Fibroblast growth factor (Fgf) 10 is a critical regulator of bud formation during lung morphogenesis. fgf10 is expressed in distal lung mesenchyme at sites of prospective budding from the earliest developmental stages and signals through its epithelial receptor Fgfr2b. Experiments in intact lung organ cultures demonstrate that Fgf10 is a chemotactic factor for distal, but not for proximal, epithelium. This differential response suggests the involvement of an additional mechanism regulating Fgf10-Fgfr2b interactions, because Fgfr2b is uniformly expressed throughout the respiratory tract. Here we use an immunohistochemistry-based binding assay to show that O-sulfated heparan sulfates (HS) are critical for Fgf10 binding to the distal epithelium. We show that altering endogenous gradients of HS sulfation with sodium chlorate or over-O-sulfated synthetic heparin in lung organ cultures dramatically decreases Fgf10 binding. Moreover, we show that under these conditions epithelial binding is not improved by providing exogenous FGF10. Our data suggest that, not only ligand availability, but also the presence of specific patterns of HS modification in the distal lung epithelium are critical determinants of Fgf10 binding to the epithelium and signaling.
- fibroblast growth factor 10
- fibroblast growth factor receptor 2
- lung development
- branching morphogenesis
fibroblast growth factor 10 (Fgf10) and its receptor Fgfr2b are major regulators of cellular activities during organogenesis. Genetic inactivation of fgf10 or fgfr2b results in multiple organ abnormalities that include agenesis of limb, pituitary gland, and foregut derivatives such as the thyroid and the lung (20, 30). During lung development fgfr2b is expressed throughout the respiratory epithelium, while fgf10 is dynamically expressed in the distal mesenchyme near branching epithelial tubules. Fgf10 induces epithelial budding, and its spatial and temporal distribution in the mesenchyme appears to be critical for the establishment of the three-dimensional pattern of the branching tubules (4-6, 22, 34).
Heparan sulfates (HS) are abundant components of cell surfaces and extracellular matrices, where they are attached to core proteins to form HS proteoglycans (HSPG). Among their many functions, HS modulate growth factor distribution and receptor binding and signaling (1). During HS biosynthesis glycosaminoglycan (GAG) moieties are modified by a variety of enzymes to generate great diversity of HS chains, which influences the affinities of HS for growth factors and their receptors (reviewed in 21, 23). The biological relevance of these modifications is illustrated by the disruption of tracheal morphogenesis and reduction of Fgf-dependent MAPK activation in Drosophila mutants lacking the HS biosynthetic enzymes sulfateless, sugarless or 6-O-sulfotransferase (16, 19). Differentially sulfated HS have been described in basement membranes of the adult lung in association with type I and type II alveolar cells (28, 29). Alveolar type II cells of the adult rat express the HS-synthesizing enzymes 3-O-sulfotransferase and N-deacetylase/N-sulfotransferase (NDST) (18).
Moreover, genetic inactivation of NDST-1 in mice results in pulmonary hypoplasia, disruption of type II cell differentiation, and neonatal death (9, 25). No changes in branching, however, have been described.
We have recently shown that Fgf10-mediated bud formation and gene expression in lung epithelial cultures depend on the presence of HS (15). The mechanism by which HS influence these responses is not clear. Here we show that regulation of Fgf10 binding to the epithelium is a major component of this mechanism. We performed binding assays in situ using an anti-Fgf10 antibody in histological sections of lungs at different developmental stages and in cultures in which we altered the patterns of HS sulfation. We present data suggesting that O-sulfated HS expressed on the surface of the distal epithelium may control Fgf10 binding and, consequently, its biological activity.
MATERIALS AND METHODS
Whole lung cultures. Embryonic day 11.5 (E11.5) mouse lungs were cultured on Millipore 0.8-μm filters on the top of metal meshes in 60-mm organ culture dishes (Fisher Scientific) in an atmosphere of 95% air and 5% CO2 for 72 h in BGJb media (GIBCO-BRL) containing 1% fetal calf serum (GIBCO-BRL) and 20 mg/ml vitamin C (Sigma) (control conditions). To disrupt endogenous HS we initially cultured lungs for 24 h in media containing 50 mU/ml bacterial heparitinase followed by 48 h of culture in media containing 25 mM sodium chlorate (NaChl). These concentrations have been previously shown not to be toxic (10). In some experiments heparin beads soaked in human recombinant FGF10 (100 ng/μl, R&D Systems) were grafted onto E11.5 lung after 24 h. Specimens were fixed in 4% paraformaldehyde at 4°C overnight and processed for paraffin embedding (immunohistochemistry or isotopic in situ hybridization) or stored at -20°C in 100% methanol (whole mount procedures).
Modified heparins. We used selectively sulfated heparins in binding assays in the epithelial cell line and in lung organ cultures (below). Selectively modified heparins (over-O-sulfated and de-O-sulfated) from Neoparin (San Leandro, CA) were dissolved in BGJb medium and added to E11.5 lung explants (25-125 μg/ml) for 48 h. In de-O-sulfated heparin all O-sulfated groups in glucosamine and uronic acid residues were chemically removed, but N-sulfated groups of glucosamine residues remained intact. In over-O-sulfated heparin all hydroxyl groups of disaccharide units were substituted by sulfate groups. Heparins have a molecular mass of ∼13 kDa.
Antibodies. To identify sites of Fgf10 bound to the lung epithelium we used the goat polyclonal antibody C17 (1:400) from Santa Cruz Biotechnology in immunohistochemical assays. This antibody has been previously shown to specifically block Fgf10 effects in adipocyte differentiation and in tooth organogenesis (12, 27). C17 recognizes an epitope near the carboxy terminus of Fgf10 that is identical for mouse and human proteins. Specificity of the antibody was confirmed by Western blot analysis of human recombinant FGF10 (R&D Systems), NIH3T3 fibroblasts, and E12 and E14 lung homogenates (see results).
We also assessed C17 specificity by performing immunohistochemistry with antibody that had been precleared with immobilized FGFs and blocking peptide. In brief, Affigel-10 beads (Bio-Rad) were coupled with human recombinant FGF2, FGF7, FGF10, or antibody-specific blocking peptide (Santa Cruz Biotechnology) according to the manufacturer's protocol. C17 antibody was added to the beads to a final volume of 100 μl in microtubes and incubated at room temperature for 2 h. Samples were centrifuged briefly, and supernatants were applied to E14 lung histological sections. Immunohistochemistry was performed as described in Immunohistochemical analysis. For positive control we used supernatants in which C17 antibody was incubated with Affigel beads preblocked with ethanolamine.
HS distribution was visualized using the anti-HS mouse monoclonal antibody 3G10 (“anti-stub,” 1:500) obtained from Seikagaku. 3G10 identifies a neoepitope generated by heparinase III digestion of HS (7). 3G10 immunostaining was performed in paraffin sections and in whole mount specimens previously fixed in paraformadehyde and pretreated with heparinase III (400 mU/ml for 2 h) immediately before immunohistochemistry.
Immunohistochemical analysis. For analysis of paraffin sections, anti-goat IgG and anti-mouse IgG kits from Vector Laboratories were used according to the manufacturer's protocol. Whole mount immunohistochemistry was performed as described elsewhere (13) with minor modifications. In brief, explants were rehydrated preblocked in PBS containing 2%(wt/vol) dry nonfat milk (PBSTM solution) and incubated overnight at 4°C in PBSTM with appropriate dilution of primary antibody, followed by extensive washing in PBSTM at room temperature. Biotinylated secondary antibody was added, and specimens were incubated in PBSTM overnight at 4°C. Antibody staining was visualized by diaminobenzidine peroxidase detection kit (Vector Laboratories).
In situ hybridization. We synthesized digoxigenin (DIG)-labeled RNA probes using Maxi Script kits from Ambion. Proteinase K, DIG, anti-DIG alkaline phosphatase conjugate, and BM Purple substrate solution were from Roche Diagnostics. Briefly, lung explants were rehydrated, digested with Proteinase K, prehybridized for 1 h at 70°C in buffer containing 50% formamide, 5× SSC, 1% SDS, 50 mg/ml yeast RNA, and 50 μg/ml heparin followed by overnight hybridization with DIG-labeled RNA probes. After blocking and incubation with anti-DIG alkaline phosphatase conjugate overnight at 4°C, signal was visualized with BM Purple substrate solution according the manufacturer's protocol. Isotopic in situ hybridization experiments were performed using 35S-labeled riboprobes as described in Ref. 6. In brief, sections were rehydrated, digested with Proteinase K, hybridized overnight at 50°C, and washed. Sections were dehydrated by being washed in increasing concentration of methanol in PBS, dried, dipped in photographic emulsion (Kodak NTB-2), incubated at 4°C for 3 wk, and developed.
Binding assay in 20-3 cells. Binding of 125I-FGF10 to confluent 20-3 rat embryonic distal lung epithelial cells (17) was conducted as previously described (10). In brief, confluent 20-3 monolayers in 24-well plates (2 cm2/well) were generated by plating 5 × 104 cells per well in 1 ml of DMEM containing 10% calf serum and penicillin (100 units/ml) and streptomycin (100 μg/ml). Binding assays were conducted 3 days after plating. Cells were washed once with cold binding buffer (DMEM, 0.05% gelatin, and 25 mM HEPES, pH 7.4) and then incubated for 10 min at 4°C in fresh binding buffer (0.5 ml/well) to inhibit receptor internalization. De-O-sulfated or over-O-sulfated heparin was added directly to the binding medium to generate the final indicated concentrations, and then 125I-FGF10, prepared using a modified Bolton-Hunter procedure (10), was added at a final concentration of 10 ng/ml. The cells were incubated for 2.5 h at 4°C to allow binding to reach steady state. At the end of the incubation period the cell layers were washed three times with cold binding buffer, and the 125I-FGF10 bound to HSPG sites on the surface of the 20-3 cells was extracted in high-salt buffer (2 M NaCl and 20 mM HEPES, pH 7.4) for 5-10 s. The amount on 125I-FGF10 in the high-salt extract was quantitated using a Cobra Auto-Gamma 5005 gamma-counter (Packard Instruments, Meridian, CT). The amount of 125I-FGF10 bound under each condition was normalized to that observed in the absence of the synthetic heparin (100%). The concentration of de-O-sulfated and over-O-sulfated heparin required to inhibit 125I-FGF10 binding by 50% (IC50) was estimated. Means and SE were determined from triplicate measurements. Differences between groups were analyzed by Student's t-test; changes were considered significant at P < 0.05.
RESULTS AND DISCUSSION
Specificity of the C17 antibody for fgf10 epitope. We used the anti-Fgf10 blocking antibody C17 in sections of embryonic lungs to map sites where Fgf10 binds to the epithelium. The ability of this antibody to specifically block Fgf10-mediated effects has been shown in other systems (12, 27). We further validated the use of C17 in our study by different approaches. First we performed Western blot analysis using C17 against various recombinant proteins and homogenates of cells or tissue known to express fgf10 mRNA (3, 4). Figure 1A shows that when incubated with increasing concentrations of human recombinant FGF10 (10, 50, and 100 ng/ml) C17 recognized a single FGF10 species of ∼22 kDa. This band is clearly identified in homogenates of NIH3T3 cells. The 22-kDa band is barely detectable in homogenate of E12 lungs (n = 10), but signals are enhanced when 10 ng of human recombinant FGF10 is added to the sample; C17 expression was further confirmed in E14 lung homogenate (Fig. 1B). C17 incubation with recombinant FGF1, -2, -7, or EGF (all from R&D Systems, at 50 ng/ml) revealed no cross-reactivity.
Immunohistochemistry of C17 in paraformaldehyde-fixed, paraffin-embedded tissue sections of E14.5 lungs showed signals in distal epithelial tubules, but not in proximal epithelium (Fig. 1C). Sites of C17 staining corresponded to sites where fgf10 mRNA has been described in the adjacent distal mesenchyme (4). C17 signals, however, were nearly undetectable in the mesenchyme. Thus under our experimental conditions the C17 antibody seemed to predominantly label epithelium-associated Fgf10. We ascribed the weak mesenchymal labeling to potential loss of ligand due to tissue fixation or processing, or antigen masking of secreted molecules by extracellular components (11). Preferential labeling of the epithelium, in turn, likely reflects accumulation of ligand or ligand-receptor complexes on the cell surface or in intracellular compartments. This is supported by observations of HS interactions with other FGF members. For example, HS have been shown to cause prolonged retention of FGF2 inside cells (32). Discrepancies between sites of immunolocalization and sites of mRNA expression have been reported in the lung for other secreted factors. In embryonic rat lungs FGF7 has been immunolocalized to the epithelium, whereas mRNA is expressed in the mesenchyme (24).
To further test whether C17 staining in E14 lung epithelial tubules was specific for Fgf10, we performed immunohistochemistry using antibody previously incubated with FGF2-, FGF7-, FGF10-, or blocking peptide-coupled beads (see materials and methods). As described above, we found strong C17 signals in distal epithelial tubules of control antibody (antibody alone or incubated with ethanolamine-preblocked beads) (Fig. 1C). Similarly strong signals were found in sections stained with antibody incubated with beads coupled with human FGF7 or FGF2, indicating that C17 was not retained by nonspecific binding (Fig. 1E and data not shown). By contrast, no signal was observed in sections treated with antibody incubated with beads coupled with human FGF10 or C17-specific blocking peptide (Fig. 1, D and F). These data indicate that the pattern of C17 immunostaining detected in lung epithelial tubules was specific for Fgf10.
Developmental expression of the C17 epitope. We determined the pattern of expression of C17 in sections of E11.5-E16 lungs. Figure 2 shows that at E12, E14, or E16 (Fig. 2, A, B, and C, respectively), when airways are actively branching, C17 signals are restricted to distal epithelial buds. We performed in situ hybridization of Fgf10 in whole mounts or in histological sections at equivalent developmental stages for comparisons. Results in Fig. 2 confirm that at all times C17 epithelial labeling is restricted to areas where fgf10 mRNA is expressed in the mesenchyme (see right panels in Fig. 2, A-C).
Ligand availability and the ability of FGF10 to bind to the proximal epithelium. We asked whether the proximal-distal pattern of C17 staining in the developing lung epithelium was solely determined by the differential distribution of Fgf10 with the highest levels in distal mesenchyme. To approach this question we generated a proximal gradient of FGF10 in E11.5 lung explants by implanting FGF10-soaked heparin beads near the trachea. Beads were similarly implanted near distal buds for comparison. Explants were cultured for 72 h in a control medium, and C17 expression was determined by immunostaining of histological sections. Figure 3A confirms that distal, but not proximal, epithelium responded to a localized source of exogenous FGF10 by engulfing the bead, as previously reported (22). C17-labeled epithelial buds in the periphery of the explant, including the epithelium that surrounded the FGF10 bead (Fig. 3, A-C). No significant difference in the intensity of staining was observed between buds near or far from the bead, suggesting that under these conditions excess ligand does not necessarily increase FGF10 binding to epithelium. Interestingly, no binding of FGF10 to proximal lung epithelium is found even in the presence of the local source of FGF10. Figure 3, D and E, shows that C17 immunostaining of the tracheal epithelium or main bronchus near the bead is undetectable. C17-negative epithelium did not move toward the FGF10 bead. Thus staining positively correlated with the ability of the epithelium to bind FGF10 and activate a response. Our data suggest that distal buds have a component that allows binding of FGF10 to the epithelium. This component is missing in proximal airways.
HS are present in proximal and distal airways at comparable levels. Among the potential reasons for the C17 expression pattern is a differential distribution of high- or low-affinity receptors for Fgf10 along the respiratory tract epithelium. FGF10 mediates its effects through binding and activation of Fgfr2b, an epithelial receptor highly expressed in the early lung. However, Fgfr2b shows no proximal-distal gradient (6). Because HS are critical for Fgf-Fgfr interactions, we examined the possibility that differences in levels or distribution of HS could be influencing Fgf10 binding. To identify sites of HS expression we stained E12-14 lung with the 3G10 antibody. 3G10 does not react with intact HS but recognizes a neoepitope generated by heparinase III-cleaved HS (7). Therefore, 3G10 serves as a marker of HS in the tissue without providing information about their structure. Whole mount immunohistochemistry of E12 lungs shows that 3G10 staining was ubiquitous (Fig. 4A). Staining of histological sections shows that, although present in epithelium and mesenchyme, labeling appeared to be prominent underlining basement membranes (Fig. 4, B and C). The absence of an obvious proximal-distal gradient of 3G10 signals suggests that HS chains are uniformly distributed in the lung. This pattern is similarly observed in cultured lungs. Figure 4, D and E, shows 3G10 signals in distal and proximal airways adjacent to FGF10 beads at comparable levels. We concluded that HS are expressed in proximal and distal lung at comparable levels. However, it is possible that HS have distinct structural characteristics which allows differential binding of FGF10 to the epithelium.
HS sulfation is required for FGF10 binding to the epithelium. Our data suggest that availability of ligand and HS are necessary but not sufficient for Fgf10 to bind to the lung epithelium. We reasoned that not only proper levels but also specific patterns of HS sulfation influence Fgf10 binding and downstream effects. We cultured E11.5 mouse lungs in medium containing heparinase III for 24 h to disrupt endogenous HS and subsequently with NaChl (25 mM) to prevent HS sulfation of newly formed HS. NaChl has been shown to preferentially block O-sulfation to a greater extent than N-sulfation (26). Similarly as in the previous experiments, FGF10-containing heparin beads were grafted onto these cultures at proximal and distal sites. Efficient disruption of HS sulfation was demonstrated by inhibition of branching (Fig. 5, A and B; Ref. 15). Interestingly, C17 immunostaining indicates that NaChl treatment nearly abolishes Fgf10 signals in the lung (Fig. 5B). 3G10 staining shows that HS expression is preserved. Strong HS signals are shown in basement membranes of epithelial tubules (Fig. 5C). A previous study demonstrates that endogenous fgf10 expression, as assessed by in situ hybridization, is preserved under NaChl treatment (data not shown, Ref. 15). Inhibition of Fgf10 signaling was indicated by disruption of stereotypical responses of the distal epithelium to the FGF10 bead. We found that signals were not increased in the epithelium associated with the FGF10 beads (Fig. 5, D-F). Overall the data are consistent with the idea that inhibition of HS sulfation by NaChl prevents FGF10 from binding to the distal epithelium.
O-sulfated groups are required to interfere with FGF10 binding. Experiments from the previous section provided initial evidence that sulfated HS are likely to interact with Fgf10 when complexes are assembled in the epithelium. To gain insights into the specific patterns of sulfation required for Fgf10 binding, we performed competition binding assays in epithelial 20-3 cells. 20-3 is a rat lung epithelial cell line that expresses both HS and Fgfr2b (Ref. 17 and data not shown). We compared the ability of N-sulfated and O-sulfated HS to compete with endogenous HS for Fgf10 binding sites in lung epithelial cells. Binding assays were performed with radiolabeled FGF10 in the presence of various concentrations of heparins modified to contain the maximal amount of O-sulfates (over-OSO4) or no O-sulfates with preserved N-sulfates (de-OSO4). Figure 6A shows that over-O-sulfated heparin competes most efficiently for FGF10 and inhibits FGF10 binding to the lung epithelial cells at much lower concentrations than de-OSO4 heparin does. Regular heparin has an intermediate ability to compete for FGF10, when analyzed under the same conditions (data not shown).
Next, we used lung organ cultures to determine whether fgf10 binding to the epithelium, as assessed by C17 staining, could be altered by these heparins. We found that although fgf10 signals were detected in distal tubules of control and in the presence of de-O-sulfated heparin, low or nearly absent labeling is seen in explants cultured in over-O-sulfated heparin at all concentrations (25-125 μg/ml; Fig. 6, B-G). The effect of over-O-sulfated heparin on fgf10 binding was similar to that seen in NaChl-treated cultures, although the effects on morphology and gene expression differed. At a similar concentration, the patterning effects of de-O-sulfated and over-O-sulfated heparins were remarkably different and reflected the ability of these heparins to compete with endogenous HS for fgf10. As shown in Fig. 6E, at 125 μg/ml, some changes in the distal morphology are seen in de-O-sulfated-treated explants; more dramatic changes in the pattern of branching that correlates with loss of C17 staining are found in over-O-sulfated-treated (125 μg/ml) explants. These alterations likely result from competitive binding and sequestration of HS-dependent molecules that regulate branching morphogenesis and are described elsewhere (15). We concluded that properly O-sulfated HS are critical for FGF10 binding and signaling.
Concluding remarks. In summary, we provide evidence that, besides availability of ligand and Fgfr2b expression, it is critical that O-sulfated HS are present in epithelial tubules to allow binding of Fgf10 to the lung epithelium. We show that, in lung organ cultures in which FGF10-soaked beads are implanted adjacent to proximal (trachea) or distal airways, Fgf10 labeling is detected only in the distal epithelium surrounding the bead. Under treatment with NaChl or oversulfated heparin, neither proximal nor distal epithelium is stained. The impaired response of the epithelium to exogenous FGF10 results from failure of Fgf10 to bind to and to form a stable complex with Fgfr2b even when Fgf10 is provided in excess. This simple model illustrates the critical role of HS in Fgf10-mediated lung morphogenesis. Together these data suggest that differences in the profile of HS sulfation throughout the respiratory epithelium may underlie the proximal-distal differences in the ability of the epithelium to bind to and respond to Fgf10.
A number of studies have shown that chondroitin sulfates (CS) are also abundant in the embryonic lung and are likely to regulate developmental processes (2, 33). Although we have not assessed the function of CS here, we have evidence that highly sulfated CS are diffusely expressed in the lung mesenchyme and do not seem to be associated with local gradients of FGF10 or sites of bud formation (15). Moreover, a recent study shows that FGF10 binds to CS with significantly lower affinity than to similarly sulfated heparin (8). These observations and our current data suggest that HS are most likely critical GAGs involved in FGF10 binding during lung bud formation.
Classic and recent grafting experiments in organ cultures show that the distal mesenchyme is able to induce budding and distal lung gene expression in proximal epithelium (31, 35). There is also evidence that the proximal mesenchyme acts as a physical barrier or produces factors that prevent tracheal or bronchial budding. The presence of an FGF10 bead adjacent to proximal epithelium of intact lung organ cultures is not able to elicit a budding response, as distally placed beads do (22). In contrast, experiments in mesenchyme-free epithelial cultures show that both proximal and distal epithelia are competent to respond to exogenous FGF10 (14, 34). This observation and the proximal-distal pattern of C17 staining reported here suggest that the proximal mesenchyme may selectively produce inhibitory factors that prevent FGF10 from binding to proximal epithelium. Alternatively, the proximal mesenchyme may have diffusible factors that interfere with HS O-sulfation and prevent proximal epithelial binding. We have recently demonstrated that the HS sulfotransferase HS 6OST1 is expressed in a proximal-distal gradient with high levels in distal epithelial tubules during branching (15). It is possible that this gradient is controlled by mesenchymal factors. Additional experiments are necessary to test this hypothesis.
This work was supported by National Heart, Lung, and Blood Institute Grant PO1 HL-47049.
We thank Mary Williams, Jerry Brody, and Jining Lu for thoughtful discussions. We thank Dr. Noboyuki Itoh for providing us with the FGF10 cDNA clone. We are grateful to Xiaoqing Qi and Yu Yang for excellent technical assistance.
The costs of publication of this article were defrayed in part by the payment of page charges. The article must therefore be hereby marked “advertisement” in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
- Copyright © 2003 the American Physiological Society