Fgf10 expression identifies parabronchial smooth muscle cell progenitors and is required for their entry into the smooth muscle cell lineage

Despite the wealth of knowledge about the origin and differentiation of lung epithelial cells, extremely little is known about the origin and differentiation of the various lung mesenchymal cell types, i.e. the parabronchial and alveolar smooth muscle cells (SMCs), the endothelial cells,the pericytes, lipocytes and stromal fibroblasts. Yet, insight into the origin and generation of these cell types is extremely important, as they are crucial for normal respiratory function. This is illustrated by an overproliferation of bronchial SMCs associated with asthma (reviewed by Lazaar, 2002) and broncho-pulmonary dysplasia (reviewed by Hershenson et al., 1997). In addition, overproliferation of lung fibroblasts causes the respiratory problems associated with lung fibrosis(Raghu et al., 1998).

Given the significance of the distal lung tips for the establishment of lung epithelial lineages (reviewed by Warburton et al., 2000), we considered the possibility that the distal tips have a similar significance for the lung mesenchymal lineages. We focused in particular on a putative role for fibroblast growth factor 10 (FGF10) in this process, as it is specifically expressed in the distal lung mesenchyme(Bellusci et al., 1997b). Homozygous null mutants for Fgf10 display complete lung agenesis(Min et al., 1998; Sekine et al., 1999), and can therefore not be used to study the role of Fgf10 in the establishment of lung mesenchymal lineages. Instead, we took advantage of the recently reported Mlcv1v-nLacZ-24 transgenic mouse strain(Kelly et al., 2001). In this strain the transgene was inserted 120 kb upstream of the Fgf10 gene,and due to positional effects, LacZ expression appeared to be a true reporter for Fgf10 expression in the developing heart(Kelly et al., 2001). Given that tissue-specific enhancer sequences drive tissue-specific gene expression,it remains to be determined whether this mouse strain is also a useful reporter for Fgf10 expression in other organs, in this case the lung.

Here, we first demonstrate that LacZ expression in Mlc1v-nLacZ-24 mice faithfully mimicked Fgf10 expression in the developing lung, and excluded a contribution of the Mlc1vpromoter sequences to LacZ expression. Analyzing the expression profile of Mlc1v-nLacZ-24 (hereafter named Fgf10^LacZ for simplicity), we show that Fgf10-positive cells in the distal mesenchyme were progenitors for the parabronchial SMCs. Next, we demonstrate that the integration of the transgenic cassette resulted in decreased expression from the Fgf10gene. We obtained Fgf10 hypomorphic mutants, by generating heterozygous Fgf10 embryos that were hemizygous for transgenic insertion: Fgf10^+/–;Mlc1v-nLacZ-24^+/–(hereafter called Fgf10^LacZ/–). In Fgf10^LacZ/– lungs, part of theβ-galactosidase/Fgf10-positive cells failed to leave the distal tip mesenchyme to become SMCs. Therefore, we postulate that Fgf10-positive cells in the distal lung mesenchyme are progenitors for parabronchial SMCs and, moreover, that Fgf10 is required for their entry into the smooth muscle cell lineage.

Whole-mount in-situ hybridization protocols were based on previously described methods (Winnier et al.,1995). The following mouse cDNAs were used as templates for the synthesis of digoxigenin-labeled riboprobes: a 2 kb full-length nLacZcDNA (Kelly et al., 2001); a 584 bp Fgf10 cDNA (Bellusci et al., 1997b); a 948 bp full-length mouse Spry2 cDNA(Mailleux et al., 2004); a 642 bp Shh cDNA(Bellusci et al., 1997a); a 1.5 kb full-length mouse Bmp4 cDNA; and a 90 bp Mlc1v cDNA(Lyons et al., 1990). To determine relative expression levels in mutants compared with wild-type embryos, they were all processed in the same tube. LacZ expression on whole-mount lungs was monitored by detecting β-galactosidase activity as described by Kelly et al. (Kelly et al.,1995). Sections were subsequently cut at 35 μm with a vibratome or at 5 μm with a microtome after paraffin embedding.

The Mlc1v-nLacZ-24 line is bred in a mixed background and has been previously described (Kelly et al.,2001). The transgene containing an nLacZ reporter gene is integrated upstream of the Fgf10 gene. Fgf10^LacZ/– mouse embryos were generated by crossing Fgf10^+/– on a C57Bl/6 background(Sekine et al., 1999) with Mlc1v-nLacZ-24^+/– mice(Kelly et al., 2001). Wild-type littermates were used as control embryos at different developmental stages. The Fgf10^– and Mlc1v-nLacZ-24⁺alleles were genotyped as described previously(Mailleux et al., 2002; Kelly et al., 2001). The number of Fgf10^LacZ/– embryos used in this study (52 in total) at the different stages was as follows: embryonic day (E) 12.5(n=7); E13.5 (n=3); E14.5 (n=6); E16.5(n=2); E17.5 (n=10); E18.5 (n=7); P₀(n=17). For simplicity, the Mlc1v-nLacZ-24^+/– is called Fgf10^LacZ/+; the Mlc1v-nLacZ-24^+/+ is called Fgf10^LacZ/LacZ; the Fgf10^+/+;Mlc1v-nLacZ-24^+/– is called Fgf10^LacZ/+ while the Fgf10^+/–;Mlc1v-nLacZ-24^+/–is called Fgf10^LacZ/–. The SpC-Bmp4(n=3 at E16.5) and SpC-Shh (n=4 at E16.5)transgenic embryos have been generated previously(Bellusci et al., 1996; Bellusci et al., 1997a).

Embryonic lungs were removed at E11.5 and placed on Nuclepore Filters (8μm pore diameter) in 30 μl of a 1:1 mixture of Matrigel and culture medium (DMEM/F12 medium containing penicillin/streptomycin and 0.1%heat-inactivated fetal bovine serum). After 30 minutes at 37°C to allow the Matrigel to polymerize, these filters were laid on the surface of 500μl culture medium containing cyclopamine (TRC Biomedical Research Chemicals, Canada) in Nunclon dishes (technique adapted from Lebeche et al., 1999). Concentrations of 1, 2, 5, 10 and 15 μmol/l cyclopamine were tested as previously described (Yao et al.,2002). At 10 and 15 μmol/l, cyclopamine was toxic. Explants were cultured for 1 day in the presence of cyclopamine and then fixed in 4%PFA (15 minutes at 4°C) before X-gal staining. For distal mesenchyme grafting, the distal part (mesenchyme and mesothelium) of the accessory or median lobes of wild-type and Fgf10^LacZ/+ lungs at E12.5 were dissected with tungsten needles in culture medium. Fgf10^LacZ/+ and wild-type lungs were reciprocally grafted in 1:1 Matrigel:culture medium on Nuclepore Filters using tungsten needles. After 30 minutes polymerization of the Matrigel at 37°C, 500 μl culture medium was added under the filter. Grafts were cultured for 44 hours(37°C, 6% CO₂), fixed in 4% PFA and X-gal stained as described above.

The distal part of wild-type accessory lobe mesenchyme was dissected and incubated with the Cell Tracker Green CMFDA (Molecular Probe; 20 μmol/l in DMEM/F12 medium) for 20 minutes in the dark at room temperature. The labeling mixture was prepared following the manufacturer's instructions. The distal parts were then grafted back at the tip of the accessory lobe they were derived from. Time-lapse video-cinematography was started 18 hours post-grafting using a LEICA inverted microscope equipped with a temperature and CO₂ controlled chamber and a Princeton Micromax CDD camera as described previously (Murase and Horwitz,2002). The experiments have been carried out over a 24-48 hour period. With this in-vitro system, the position of the most external edges of the lung mesenchyme did not significantly change and was used as the reference point in order to follow the putative migration of the mesenchymal cells within this time frame.

Total RNA was extracted from a pool of three lungs at E14.5 per genotype using the RNeasy Total RNA kit (Qiagen). The RNA was reverse-transcribed using hexamers with TaqMan Transcript Reagent kit (Applied Biosystem) according to the manufacturer's conditions. One-fiftieth of the cDNA prepared from 1 μg RNA was subjected to PCR using murine gene-specific primers for Fgf10(5′-TGTTTTTTTGTCCTCTCCTGGGAG-3′ and 5′-GGATACTGACACATTGTGCCTCAG-3′), Bmp4(5′-GAACAGGGCTTCCACCGTA-3′ and 5′-TGAGGTGTCCAGGAACCAT-3′), Spry2(5′-CTCCACTCAGCACAAACAT-3′ and 5′-TTGTCCTTGTATGCTCCGA-3′) and tubulin(5′-TGGCCAGATCTTCAGACCAG-3′ and 5′-GTAAGTTCAGGCACAGTGAG-3′) as internal control. No amplicons were detected in water and minus RT control. For semiquantitative PCR, target sequences were amplified within the linear amplification range between 25 and 34 cycles at 55°C in order to yield visible products. PCR products were separated by electrophoresis on a 1.5% agarose gel and stained with ethidium bromide. Photographs were taken with a Vilbert-Loumart apparatus and the intensity of the bands was determined by densitometry with NIH Image software(http://rsb.info.nih.gov/nih-image/)using Gel Plotting Macros(ftp://rsbweb.hih.gov/pub/nih-image/macros/). Expression of Fgf10, Bmp4 and Spry2 were determined relative to tubulin expression.

Explants consisted of the total mesenchyme distal to the primary bronchi of wild-type and Flk1^LacZ/+(Shalaby et al., 1995) E11.5 lungs. They were cultured in Matrigel™(Bellusci et al., 1997b) for 48 hours in the presence of 0, 50, 100 and 150 ng/ml BMP4 (R&D). BMP4 (100 ng/ml) was the optimal dose to induce α-SMA expression in the mesenchyme, and this dose was used throughout the study (n=4). The explants were then fixed in PFA 4%, dehydrated, embedded in paraffin and sectioned at 7 μm. Flk1^LacZ/+ explants were stained with X-gal before embedding.

Whole lungs were dissected at E13.5 and subjected to trypsin digestion to give rise to single cells. Mesenchymal cells were separated from epithelial cells by differential adhesion as described previously(Lebeche et al., 1999; Yang et al., 1999). Mesenchymal cells were cultured for 4 hours (undifferentiated cells) or 48 hours (differentiated cells) in DMEM/F12 medium containing penicillin/streptomycin and 0.5% heat-inactivated fetal bovine serum for starvation. They were subsequently cultured during 20 minutes with 100 ng/ml of recombinant FGF1, FGF7 and FGF10, respectively (R&D Systems). Protein extracts and immunoblot analysis were performed as described previously(Yang et al., 1999).

A Cy3-conjugated mouse monoclonal antibody against α-SMA, (Sigma,C-6198), pan-Cytokeratine (Dako), β-galactosidase (US Biological,G1041-42) were used at a dilution of 1/200 for immunohistochemistry. For immunoblot analysis, rabbit polyclonal antibodies against phospho-AKT (Ser473,#9271), AKT (#9272), phospho-p44/42 MAP kinase (#9101) and p44/42 MAP kinase(#9102) were obtained from Cell Signalling Technology, and used according to the manufacturer's instructions.

We first analyzed whether Mlc1v-nLacZ-24 transgenic expression represents expression from the Fgf10 locus in the developing lung,similar to previous observations in the heart(Kelly et al., 2001). By whole-mount in-situ hybridization we could not detect any endogenous Mlc1v mRNA in the lung at E12.5 and 13.5, whereas we did detect Mlc1v mRNA expression in the internal control tissue: the cardiac muscle (data not shown). This strongly suggested that the endogenous Mlc1v gene promoter is inactive in the developing lung, and that it may therefore be concluded that the Mlc1v promoter sequences in the transgenic cassette do not drive LacZ expression in the lung. To further support this, we used a model of in-vitro smooth muscle cell differentiation in primary cultures of embryonic lung mesenchymal cells(Yang et al., 1998). If the myosin light chain ventricular 1 promoter of the transgene drives a residual expression in the smooth muscle cells of the Mlc1v-nLacZ-24^+/– lungs, we would expect 100% of the cells to be simultaneously positive for β-galactosidase and alpha smooth muscle actin (α-SMA). Indeed, all the mesenchymal cells expressedα-SMA after 44 hours of culture. However, only a fraction of these cells were β-galactosidase-positive (data not shown), suggesting that the Mlc1v promoter is indeed inactive in the Mlc1v-nLacZ-24^+/– line.

We next compared the expression pattern of the transgene in lungs of Mlc1v-nLacZ-24 hemizygous embryos to Fgf10 expression in wild type lungs, at the level of either LacZ mRNA expression orβ-galactosidase activity. At E10.5, the distal tip of the right lobe(black arrow), but not the left lobe, expressed Fgf10 mRNA(Fig. 1A). One day later, the right lobe had subdivided into four lobes. Each of these lobes expressed Fgf10 mRNA at the distal tips(Fig. 1C). At this time, the left lobe had also started to express Fgf10 mRNA (dashed boxed area in Fig. 1C). We have previously shown that the distal Fgf10 expression is exclusively mesenchymal(Bellusci et al., 1997b). We detected β-galactosidase activity in a very similar pattern at E10.5(Fig. 1B) and E11.5(Fig. 1D). The absence ofβ-galactosidase activity in the left lobe at E11.5 (boxed area in Fig. 1D) may reflect the delayed onset of expression of Fgf10 in the left lobe, as already observed at E10.5. Analysis of LacZ expression as mRNA(Fig. 1G,H) or asβ-galactosidase activity (Fig. 1I,J) at E12.5 revealed that its expression was still highly similar to the expression pattern of Fgf10 mRNA at the mesenchymal edges of each lobe adjacent to the distal ends of the bronchi(Fig. 1E,F)(Mailleux et al., 2001). However, at this stage (Fig. 1I,J) and later at E14.5, β-galactosidase activity was additionally observed at the level of the secondary bronchi (arrow in Fig. 1L) by contrast to Fgf10 mRNA expression (Fig. 1K). This expression corresponds to the mesenchyme adjacent to the bronchial epithelium (Fig. 2E,F). Importantly, LacZ expression was essentially not found in the mesenchyme of the primary bronchi except for the very distal part(black arrowhead in Fig. 1I and white arrowhead in Fig. 1L).

The difference between LacZ expression at the mRNA and protein level is most probably due to a higher stability of the β-galactosidase protein compared with LacZ mRNA, as previously observed during heart development (Kelly et al.,2001). It may indicate that cells, expressing Fgf10 mRNA while located in the distal mesenchyme, or their daughter cells, populate the more proximal areas. In conclusion, the similarity of expression patterns of Fgf10, LacZ and β-galactosidase activity, combined with the lack of detection of Mlc1v transcripts in the developing lung, strongly suggests that expression of the Mlc1v-nLacZ-24 transgene is exclusively driven by Fgf10 regulatory sequences and completely and solely represents the Fgf10 expression pattern in the developing lung.

To further validate the use of the Mlc1v-nLacZ-24 transgene as a reporter for Fgf10 expression, we analyzed the control of LacZ expression by Fgf10 regulatory elements in a functional assay. We made use of the knowledge that sonic hedgehog (SHH) downregulates Fgf10 expression in the lung(Bellusci et al., 1997b; Pepicelli et al., 1998; Lebeche et al., 1999), and that cyclopamine is a specific inhibitor of SHH function(Yao et al., 2002). Lungs derived from E11.5 hemizygous Mlc1v-nLacZ-24 embryos, and cultured for 28 hours in the presence of 5 μmol/l cyclopamine displayed dilated epithelium (arrows in Fig. 1S,O). A similar phenotype was previously observed upon addition of recombinant FGF10 on lung grown in vitro(Bellusci et al., 1997b),suggesting that cyclopamine treatment triggers Fgf10 upregulation. An increase in β-galactosidase activity was observed compared with the untreated lung (compare Fig. 1P-U). The effect of cyclopamine upon β-galactosidase activity was particularly apparent in the left lobe (arrows in Fig. 1P,T). These observations support our expression data and indicate that LacZ expression in the Mlc1v-nLacZ-24 strain can be used as a reporter for Fgf10expression. In accordance with this conclusion, we will hereafter refer to the mice heterozygous for the Mlc1v-nLacZ-24 cassette as Fgf10^LacZ/+ mice.

The β-galactosidase expression domain expanded progressively from the peripheral distal mesenchyme toward the distal epithelium between E11.5 and 12.5 (Fig. 2A-D), whereas the Fgf10 and LacZ mRNAs remained expressed only in the peripheral mesenchyme at this time (Fig. 1D and double pointed arrows in Fig. 1F,H). This supports the presumption that cells from the peripheral distal mesenchyme or their daughter cells can relocate to more proximal areas. Vibratome sections of the accessory lobe at E12.5 illustrate that the β-galactosidase-positive cells were exclusively located in the mesenchyme throughout the distal tip mesenchyme and as one layer adjacent to the proximal epithelium(Fig. 2E). Transversal sections in the most proximal part of E12.5^LacZ/+ accessory lobe show that β-galactosidase expression entirely surrounded the epithelium of the secondary bronchi (Fig. 2E'). In addition, at this stage not all the mesenchymal cells directly adjacent to the epithelium were positive for β-galactosidase(Fig. 2E, red arrows), in harmony with the patchy β-galactosidase expression pattern in whole-mount staining of the accessory lobe (arrow in Fig. 2D). By E13.5, theβ-galactosidase-positive cells had formed a continuous layer adjacent to the epithelium (Fig. 2F).

The location of these cells suggested that they could be parabronchial SMCs. Indeed, they co-expressed the smooth muscle cell marker α-SMA around the bronchial epithelium (Fig. 2G). Fig. 2H,I show additionally that α-SMA was not expressed in the distal tip mesenchyme,but was restricted to the layer of mesenchymal cells directly in contact with the proximal bronchial epithelium.

In conclusion, the dynamics of the β-galactosidase expression pattern strongly suggests that the mesenchymal Fgf10/β-galactosidase-positive cells in the distal tip represent smooth muscle cell progenitors.

In order to assess whether these proximal β-galactosidase-positive cells are indeed derived from the distal mesenchyme, we carried out reciprocal heterotypic grafts of pulmonary distal mesenchyme on lungs of wild-type and Fgf10^LacZ/+ embryos. The grafted lungs were cultured for 2 days and compared to unmanipulated Fgf10^LacZ/+ cultured lungs. In the unmanipulated lungs, the β-galactosidase-positive cells were located in the distal tip as well as in a single layer of cells around the more proximal epithelium (Fig. 3A-D).

An identical patterning of β-galactosidase-positive cells was observed in cultures of wild-type lungs grafted with Fgf10^LacZ/+distal lung mesenchyme (Fig. 3E-H). By contrast, in cultures of Fgf10^LacZ/+lung grafted with wild-type mesenchyme(Fig. 3I), no or fewβ-galactosidase positive-cells were observed proximally(Fig. 3K,L). These results demonstrate unequivocally that the proximal mesenchymalβ-galactosidase-positive cells originate from the Fgf10-positive distal tip mesenchyme. They additionally support our notion that the Mlc1v promoter sequences do not drive LacZ expression in the developing lung.

To assess how distal mesenchymal cells populate the more proximal areas,the distal part of a wild-type lung (mesenchyme and mesothelium) was dissected out from the accessory lobe, re-implanted after staining with the fluorescent marker CMFDA and followed up for 25 hours by video-cinematography. Our results show no relative movement of the labeled cells with respect to the epithelium and no cells leaving the labeled explant(Fig. 3M,N). This suggests that mesenchymal cells do not actively migrate along the epithelium but passively relocate during epithelial outgrowth into the distal mesenchyme.

We noted that Fgf10^LacZ/+ mice, similar to Fgf10 heterozygous mice, have smaller eyelids and a moderate nervous behavior compared with wild-type mice. We therefore tested whether the insertion of the Mlvc1v-nLacZ cassette 120 kb upstream of the transcriptional start site of the Fgf10 gene reduces Fgf10expression. Indeed, whole-mount in-situ experiments showed decreased Fgf10 expression in Fgf10^LacZ/LacZ embryos when compared with wild-type littermates (Fig. 4A,B). This indicates that the insertion of the transgenic cassette creates a hypomorphic Fgf10 allele. It furthermore suggests that an allelic series can be generated by crossing hemizygous or homozygous Fgf10^LacZ mice with heterozygous Fgf10 null mice.

Fifty-two Fgf10^LacZ/– compound heterozygous embryos were obtained at Mendelian ratio between E12.5 and P0. Semi-quantitative RT-PCR was performed on RNA isolated from a pool of three lungs per genotype at E14.5 (Fig. 4C). The results indicate a gradual decrease of Fgf10mRNA expression in Fgf10^LacZ/+, Fgf10^+/–and Fgf10^LacZ/– compound heterozygous mice,confirming the generation of an allelic series of pulmonary Fgf10expression.

At E12.5, Fgf10^LacZ/– embryos have a slightly smaller lung with decreased branching morphogenesis compared with wild-type or Fgf10^LacZ/+ lungs (Fig. 4D,E). In addition, α-SMA expression is decreased specifically around the epithelium of the secondary bronchi (n=2; Fig. 4F,G). The deposition of laminin, an extracellular matrix protein synthesized by the lung epithelium and involved in the differentiation of the SMC(Zhang et al., 1999), was not perturbed around the bronchial epithelium of Fgf10^LacZ/– lungs (data not shown).

A similar decrease in α-SMA expression was observed at E13.5(n=2, data not shown), E14.5 (n=2, data not shown) and E17.5(n=3, Fig. 4I,K)compared with control lungs (Fig. 4H,J), as well as at E18.5 (n=3, data not shown) and after birth (n=2, data not shown). Elastin deposition by SMCs around the bronchi was also reduced in Fgf10^LacZ/– lungs at birth (data not shown). In addition, a specific decrease in gelatinase activity at birth in Fgf10^LacZ/– lung mesenchyme,especially in parabronchial SMCs, was revealed by in-situ zymography (data not shown). Therefore, decreased Fgf10 expression in the distal mesenchyme clearly leads to a defect in the proper establishment of the parabronchial SMC population from early embryonic stages.

Interestingly, less β-galactosidase activity was observed around the epithelium of the secondary bronchi, while more activity was present in the distal mesenchyme of Fgf10^LacZ/– lungs compared with Fgf10^LacZ/+ lungs (compare Fig. 4L,M). By TUNEL or PCNA analyses (data not shown), we did not see any significant changes in apoptosis or in proliferation of the parabronchial SMCs between E14.5 and 17.5.

These results suggest that FGF10 is involved directly or indirectly in the entry of the Fgf10/β-galactosidase-positive SMC progenitors into the SMC lineage rather than in their survival or proliferation.

To test whether FGF10 acts in an autocrine fashion on the mesenchyme, or in a paracrine fashion via the epithelium, primary cultures of mesenchymal cells derived from the entire E13.5 wild-type lung were established. These were either cultured for 4 hours, sufficient for the cells to start to differentiate, or for 48 hours to lead to fully differentiated SMCs, before addition of recombinant FGF1, FGF7 or FGF10. Unlike FGF1, FGF10 and FGF7 did not upregulate ERK phosphorylation (Fig. 5A) nor AKT phosphorylation (data not shown) in undifferentiated or differentiated cultures. These results suggest that FGF10 does not act on mesenchymal cells in an autocrine fashion but via an epithelial intermediate. Previous reports have shown that Bmp4 is a downstream epithelial target of FGF10 in the lung (Lebeche et al., 1999; Weaver et al.,2000; Mailleux et al.,2001). In harmony with these results, the expression of Bmp4 was decreased in Fgf10^LacZ/– lungs(Fig. 5C,E). The expression of Sprouty2, another FGF10 downstream target(Mailleux et al., 2001) was also decreased (Fig. 5D,E),while the expression of Shh, an upstream negative regulator of Fgf10 expression (Bellusci et al.,1997b), was unchanged (Fig. 5B).

To test whether Bmp4 is indeed involved in the formation of SMCs in vivo, we overexpressed it under the surfactant protein C promoter specifically in the distal lung epithelium. As a positive control, we also overexpressed Shh under the same promoter, as it has been shown to trigger the induction of α-SMA expression in isolated lung mesenchymal explants cultured in vitro (Weaver et al.,2003). To our surprise, SpC-Shh overexpression did not have an apparent effect on localization or levels of α-SMA expression in the mesenchyme (Fig. 5H),whereas SpC-Bmp4 induced ectopic expression of α-SMA in the mesenchyme as expected (Fig. 5I). In accordance with this result, we also showed that recombinant BMP4 (100 mg/ml) directly upregulates α-SMA expression lung in mesenchyme explants (n=4) cultured in Matrigel™(Fig. 5J,L). Not all the cells were positive for α-SMA after BMP4 treatment. These cells may be endothelial cells, as suggested by the presence ofβ-galactosidase-positive cells in mesenchymal explants from E12.5 Flk1^LacZ/+ lungs in the absence or presence of BMP4(Fig. 5K,M). Taken together,these data strongly suggest that epithelial BMP4, a target of FGF10, controls the differentiation of cells arising from the distal mesenchymal Fgf10-expression domain into the parabronchial SMC lineage.

By expression analysis at the RNA and protein level combined with a functional assay, we here demonstrate that expression from the Mlc1v-nLacZ-24 transgene exclusively and completely reflects Fgf10 expression in the murine embryonic lung. These findings are in accordance with a previous report showing that expression of this transgene reflects Fgf10 expression in the developing heart, attributed to its integration 120 kb upstream of the Fgf10 gene(Kelly et al., 2001). By screening the genome using the Ensembl database at http://www.ensembl.org/Mus_musculus/,we found that only one other gene was present near this integration site at the telomere of chromosome 13. It concerns the mitochondrial ribosomal protein S30 (Mrps30) gene, which is located 200 kb upstream of insertion of the transgenic cassette. Furthermore, unpublished work in our group has demonstrated that Fgf10^LacZ/–, embryos display a hypomorphic Fgf10 phenotype in the gut, limb and mammary gland (S.B.,unpublished). This strongly suggests that the Mlc1v-nLacZ-24transgene is expressed in a pattern similar to that of Fgf10 in these organs. We therefore hypothesize that the Mlc1v-nLacZ-24 mouse strain may be a useful reporter for Fgf10 expression in any organ.

Our chimaeric lung cultures demonstrated that the distal mesenchyme gives rise to the parabronchial SMC population in more proximal areas. To date, this is the first evidence for the location of progenitors for parabronchial SMCs,as well as for their identification by the expression of Fgf10.

Our data indicate that β-galactosidase is essentially not present in the mesenchyme around the primary bronchi of the Fgf10^LacZ/+ lungs, while α-SMA is detected around the primary bronchi at E12.5. This difference suggests that the smooth muscle cells around the epithelium of the primary bronchi and the smooth muscle cells around the epithelium of the secondary bronchi may originate from different pools of progenitors. This is supported by the presence of α-SMA in the secondary bronchi at E11 (Tollet et al.,2001), even before Fgf10/β-galactosidase-positive cells are relocalized around the secondary bronchi (this report). However,from E13.5 onward, the smooth muscle cells around the secondary bronchi originate mostly from the Fgf10-expression domain, as indicated by the homogenous expression of β-galactosidase in the mesenchyme adjacent to the bronchial epithelium.

The distal location of this new subset of parabronchial SMC progenitors in the lung is reminiscent of the distal location of the progenitors for the various epithelial cell lineages of the lung (reviewed by Warburton et al., 2000). It is now of interest to determine whether the other mesenchymal lineages, including other types of SMCs (e.g. the vascular SMCs), lipocytes and endothelial cells,also originate in this domain. Interestingly, this domain of progenitor cells is also the area where budding morphogenesis is initiated(Bellusci et al., 1997b) and where many signaling molecules are expressed. This suggests that the distal lung buds act as signaling centers that synchronize morphogenetic events with differentiation.

A difference in parabronchial SMC formation was observed in Fgf10^LacZ/– lungs at all the stages investigated. The decreased β-galactosidase expression around the bronchial epithelium and the increased expression in the distal mesenchyme in Fgf10^LacZ/– lungs strongly suggest that FGF10 regulates the entry of the SMC progenitors into the SMC program, rather than controlling the progenitor population. This is a new role for Fgf10,as to date its major role has seemed to be the regulation of budding and branching morphogenesis in a variety of organs(Min et al., 1998; Sekine et al., 1999) including the lung (Bellusci et al.,1997b). Interestingly, our in-vitro data indicate that the relocalization of the Fgf10-positive cells in the distal mesenchyme along the epithelium is essentially due to the growth of the distal epithelium into the distal mesenchyme. However, a comparison of Fig. 3D with Fig. 2E demonstrates that the domain of β-galactosidase-positive cells does not expand as proximally in vitro as it does in vivo, and that the branching pattern is disturbed. Therefore, we cannot exclude the fact that active migration of the mesenchyme does occur in vivo. It is therefore possible that the epithelium exerts a chemoattractive activity on cells derived from the distal mesenchyme, similar to the way the mesenchyme acts as a chemoattractant for the distal lung epithelium (Park et al.,1998).

Our cultures of primary lung mesenchyme in the presence of various recombinant FGFs demonstrated that FGF10, in contrast to FGF1, was not able to phosphorylate ERK and AKT. This indicates that FGF10 does not act in an autocrine fashion on the mesenchyme to fulfill its function in parabronchial SMC formation. At present, only FGF9 secreted by the mesothelium is implied in vivo in the biogenesis of parabronchial SMCs. It is proposed to signal through the mesenchymal receptor FGFR2c, and to maintain the mesenchymal progenitors in a proliferative and undifferentiated state(Weaver et al., 2003). By contrast, we propose that FGF10 uses an epithelial intermediate, BMP4.

Our results indicate a decrease in Bmp4 expression in Fgf10^LacZ/– embryos. This reduction in Bmp4expression seems to primarily occur in the epithelium. These results are consistent with previous reports showing that FGF10 upregulates epithelial Bmp4 transcription (Lebeche et al., 1999; Weaver et al.,2000). Overexpression of Bmp4 in the distal lung epithelium using the surfactant protein C promoter led to ectopic expression of α-SMA in the distal mesenchyme. While addition of recombinant SHH induces α-SMA expression on isolated lung mesenchymal explants(Weaver et al., 2003), we show that overexpression of Shh in the distal lung epithelium in vivo does not modify α-SMA expression. This may be explained by the lack of upregulation of Bmp4 in the epithelium or the mesenchyme upon overexpression of Shh in vivo(Bellusci et al., 1997a), by contrast to the induction of Bmp4 expression by SHH in vitro(Weaver et al., 2003).

Consistent with a major role for Bmp4 in SMC differentiation,recombinant BMP4 induced α-SMA expression in lung mesenchyme explants in vitro after 48 hours of culture. These results strongly suggest that BMP4 induces SMC formation by acting directly on the mesenchyme. We therefore propose that FGF10 expressed by the distal mesenchyme may contribute to parabronchial SMC formation via the upregulation of BMP4 synthesis by the epithelium. The failure to induce α-SMA expression in all cells can be explained by the presence of other cell types in the mesenchymal explants,e.g. the endothelial cells. In addition, these explants also contain a layer of mesothelium, producing FGF9, which has been shown to prevent the differentiation of the smooth muscle cells(Weaver et al., 2003).

Independent reports support the proposed role of BMP4 in smooth muscle cell differentiation. In the kidney, periureteral mesenchymal cells differentiate into smooth muscle cells at a site where Bmp4 is highly expressed. In addition, E15.5 Bmp4^+/– ureters have fewerα-SMA-positive cells (Miyazaki et al., 2003). Furthermore, in human lung fibroblast cultures,exogenous BMP4 inhibits proliferation and promotes smooth muscle cell differentiation, as indicated by increased expression of α-SMA and smooth muscle myosin (Jeffery et al.,2004).

The expression of the BMP antagonist noggin in parabronchial SMC(Weaver et al., 2003) suggests that regulation of BMP signaling may be important to finely tune SMC differentiation.

In conclusion, we demonstrate that the mesenchyme of the distal lung tip contains a set of progenitors for parabronchial SMCs that can be identified by Fgf10, and that normal transcription levels of at least one Fgf10 allele are required for their entry into the parabronchial SMC lineage.

We are grateful to Drs Morgan Matthew, Klaus Kratochwil and Vesa Kaartinen for helpful comments on the manuscript, Daniel Meur and Dominique Morineau for the photographic reproductions, and Martine Blanche for skilled technical assistance in the mouse facility. We would also like to thank Dr Margaret Buckingham for making the Mlc1v-nLacZ-24 line available to us. This work was supported by the Association pour la Recherche sur le Cancer (ARC)grant 5214, the Action Concertée Incitative 81/2001 – Biologie du Développement et Physiologie Intégrative, NIH RO1 HL074832-01 and HL75773-01 grants and the American Lung Association (CI-36-N) (S.B.). R.K. is an INSERM research fellow. A.M. received financial support from the French Ministry of Research and Technology as well as the ARC. J.V. received financial support from the European Commission (contract QLGA-CT-2000-51230),the CHLA Institutional award and the California Breast Cancer Research Program fellowship (10FB-0116).