Self-organized and directed branching results in optimal coverage in developing dermal lymphatic networks

Article https://doi.org/10.1038/s41467-023-41456-7 Self-organized and directed branching results in optimal coverage in developing dermal lymphatic networks Received: 26 July 2022 Check for updates 1234567890():,; 1234567890():,; Accepted: 5 September 2023 Mehmet Can Uçar 1,4, Edouard Hannezo 1,4 , Emmi Tiilikainen 2, Inam Liaqat 2, Emma Jakobsson2, Harri Nurmi2,3 & Kari Vaahtomeri 2,3 Branching morphogenesis is a ubiquitous process that gives rise to high exchange surfaces in the vasculature and epithelial organs. Lymphatic capillaries form branched networks, which play a key role in the circulation of tissue fluid and immune cells. Although mouse models and correlative patient data indicate that the lymphatic capillary density directly correlates with functional output, i.e., tissue fluid drainage and trafficking efficiency of dendritic cells, the mechanisms ensuring efficient tissue coverage remain poorly understood. Here, we use the mouse ear pinna lymphatic vessel network as a model system and combine lineage-tracing, genetic perturbations, whole-organ reconstructions and theoretical modeling to show that the dermal lymphatic capillaries tile space in an optimal, space-filling manner. This coverage is achieved by two complementary mechanisms: initial tissue invasion provides a non-optimal global scaffold via self-organized branching morphogenesis, while VEGF-C dependent side-branching from existing capillaries rapidly optimizes local coverage by directionally targeting low-density regions. With these two ingredients, we show that a minimal biophysical model can reproduce quantitatively whole-network reconstructions, across development and perturbations. Our results show that lymphatic capillary networks can exploit local selforganizing mechanisms to achieve tissue-scale optimization. Tubular epithelial and endothelial organs play key roles in the transport and exchange of fluids. For efficient function, these organs need to have a high surface-to-volume ratio (e.g., lungs and kidney) and/or even coverage of large tissue volumes (blood and lymphatic vasculature), which can be achieved by complex branched structures. The organ function and tissue environment set challenges to the regulatory mechanisms of branching morphogenesis. Although the molecular players controlling branching morphogenesis vary substantially from one organ to the next, several common cellular events and gene regulatory network motifs have been identified for branching1,2. From a biophysical perspective, different models of branching morphogenesis have been proposed in the past3–5, ranging from stereotypic fractal-like dynamics such as in the mammalian lung6, which allows for optimal gas transport and exchange, to a self-organized process relying on local cues such as local tubule density7–9. Although such selforganization from local cues provides a simple mechanism to tile space, it has been shown to do so at the expense of efficiency, showing large spatial fluctuations and poor space-filling properties9. Beyond such mechanisms for generating tree-like structures, several works, in particular in plant and animal vasculature which often form complex 1 Institute of Science and Technology Austria (IST Austria), Am Campus 1, 3400 Klosterneuburg, Austria. 2Translational Cancer Medicine Research Program, University of Helsinki, Biomedicum Helsinki, Haartmaninkatu 8, 00290 Helsinki, Finland. 3Wihuri Research Institute, Biomedicum Helsinki, Haartmaninkatu 8, 00290 Helsinki, Finland. 4These authors contributed equally: Mehmet Can Uçar, Edouard Hannezo. e-mail: ; kari.vaahtomeri@helsinki.fi Nature Communications | (2023)14:5878 1 Article loopy networks, have concentrated on rationalizing their final geometry and topology, with an emphasis on potential optimality in terms of homeostatic function10–14. Furthermore, numerous theoretical models have been proposed to describe the formation of vascular networks, based on both coarse-grained differential equations15–18, and using discrete simulation frameworks at the cellular scale19–21 as well as “hybrid” approaches combining (sub)cellular- and tissue-scale dynamics22. However, how the dynamical mechanisms of complex network growth relate to their final homeostatic structures and functions remain poorly understood. The lymphatic vasculature forms a branched network, which is characterized by blind-ended lymphatic capillaries (the treetop) that act as a site for tissue fluid and leukocyte entry. From the lymphatic capillaries, tissue fluid-derived lymph and leukocytes run into lymphatic collector vessels (the stalk) and thereafter to a series of lymph nodes, before entering the blood circulation at the subclavian vein. Most of the lymphatic vessels (LVs) grow via branching morphogenesis (i.e., lymphangiogenesis), which proceeds from the stalk to the treetop direction23,24. Multiple mouse models and correlative patient data demonstrate that the lymphatic capillary density and network maturation are essential for efficient lymphatic function23. However, how optimal is the coverage provided by lymphatic capillaries, and what are the mechanisms ensuring such coverage, remain poorly characterized at the molecular and cellular level. To study the mechanisms of branching morphogenesis of lymphatic capillaries, we chose the widely used model system of mouse ear pinna dermal lymphatic capillaries. The flat, quasi-two-dimensional, dermal tissue of the ear pinna allows whole network imaging and, thus, the reconstruction of the full developmental process of lymphatic capillary network formation. Here we show, by a combination of morphometric timepoint analyses, lineage tracing, genetic perturbations, quantitative reconstructions and theoretical modeling that developing lymphatic capillary networks exploit local selforganizing mechanisms to achieve tissue-scale optimization. Results Quantitative modeling of morphogenesis of the mouse ear pinna dermal lymphatic capillary network To characterize the postnatal morphogenetic process of the dermal lymphatic capillary network, we analyzed mouse ear pinna size and ventral lymphatic capillary network structure throughout its postnatal development and upon maturity at P21 (Supplementary Fig. 1A-C). Mouse ear pinna consists of cartilage and adipocyte layer-separated ventral and dorsal halves. In the first postnatal days, pre-existing LV network is located in the deep dorsal dermis, whereas the ventral side is devoid of LVs (Supplementary Movie 1). At P4, several trees of LYVE1+ LVs invade the ventral dermis from the ear pinna stalk and grow towards the tip of the ear pinna (Fig. 1A). At P6, the pre-existing deep dorsal LV network sprouts to the ventral dermis at few locations close to the ear pinna tip, and expands horizontally on the superficial dermal layers to form large tree-like structures, which are linked to deep dorsal LV network (Fig. 1A, B, Supplementary Fig. 1D–F, and Supplementary Movies 2 and 3). Some of the ventral tip-tre (...truncated)