Introduction

Top Abstract Introduction Materials and Methods Results Discussion References

The extensive innervation seen in adult mammalian airways arises early in the pseudoglandular stage of lung development. In humans, pigs, and mice, a network of ganglia, nerve trunks, and small bundles follow the smooth muscle-covered epithelial tubules with varicose fibers descending into the smooth muscle layer (1). This close association, both spatially and temporally, of the growing nerve endings and the airway smooth muscle (ASM) was a feature observed in all three species and strongly suggested that trophic signaling between nerves and ASM was taking place. To investigate which trophic factors regulate nerve/target organ interaction in the developing lung, an in vitro model was needed, where growth and distribution of neural tissue and its principal target, the ASM, simulated the situation in vivo.

The in vitro model chosen for this purpose was the mouse lung explant, which has been extensively used in experiments on epithelial-mesenchymal interactions and branching morphogenesis (4). Neuroblasts have occasionally been observed in tissue sections of cultured lungs (7, 8) but were quite unrepresentative of the abundant innervation that we have previously shown to be present in the developing lung in vivo by using immunohistochemistry and confocal laser scanning microscopy (1). In the embryonic mouse, neural crest-derived cells (NCC) that will differentiate into neurons and glia of the enteric nervous system (ENS) are already present in the foregut (9) at the time when the lung buds arise at embryonic Day 9.5 (10), and it is assumed that some of these cells will migrate into the lung as it is forming (11). Recently we showed that neuronal precursors, positive for p75^NTR, a marker for enteric NCC (9, 12), are present in the lung at E11 and are at this stage also migrating from the vagus to the trachea and primary bronchi. At E12, nerve trunks positive both for p75^NTR and the pan-neuronal marker, protein gene product 9.5 (PGP 9.5) (13), mainly follow the smooth muscle- covered tubules to the base of the epithelial buds, and by E13 some nerves extend out into the mesenchymal cap beyond the buds. At E13 and E14, PGP 9.5-positive ganglia innervated by the vagus and interconnected by nerve trunks cover the future trachea and are present at branch points of the primary bronchi (2).

No studies have hitherto been undertaken that investigate the migration, guidance, differentiation, and survival of neuronal precursors and nerves in the developing lung. In contrast, the development of the ENS has been extensively studied. Both isolated NCC (12) and NCC in explants of vagus nerves and bowel (14) have been shown to migrate and develop into neurons in vitro. Enteric NCC of vagal crest origin colonize the entire gastrointestinal tract (15) and it is therefore possible that they will also migrate with the lung buds as they are forming or from the vagus soon after (2). These cells are dependent on glial derived neurotrophic factor (GDNF), which has been identified as the most important neurotrophic factor in the development of the ENS, because mice lacking GDNF or receptor for GDNF (RET) lack all neurons below the esophagus and proximal stomach (16). GDNF applied to neural crest cells isolated from the gut induces proliferation and differentiation (12, 20, 21), and it also has a chemotactic influence on neural crest cells in cultured gut explants (22). The mRNA of GDNF and its receptors are present in the mouse lung (23) at the time when neuronal precursors are migrating and differentiating along the smooth muscle- covered tubules. It is therefore possible that GDNF may be involved in guidance of pulmonary neuronal precursors.

The aim of this work was to characterize the spatial and temporal distribution of neural tissue and its primary target tissue, ASM, in cultured lung explants and further to evaluate whether this in vitro model was suitable for investigating trophic signals that affect migration and differentiation of pulmonary neuronal precursors. To this end, we first compared the growth rate of explants in culture with the lung in vivo to select an appropriate time to commence culture. Neural tissue and smooth muscle was imaged in the whole left lobe by confocal microscopy after immunohistochemical staining. We further used GDNF-supplemented medium and GDNF-impregnated beads to show that this in vitro system can be used to study the behavior of pulmonary neuroblasts in response to neurotrophic factors.

	Materials and Methods

Top Abstract Introduction Materials and Methods Results Discussion References

Organ Culture

Timed-pregnant mice (outbred Quackenbush) at 11, 12, and 13 d of gestation were obtained from the Animal Resources Centre (Canning Vale, WA, Australia). Mice had been mated overnight and the morning of finding the vaginal plug was considered as embryonic Day 0. The mice were killed by inhalation of a 70% CO₂/30% O₂ gas mixture for 30 s, followed by cervical dislocation. Embryos were dissected from the uteri, placed in sterile Hanks' balanced salt solution (HBSS) on ice, and decapitated. The primordial lungs were excised in HBSS using sterile microdissection tools. Left lobes of mice at embryonic day (E) 12 were used unless otherwise specified. Falcon 12-well culture inserts with transparent polyethylene terephthalate track-etched membranes, 4 µm pore size (Becton Dickinson Labware, Franklin Lakes, NJ), were used. One or two lobes were placed on each filter. Each well contained 1 ml of medium, and inside the well only one drop of medium to keep the lobe moist but still exposed to the air. The culture medium consisted of DMEM/F12 with 0.048 mg/ml penicillin, 0.67 mg/ml streptomycin, and 0.205 mg/ml amphoterizin B (Fungizone). Fetal bovine serum (FBS), 10%, was also added unless otherwise specified. The medium was changed after 3 d in culture. All culture reagents were obtained from GIBCO (Life Technologies Pty Ltd., Mulgrave, VIC, Australia). Lobes were cultured for 2-5 d in a humidified, 5% CO₂ environment at 37°C. Video micrographs of the left lobes were captured at the start and at each day of culture using a Sony CCD-Iris/RGB color video camera, connected to an inverted microscope (Olympus SZH10, Olympus Optical Co., Ltd., Tokyo, Japan) and digitalized using the software Apple Video Player on a PowerMac. Digital images of left lung lobes were analyzed using PhotoShop 5; the areas of the left lobes were measured and the peripheral buds counted. The number of cultured lobes measured were 5 at E11, 27 at E12 (four separate experiments) and 10 at E13. The number of in vivo lobes were as follows: E11: n = 19, E12: n = 57, E13: n = 30, E14: n = 9, E15: n = 9. The area of the cultured lobes were plotted as a function of time and compared with the equivalent day in vivo (i.e., E12 lobes cultured for 2 d [E12+2] were compared with E14 in vivo lobes). Means and standard errors of the mean (SEM) are shown. In vivo, the number of peripheral buds were counted after flattening the lobe with a coverslip. This could not be done in vitro, so buds could only be counted at the early days of culture. In E11+5 (n = 3) and E12+5 (n = 3) lobes, buds were counted after immunohistochemical processing with PGP 9.5 which stains the undifferentiated epithelial tubules and buds.

To determine the effect of GDNF on nerve growth, GDNF was added to the medium (100 ng/ml), or GDNF-impregnated agarose beads (Cibacron, 80-110 µm diameter; Sigma Chemical Co., St. Louis, MO) were placed on the filters. Beads were placed near the proximal part of the lobe (where the lobar bronchus had been severed). The positioning of beads in exact spots was difficult but they were placed so they touched the lobe or were less than 500 µm away. Beads would fall off unless they had been incorporated within the tissue of the explants, i.e., epithelial buds, spreading mesenchymal cells, or neural tissue. Due to the risk of dislodgment during immunohistochemical processing, several beads were used with each explant. The agarose beads were washed in sterile HBSS. 2 µl of GDNF (100 µg/ml) was added to 2 µl of bead suspension and incubated for 1 h at 4°C (22). Control beads were treated the same way but incubated in HBSS without addition of GDNF.

Immunohistochemistry

E12 lobes were removed after 2 or 5 d in culture and treated as whole mounts while still attached to the filters. The tissue was fixed in the culture inserts overnight (ON) in freshly prepared 4% paraformaldehyde. The transparent filters were removed from their casings and the tissue was cleared in dimethyl sulfoxide (DMSO; Sigma) for 3 × 10 min, washed in phosphate-buffered saline (PBS), and blocked in PBS with 1% bovine serum albumin (PBS-BSA) (Sigma) for 30 min. Antibodies (ab) were diluted in PBS-BSA. The tissue was incubated in the primary ab ON at 4°C, washed in PBS, and then incubated in secondary ab for 4 h at room temperature or ON at 4°C. After further washing in PBS, the filters with the attached tissue were then mounted in glycerol (90%) with antifade (0.1% p-phenylenediamine; Sigma) between two coverslips. All washing steps were performed with the tissue shaking in PBS for at least 3 h, with several changes of buffer. Neural tissue was detected by using polyclonal rabbit antibodies to PGP 9.5 (1/200) (13, 24) (Ultraclone, Isle of Wight, UK) and synapsin (1/200) (Calbiochem-Novabiochem Corporation, San Diego, CA) (25) or monoclonal rat-anti-p75^NTR (1/50) (Chemicon International Inc., Temecula, CA). Smooth muscle was recognized by a monoclonal mouse antibody to -actin (1/2,000) (Sigma) (26). Secondary antibodies used to visualize the labeled tissue were anti-rabbit or anti-mouse, conjugated to CY3 (1/40) (Zymed Laboratories, San Francisco, CA) or Oregon green 488 (Og 488, 1/200), or an anti-rat conjugated to Alexa568 (1/200) (Molecular Probes, Eugene, OR). As a control the primary antibody was omitted with no staining above background as a result. The number of E12 lobes that were stained was at least five in four separate experiments.

Confocal Microscopy and Analysis

Fluorescent images of nerves and smooth muscle were obtained using a confocal laser scanning microscope (MRC 1000; Bio-Rad, Hemel Hempstead, UK) with COMOS software (version 7.0; Bio-Rad). The fluorescent markers were detected by a krypton/ argon laser with the excitation wavelengths of 488 nm for Og 488 and 568 nm for CY3 and Alexa568. The whole mounts were optically sectioned by scanning at increasing depths of focus (in steps of 1 to 10 µm depending on the magnification used). The stacks of images obtained were projected into two-dimensional (2-D) images with the aid of Confocal Assistant, a software program that uses the maximum intensity of the corresponding pixels in each optical section to produce a 2-D image. In some cases, only a partial projection (i.e., a selected sequence of optical sections) was made to show more clearly the structures of interest. With larger whole mounts, several fields of view were captured and made into a montage representing the complete whole mount. After double staining, the green and red images were captured separately, colorized, and merged to show a composite nerve/ smooth muscle image or to show the overlapping pattern of PGP 9.5 and p75^NTR. Image processing (montaging, colorizing, merging, and analysis) was performed using Adobe PhotoShop 5.0 software. To quantify the area covered by neural tissue, images of p75^NTR-immunoreactivity were captured at 5 µm steps, using the same laser power, iris, and gain for control and GDNF-treated lobes. The z-series were projected and montaged for each lobe to include the entire amount of p75^NTR-positive tissue. The lobe was blotted out (so only the area outside the lobe would be measured), followed by threshholding of each image and subsequent measuring of the p75^NTR-positive area using Adobe PhotoShop 5.0.

	Results

Top Abstract Introduction Materials and Methods Results Discussion References

Growth of Left Lung Lobes in Culture

To characterize the growth of the in vitro system and to compare cultured lung explants with those grown in vivo, left lobes at E11, E12, and E13 were cultured and monitored for growth (area and number of peripheral buds, Figure 1A) by video microscopy at the start and at each day of culture. Lobes were grown at the air/liquid interface on membranes in culture inserts and therefore grew relatively flat. Figure 1B shows the lobe area at each embryonic day in vivo as well as growth curves for explants started at E11 (5 d in culture: E11+5), at E12 (5 d in culture: E12+5), and at E13 (4 d in culture: E13+4). The area of lobes started at E11 increased very slowly, whereas the areas of E12 and E13 lobes continued to increase steadily, though not as rapidly as in vivo. The number of epithelial buds around the periphery (Figure 1A), a measure of structural development (2, 5), was counted for the first days of culture and it was found that E12 lobes had an average of 11 ± 1.5 buds at the start, which increased to 23 ± 1.7 after 1 d, and 44 ± 3.4 after 2 d in culture (Figure 1B, numbers of buds next to symbols). These numbers of peripheral buds are the same as seen in lobes at E13 (23 ± 0.4) and E14 (44 ± 0.8), respectively, showing that the epithelial tubules of E12 lobes have branched in culture at a similar rate to that seen in vivo. In contrast, lobes started at E11 showed only a modest increase in the number of epithelial buds. The peripheral buds could only be reliably counted under the light microscope in the early stages of lung culture, because subsequently definition was lost due to the great increase in the number of buds. It was, however, possible to determine the number of buds at the end of culture by using an antibody to PGP 9.5 which stains the undifferentiated epithelium. Using this method, we found that the number of peripheral buds displayed by E11+5 lobes was 27 ± 4.1 (Figure 1, E11+5), whereas E12+5 lobes displayed 93 ± 5.7 (Figure 1, E12+5). We wished to establish an in vitro system with lobes of a young age, but because the cultured lobes at E11 grew poorly we chose to use E12 lobes throughout the study.

View larger version (44K): [in this window] [in a new window]   Figure 1.   (A) Video micrograph of a left lobe started at embryonic day (E) 12 and grown in culture for 2 d showing the lobe area (dashed line) and the epithelial buds around the periphery (dots). (B) Growth curves, expressed as area, for left lobes of fetal mouse lungs. The numbers next to the symbols show the number of peripheral buds at each stage. Explants in vitro are compared with the corresponding stage in vivo. In vivo: area of left lobes at E11-15. E11+5: cultures started at E11 and grown for 5 d (n = 5). E12+5: cultures started at E12 and grown for 5 d (n = 27). E13+4: cultures started at E13 and grown for 4 d (n = 10). The positions of the symbols have been slightly shifted sideways to prevent them from obscuring each other. The values are presented as means ± SEM. The areas of E12 and E13 explants increased steadily during the culture period, though not as fast as in vivo, and they also branched at a similar rate to that seen in vivo during the first 2 or 1 d, respectively.

Neural Tissue Survives and Continues to Grow In Vitro

The left lobes of fetal mouse lungs at E12 were cultured for 2 or 5 d, followed by immunohistochemical processing for neural tissue with antibodies to PGP 9.5 and synapsin. Anti-PGP 9.5 is a pan-neuronal marker (2, 3, 13) that also stains the undifferentiated epithelium of the tubules (3, 24), making it possible to visualize nerves and tubules concurrently. At the start of culture, E12 lobes displayed ~ 12 peripheral buds. A few PGP 9.5-positive ganglia were present in the proximal part of the lobe and nerve trunks followed the lobar bronchus and secondary branches (Figure 2A). After 2 d in culture the area of the lobes had more than doubled, further branching had taken place, and the number of PGP 9.5-positive buds around the periphery had increased greatly (Figure 2B). Ganglia and nerves were especially prominent in the proximal part of the lobe and PGP 9.5-positive neural tissue extended along the lobar bronchus, but could not be detected on the more distal branches.

View larger version (152K): [in this window] [in a new window]   Figure 2.   (A) Confocal projection of a left lung lobe at E12 in vivo, a typical lobe at the start of culture. PGP 9.5 immunoreactivity reveals nerve trunks along the lobar bronchus (br) and smaller bundles along the laterals. A few ganglia are present in the proximal part of the lobe (arrows). The epithelial buds are also positive for PGP 9.5. (B) Confocal projection of an E12 lobe cultured for 2 d. The lobe is more than twice the size compared with a lobe at the start of culture. PGP 9.5 immunoreactivity shows the undifferentiated epithelial tubules and buds as well as nerves and ganglia (arrows) along the lobar bronchus (br). Scale bars: 250 µm.

After 5 d in culture, the areas of the lobes had increased ~ 4-fold. Staining for PGP 9.5 revealed that extensive branching had occurred with numerous epithelial buds around the periphery of the lobes (Figure 3A). Further innervation had taken place, especially prominent in the lobar bronchus and secondary branches. Only a few fine nerves were observed to extend to the edge of the lobes (Figure 3A, inset). Large nerve trunks ran along the length of the bronchus. Ganglia, which were not seen at the start of culture (Figure 2A), were situated along the lobar bronchus and secondary branches (Figure 3A). They varied in size and shape, with some containing only a few neurons and others consisting of groups of loosely packed cells (Figure 3B). Large ganglia with shapes varying from elongated (Figure 3C) to rounded and compact (Figure 3D) contained more than 100 cells. Single optical sections through ganglia revealed cell bodies with PGP 9.5-positive cytoplasm and nuclei, whereas the nucleoli remained unstained (Figure 3D'). Occasionally only the cytoplasmic rim was PGP 9.5- positive (not shown). An antibody to synapsin was also used to stain for neural tissue. Whereas PGP 9.5 was present in both cell bodies and fibers (Figure 3E), synapsin was present only in the fibers (Figure 3E').

View larger version (155K): [in this window] [in a new window]   Figure 3.   Left lung lobes at E12 cultured for 5 d. (A) A montage of confocal projections showing immunoreactivity to PGP 9.5. Observe the many epithelial buds around the periphery of the lung. Ganglia and nerve fibers are situated along the lobar bronchus and branches. Inset: Nerve originating from a small ganglion, extending to the periphery. (B) Loose cluster of PGP 9.5-positive cells in a lobar bronchus. (C) Elongated ganglion containing many PGP 9.5- positive cell bodies. (D) Round compact ganglion. (D') Single section through the ganglion in D revealing PGP 9.5 immunoreactivity in the cytoplasm and nucleus. The nucleoli are unstained. (E) Ganglion stained for PGP 9.5, revealing nerves and cell bodies. (E') Ganglion in E stained for synapsin, showing only nerve fibers. Scale bars: 200 µm in A and 25 µm in B-E.

ASM Is Maintained and Continues to Grow in Culture

An antibody to -actin was used to detect smooth muscle; Figure 4A shows a lobe cultured for 2 d without FBS. The -actin-positive smooth muscle covers the tubules to the base of the epithelial buds. The buds appear as holes in the smooth muscle but can be seen protruding in Figure 4B, which shows staining of the undifferentiated epithelium by the antibody to PGP 9.5. Lobes cultured without FBS for more than 2 d displayed a morphology of dilated tubules and did not exhibit narrowing of the tubules, which is considered to be important for lung development (27). Therefore all lobes cultured for 5 d were grown in media supplemented with 10% FBS. Smooth muscle was maintained after 5 d in culture. Figures 5A and 5B show the lobar bronchi of two different lobes. Double staining using antibodies to PGP 9.5 (green) and -actin (red) show that nerves and ganglia, interconnected by nerve trunks, overlie the smooth muscle-covered tubules. The morphology of smooth muscle was similar to that seen in vivo (2) with bundles encircling the tubules.

View larger version (127K): [in this window] [in a new window]   Figure 4.   Confocal projection of lobe at E12 after 2 d in culture grown without fetal bovine serum. (A) Immunoreactivity to -actin reveals smooth muscle encircling the epithelial tubules to the base of the growing buds. The buds are unstained, and their whereabouts are only apparent as holes in the smooth muscle. (B) PGP 9.5 immunoreactivity reveals the presence of epithelial tubules under the smooth muscle, and the protruding buds. PGP 9.5-positive ganglia (arrows) are present in the proximal part of the lobar bronchus (br).

View larger version (113K): [in this window] [in a new window]   Figure 5.   Confocal projections of left lobes at E12 cultured for 5 d. (A) A left lobar bronchus. The antibody to -actin (red) shows smooth muscle bundles with a cylindrical orientation. PGP 9.5 immunoreactivity (green) is present in ganglia and nerves along the length of the smooth muscle-covered tubule. (B) Large, PGP 9.5-positive ganglia (green) at the junction between the smooth muscle- covered lobar bronchus (red) and one of the secondary branches. (C) Elongated, -actin-positive cells with filamentous projections outside a cultured lobe at the edge of the cell layer. (D) Cells positive for -actin (red) outside the proximal part of the lobe. PGP 9.5- positive nerves (green) grew into the layer of -actin-positive cells from the area of the severed lobar bronchus. Few small ganglia (2-3 neurons) or individual neurons (arrows) were present outside the lobe. The epithelium of the lobar bronchus (lobe) is PGP 9.5-positive (green). The -actin staining of the cells outside the lobe was so strong that the laser power had to be turned down; therefore, the more weakly stained smooth muscle of the tubule does not show in this overview. E-I show lobes cultured for 5 d in GDNF-supplemented media. (E) PGP 9.5-positive nerves (green) reach all the way to the edge of the layer of -actin-positive cells (red) but does not grow outside the cells and onto the filter. (F) Increased amount of PGP 9.5-positive (green) nerves and many large ganglia (arrows) cover the -actin-positive cells (red) outside the proximal part of the lobe in response to GDNF. (G) Increased amount of neural tissue in response to GDNF. Lobe surrounded by p75NTR-positive nerves and migrating neuronal precursors that have formed ganglia. (H) Enlargement of area in G showing immunoreactivity to p75NTR (H, red), PGP 9.5 (H', green), and the two images merged (H''). Yellow represents colocalization, but the true relationship between the two markers can only be discerned at higher power. (I, I', and I'') Single optical section through a ganglion in H (boxed area) double stained for p75NTR and PGP 9.5, showing p75NTR-positive membranes (I, red) around PGP 9.5-positive cytoplasm and nucleus (I', green). The nucleoli are unstained. (I'') The merged images shows the relationship of the markers. A yellow color means that they are colocalized in that particular area which especially occurs in the nerves (arrows). Scale bars: 100 µm in A, B, C, E, and H; 200 µm in D, F, and G; 25 µm in I.

Nerves Grow on -Actin-Positive Cells Outside the Explants

Mesenchymal cells, outside the epithelial buds, migrated out from the lobe and formed a wide layer of cells around the periphery, when cultured in medium with 10% FBS. These cells were elongated with filamentous projections and stained strongly for -actin (Figure 5C), suggesting that these mesenchymal cells may have differentiated into smooth muscle precursors. PGP 9.5-positive nerves grew from ganglia on the proximal part of the lobar bronchus (where the bronchus, and hence the nerve fibers, had been cut from the lung), out into the layer of -actin-positive cells (Figure 5D, overview). Some PGP 9.5-positive cells were positioned outside the lobe but the neural tissue consisted mainly of dispersed nerve bundles. When FBS was omitted from the medium, the edges of the lobes were more clearly defined, with little or no growth of -actin- positive cells outside the perimeter. Nerves did not grow outside the edges of the lobes unless the -actin-positive cells were present, suggesting that these cells constitute a favorable substrate for nerve growth and also that they may express a signal which attracts nerves and neuronal precursors.

GDNF Increases Nerve Growth and Acts as a Chemoattractant

To investigate the influence of GDNF on neuronal precursors of the fetal lung, E12 lobes were cultured for 5 d in GDNF-supplemented medium, followed by immunohistochemical detection of neural tissue and smooth muscle. The amount of neural tissue was greatly increased in response to GDNF. PGP 9.5-positive nerves reached all the way to the edge of the -actin-positive cells (Figure 5E), whereas the majority of neural tissue grew most densely near the proximal part (Figure 5F). The neural tissue consisted of a high density of nerve trunks as well as many PGP 9.5-positive cells formed into large ganglia, indicating that migration, proliferation, and differentiation of neuronal precursors, as well as neurite extension, had taken place. To stain for neural tissue, we used both an antibody to PGP 9.5, a pan-neuronal marker, and an antibody to p75^NTR, the low-affinity neurotrophin receptor, which has been shown to stain nerves and neuronal precursors of the gut (9) and lung (2). Figure 5G shows a lobe surrounded by p75^NTR-positive neural tissue after exposure to GDNF. Double staining with antibodies to both p75^NTR (Figure 5H) and PGP 9.5 (Figure 5H') revealed that both antibodies labeled similar tissues even though their distribution is not identical (Figure 5H''). Due to the density of nerves and large ganglia outside the lobes, it is difficult to make out individual nerves and neurons in projected images. Single optical sections through the tissue at higher magnification revealed that p75^NTR-positive membranes (Figure 5I) enveloped PGP 9.5-positive cell bodies (cytoplasm and nucleus) (Figure 5I'). It was evident that nerves were both p75^NTR- and PGP 9.5-positive (yellow, Figure 5I''), although some cells contained p75^NTR-positive membranes without PGP 9.5 staining and vice versa. To determine the effect of GDNF on neural tissue, the areas outside the lobes, covered in p75^NTR-positive neural tissue, were quantified, using six lobes cultured in GDNF-supplemented media and seven controls cultured without GDNF (Figure 6). It was found that there was a significant, 14-fold increase in the area that was p75^NTR immunoreactive after explants had been cultured in GDNF (unpaired t test, P < 0.01).

View larger version (15K): [in this window] [in a new window]   Figure 6.   Effects of GDNF on the size of the area covered by p75-positive nerves and neuronal precursors outside the explants. Data are from six explants cultured in GDNF and seven controls, expressed as means ± SEM. There was a significant increase in p75-positive staining in GDNF-treated cultures (unpaired t test, P < 0.01).

To see whether GDNF acts as a chemoattractant, a few agarose beads, impregnated with GDNF, were placed next to each lung lobe (E12, proximal end) after 1 d in culture (Figure 7A). After another 4 d, the lobes were removed and immunohistochemically processed for PGP 9.5 immunoreactivity. If beads were placed so they touched the lobe, the increased nerve growth was concentrated in the area around the beads (Figure 7B). When beads were placed further away (~ 200-500 µm), large nerve trunks were observed to reach out toward the beads and surround them with nerve fibers (Figure 7C). Occasionally, a bead would be surrounded by PGP 9.5-positive cell bodies (Figure 7D), indicating that migration of neuronal precursors toward the bead and/or proliferation around it had taken place. In all cultures (n = 5) with GDNF-impregnated beads, the beads were surrounded by nerves and/or PGP 9.5-positive cell bodies, whereas control beads without GDNF showed no apparent attraction to nerves (Figure 7E, n = 3).

View larger version (117K): [in this window] [in a new window]   Figure 7.   (A) Videomicrograph of E12 left lobe cultured for one day. Beads (arrow) were placed near the proximal part of the lobar bronchus. (B) Boxed area in A showing the same lobe after 5 d in culture. Confocal projection showing that PGP 9.5-positive nerves are concentrated to the area around the beads (b). The severed lobar bronchus (br) is also surrounded by nerves. (C) Beads (b) were placed a little further away from the proximal lobar bronchus. A large PGP 9.5-positive nerve trunk extended out toward the beads and encircled them after 5 d in culture. (D) A bead (dotted line) engulfed in a ganglion made up of many cell bodies positive for PGP 9.5. (E) Control beads (b) incubated in media without GDNF did not attract nerves or neurons. Scale bars: 500 µm in A and 100 µm in B, C, D, and E.

	Discussion

Top Abstract Introduction Materials and Methods Results Discussion References

We describe a culture of lung explants where neural tissue survives and continues to grow, and neuronal precursors migrate, mature, and extend neurites. Left lobes grew relatively flat on the surface of transparent membranes, which improved visualization of the separate tubules and hence the structure and distribution of nerves and smooth muscle. The lobes were immunohistochemically processed while still attached to the membranes, thus avoiding injury to the tissue. The explanted lung has been extensively utilized in studies on epithelial-mesenchymal interactions and branching morphogenesis (4), but in these studies the abundance of nerves and their distribution over the ASM covering the tubules has not been recognized. We started the cultures as early as possible, when the differentiation of neuronal precursors and the innervation of tissues was progressing. E11 is the earliest time when the lung can be separated from the foregut as an entity but these lobes grew slowly in culture. Lobes started at E12, however, displayed a steady increase in lobe area and branching of the epithelial tubules that was closely comparable to that in vivo.

The smooth muscle-covered epithelial tubules became innervated in a manner similar to that seen in vivo (2). Lobes at E12, the time at which the cultures were started, displayed few weakly staining PGP 9.5-positive ganglia in the proximal lobar bronchus. After 2 d, clusters of PGP 9.5-positive cells and nerves had appeared along the length of the lobar bronchus, and after 5 d neurons were also found on the secondary and tertiary branches. PGP 9.5 stains both the undifferentiated epithelium (2, 3, 24) and neurons (2, 3, 13). The different tissues, however, can easily be identified by their morphology and location; the epithelial cells lying under the smooth muscle in the tubular wall and the neurons, collected into distinct clusters, situated above the smooth muscle. The neurons also have p75^NTR-positive membranes and processes which the epithelial cells lack. PGP 9.5 was mostly present in both the nucleus and cytoplasm but was occasionally seen only in the cytoplasm. The reason for this is not known but has previously been observed in vivo in humans (1), mice (2), and pigs (3). The number of PGP 9.5-positive neurons had increased during culture but it was not possible to quantitatively evaluate whether this increase is similar to that seen in vivo. This is due to the increase in parenchymal thickness in vivo at the later stages of lung development (from E14), causing antibodies to get trapped in the mesenchyme, resulting in high background staining (2). Nevertheless, the evidence presented here, in conjunction with the information about the distribution of neurons in pig lungs during the canalicular stage (3), indicates that neural development is relatively similar in vivo and in vitro. It is unlikely that the migration of NCC into the left lobe is complete by the time we commence the cultures at E12, because migration into the lobes is most active at E11 (2) and E12, and is still ongoing at E13 (unpublished observations). However, all lung explants display a large number of PGP 9.5-positive neurons after 5 d in culture, probably due to proliferation of NCC (12) and their further migration within the lobe.

Neurons and their axons are predominantly found associated with the smooth muscle-covered tubules during normal lung development in vivo, indicating that smooth muscle and smooth muscle precursors may secrete neurotrophic factors or provide nerves with a favorable substrate for growth. Furthermore, large nerve trunks run along the length of the tubules, giving off fine varicose fibers to the smooth muscle (1). Here we show that this is also seen in vitro with nerves mainly following the smooth muscle- covered tubules, with only few nerves coursing through the mesenchyme to the periphery. Further evidence of the attraction of nerves to smooth muscle is that nerves and neurons did not only follow the tubules but also grew out into a wide layer of -actin-positive cells that had appeared outside the perimeter of the cultured lobes. The migration and differentiation of these cells were aided by FBS in the culture medium because lungs cultured without it contained few or no such cells. The -actin-positive cells had an elongated shape with filamentous projections, indicating that they may be smooth muscle precursors. Undifferentiated mesenchymal cells, isolated from embryonic lungs and given the opportunity to spread out and elongate, have been shown to express many smooth muscle markers; -actin, smooth muscle myosin, desmin, and SM22 (28). They also develop membrane potentials of -60 mV and voltage-dependent Ca²⁺ currents (29), all indicative of smooth muscle. Neurons and their axons grew into the layer of -actin-positive cells, but not outside it onto the filter. This suggests that these cells provide a good substrate for nerve growth and also that they may express a neurotrophic factor or factors that attract neuronal precursors and promote neurite extension. The expression of neurotrophic factors in smooth muscle has been reported in cultured rat vascular smooth muscle cells, which express nerve growth factor (NGF), neurotrophin 3 (NT-3), and brain-derived neurotrophic factor (BDNF) mRNA (30). GDNF mRNA has been found in the wall of the fetal and adult gut, possibly in the smooth muscle (31). Neurturin (NRTN, a member of the GDNF-family) mRNA is expressed by the smooth muscle of penile blood vessels and corpus cavernosum in adult rat (34) and in the circular smooth muscle layer of the fetal intestine (35). A few of these factors have also been found in the developing lung; NRTN mRNA is present in the airway smooth muscle of the mouse at E17 (earliest age tested, 35), whereas GDNF mRNA is present in the lung as early as E10 and found by in situ hybridization to be expressed by the mesenchyme closely associated with the epithelial tubules at E13 (23). The latter study did not attempt to recognize smooth muscle cells, so it is possible that the GDNF was actually expressed by the airway smooth muscle or mesenchymal cells closely associated with it.

GDNF and/or other ligands of the same family are likely to be important in guidance, differentiation, and survival of pulmonary neurons. In this study we used an in vitro model to show that GDNF induces a striking increase in the amount of pulmonary neural tissue positive for p75^NTR and PGP 9.5. This suggests that neuronal precursors of the lung respond in a very similar way to those of the gut, because GDNF has been shown to induce proliferation and differentiation of isolated enteric neural crest cells (12, 20, 21). We further show that GDNF is a chemoattractant to pulmonary neuronal precursors and influences the direction of neurite extension. This has previously been shown in the gut where GDNF attracts enteric NCC and their axons (22). GDNF signals through a receptor complex of RET and GFR1. Both these receptors are present in enteric neurons (12, 22) and their mRNA is present in the developing lung (23), although no studies have specifically shown their location in pulmonary neurons. Neurons from lumbar dorsal root ganglia and sympathetic ganglia also express Ret and GFR1 but show no preference for outgrowth toward GDNF-impregnated beads (22). The similar behavior of neuronal precursors of the gut and the lung indicates that GDNF is an important neurotrophic factor in both these organs and provides evidence that the same population of NCC that colonize the gut also migrate into the lung.

This study shows that the development of neural tissue and smooth muscle of the fetal lung can be investigated in vitro. It is known that GDNF mRNA is present in the lung (23) at the time when neuronal precursors are migrating along the branching tubules and sending out axons to target tissues like smooth muscle (2), and here we show that GDNF did indeed exert a striking influence on pulmonary neural tissue in vitro. It is by no means the only factor that could guide innervation of the developing lung, but if information on the in vivo distribution of factors that may influence innervation is combined with experiments in vitro which confirms their neurotrophic role, we can learn both how and when the nervous system of the lung is established.