Anxa4 mediated airway progenitor cell migration promotes distal epithelial cell fate specification

Genetic studies have shown that FGF10/FGFR2 signaling is required for airway branching morphogenesis and FGF10 functions as a chemoattractant factor for distal epithelial cells during lung development. However, the detail downstream cellular and molecular mechanisms have not been fully characterized. Using live imaging of ex vivo cultured lungs, we found that tip airway epithelial progenitor cells migrate faster than cleft cells during airway bud formation and this migration process is controlled by FGFR2-mediated ERK1/2 signaling. Additionally, we found that airway progenitor cells that migrate faster tend to become distal airway progenitor cells. We identified that Anxa4 is a downstream target of ERK1/2 signaling. Anxa4−/− airway epithelial cells exhibit a “lag-behind” behavior and tend to stay at the stalk airways. Moreover, we found that Anxa4-overexpressing cells tend to migrate to the bud tips. Finally, we demonstrated that Anxa4 functions redundantly with Anxa1 and Anxa6 in regulating endoderm budding process. Our study demonstrates that ERK1/2/Anxa4 signaling plays a role in promoting the migration of airway epithelial progenitor cells to distal airway tips and ensuring their distal cell fate.

tips are able to generate both distal and stalk airway epithelial cells at least up to E13.5 19 . It has been showed that FGF10 signaling is essential for preventing distal airway progenitor cell from differentiating into stalk airway epithelial cell 20,21 . However, it is less clear how airway branching morphogenesis is orchestrated with the airway epithelial cell fate specification.
Annexin proteins are found in species ranging from fungi to higher vertebrates. They are a highly-conserved superfamily of proteins that bind with membrane phospholipids in a calcium-dependent manner and this binding links them to many membrane-related processes (e.g., basic membrane organization, membrane trafficking) [22][23][24] . Further, it has been shown that several members of the Annexin family are able to bind with and "bundle" F-actin filaments [25][26][27][28] . In vitro studies including both gain-and loss-of-function experiments have shown that Annexins play a role in promoting cell migration 29,30 . Despite the abundance and conservation of Annexins in most eukaryotic species, relatively little is known about the regulation of Annexin gene expression and little is known about the function of Annexin proteins during embryonic lung development.
Here, using a combination of live imaging, mouse genetics and lung endoderm culture system experiments, we found that tip airway epithelial progenitor cells migrate faster than cleft cells during airway bud formation. We identified Anxa4 (encoding Annexin A4) as a downstream target of ERK1/2 signaling and found that the expression level of Anxa4 is positively regulated by the activity of ERK1/2 signaling. We showed that Anxa4 is required for airway epithelial cell migration, both in vitro and in vivo. Furthermore, we found that Anxa4 −/− epithelial cells exhibit a "lag-behind" behavior and tend to contribute to the stalk airways. In contrast, Anxa4-overexpressing cells tend to remain in bud tips during bud formation. We also found that Anxa4 functions redundantly with Anxa1 and Anxa6 in regulating the endoderm budding process. Our study establishes that FGF10-activated ERK1/2/Anxa4 signaling plays a role in promoting airway progenitor cell migration and ensuring their distal airway epithelial cell fate by regulating Anxa4 expression during airway bud formation.

Results
Airway progenitor cells that migrate faster tend to commit to distal airway cell fate. To investigate the cellular behaviors during airway bud formation, we conducted an ex vivo time-lapse imaging experiment with E12.5 lungs to monitor cell behaviors during airway bud formation. Pregnant females carrying Shh CreER/+ ; Rosa26-mTmG embryos were injected with one dose of tamoxifen at E9.5 (Fig. 1A). After a tamoxifen injection, membrane GFP was expressed in a mosaic manner in the airway epithelial cells of Shh CreER/+ ; Rosa26-mTmG lung, enabling tracing of individual airway epithelial cells. Airway epithelial progenitor cells at the tip of the first lateral branch in the right caudal lobe (RCd.L1) of lung at E12.5 were imaged (Fig. 1B). In wild type mice, this lateral branch undergoes a planar bifurcation that leads to the formation of two new buds 31 . The region between these two newly formed bud tips is referred to as the cleft (Fig. 1B). The RCd.L1 bud tip was imaged every 10 minutes for 5 hours. After five-hour live imaging, airway progenitor cells at the bud tip directly migrated out and formed a new bud tip, while airway progenitor cells at the cleft region migrated slower and formed the cleft ( Fig. 1C and Supplementary Video 1).
To analyze cell migration of airway progenitor cells, we manually traced the positions of GFP cells at 31 time-points over a 5-hour-time-course live imaging by using Imaris software. Monitoring of the position data for each cell at each time points allowed us to directly analyze the migration tracks of cells and quantify both cell migration displacement and cell migration velocity (Fig. s1A). The cell migration displacement was calculated as the linear distance (in μm) between the current position of a given cell at a given time point and its starting position. The cell migration velocity was calculated as the ratio of cell migration distance (in μm) to migration time (in hours) (Fig. s1A,B). By analyzing migration tracks of airway progenitor cells at bud tip (tip cells) and airway progenitor cells at cleft region (cleft cells), we found that the tip cells migrated over longer distance and more directionally as compared to cleft cells (Fig. 1D). The average migration displacement over the 5 hours of the experiment was higher for tip cells as compared to cleft cells (Fig. 1E). Moreover, we found that the tip cells have higher migration velocity and longer apical-to-basal cell length as compared to cleft cells (Fig. 1F,G).
We next sought to identify the molecular mechanisms underlying our observation that tip cells migrate faster than cleft cells during planar bifurcation. Given that ERK1/2 signaling can be activated by FGF10/FGFR2 signaling and is essential for cell migration during development 32,33 , we hypothesized that FGF10/FGFR2 may regulate cell migration of tip cells via ERK1/2 signaling during airway bud formation. Indeed, the level of p-ERK1/2 was higher in cells of the two nascent bud tips than in cells of the cleft region during the planar bifurcation process, suggesting that ERK1/2 signaling may control tip cell migration (Fig. 1H).
Similar to lung, kidneys are highly branched organs and renal branching morphogenesis is known to requires GDNF and its receptor Ret 34,35 . It has been demonstrated that Ret is required for directed movement of progenitor cells during renal development and Ret −/− cells tend to lag behind and contribute to the renal trunks 17 .
To investigate the effect of loss of FGFR2 signaling on cell migration, we conditionally knocked out Fgfr2 in airway epithelial cells at E9.5 using Shh CreER/+ mice and live imaged the RCd.L1 bud tips of Shh CreER/+ ; Fgfr2 F/F ; Rosa26-mTmG (Fgfr2 cKO ) lungs at E12.5 ( Fig. s1C and Supplementary Video 2). We found that Fgfr2 −/− cells showed shorter migration track and smaller migration displacement than control cells and that their migration velocity was much slower than control cells (Fig. s1D-F and Supplementary Video 2). Additionally, we found that some Fgfr2 −/− ;GFP + cells underwent apoptosis (Fig. s1G), consistent with previous study that loss of Fgfr2 induces cell death 36 .
ScIeNtIfIc RepoRts | (2018) 8:14344 | DOI:10.1038/s41598-018-32494-z ERK1/2 signaling regulates the expression of Anxa4. Seeking to identify downstream target(s) of ERK1/2 signaling in airway epithelial cells, we set up a mesenchyme-free lung endoderm culture system in which FGF10 is the only supplemented growth factor. Distal lung endoderm buds were dissected out from whole lungs at E11.5 and were subsequently cultured with serum-free medium supplemented only with FGF10 (800 ng/ml) ( Fig. 2A). Based on our observation of both morphology and cell behavior, we defined the developmental process of the in vitro cultured endoderm explants into two stages: (i) within the initial 24 h of the culture period ("pre-budding stage"), the lung endoderm bud became sealed, grew and expanded into a cyst, progressing toward bud formation; (ii) from 24 h to 48 h ("budding stage"), the lung endoderm underwent branching and formed many buds at the cyst surfaces (Fig. 2B). We noted that the phosphorylation level of ERK1/2 was significantly increased at 24 h and at 48 h as compared to 0 h in the cultured lung endoderm explants (Fig. 2C). Similar to our finding that tip cells exhibit high p-ERK1/2 levels during planar bifurcation (Fig. 1H), we found that p-ERK1/2 levels were high in the bud tips of cultured lung endoderm explants (Fig. 2D).
Annexin proteins are highly-conserved calcium-dependent phospholipid binding proteins 22,37 and several Annexins are known to bind with F-actin to regulate cell migration 25,28,29 . A previous study has shown that several Annexin family members may be downstream targets of FGF10 and may be involved in endoderm budding process 38 . Consistently, we here found that the expression levels of Anxa gene family members (encoding Annnexins) showed a steady increase from 0 h to 48 h in cultured lung endoderm explants (Fig. s2A). This increasing expression trend was positively correlated with increased budding activity in lung endoderm cultures, suggesting that Annexins may be downstream targets of FGF10 and, further, that these proteins may be involved in FGF10-induced lung endoderm budding process. However, only Anxa1, Anxa4 and Anxa6 showed significantly increased expression from 24 h to 48 h, the period during which lung endoderm budding occurs. Whole-mount RNA in situ hybridization experiments revealed that Anxa1 is highly expressed in the stalk airway epithelium and showed that Anxa4 is highly expressed in the distal airway epithelium; Anxa6 is expressed in both the epithelium and the mesenchyme (Figs 2E and s2B). Given our finding that Anxa4 is highly expressed in the distal airway epithelium which also has high p-ERK1/2 level (Figs 1H and 2E), we next set out to investigate whether the expression of Anxa4 is regulated by ERK1/2 signaling. We cultured wild type lung endoderm explants with DMSO or PD0325901 (inhibitor of MEK, upstream of ERK1/2) at 0 h and then analyzed the expression levels of Anxa4 at 48 h using qPCR. We found that inhibition of ERK1/2 signaling decreased the expression of Anxa4; this inhibition also reduced the expression levels of Etv4 and Spry2, two well-known downstream targets of ERK1/2 signaling (Figs 2F and s2C). Whole-mount in situ analysis of Anxa4 in lung endoderm explants confirmed that the expression of Anxa4 is inhibited by treatment with PD0325901 (Fig. 2G). Taken together, these results establish that the expression of Anxa4 is regulated by ERK1/2 signaling.
After live imaging the airway progenitor cells in the RCd.L1 bud tips of both control and Anxa4 cKO lungs, we found that both control and Anxa4 cKO lung buds are able to grow and to form two new buds ( Fig. 3B and Supplementary Video 3 and 4). However, the migration tracks of tip cells of control and Anxa4 cKO lungs revealed that the tip cells of Anxa4 cKO lungs have shorter migration tracks and lesser displacement as compared to tip cells of control lungs (Fig. 3C,D). Analysis of cell migration velocity showed that tip cells of Anxa4 cKO lungs migrate slower than did tip cells of control lungs (Fig. 3E). However, the apical-to-basal cell length of tip cells of Anxa4 cKO lungs is similar to that of tip cells of control lungs (Fig. 3F), suggesting that loss of Anxa4 does not affect the tip cell shape change. These results demonstrate that Anxa4 is required for airway epithelial progenitor cell migration during airway bud formation.
Loss of Anxa4 negatively affects distal airway epithelial cell fate specification. As Anxa4 −/− airway progenitor cells migrate slower than wild type cells, we next assessed the ability of Anxa4 −/− airway progenitor cells to become Sox9 + distal airway epithelial cells. Pregnant females carrying both control and Anxa4 cKO embryos were injected with TAM at E9.5 and these embryos were sampled at different time points from E12.5 to E15.5 (Fig. 4A). Distal and stalk airway epithelial cells were distinguished based on the expression of Sox9 and Sox2, respectively. The proportions of GFP + cells in the distal airways (Sox9 + ) of Anxa4 cKO lungs decreased over time as compared to that of control lungs, while the proportions of GFP + cells in the stalk airways (Sox2 + ) of Anxa4 cKO lungs increased: at E12.5, the proportions of GFP + cells in the distal airways and stalk airways of Anxa4 cKO lungs were similar to that of control lungs; at E13.5, the proportion of GFP + cells in the distal airways (Sox9 + ) of Anxa4 cKO lungs slightly decreased as compared to that of control lungs; at E14.5 and E15.5, the proportion of GFP + cells in the distal airways of Anxa cKO lungs significantly decreased as compared to that of control lungs (Fig. 4B,C).
We next examined the effect of loss of Anxa4 in lung epithelial cells on cell proliferation and apoptosis using immunostaining against pH3 and Caspase3 in control and Shh Cre/+ ; Anxa4 F/F lungs. In Shh Cre/+ ; Anxa4 F/F lungs, Anxa4 was knocked out in all airway epithelial cells. The proliferation rate of airway epithelial cells did not differ significantly between littermate control and Shh Cre/+ ; Anxa4 F/F lungs (Fig. s4A,B). We did not detect Caspase3 + cells in either control or Shh Cre/+ ; Anxa4 F/F lungs (Fig. s4A). The proportions of GFP + airway epithelial cells to the total airway epithelial cells were similar between control and Anxa4 cKO lungs at both E11.5 and E15.5 (Fig. s4C-E), supporting that loss of Anxa4 doesn't decrease cell proliferation or induce apoptosis. Transwell migration assays with isolated primary lung epithelial cells showed that the migration of Anxa4 −/− epithelial cells toward FGF10 was significantly decreased as compared to control epithelial cells (Fig. s4F,G). These experiments demonstrated that loss of Anxa4 decreases distal cell fate commitment and that this decrease is most likely caused by decreased cell migration, as loss of Anxa4 does not affect cell proliferation or apoptosis.
To further investigate the role of Anxa4 in cell fate commitment, we generated an Anxa4-overexpression (Anxa4 OE ) lentivirus carrying a nuclear H2B-GFP reporter and then co-cultured lung endoderm explants with this lentivirus to overexpress Anxa4. After 48 h co-culture, we did whole-mount immunostaining using antibodies against GFP and Sox2. Our immunostaining results showed that more GFP + cells remained in the endoderm bud tips (Sox2 − ) of Anxa4 OE endoderm explants as compared to vector-lentivirus infected endoderm explants, while fewer GFP + cells were detected in the proximal endoderm (Sox2 + ) of Anxa4 OE endoderm explants (Fig. 4D,E). Taken together, our findings show that Anxa4 controls cell migration and promotes distal airway epithelial cell fate specification (Fig. 4F).
ScIeNtIfIc RepoRts | (2018) 8:14344 | DOI:10.1038/s41598-018-32494-z Anxa4 functions redundantly with Anxa1 and Anxa6 in regulating endoderm budding process. We next used Shh Cre/+ mouse line to knockout Anxa4 in all lung epithelial cells to investigate the effect of the loss of Anxa4 on airway branching morphogenesis and found that neither the branching pattern nor the tube shape of Shh Cre/+ ; Anxa4 F/F mouse lungs differed significantly from their somite-matched littermate controls (Fig. s3B). We therefore hypothesized that the normal lung development phenotype that we observed in Shh Cre/+ ; Anxa4 F/F mice may result from functional redundancy among the Annexins 22,24 . Similar to a previous study that reported increased expression of other Anxa genes in Anxa1 −/− embryos 39 , we found that the expression levels of Anxa1 and Anxa6 were increased in Shh Cre/+ ; Anxa4 F/F mutant lungs (Fig. 5A). It was also notable that the expression levels of Anxa1 and Anxa6 increased significantly between 24 h and 48 h in lung endoderm culture (Fig. s2A).
We therefore knocked down either Anxa1 or Anxa6 in control and Anxa4 −/− endoderm explants with lentivirus carrying shRNA targeting Anxa1 or Anxa6. The RNAi knockdown efficiency for shAnxa1 and for shAnxa6 in the lung endoderm explants was about 40% at 48 h (Fig. 5B). Note that a slight increase in the Anxa4 expression level was detected in Anxa1 and Anxa6 knock-down endoderm explants (Fig. 5B). In control lung endoderm explants, knockdown of Anxa1 or Anxa6 by lentivirus-mediated RNAi did not impair their budding process (Fig. 5C,D). We also co-cultured Anxa4 −/− endoderm explants with Scramble shRNA lentivirus and found that Anxa4 −/− endoderm explants showed slight (not significant) decreases in the number of buds as compared with control endoderm explants (Fig. 5C,D). However, in the Anxa4 −/− endoderm explants, knockdown   of Anxa1 or Anxa6 resulted in significantly fewer buds as compared to Scramble-shRNA-treated control or Scramble-shRNA-treated Anxa4 −/− lung endoderm explants (Fig. 5C,D). Immunostaining against pH3 and Caspase3 showed that loss of Anxa1, or Anxa4, or Anxa6 did not impair the relative proportions of pH3 + and Caspase3 + cells in these lung endoderm explants (Fig. s5A,B), suggesting that the impaired endodermal budding process was not caused by impaired cell proliferation or by apoptosis. These experiments indicate that Anxa1, Anxa4 and Anxa6 function redundantly in regulating lung endoderm budding process.

Discussion
Here, by combining live imaging, mouse genetics and mesenchyme-free lung endoderm culture system, we found that airway epithelial progenitor cells that migrate faster are more likely to become Sox9 + distal airway epithelial cells. This process is controlled by FGFR2-mediated ERK1/2 signaling. Using the lung endoderm culture system, we identified Anxa4 as a downstream target of ERK1/2 signaling. We further demonstrated, both in vitro and in vivo, that Anxa4 promotes airway epithelial cell migration. Loss of Anxa4 decreases airway epithelial cell migration during bud formation, while overexpression of Anxa4 promotes cell migration. We also found that Anxa4 −/− epithelial cells tend to exhibit a lag-behind behavior and contribute to stalk airways in vivo, suggesting that Anxa4 plays a role in promoting distal cell fate commitment. This lag behind behavior is most likely caused by decreased cell migration, as loss of Anxa4 does not affect cell proliferation or apoptosis. Attempts have been made to investigate the underlying mechanisms of branching morphogenesis at cellular level over the past decades. Studies combining fluorescent reporters, mouse genetics and live imaging have revealed the dynamics and kinematics of branching morphogenesis in a variety of model organs. These studies have shown that a variety of cellular behaviors, including local proliferation, cell migration, cell invasion, apical constriction, can contribute to branching morphogenesis in different contexts 12,14,16,18 . FGF10/FGFR2 signaling is essential for cell proliferation and cell migration during lung development 6,8,36 . Studies have shown that localized cell proliferation is not required for the initiation of bud formation 12,13 . Other studies have shown that ERK1/2-signaling-controlled cell migration is involved in lung endoderm budding and renal branching morphogenesis 17,40 . In the present study, we found that cell migration is involved in airway bud formation and that tip cells migrate faster than cleft cells. Further, changes in cell shape are known to accompany cell migration 41 , we found here that tip cells are more elongated than cleft cells, a result suggesting that we are here observing active cell migration.
A previous study used microarrays to profile the transcriptomes of cultured mesenchyme-free lung endoderm explants with the goal of identifying downstream targets of FGF10/FGFR2 signaling in the lung epithelial cells during branching morphogenesis 38 . This microarray study implicated approximately 200 genes, including several members of the Anxa family, in the initial stages of bud formation. Consistently, we here found that the expression levels of Anxa family members are increased over time in lung endoderm culture and identified Anxa4 as a downstream target of ERK1/2 signaling. We also showed that loss of Anxa4 impairs airway epithelial cell migration, without affecting cell proliferation. Mice deficient in Anxa1, Anxa2, Anxa5, or Anxa6 have been generated and used to evaluate the physiological roles of Annexins 39,[42][43][44] . However, all of the mice strains that lack a single Annexin are viable and exhibit normal development. It has been shown that the expression levels of other members of the Annexin family are altered in the tissues of Anxa1 −/− mice, suggesting the existence of reciprocal regulation between Annexin family members and of functional redundancy among Annexins 39 . Future studies that achieve the deletion of multiple Annexin genes will help to elucidate the precise functions of particular Annexin genes in embryonic lung branching morphogenesis.

Materials and Methods
Mice. The Shh Cre/+ 45 , Shh CreER/+ 45 , Rosa26-mTmG 46 and Fgfr2 F/F 47 mice have been described previously. All mice experiments were performed in accordance with the guidelines for the use and care of laboratory animals of the National Institute of Biological Sciences, Beijing. The experimental protocol was approved by the National Institute of Biological Sciences, Beijing (protocol number NIBS2012M0017). Mice were housed under standard environmental conditions (20-22 °C, 12-12 hr light-dark cycle) and provided food and water ad libitum. Animals were anesthetized by using Pentobarbital sodium before sacrifice. For live imaging experiments, pregnant females carrying Shh CreER/+ ; Rosa26-mTmG or Shh CreER/+ ; Anxa4 F/F ; Rosa26-mTmG or Shh CreER/+ ; Fgfr2 F/F ; Rosa26-mTmG were injected with one dose (30 μg/g) of Tamoxifen at E9.5. For the cell fate lineage tracing experiments, pregnant females were injected with one dose (75 μg/g) of Tamoxifen at E9.5.
Generation and genotyping of Anxa4 flox mice. A conditional, replacement-type targeting vector was produced by inserting one LoxP site 210 bp upstream of Anxa4 exon 3. A fragment containing a Loxp-flanked neo-cassette was cloned into a site 244 bp downstream of Anxa4 exon 3. The targeting construct was linearized and electroporated into C57/BL6 ES cells and was then selected with G418 on embryonic fibroblast feeder cells. Recombinant clones containing a floxed Anxa4 gene were identified by PCR using primers P1 (GGTGAACCATCTCTCGTCCTAAGCTCG) and P2 (CTGCTAACACATTCTCCCATCCGTCAC). The targeted clones were injected into C57BL/6J blastocysts, yielding 3 lines of chimeric mice that transmitted the Anxa4 flox allele through the germ line.

Time-lapse imaging of embryonic lung explants and imaging analysis. The procedures for
ex vivo time-lapse imaging have been described previously 48 . Briefly, lungs at E12.5 were dissected out and immediately embedded in 0.4% low melting-point agarose (Lonza) dissolved in the culture medium (1% insulin-transferrin-selenium +10 μM vitamin C and 1% penicillin/streptomycin + BGJb media). the culture dish with embedded lung explants were firstly cultured in a cell incubator (5% CO 2 , 37 °C) for 1 h, then were carried out and placed on a 37 °C heated platform for time-lapse images imaging. Time-lapse imaging was taken with a two-photon microscope (FV1000, Olympus) using a 25× water immersion objective. Imaging stacks of 512 × 512 pixels × 25 optical sections (xyzt sampling: 0.994 × 0.994 × 5 μm × 10 min) were acquired every 10 min for 5 h. For live imaging analysis, the images were opened by Imaris software, cells in the bud tip or cleft were distinguished by the end time point of live imaging: if the cells were in the two newly formed bud tips at the end time point, then we defined them as "tip cells"; if the cells were between the two newly formed bud tips at the end time point, then we defined them as "cleft cells". We traced the xy positions of GFP + tip cells and cleft cells at all time-points. All these xy position data were used for analyzed in cell migration trace plot, cell displacement to starting position and cell migration velocity. E-cadherin whole-mount staining. Lungs were dissected out from mouse embryos fixed with 4% PFA in PBS for 1 h at 4 °C. To facilitate the analysis of branching morphogenesis, whole lungs were stained with E-cadherin to visualize all airway epithelial cells. Whole-mount immunostaining was performed as previously described 31 .
Lung endoderm isolation and culture. Lung endoderm isolation was performed according to Weaver et al. 49 . Briefly, lung explants were dissected from ICR mice at E11.5 in HBSS, washed three times in Tyrode-Ringer's solution and then incubated in a pancreatin-trypsin solution for 5 min on ice. Lung endoderm explants were isolated from the mesenchyme using tungsten needles in DMEM/F12 media with 10% fetal bovine serum. The distal lung endoderm buds were cut off and embedded in 50% growth factor reduced Matrigel (Corning) and cultured in DMEM/F12 media supplemented with 800 ng/ml human recombinant FGF10 (Sino Biological). In some experiments, a MEK inhibitor (PD0325901, 1 μM, Selleck) was added at 1 h after culture initiation. In shRNA knockdown or Anxa4-overexpressing experiments, the lung endoderm was co-cultured with lentivirus added in 50% Matrigel and culture media at the time of initial culturing.

Isolation of primary lung epithelial cells and Transwell migration assays.
The procedures for primary lung epithelial cell isolation have been described previously 51 . Briefly, after removing the heart and trachea, E14.5 lung lobes were minced into very small pieces and digested in DMEM media with neutral protease (Worthington) and DNase I (Roche) at 37 °C for 20 min. After stopping digestion and passing cells through 40 μm filter, we next centrifuged samples for 10 min at 350 × g, re-suspended the cell pellet in RBC lysis buffer (BD Biosciences) to remove red blood cells and centrifuged samples again for 10 min at 350 × g. The cell pellet was suspended with DMEM/F12 plus 10% FBS and cultured in 35 mm dish for 1.5 hours. The supernatant was collected and centrifuged for 3 min at 350 × g to obtain purified primary lung epithelial cells, which were suspended with DMEM/F12 plus 1% FBS and quantified (cell number) using a hemocytometer. For the Transwell ® migration assays, 5 × 10 4 cells in 200 μl were seeded into the upper well of the Transwell ® apparatus (6.5 mm diameter, 8 μm pore size, Corning Costar); 600 μl of DMEM/F12 medium supplemented with 400 ng/ml human recombinant FGF10 was added into the lower chamber. Following incubation for 48 hours at 37 °C to allow cell migration, cells were fixed with 4% PFA and perform crystal violet staining.