| We developed an immunostaining protocol that allows full visualization in 3-D of the entire bronchial tree in fixed, whole-mount lungs. We collected and stained lungs from hundreds of wild type mouse embryos between embryonic day 11 (E11) and E15 and determined the airway pattern, allowing the dynamic branching process to be reconstructed. This information was then used to construct a branch lineage representing the developmental history of the ∼5,000 branches of the bronchial tree.; Reconstructing the lineage revealed three local modes of branching, each generating a distinct arrangement of branches, which is used multiple times at stereotyped positions to generate the three-dimensional pattern of the lung. In domain branching, daughter branches form in proximal-to-distal order in rows (domains) along a parental branch. The position of these domains and the order in which they are used is fixed. While branching proceeds in proximal-to-distal order in one domain, branching begins again in the next domain to fill in the airway tree. In planar and orthogonal bifurcation, daughter branches form exclusively at the tip of the parent branch. In both modes of branching, daughter branches form by sequential bifurcation but in planar bifurcation each daughter branch bifurcates in the same plane as its parent, whereas in orthogonal bifurcation, daughter branches bifurcate along an axis orthogonal to that along which its parent branched. The branching process is stereotyped, but there are occasional errors, including "displacement" errors, in which a branch originates from an ectopic location; and, "skipping a generation," in which a branch is missing and a granddaughter branch originates directly from the grandparent.; These three distinct modes of branching serve as units of pattern, with different, presumably physiologically relevant, design properties. We propose that each mode of branching is controlled by a locally operative, genetically encoded subroutine, a series of discrete morphogenesis steps that generates the mode-specific branch patterns. Some of these steps are shared among different subroutines---branch initiation and outgrowth, for example. But some differ. For example, periodic proximal-to-distal branching and sequential use of domains are unique to domain branching. And there must be a difference in how the axis of bifurcation is set in orthogonal and planar bifurcation.; Subroutines are genetically separable. In a limited pilot screen, the branch patterns of lungs from five inbred mouse strains and from knockouts of members of the Sprouty family of receptor tyrosine kinase feedback inhibitors were analyzed. In lungs from Sprouty2 mutant embryos, the number of branches in specific domains is increased, and branching in these domains initiates more proximally than in wild-type: in the absence of Sprouty2 there is a homeotic-like transformation in which normally non-branching regions along each parent branch acquire the branching identity of more distal regions. In one of the inbred strains, C57BL/6J, specific domains off a parental branch are shifted distally with respect to the other domains. Thus, there is remarkably precise genetic control of branching---down to the level of individual domains.; Subroutines are used at defined positions in three fixed sequences. Remarkably, these three sequences of deployment describe the complete lineage of the bronchial tree. When and where a subroutine is used does not correlate simply with branch generation, lobe, or developmental time, suggesting that subroutine deployment is controlled by a more global master routine. There are also stereotyped local differences in how subroutines are used---for example, in the number and position of domains around branches that use domain branching---that are presumably also controlled by the master routine. Each use of a subroutine generates multiple branches, allowing the pattern of a large number of branches to be genetically encoded b... |
