The Fgf9-Nolz1-Wnt2 axis regulates morphogenesis of the lung

The morphogenesis of the lung is coordinated by complex genetic cascades that occur in endoderm-derived epithelia and mesoderm-derived mesenchyme (Morrisey and Hogan, 2010; Warburton et al., 2000; Zepp and Morrisey, 2019). Transcription factors that are selectively expressed in the mesenchyme or epithelium are important for lung development (Costa et al., 2001; Maeda et al., 2007). For example, Foxf1, Tbx4/5, and Hox5 are expressed specifically in mesenchymal cell lineages (Chapman et al., 1996; Hrycaj et al., 2015; Ustiyan et al., 2018). Deletion of the mesenchyme-specific genes Foxf1, Tbx4/5 or Hox5 genes (Hoxa5, Hoxb5 and Hoxc5) results in a prominent phenotype of reduced lung buds (Cardoso and Lü, 2006; Morrisey and Hogan, 2010). Foxf1, Tbx4/5 and Hox5 regulate not only tracheal mesenchyme differentiation, but also epithelial cells. Knockout (KO) mouse studies show that Foxf1, Tbx4/5 and Hox5 regulate lung development through Wnt2/2b and Bmp4 signaling (Arora et al., 2012; Hrycaj et al., 2015; Ustiyan et al., 2018).

Epithelium-specific transcription factors include TTF1 (Nkx2-1), Foxa1/a2 and Sox2 (Maeda et al., 2007). Deletion or overexpression of these epithelium-specific transcription factors results in deficits in branching morphogenesis, differentiation of epithelial cells and mesenchymal cells, along with altered BMP or Shh signaling molecules (Gontan et al., 2008; Minoo et al., 1999; Wan et al., 2005). Because deletion of mesenchyme-specific transcription factors could non-cell autonomously affect epithelial development and vice versa, it highlights the importance of signal crosstalk between the mesenchymal and epithelial compartments. The signal molecules Fgf, Wnt, Shh and BMP have been shown to underlie mesenchyme-epithelium interactions in lung development (Han et al., 2016).

Nolz1 (also known as Znf503 and Zfp503) is a murine member of the nocA/elB/tlp-1 (NET) gene family. The NET family is an evolutionarily conserved gene family that also includes tlp-1 (Caenorhabditis elegans), elbow (elB) and no ocelli (noc) (Drosophila), nlz1 (znf703) and nlz2 (znf503) (zebrafish), Nolz1 (chick), and Nolz2 (Zfp703) (mouse) (Chang et al., 2004; Chen et al., 2020; Dorfman et al., 2002; Ji et al., 2009; Runko and Sagerström, 2003; Zhao et al., 2002). Members of the NET family contain three core motifs, including the buttonhead box, Sp and C₂H₂ zinc-finger motifs (Nakamura et al., 2004; Pereira et al., 2016). Nolz1 and its homologs have been shown to function as a transcriptional repressor by interacting with the Groucho corepressor. ElB is associated with Groucho via the FKPY motif, whereas the interacting domain of Nlz1/Nlz2 with Groucho is mapped to the region between the buttonhead box and the C₂H₂ zinc-finger motif (Pereira-Castro et al., 2013; Runko and Sagerström, 2004). Chick Nolz1 has also been shown to interact with Grg5. Functionally, Nolz1 and its homologs play a significant role in the developmental regulation of patterning, cell fate specification and differentiation of neuronal and non-neuronal systems. Tlp-1 determines the asymmetric cell of tail cells that responds to Wnt signals in C. elegans (Zhao et al., 2002). ElB/noc regulate the development of eyes, legs and trachea in Drosophila (Dorfman et al., 2002). Nlz1 and Nlz2 are involved in the patterning of rhombomeres in the development of the hindbrain and the closure of the optic fissure in zebrafish (Brown et al., 2009; Runko and Sagerström, 2003). Chick Nolz1 is required for the specification of motor neurons (Ji et al., 2009). Murine Nolz1 regulates the formation of striatal subdivisions in the mouse forebrain (Chen et al., 2020). The functional diversity of the members of the NET family implicates important roles of NET proteins in the developmental regulation of different cell types in organogenesis.

The Nolz1 Drosophila ortholog elB/noc has been shown to regulate the development of the trachea, a respiratory organ in Drosophila (Dorfman et al., 2002). Given the importance of elB/noc in the specification of tracheal branches in Drosophila, Nolz1, as an evolutionarily conserved NET protein, could play a role in mammalian lung development. Here, we first identified Nolz1 as a mesenchyme-specific transcriptional regulator in mouse lung development. We then investigated the biological function of Nolz1 in the regulation of lung morphogenesis.

To investigate the temporal expression profile of Nolz1 protein in the developing mouse lung, we performed the time course study of Nolz1 protein expression by western blotting. The results showed that Nolz1 protein was highly expressed at embryonic day (E) 12.5 and E14.5 and was downregulated at E16.5, with the expression drastically decreasing after birth and barely detectable in adult mouse lungs (Fig. 1A). These findings suggest that Nolz1 may play a role in the early stages of lung development.

To investigate the expression pattern of Nolz1 protein, we performed immunostaining of Nolz1. Consistent with the western blotting results, we found that Nolz1 protein was detected in the pseudoglandular (E12.5-E16.5), canalicular and saccular (E17.5-E18.5) stages of lung development (Fig. 1B) (Costa et al., 2001; Maeda et al., 2007; Morrisey and Hogan, 2010; Warburton et al., 2000). Nolz1 immunoreactivity was localized in the nucleus (Fig. S1B). Double immunostaining of Nolz1 and TTF1, a marker of epithelial cells (Bohinski et al., 1994), showed that Nolz1 was not co-localized in TTF1-positive epithelial cells at E11.5, E12.5, E14.5, E16.5 and E18.5 (Fig. 1B). Nolz1 was expressed in the majority of TTF1-negative mesenchymal cells at E11.5-E18.5 (Fig. 1C). These findings indicated that Nolz1 was specifically expressed in the TTF1-negative mesenchyme of developing lungs.

We next determined whether Nolz1 was expressed in proliferating progenitors of the mesenchyme. Double immunostaining of Nolz1 and Ki67, a proliferating marker, showed colocalization of Nolz1 and Ki67 in mesenchymal cells at E12.5 (Fig. 1D,E). Nolz1 was co-expressed with PECAM1, a marker of progenitors and differentiated endothelial cells (White et al., 2007), in E12.5 mesenchymal cells (Fig. 1D,E). Nolz1 was co-expressed with SM22α (Tagln), a marker of progenitors and differentiated smooth muscle cells (Goss et al., 2011; Solway et al., 1995), in E12.5 mesenchymal cells (Fig. 1D,E). These findings identified Nolz1 as a previously unreported marker for mesenchymal cells in the developing mouse lung.

To investigate the function of Nolz1 in lung development, we studied Nolz1 germline KO mice (Fig. S1). Nolz1 KO mice died immediately after birth, probably because of lung failure, because lung size was dramatically decreased in E18.5 KO mice. Morphologically, the mutant lung had five lobes with proper orientations, but the mutant lobes collapsed and were significantly smaller than those of the wild type (Fig. 2A). The time-course study showed that mutant lungs were morphologically normal at E11.5 compared with the wild-type lungs. The impaired growth of mutant lungs became evident at E12.5, as the number of epithelial branches was reduced in mutant lungs (Fig. 2B). By E14.5 and E16.5, the size of mutant lungs was about one-third that of wild-type lungs. The mutant lungs were much more compact and smaller than wild-type lungs at E18.5.

Hematoxylin and eosin (H&E) stains showed that the mesenchyme and epithelium appeared to be similar between wild-type and KO lungs at E12.5 (Fig. 2C). By E14.5, reduced and less compact mesenchyme was observed in mutant lungs (Fig. 2C,D). Notably, enlarged mutant epithelial tubules were found at E14.5 (Fig. 2C,D), but this phenotype was not observed after E14.5. At E16.5, which was the beginning of the canalicular stage, the epithelium began to form the saccular structure in wild-type lungs. By contrast, immature epithelial columns remained in mutant lungs at this stage (Fig. 2C). Furthermore, the E18.5 mesenchymal septation between the air sacs was thicker in mutant lungs than in wild-type lungs (Fig. 2C,D).

To examine whether abnormal cell death accounted for the hypoplasia of the Nolz1 null mutant lung, we assayed apoptosis using immunostaining of active caspase 3 (AC3), an apoptosis marker. The results showed that no, or at most a few, AC3-positive cells were observed in wild-type and KO lungs, and no apparent differences were detected at E12.5 and E13.5 (Fig. S2A,B). Note that AC3 immunostaining was validated by positive controls in which many AC3-positive cells were present in the dorsal root ganglion (DRG) in the same sections at E12.5 and E13.5.

Similar results were obtained using the terminal deoxynucleotidyl transferase-mediated dUTP nick end labeling (TUNEL) assay. Few TUNEL-positive cells were present in wild-type and Nolz1 KO lungs, with no significant differences between the two genotypes at E12.5 and E13.5 (Fig. S2C,D). TUNEL-positive cells were present in the DRG of the same sections of E12.5 embryos as a positive control. These findings suggest that Nolz1 null mutation does not induce abnormal apoptosis at the early stages of lung development.

Given the hypoplasia phenotype and specific expression of Nolz1 in the mesenchyme, Nolz1 null mutation might cause defective proliferation of mesenchymal cells. To investigate, we assayed cell proliferation with the BrdU incorporation assay. Pregnant mice were pulse-labeled with 5-bromo-2′deoxyuridine (BrdU) (50 mg/kg) for 1 h before embryo culling at E12.5 or E13.5. BrdU-positive cells were decreased by 33% and 20%, respectively, in the mutant mesenchyme at E12.5 and E13.5 compared with wild type, but not in the mutant epithelium (Fig. 3A). Because branching morphogenesis was impaired in Nolz1 KO lungs, defective cell proliferation might occur at the distal lung buds. Double immunostaining of BrdU and Sox9, a distal epithelium marker, showed that the percentage of BrdU and Sox9 double-positive cells was not different between E12.5 wild-type and KO lungs (Fig. S3A).

We also immunohistochemically examined the level of phosphohistoneH3 (PH3), a mitotic marker for G2/M phases (Hendzel et al., 1997; Yin et al., 2008). At E12.5, PH3-positive cells were significantly reduced by 51% and 44%, respectively, in the mesenchyme and epithelium of mutant lungs. At E13.5, PH3-positive cells were decreased by 58% and 36%, respectively, in the mesenchyme and epithelium of mutant lungs (Fig. 3B).

Consistent with the decreases in BrdU- and PH3-positive cells in Nolz1 KO lungs, cyclin D1, a G1 cyclin protein required for the progression of the G1 phase (Baldin et al., 1993), encoded by Ccnd1, was reduced in the mesenchyme of mutant lungs at E12.5 and E14.5, but not in the mutant epithelium (Fig. 3C). Moreover, quantitative reverse transcription-polymerase chain reaction (qRT-PCR) showed that c-myc (Myc) mRNA, a key regulator of proliferation (Kioussi et al., 2002), was reduced by 43% in E12.5 mutant lungs compared with wild-type lungs (Fig. 3D). Note that qRT-PCR did not show changes in Ccnd1 mRNA in mutant lungs (Fig. S3B). As whole lung tissue was collected for qRT-PCR, the predominant expression of cyclin D1 in the epithelium could have overshadowed the low level of cyclin D1 in the mesenchyme.

Taken together, the reductions in BrdU-, PH3- and cyclin D1-positive cells and c-myc mRNA in KO lungs suggest that Nolz1 not only cell-autonomously regulates the proliferation of mesenchymal cells, but also non-cell autonomously affects epithelial cells at the mitotic phase in developing lungs.

The progenitor cells in lung mesenchyme give rise to different cell types, including smooth muscle cells and vascular endothelial cells. Immunostaining showed that smooth muscle actin (SMA; encoded by Acta2) expression was transiently decreased in Nolz1 KO lungs at E11.5. The SMA expression pattern was, however, not significantly altered at E12.5, E14.5, E16.5 and E18.5 in mutant lungs compared to that of wildtype lungs (Fig. S4A). The expression of Acta2 and SM22a mRNAs, progenitor markers of smooth muscle cells, was not changed in E12.5 mutant lungs (Fig. S4B). The expression pattern of PECAM1, a marker of vascular endothelial cells, appeared to be not changed in E13.5, E16.5 and E18.5 mutant lungs (Fig. S5A). Quantification of PECAM1 immunofluorescent intensity showed that there was no significant difference between wild-type and Nolz1 mutant lungs (Fig. S5B). Nor was Foxp1, which was expressed in the mesenchyme and distal parts of the epithelium (Shu et al., 2007), altered in E13.5 mutant lungs (Fig. S5C).

Immunostaining showed that the TTF1 expression was not changed in E11.5 Nolz1 KO lungs (Fig. 4A), but Foxp2, a marker of the distal part of epithelial cells (Shu et al., 2007), was ectopically expressed in the mesenchyme of Nolz1 mutant lungs (Fig. 4B). Notably, the ectopic expression of Foxp2 was domain-specific, i.e., Foxp2 was ectopically expressed in mesenchyme primarily surrounding the proximal parts of the epithelium, suggesting that Nolz1 suppresses Foxp2 in the mesenchyme. Sox9 is also known to be expressed in the distal epithelial progenitors and deletion of Sox9 affects lung branching (Rockich et al., 2013). However, the Sox9 expression pattern appeared to be unchanged in mutant lungs at E12.5 and E14.5 compared with wild type (Fig. 4C). The density of Sox9-positive cells was not altered in E12.5 mutant lungs (Fig. 4D).

We performed qRT-PCR to assay different differentiated cell types in E18.5 lung epithelium using the following epithelial cell markers. In alveolar epithelial type I cell (AEC1) cells: aquaporin 5 (Aqp5); T1α (Pdpn); in AEC2 cells: SpA (SftpA), SpB (SftpB) and SpC (SftpC); in Clara cells: CC10 (Scgb1a1); in ciliated cells: Foxj1. The results showed that Aqp5 and Pdpn mRNAs were decreased in E18.5 mutant lungs, whereas Foxj1 mRNA was increased in mutant lungs (Fig. 4E). SftpA, SftpB, SftpC and Scgb1a1 mRNAs were not changed in mutant lungs.

We further performed immunostaining and quantified the expression levels of Aqp5, T1α, CC10 and SPC in the lung tissue of E18.5 wild-type and Nolz1 KO mice. Consistent with the qRT-PCR results, the immunostaining showed that Aqp5 and T1α expression was decreased in Nolz1 KO lungs (Fig. 4F). CC10 and SPC expression was not changed in Nolz1 KO lungs, but collapsed CC10-positive epithelia were evident in KO lungs. These results suggest that Nolz1 may non-cell autonomously regulate the differentiation of epithelial cells.

Several signaling molecules are known to be important for lung development, including Wnts, Fgfs, Pdgfs, Shh, Bmps and Tgfβ2 (Maeda et al., 2007; Morrisey and Hogan, 2010; Warburton et al., 2000). Hypoplasia of mouse lung has previously been documented in Shh, Fgf9, Fgf10 and Wnt2 null mutant mice, and defective cell proliferation of epithelial and mesenchymal cells was also found in these mutant mice (Colvin et al., 2001; Goss et al., 2009; Pepicelli et al., 1998; Sekine et al., 1999). We examined whether these signal molecules were altered in E12.5 Nolz1 KO lungs using qRT-PCR. The results showed that Wnt2, Lef1, Fgf10, Gli3 and Bmp4 were decreased in mutant lungs. In contrast, Pdgfra and Tgfb2 were increased in mutant lungs. The expression of Wnt2b, Wnt7b, Axin2, Fgf9, Fgfr2, Pdgfc, Pdgfrb, Shh and Gli2 mRNAs were not changed in mutant lungs (Fig. 5A). These results suggest that Nolz1 may regulate Wnt, Shh, Bmp, Pdgf and Tgf signaling in developing lungs.

Previous studies have shown that Wnt2 regulates the proliferation of mesenchymal cells in developing mouse lungs (Goss et al., 2009). Because Wnt2 expression was drastically reduced by 66% in Nolz1 mutant lungs, we asked whether Wnt2 signaling was involved in the regulation of cell proliferation in Nolz1 mutant lungs. We validated the reduction in Nolz1 mutant lungs using in situ hybridization. Wnt2 mRNA was mainly expressed in the distal parts of mesenchyme of E11.5 and E13.5 wild-type lungs (Fig. 5B). Consistent with the result of qRT-PCR, Wnt2 mRNA was markedly decreased in the mutant mesenchyme.

The loss-of-function study showed that Wnt2 expression was decreased in the mesenchyme of Nolz1 mutant lungs. We then asked whether Nolz1 was sufficient to induce Wnt2 in lung mesenchymal cells. We cultivated primary mesenchymal cells from E14.5 wild-type mouse lungs. Cultured cells were electroporated with pcBIG-myc-Nolz1-ires-EGFP and pcBIG-ires-EGFP control plasmids (Fig. S6). Electroporated cells were cultivated for 48 h and were then harvested for the qRT-PCR assay. Nolz1 overexpression markedly increased Wnt2 mRNA by 60% compared with the mock control pcBIG-ires-EGFP. It also significantly increased Ccnd1 and c-myc mRNAs by 104% and 24%, respectively, compared with mock controls in mesenchymal cells (Fig. 5C). Collectively, the loss-of-function and gain-of-function studies suggest that Nolz1 is sufficient and required for the expression of Wnt2, Ccnd1 and c-myc in mesenchymal cells of developing lungs.

We further asked whether Nolz1 directly regulated Wnt2 in lung mesenchymal cells. Previous studies have identified the DNA sequence 5′-AGGAT-3′ as a Nolz1 putative binding motif in zebrafish (Runko and Sagerström, 2003; Siggers et al., 2014). We searched the 5′ flanking region of the Wnt2 mouse gene and found two AGGAT motifs at −788 (N1) and −2194 (N2) in the 5′ flanking regions (+1: translation start site) (Fig. 5D). We performed the chromatin immunoprecipitation (ChIP) assay in E15.5 wild-type and Nolz1 KO lung tissue using Nolz1 antibody. A strong ChIP-qPCR signal was detected at N1 and N2 loci in the wild-type lung as opposed to the weak signal in the Nolz1 KO lung (Fig. 5E). In another set of experiments, we electroporated pcBIG-myc-Nolz1-ires-EGFP into the primary culture of mesenchymal cells of E14.5 lungs. Two days after electroporation, cultured cells were harvested for the ChIP assay. A PCR band with the expected size of 187 bp was amplified from the N2 locus with the anti-myc epitope-tag antibody, but not with the control rabbit IgG (Fig. S7).

We then cloned the 187 bp DNA containing the N2 motif into the pGL3-basic luciferase (Luc) reporter plasmid. The pGL3-N2-Luc or its control pGL3-Luc plasmids were transfected into E14.5 primary culture of mesenchymal cells. Cells were cultured for 48 h before harvest for the reporter gene assay. The luciferase activity of the pGL3-N2 group was increased by 2.9-fold compared with the control pGL3 group (Fig. 5F). These results suggest that Nolz1 can bind to the AGGAT of the N2 motif in the promoter region of Wnt2 to transcriptionally activate the expression of Wnt2 in lung mesenchymal cells.

Previous studies have demonstrated that Wnt2/2b regulates lung development through the canonical Wnt/β-catenin pathway (Goss et al., 2011, 2009). Given that Wnt2 was a target gene of Nolz1, we asked whether Nolz1 was involved in the Wnt/β-catenin signal pathway. We intercrossed Nolz1^+/− mice with BATGAL or TOPGAL Wnt/β-catenin transgenic reporter mice to generate Nolz1^−/−;BATGAL or Nolz1^−/−;TOPGAL mice. β-Galactosidase (β-gal; encoded by lacZ) staining in the wholemount showed a reduction in β-gal signals in Nolz1^−/−;BATGAL and Nolz1^−/−;TOPGAL lung. In BATGAL lungs, β-gal signals were expressed in epithelial columns and the mesenchyme. β-Gal-positive signals were decreased in both the epithelium and mesenchyme of Nolz1^−/−;BATGAL mice (Fig. 5G). In TOPGAL lungs, β-gal signals were restricted only to epithelial columns. β-Gal signals were reduced in epithelial columns of Nolz1^−/−;TOPGAL mutant lungs (Fig. S8A). Furthermore, immunostaining of β-catenin showed decreased β-catenin immunoreactivity in E13.5 Nolz1 null mutant mesenchyme (Fig. S8B). Axin2 immunoreactivity was reduced in the distal parts of E13.5 Nolz1 mutant mesenchyme (Fig. 5H). However, Axin2 expression appeared to be normal in the mesenchymal regions surrounding the airways of mutant lungs (Fig. S8C).

We next examined whether Nolz1 activated Wnt/β-catenin signaling with the TOPFlash/FOPFlash assay in lung mesenchymal culture. TOPFlash and FOPFlash are luciferase reporters that contain normal and mutant TCF-binding sites, respectively. The ratio of TOPFlash/FOPFlash represents the specific activity of Wnt/β-catenin signaling. Overexpression of Nolz1 significantly increased the TOPFlash/FOPFlash ratio by 71% compared with the mock control (Fig. 5I).

The above findings suggest that Nolz1 may regulate mesenchymal cell proliferation through the canonical Wnt2/β-catenin signal pathway. As deletion of Nolz1 resulted in the reductions in Wnt2 expression and proliferation of mesenchymal cells, we asked whether exogenously applied Wnt2 protein could rescue defective proliferation of mesenchymal cells in Nolz1 KO lungs. We cultivated E12.5 lung explants with recombinant human Wnt2 protein (rWnt2, 150 ng/ml) for 48 h (Fig. 6A). Lung explant culture was pulse-labeled with BrdU for 1 h before tissue harvest (Fig. 6B). The number of epithelial branches at 48 h after culture was normalized with that at the beginning of culture (0 h). In wild-type lung explants, there was no difference between vehicle and rWnt2 treatments in epithelial branches. In Nolz1 KO lung explants, rWnt2 protein, but not heat-inactivated rWnt2 protein, markedly increased epithelial branches by 1.3-fold compared with the vehicle group (Fig. 6C). Treatments with rWnt2, but not heat-inactivated rWnt2, increased BrdU-labeled cells in the epithelium of Nolz1 KO lungs (Fig. 6D). rWnt2 treatments also increased BrdU-labeled cells in the mesenchyme of KO lungs. In wild-type lung explants, rWnt2 treatments did not increase BrdU-positive cells in the epithelium and mesenchyme (Fig. 6D). The endogenous Wnt2 protein in wild-type lung explants may have diminished the effects of exogenous Wnt2. Nolz1 deletion decreased the endogenous Wnt2 level, which may allow exogenous rWnt2 to promote the proliferation of mesenchymal cells and the growth of epithelial branches in mutant lung explants. The rescue experiment supports the causality that Nolz1 regulates mesenchymal cells and epithelial branches through Wnt2 signaling.

Finally, we attempted to identify upstream signaling molecules that regulate Nolz1 expression in developing lungs. We focused on Fgf9 and Fgf10, because Fgf9 and Fgf10 promote the proliferation of mesenchymal cells in the developing lungs (del Moral et al., 2006; Jean et al., 2008). Wild-type lung explant culture was treated with mouse recombinant FGF9 (rFgf9) or mouse recombinant FGF10 (rFgf10) for 48 h. Treatments with rFgf10 did not alter the epithelial structure, nor did they change Nolz1 and Wnt2 mRNAs (Fig. 7A). By contrast, rFgf9 treatments not only resulted in enlarged epithelia, but also significantly increased Nolz1, Wnt2 and Lef1 mRNAs compared with vehicle treatments. Consistently, western blotting showed a marked increase in Nolz1 protein with rFgf9 compared with the vehicle group (Fig. 7B).

We further determined whether Nolz1 expression could be reduced by inhibition of FGF signaling in lung culture. The treatment of SU5402 (Yi et al., 2009) resulted in significant decreases in Nolz1, Wnt2 and Lef1 expression in lung cultures as assayed by qRT-PCR (Fig. 7C). Taken together, these findings suggest that Fgf9 signaling acts upstream of Nolz1 in developing lungs. Note that rFgf9 could increase Wnt2 and Lef1 expression in Nolz1 KO lungs compared with wild-type lungs (Fig. 7D), suggesting that Fgf9 can promote Wnt2 transcription independently of Nolz1.

Nolz1 KO mice exhibited marked deficits in the development of the nervous system. Our previous study has shown that deletion of Nolz1 resulted in defective cell migration and differentiation of striatal neurons in the forebrain (Chen et al., 2020). As a consequence, the parcellation of the striatum into the dorsal and ventral divisions was impaired.

In the present study, we have identified Nolz1 as a mesenchyme-specific transcriptional regulator that plays an essential role in lung morphogenesis. Nolz1 regulates mesenchymal cell proliferation and the growth of epithelial branches. Loss-of-function and gain-of-function studies identified Wnt2 as a candidate molecule for mediating Nolz1 function in developing lungs. Furthermore, we showed that Fgf9 is an upstream regulator of Nolz1. We propose that the Fgf9-Nolz1-Wnt2 signal axis plays an important role in controlling lung morphogenesis (Fig. 7E).

Previous studies have shown that several morphogens, including Shh, Fgfs, Wnts and BMP proteins, can regulate the proliferation of mesenchymal cells in developing lungs (Morrisey et al., 2013). Epithelial Shh promotes cell proliferation and BMP4 expression in the mesenchyme during lung bud formation. Null mutation of Fgf9 derived from the epithelium and mesothelium causes defective cell proliferation in the lung mesenchyme (White et al., 2007, 2006). Epithelium-derived Wnt7b promotes cell proliferation in the mesoderm of the lung through Axin2 and Lef1 (Rajagopal et al., 2008; Shu et al., 2002). The Wnt2 null mutation results in significant lung hypoplasia due to decreased cell proliferation in both the epithelial and mesenchymal compartments (Goss et al., 2009). In most cases, these secreted morphogenic molecules regulate cell proliferation in both compartments of the epithelium and mesenchyme.

We have identified Nolz1 as an important transcription factor that regulates cell proliferation in the mesenchyme. In cell cycle progression, cyclin D1 marks proliferative cells in the transition of G1 to S phases. BrdU labels proliferative cells in the S phase. PH3 marks cells in the mitotic phase. Deletion of Nolz1 resulted in a decrease in PH3⁺ epithelial cells but no changes in cyclin D1⁺ and BrdU⁺ epithelial cells. These results suggest that Nolz1 is involved in cell cycle progression at the late phase but not the early phase. As Nolz1 is not expressed in the epithelium, the non-cell autonomous regulation of epithelial cells by Nolz1 may be different from the cell-autonomous regulation of mesenchymal cells by Nolz1.

Because Nolz1 was expressed in the lung primordium as early as E11.5, at which time lung morphogenesis begins to occur, Nolz1 may regulate morphogenic molecules, which in turn control cell proliferation in the lung mesenchyme. We examined Shh, Fgfs, Wnts and BMP signaling molecules to determine whether Nolz1 regulated cell proliferation through these morphogenic molecules. The qRT-PCR assay showed that the expression of Wnt2 and Fgf10 was decreased in E12.5 mutant lungs (Fig. 5A). The concurrent decrease in Wnt2 and Fgf10 expression was interesting, because Wnt2 null mutation has been shown to decrease Fgf10 in developing mouse lungs (Goss et al., 2011). Therefore, deletion of Nolz1 leads to Wnt2 reduction, which may in turn downregulate Fgf10 in the Nolz1 mutant lungs.

Our ChIP and reporter gene data suggest that Nolz1 transcriptionally regulates Wnt2 in developing lungs. We postulate that Nolz1 may regulate mesenchymal cell proliferation and epithelial branching through Wnt2 signaling. This hypothesis is supported by the rescue experiment in which the exogenous rWnt2 protein could partially rescue defective cell proliferation and epithelial branching in mutant lung culture (Fig. 6). Consistent with this hypothesis, the Nolz1 KO lung phenotypes were similar to those of Wnt2 KO mice (Goss et al., 2009). Wnt2 deletion induces lung hypoplasia in which reduced cell proliferation occurs in both the epithelial and mesenchymal compartments. In Wnt2/2b double KO mice, TTF1-positive lung endoderm progenitors are lost and the trachea and epithelial branches fail to develop in the mutant lungs. Note that deletion of Nolz1 selectively decreased Wnt2 but not Wnt2b in developing lungs. Wnt2 is expressed in the submesothelial layer at the periphery of the lung; however, Nolz1 expression appears evenly in the developing lung, suggesting that Nolz1 may function as a repressor by interacting with other cofactors in different cellular contexts in lung development (see below).

Deletion of Wnt2 also results in decreased SMA expression in smooth muscle cells, and Wnt2 can promote the development of smooth muscle of the lung airway through myocardin/Mrtf-B and Fgf10 (Goss et al., 2011). Notably, a transient decrease in SMA expression was detected in Nolz1 KO lungs at E11.5, although no significant changes in SMA immunoreactivity were found in later stages (Fig. S4). Given that Nolz1 positively regulates Wnt2, Nolz1 may act through Wnt2 signaling to regulate not only cell proliferation but also smooth muscle cell differentiation in the lung mesenchyme. Our study also showed that Nolz1 was engaged in canonical Wnt/β-catenin signaling in lung morphogenesis. The data of BATGAL and TOPGAL reporter mice and the TOPFlash/FOPFlash assay indicated that Nolz1 is engaged in canonical Wnt/β-catenin signaling in developing lungs. Wnt2 is known to act through the canonical Wnt/β-catenin pathway in lung mesenchymal cells (Goss et al., 2011; Yin et al., 2008). Nolz1 KO not only decreased Wnt2, but also reduced downstream target genes of canonical Wnt/β-catenin signaling, including Lef1, c-myc and Ccnd1 (De Langhe et al., 2008; Juan et al., 2014; Shtutman et al., 1999; Tetsu and McCormick, 1999; Zhang et al., 2012).

Previous studies have identified several transcription factors that are specifically expressed in the lung mesenchyme, including Foxf1, Tbx4/5 and Hox5 (Cardoso and Lü, 2006; Morrisey and Hogan, 2010). Deletion of Foxf1 inhibits mesenchymal cell proliferation and delays lung branching with altered expression of Wnt2, Hoxb7 and Wif genes (Ustiyan et al., 2018). Tbx4/5 mutants exhibit defective lung branching and downregulation of the mesenchymal markers of Wnt2 and Fgf10 and the epithelial markers of Bmp4 and Spry2 (Arora et al., 2012). Hoxa5;Hoxb5;Hoxc5 triple-mutant embryos show deficits in lung branching and proximal-distal patterning. Deletion of Hox5 leads to loss of Wnt2/2b expression in the distal lung mesenchyme and reduction of downstream Wnt2 signaling molecules of Lef1 and Axin2 in the mesenchyme and Bmp4 in the epithelium (Hrycaj et al., 2015).

It is of particular interest that Nolz1 KO lungs share some phenotypes, including decreased cell proliferation in lung buds and downregulation of mesenchymal Wnt2 signaling, with Foxf1, Tbx4/5 and Hox5 KO lungs, suggesting that Nolz1 may interact with these mesenchyme-specific transcription factors in the regulation of lung development. Indeed, we have found that the expression of Hoxa5 and Hoxb5 was reduced in Nolz1 KO lungs (S.-Y.C. and F.-C.L., unpublished).

Because Nolz1 was expressed as early as E11.5, when cell proliferation was activated by morphogen molecules, the earliest time point examined in the present study, the proliferative effects of morphogenic molecules may be mediated by Nolz1 signaling. Fgf9 is required for the proliferation of mesenchymal and epithelial cells (del Moral et al., 2006; White et al., 2006). A previous study has shown reductions in Wnt2 and cell proliferation in Fgf9 KO lungs and that transgenic overexpression of Fgf9 in the epithelium increases Wnt2 in the lung mesenchyme (Yin et al., 2008). Because Nolz1 is required for Wnt2 expression and cell proliferation in the lung mesenchyme, we postulate that Fgf9 acts upstream of Nolz1. Consistent with this hypothesis, treatments with rFgf9 significantly induced Nolz1, Wnt2 and Lef1 in E12.5 lung explant culture (Fig. 7B). In addition to Fgf9, Fgf10 appears to act upstream of Wnt2, as Wnt2 is decreased in the lung primordium of Fgf10 KO mice (Sekine et al., 1999). However, treatments with Fgf10 did not upregulate Nolz1 in the E13.5 lung explant culture (Fig. 7A). Therefore, we propose that Fgf9 secreted from the epithelium and mesothelium (White et al., 2006) may regulate Nolz1 expression in developing lung mesenchyme. Fgf9-Nolz1-Wnt2 signaling may represent a previously unreported axis in the regulation of lung morphogenesis (Fig. 7E).

The developing mouse lung contains two compartments, the epithelium and mesenchyme. The epithelium of the developing lung buds is derived from the endoderm, whereas the mesenchyme surrounding the buds is derived from the mesoderm at E9.5-E10 in mouse embryogenesis (Maeda et al., 2007; Morrisey and Hogan, 2010). Transplantation experiments have shown that cross-talk signaling between the epithelium and mesenchyme is important for branching morphogenesis and cyto-differentiation (Alescio and Cassini, 1962; Sakakura et al., 1976). The signaling molecules secreted from the mesenchyme regulate the growth and differentiation of epithelial cells and vice versa (Zepp and Morrisey, 2019). For example, Fgf10 released from the distal mesoderm acts on the epithelium through Fgfr2IIIb to promote cell proliferation and bud morphogenesis (De Moerlooze et al., 2000). Fgf9 is produced by the lung mesothelium and epithelium, and Fgf9 signaling through mesenchymal Fgfr1c and Fgfr2c maintains mesenchymal cell proliferation (Colvin et al., 2001; White et al., 2007). Wnt7b is expressed in the epithelium to promote cell proliferation in the mesoderm through Axin2 and Lef1 (Rajagopal et al., 2008; Shu et al., 2002) and the epithelial Shh promotes cell proliferation and Bmp4 expression in the mesenchyme. Induction of epithelial branching also requires the interaction of epithelial cells with mesenchymal cells. The co-culture experiment demonstrates that the extent of lung bud induction depends on the amount of co-cultured mesenchyme (Masters, 1976).

Although Nolz1 is specifically expressed in the compartment of mesenchymal cells, defective development of epithelial cells was observed in Nolz1 KO lungs. A decrease in AEC1 differentiation without altering AEC2 differentiation was observed in Nolz1 KO lungs. Recent studies have identified a population of bipotential progenitors that give rise to AEC1 and AEC2 cells (Volckaert and De Langhe, 2015; Whitsett et al., 2019). A possibility to account for the observation of a decrease in AEC1 differentiation without altering AEC2 differentiation is that there is a decrease in the differentiation of AEC1 cells from bipotential progenitors without affecting the differentiation of AEC2 cells in Nolz1 KO lungs. Aqp5 and T1α, markers of AEC1, were decreased, but Foxj1, a marker of ciliated cells, was increased in Nolz1 mutant lungs. Foxp2, a marker of the distal part of epithelial cells (Shu et al., 2007), was ectopically expressed in the mesenchyme of E12.5 Nolz1 KO lungs. The non-cell-autonomous effects of Nolz1 mutation may result from defective signal transduction across the mesenchymal and epithelial compartments. As early as E12.5, a reduction in epithelial proliferation was found in Nolz1 mutant lungs. It is of interest that defective epithelial proliferation also occurs in Wnt2 and Fgf10 mutant mice (Goss et al., 2011, 2009; Sala et al., 2011). Because deletion of Nolz1 downregulated Wnt2 and Fgf10, it is possible that the comprised Wnt2 and Fgf10 signaling in the mesenchymal compartment may non-cell autonomously affect epithelial proliferation in Nolz1 mutant lungs. This possibility awaits to be examined in future studies.

Previous biochemical studies have revealed that mouse Nolz1 and its homologs, zebrafish Nlz1 and Nlz2 and chick Nolz1, can interact with the Groucho-TLE co-repressor to transcriptionally repress gene expression in neurons and optic pigment cells (Brown et al., 2009; Ji et al., 2009). We have also shown that Nolz1 transcriptionally represses Dlx1/2 genes in developing striatal neurons (Chen et al., 2020). Our present study of Nolz1 in developing mouse lungs, however, indicates that Nolz1 functions as a transcriptional activator to regulate Wnt2 expression. Note that the pGL3-N2-Luc reporter gene activity was increased in lung mesenchymal cells, but was suppressed in HEK293T cells and embryonic sarcoma-derived C3H 10T1/2 cells (S.-Y.C. and F.-C.L., unpublished). These results suggest that Nolz1 can function as a cell context-dependent transcriptional regulator. Nolz1 may, in conjunction with different co-factors, regulate gene expression in different cell types.

Our study shows that Nolz1/Znf503 regulates mesenchymal cell proliferation in lung morphogenesis. The ability of Nolz1 to promote cell proliferation suggests a pathological role for Nolz1 in tumorigenesis. Indeed, previous studies have shown that NOLZ1/ZNF503 promotes the proliferation and invasion of mammary epithelial cells. ZNF503 enhances aggressive breast cancer progression by repression of GATA3 (Shahi et al., 2015, 2017) and promotes the migration, invasion and epithelial-mesenchymal transition process of hepatocellular carcinoma cells by suppressing GATA3 (Yin et al., 2019). ZNF503 is also involved in lung cancer.

ZNF503 mRNA level is upregulated in non-small cell lung cancer (NSCLC). Furthermore, microRNA and long noncoding RNA (lncRNA) can regulate tumorigenesis through ZNF503. For example, miR-340-5p acts through ZNF503 to inhibit proliferation and invasion in an NSCLC cell line (Lu and Zhang, 2019). The TTN-AS1 lncRNA promotes the progression of NSCLC via the miR-491-5p/ZNF503 axis (Qi and Li, 2020). Therefore, Nolz1 appears to play a pathogenic role in the regulation of abnormal cell proliferation in tumorigenesis.

It is of interest that the mouse Nolz1 gene is one of the evolutionarily conserved genes that have high degrees of conservation of the 5′ flanking promoter regions in different species (Chang et al., 2013, 2011). It implies that the expression pattern of Nolz1 in different organs may be conserved between different species. Consistent with this hypothesis, a previous study has shown that the Drosophila homologs of Nolz1, elB/noc, are essential for the morphogenesis of the dorsal tracheal branches during development (Dorfman et al., 2002). ElB is expressed specifically in the dorsal branches of the trachea. The elB mutation induces abnormal migration of the dorsal branches, whereas the misexpression of elB in the trachea causes loss of the visceral branch and increased dorsal branch. As a mammalian member of the NET family, Nolz1 plays an essential role in lung morphogenesis. Therefore, Nolz1 and its homologs are part of transcriptional regulator toolkits that are essential for the development of respiratory organs.

The BATGAL mice (stock number 005317) and TOPGAL mice (stock number 004623) were obtained from The Jackson Laboratory. Nolz1^+/− mice (Chen et al., 2020) and BATGAL and TOPGAL transgenic reporter mice were housed in a 12 h light/12 h dark cycle room at the Animal Center of National Yang Ming Chiao Tung University (NYCU). The protocol of animal use was approved by the Institutional Animal Care and Use Committee of NYCU. The day the mating plug was found was identified as E0.5.

The genotypes of transgenic mice were identified by PCR with genomic DNA. Genomic DNA was prepared according to the method described at The Jackson Laboratory (http://jaxmice.jax.org). Briefly, the toes or tails of mice were lysed in NaOH buffer (25 mM NaOH, 0.2 mM EDTA) and heated to 98°C for 30 min and then cooled down to room temperature. An equal volume of 40 mM Tris-HCl (pH 5.5) was added to the lyse buffer to neutralize NaOH. After spinning down briefly, the genomic DNA was collected and used as templates for PCR reaction with pairs of genotype-specific primers in the following: Nolz1 floxed-Forward (5′-CCAATGTGGAAAGATAGTAGCC-3′), Reverse (5′-TCCAGCAGGAAGAAGACAGG-3′); Nolz1 null-Forward (5′-GTCGCCTTCCTCAGAAGCTA-3′), Reverse (5′-GATCTGGACAGGGAAAAGCA-3′); lacZ Forward (5′-CGTGGCCTGATTCATTCC-3′), Reverse (5′-ATCCTCTGCATGGTCAGGTC-3′).

Time-pregnant mice or adult mice were anesthetized with isoflurane. Lung tissues were dissected from E11.5, E12.5, E13.5, E14.5, E16.5, E18.5, postnatal day (P) 0, P7, P14 and adult mice. The surrounding organs were removed from dissected lung tissue. The dissected E11.5-E13.5 lung tissues were placed on top of the culture insert (0.4 μm pore size, PIHP01250, Merck Millipore). Explants were cultured with the BGjb medium (12591038, Thermo Fisher Scientific) supplemented with ascorbic acid (0.1 mg/ml, A92902, Sigma-Aldrich) for 48 h. For the rescue experiments, lung tissues were dissected from E11.5-E12.5 wild-type or Nolz1 null mutant embryos. The lung explants were cultured with recombinant human WNT2 protein (rWnt2; 150 ng/ml, H00007472-P01, Novus Biologicals). rWnt2 protein was inactivated by heating at 80°C for 10 min. The heat-inactivated rWnt2 protein was used as a control for the biological activity of the rWnt2 protein. For the induction of Nolz1 experiments, the lung tissues were dissected from E13.5 wild-type embryos. The lung explants were cultivated with recombinant mouse Fgf9 protein (200 ng/ml, 7399-F9-025, R&D Systems) or Fgf10 protein (200 ng/ml, 6224-FG-025, R&D Systems) or the Fgf9 inhibitor, SU5402 (10 μM, SML0443, Merck). Lung explants were cultured for 48 h, with the change of fresh medium containing the recombinant proteins 24 h after cultivation. For each experimental condition, at least four lung explants from different embryos were used.

The culture of lung mesenchymal cells was performed according to Lebeche et al. (1999). The lungs of E14.5 littermate embryos were dissected and were then digested with collagenase solution containing 0.1% collagenase (C9697, Sigma-Aldrich) and 0.01% DNAase I (D5025, Sigma-Aldrich) in PBS at 37°C for 30 min. Enzymatic digestion was stopped by adding 2% fetal bovine serum (FBS; 16000044, Gibco). The digested lung tissue was triturated by pipetting 30 times with pipet-aid. The dissociated cells were spun in a KUBOTA 2010 centrifuge at 1000 rpm (167 g) for 5 min at room temperature, and the cell pellet was resuspended in 10 ml of DMEM (12100046, Gibco) supplemented with 2% FBS. The cells were plated with the density of 2×10⁶ cells in six-well dishes, and were then incubated with 5% CO₂ at 37°C for differential cell adhesion. After 50 min of incubation, the medium containing epithelial cells was removed by washing twice with PBS followed by adding fresh medium while the mesenchymal cells remained adherent to the culture plates. The enrichment of mesenchymal cells in the culture was validated by immunostaining of Nolz1 and TTF1. The immunostaining indicated that less than 1% of the cultured cells were TTF1-positive epithelial cells and 97% of the cultured cells were Nolz1-positive mesenchymal cells.

E14.5 lung tissue was dissociated into single cells (5×10⁶). Cells were resuspended in 50 μl of electroporation solution containing 6 μg DNA plasmids, 2 mM CaCl₂ and 55 mM glucose in PBS and were then transferred to an electroporation cuvette (2 mm, BTX). Electroporation was performed using an electroporator (170 V, 5 ms of duration, one pulse, BTX ECM 830). Electroporated cells were plated at the density of 4×10⁶ cells in six-well plates. The cells were cultured to enrich mesenchymal cells as described above. The cells were harvested 48 h after seeding for qRT-PCR analysis and ChIP assay.

E11.5, E12.5, E13.5, E14.5 whole embryos and E16.5, E18.5 whole lungs were fixed with 4% paraformaldehyde (PFA) overnight and were embedded with paraffin before sectioning at 5 μm. For frozen sections, after cryoprotection with 30% sucrose for 2 days, embryos were sectioned using a cryostat (Leica, CM3050 S) at the thickness of 10 μm. For histological analysis, sections were stained with H&E (MHS16, HT110216, Sigma-Aldrich). For immunohistochemistry, paraffin-embedded sections were dewaxed by heating to 65°C for 30 min and then immersed in xylene for 3×10 min. The dewaxed sections were rehydrated by successive immersion in 100% ethanol for 5 min, 80% ethanol for 5 min and 70% ethanol for 5 min followed by rinses in PBS. For cryosections, antigen retrieval was performed as needed. Briefly, sections were incubated with retrieval buffer [10 mM citrate acid (pH 6.0), 0.05% Tween 20] at 95°C for 10 min. For immunohistochemistry, the sections were pre-treated with 0.2% Triton X-100 in 0.1 M PBS and 10% methanol, 3% H₂O₂. After washing with PBS, sections were incubated in a blocking solution containing 3% serum in 0.1 M PBS for 1 h. The sections were incubated with primary antibodies diluted in 0.1 M PBS containing 1% serum, 0.2% Triton X-100 and 0.1% sodium azide overnight at room temperature or for 2 days at 4°C. The following primary antibodies were used: rabbit anti-Nolz1 (1:1000; Ko et al., 2013), mouse anti-TTF1 (1:500, MS-699-P0, NeoMarkers), rabbit anti-Sox9 (1:2000, ab185966, Abcam), rabbit anti-Axin2 (1:500, ab107613, Abcam), mouse anti-β-catenin (1:1000, 610153, BD Biosciences), rat anti-BrdU (1:1000, OBT0030G, Accurate Chemical), mouse anti-Ki67 (1:500, 550609, BD Pharmingen), rabbit anti-PH3 (1:1000, 06-570, Merck Millipore), rabbit anti-active Caspase3 (AC3) (1:500, 9661, Cell Signaling Technology), mouse anti-cyclin D1 (1:1000, sc-8396, Santa Cruz Biotechnology), rabbit anti-SPC (1:500, sc-13979, Santa Cruz Biotechnology), mouse anti-CC10 (1:500, sc-130411, Santa Cruz Biotechnology), rabbit anti-Aquaporin 5 (1:500, ab78486, Abcam), hamster anti-T1α (1:500, ab11936, Abcam), mouse anti-SMA (1:2000, A2547, Sigma-Aldrich), mouse anti-SM22α (1:500, sc-53932, Santa Cruz Biotechnology), rabbit anti-PECAM1 (1:500, A19014, ABclonal) and mouse anti-myc (1:1000, 05-724, Merck Millipore), rabbit anti-Foxp1 (1:2000, ab16645, Abcam), rabbit anti-Foxp2 (1:2000, ab16046, Abcam). Secondary antibodies conjugated with chromophore Alexa-488 [1:500, donkey anti-mouse (A-21202), donkey anti-rabbit (A-21206), Thermo Fisher Scientific], with Alexa-555 [1:500, donkey anti-mouse (A-31570), donkey anti rabbit (A-31572), goat anti-hamster (A78959), goat anti-rat (A-21434), Thermo Fisher Scientific] or with Biotin (1:500, donkey anti-mouse, 715-065-150; donkey anti-rabbit, 711-065-152; donkey anti-rat, 712-065-150, Jackson ImmunoResearch Laboratories) were used. For immunostaining of Nolz1, TTF1, AC3 and Ki67, immunostaining signals were further amplified by tyramide-FITC or tyramide-Cy3 (NEL753001KT, Perkin-Elmer), 3-Amino-9-ethylcarbazole (AEC, A5754, Sigma-Aldrich) or 3,3′-Diaminobenzidine (DAB, D12384, Sigma-Aldrich) after avidin-biotin complex incubation (Elite ABC kit, PK-6100, Vector Laboratories). Immunostained sections were counterstained with DAPI (1 μg/ml, D9542, Sigma-Aldrich) or Hematoxylin (H).

E14.5 Nolz1^+/+;BATGAL, Nolz1^−/−;BATGAL, Nolz1^+/+;TOPGAL and Nolz1^−/−;TOPGAL lungs were dissected in ice-cold PBS, briefly washed in PBS twice and fixed with 4% paraformaldehyde for 1 h. The tissues were then pre-incubated with β-gal staining solution (150 mM NaCl, 1 mM MgCl₂, 3.5 mM potassium ferrocyanide, 3.5 mM potassium ferricyanide, 0.3 mM chloroquine, 0.01% sodium deoxycholate, 0.2% NP-40 in 0.1 M PBS for 1 h at 37°C followed by incubation in β-gal staining solution containing X-gal (1 mg/ml, B4252, Sigma-Aldrich) for 16 h at 37°C. After the whole-mount β-gal-stained lungs were photographed, tissues were cryosectioned for further analysis.

E11.5 and E13.5 lung sections were mounted on the slides coated with gelatin (G2500, Sigma-Aldrich) and poly-D-lysine (P6407, Sigma-Aldrich). The pCMV-Sport6-Wnt2 plasmid (NCBI accession number BC026373.1; IMAGE:4162686) was linearized with the restriction enzyme EcoRI-HF (#R3101, New England Biolabs) to generate the antisense probe. The digoxigenin-labeled antisense cRNA probe was synthesized by in vitro transcription with T7 polymerase (P2075, Promega), RNA labeling mix (11277073910, Roche) and the linearized pCMV-Sport6-Wnt2 plasmid. After fixation with 4% PFA and rinsing with PBS, the sections were treated with 0.2 N HCl and proteinase K (10 μg/ml, P2308, Sigma-Aldrich), and were then prehybridized with 50% formamide in 2× SSC for 90 min at 65°C. The sections were hybridized with the Wnt2 cRNA probe solution [10.6% dextran sulfate, 53% formamide, 1 mM EDTA, 10.6 mM Tris-HCl (pH 8.0), 318 mM NaCl, 1.06× Denhart's solution, 200 μg/ml tRNA and 10 mM DTT] for 16 h at 65°C. Sections were then washed with 2× SSC containing 50% formamide for 1 h at 65°C, and were then treated with RNase A (20 μg/ml, R6513, Sigma-Aldrich). The sections were washed with 2× SSC and 0.2× SSC for 20 min followed by incubation in a blocking solution containing 2% blocking reagent (11096176001, Roche) and 20% sheep serum (16070096, Thermo Fisher Scientific) for 1 h. The sections were incubated with horseradish peroxidase (HRP) conjugated sheep anti-digoxigenin antibody (1:100, 11633716001, Roche) overnight at room temperature. The signals were detected using a TSA amplification kit (NEL753001KT, PerkinElmer) following the manufacturer's protocol. Sections were counterstained with DAPI.

The pGEM-T-Nolz1 plasmid was linearized with SalI to generate antisense probes. The production of cRNA probes was carried out in the presence of ³⁵S-UTP (NEG039H250UC, PerkinElmer) with T7 RNA polymerase by in vitro transcription. Brain sections were incubated with 10 μg/ml proteinase K at 37°C for 5 min, RNA hybridization was performed with the ³⁵S-UTP-cRNA probe (107 cpm/ml) in the hybridization buffer containing 50% formamide, 10% dextran sulfate, 0.3 M NaCl, 1× Denhart's solution, 0.01 M Tris (pH 8.0), 1 mM EDTA. 0.5 mg/ml yeast tRNA and 1 mM DTT at 58°C for 16 h. After several rinses in 4× SSC, the slides were treated with RNase A (10 μg/ml) and washed with 2× SSC, 1× SSC, and 0.5× SSC at room temperature. The slides were then washed with 0.1× SSC at 50°C for 30 min followed by another 0.1× SSC wash at room temperature for 5 min. Ethanol dehydration was carried out with serial rinses with 50%, 75%, 95% ethanol, and then 100% ethanol twice, each for 3 min. The slides were vacuum-dried for 1 h, and were exposed to X-ray film. Signals were detected by autoradiography.

Time-pregnant mice were injected with BrdU (50 mg/kg body weight, B5002, Sigma-Aldrich) 1 h before the harvest of embryos. Embryos were sectioned and processed for immunohistochemistry as described above. The lung explant culture was incorporated with BrdU (10 µM) for 1 h and then harvested for further analysis.

The TUNEL assay was performed in cryosectioned lung tissue using the In Situ Cell Death Detection Kit (11684795910, Roche) according to the manufacturer's instructions. Briefly, sections were treated with 0.1 M glycine in PBS buffer for 30 min and then incubated with 0.1% Triton X-100 in 0.1% sodium citrate buffer for 3 min at 4°C. Following 3×5 min rinses with PBS, the sections were incubated with the solution containing 10% terminal transferase (TdT) for 1 h at 37°C to label the free 3′-OH ends in fragmented DNA with fluorescein-conjugated dUTP. The sections were counterstained with DAPI.

Lung tissues and cultured cells were lysed in Trizol reagent (15596018, Life Technologies), followed by RNA extraction. A two-step quantitative reverse transcription-polymerase chain reaction (qRT-PCR) was performed. Total RNA (1 μg) was mixed with oligo-dT₁₅ (10814270001, Roche) and dNTP (FYT013, Yeastern Biotech) and then denatured at 65°C for 5 min. After the addition of 0.1 M DTT, RNase inhibitor (RNasin, N2111, Promega), 5× buffer, and SuperScript III transcriptase (18080093, Thermo Fisher Scientific), reverse transcription of RNA into cDNA was carried out at 50°C for 2 h. The reaction was stopped by incubation at 70°C for 15 min. The cDNA was used for the real-time PCR reaction (Applied Biosystems StepOne system). The PCR reaction solution contained 1/10 cDNA and 2× TaqMan Fast Universal PCR Master Mix (A46109, Thermo Fisher Scientific). PCR reactions were carried out under the following conditions: 95°C for 10 min followed by 40 cycles at 95°C for 10 s, 60°C for 1 min with the following sets of PCR primers: Aqp5-Forward (5′-ATGAACCCAGCCCGATCTTT-3′), Reverse (5′-ACGATCGGTCCTACCCAGAAG-3′); Axin2-Forward (5′-CTCCCCACCTTGAATGAAGA-3′), Reverse (5′-ACATAGCCGGAACCTACGTG-3′); Bmp4-Forward (5′-GCCATTGTGCAGACCCTAGT-3′), Reverse (5′-ACCCCTCTACCACCATCTCC-3′); c-myc-Forward (5′-TCGTGAGAGTAAGGAGAA-3′), Reverse (5′-CAAGGTTGTGAGGTTAGG-3′); cyclin D1-Forward (5′-TGTTCGTGGCCTCTAAGATGAAG-3′), Reverse (5′-AGGTTCCACTTGAGCTTGTTCAC-3′); Fgf9-Forward (5′-TGCAGGACTGGATTTCATTTAGAG-3′), Reverse (5′-CGAAGCGGCTGTGGTCTTT-3′); Fgf10-Forward (5′-GAGAAGAACGGCAAGGTCAG-3′), Reverse (5′-CCCCTTCTTGTTCATGGCTA-3′); Fgfr2-Forward (5′-AAGATGATGCCACAGAGA-3′), Reverse (5′-CCAGGAGGTTGATAATGTTC-3′); Foxj1-Forward (5′-TGGACTGGCTCCATTTTA-3′), Reverse (5′-ATGCATTTAGACACCGGA-3′); GAPDH-Forward (5′-CGTGGAGTCTACTGGTGTCTTC-3′), Reverse (5′-TGCATTGCTGACAATCTTGAG-3′); Gli2-Forward (5′-TCCCTCAGCAAATGGAAGTTG-3′), Reverse (5′-CCGTGCTCCCGTTGATG-3′); Gli3-Forward (5′-GTGTGACTTCTCCTCTTA-3′), Reverse (5′-GCATCCTTCCTATTACCT-3′); Lef1-Forward (5′-TTACTCTGGCTACATAATGAT-3′), Reverse (5′-ACGGGCACTTTATTTGAT-3′); Nolz1-Forward (5′-TTACTCTGGCTACATAATGAT-3′), Reverse (5′-GGCTCCTTCTTATCTGAACC-3′); Pdgfra-Forward (5′-TGGCATGATGGTCGATTCTA-3′), Reverse (5′-CGCTGAGGTGGTAGAAGGAG-3′); Pdgfrb-Forward (5′-CCGGAACAAACACACCTTCT-3′), Reverse (5′-TATCCATGTAGCCACCGTCA-3′); Pdgfc-Forward (5′-CTTAGTTGTCTTGATATGG-3′), Reverse (5′-GAGCATCTGTCTATCTAT-3′); Scgb1a1-Forward (5′-TCCTAACAAGTCCTCTGTGTAAGA-3′), Reverse (5′-AGGAGACACAGGGCAGTGACA-3′); Shh-Forward (5′-CCCAATTACAACCCCGACAT-3′), Reverse (5′-GTCTTTGCACCTCTGAGTCATCA-3′); Acta2-Forward (5′-TGGCATCAATCACTTCAA-3′), Reverse (5′-CCTATCTGGTCACCTGTA-3′); SM22a-Forward (5′-GCCCAGCCTCTACATCTT-3′), Reverse (5′-GAATGCTAACAGGAGTGACAA-3′); SftpA-Forward (5′-CTCCAGACCTGTGCCCATATG-3′), Reverse (5′-ACCTCCAGTCATGGCACAGTAA-3′); SftpB-Forward (5′-ACGTCCTCTGGAAGCCTTCA-3′), Reverse (5′-TGTCTTCTTGGAGCCACAACAG-3′); SftpC-Forward (5′-ACCCTGTGTGGAGAGCTACCA-3′), Reverse (5′-TTTGCGGAGGGTCTTTCCT-3′); Pdpn-Forward (5′-AGGTACAGGAGACGGCATGGT-3′), Reverse (5′-CCAGAGGTGCCTTGCCAGTA-3′); Tgfb2-Forward (5′-TATTGATGGCACCTCTAC-3′), Reverse (5′-ACAACATTAGCAGGAGAT-3′); Wnt2-Forward (5′-TCTTGAAACAAGAATGCAAGTGTCA-3′), Reverse (5′-GAGATAGTCGCCTGTTTTCCTGAA-3′); Wnt2b-Forward (5′-GCAGACAACAGACTAGATTC-3′), Reverse (5′-TACGATTGGATGAAGAGGAA-3′); Wnt7b-Forward (5′-GCAGTGTGGATGGATGTT-3′), Reverse (5′-GGCTACCCAGTCGTTAGTA-3′). The Ct value of genes were normalized to the GAPDH internal control.

The lung tissues from E12.5, E14.5, E16.5, E18.5 embryos and P0, P7, P14 and adult mice were lysed in RIPA buffer [20 mM HEPES (pH 7.8), 150 mM NaCl, 1 mM EDTA, 1% NP-40, protease inhibitor cocktail (04693132001, Roche)]. Tissue lysates (20 μg) of embryonic lungs were subjected to SDS-PAGE and transferred to a polyvinylidene difluoride membrane (PVDF; IPVH00010, Merck Millipore). Immunoblotting was performed using the anti-Nolz1 antibody (1:1000) and mouse anti-actin (1:10,000, A1978, Sigma-Aldrich). Secondary antibodies were HRP-conjugated goat anti-rabbit IgG and goat anti-mouse IgG (1:10,000, 111-035-003, 115-035-003, Jackson ImmunoResearch Laboratories). For testing the specificity of the Nolz1 antibody, it was incubated with 50 μg Nolz1 recombinant protein overnight at 4°C. The preabsorbed antibodies were then used for immunoblotting. Immunoblotting ECL (PK-NEL113, PerkinElmer) signals were detected with a Luminescence Imaging System (Fuji Film, LAS-4000).

For the lung tissues, E15.5 wild-type and Nolz1 mutant lungs were collected for ChIP assay according to the manufacturer's instructions (EZ-Magna ChIP A/G kit, 17-10086, Merck Millipore). In brief, lung tissues were fixed with 1% formaldehyde for 20 min at room temperature followed by quenching formaldehyde with 1/20 volume of 2.5 M glycine. After washing twice with 5 ml of PBS, lung tissues were homogenized with 0.5 ml of cell lysis buffer containing 2.5 μl of protease inhibitor cocktail II (20-283, Merck Millipore). Dissociated cells were spun down at 800 g for 5 min at 4°C. The pellet was resuspended in 0.5 ml nuclei lysis buffer and subjected to sonication to shear DNA with a sonicator (30 s on, 30 s off, 15 cycles, two rounds, Bioruptor UCD-200). The debris was removed using a KUBOTA 3300 centrifuge at 13,000 rpm (15,400 g) for 10 min at 4°C. Rabbit anti-Nolz1 antibody (5 μg) (Chen et al., 2020) or rabbit IgG (5 μg) were incubated with protein A/G Magna beads (20 μl) and the sheared DNA lysate overnight at 4°C. Antibody-DNA complexes were washed sequentially with low salt, high salt, LiCl and TE buffers followed by incubation in the ChIP elution solution containing proteinase K at 65°C for 2 h. DNA purification was performed using spin columns. The purified DNA products were used as templates for PCR reaction with N1 primers [Forward (5′-GAGGAGCTGTGTGTGCGTAA-3′), Reverse (5′-CCAGTGTCTGACCAGGTTGA-3′)] and N2 primers [Forward (5′-CAGGTCAGAAGGGGTGTTTG-3′), Reverse (5′-GCCTCCTAGAAAAGGGATGG-3′)]. The qPCR reactions were performed under the following conditions: 95°C for 10 min followed by 40 cycles at 95°C for 10 s, 60°C for 1 min by Applied Biosystems StepOne system.

For the primary mesenchymal culture, the pcBIG-myc-Nolz1-ires-EGFP plasmids were electroporated into the lung mesenchymal cell culture. The ChIP assay was performed as described above. Mouse anti-myc antibody (1 μg) was used for capturing myc-Nolz1 and mouse IgG (1 μg) was used as the control group. The purified DNA products were used as templates for PCR reactions with N1 and N2 primers. PCR was performed under the following reaction conditions: 95°C for 10 min followed by 30 cycles at 95°C for 10 s, 58°C for 10 s and 72°C for 20 s. An additional 72°C was run for 3 min at the end of the PCR reaction. The PCR amplicons of the ChIP assay were analyzed by 2% agarose gel electrophoresis.

The genomic region of the mouse Wnt2 gene (NM_023653.5) containing the putative Nolz1 binding motif N1 (−722 to −904, +1 translation start site) and the N2 motif (−2149 to −2335) was amplified by PCR with the N1-ChIP and N2-ChIP primers. The amplicons of N2 (187 bp) were constructed into the pGL3-basic plasmid (E1751, Promega).

Primary lung mesenchymal cells were cotransfected with pGL3-basic or pGL3-N2 and pGL4.74 Renilla luciferase plasmids (E6921, Promega) at a ratio of 1:0.1 (4 μg/six-well dishes) with Lipofectamine LTX PLUS (15338100, Thermo Fisher Scientific). The luciferase activity was assayed using the Dual-Luciferase Reporter Assay Kit (E1910, Promega). The luciferase signals were detected with a luminescence reader (VICTORTM ×2, PerkinElmer). The TOPFlash or FOPFlash plasmids (gifts from Dr C.-M. Chen of NYCU, Taiwan) were cotransfected with pcBIG-myc-Nolz1-ires-EGFP (or mock control pcBIG-ires-EGFP) and pGL4.74 Renilla luciferase plasmids in a ratio of 0.5:1:0.1 (4 μg/six-well dishes) using Lipofectamine LTX PLUS. The luciferase activity was assayed using the Dual-Luciferase Reporter Assay Kit. The luciferase signals were detected with a luminescence reader (VICTORTM ×2, PerkinElmer).

Photomicrographs of the dissected lungs were taken under a dissecting microscope (Nikon, SMZ800) with the aid of a digital camera (Nikon, DXM 1200C). Photomicrographs of lung histology were acquired with the aid of fluorescence microscopes (BX51, BX53, BX63, Olympus) and a confocal microscope (Zeiss, LSM700). Photomicrographs of lung sections and lung explant cultures were selected for morphological quantification. At least three pairs of wild-type and Nolz1 null mutant lungs from independent littermates were used for analysis. The number of epithelial branches of the left lobe in lung explants at 48 h after culture was normalized with the number of epithelial branches at 0 h. The numbers of Nolz1⁺, Ki67⁺, PECAM1⁺, SM22α⁺, Sox9⁺, BrdU⁺, PH3⁺, cyclin D1⁺ and DAPI⁺ cells and the immunofluorescent intensity of Aqp5, T1α, CC10, SPC and PECAM1 were quantified using ImageJ software (National Institutes of Health). Mesenchymal cell density and the epithelial tubule areas in E14.5 lungs and the distance between sacs in E18.5 lungs were also quantified by ImageJ. The intensity of PCR amplicons of the ChIP assay was quantified by ImageJ. The data for statistical analysis were first assessed for normality. For the normally distributed data, unpaired two-tailed Student's t-test and one-way ANOVA with Tukey multiple comparison tests was used for statistical analyses. Error bars represent mean±s.e.m. for at least three independent experiments. GraphPad Prism 8 was used for making graphs and calculating statistical significance.

We thank Dr C.-M. Chen and Dr L.-R. You for teaching embryonic lung dissection, providing reagents, transgenic mice and insightful discussion, T.-H. Kuo for help in data analysis and the NYMU Genomic Research Center for providing cDNA clones.