ROBO2 signaling in lung development regulates SOX2/SOX9 balance, branching morphogenesis and is dysregulated in nitrofen-induced congenital diaphragmatic hernia

This study was carried out in strict accordance with FELASA guidelines [32] and with European (European Union Directive 86/609/EEC) regulation. The protocol was approved by the Direção Geral de Alimentação e Veterinária (DGAV 021328).

Female Sprague–Dawley rats (225 g; Charles-River; Spain) were maintained in appropriate cages under controlled conditions and fed with commercial solid food. The rats were mated and checked daily for vaginal plug. The day of plugging was defined as embryonic day (E) 0.5 for time dating purposes. According to the nitrofen-induced CDH rat model [33, 34], at E9.5, randomly selected pregnant rats were exposed to 100 mg of nitrofen (2,4-dichlorophenyl-p-nitrophenylether). At different time points (E13.5, E15.5, E17.5, E19.5, and E21.5), fetuses were harvested by cesarean section. After fetal decapitation, a thoracic-laparotomy was performed under a binocular surgical microscope (Leica, Wild M651.MSD, Switzerland) to inspect the diaphragm and harvest the organs.

The fetuses were divided into two groups: a control group (control) with fetuses exposed to olive oil alone, and hypoplastic group (hypoplastic) with those exposed to nitrofen. Regarding this experimental design, the assessment of diaphragmatic defects by surgical inspection at E13.5 and E15.5 is impractical. Therefore, for early gestational age, the hypoplastic group refers to the fetuses exposed to nitrofen (independently of CDH development), while at later gestational ages (E17.5, E19.5, and E21.5), the hypoplastic group refers to fetuses exposed to nitrofen which developed CDH. Lungs were either snap-frozen in liquid nitrogen for protein extraction or fixed in 4% paraformaldehyde for immunohistochemistry. Lungs of fetuses at 13.5 days post-conception were also dissected to perform fetal lung explants cultures and posterior Western blot analysis. GPower 3.1.9.4 (Franz Faul, Universitat Kiel, Germany) was used for sample size calculation. In total, 18 dams and 165 embryonic rats were used in this study.

Normal and hypoplastic lungs of different gestational ages (E13.5–21.5) were fixed in 4% paraformaldehyde and embedded in paraffin as previously described [35]. 4 µm sections were placed onto glass microscope slides. Heat-induced antigen retrieval was performed with a citrate buffer. Sample sections were blocked with 4% fetal bovine serum for 1 h at room temperature and incubated with primary antibodies [ROBO1 (Cat No. sc25672) and ROBO2 (Cat No. sc16615), 1:200, overnight (ON), 4 ºC, Santa Cruz Biotechnology Inc, USA; SOX2 (Cat No. AF2018) and SOX9 (Cat No. AF3075), 1:100, ON, 4 ºC, R&D systems, USA] diluted in phosphate-buffered saline (PBS1x). Negative control reactions included omission of the primary antibody for which the immunoreactive staining was not observed. Tissue sections for negative control, ROBO1, SOX2, and SOX9 were labeled with a streptavidin–biotin immunoenzymatic antigen detection system (Cat No. TL-125-QHD, Thermo Scientific, USA) according to the manufacturer’s instructions and visualized with a diaminobenzidine tetrahydrochloride solution (Cat No. TA-125-QHDX, Thermo Scientific, USA) [36]. Tissues sections for ROBO2 and respective negative control were labeled with biotinylated anti-goat IgG (Cat No. BA-9500, Vector Laboratories, UK) followed by streptavidin–horseradish peroxidase incubation and visualized with a diaminobenzidine tetrahydrochloride solution (DAB, Cat No. TA-125-QHDX, Thermo Scientific, USA). Sections with and without primary antibodies were simultaneously processed and analyzed. The time expended in DAB solution was variable according to the developmental stage, but equally matched between control and hypoplastic sections, allowing the quantification of immunohistochemical signals. The percentage of stained cells per microscopic field was scored in four independent areas per section (four sections per each experimental group) by two blinded observers. Scoring was as follows: 0, 0–1% cells/pulmonary structure; 1, 1–25% cells/pulmonary structure; 2, 25–50% cells/pulmonary structure; 3, 50–75% cells/pulmonary structure; 4, 75–100% cells/pulmonary structure in accordance with [37]. At least three independent experiments were performed for each antibody tested. In each experiment, a different set of slides comprising the whole range of gestational ages was used. Different and unrepeated animal samples were selected for each group (gestational age). Six different animals were examined for each group per studied antibody. All sections were scanned with Olympus BX61 Upright Microscope (Olympus corporation, Japan) and independently evaluated by two investigators.

Normal and hypoplastic lungs from different gestational ages (E13.5–E21.5) were processed for Western blot analysis in accordance with previously described methods [38, 39]. Briefly, 15 µg of protein were loaded onto 10% acrylamide mini gels, electrophoresed at 100 V at room temperature, and then transferred to nitrocellulose membranes (HybondTM-C Extra, GE Healthcare Life Sciences, UK). Blots were blocked in 5% bovine serum albumin and probed with primary antibodies to ROBO1 (1:500, ON, 4 ºC; Cat No. sc25672, Santa Cruz Biotechnology Inc., USA), ROBO2 (1:250, ON, 4 ºC; Cat No. sc16615, Santa Cruz Biotechnology Inc. USA), SOX2 (1:250, ON, 4 ºC; Cat No. AF2018, R&D system, USA), SOX9 (1:250, ON, 4 ºC; Cat No. AF3075, R&D system, USA), non-phospho (Active) β-Catenin (Ser33/37/Thr41) (1∶5000; Cat No. #4270, Cell Signaling Technology Inc., USA), total β-Catenin (1∶30,000; Cat No. #NBP1-54,467, NOVUS Biologicals, USA), and BMP4 (1:250, ON, 4 ºC; Cat No. sc-6896, Santa Cruz Biotechnology Inc., USA) according to the manufacturer's instructions. For loading control, blots were probed with β-Tubulin (1∶150.000; Cat No. ab15568, Abcam, USA). After this, membranes were incubated with a secondary horseradish peroxidase conjugate, developed with Clarity West ECL substrate (Cat No. 1705060, Bio-Rad, USA) and the chemiluminescent signal was captured using the Chemidoc XRS (Bio-Rad, USA) [36].

Quantitative analysis was performed with Quantity One 4.6.5 1-D Analysis Software (Bio-Rad, USA). Three independent experiments were performed (n = 3). In total, nine animals were used in each group (gestational age/condition) per antibody.

The fetal lung explants were incubated in a 5% CO₂ incubator at 37 °C for 96 h, and the medium was replaced every 48 h. The branching morphogenesis was monitored daily by photographing the explants. At day 0 (D0: 0 h) and day 4 (D4: 96 h) of culture, the total number of peripheral airway buds (branching) of all lung explants was determined by counting the number of peripheral airway epithelial buds of the developing respiratory tree [37, 40].

Ex vivo cultures of normal lung explants were complemented daily with distinct doses of recombinant rat ROBO1 Fc Chimera (0.04, 0.4, 4 and 40 ng/mL, R&D system; Cat No. 1749-RB-050), recombinant human ROBO2 Fc Chimera (0.04, 0.4, 4 and 40 ng/mL, R&D system; Cat No. 3147-RB-050), or recombinant human IgG1 Fc (used as control at higher dose, R&D; Cat No. 110-HG-100). These doses were selected according to the literature [41, 42]. Lung explants were obtained in three independent experiments (n ≥ 9 for each dose tested). After 4 days in culture, control, ROBO1 and ROBO2 inhibited lung explants were analyzed for branching morphogenesis in terms of area, external and internal epithelial perimeter, and the number of peripheral airway buds. In addition, SOX2, SOX9, BMP4, total and non-phospho (active) β-Catenin were quantified by Western blot at day 4 in all experimental groups.

All quantitative data are presented as mean ± standard error of the mean (SEM). The statistical analysis was performed by two-way ANOVA for lung condition (normal and hypoplastic) and embryonic day (E13.5, E15.5, E17.5, E19.5, and E21.5) on protein expression level. One-way ANOVA was performed for the number of peripheral airway buds and protein expression levels on recombinant rat ROBO1 Fc Chimera (0.04, 0.4, 4, 40 ng/mL), and recombinant human ROBO2 Fc Chimera (0.04, 0.4, 4, 40 ng/mL). The parametric test assumptions were previously verified, and an additional LSD test was used for post-test analysis. Statistical analysis was performed using the statistical software IBM SPSS Statistics 24.0. Statistical significance was confirmed at p < 0.05.

To determine the molecular changes for ROBO1, ROBO2, SOX2 and SOX9 in experimental-CDH, the relative expression levels of each protein were analyzed by Western blot at the distinct gestational ages: E13.5, E15.5, E17.5, E19.5, and E21.5. Subsequent analysis revealed distinct molecular profiles for receptors (Fig. 1a–c) and epithelial progenitor markers (Fig. 1d and e), in normal and hypoplastic fetal lung development. Specifically, the progress of normal fetal lung development was defined by unchanged ROBO1 (Fig. 1b) and SOX2 (Fig. 1d) expression levels at E13.5, E15.5, E17.5 and E21.5, with E19.5 as the only affected stage in which the overexpression of ROBO1 (Fig. 1b) and SOX2 (Fig. 1d) was visible. In contrast, separate molecular profiles for ROBO1 (Fig. 1b) and SOX2 (Fig. 1d) were visualized in the experimental-CDH. While on the one hand, the ROBO1 (Fig. 1b) was decreased at E19.5 and overexpressed at E21.5, the SOX2 (Fig. 1d) was revealed to be downregulated at E15.5 and overexpressed at later gestational ages (E17.5-, E19.5-, and E21.5-hypoplastic versus control).

Regarding the relative expression levels of ROBO2 (Fig. 1c) and SOX9 (Fig. 1e), Western blot analysis identified an opposite effect on ROBO2 and SOX9 at later (E17.5, E19.5 and E21.5) versus early (E13.5, E15.5) developmental stages, in which the significant overexpression of ROBO2 and SOX9 depletion marked the normal lungs. In contrast, after experimental-CDH induction, ROBO2 (Fig. 1c) was significantly increased at E13.5 and E15.5 and downregulated at E17.5, while SOX9 (Fig. 1e) showed an important decrease at E15.5 and was overexpressed at E17.5 and E19.5 in experimental-CDH versus normal lungs.

To further explore these molecular profiles across hypoplastic versus normal fetal lung development, in vivo spatiotemporal distributions for ROBO1, ROBO2, SOX2, and SOX9 were analyzed by immunohistochemistry and quantified by pulmonary structure.

Our results showed ROBO1 first expressed in mesenchymal and epithelial cells defining the undifferentiated tissues at E13.5 (Fig. 2aA) and E15.5 (Fig. 2aB). ROBO1+ was next expressed in bronchi and primordia of bronchiole (Fig. 2aC and aD) at E17.5, characterizing the first differentiated pulmonary structures in normal lungs. As lung development progressed, ROBO1 was observed in bronchi, terminal bronchiole, and bronchioalveolar junction (BADJ) at E19.5 (Fig. 2aE and aF) and E21.5 (Fig. 2aG and aH), with important gain-or-loss expression after experimental-CDH induction (Fig. 2aa–ah). Indeed, quantification of immunohistochemistry signals demonstrated the decrease of ROBO1 expression in bronchi at E17.5 and E19.5; primordia of bronchiole at E17.5; and terminal bronchiole at E19.5 (Fig. 2b). Conversely, the overexpression of ROBO1 was visualized in bronchi, terminal bronchiole, and BADJ at E21.5 in experimental-CDH versus normal lungs (Fig. 2b). Finally, unexpected new ROBO1+ cells were observed in the alveolar duct at E21.5 after induction of experimental-CDH (Figs. 2ag–ah and 2b).

Regarding ROBO2, the immunohistochemistry assay demonstrated restricted epithelial ROBO2 expression in undifferentiated tissues, with ROBO2+ cells staining all pulmonary structures in normal (Fig. 3aA–aH) and hypoplastic (Fig. 3aa–ah) fetal lungs. Interestingly, compared to normal lungs, the quantification of stained cells indicated ROBO2 overexpressed in epithelium at E13.5 and E15.5; bronchi and primordia of bronchiole at E17.5; and in the alveolar duct and mesenchyme at E21.5 (Fig. 3b) in the nitrofen-induced CDH rat model. In contrast, a significant decrease in ROBO2 was observed in bronchi at E19.5; terminal bronchiole and BADJ at E19.5 and E21.5 (Fig. 3b) after experimental-CDH induction. New ROBO2+ cells were observed in mesenchyme at E13.5, E15.5, and E17.5 in experimental-CDH versus normal lungs.

To define the cellular profile of proximal and distal epithelial progenitors, the spatiotemporal distribution for SOX2 and SOX9, respectively, were analyzed across the distinct gestational ages (from E13.5 to E21.5) in normal and hypoplastic lungs. Immunohistochemistry revealed initial SOX2 (Fig. 4aA and aB; and Fig. 4aa and ab) expression in undifferentiated epithelial cells that progressively populate the proximal pulmonary structures in normal fetal lung development. Specifically, SOX2 was visualized in bronchi at E17.5 (Fig. 4aC and aD), E19.5 (Fig. 4aE and aF) and E21.5 (Fig. 4aG and aH); primordia of bronchiole at E17.5; terminal bronchiole and BADJ at E19.5 and E21.5 across normal fetal lung development. Regarding experimental-CDH, overexpression of SOX2 (Fig. 4b) was detected in the undifferentiated epithelium at E13.5 (Fig. 4aa) and E15.5 (Fig. 4ab); terminal bronchiole at E19.5 (Fig. 4ae and af) and E21.5 (Fig. 4ag and ah); and bronchi at E21.5. Interestingly, no SOX2+ cells were observed in primordia of bronchiole at E17.5 in experimental-CDH, whereas significant SOX2 loss characterized the BADJ at E19.5 and E21.5 (Fig. 4b).

Concerning SOX9, the immunohistochemistry assay identified an initial SOX9 expression in epithelium at E13.5 and E15.5 and mesenchyme at E15.5 that give rise to wide staining in all pulmonary structures in normal (Fig. 5aA–aH) and hypoplastic lungs (Fig. 5aa–ah). Indeed, SOX9 was detected in bronchi at E17.5, E19.5, and E21.5; primordia of bronchiole at E17.5; terminal bronchiole and BADJ at E19.5 and E21.5; and in alveolar duct at E21.5 in normal (Fig. 5aC–aH), and hypoplastic lungs (Fig. 5ac–ah) for which the more significant differences were observed after quantification of immunohistochemical signals (Fig. 5b). Particularly in experimental-CDH, SOX9 was downregulated in epithelium at E15.5 and in alveolar duct and mesenchyme at E21.5, whereas their overexpression was simultaneously observed in bronchi at E17.5, E19.5, and E21.5; primordia of bronchiole at E17.5; terminal bronchiole at E19.5 and E21.5; BADJ at E19.5; and mesenchyme at E17.5 and E19.5 (Fig. 5b).

These new findings urged us to explore the function of ROBO1 and ROBO2 in fetal lung development, particularly in branching morphogenesis and in SOX2/SOX9 expression profiles. For that, lung explant cultures at E13.5, previous reported as comparable to the in vivo lung both in structure and function, were selected as a model, in which restricted inhibition of ROBO1 or ROBO2 was performed followed by branching morphogenesis analysis and SOX2, SOX9, total and non-phospho (active) β-Catenin and BMP4 quantification at day 4.

Ex vivo fetal lung explant cultures, at E13.5, were supplemented with IgG; recombinant ROBO1, or ROBO2 chimera proteins. When added to the culture medium, ROBO1 and ROBO2 chimera proteins selectively inhibit ROBO1 and ROBO2 function [41, 42]. In addition, apart from the ROBO1 and ROBO2 quantification, Western blot analysis was also performed for SOX2, SOX9, total, and non-phospho (active) β-Catenin and BMP4 at day4.

Our results showed inhibition of ROBO2 as stimulator of branching morphogenesis from 0.04-to-40 ng/mL (Fig. 6a, b), while no significant differences in terms of number of peripheral airway buds were observed after ROBO1 inhibition (Fig. 6a, b). Molecular analysis showed significant ROBO1 and ROBO2 downregulation after supplementing the medium with recombinant rat ROBO1 Fc Chimera and recombinant human ROBO2 Fc Chimera, respectively, both versus recombinant human IgG Fc used as control (Fig. 6c, d). In addition, the overexpression of SOX2 (Fig. 6e) and total β-Catenin (Fig. 6g) with downregulation of BMP4 (Fig. 6i) and unchanged non-phospho (active) β-Catenin (Fig. 6h) and SOX9 (Fig. 6f) was identified after ROBO1 inhibition contrast, with a significative decrease in both SOX2 (Fig. 6e) and BMP4 (Fig. 6i); overexpression of SOX9 (Fig. 6f), non-phospho (active) (Fig. 6h) and total β-Catenin (Fig. 6g) observed after ROBO2 inhibition in promotion of fetal lung branching morphogenesis.