Main

TGF-α and EGF are two members of the EGF family of growth factors which play an important role in embryogenesis(1) and in fetal development(2, 3). These factors have similar biologic properties and show 42% sequence homology(4, 5). The molecular analysis of murine fetal tissue has demonstrated that TGF-α is more highly expressed than EGF during fetal development(6–8). Both TGF-α and EGF mediate their actions by binding to the extracellular domain of a common transmembrane glycoprotein receptor (EGF receptor) with activation of the tyrosine kinase in the intracellular domain(9, 10). These growth factors were first shown to be mitogenic for epithelial cells and fibroblasts in a variety of tissues, including lung(4, 9, 11) and appeared to be most active in the least differentiated epithelial cells(12). The role of EGF in lung organogenesis was clearly demonstrated in 1980 by Goldin and Opperman(13): grafting agarose pellets containing EGF along-side the embryonic chick tracheal epithelium induced supernumerary tracheal buds. Furthermore, in rat or mouse embryonic lung explants, EGF was shown to stimulate cell proliferation in both mesenchyme and epithelium, resulting in an increase in branching activity(14, 15) or an enlargement of the lung(16).

There is considerable evidence that these growth factors are also involved in lung maturation. EGF has proved to have promoted maturation of distal airways in rabbits(17) and in lambs(12) and to accelerate alveolar type II differentiation in both the lung of the fetal rhesus monkey(18) and in cultured rat type II pneumocytes(19). EGF stimulated surfactant production in human explants(20) and increased both antioxidant enzyme and surfactant system development during hyperoxia(21). Injection of EGF into the amniotic fluid of fetal rhesus monkeys during late gestation accelerated the differentiation of tracheal mucus secretory cells and also increased the amount of secretory product released in the airway lumen, but had no further effect on cell proliferation(22).

EGF and TGF-α have both been localized by immunohistochemistry in rat and mouse lung(2, 15), and in human fetal lung and trachea(23, 24). TGF-α mRNA was found in RNA extracts of human fetal lungs(24) and was reported higher in the canalicular rather than saccular fetal rat lung(25). However, the source of TGF-α in the lung is still unclear. EGF and TGF-α bind to EGF receptor which was detected by immunohistochemistry in mouse and ovine lung(15, 26) and in human fetal respiratory epithelial cells(27). However, most of these studies focus on only part of the gestational development, and the results can vary with species. The study of the localization and the synthesis of both EGF and TGF-α together with their site of activity through EGF receptor has never been carried out in human fetuses nor throughout the whole period of gestation.

The role of EGF and TGF-α has been suggested in pathologic conditions, such as acute and chronic lung disease in the neonate(23). However, their exact site of production and binding still has to be evaluated in normal lung development. The aim of the study was to identify the site of synthesis of EGF and TGF-α usingin situ hybridization and to analyze the cellular distribution of EGF, TGF-α proteins, and EGF receptor using immunohistochemistry in both the human trachea and in the lung throughout the whole period of fetal development.

METHODS

Fetal tissue material. Fourty-seven embryos and fetuses, with a GA (menstrual age) ranging from 10 to 41 wk, were obtained from spontaneous abortions or medical inductions. The age distribution is indicated inTable 1. All of the fetuses were well preserved, and none was shown to have any respiratory abnormality or infection. They were associated with neither polyhydramnios nor oligohydramnios. Different airway tissue specimens were collected from the trachea and the lung. From 39 fetuses between 10 and 41 wk of GA, tissue samples were immediately fixed in 15% phosphate-buffered formalin and embedded in paraffin for immunohistochemistry. In eight fetuses ranging from 12 to 33 wk of GA, one part of the lung was immediately frozen in liquid nitrogen for in situ hybridization, and another part was fixed in formalin for immunohistochemistry. One these frozen fetal lungs (25 wk of GA) was also used for tissue extraction and immunoblotting.

Table 1 Immunolocalization of EGF, TGFα, and EGF receptor proteins and expression of EGF and TGFα mRNA transcripts
Full size table

In situ hybridization. Frozen sections of the trachea and lung tissue 5 μm thick were collected on chrome-alum (0.4%) gelatin(0.5%)-coated microscope slides which were immediately fixed in 4% paraformaldehyde (pH 7.4) for 10 min, washed in PBS (20 mM sodium phosphate-0.7% NaCl pH 7.4), and then dehydrated in ethanol and stored in ethanol 70% at 4°C before use.

The molecular probes used were TGF-α cDNA, 900 bp long, cloned in theEco RI site of the PBR 327 and EGF cDNA 1700 bp long, cloned in theEco RI site of the PUC kindly provided by Dr. Bell (Howard Hugues Institute, Chicago). The radiolabeled cDNA probes were prepared by the random priming technique (Amersham Corp., Little Chalfont, UK), using35 S-labeled-dCTP (specific activity: 650 Ci/mM) (Amersham) and were then purified through a Sephadex G50 column. The filtration was followed by ethanol precipitation. Specific activity of the resulting 35S-labeled DNA was 3.108 cpm/μg. The slides were pretreated by first heating them to 70°C in 2× SSC (1 × SSC = 0.15 M sodium chloride and 0.015 M sodium citrate) for 10 min to facilitate probe penetration and were then dipped in a solution which contained triethanolamine (0.1 M, pH 8) and acetic anhydride (0.25%) at room temperature for 10 min and shaken. They were carefully rinsed. The denatured labeled DNA was mixed with 50% formamide, 0.6 M NaCl, 10 mM Tris, 1 mM EDTA, 1 × Denhart's solution, 250 μg/mL denatured salmon sperm DNA, 500 μg/mL tRNA, 10 mM DTT, and with 10% dextran sulfate. The hybridization was performed using 10 μL of hybridization solution on each slide, corresponding to 150 000 cpm. Hybridization was carried out for 18 h at 42°C in a humidified chamber. After hybridization, the slides were first washed in SSC of degrading concentration, dehydrated, and then air-dried. Finally, sections were dipped in K5 emulsion (Ilford lim., Mobberley, Cheshire, UK) for autoradiography, exposed at 4°C for 6 wk, developed, counterstained with hematoxylin, and then photographed.

As the probes were cDNA, RNase was used as a negative control. In each case, and for each probe, the control slides were incubated with 10 μg/mL RNase for 1 h at 37°C, before pretreatment and hybridization. Frozen sections of bronchial tumors were taken as positive controls.

Tissue extraction and immunoblotting. Frozen lung tissue was reduced to powder in liquid nitrogen, washed in PBS containing 1 mM phenylmethylsulfonyl fluoride, and 5 mM EDTA and centrifuged (10 000 ×g, 10 min at 4°C). Proteins were extracted overnight at 4°C from the resulting pellet in 50 mM Tris-HCl, pH 7.5 containing 2 mM urea and 1 M NaCl. After centrifugation (10 000 × g, 10 min at 4°C) the supernatant was dialyzed against water and lyophilized. Proteins were finally dissolved in electrophoresis sample buffer.

Proteins were separated by SDS-PAGE in 20% polyacrylamide gels (Phast system, Pharmacia Biotech Inc.). The resulting gels were equilibrated in the transfer buffer: 25 mM Tris-HCl, 192 mM glycine, 20% (vol/vol) methanol, pH 8.3. The proteins were then electrophoretically transferred to nitrocellulose membranes. Proteins were fixed for 15 min in 0.2% glutaraldehyde. Membranes were incubated for 1 h in 5% (wt/vol) fat-free dry milk in PBS + 0.05% Tween 20 (Tween-PBS) and incubated overnight at 4°C, with the relevant antibody: the monoclonal mouse anti-human TGF-α (Ab-2) and the polyclonal rabbit anti-human EGF (Ab-3), both purchased from Oncogene Science, Inc. (Manhasset, NY) and diluted in Tween-PBS at 1 μg/mL. Membranes were incubated for 1 h with an anti-mouse IgG alkaline phosphatase conjugate (1/1000, Sigma Chemical Co., St. Louis, MO) or with an anti-rabbit IgG alkaline phosphatase conjugate(1/1000; Chemicon, Temecula, CA) in Tween-PBS and then developed with the 5-bromo-4-chloro-3-indolyl phosphate/nitro blue tetrazolium substrate(Chemicon). Control included highly purified TGF-α and EGF (Sigma).

Immunohistochemistry. Paraffin sections were cut (3 μm), mounted on gelatin-coated slides, and then dried overnight at 50°C. The tissue sections were then deparaffinized with xylene and rehydrated first in graded ethanol baths and then in distilled water and PBS, pH 7.2. Each section was pretreated with saponin in distilled H2O, for 50 min at 37°C, and then incubated in a 0.3% hydrogen peroxide bath for 5 min at room temperature, to remove endogenous peroxidase activity, and also rinsed in PBS. A blocking reagent (6% goat serum) was added for 5 min. The sections were then rinsed twice with PBS before incubation with the different primary antibodies.

The monoclonal mouse anti-human TGF-α (Ab-2) (Oncogene Science), reacted with the COOH-terminal 34-50 residues and showed no cross reactivity with EGF(28). The monoclonal mouse anti-human TGF-α and the polyclonal rabbit anti-human EGF (Ab-3) (Oncogene Science) primary antibodies were diluted to 1:80 in PBS and incubated overnight at 4°C. The monoclonal mouse antibody raised against EGF-R (MAb) was obtained from Amersham Corp. (Buckinhamshire, UK) and used at a concentration of 1:15 in PBS for 30 min at 37°C.

The primary antibodies were revealed with a Kit DAKOLSAB (K680) as follows. After two rinses, the sections were incubated for a further 10 min in an anti-mouse or anti-rabbit IgG biotinylated antibody, and then for 10 min in labeled streptavidin. The sections were exposed to a chromogen substrate solution (3% 3-amino-9-ethylcarbazole) for a further 10 min in labeled streptavidin. After rinsing, the sections were counterstained with hematoxylin, then dehydrated and cover-slipped. Negative controls were carried out using the same procedure and by 1) omitting the primary antibody and 2) using nonimmune mouse or rabbit serum.

RESULTS

The specificity of TGF-α and EGF antibodies were tested using a Western blot technique (Fig. 1). In a protein extract derived from a lung tissue at 25 wk of GA, the anti-EGF antibody was shown to identify a peptide with a similar Mr than that for pure EGF. This antibody did not cross-react with TGF-α. Similarly the anti-TGF-α antibody recognized pure TGF-α, not EGF, and a lung protein with the same Mr as that of TGF-α.

Figure 1
figure 1

Western blot. Purified EGF (0.1 μg, lanes 1), TGF-α (0.1 μg, lanes 3), or lung protein extract from a fetus at 25 wk of GA (lanes 2) were electrophoresed in 20% polyacrylamide gels and blotted onto nitrocellulose membranes. Nitrocellulose blots were developed with the anti-EGF (left) or the anti-TGF-α (right). Molecular masses in kilodaltons of standard proteins (Pharmacia) are reported on theright.

Full size image

The expression of TGF-α and EGF genes was analyzed according to the different stages of human fetal lung development: 1) pseudoglandular period between 10 and 16 wk of GA, 2) canalicular period extending to 24 wk of GA, and 3) terminal sac-alveolar period exceeding 24 wk of GA. The localization of both TGF-α and EGF proteins and their common receptor was studied through these three phases. Particular attention was paid to the presence of these factors according to the degree of maturation of both the surface respiratory epithelium and of the glands along the large airways.

Pseudoglandular stage. During the pseudoglandular stage, both branching morphogenesis and cell proliferation were predominant, and most bronchial branches were formed. Along the large airways, the first ciliated and secretory cells appeared after 12-13 wk, and the epithelium began to grow and bulge into the mesenchyme to form the first tubular glands. A few cells containing TGF-α and EGF mRNA were found around the cartilaginous rings and in chondrocytes of the proximal airways, as well as in the pleura and in the interlobar septa (not shown). In the clefts of the branching epithelial tubes, mesenchymal cells expressed EGF and TGF-α mRNA transcripts, but epithelial cells remained unlabeled throughout the period.

TGF-α protein was present with a higher immunoreactivity than EGF protein, but both exhibited a similar distribution which predominantly concerned epithelial cells. From 9 to 12 wk of GA, faint submembrane or diffuse immunostaining was seen in the undifferentiated columnar epithelium of the large airways and of the branching tubules (Fig. 2). During the following weeks, the epithelium became more differentiated, after a progressive cranio-caudal maturation. Both growth factors were present in ciliated, secretory, and undifferentiated cells lining the tracheal and bronchial lumens and the growing glands. Throughout this period, the intensity of immunostaining was higher in the proximal part of the airways than in the distal branching buds. The perichondral mesenchyme, a few chondrocytes, the smooth muscle along the proximal airways, and the vessels were all immunostained. In the distal mesenchyme, both growth factors were detected in a few fibroblasts, as well as in mesothelial and endothelial cells. Throughout this period, the EGF receptor was detected in the same cells and with the same intensity of immunostaining as that of TGF-α (Fig. 2).

Figure 2
figure 2

Immunohistochemistry in fetal human lung during the pseudoglandular period. At 10 wk of gestation, immunoreactivity to TGF-α(A), EGF (B), and EGF receptor (C) is localized in the epithelium of the branching channels and in a few mesenchymal cells. Negative control (D). Bar equals 100 μm.

Full size image

Canalicular period. During the canalicular period, the fluidfilled distal airspaces were lined by flattened epithelial cells. Capillaries invaded the mesenchyme and began to approach the cuboidal airway epithelium. The large airways were lined by numerous ciliated and secretory cells, and the glands exhibited mucus secretion. TGF-α and EGF mRNA appeared to be confined to the mesenchymal cells and were more particularly present in the dense connective tissue of the pleura, of the interlobular septa, and of the large airways and vessels (Fig. 3). Both transcripts were more abundant in the mesenchyme, rather than during the pseudoglandular period. TGF-α and EGF mRNA were also observed in a few chondrocytes of the tracheal and bronchial cartilage, but could not be identified in neither epithelial cells nor smooth muscle cells.

Figure 3
figure 3

In situ hybridization of TGF-α(A) and EGF (B) in fetal human lung during the canalicular period. At 20 wk of gestation, pleural, peribronchiolar, and perivascular mesenchyme exhibit TGF-α, EGF mRNA expression. Negative control(C). Pleura (arrow), vessel (v), small bronchus(b). Bar equals 100 μm.

Full size image

TGF-α and EGF immunoreactivity was easily demonstrated in the respiratory epithelium lining the trachea and the large bronchi(Fig. 4). Both growth factors and receptor were diffusely present in the cytoplasm of each type of epithelial cell, and particularly in the apical domain of ciliated cells. TGF-α, EGF, and their receptor were localized in the collecting ducts and in the growing glands. Only a few mesenchymal cells were immunostained in the submucosa, but all the endothelial and muscle cells of the vessels, as well as the tracheal and bronchial muscles, appeared positive. In the distal airways, the epithelial immunoreactivity was lower than that in the more mature proximal airways(Fig. 5). The mesenchyme contained numerous positive capillaries along the canalicular epithelium. The mesothelium was immunostained with the TGF-α, EGF, and EGF receptor antibodies.

Figure 4
figure 4

Immunolocalization of TGF-α in the human trachea at 18 wk of gestation (A). The ciliated and basal cells in the surface epithelium and the cells budding in the mesenchyme to form glands are immunostained, whereas only a few mesenchymal cells are positive. Negative control (B). Bar equals 100 μm.

Full size image
Figure 5
figure 5

Immunolocalization of TGF-α (A), EGF(B), and EGF receptor (C) in fetal human lung during the canalicular period. At 20 wk of gestation, immunoreactivity is present in the flattened epithelial cells of the distal airways, in mesothelial cells(arrowhead) along the pleura and in rare mesenchymal cells. Note the higher immunoreactivity in the proximal airways (Fig. 4) where epithelial cells are more differentiated. Negative control(D). Bar equals 100 μm.

Full size image

Saccular and alveolar period. During the saccular and alveolar period, a progressive thinning of the epithelium and a protrusion of capillaries into the airspaces resulted in the development of the first blood-air barriers necessary for extrauterine survival. The large airways were lined by a mature respiratory epithelium and exhibited mature glands with both mucous and serous cells. TGF-α and EGF mRNA was located in the mesenchymal cells of the tracheal and bronchial submucosa, along the cartilaginous rings, and around the growing glands. No glandular or surface epithelial cells synthesized TGF-α and EGF mRNA. In the lung, the septa between the alveolar structures were covered with hybridization grains. Under light microscope, and using frozen sections, the alveolar septa were too thin to differentiate alveolar, endothelial and mesenchymal cells, and to detect the exact site of synthesis (Fig. 6).

Figure 6
figure 6

In situ hybridization of TGF-α(A) in fetal human lung during the alveolar period. At 32 wk of gestation, high levels of TGF-α mRNA expression are detected in the alveolar septa. Negative control (B). Bar equals 100 μm.

Full size image

During the period exceeding 24 wk (GA), no significant changes were detected in the immunoreactivity of TGF-α, EGF, and EGF receptor in the respiratory epithelium of the large airways. In the glands, most of the labeled cells were located at the periphery of the acini and appeared to be serous cells (Fig. 7A). In the distal airspaces(Fig. 7B), the bronchiolar epithelium and alveolar cells were immunostained with TGF-α, EGF, and their receptor. In the mesenchyme, the vessels still remained positive, in contrast to the immunoreactivity of mesenchymal and mesothelial cells which became negative.

Figure 7
figure 7

Immunolocalization of TGF-α in fetal human trachea and lung at 38 wk of gestation. TGF-α is observed in the mature surface epithelium of the trachea (A) with a higher reactivity in the apical domain of epithelial cells. In the glands immunostaining is most strongly detected in serous cells (arrows). The degree of immunoreactivity is higher in the large airway epithelial cells than in respiratory bronchiolar epithelium (b) (B). Alveolar cells(arrowheads) and muscle cells in the vessels (v) are also labeled. Negative control (C). Bar equals 100 μm.

Full size image

DISCUSSION

Over the last 15 y the importance of the EGF growth factor family has been shown in lung development(11, 21). This is the first study that reports the site of the synthesis of TGF-α and EGF, as well as their interaction with target cells through their common receptor in the course of the complete human fetal lung development. TGF-α and EGF mRNA appeared to be confined to the mesenchymal cells, and mainly to the dense connective tissue surrounding the large bronchial and vascular structures and the pleura. Using in situ hybridization, mesenchymal cells expressed only low levels of RNA transcripts during the pseudoglandular period, although cell proliferation was active, and the labeling was intense during the following canalicular and alveolar periods. In previous reports, the localization of transcripts had been studied only in mice, and hybridization signals for EGF precursor mRNA had also been identified in the mesenchymal cells of the lung between 13 and 16 d of gestation(2). In humans, TGF-α mRNA was recently analyzed from fetal lung homogenates and detected throughout the period of examination (10-24 wk). However, the alveolar and gland development was not studied, and the site of synthesis was not identified(24). By using cDNA probes we could loose a few signals in any type of cells; however, the main synthesis is obviously in the mesenchyme.

TGF-α and EGF proteins, as well as their common receptor, were localized with immunohistochemistry in airway epithelial cells, at all levels, from the trachea to distal airspaces and during the whole period of fetal development. This suggests that in humans, as in experimental models, EGF and TGF-α play an important role in lung development, not only in cell proliferation and branching morphogenesis, but also later on in epithelial cell differentiation. A similar distribution of TGF-α was already reported in the surface epithelium of human fetal lung at midgestation(24). The absence of modification of TGF-α during late gestation is not surprising, because the tracheal surface epithelium is already mature and ciliated at 24 wk(29). Curiously, in a previous report, TGF-α was not detected in either the fetal nor the adult respiratory epithelium, but was observed in the smooth muscle of vessels and in glandular cells of adult trachea(30). The results concerning the distribution of EGF are most discordant. In the present study, EGF immunostaining was found in all kinds of respiratory epithelial cells lining the airways, whatever the degree of cell differentiation. These findings support those reported in rat(31) and fetal mouse(2, 15), but are in conflict with a previously reported study on human fetal lung(23). The distribution of immunostaining of EGF in the surface epithelium was shown to be limited to nonciliated cells in early fetal trachea or to small clusters of nonciliated cells in older trachea, but never detected in bronchiolar epithelium or in distal airspaces at any age of development(23). In the undifferentiated cells of the growing glands and in the vascular or bronchiolar smooth muscle, we observed a high immunoreactivity with EGF and TGF-α antibodies. The distribution of TGF-α was not yet reported in human fetal glands, but was already noted in adults(30). In older fetuses, as well as in neonates(23), EGF and TGF-α are mainly detected in serous cells.

We observed that the distribution of EGF receptor was mainly in epithelial cells whatever age of development. These results agree with those reported in human(27, 32), ovine(26), and mouse(15) lungs. Because EGF and TGF-α are known to bind to this common receptor, it is therefore not surprising to observe the same site of immunoreactivity. In mouse lung explants, EGF receptor was also identified in fibroblasts and in alveolar type II epithelial cells(14, 15). We can confirm these data which differ from previous reports(26, 27). The discrepancies may reflect differences in antibody specificity rather than different expression in culture or species, as it was propounded(27).

Inasmuch as EGF, TGF-α proteins, and their receptor are present in the airways at different degrees of growth and maturation, their role probably varies during development. EGF initially enhances cell multiplication in undifferentiated cells, but has little effect on fully differentiated cell proliferation(12, 22). In the organ culture model, EGF, unlike TGF-α, stimulates branching morphogenesis in a dose-dependent manner(14). These two growth factors seem to induce metalloproteinase activity(16), and their role in branching morphogenesis appears to depend on the balance between metalloproteinase and specific tissue inhibitors of metalloproteinase. In the absence of metalloproteinase inhibitor, TGF-α induces only cell proliferation with a dilatation of end buds and an enlargement of the lungs without any further branching(16). TGF-α was shown to promote angiogenesis(33), and this seems to take place in the modeling of the first blood-air barrier.

EGF and TGF-α are probably more important in late gestation, when they are synthesized at a high level. At that period, they induce cell maturation. EGF can accelerate differentiation of mucous secretory cells and stimulate the secretory product released into the conducting airway lumen(22). In the distal airways and in normal conditions, EGF enhances alveolar type II differentiation and surfactant synthesis(18), but also induces antioxidant enzyme maturation under hyperoxic conditions(21) and protects fetal rat lung from hyperoxic toxicity. Prenatal exposure to EGF stimulates biochemical and histologic maturation of the lung and markedly attenuates the clinical severity of respiratory disease in prematurely delivered rhesus monkeys(34). The presence of EGF family growth factor in ciliated cells is intriguing, and the role of EGF and TGF-α in these mature undividing cells is still to be determined.

EGF and TGF-α transcripts are found mainly in mesenchyme whereas the proteins are identified in a few mesenchymal cells and mostly in epithelial cells. This suggests that both EGF and TGF-α are mainly mediated by paracrine interactions between epithelial and mesenchymal cells in the fetal lung. In vitro, primary rat tracheal epithelial cells were shown to use TGF-α as an autocrine growth factor to proliferate(35). In vivo the mechanisms controlling lung cell proliferation are more complex and depend upon epithelial-mesenchymal interactions and many other growth factors. Any quantitative alteration of their production can induce abnormal development. We demonstrated the importance of the mesenchyme in the production of transcripts, especially under the pleura. In fetal hydrothorax, bilateral lung hypoplasia is probably not only the result of lung compression by pleural effusions, but may also be the consequence of a defect in EGF and TGF-α synthesis in the damaged mesenchyme. On the contrary, overexpression of TGF-α can induce fibrotic lesions and alveolar damages as shown in transgenic mice(36) whose lungs showed marked similarity to those of premature infants with bronchopulmonary dysplasia. The present finding of EGF and TGF-α expression in lung development of fetal humans further implies the important role for these peptides in the developing lung before and after birth. Introduction of EGF-like peptides into the treatment of very premature newborn infants may in the future help to prevent respiratory problems due to immaturity.