChips using default scan settings. Intensity data was collected from each chip and scaled to a target intensity of 1500. The results were analyzed using GeneSpring 5.0 (Silicon Genetics, Inc.), JMP4 (SAS Institute, Inc.), and Spotfire 7.12 (Spotfire, Inc.) software.
A total of 18 chips were used in this experiment. Hybridization data (ϳ216,000 data points) were sequentially subjected to normalization, transformation, filtering, clustering, and function classification as previously described (25). Data were normalized to enable the direct comparisons across chips and across genes. Statistical differences between Titf1 PM/PM and control littermates were identified by distribution analysis and Welch's t test at p value Յ 0.05. Variations related to processing and biological replicates were calculated and separated from the candidate genes to identify primary genotype response. Fold changes were calculated for each gene against its specific control to determine relative gene expression. Additional filters included minimal absolute intensity Ն 30, a minimum of 4 detectable judgments for A-set (12 chips) and 2 detectable judgments for B and C-sets (6 chips), and coefficient of variation among replicates Յ 50%. Genes with average fold changes Ն 2 and genes that were cross-validated via different probes on the same chip, or the same gene on different chips (A, B, and C sets), were prioritized. Differentially expressed genes were classified into functional categories based on gene ontology definitions. To determine representation of functional categories in the selected gene list, the binomial probability was calculated for each category using corresponding U74Av2 genome as the reference dataset. Hierarchal clustering was applied to visualize and further group the selected genes based on their expression similarity. Pearson correlation was used for similarity measure. Clusters were constructed by the unweighted pair group method with arithmetic mean.
Mouse and human promoter sequences were downloaded from the Harvard-Lipper Center for Computational Genetics (www.arep.med. harvard.edu). Upstream sequences (1 kb) of selected genes were retrieved from above promoter sequence data base and searched for potential TTF-1 regulatory sites. A 15-bp sequence NNWCTCAAGTRY-WNN from the Genomatix matrix library was used as the TTF-1 consensus binding site with the core similarity (CAAG) setting as 1 and maximal of 2 mismatches (Genomatix, Inc.).

RESULTS
Heterozygous Titf1 PM/ϩ mice have been maintained in the vivarium for more than 1 year without apparent abnormalities in activity or reproduction. Although Titf1 PM/PM fetuses were present at E18 in numbers consistent with that expected by Mendelian inheritance, Titf1 PM/PM pups were never observed postnatally (Table I). At E18 -18.5, homozygous Titf1 PM/PM mice were alive at birth, but died rapidly of respiratory failure. At E18, lung mass was visibly reduced. Total RNA and protein content was decreased, indicating pulmonary hypoplasia. The trachea and bronchi were intact and without the tracheoesophageal fistulae characteristic of Titf1 null mice. In contrast to the lack of peripheral lung tissue characteristic of Titf1 null mice, bronchiolar and acinar regions of the lung were formed in Titf1 PM/PM mice (Fig. 2, A and B). Staining for TTF-1 was readily detected in the nuclei of respiratory epithelial cells. The intensity and distribution of staining was similar in Titf1 PM/PM and Titf1 ϩ/ϩ controls (Fig. 2, C-F). Epithelial cell types characteristic of the conducting and peripheral airways were observed in proper proximal-distal gradients along the airway, although squamous cell differentiation was lacking in the lung periphery, perhaps indicating arrested terminal differentiation of type II cells. Abnormally dilated peripheral lung tubules were observed, however, in all of the mutant mice, indicating defects in formation of the lung parenchyma. Lobulation was normal as evidenced by four right lobes and one left lobe, supporting the concept that the TTF-1 PM is sufficient to direct early lung morphogenesis. Deficits in vasculogenesis were indicated by decreased PECAM staining surrounding dilated peripheral lung tubules, in Titf1 PM/PM mice (Fig. 3, A and B). Extensive staining for ␣SMA was observed surrounding the abnormally dilated peripheral tubules (Fig. 3, C and D), indicating failure of peripheral mesenchymal differentiation. Staining for CCSP was decreased in the conducting airways of the mutant mice (Fig. 3, E and F).
VEGF-A mRNA expression in the Titf1 PM/PM was also reduced, especially in the distal-most peripheral, acinar tubules and terminal saccules (Fig. 3, G and H). The reduction in both CCSP and VEGF-A mRNA expression appeared to be a result of decreased expression per cell and decreased numbers of cells expressing these RNAs.
Changes in Expression of TTF-1-regulated Genes-Because TTF-1 is known to be required for the expression of a number of proteins selectively expressed in the respiratory epithelium, immunohistochemistry or in situ hybridization was performed for SP-A, SP-B, proSP-C, and CCSP (Fig. 4). Immunostaining for proSP-C and SP-B, markers selective for acinar and type II epithelial cells of the peripheral lung, was detected in Titf1 PM/PM mice (data not shown). The proximal-distal distribution and intensity of surfactant protein staining for SP-B and proSP-C was similar to that in control littermates, al-

FIG. 2. Lung histology and TTF-1 expression.
Reduced numbers of terminal alveolar saccules and abnormally dilated, or cystic, peripheral structures (arrowheads) were observed in lungs from Titf1 PM/PM mice at E18 (A and B). Immunostaining for TTF-1 in lungs from wild type (C and E) and mutant mice (D and F) was detected in the nuclei of pulmonary epithelial cells lining the bronchioles (br) and peripheral respiratory structures (*). Peripheral respiratory structures were well developed in control mice, forming terminal alveolar ducts and saccules (*) with relatively thin interalveolar septa lined by TTF-1-positive type II cells (arrow) and thin, squamous, type I cells lacking TTF-1 staining. In Titf1 PM/PM mice, the peripheral respiratory parenchyma was composed of 2-3 generations of abnormally branched, dilated, acinar tubules (*), ending in smaller acinar buds (arrowheads) (D and F), which were lined by TTF-1-positive epithelial cells (arrow) and surrounded by abundant mesenchyme. Br, bronchiole; t, terminal bronchiole; *, acinar tubule/alveolar duct. Illustrations are representative of n ϭ 8 -9 for each genotype. Bars equal 1 mm (A and B), 100 m (C and D), and 50 m (E and F).
though SP-B staining was generally extracellular in wild type and intracellular in mutant mice. Expression of SP-A mRNA was absent in the conducting airways and dramatically decreased in the peripheral airways of the mutant mice, whereas SP-C and SP-B mRNAs were relatively preserved (Fig. 4

, E-H).
Nuclease protection assays were utilized to quantify the effect of the Titf1 PM/PM on surfactant proteins and CCSP mRNAs (Fig. 5). Abundance of CCSP and SP-A were decreased ϳ5-8fold, whereas that for SP-B was decreased ϳ50%. Although undetectable in Titf1 null mutant mice (data not shown), expression of surfactant proteins and CCSP was present in the Titf1 PM/PM mice. Thus, the decrease in expression of these target proteins was variable, and numbers of cells and levels of SP-A and CCSP were most decreased; in contrast, levels of expression of SP-B and SP-C were maintained in some cells, but the proportion of cells expressing these RNAs was altered in the mutant mice.
Decreased Surfactant Protein and Processing-Decreased but variable expression of known TTF-1 target genes suggested that the Titf1 PM/PM may function, at least in part, as a hypo-morphic allele. Reduction of SP-B to less than 50% of normal levels causes lung dysfunction in mice (26,27); therefore, reduction in SP-B expression may contribute to lack of postnatal FIG. 3. PECAM: ␣SMA, CCSP, and VEGF expression. In the wild type mice (A), PECAM immunostaining was detected in large vessels (v) and in the extensive capillary network located in the alveolar septa (arrowheads). In the Titf1 PM/PM mice (B), relatively few PECAM-positive capillaries (arrowheads) were detected in the mesenchyme surrounding dilated, bronchiolar-like, peripheral tubules (*), often found at the tips of the lung lobes. In the wild type control (C), ␣SMA-positive cells (arrows) were found adjacent to the basal side of the bronchiolar epithelium (br), in the wall of small veins (v), and in scattered, individual myofibroblasts located in the alveolar septa (arrowheads). In mutant mice (D), ␣SMA-positive cells (arrowheads) completely surrounded each of the dilated, bronchiolar-like, peripheral tubules (*). In controls, CCSP-staining cells were detected in abundance throughout the conducting airways (arrowheads) (E). In the mutant mice, the staining intensity and the number of CCSP-positive cells (arrowheads) were reduced in the conducting airways (F). The number of VEGF mRNA-positive cells detected by in situ hybridization was reduced in the Titf1 PM/PM mice (G and H), especially in dilated acinar tubules located at the periphery of the lung (arrowheads). Br, bronchiole; v, vessel; alv, alveolar saccule; *, dilated, bronchiolar-like, peripheral tubule. Illustrations are representative of n ϭ 8 -9 mice of each genotype.

FIG. 4. SP-A, SP-B, and SP-C mRNA expression.
Toluidine bluestained, bright field images of wild type (A) and Titf1 PM/PM (B) mice illustrating the underlying histology for the dark field in situ hybridization images below (C-H). Note the reduced number of terminal alveolar saccules and abnormally dilated, or cystic, peripheral structures (*) in the mutants (B). SP-A, SP-B, and SP-C mRNAs were found in the alveolar ducts and saccules, whereas SP-B mRNA was also detected in the bronchiolar epithelium (br) in controls (A, C, E, and G). Weak, scattered, hybridization signals for SP-A mRNA were found in the peripheral respiratory tubules of mutant mice (D). SP-B mRNA was detected throughout the conducting airway, in the peripheral respiratory tubules, and in the abnormally dilated, peripheral, bronchiolar-like structures (F). SP-C mRNA was detected in the peripheral respiratory tubules, but not in the dilated, bronchiolar-like, peripheral structures (H). Br, bronchiole; v, vessel; *, dilated, bronchiolar-like, peripheral tubule. Panels are representative of n ϭ 8 -9 for each genotype. Bars equal 100 m (A-H).
FIG. 5. S1 nuclease protection assay for SP-A, SP-B, SP-C, and CCSP mRNA. Total lung RNA was isolated from control and Titf1 PM/PM littermates and subjected to S1 nuclease protection assay (A, 3 g/lane). Gels were quantified by phosphorimaging and SP gene expression was normalized to corresponding control L32 mRNA. The mean values for wild type mice were set to 1 of each SP, and relative expression was plotted (B) as mean Ϯ S.D. (n ϭ 3); *, p Ͻ 0.05. survival of Titf1 PM/PM mice. Because processing of proSP-B and proSP-C is known to be cell-specific and proteolytically processed SP-B peptide is required for surfactant function, processing of proSP-B was assessed in lung homogenates from Titf1 PM/PM mice. The active SP-B peptide (18-kDa dimer) was markedly decreased in lung homogenates from the mutant mice, whereas abundance of the 42-kDa proSP-B precursor was increased, indicating that proteolytic processing of SP-B was deficient (Fig. 6). RNA microarray data indicated a 21-fold decrease in expression of napsin (Kdap) (see below). Napsin, a type II epithelial cell-selective aspartyl protease, was decreased in lung homogenates of the Titf1 PM/PM mice (Fig. 6). Together with the previous report that an aspartyl protease is required for maturation of proSP-B, these findings support the likelihood that napsin plays a role in the processing of SP-B. Deficient processing of SP-B may contribute to the respiratory failure in the Titf1 PM/PM mice, although the observed alterations in lung morphogenesis likely contribute to respiratory failure after birth.
Identification of Genes Influenced by the TTF-1 Phosphorylation Mutation-To identify genes responsive to Titf1 PM/PM , lung RNAs from Titf1 PM/PM mice (E18) and their wild type control littermates were compared using the Affymetrix murine genome U74Av2, which contains three GeneChips and ϳ36,000 full-length mouse genes/ESTs. Data from 18 chips were normalized, and statistical differences between Titf1 PM/PM and control mice were identified with p values Յ 0.05 and fold change Ն 2. Using these criteria, 98 known genes were identified from the A chip (Tables II and III). Forty-nine mRNAs were increased and 49 were decreased in the lungs from Titf1 PM/PM pups. There were 97 ESTs in B and C chip that also met these criteria. Annotations for those ESTs were collected through a combination of homology searches against known mouse genes and retrieved ortholog information from rat and human genomes (NetAffx). Forty-seven genes with known annotations are listed in Table IV. Hierarchical clustering of these differentially regulated genes is shown in Fig. 7. Data are shown in two-dimensional matrix, and remarkably ordered gene expression profiles were displayed on genes se-lected from A chip (Fig. 7A) and B and C chips (Fig. 7B). At the chip level (top dendrogram), RNAs influenced by Titf1 PM/PM formed two distinct groups. Within the mutant group, samples collected from the same littermates were more closely related than those from different litters, whereas no differences among litters were observed in wild type controls. At the RNA level (see the dendrogram at the left of Fig. 7), genes were clearly separated into those mRNAs increased or decreased corresponding to the Titf1 genotypes.
Because TTF-1 plays critical roles in lung development and morphogenesis, we specifically inspected the effects of the Titf1 PM/PM mutant on genes that are considered important for lung formation and/or function, including some known TTF-1 transcriptional targets (Fig. 8). Among them Kdap (napsin), Calb3, Sftpa, Vegf-A, Scgb1a1 (also known as CCSP), Aqp5, Sox17, Lzp-s, Fgf1, and Pdgfra were decreased in Titf1 PM/PM mice; Sftpb, Sftpc, and Bmp4 were significantly, but moderately, decreased. The decrease in Sftpc observed in the array was not great as seen by S1 nuclease assay, but was statistically significant. Other genes including Sftpd, Mdk, Evi1, Clu, Znfn1a1, and FoxM1 were increased in the mutant mice. To further test whether the Titf1 PM/PM influenced expression of these genes directly or indirectly, we searched for potential TTF-1 binding sites within 1 kb upstream of the start of transcription. Genes expressed selectively in the respiratory epithelium versus lung mesenchyme were identified. Genes 1) influenced by Titf1 PM/PM , 2) selectively expressed in respiratory epithelium, and 3) containing TTF-1 binding site(s) within 1 kb of the start of transcription were considered as possible direct transcriptional targets of TTF-1. A number of genes fit these criteria, including Sftpa, Sftpb, Sftpc, Scgb1a1, Clu, Tcf7, Lef1, ␤-catenin, Sox17, Aqp1, Aqp5, Bmp4, Lzp-s, Zfp386 (Kruppel-like), H 2 -Q 1 , Calb3, Gsta4, Mdk, and Evi1.
Differentially expressed genes were further classified according to their known or predicted functions. Each gene was annotated and assigned to a functional category. To simplify the calculation, we assumed that genes in each category could be fit to a binomial distribution. The binomial probability was calculated for each category using the entire U74Av2 as the reference dataset. The "defense response," which includes immune, inflammatory, and stress responses, was the most represented category of those RNAs increased in the Titf1 PM/PM mice. Among RNAs for which abundance was decreased, those involved in lipid metabolism, signal transduction, and defense response were most highly represented (Tables V, parts a and b).

Titf1 PM/PM Supports Early Branching Morphogenesis but
Not Maturation of Acinar Saccules-Lobulation and early branching morphogenesis were maintained, whereas formation and differentiation of peripheral bronchioles and acini were deficient in the Titf1 PM/PM mice. Peripheral pulmonary vessels, as indicated by regional differences in PECAM staining, were perturbed in the abnormal lung saccules and the numbers of cells expressing VEGF mRNA decreased. Deficient formation of acinar buds and terminal saccules, as well as of the alveolar capillary bed, in the mutant mice indicates a critical role for TTF-1 in the regulation of genes required for reciprocal interactions between the epithelium and mesenchyme during formation of the peripheral lung. These findings are consistent with recent studies supporting the concept that early vascularization is required for normal morphogenesis of the developing pancreas (28).
Lack of Terminal Differentiation in Peripheral Lung Structures-Peripheral lung tubules in the Titf1 PM/PM mice were poorly developed, and squamous type I cell differentiation, typical of the normal E18 lung, was not observed. Likewise, there was a paucity of small blood vessels that normally come into close apposition with the respiratory epithelium at this time. Fewer terminal saccules were observed in the mutant mice, indicating arrest of late branching morphogenesis, resulting in fewer acinar tubules. Taken together, lung morphology in the Titf1 PM/PM mice is consistent with abnormalities in branching morphogenesis and/or delay in cytodifferentiation that normally occur in the late pseudoglandular, canalicular, saccular stages of development (E14 -E18) and is termed acinar dysplasia. These findings are distinct from those in Titf1 Ϫ/Ϫ mice in which trachea and main bronchi form, but the bronchial tubules fail to undergo further branching with resultant loss of lobulation and absence of peripheral lung parenchyma (2).
Epithelial Cell Differentiation in the Titf1 PM/PM Mice-TTF-1 regulates the expression of surfactant proteins SP-A, SP-B, SP-C, and CCSP, which mark distinct and overlapping subsets of respiratory epithelial cells. In the present studies, the mutant Titf1 PM/PM had differential effects on expression of known transcriptional targets. Immunohistochemical analysis and in situ hybridization demonstrated that SP-C and SP-B mRNAs were expressed in the peripheral lung structures of the mutant mice at levels similar to that expressed by individual cells in control mice. However, SP-B-and SP-C-positive cells were not observed in many of the larger, dilated peripheral tubules of the Titf1 PM/PM mice. Changes in relative numbers of cells expressing these RNAs rather than transcriptional activity of the genes were influenced by TTF-1 PM . Expression of CCSP and SP-A mRNA was more markedly decreased in all cells in the mutant mice, likely indicating an effect of TTF-1 PM on their transcription. Thus, proximal/peripheral patterning of the epithelial cell differentiation was generally maintained, but the level of expression of these TTF-1 target genes was variably decreased. Because the numbers of peripheral tubules were decreased, reduction in peripheral lung markers may indicate reduction in numbers of specific cell types, decreased transcription of target genes, or both. Although type II epithelial cell differentiation was observed, as indicated by the expression of SP-C mRNA and proSP-C staining, squamous alveolar type I cells were not present in the abnormal lung tubules. SP-B mRNA, normally expressed in both conducting and peripheral airways at E18 -18.5, was present throughout the pulmonary Expression of genes known to be direct targets of TTF-1, including surfactant proteins and CCSP, was significantly, but variably, decreased in the Titf1 PM/PM mice, supporting the concept that the Titf1 PM/PM mutant represents, in part, a hypomorphic TTF-1 allele. Because lung structure was perturbed in the mutant mice, differences in the proportions of specific subsets of cells may be influenced by changes in RNA concentration related to differences in cell types rather than by transcriptional mechanisms. For example, peripheral tubules were decreased in number, and squamous cells (type I cells) failed to form in the mutant lung. Thus, decreased aquaporin-1 mRNA (a marker of type I epithelial cells) may also reflect changes related to the absence of cell type and/or regulation by TTF-1.
Decreased Processing of ProSP-B-Immunostaining indicated the lack of the active SP-B protein in the airways and increased intracellular staining for SP-B in the mutant mice (data not shown), indicating lack of secretion or proteolytic processing of SP-B that normally occurs in the perinatal period.
Consistent with this observation, SP-B processing was decreased and the abundance of proSP-B increased in the Titf1 PM/PM mice. Napsin mRNA and protein were also decreased, indicating a potential role for this protease in cellspecific processing of SP-B.

mRNAs Relevant to Perinatal Lung Function Were Decreased in Lungs from Titf1 PM/PM Mice: Surfactant Protein and Lipid
Homeostasis-Expression of a number of known and potential TTF-1 target genes was reduced in the Titf1 PM/PM mice, including the known TTF-1 target genes, Sftpa, Sftpb, Sftpc, and Scgb1a1. Kidney-derived aspartyl protease-like protein (napsin) was dramatically decreased (21-fold). Napsin is expressed selectively in type II epithelial cells of the lung, and in a subset of renal tubular epithelial cells in the kidney (29). Decreased napsin and deficient proteolytic processing proSP-B were observed in the lungs of Titf1 PM/PM mice. Stearyl-coenzyme A desaturase-1, phospholipid scramblase 2, low density lipoprotein receptor 2, fatty acid synthase, and pyruvate carboxylase mRNAs were significantly decreased in the mutant mice, predicting a potential role for TTF-1 in the regulation of lipid homeostasis that is required for surfactant production prior to birth. The expression of ␤-adrenergic receptor 2, known to regulate surfactant secretion and ion transport in the neonatal lung, was decreased ϳ3-fold. The numbers and activity of ␤ 2adrenergic receptors increase dramatically in the perinatal and postnatal period (30), a process induced by perinatal exposure to glucocorticoid, consistent with its role in surfactant secretion at birth. Genes Regulating Fluid and Electrolyte Transport-Aquaporin-1 and -5 mRNAs were decreased (4-and 3-fold, respectively) in the Titf1 PM/PM mice. Although the functional significance of these findings is unclear, TTF-1 may influence airway reactivity; aquaporin-5-deficient mice develop airway hyperactivity in response to cholinergic challenge (31). Expression of a number of solute carriers (neurotransmitter transporter, sodium/sulfate supporter, and organic anion transporter), ion channels (sodium channel non-voltage-gated 1␥, and voltage-gated type IV ␣ polypeptide) were significantly decreased (2-and 9-fold), indicating that TTF-1 may regulate these genes to maintain fluid and electrolyte balance in the lung.
Regulation of Genes Modulating Host Defense Functions-Genes involved in host defense and inflammation were most influenced by the Titf1 PM/PM . Calbindin D9K (a neutrophil chemoattractant molecule), hemolytic complement (Hc), Scgb1a1, Sftpa, Kit ligand (stem cell factor), and lysozyme were decreased. Some of these genes are known to be expressed in respiratory epithelial cells and are direct transcriptional targets of TTF-1, e.g. Sftpa and Scgb1a1. Scavenger receptor class A, IL-7, leukocyte cell-derived chemotaxin-1, clusterin, peptoglycan recognition protein, chitinase (acidic), glutathione Stransferase, trefoil factor 2,3, small proline-rich polypeptide, serum amyloid, CD14, and others were increased. TTF-1 staining was decreased in human lung tissues following lung injury and infection (32); thus, decreased activity of TTF-1 may influence transcription of host defense genes involved in protection from lung injury and in repair. The observed increase in the expression of mRNAs selectively expressed in lymphocytes may represent the presence or absence of thymic tissue that may be Because these changes were found in each of the individual lungs assayed, it is unclear whether changes in lymphocytes or the relative proportion of thymic tissues contributing to the RNA pools were influenced by TTF-1.
The increased representation of genes involved in host defense in the Titf1 PM/PM mice supports the concept that TTF-1 phosphorylation plays a role in host defense responses in the lung. It is of interest that increased expression of TTF-1 in the postnatal lung in transgenic mice caused marked inflammation, emphysema, and eosinophilic infiltration (21). It is also possible that increased expression of some of these genes represent cell injury responses related to the TTF-1 mutant protein (e.g. glutathione S-transferase and serum amyloid, etc.), a possibility that cannot be excluded. However, there was no observable histologic evidence of cell necrosis or inflammation in wild type or Titf1 PM/PM mice.
Alterations in Transcriptional Pathways Modulating Respiratory Epithelial Cell Differentiation-Significant differences were observed in the abundance of mRNAs encoding a number of transcriptional proteins known to be expressed in the developing lung. Such changes may indicate that these proteins are direct or indirect targets of TTF-1 phosphorylation or TTF-1 per se. The increase in myb may represent a compensatory response to decreased TTF-1 activity, because myb is known to act synergistically with TTF-1 and binds to elements in the Sftpa gene (33). Expression of several transcription factors were decreased, perhaps representing potential transcriptional targets of TTF-1 phosphorylation. Forkhead F2 (known to be expressed in lung mesenchyme), naked cuticle homologue, androgen-induced basic leucine zipper, metal response element binding transcription factor-1, homeobox2, forkhead boxQ, paired related, thymocyte-selective HMG box, EVI-1, and FoxM 1 (a Fox family member regulatory cell cycle and expressed in the lung mesenchyme) mRNAs were increased. Although these changes may represent reciprocal or compensatory responses to the lack of TTF-1 activity or phosphorylation, changes in their abundance may reflect changes in cell populations in which they are expressed.
Several mRNAs in the Wnt signaling cascade were altered in the Titf1 PM/PM mice, including Wnt-4 and -11, ␤-catenin, Tcf-7 (Tcf-1), and Lef-1, which were increased 2-3-fold. In contrast, expression of Wnt-3a and Wnt-5b were decreased in the mutant lungs. Nuclear ␤-catenin is present in epithelial cells of the developing lung during the embryonic period, at sites overlapping with TTF-1 (34). The present findings support the concept that   7. Two-dimensional hierarchical clustering of selected genes/EST. Shown are data from U74Av2 (A) and U74Bv2 (B) and Cv2 chips identifying mRNAs significantly regulated in response to Titf1 PM/PM . Intensity in the red and green color range indicate the increase and decrease in mRNA abundance, respectively. Each row represents a single gene; each column represents a particular experimental sample; each box represents a normalized gene expression value. Clustering method, unweighted pair group method with arithmetic mean. Similarity measure, Euclidean distance. TTF-1-and ␤-catenin-dependent pathways interact in the peripheral lung, directly or indirectly, during lung morphogenesis.
Genes Modulating Lung Vasculogenesis-Regional decreases in vascularity of the abnormal peripheral lung tubules, as detected by PECAM staining, were associated with decreased VEGF-A mRNA, indicating that TTF-1 phosphorylation is required for normal levels of expression of VEGF in the developing respiratory epithelium. Surprisingly, PECAM RNA was increased in the mutant mice, perhaps related to the extensive tissue remodeling. BMP-4, ECAM, carbonic anhydrase, VEGFR1, and ephrine A2, proteins known or considered to be markers or regulators of pulmonary mesenchyme differentiation and vasculature formation, were perturbed in the Titf1 PM/PM mice.
Identification and Mapping of Known and Predicted TTF-1 Response Elements-Genes known to be expressed in a respiratory epithelial cell-specific manner under direct transcriptional control of TTF-1 were subjected to a computer-assisted analysis of their regulatory regions. Consensus elements for TTF-1 binding were readily detected within the regulatory regions of Sftpa, Sftpb, Sftpc, and Scgb1a genes. A number of these elements were previously validated by direct site-specific mutagenesis, transfection assays, and gel retardation analyses. A consensus for a TTF-1 binding sequence was utilized to identify potential TTF-1 regulatory elements in the subset of genes for which expression was influenced in the Titf1 PM/PM mice. A distinct subset of genes expressed in the lung mesenchyme did not contain the element, but were consistently influenced by the Titf1 PM/PM , supporting the likelihood that TTF-1 influences their expression indirectly, via reciprocal tissue interactions between the epithelium and the mesenchyme or by changing the proportions of cells expressing the gene.
Conclusion-The lung developed relatively late during vertebrate evolution, representing a singular solution to the problem of air-breathing. TTF-1 is required for normal formation of the peripheral lung at birth. Perhaps it is not surprising that Titf1 PM/PM influences lung structure and the expression of subsets of genes regulating biological functions uniquely required for adaptation following birth, including host defense, fluid balance, surfactant homeostasis, and the formation of an extensive interface between the peripheral-vascular bed and the alveolar surfaces upon which gas exchange depends.