The lung develops under the control of FGF10-FGFR2IIIb signaling, just like the salivary and prostate glands, in which neural regulation of epithelia has been identified [6],[7]. A model of lung development was selected in which neural development is visible without the need for antibody staining. The microtubule-associated protein tau eGFP (MAPT) knockin mouse line has enhanced green fluorescent protein (eGFP) inserted in the tau protein locus. This labels all neuronal cell bodies and axons with eGFP [8]. Hence, neural structures can be seen and followed over time in lung explants.
We found that epithelial budding and neural arborisation proceed in tandem over time in culture (Figure 1; n >15) as they do in situ [9]. This arborisation arises from post-synaptic parasympathetic neurons whose cell bodies reside within the lung and which survive explantation. Indeed, they continue to elaborate nerve axons along the epithelial tubes as airway branching proceeds. In contrast, sympathetic or sensory neurons do not persist in lung explants as their cell bodies lie outside the lung.
Laser ablation of intrinsic pulmonary neurons was developed to test their local roles during morphogenesis, and to do so with single-cell precision, which contrasts with alternative methods, such as wholesale ganglion excision or pharmacological blockade [1],[2]. Two-photon laser-scanning microscopy provided an efficient and three-dimensionally localised ablation of eGFP-labeled neural cell bodies and axons. The energy from the laser is confined to a volume of approximately 1 μm3, permitting focal destruction of individual cells and axons (Figure 2A-B). Hence, lasing a single cell with two-photon energy destroys the target cell, but without imparting any collateral damage to cells immediately adjacent to, in front of, or behind the target (Figure 2C-F). Confocal imaging confirmed that no acute damage is done to bystander cells, including adjacent epithelium, endothelium, nerves or mesenchyme.
Subacute damage to surrounding cells was sought using Topro3-iodide as a dead-cell indicator dye. Even hours after targeted two-photon laser ablation, there were no delayed effects on adjacent bystander cells (Figure 2G-I), reaffirming that the technique provides ablation with single-cell precision. In contrast to prior approaches, such as wholesale ganglion excision, this selective neural ablation avoids gross disruption of organ anatomy, which has major effects on non-neural bystander cells and requires reconstitution of the disrupted explant.
To determine the effects of denervation on epithelial budding, this precise two-photon laser ablation technique was applied to one lung of each explanted pair. Confocal microscopy identified the eGFP labeled neuronal cell bodies and axons within the cultured left lung. At this stage, the explants are fractions of a millimeter in thickness, allowing the complete innervation of the lung to be visualised. Each neuronal cell body was individually destroyed by laser ablation. Likewise, each axon bundle was severed using laser ablation at the level of the proximal airway branches within the left lung. Thus, selective denervation of the left lung was achieved to determine its effects on epithelial budding. A non-ablated control for this intervention was to culture the paired right lung without any laser ablation (Figure 3A-C). A control ablation was performed in a stage-matched left lung explanted from a sibling embryo by imparting identical laser irradiation but applying it to non-neural cells in the mesoderm immediately adjacent to the neuronal cell bodies and axons (Figure 3D-F). Just like target regions for the neuronal ablation, these control ablations avoided the epithelium and endothelium. These nerve-ablated and control ablated explants and their partner right lungs were then cultured for two days.
The laser-denervated left lung added few if any new buds (Figure 3A-C). In contrast, the right lung without any intervention (non-ablated control) developed new buds as usual (n = 7 per group). Laser ablation of non-neural mesodermal cells in the left lung (control ablation) saw the left lung bud as profusely as unperturbed explants (Figure 3D-F). Likewise, budding halted after laser denervation of the right lung, but proceeded normally in the non-intervened left lung (Figure 3G-I) or when laser ablation was targeted only at non-neural mesodermal cells (Figure 3J-L). Overall, bud numbers more than tripled after control ablations, while they flat-lined after neural ablation (Figure 3M). Therefore, it is selective neural ablation rather than any non-specific effect of laser irradiation that abrogates local epithelial lung budding.
Previous studies reported that parasympathetic cholinergic signaling regulated salivary gland budding and epithelial stem cells [1],[2]. To test if neural ablation stopped lung budding due to interruption of parasympathetic signaling, lung explants were cultured with and without the muscarinic antagonist 1,1-dimethyl-4-diphenylacetoxypiperidinium iodide (4-DAMP, 10 to 20 M; n ≥12 per treatment). In contrast to reports in the salivary gland, muscarinic blockade (M3) did not affect lung budding at embryonic day 12.5 to 13 when early branching is under way (Figure 3N and Additional file 1: Figure S1 A-D). Similarly, pan-cholinergic blockade with atropine (20 μm, 40 μm; n ≥8 per treatment) failed to alter budding (see Additional file 1: Figure S1E). Even when synaptic release was blocked using botulinum toxin (50 ng/mL, 100 ng/mL, 200 ng/mL; n ≥12 per treatment), lung explants budded normally in culture (see Additional file 1: Figure S1F-J). Together, this shows that the failure of local lung budding after selective pulmonary denervation cannot be attributed to a loss of cholinergic neurotransmitter signaling.
Immunohistochemistry on lung explants confirmed that the laser ablation achieved selective neural destruction and also established the effects of denervation on pulmonary cell fates (Figure 4). The number of neural cells and endothelial cells were identified by beta3-tubulin and vascular epithelial growth factor receptor 2 (VEGFR2) staining, respectively (Figure 4A and B). The relative number of neural cells per explant was greatly reduced by neural ablation (Figure 4C; for all plots in Figure 4, n = 3 different samples, with 10 or more tissue sections per treatment).
Pulmonary cell proliferation, measured using phosphohistone H3 (pH3) immunostaining, was reduced after laser denervation (Figure 4D), while apoptosis rates as measured by cleaved caspase 3 immunohistochemistry were unchanged (mean, SEM; Figure 4E). Following neural ablation, VEGFR2 positive vascular endothelial cells numbered about a third of that of normal explants (Figure 4F), and the numbers of cells in endothelial cell clusters were reduced, indicating reduced endothelial proliferation following neural depletion (Figure 4G). Hence, pulmonary neural tissue regulates airway branching as well as epithelial and endothelial proliferation.
Next, we sought to determine to what extent neural regulation of airway morphogenesis is separable from neural regulation of epithelial and endothelial proliferation. In the mammalian model, this is not readily testable because epithelial and endothelial proliferation are themselves fundamental requirements for normal airway morphogenesis. Instead, airway morphogenesis was examined in Drosophila embryos (Figure 5A) because fly tracheogenesis shares genetic homologies with mammalian lung development but is completed via epithelial migration without an adjacent vasculature or epithelial proliferation (once the requisite number of tracheoblasts has been produced) [10]-[14].
Selective ablation of nerves in Drosophila embryos, therefore, tests whether neural control of airway morphogenesis is conserved between vertebrates and invertebrates and whether neural regulation persists independently of endothelial or epithelial proliferation [15]. Starting at stage 12, the elav driver is expressed in developing neural tissue in Drosophila embryos (Figure 5B) [16]. Using this driver, we selectively expressed the RicinA toxin in the nervous system (n >30). RicinA is a potent intracellular toxin that kills cells that express it. Lacking a B subunit, it cannot enter bystander cells to kill them [17]. In this model, toxin expression starts to kill neural cells after completion of tracheoblast proliferation but as tracheoblasts are beginning their migration to elaborate the tracheal airways. Therefore, any observed disruptions of the tracheal system derive from a requirement for nerves for the proper migration of tracheoblasts and the structuring of the tubular segments.
Tracheoblast migration is regulated by mitogen signaling and topographical guidance [12],[18]. Dorsal trunks form early, while the smaller tracheal branches develop later. Immunohistochemical labeling of the trachea and nervous system showed that selective RicinA-mediated lesioning of neurogenesis led to substantial disruption of airway morphogenesis (Figure 5C-F). Otherwise, the embryos looked grossly normal with appropriate formation of other organs such as the gut. Most often, dorsal trunks were relatively spared but were discontinuous and transposed ventral to their usual location. In contrast, smaller branches including the transverse connectives, lateral trunk, visceral, ganglionic and dorsal branches were smaller or absent. In severe examples, tracheal structure was lost, with the tracheoblasts clumped in spheres (Figure 5F). This suggests that as more neurons are killed by expressing RicinA, tracheoblast migration and segmental fusion are progressively impaired, due in part to loss of neuronal guidance. This explains the relative preservation of sections of a dorsal trunk structure, albeit often displaced from its expected location within the embryo, with more marked loss of the smaller branches that form later. Our tracheal phenotypes cannot be explained solely by indirect effects on other tissues, such as muscle, because they are distinct from phenotypes reported when these non-neuronal lineages are ablated [19],[20].
Immunostaining for the epithelium matched that of the 2A12 tracheal lumen marker confirming that the neural ablations cause true disruptions of tracheal structure rather than just perturbed levels of the lumen protein (Figure 5G-I). Together, these mammalian and Drosophila studies show that neural control of airway morphogenesis is conserved between mammals and invertebrates and that impacts on airway branching are independent of neural regulation of endothelial cell behaviors.