Dynamics of branched tissue assembly

The assembly of cells into tissues is a complex process controlled by numerous signaling pathways to ensure the fidelity of the final structure. Tissue assembly is also very dynamic, as exemplified by the formation of branched organs. Here we present two examples of tissue assembly in branched systems that highlight this dynamic nature: formation of the tracheal network in Drosophila melanogaster and the ducts of the mammary gland in mice. Extension of the branches during tracheal development is a stereotyped process that produces identical organ geometries across individuals, whereas elongation of the ducts of the pubertal mammary gland is a non-stereotyped process that produces unique patterns. By studying these two organs, we can begin to understand the dynamic nature of development of other stereotyped and non-stereotyped branching systems, including the lung, kidney, and salivary gland.

Th e self assembly of cells into tissues and organs is an elegant and intricate process that is vital for development and homeostasis. During organogenesis, the assembly of cells is controlled genetically as well as through cues from cell-cell and cell-matrix interactions [1,2]. Cells assemble into higher-order patterns that are either stereotyped such that there is little variation in the timing and fi nal architecture between individuals, or non-stereotyped, which leads to diff erences in the details [3]. Th e branching of the ducts of the pubertal mammary gland is a non-stereotyped process in which the interactions between the ducts and their surrounding micro environment produce unique tree-like architectures. In contrast, branching of the trachea in Drosophila melanogaster is a stereotyped process that is under strict genetic control. Th ese two systems serve as excellent models with which to investigate the dynamic interplay between cells during tissue formation, under two diff erent modes of control.

Formation of the trachea in D. melanogaster
Th e trachea of the fruit fl y is a ductal structure responsible for the delivery of oxygen to tissues. Th is organ forms during embryonic development and involves invagination, division, extension and fusion of select cells of placodes along the lateral ectoderm ( Figure 1). Th e determination and positioning of the placodes as well as the choreographed steps that lead to the formation of the trachea is predominantly under the control of the gene trachealess (Trh) [4,5]. At embryonic stage 11, through the activation of the gene rhomboid (Rho) by Trh, the 20 placodes containing approximately 40 cells each invaginate through apical constriction and undergo mitotic division to form 80-cell tracheal sacs [6][7][8]. Th e internalized cells then extend from the sites of invagination to form six distinct branches: the dorsal branch (DB), dorsal trunk (DT), visceral branch (VB), lateral trunk (LT), ganglionic branch (GB), and transverse connective (TC). Th e TC forms from the mid-region of each sac, and DTs extend along the anterior-posterior axis and fuse with DTs of the neighboring sacs to form the main throughway of the trachea. Th e cells of the DB migrate dorsally and select fusion cells that bind with their counterpart cells of DBs at the opposite side of the embryo [9], while the cells of the LT migrate ventrally and bind with their counterpart LT cells of neighboring metameres to the anterior and posterior, thus fusing the branches and forming additional contiguous pathways for circulation. Th e cells of the VB and GB do not fuse with those of the neighboring sacs but still branch and extend into the surroundings to complete the tracheal structure [7]. Th e extensions of the branches are guided by various signaling pathways, including Breathless (Btl) [10], Decapentaplegic (Dpp) [11] and Slit [12]. Even in a strictly stereotyped branching system such as this, however, the role of cellular dynamics is vital and can be seen in the extension and fusion of the DB.
Th e DB consists of approximately six cells and these cells migrate dorsally away from the sac toward the morphogen Branchless (Bnl), which acts as a chemoattrac tant secreted by the surrounding cells [13]. However, the branch is able to extend normally even when all

Abstract
The assembly of cells into tissues is a complex process controlled by numerous signaling pathways to ensure the fi delity of the fi nal structure. Tissue assembly is also very dynamic, as exemplifi ed by the formation of branched organs. Here we present two examples of tissue assembly in branched systems that highlight this dynamic nature: formation of the tracheal network in Drosophila melanogaster and the ducts of the mammary gland in mice. Extension of the branches during tracheal development is a stereotyped process that produces identical organ geometries across individuals, whereas elongation of the ducts of the pubertal mammary gland is a non-stereotyped process that produces unique patterns. By studying these two organs, we can begin to understand the dynamic nature of development of other stereotyped and nonstereotyped branching systems, including the lung, kidney, and salivary gland. but one of the cells is devoid of Btl, the receptor for Bnl [14]. Th e cell with functional Btl receptors acts as a leader and is able to guide the rest of the cells. In wild-type embryos, a leader cell emerges even when every cell expresses Btl. At this stage of development, the cells of the DB are nearly identical and there is no predetermined leader; however, there are small diff erences in the expression level of Btl. By exploiting these small diff erences, the cells with the highest expression levels of Btl jockey for the lead position in the newly forming branch, while simultaneously amplifying the diff erence in the activation of Btl among the population through Notch-mediated lateral inhibition [14][15][16][17]. Th e activation of Btl by Bnl induces the production of Delta, a ligand that binds to and activates Notch in the neighboring cells. Activated Notch inhibits signaling downstream of the Btl receptor and production of Delta within that cell. Over time, this results in one cell with substantially higher Btl activity than its neighbors [16,17]. Th e successful cell assumes the role of the leader at the tip of the branch while the other cells form the stalk. Th is establishes a hierarchy and the leader cell is able to effi ciently move the branch up the Bnl gradient. As the dorsal branch extends, one of the cells of the branch comes in contact with a cell from the neighboring branch and binds to it, thus fusing the branches to form a contiguous airway [9]. Notch signaling again plays a similar role by aiding Bnl and Dpp and ordaining the fusion cell fate [16,18,19].
During metamorphosis, the trachea once again undergoes signifi cant remodeling. Posterior tracheal branches are lost while anterior branches expand to cover that space and new cells replace most of the cells in the existing branches. Progenitor cells called tracheoblasts that have remained quiescent during embryonic and larval development begin to proliferate and travel along the TC and VB under the guidance of Btl and replace the old cells [20,21]. However, the DBs are replaced not by progenitor cells but by diff erentiated DB cells that reenter the cell cycle and proliferate to replace the existing branch [21].

Formation of the mammary ducts
Th e mammary gland is a ductal structure responsible for production and secretion of milk in mammals. Th e ducts of the mammary gland are composed of a hollow lumen surrounded by a layer of luminal epithelial cells, which are themselves surrounded by a layer of myoepithelial cells that create the basement membrane that separates the epithelium from the stroma. During puberty in rodents, hormones from the ovary and the pituitary gland induce the formation of bulbous terminal end buds (TEBs) at the tips of the rudimentary epithelial ducts [22]. Extension and branching of the TEBs into the surrounding mesenchyme sculpt the highly branched  [23]. Th e cap cells and the preluminal epithelial cells are highly proliferative. Th e force generated from the rapid proliferation, along with the construction of basement membrane laterally and destruction of extracellular matrix at the front, is thought to propel the ducts forward at an average rate of 500 μm per day [24,25]. As the TEBs extend, the cap cells move laterally along the duct and diff erentiate into myoepithelial cells, while the preluminal epithelial cells give rise to luminal epithelial cells [26]. Th e structure of the TEB is maintained through E-cadherin-and P-cadherin-containing adherens junctions and netrin-1/neogenin interactions. Preluminal cells interact through E-cadherin, while the cap cells are connected by P-cadherin [27]. Local interactions between netrin-1, a neural guidance cue present on preluminal cells, and neogenin, a netrin receptor present on the cap cells, help maintain the proximity of the preluminal cells to the cap cells [25]. Loss of either netrin or neogenin results in the separation of the cap cell layer from the preluminal layer and destabilization of the TEB [25].
Estrogen and growth hormone initiate ductal elongation by activating the estrogen receptor (ERα) and the growth hormone receptor, respectively, in the stromal cells surrounding the epithelium [27][28][29]. Th e activation of growth hormone receptor enables TEB formation through the insulin-like growth factor pathway [30]. Estrogen signaling leads to the activation of progesterone receptors in the epithelium, which induces epithelial proliferation and ductal movement. Estrogen signaling also stimulates the stromal cells to activate matrix metallo proteinases (MMPs) through amphiregulin and epidermal growth factor receptor [22]. Stromal cells activate several MMPs, including MMP3 and MMP14. MMP14 activates MMP2, which aids in ductal elongation by preventing apoptosis of the cells within the TEB. MMP2 and MMP3 also play a role in determining sites of lateral branching along the duct. High MMP2 prevents the initiation of lateral branches by stimulating the production of transforming growth factor (TGF)β, whereas MMP3 enables lateral branching by degrading the basement membrane and allowing the cells to extend through the fat pad [31].
Unlike tracheal branching, there is no evidence for a global chemoattractant that guides the TEBs through the fatpad [32]. Th e TEBs weave through the fatpad, extending lateral branches and bifurcating along the way, by integrating the global stimulatory cues with more local cues from neighboring cells, their environment and their own geometry [32]. Patterns of endogenous mechanical stresses arise along the ducts due to asymmetries in their geometry coupled with isometric contraction of the cells and their ability to transmit these forces to their neighbors through adherens junctions [33]. Th ese mechanical stress patterns appear to regulate the sites at which branches initiate. Indeed, three-dimensional organo typic culture models revealed that branches initiate from sites of highest mechanical stress [32], and that cells in these regions of high mechanical stress adopt a more motile mesenchymal phenotype [34,35], increase signaling through focal adhesion kinase (FAK) [32], and increase production of MMPs to aid in degradation of surrounding matrix, all of which enable branch extension [36]. However, while high mechanical stress is necessary for branch initiation, it is not suffi cient [32]. Branch initiation and extension are also controlled by the local concen tration of morphogens such as TGFβ [37]. High con centrations of TGFβ prevent branching of the ducts, and autocrine secretion of this inhibitory morphogen by the epithelial cells is thought to prevent aberrant branching and maintain proper spacing between ducts [37,38]. By responding to these various stimulatory and inhibitory cues, the TEBs navigate through the stroma until they reach the limits of the fat pad, where they regress and form terminal ducts. However, the ducts still maintain a certain population of undiff erentiated stem cells that are capable of self-renewal and repopulation of the entire mammary gland [39,40]. Th ese undiff erentiated cells are thought to aid in the remodeling of the mammary gland during and after pregnancy, where further ductal expansion and regression occur, respectively [40].

Conclusion
Branching of the trachea and the mammary gland highlight the dynamic process of tissue formation. Th ese two systems show how stem cells, progenitor cells and diff erentiated cells come together, and through a combination of global and local guidance cues specify the fi nal architecture of the tissue. During puberty in mammals, the pluripotent cap cells combined with other cell types of the TEB and through the regulation of the microenvironment produce the unique pattern of the ductal tree. In the formation of D. melanogaster trachea, even though the fi nal architecture is fi xed, the cells of the invaginated tracheal placodes have a high degree of freedom to determine the leaders, the fusion cells and those that will remain quiescent during the various branch formations. In addition to normal development, the dynamics of stem cells also come into play during tumor formation. Cancer stem cells (CSCs) are specialized cells within tumors that are capable of selfrenewal and initiation of tumors [41]. CSCs have all the characteristics of normal stem cells and are even capable of generating non-malignant structures [42,43]. Th e origin of CSCs is still hotly debated. While the similarities between CSCs and normal stem cells might strongly suggest that CSCs arise through aberrant diff erentiation of normal stem cells, there is also evidence to suggest that CSCs might arise from progenitors or even through dediff erentiation of fully diff erentiated cells. Under standing the dynamics of cells during normal development will help in understanding CSCs and malignant development in general, as many of the processes involved in normal development are conserved in malignant development.
Th anks in large part to advances in technology, we are fi nally able to glimpse at the complexities of developmental processes. However, better in vivo imaging approaches are needed to fully answer most of the open questions, as it is near impossible to investigate dynamic processes using still images. Advances in live imaging could help, for example, to study the formation and movement of the TEBs. Studies in culture have suggested that organization of the TEBs might involve cellular sorting to optimize the structure for invasion through the fatpad [44]; however, this has yet to be shown in vivo. Furthermore, the mechanism by which the TEB moves through the fatpad is still unclear [45]. In addition to imaging, advances in techniques to manipulate in vivo microenvironments might prove to be the most fruitful venture. Th e mechanics of the microenvironment integrate with molecular cues in the regulation of cell division, motility, branching, and stem cell diff erentiation [46]. Laser ablation studies in D. melagonaster have been used to show the involvement of tensile forces in tracheal branching [47][48][49]. At present, however, culture models still provide one of the best means of studying the eff ects of the mechanics of the microenvironment on these processes in a controlled and quantifi able manner. While they do suff er from drawbacks of artifi ciality, sophis ticated three-dimensional culture models can aid to bridge the gap between the plastic world and the in vivo world, and could play an important role in studying the interplay between multiple cell types during development [50,51]. A better appreciation of the dynamic nature of tissue assembly will have a great impact on the study of normal and malignant development and may also will help advance the fi eld of tissue engineering.