Distinct fibroblast subsets regulate lacteal integrity through YAP/TAZ-induced VEGF-C in intestinal villi AbstractEmerging evidence suggests that intestinal stromal cells (IntSCs) play essential roles in maintaining intestinal homeostasis. However, the extent of heterogeneity within the villi stromal compartment and how IntSCs regulate the structure and function of specialized intestinal lymphatic capillary called lacteal remain elusive. Here we show that selective hyperactivation or depletion of YAP/TAZ in PDGFR尾+ IntSCs leads to lacteal sprouting or regression with junctional disintegration and impaired dietary fat uptake. Indeed, mechanical or osmotic stress regulates IntSC secretion of VEGF-C mediated by YAP/TAZ. Single-cell RNA sequencing delineated novel subtypes of villi fibroblasts that upregulate Vegfc upon YAP/TAZ activation. These populations of fibroblasts were distributed in proximity to lacteal, suggesting that they constitute a peri-lacteal microenvironment. Our findings demonstrate the heterogeneity of IntSCs and reveal that distinct subsets of villi fibroblasts regulate lacteal integrity through YAP/TAZ-induced VEGF-C secretion, providing new insights into the dynamic regulatory mechanisms behind lymphangiogenesis and lymphatic remodeling. IntroductionEach small intestinal villus contains a highly specialized lymphatic capillary called a lacteal, which is essential for dietary fat uptake and gut immune surveillance1,2,3. The lymphatic endothelial cells (LECs) that make up lacteals constantly but slowly proliferate under vascular endothelial growth factor (VEGF)-C鈥揤EGF receptor (VEGFR)3 and delta-like 4 (DLL4)-Notch signaling, whereas most LECs forming the lymphatic vessels (LVs) in other organs are relatively quiescent in adults4,5. Moreover, adrenomedullin-calcitonin receptor and VEGF-A-VEGFR2 signaling have been demonstrated as regulators of lacteal integrity and function, and vascular endothelial (VE)鈥恈adherin is also required for lacteal morphogenesis and maintenance6,7,8. Renewal capacity is unique to certain organ-specific LECs, which implies that distinct regulatory processes may support lacteals to cope with the dynamic microenvironment and highly repetitive mechanical forces of intestinal peristalsis1,2,3. A lacteal is surrounded by the villus stroma, which is composed of capillaries, pericytes, smooth muscle cells (SMCs), fibroblasts, immune cells, and extracellular matrix residing under the rapidly renewing intestinal epithelial cells1,2,3. In this milieu, a subset of SMCs has been suggested as an important source of VEGF-C for lacteal maintenance5. On the other hand, emerging evidence suggests that the mesenchymal compartment of the intestinal stroma provides Wnt ligand for intestinal stem cells and is important in maintaining gut homeostasis by tightly coordinating the renewal capacity of intestinal stem cells9,10,11,12. Indeed, mesenchymal stromal cells are also being recognized to have diverse roles in microenvironmental remodeling in various organs13,14,15,16. However, although lacteals are compactly surrounded by mesenchymal stromal cells, their contributions to lacteal maintenance remain unknown.The mammalian Hippo signaling pathway is the key regulator that controls various biological processes including cellular differentiation, proliferation, tissue homeostasis, and stem cell renewal17,18,19. The key components of the Hippo kinase cascade are LATS1 and LATS2 (LATS1/2), and the final transcriptional regulators YAP and TAZ (YAP/TAZ) primarily bind to TEA domain transcription factors (TEAD) to exert their activities18. Recent studies have demonstrated that the Hippo pathway regulates stromal cells, including blood endothelial cells (BECs), LECs, SMCs, and pericytes, to promote tissue development and function in a highly context-dependent manner20,21,22,23,24. Indeed, intimate Hippo-dependent crosstalk among these cell types is essential for tissue-specific formation and specialization of vessels, as well as their remodeling20,21,22,23,24. YAP/TAZ are regulated by mechanical stress, and they are well-known mediators and effectors of mechanical cues for the regulation of cell proliferation and differentiation25,26,27,28,29. In particular, mesenchymal stromal cells in intestinal villi are placed under high levels of repetitive mechanical stress and extreme variations in osmotic pressure1,2,3 that can modulate YAP/TAZ.Here, we explored the role of intestinal villi stromal cells by generating PDGFR尾-specific YAP/TAZ hyperactivation or depletion mouse models. We show that PDGFR尾+ intestinal stromal cells (IntSCs) play pivotal roles in lacteal maintenance and function by YAP/TAZ-dependent secretion of VEGF-C. We demonstrate that YAP/TAZ of PDGFR尾+ IntSCs integrate mechanical cues and VEGF-C secretion to support lacteal integrity. Using single-cell RNA sequencing, we reveal the heterogeneity of PDGFR尾+ IntSCs by identifying five distinct clusters of fibroblasts in addition to enteric SMCs and mural cells. Furthermore, we visualize the unique location of each subtype using differentially expressed marker genes identified in our analysis. Our results show that certain subsets of fibroblasts surrounding a lacteal secrete VEGF-C upon YAP/TAZ activation. These findings propose the novel and dynamic regulatory mechanism of lacteal maintenance and function by distinct subsets of fibroblasts, which constitute the peri-lacteal microenvironment.ResultsYAP/TAZ hyperactivation in PDGFR尾+ IntSCs promotes aberrant lacteal spouting and branching network formation in the small intestinal villiLATS1/2 constitute the core of Hippo pathway, which inhibits YAP/TAZ activities through phosphorylation-dependent cytoplasmic retention and degradation18. To gain insights into the role of Hippo pathway in PDGFR尾+ cells, we hyperactivated YAP/TAZ by deleting both Lats1/2 in these cells in a tamoxifen-dependent manner using the Lats1/2i鈭哖尾C mutants, which were generated by crossing Pdgfrb-Cre-ERT2 mouse30 with Lats1fl/fl/Lats2fl/fl mouse31,32 (Fig.1a). Cre-ERT2-negative but flox/flox-positive littermates were defined as wild-type (WT) mice in each experiment.Fig. 1: YAP/TAZ hyperactivation in PDGFR尾+ IntSCs induces lacteal sprouting and branching.a Diagram depicting the generation of Lats1/2i螖P尾C mouse and PDGFR尾+ cell-specific depletion of Lats1/2 in 8-week-old mice and analyses at 2 weeks later. b, c Representative images of LYVE-1+ lacteals and CD31+ capillary plexus under the E-cadherin+ epithelial cells in intestinal villi of the indicated part of small intestine and comparisons of the lacteal surface area per villi, lacteal length, villi length, and villi width in the duodenum (DD), jejunum (JJ), and ileum (IL) in WT and Lats1/2i螖P尾C mice. Dots indicate values of 101鈥?17 villi/group in n鈥?鈥? mice/group pooled from three independent experiments. Horizontal bars indicate mean 卤 SD and *** P鈥?lt;鈥?.0001 versus WT by two-tailed Mann鈥揥hitney U test. n.s., not significant. Scale bars, 100鈥壩糾. d, e Representative images and comparisons of the number of lacteal sprouts and branches per 100鈥壩糾 of lacteal length in WT and Lats1/2i螖P尾C mice. Each dot indicates a mean value of 10鈥?0 villi/mouse and n鈥?鈥? mice/group pooled from three independent experiments. Horizontal bars indicate mean 卤 SD and P value versus WT by two-tailed Mann鈥揥hitney U test. Scale bars, 50鈥壩糾. f, g Representative images and comparisons of the number of Prox1+ lymphatic endothelial cells (LECs) and Prox1+EdU+ proliferative LECs (white arrowheads) per 100鈥壩糾 of lacteal length in WT and Lats1/2i螖P尾C mice. Each dot indicates a mean value of 10鈥?0 villi/mouse and n鈥?鈥? mice/group pooled from three independent experiments. Horizontal bars indicate mean 卤 SD and P value versus WT by two-tailed Mann鈥揥hitney U test. Scale bars, 50鈥壩糾.Full size imageThe most striking finding was observed along the whole small intestine including duodenum, jejunum, and ileum of Lats1/2i鈭哖尾C mice, which revealed aberrant lacteal sprouting and branching network formation, while lacteals of WT mice showed single, blunt-ended, and finger-like round tube that is representative of a normal lacteal (Fig.1b, c). The surface area and length of lacteals increased by 鈭?/span>1.6 and 鈭?/span>1.3-fold, respectively, while villi width was reduced by 鈭?/span>23% in all portions of small intestine in Lats1/2i鈭哖尾C mice compared with WT; villi length was not significantly changed (Fig.1b, c). Detailed analyses of lacteals in jejunum revealed presence of 8鈥? lymphatic sprouts and 11鈥?2 lymphatic branching networks in 100鈥壩糾 length of each lacteal of Lats1/2i鈭哖尾C mice, which were rarely visible in WT mice (Fig.1d, e). Besides, numbers of Prox1+ LECs and EdU+Prox1+ proliferating LECs increased by 2.4-fold and 18.1-fold in the lacteals of Lats1/2i鈭哖尾C mice compared with those of WT mice (Fig.1f, g), indicating that the lacteal LECs are actively proliferating in Lats1/2i鈭哖尾C mice.When we examined the expressions and distributions of YAP/TAZ in the mouse intestinal villi, YAP/TAZ proteins were moderately enriched in PDGFR尾+ IntSCs as a nuclear (N) and cytoplasmic (C) localization type (N鈥?鈥塁 type) in WT (Supplementary Fig.1a, b). In comparison, their levels were increased and distribution pattern was changed to a predominant nuclear localization type (N鈥?gt;鈥塁 type) in PDGFR尾+ IntSCs of Lats1/2i鈭哖尾C mice (Supplementary Fig.1a, b). Moreover, to confirm that the LATS1/2 deletion induces aberrant lacteal phenotypes via YAP/TAZ, we observed the lacteal phenotypes using Lats1/2-Yap/Tazi鈭哖尾C mice (Supplementary Fig.2a). The aberrant sprouting and branching phenotypes in lacteals of Lats1/2i鈭哖尾C mice were restored in Lats1/2-Yap/Tazi鈭哖尾C mice (Supplementary Fig.2b). This result indicates that LATS1/2 deletion alters lacteal morphogenesis through YAP/TAZ.YAP/TAZ hyperactivation in PDGFR尾+ IntSCs alters lacteal LEC junctions and impairs lipid drainage into lactealsLVs are lined by LECs and they consist of initial lymphatic capillaries with button-like junctions and collecting LVs with zipper-like junctions33. Similar to button-like junctional organization of LECs in lymphatic capillaries of peripheral organs33, lacteal junctions consisted mostly of discontinuous button-like (鈭?/span>54%) junctions (Fig. 2a, b). However, the proportion of button-like junction was reduced by 鈭?/span>83%, while the proportion of zipper-like junction increased by 鈭?/span>4.0-fold in lacteals of Lats1/2i鈭哖尾C mice (Fig.2a, b). Further analysis with transmission electron microscopy revealed aberrantly overlapped, large vacuole-containing complicated and thickened lacteal LECs with closed junctions, and chylomicrons were rarely detected in lacteal lumen of Lats1/2i鈭哖尾C mice. On the other hand, lacteal junctions were paracellularly open and chylomicrons were observed inside and outside of the lacteal lumen in WT mice (Fig.2c). These results imply that the transport of chylomicrons into the lacteal is largely impaired in Lats1/2i鈭哖尾C mice compared with WT.Fig. 2: Button-to-zipper junctional transition and impaired lipid absorption in Lats1/2i螖P尾C mice.a, b Representative images and comparison of VE-cadherin+ lymphatic endothelial cell (LEC) junctions of LYVE-1+ lacteals in WT and Lats1/2i螖P尾C mice. Black dotted box is magnified in the right panel and arrowheads indicate the lacteal junctional pattern that is dominant in each group with the same color described in b. Horizontal bars of each colored segment represent mean 卤 SD of 5鈥?0 villi/mouse and n鈥?鈥? mice/group pooled from three independent experiments. *** P鈥?lt;鈥?.0001 versus WT by two-way ANOVA with Holm鈥揝idak鈥檚 multiple comparisons test. Scale bars, 20鈥壩糾. c Representative transmission electron micrographs (EM) of a lacteal in WT and Lats1/2i螖P尾C mice. Black dotted box is magnified in the right panel. LL, lacteal lumen; CM, chylomicron; IS, interstitium of lamina propria. Similar findings were observed in n鈥?鈥? mice/group from two independent experiments. Scale bars, 1 渭m. d Diagram depicting the schedule of intravital imaging of lipid clearance from lamina propria via lacteals after supplying the BODIPY-fatty acid (FA) into the WT or Lats1/2i螖P尾C mice and their analyses until after 46鈥塵in. e, f Representative intravital images and comparisons of the drainage of BODIPY-FA from the lamina propria (white asterisks) through the LYVE-1+ lacteal in WT and Lats1/2i螖P尾C mice. Each dot indicates a value of 4鈥? villi/mouse and n鈥?鈥? (WT) or n鈥?鈥? (Lats1/2i螖P尾C) mice pooled from four independent experiments. Horizontal bars indicate mean 卤 SD and P value versus WT by two-tailed Mann鈥揥hitney U test. FI, fluorescence intensity; AU, arbitrary unit. Scale bars, 20鈥壩糾. g Diagram depicting the schedule of olive oil gavage at 12鈥塰 prior to the intestinal sampling in WT or Lats1/2i螖P尾C mice. h, i Representative images and comparison of Oil red O staining in intestinal villi of WT and Lats1/2i螖P尾C mice. Each dot indicates a mean value of 5鈥? villi/mouse and n鈥?鈥? mice/group pooled from two independent experiments. Horizontal bars indicate mean 卤 SD and P value versus WT by two-tailed Mann鈥揥hitney U test. Scale bars, 100鈥壩糾.Full size imageTo ensure this finding, we measured lipid clearance from the lamina propria of small intestinal villi into the lacteal after intraluminal loading of boron-dipyrromethene fluorescent-conjugated fatty acid (BODIPY-FA) using an intravital imaging system34. Normalized residual signals of BODIPY-FA in the lamina propria at 36鈥塵in and 46鈥塵in after BODIPY-FA loading were 鈭?/span>20鈥?0% higher in Lats1/2i鈭哖尾C mice compared with WT mice (Fig.2d鈥揻). Moreover, oil red O staining of Lats1/2i鈭哖尾C mice small intestine at 12鈥塰 after olive oil gavage revealed 5.5-fold increased accumulation of lipid in the lamina propria compared with WT mice (Fig.2g鈥搃). These results indicate that alterations in lacteal junction is accompanied by impaired dietary lipid uptake through the lacteal in Lats1/2i鈭哖尾C mice.To delineate the cause of lacteal impairment, we performed detailed morphological analysis of the intestinal villi. Lacteals are surrounded by longitudinal bundles of SMCs that facilitate transport of lymph containing chylomicron by periodic squeezing under the control of autonomic nervous system1,3,34. Notably, alignment of villus SMCs was skewed, distorted, and even aberrantly intermingled with the lacteals, and their coverage area was decreased by 55.9% in Lats1/2i鈭哖尾C mice compared with WT mice (Supplementary Fig.3a鈥揷). Also, while VEGFR3 expression in lacteal was increased by 1.6-fold, no significant change in VEGFR2+ capillary plexus density was observed in intestinal villi of Lats1/2i鈭哖尾C mice compared with WT mice (Supplementary Fig.3d, e). Although the overall shape of the PGP9.5+ neural network was slightly altered, the proportion of immune cells including CD3+ T cells and F4/80+ macrophages and VE-cadherin+ alignment and junctional pattern of the BECs of capillary plexus in the lamina propria were not visibly altered in Lats1/2i鈭哖尾C mice compared with WT mice (Supplementary Fig.3d鈥揼). In addition, F4/80+ macrophages, previously suggested as source of VEGF-C35, showed no significant changes in their numbers between WT and Lats1/2i鈭哖尾C mice. (Supplementary Fig.4a鈥揷). Furthermore, no notable alterations were detected in LVs of other organs including small intestinal submucosa, ear skin, tracheal mucosa, or central tendon of the diaphragm (Supplementary Fig.5a鈥揷). Collectively, these results suggest that lacteal phenotypes are primary effect by hyperactivation of YAP/TAZ in PDGFR尾+ IntSCs.VEGF-C is upregulated in PDGFR尾+ IntSCs of Lats1/2i鈭哖尾C miceThese findings led us to investigate how activation of YAP/TAZ in PDGFR尾+ IntSCs causes lacteal spouting, branching, and proliferation. We generated tdTomatorP尾C reporter mice by crossing Pdgfrb-Cre-ERT2 mouse with Rosa26-tdTomato reporter mouse to trace PDGFR尾+ cells in the small intestinal villi, where we observed confined and distinct distributions of PDGFR尾+ signals in subsets of longitudinal SMCs surrounding the lacteal, fibroblasts, and pericytes along the capillary network in the small intestinal stroma (Supplementary Fig.6a鈥揷). We then generated Lats1/2i螖-tdTomatorP尾C mouse by crossing tdTomatorP尾C mouse with the Lats1fl/fl/Lats2fl/fl mouse and isolated PDGFR尾+ IntSCs from the small intestine of tdTomatorP尾C and Lats1/2i螖-tdTomatorP尾C mice for transcriptomic analysis with focus on comparison of secretory proteins (Supplementary Fig.6d, e). Of note, heat map of RNA sequencing data demonstrated that genes including Wisp2 (145.1-fold), Tnfsf15 (14.2-fold), Bdnf (8.9-fold), Ebi3 (8.4-fold), Angpt2 (6.3-fold), Vegfc (6.1-fold) and Ctgf (5.9-fold) were highly upregulated among the top 23 differentially expressed secretory molecules between tdTomatorP尾C and Lats1/2i螖-tdTomatorP尾C mice (Supplementary Fig.6f). To further delineate the secretory molecule responsible for lacteal sprouting, we employed sprouting lymphangiogenesis assay using the spheroidal aggregates of LECs36. While VEGF-C alone induced robust sprouting of LEC spheroid, WISP2, TNFSF15, BDNF, EBI3, or ANGPT2 exhibited no significant effect on sprouting regardless of the presence or absence of VEGF-C (Supplementary Fig.6g, h), implying that VEGF-C is likely the main factor responsible for the lacteal spouting. Indeed, mRNA levels of lymphangiogenic factors (Vegfc, 3.6-fold; and Vegfd, 2.4-fold) and VEGF-C activation factor (Ccbe1, 2.8-fold) were increased, although Vegfa, Adamts3, Tgfb, and Ifng were not significantly altered in the small intestinal lysates of Lats1/2i鈭哖尾C mice compared with WT mice (Supplementary Fig.7a). Moreover, gene sets related to actin filament branching, apical surface cell polarity, and maintenance of cell polarity were remarkably altered according to gene set enrichment analysis of the isolated small intestinal LECs in Lats1/2i螖-tdTomatorP尾C mice compared with those of tdTomatorP尾C mice (Supplementary Fig.7b). Ingenuity Pathway Analysis also revealed changes that were mostly related to the endothelial growth signaling and cell cytoskeleton regulation in small intestinal LECs of Lats1/2i螖-tdTomatorP尾C mice (Supplementary Fig.7c). Overall, our data indicates that VEGF-C level is not only upregulated but it could also be the main factor for lacteal phenotypes upon YAP/TAZ hyperactivation in the PDGFR尾+ IntSCs.VEGFR3 blockade completely abrogates aberrant lacteal sprouting and hyperbranching in Lats1/2i鈭哖尾C miceAs Vegfc was highly upregulated in PDGFR尾+ intestinal stromal cells by YAP/TAZ hyperactivation, we sought to determine whether aberrant lacteal sprouting and hyperbranching are dependent on the major lymphangiogenic signaling, VEGF-C鈥揤EGFR3 pathway. To interfere with this signal, we intraperitoneally (i.p.) injected 1012 viral particles of adeno-associated viral vectors (AAVs) encoding soluble VEGFR3 (AAV鈥搈VEGFR31-4-Ig) or control vector (AAV鈥搈VEGFR34-7-Ig)37 on the day of tamoxifen delivery and examined the lacteal phenotypes of WT and Lats1/2i鈭哖尾C mice 2 weeks later (Fig.3a). The vector-encoded VEGFR3 protein was still detected in the serum at 2 weeks after AAV delivery (Fig.3b). Strikingly, aberrant lacteal phenotypes of Lats1/2i鈭哖尾C mice including increased Prox1+ lacteal LEC number, hypersprouting, and hyperbranching were completely blocked by administration of AAV鈥搈VEGFR31-4-Ig (Fig.3c鈥揺). Moreover, lacteal LEC junctional alteration from predominant button-like pattern to zipper-like pattern in Lats1/2i鈭哖尾C mice was completely blocked (Fig.3f, g). In comparison, our approach to block VEGFR2 signaling by 伪-VEGFR2 blocking antibody, DC101 (50鈥塵g/kg every alternative day for 2 weeks), could only partially restore the aberrant lacteal sprouting and branching, but almost no changes were observed in the number of lacteal Prox1+ LECs and lacteal junctional transition compared with IgG control in Lats1/2i鈭哖尾C mice (Supplementary Fig.8a鈥揻). These results indicate that activation of VEGF-C鈥揤EGFR3 signaling plays a greater role than VEGFR2 signaling in aberrant lacteal sprouting and branching led by YAP/TAZ hyperactivation in PDGFR尾+ IntSCs.Fig. 3: VEGFR3 blockade abrogates lacteal sprouting and branching in Lats1/2i螖P尾C mice.a Diagram for analyses of adult WT and Lats1/2i螖P尾C mice two weeks after single intraperitoneal (i.p.) injection of AAV鈥搈VEGFR3(4-7)-Ig (mVR3(4-7)-Ig, control) or AAV-mVEGFR3(1-4)-Ig (mVR3(1-4)-Ig, VEGFR3 blockade). b Representative immunoblot analysis showing mVEGFR3-Ig proteins in serum two weeks after single i.p. injection of AAV鈥搈VEGFR3(4-7)-Ig or AAV-mVEGFR3(1-4)-Ig (1鈥壝椻€?012 viral particles in 150鈥?00鈥壩糽 PBS). kDa, kilodalton. Similar findings were observed in n鈥?鈥? mice/group from two independent experiments. c鈥?b>e Representative images of CD31+ capillary plexus and LYVE-1+ lacteals in the villi of Lats1/2i螖P尾C mice treated with mVR3(4-7)-Ig or mVR3(1-4)-Ig and comparisons of the number of Prox1+ lymphatic endothelial cells (LECs), lacteal sprouts, and lacteal branches of LYVE-1+ lacteals per 100鈥壩糾 of lacteal length in the villi of WT and Lats1/2i螖P尾C mice treated with mVR3(4-7)-Ig or mVR3(1-4)-Ig. White dotted-line outlines the lacteal. Each dot indicates a mean value of 10鈥?0 villi/mouse and n鈥?鈥? mice/group pooled from two independent experiments. Horizontal bars indicate mean 卤 SD and P value versus WT or mVR3(4-7)-Ig by two-tailed Mann鈥揥hitney U test. n.s., not significant. Scale bars, c 100鈥壩糾; d 50鈥壩糾. f, g Representative images and comparison of VE-cadherin+ LEC junctions of LYVE-1+ lacteals in the villi of WT and Lats1/2i螖P尾C mice treated with mVR3(4-7)-Ig or mVR3(1-4)-Ig. Horizontal bars of each colored segment represent mean 卤 SD of 5鈥?0 villi/mouse and n鈥?鈥? mice/group pooled from two independent experiments. ***P鈥?lt;鈥?.0001 versus WT or mVR3(4-7)-Ig treated Lats1/2i螖P尾C mice by two-way ANOVA with Holm鈥揝idak鈥檚 multiple comparisons test. Scale bars, 25鈥壩糾.Full size imageYAP/TAZ-induced VEGF-C secretion by PDGFR尾+ IntSCs could contribute to lacteal maintenance and lipid absorptionTo address the roles of YAP/TAZ in PDGFR尾+ IntSCs, we generated Yap/Tazi鈭哖尾C mouse by crossing a Pdgfrb-Cre-ERT2 mouse with a Yapfl/fl/Tazfl/fl mouse38 (Fig.4a). YAP/TAZ depletion in PDGFR尾+ IntSCs was confirmed by immunohistochemical analysis (Supplementary Fig.9a, b). Notably, lacteal length was decreased by 鈭?/span>35.6% while lacteal width was relatively increased by 鈭?/span>1.3-fold, but there were no other major alterations in the overall structure of the intestinal villi at 3 months after tamoxifen delivery in Yap/Tazi鈭哖尾C mice compared with WT mice (Fig.4b, c). Moreover, the lacteal junctional pattern of Yap/Tazi鈭哖尾C mice exhibited 鈭?/span>83% decreased proportion of button-type, while the proportion of zipper-type was increased by 鈭?/span>4.0-fold compared with WT mice (Fig. 4d, e). However, the junctional pattern of villus blood capillary was not altered (Supplementary Fig.9c), and no remarkable changes in LVs of other organs including small intestinal submucosa, ear skin, tracheal mucosa, and diaphragm were observed (Supplementary Fig.10a鈥揷). Nonetheless, plasma triglyceride concentration was reduced by 29.7鈥?4.1% at 1~3鈥塰 after olive oil gavage in Yap/Tazi鈭哖尾C mice compared with WT mice (Fig.4f). Importantly, Vegfc mRNA was decreased by 44.9%, while no changes in mRNA of Vegfd, Vegfa, Ccbe1, Adamts3, Tgfb, and Ifng were observed in the small intestinal lysates of Yap/Tazi鈭哖尾C mice compared with WT mice (Fig.4g and Supplementary Fig.9d). We also investigated whether lacteal dysfunction in Yap/Tazi鈭哖尾C mice has an effect on dietary lipid absorption in Yap/Tazi鈭哖尾C mice with high-fat diet (HFD) challenge for 8 weeks. While we observed no difference in weight gain during normal chow (NC) diet, Yap/Tazi鈭哖尾C mice gained significantly less weight, exhibited less fat mass and less plasma triglycerides, and had less ectopic fat accumulation in the liver compared with WT mice (Supplementary Fig.11a鈥揼). These findings suggest that Yap/Tazi鈭哖尾C mice have impaired lipid absorption, presumably due to decreased VEGF-C and lacteal dysfunction.Fig. 4: Regression of lacteal and attenuated Vegfc inintestinal villi of Yap/Tazi螖P尾C mice.a Diagram depicting the generation of Yap/Tazi螖P尾C mouse and PDGFR尾+ cell-specific depletion of Yap/Taz in small intestinal villi starting at 6 weeks and their analyses at 3 months after tamoxifen treatment. b, c Representative images of LYVE-1+ lacteals and CD31+ capillary plexus under the E-cadherin+ intestinal epithelial cells and comparisons of the lacteal length, lacteal width, villi length, and villi width in duodenum (DD), jejunum (JJ), and ileum (IL) of the indicated part of small intestine in WT and Yap/Tazi螖P尾C mice. Dots indicate values from 134~159 villi/group from three independent experiments using n鈥?鈥? mice/group. Horizontal bars indicate mean 卤 SD and *** P鈥?lt;鈥?.0001 versus WT by two-tailed Mann鈥揥hitney U test. n.s., not significant. Scale bars, 100鈥壩糾. d, e Representative images and comparison of VE-cadherin+ lymphatic endothelial cell (LEC) junctions of LYVE-1+ lacteals in the villi of WT and Yap/Tazi螖P尾C mice. Arrowheads indicate the button-type (WT) or zipper-type dominant junctional pattern (Yap/Tazi螖P尾C). Horizontal bars of each colored segment represent mean 卤 SD of 5~10 villi/mouse and n鈥?鈥? mice/group pooled from three independent experiments. *** P鈥?lt;鈥?.0001 versus WT by two-way ANOVA with Holm鈥揝idak鈥檚 multiple comparisons test. Scale bars, 20鈥壩糾. f Plasma triglyceride concentration (mg/dl) at the indicated time points after gavage of 200鈥壩糽 of olive oil following fasting for 12鈥塰 in WT and Yap/Tazi螖P尾C mice. Each dot indicates a value from n鈥?鈥? (WT) or n鈥?鈥? (Yap/Tazi螖P尾C) mice pooled from three independent experiments. Horizontal bars indicate mean 卤 SD and P value versus WT by two-way ANOVA with Bonferroni鈥檚 multiple comparisons test. g Comparison of Vegfc mRNA expression in the intestinal villi lysates of WT and Yap/Tazi螖P尾C mice. Fold changes in mRNA expressions relative to the levels of WT mice are presented. Each dot indicates a value obtained from one mouse and n鈥?鈥? mice/group pooled from three independent experiments. Horizontal bars indicate mean 卤 SD and P value versus WT by two-tailed Mann鈥揥hitney U test.Full size imageYAP/TAZ-induced VEGF-C secretion from PDGFR尾+ IntSCs is regulated by mechanical cues and osmotic stressIntestinal villi are constantly exposed not only to high mechanical forces by peristaltic contractility but also to dynamically changing osmolarity during food digestion and absorption1,3,4,34. Of note, villi subepithelial fibroblasts have been suggested to respond to such high strain1,3,4,34. Because YAP/TAZ serve as mechanosensors and mechanotransducers18,19,39, we analyzed whether YAP/TAZ are dictated by the mechanical stimuli for the secretion of VEGF-C in fibroblasts. We mechanically stretched (4% of linear stretch at 10 cycles/min) primary cultured mouse embryonic fibroblasts (MEFs) for 3鈥塰 to mimic the mechanical stimulus induced by intestinal peristalsis. This stimulus induced nuclear localizations of YAP/TAZ and increased the mRNA expressions of Vegfc (~2.0-fold) as well as YAP/TAZ target genes such as Ctgf (~5.7-fold) and Ankrd1 (~2.6-fold) (Supplementary Fig.12a, b). In contrast, high osmolarity (sorbitol 0.4鈥塎) induced cytoplasmic localizations of YAP/TAZ40 and attenuated the expresions of Vegfc (55.4%), Ctgf (80.9%), and Ankrd1 (77.7%) in the MEFs (Supplementary Fig.12c, d). These findings led us to corroborate the molecular interaction between YAP/TAZ and the Vegfc gene. We found several consensus TEAD-binding sequence (GGAAT) in the Vegfc promoter locus (Fig.5a)24. To further validate these regions as TEAD-binding sites, we performed chromatin immunoprecipitation (ChIP) analysis followed by q-PCR with primers encompassing indicated TEAD-binding sequences (Fig.5a). Our ChIP q-PCR results revealed a significant endogenous binding of the TEAD protein near the TEAD-binding sequence (termed R3) found approximately 2.2鈥塳b upstream from the Vegfc transcription start site (Fig.5b).Fig. 5: Mechanical and osmotic stress regulate YAP/TAZ and Vegfc in IntSCs.a Diagram depicting the location of mouse Vegfc promoter element upstream of the transcription start site (ATG) relative to the Vegfc gene with presence and arrangement of consensus TEAD-binding sequences (GGAAT, red lines). ChIP-qPCR primers of Vegfc gene promoter around the TEAD-binding sequences are depicted as R1, R2 or R3. b ChIP-qPCR in MEFs indicating the endogenous interaction between TEAD and the Vegfc promoter located 鈭?.2鈥塳b upstream of the transcription start site of Vegfc gene. Dots indicate values from four independent experiments, and horizontal bars indicate mean 卤 SD. P value versus negative control (NC) by two-tailed Mann鈥揥hitney U test. n.s., not significant. c, d Representative images of YAP and TAZ and comparison of mRNA expressions of Vegfc, Ctgf, and Ankrd1 in isolated IntSCs after treatment with control (DMSO), latrunculin A (1鈥壩糓), or blebbistatin (50鈥壩糓) for 6鈥塰. Similar results were observed in four independent experiments. Scale bars, 50鈥壩糾. Dots indicate data from four independent experiments and horizontal bars indicate mean 卤 SD. P value versus control by two-tailed Mann鈥揥hitney U test. e, f Representative images of YAP and TAZ and comparison of mRNA expressions of Vegfc, Ctgf and Ankrd1 in isolated IntSCs plated on 1.5鈥塳Pa and 28鈥塳Pa matrix. Similar results were observed in four independent experiments. Scale bars, 20鈥壩糾. Dots indicate data from four independent experiments and horizontal bars indicate mean 卤 SD. P value versus 1.5鈥塳Pa matrix by two-tailed Mann鈥揥hitney U test. g, h Representative images of YAP and TAZ and comparison of mRNA expressions of Vegfc, Ctgf and Ankrd1 in isolated IntSCs with or without 0.4鈥塎 sorbitol-induced osmotic stress for 3鈥塰. Similar results were observed in four independent experiments. Scale bars, 50鈥壩糾. Dots indicate data from eight independent experiments and horizontal bars indicate mean 卤 SD. P value versus Control by two-tailed Mann鈥揥hitney U test.Full size imageWe further investigated whether YAP/TAZ are regulated by the mechanical and osmotic stress for VEGF-C secretion in isolated mouse IntSCs. We first disrupted actin-cytoskeleton by treatment with actin polymerization inhibitor (Latrunculin A) and non-muscle myosin inhibitor (Blebbistatin), because mechanical cues are propagated through the actin-cytoskeleton29. Perturbation of F-actin or non-muscle myosin inhibited YAP/TAZ nuclear localization and attenuated Vegfc transcription in isolated IntSCs (Fig.5c, d). In parallel, matrix stiffness enforces different degrees of cell spreading41. Therefore, we seeded isolated IntSCs on soft (1.5鈥塳Pa) or stiff (28 kPa) hydrogels to induce differential mechanical stress. IntSCs seeded on stiff (28 kPa) matrix showed more spread phenotype than those seeded on soft (1.5鈥塳Pa) matrix (Fig.5e). This mechanical cue induced nuclear localization of YAP/TAZ and enhanced mRNA expressions of Vegfc (~2.0-fold), Ctgf (~5.2-fold) and Ankrd1 (~5.6-fold) in IntSCs (Fig.5e, f). Conversely, high osmolarity led to cytoplasmic localization of YAP/TAZ and attenuated expressions of Vegfc (64.7%), Ctgf (60.1%), Ankrd1 (48.1%) in IntSCs (Fig.5g, h). Furthermore, osmotic perturbation into the ex vivo culture of mouse small intestine and in vivo42 promoted cytoplasmic translocations of YAP/TAZ in PDGFR尾+ IntSCs (Supplementary Fig.13a鈥揹). In concordance, while the protein level of VEGF-C substantially increased after mechanical stretch in the supernatant of primary IntSC culture medium from WT mice, VEGF-C level was not significantly altered in the Yap/Taz depleted primary IntSCs (Fig.6a鈥揹 and Supplementary Fig.14). Collectively, our data imply that YAP/TAZ in PDGFR尾+ IntSCs contribute to maintain lacteal integrity and function by dynamically regulating VEGF-C expression in response to mechanical cues and osmotic stress (Fig.6e).Fig. 6: Mechanical stretching induces YAP/TAZ activation and VEGF-C secretion in IntSCs.a Diagram for primary culture of IntSCs derived from WT or Yap/Tazi螖P尾C mice for 2 days and their analyses at 2 days after 100% EtOH or 5鈥壜礛 of hydroxy-tamoxifen (4-OHT) treatment. b Phase-contrast image of freshly isolated primary mouse IntSCs from WT mice without passage (P1). Similar results were observed in four independent experiments. Scale bar, 200鈥壩糾. c Representative images of YAP and TAZ with phalloidin+ actin filaments in unstretched and stretched IntSCs. Similar findings were observed in four independent experiments. Scale bars, 50鈥壩糾. d Comparison of VEGF-C protein concentration in the culture medium of unstretched and stretched IntSCs derived from WT or Yap/Tazi螖P尾C mice. Dots indicate value from four independent experiments and horizontal bars indicate mean 卤 SD. P value versus unstretched by two-tailed Mann鈥揥hitney U test. e Schematic images depicting upregulation and secretion of VEGF-C and other factors upon nuclear translocation and activation of YAP/TAZ and subsequent binding to TEAD after mechanical stimuli in PDGFR尾+ IntSCs.Full size imageSingle-cell RNA sequencing of PDGFR尾+ IntSCs unravels their heterogeneity and identifies distinct subsets of fibroblasts that secrete VEGF-CVEGF-C was suggested to be secreted from the enteric SMCs, which are surrounding the lacteal5. As we observed expression of PDGFR尾 in enteric SMCs of the intestinal villi using tdTomatorP尾C mice (Supplementary Fig.6c), we investigated whether the lacteal phenotypes of Lats1/2i鈭哖尾C mice are induced by the VEGF-C secreted from the surrounding PDGFR尾+ SMCs or PDGFR尾+ non-SMCs. We generated tdTomatorSMC by crossing Myh11-Cre-ERT2 mouse with the Rosa26-tdTomato reporter mouse and confirmed co-distribution of 伪SMA staining on Myh11+ SMCs in the small intestinal villi (Supplementary Fig.15a, b). By generating Lats1/2i鈭哠MC mice by crossing Myh11-Cre-ERT2 mouse with Lats1fl/fl/Lats2fl/fl mouse, we found that lacteal phenotypes were not so remarkable compared with those of Lats1/2i鈭哖尾C mice; still, the alignment and shape of SMCs along the villi were altered and villi structure and capillary plexus were shrunk overall (Supplementary Fig.15c鈥揻). Notably, the increase in the number of lacteal Prox1+ LECs, aberrant lacteal sprouting, and hyperbranching that were observed in Lats1/2i鈭哖尾C mice were not fully recapitulated in Lats1/2i鈭哠MC mice (Supplementary Fig.15g, h). These findings suggest that YAP/TAZ hyperactivation in PDGFR尾+ enteric SMCs alone endows a certain but minor effect on the aberrant lacteal phenotype compared with Lats1/2i鈭哖尾C mice, which are targeting both PDGFR尾+ SMCs and PDGFR尾+ non-SMCs.To delineate the detailed source of VEGF-C among the heterogeneous PDGFR尾+ IntSCs, we performed droplet-based single-cell mRNA sequencing of PDGFR尾+ IntSCs isolated from adult tdTomatorP尾C mice (Fig.7a, b). We retained profiles of 7906 cells with a median of 3,672 detected genes per cell. Unsupervised clustering of the PDGFR尾+ IntSCs revealed seven distinct cell clusters visualized with uniform manifold approximation and projection (UMAP); while all cells expressed high levels of Pdgfrb (Fig.7c).Fig. 7: Single-cell RNA sequencing reveals heterogeneity and distinct subsets of fibroblasts.a Diagram depicting the generation of tdTomatorP尾C mouse and PDGFR尾+ cell-specific expression of tdTomato from 8-week old and their analyses at 10-week old. b Illustration of droplet-based single-cell RNA sequencing (scRNA-seq) of PDGFR尾+tdTomato+ IntSCs sorted from the small intestine of tdTomatorP尾C mice. c Visualization of unsupervised clustering of 7 distinct PDGFR尾+ IntSC clusters by Uniform manifold approximation and projection (UMAP) in the small intestine of tdTomatorP尾C mice. vFB, intestinal villi fibroblast; FB, fibroblast; SMC, smooth muscle cell; MC, mural cell. d Heatmap displaying the scaled expression patterns of top ten differentially expressed genes for random sampled cells (maximum thousand cells) for each indicated clusters. e List of representative marker genes of each of the seven PDGFR尾+ IntSC clusters (left) and violin plots (right) showing the expression of top-ranking marker genes for each cluster. Log normalized read counts as y-axis (normalized expression). f UMAP visualization of unsupervised clustering of seven distinct PDGFR尾+ IntSC clusters in the small intestine of Lats1/2i螖-tdTomatorP尾C mice. g Gene expression levels of Vegfc and Vegfa in tdTomatorP尾C and Lats1/2i螖-tdTomatorP尾C mice projected on UMAP plot. Note that specific subsets of PDGFR尾+ IntSCs (vFB1-3) in the small intestine of Lats1/2i螖-tdTomatorP尾C mice show higher expression of Vegfc compared with those of tdTomatorP尾C mice. h, i Representative images of PDGFR尾+ IntSCs in WT mouse reveal expressions of each fibroblast-specific markers: PAI-1+ vFB1, Serpina3n+ vFB2, P2X1+ vFB3, Ackr4+ or Grem1+ FB4, and PDGFR伪+ Sox6+ FB5. Each white box in the lower left corner is a magnified view. White asterisks indicate lacteals and white arrowheads indicate each fibroblast cluster-specific cell type stained with the indicated marker. Similar findings were observed in n鈥?鈥? mice from two independent experiments. Scale bars, 20鈥壩糾. j Schematic images depicting the anatomic distribution of indicated markers for vFB1-3, FB4, FB5, SMC, MC, capillary plexus, and lacteal in intestinal villi of adult WT mouse. vFB1-3 are uniformly distributed around the lacteal, whereas FB4 is mainly located in the submucosal area and FB5 is mostly placed under the intestinal epithelium. Black dashed boxes are magnified in the right panels.Full size imageWhen we examined differentially expressed genes to characterize each cluster, seven distinct cell clusters exhibited highly cluster-specific gene expression profiles (Fig.7d, e). Importantly, we were able to identify five distinct clusters of PDGFR尾+ fibroblasts, of which three have not been described before (Fig.7c鈥揺). Intestinal villi fibroblast (vFB) 1 expressed high levels of immediate early genes, such as Junb and Fosb. Although expression of immediate early genes could arise as an artifact due to cell dissociation procedures, PDGFR尾+Fosb+ cells were detected in vivo regardless of cellular dissociation.vFB2 shared a similar transcriptomic profile with vFB1 but without the expression of immediate early genes (Fig.7d, e). Of note, matrix metalloproteinase 10 (Mmp10) and four and a half LIM domains protein 2 (Fhl2) genes were exclusively expressed in vFB3, suggesting a role of this cluster in extracellular matrix remodeling (Fig.7d, e). FB4 discretely expressed atypical chemokine receptor 4 (Ackr4) (Fig.7d, e), which has been reported to be expressed in submucosal fibroblasts43. Finally, FB5 exclusively expressed high levels of SRY-Box 6 (Sox6) and bone morphogenic protein 7 (Bmp7), which have been shown to be expressed in fibroblasts lining the epithelium of the small intestinal villi44. Of the remaining 2 clusters, one expressed the hallmarks of enteric SMCs including calponin 1 (Cnn1) and actin gamma 2 (Actg2), while the other cluster expressed vitronectin (Vtn) which marks mural cells (Fig.7e).We also generated a single cell library from Lats1/2i螖-tdTomatorP尾C mice (KO_) to delineate the villi PDGFR尾+ cells that secrete VEGF-C upon YAP/TAZ hyperactivation, and found seven distinct PDGFR尾+ IntSC clusters as observed in tdTomatorP尾C mice (Fig.7f and Supplementary Fig.16a, b). For comparative analysis on the transcriptomic alterations among clusters, we integrated the datasets of tdTomatorP尾C and Lats1/2i螖-tdTomatorP尾C mice, each composed of two mice per group, by multiple dataset integration methods in Seurat and performed unsupervised clustering (Supplementary Fig.16c). When we measured the correlation of clusters from both datasets, each of the seven distinct PDGFR尾+ IntSCs demonstrated highly correlative expression of marker genes, with no significant changes in cell type abundances (Supplementary Fig.16c, d). Of special note, KO_vFB1-3 populations demonstrated robust upregulation of Vegfc, but not Vegfa, upon YAP/TAZ hyperactivation regardless of the duration after the tamoxifen delivery (Fig.7g and Supplementary Fig.16e). Moreover, gene ontology suggested possible discrete functions in vFB1-3 (Supplementary Fig.16f). These results indicate that distinct subsets of intestinal fibroblasts (vFB1-3) upregulate Vegfc upon YAP/TAZ hyperactivation.To define the location of these newly identified subsets of fibroblasts, we examined and validated the expressions of markers that are specific to each cluster (Fig.7h, i). Of note, vFB1-3 were distributed in close proximity to the lacteal, while Ackr4+ FB4 and Sox6+ FB5 were localized around the submucosal and subepithelial regions, respectively (Fig.7h, i), as previously reported12,43. vFB1-3 were distinguished by specific markers: P2X1 was exclusively expressed by vFB3 but not by vFB1-2; Fosb was specifically expressed by vFB1 but not by vFB2 (Fig.8a, b). In addition, through single-molecule in situ hybridization (smFISH) of Vegfc, we observed that the expression of Vegfc is upregulated in Lats1/2i螖-tdTomatorP尾C mice compared with tdTomatorP尾C mice (Fig.8c). Moreover, we performed smFISH of Vegfc with immunofluorescence staining of vFB1-3-specific markers. Vegfc expression increased in vFB1-2 (Shisa3+PDGFR尾+ cells) and vFB3 (P2X1+PDGFR尾+ cells) in Lats1/2i鈭哖尾C mice compared with WT mice (Fig.8d). Taken together, our data unravel the heterogeneity and new subtypes of IntSCs (Fig.7j) and suggest that vFB1-3 maintain lacteal integrity and function by forming a YAP/TAZ-induced VEGF-C secreting niche.Fig. 8: vFB1-3 are distinct cell types secreting VEGF-C.a Representative images of vFB3 (P2X1+Serpina3n+PDGFR尾+) and vFB1-2 (P2X1-Serpina3n+PDGFR尾+) in WT mouse. Similar findings were observed in n鈥?鈥? mice from two independent experiments. White and yellow dotted boxes are magnified in the upper right and lower right panels, respectively. Red and yellow arrowheads indicate vFB3 and vFB1-2, respectively. Scale bar, 20鈥壩糾. b Representative images of vFB1 (Serpina3n+Fosb+Shisa3+PDGFR尾+) and vFB2 (Serpina3n+Fosb-Shisa3+PDGFR尾+) in WT mouse. Similar findings were observed in n鈥?鈥? mice from two independent experiments. White and yellow dotted boxes are magnified in the upper right and lower right panels, respectively, for each image. White and yellow arrowheads indicate vFB1 and vFB2, respectively. Scale bars, 20鈥壩糾. c Representative images of Vegfc single-molecule fluorescence in situ hybridization (smFISH) in intestinal villi of tdTomatorP尾C and Lats1/2i螖- tdTomatorP尾C mice. Similar findings were observed in n鈥?鈥? mice/group from two independent experiments. White dotted boxes are magnified in the right panel. Scale bars, 20鈥壩糾. d Representative images of Vegfc smFISH in vFB1-3 of WT and Lats1/2i螖- tdTomatorP尾C mice. Similar findings were observed in n鈥?鈥? mice/group from two independent experiments. White and yellow dotted boxes are magnified in the upper right and lower right panels, respectively, for each image. Scale bars, 20鈥壩糾.Full size imageDiscussionOur most intriguing finding is that intestinal villi fibroblasts regulate lacteal integrity through YAP/TAZ-induced secretion of VEGF-C upon mechanical cues. Taking advantage of single-cell RNA sequencing and detailed 3D analysis of IntSCs, we further reveal the heterogeneity of villi fibroblasts and characterize their distinct transcriptomic features with unique distributions in the intestinal villi.The molecular signature and functional properties of lymphatic vessels are highly dependent on the stroma where they reside and VEGF-C/VEGFR3 signaling pathway plays major role in lymphangiogenesis2,3,45. SMCs may be an important source of VEGF-C, which is required to maintain lymphatic vessel architecture in the intestine5. Nevertheless, extrinsic factors governing lacteal maintenance and function require better delineation. Hypersprouting and hyperbranching phenotypes were strikingly found in lacteals after YAP/TAZ hyperactivation in PDGFR尾+ IntSCs. This was due to the paracrine activation of VEGFR3 on neighboring LECs, triggered by high VEGF-C secretion from PDGFR尾+ fibroblast subpopulations. A recent report suggested that lacteal LEC junction zippering and disassembly of cytoskeletal VE-cadherin anchors prevent dietary fat uptake4,5,6,7. In our experiments, we observed zippering of lacteal junctions and impaired dietary fat uptake with increased or decreased VEGF-C after hyperactivation or depletion of YAP/TAZ, respectively, in PDGFR尾+ IntSCs. Thus, our results indicate that VEGF-C/VEGFR3 signaling is important for maintaining the button-like lacteal junction for proper dietary fat uptake, together with VEGF-A/VEGFR2 signaling4,5,6,7.We also demonstrated the importance of mechanical cues in regulating nucleocytoplasmic shuttling of YAP/TAZ in fibroblasts, for VEGF-C secretion possibly along with other growth factors and cytokines. Owing to their unique circumstances, intestinal lymphatic capillaries as well as surrounding IntSCs must withstand high levels of repetitive mechanical stress, extreme osmotic variations, and commensal bacteria1,2,3. To support lacteal integrity and function in such environments, some mechanisms must exist for IntSCs to adapt and respond to various types of stress. We showed that YAP/TAZ respond to mechanical or osmotic stress by promoting or attenuating VEGF-C secretion in fibroblasts, enabling them to sense and adapt to high mechanical forces and osmotic stress. Collectively, these findings point to an important link between YAP/TAZ mechanotransduction and biological outcomes in the intestine, likely dictated by the physiology of the gastrointestinal tract.Our analysis uncovering the five transcriptionally distinct subsets of intestinal fibroblasts highlights the heterogeneity and novel role of intestinal fibroblasts in maintaining lacteal homeostasis. Accumulating evidence continues to indicate the importance and diverse tissue-specific roles of perivascular fibroblasts in normal physiology, as well as during tissue remodeling after injury14,16,46. In agreement, fibroblasts are not only a mechanical scaffold for capillaries but also implicated as an important source of endothelial growth factors that mediate capillary morphogenesis and vessel maintenance47. Therefore, the intimate crosstalk between fibroblasts and capillaries seems necessary to support the homeostasis of the organ they pervade in a tissue-specific manner47. However, little is understood regarding IntSC structural and functional heterogeneity and their interaction. We have uncovered the molecular identity of intestinal fibroblasts and revealed their unique anatomic locations, which reflect their discrete functions. Of note, ACKR4+ fibroblasts in the villi stroma, equivalent to the FB4 subtype here, reside exclusively in the intestinal submucosa and are suggested to control leukocyte migration43. Indeed, FOXL1+ fibroblasts in the intestinal subepithelial compartment, which correspond to the FB5 subtype in this work, have been suggested as an essential source of Wnt protein for gut epithelial renewal12.Most importantly, we identified novel distinct subsets of fibroblasts (FB1鈥?), which were distributed adjacent to the lacteals and are potential regulators of lacteal integrity and function by YAP/TAZ-induced VEGF-C secretion. We also have suggested cell type relationships among FB1鈥? and their possible functions. Many evidences suggest the importance and multiple roles of intestinal fibroblasts and mesenchymal stem cells in maintaining intestinal homeostasis and in immunity and disease48,49. However, more extensive work is needed to understand the phenotypic and functional heterogeneity within IntSCs. Our study opens the way for using Cre-expressing reporter and fate-mapping mouse lines to reveal lineage relationships and their functions of novel IntSC subtypes. This work will contribute to a better characterization of populations associated with maintaining intestinal homeostasis or mediating intestinal dysfunction.Taken together, our findings advance understanding of IntSCs and reveal a previously unknown role for villi fibroblasts as essential regulators of lacteal integrity. We highlight unique features of organ-specific lymphatic capillaries that coordinates with the surrounding fibroblasts in the intestinal stroma to accomplish functional interplay between YAP/TAZ and lymphatic maintenance. However, the resulting phenotypes of genetically modified mice and the experimental tools used to induce mechanical or osmotic stress are more extreme than those observed in physiological conditions. Nevertheless, our results reinforce the concept that the paracrine molecules secreted from adjacent fibroblasts are the major regulators of lymphangiogenesis and lymphatic remodeling. These findings fill in parts of the novel dynamic regulatory mechanism of lacteals and offer clues to the process of maintaining intestinal homeostasis.MethodsExperimental animalsAll animal care and experimental procedures were complied with all ethical regulations for animal research and testing under the approval from the Institutional Animal Care and Use Committee (No. KA2016-12) of Korea Advanced Institute of Science and Technology (KAIST). Pdgfrb-Cre-ERT2 30, Lats1fl/fl/Lats2fl/fl31,32 and Yapfl/fl/Tazfl/fl 38 mice were transferred, established, and bred in specific pathogen-free (SPF) animal facilities at KAIST. C57BL/6鈥塉, R26-tdTomato and Myh11-Cre-ERT2 mice were purchased from the Jackson Laboratory. All mice were maintained in the C57BL/6 background and fed with free access to a standard diet (PMI LabDiet) and water. In order to induce Cre activity in the Cre-ERT2 mice, 2鈥塵g of tamoxifen (Sigma-Aldrich) dissolved in corn oil (Sigma-Aldrich) was injected intraperitoneally (i.p.) at indicated time points for each experiment. Cre-ERT2 negative but flox/flox-positive mice among the littermates were defined as control (WT) mice for each experiment. Mice were anesthetized with i.p. injection of a combination of anesthetics (80鈥塵g/kg ketamine and 12鈥塵g/kg of xylazine) before being sacrificed.Histological analysesFor whole-mount staining of small intestine, transcardial perfusion was performed with 2% paraformaldehyde (PFA) after anesthesia. Small intestine was harvested and cut longitudinally to expose the lumen. After several washes with PBS, intestines were pinned on silicon plates. Samples were then post-fixed at 4鈥壜癈 in 4% PFA for 2鈥塰. Samples were washed with PBS several times and were subsequently dehydrated with 10% sucrose in PBS for 2鈥塰 and with 20% sucrose, 10% glycerol in PBS overnight. Ear skin, trachea and diaphragm were harvested without transcardial perfusion and fixed in 4% PFA for 1鈥塰 at 4鈥壜癈. Samples were washed with PBS several times before blocking. After blocking with 5% goat or donkey serum in 0.5% Triton-X 100 in PBS (PBST) for 1鈥塰, samples were incubated with the indicated primary antibodies diluted in the blocking solution at 4鈥壜癈 overnight. After several washes with PBST, the samples were incubated for 2鈥塰 at room temperature (RT) with the indicated fluorocrome-conjugated secondary antibodies diluted in the blocking buffer. The samples were washed with PBST and nuclei were stained with DAPI (Invitrogen). After washing with PBS, samples were mounted with Vecta-shield (Vector Laboratories). Images were obtained using Zeiss LSM 800 or LSM 880 confocal microscope (Carl Zeiss).The following primary and secondary antibodies were used in the immunostaining: anti-LYVE-1 (rabbit polyclonal, 11-034, Angiobio, 1:400); anti-CD31 (rat monoclonal, 557355, BD Biosciences, 1:400); anti-CD31 (hamster monoclonal, MAB1398Z, Merck, 1:400); anti-E-cadherin (goat polyclonal, AF748, R D, 1:200); anti-Prox1 (goat polyclonal, AF2727, R D, 1:400); anti-Prox1 (rabbit polyclonal, 102-PA32AG, ReliaTech, 1:400); anti-VE-cadherin (goat polyclonal, AF1002, R D, 1:200); anti-PDGFR尾 (rat monoclonal, ab91066, Abcam, 1:200); anti-PAI-1 (mouse monoclonal, sc-5297, Santa Cruz, 1:100); anti-Serpina3n (Goat polyclonal, AF4709, R D, 1:200); anti-Fosb (rabbit monoclonal, 2251, Cell Signaling Technology, 1:400); anti-Shisa3 (rabbit polyclonal, TA320118, Origene, 1:400); anti-P2X1 (rabbit polyclonal, APR-001, Alomone labs, 1:800); anti-Ackr4 (rabbit polyclonal, SAB4502137, Sigma-Aldrich, 1:200); anti-Grem1 (goat polyclonal, AF956, R D, 1:200); anti-Sox6 (rabbit polyclonal, ab30455, Abcam, 1:400); anti-PDGFR伪 (goat polyclonal, AF1062, R D, 1:200); anti-YAP (rabbit monoclonal, 14074, Cell signaling, 1:200); anti-TAZ (rabbit polyclonal, HPA007415, Sigma-Aldrich, 1:200); anti-伪SMA, Fluorescein Isothiocyanate (FITC)-conjugated (mouse monoclonal, F3777, Sigma-Aldrich, 1:1000); anti-VEGFR3 (goat polyclonal, AF743, R D, 1:200); anti-VEGFR2 (goat polyclonal, AF644, R D, 1:200); anti-PGP9.5 (rabbit monoclonal, 13179, Cell signaling, 1:400); anti-F4/80, FITC-conjugated (rat monoclonal, 1231007, Biolegend, 1:200); anti-CD3 (hamster monoclonal, 553058, BD Biosciences, 1:1000); anti-Desmin (rabbit polyclonal, AB907, Millipore, 1:400); and Alexa Fluor 488-, Alexa Fluor 594-, Alexa Fluor 647-conjugated anti-rabbit, anti-rat, anti-goat, anti-hamster secondary antibodies (diluted at a ratio of 1:1000) were purchased from Jackson ImmunoResearch. All the antibodies used in our study were validated for the species and applications.Single-molecule fluorescence in situ hybridization (smFISH)For smFISH of small intestine, transcardial perfusion was performed with 2% PFA after anesthesia. Samples were post-fixed at 4鈥壜癈 in 4% PFA for 2鈥塰. Samples were washed with PBS several times, moved to 30% sucrose, and incubated overnight. The samples embedded in OCT were sectioned, and we performed the smFISH according to the manufacturer鈥檚 protocol that is described in the RNAscope Fluorescent Multiplex Assay kit (320850, ACDBio). The RNA scope Probe (ACDBio)-Mm-vegfc (492701-C2) was used for smFISH. The following primary and secondary antibodies were used in the immunostaining with smFISH: anti-PDGFR尾 (rat monoclonal, ab91066, Abcam); anti-Shisa3 (rabbit polyclonal, TA320118, Origene); anti-P2X1 (rabbit polyclonal, APR-001, Alomone labs) and Alexa Fluor 488-, Alexa Fluor 594-, Alexa Fluor 647-conjugated secondary antibodies were purchased from Jackson ImmunoResearch. Images were obtained using Zeiss LSM 800 or LSM 880 confocal microscope (Carl Zeiss).EdU incorporation assay for proliferating LECsTo detect proliferating LECs in lacteal, 5鈥塵g of 5-ethynyl-2鈥?deoxyuridine (EdU, A10044, Invitrogen) was dissolved in 1鈥塵l of Milli-Q water as a stock solution. Then, 200鈥壩糽 of the stock solution per mice was injected i.p. every other day for a week before analysis. Small intestine was isolated and processed as described above. EdU-incorporated cells were detected with the Click-iT EdU Alexa Fluor-488 Assay Kit (Invitrogen) according to the manufacturer鈥檚 protocol.Electron microscopyTo capture ultrastructure electron microscopic images of lacteal, small intestine was sectioned after transcardial perfusion with 4% PFA and 0.25% glutaraldehyde in 0.1鈥塎 phosphate buffer (pH 7.4). Samples were then fixed overnight in 2.5% glutaraldehyde, post-fixed with 1% osmium tetroxide, and dehydrated with a series of increasing ethanol concentrations followed by resin embedding. 70-nm ultrathin lacteal sections were obtained using an ultramicrotome (UltraCut-UCT, Leica), which were then collected on copper grids. Samples were imaged with transmission EM (Tecnai G2 Spirit Twin, FEI) at 120鈥塳V after staining with 2% uranyl acetate and lead citrate.Intravital imaging of lipid clearance from lamina propriaMice were anesthetized following food removal for 12鈥塰 before the procedure. Intravital imaging was performed as we previously described34. Briefly, BODIPY lipid probes (0.1鈥塵g/ml, Thermo Fisher) were dissolved in 2.5% DMSO solution (Sigma-Aldrich). Anti-LYVE-1 (rat monoclonal, 223322, R D) conjugated with Alexa Fluor 647 (Invitrogen) antibody (0.75鈥塵g/kg) was intravenously injected at 12鈥塰 before the imaging for fluorescent labeling of lacteals. Then, the proximal jejunum was exteriorized in an imaging chamber where the temperature was maintained at 37鈥壜癈 during the imaging. After surgical opening of the intestinal lumen 1.5鈥塩m along the anti-mesenteric border, a cover glass was placed on the exposed lumen to obtain a villi view. The intravital imaging was done over three sessions. The villi were first imaged before the BODIPY-FA supply at 0鈥塵in. Then, 30鈥壩糽 of BODIPY lipid probes were supplied once before the second imaging session to observe the initial BODIPY lipid absorption at 1鈥塵in. Next, the BODIPY lipids were supplied three times with 2鈥塵in intervals before the third imaging session at 26鈥塵in, and BODIPY-FA clearing through lacteals was analyzed at 36鈥塵in and 46鈥塵in. A home-built video-rate laser-scanning confocal microscope was used34. Two continuous-wave lasers emitting at 488鈥塶m (MLD, Coherent) and 640鈥塶m (Cube, Coherent) were used as excitation sources for fluorescence imaging. Two bandpass filters (FF01-525/50 and FF01-685/40, Semrock) were used for detection of fluorescent signals. Axial resolution below 4 渭m was acquired with 100 渭m pinhole and 60x objective lens (LUMFLN, water immersion, NA 1.1, Olympus). Images (512鈥壝椻€?12 pixels) were obtained at a frame rate of 30鈥塇z. For the signal-to-noise ratio enhancement of the image, the noise over 90 frames after post-processing the real-time images (30 frames/sec) were averaged by removing the motion artifact generated from peristalsis with a custom-written MATLAB program. ImageJ software (NIH) was used for measuring the average fluorescent intensity in the lamina propria.Lipid absorption testFollowing food removal for 12鈥塰, 200鈥壩糽 of olive oil (Sigma-Aldrich) was delivered by oral gavage for plasma triglyceride measurement and Oil Red O staining. Blood was sampled via the tail vein into the blood collection tube containing heparin at 0, 1, 2, 3, and 4鈥塰 after olive oil administration. Plasma triglyceride concentration was measured using a VetTest Chemistry analyzer (IDEXX Lab). Small intestine was stained with Oil red O at 12鈥塰 after the olive oil gavage following the manufacturer鈥檚 protocol of the Oil Red O Staining Kit (MAK194, Sigma-Aldrich). High fat diet (HFD) (60% kcal fat, Research Diets, D12492) feeding started at 12-week old and continued for 8 weeks for plasma triglyceride measurement and Oil Red O staining. Blood was sampled via the tail vein into the blood collection tube containing heparin after 6鈥塰 fasting. For Oil red O staining of liver sections, liver harvested, fixed in 4% PFA overnight at 4鈥壜癈, were cut to 100鈥壩糾 vibratome sections (Leica, VT1200S), were stained with Oil red O.Transduction of AAVIndicated mice received a single i.p. dose (1012 viral particles in 150鈥?00鈥壜祃) of a recombinant AAV encoding the ligand binding domains 1鈥? of VEGFR3 fused to the IgG Fc domain (AAV鈥搈VEGFR31-4-Ig), and control mice received the same dose of AAV encoding the domains 4鈥? of VEGFR3, which do not bind VEGF-C or VEGF-D (AAV鈥搈VEGFR34-7-Ig), fused to the IgG Fc domain, as previously described37. AAV鈥搈VEGFR31-4-Ig and AAV鈥搈VEGFR34-7-Ig were detected in serum by immunoblotting analysis (anti-VEGFR3 antibody; goat polyclonal, AF743, R D) of 0.5鈥壩糽 of serum samples collected at the same day of the intestinal analysis.Treatment of antibodyFor VEGFR2 blockade, we used a VEGFR2 neutralizing antibody (DC101). A hybridoma cell line that produces DC101 was purchased from American Type Culture Collection (ATCC). Hybridoma cells were grown in serum鈥恌ree medium. Recombinant proteins in the supernatants were purified by column chromatography with Protein A agarose gel (Oncogene). After purification, the recombinant proteins were quantified using the Bradford assay and confirmed by Coomassie blue staining after sodium dodecyl sulfate (SDS)鈥損olyacrylamide gel electrophoresis (PAGE). DC101 (50鈥塵g/kg) or an equal amount of the control antibody (IgG-Fc, SAB3700546, Sigma-Aldrich) was i.p. injected into the indicated mice every 2 days for 2 weeks.Intestinal LEC enrichment and IntSC isolation from the small intestineThe serosa, muscle layer, and mesenteric adipose tissues were removed from the small intestine. After opening gut fragments longitudinally and washing with PBS, samples were cut into 2鈥塩m pieces, and Peyer鈥檚 patches were removed. Samples were washed with PBS and incubated with 10鈥塵M EDTA with calcium- and magnesium-free DMEM (Gibco) on ice for 20鈥塵in. Tissues were vortexed with calcium-free PBS until obtaining a clear supernatant that was devoid of epithelial cells. Sample pieces were cut further into 1鈥塵m fragments and dissociated with dissociation buffer containing 2鈥塵g/ml collagenase II (Worthington), 1鈥塵g/ml Dispase (Gibco), and 1 U/ml DNase (Invitrogen) in DMEM at 37鈥壜癈 for 30鈥塵in. To help dissociation, tissues pieces were gently pipetted up and down every 10鈥塵in. The samples were then filtered through a 70 渭m strainer to mechanically disaggregate the remaining intestinal fragments. After collecting the supernatants, an equal volume of DMEM containing 10% fetal bovine serum (FBS) was added. After centrifugation, cells were resuspended in PBS. To enrich the stromal cell fraction and LECs, hematopoietic cells and epithelial cells were depleted using AutoMACS (Miltenyi) after incubation for 15鈥塵in on ice with anti鈥怌D45 and anti-CD326 Microbeads (Miltenyi). For IntSC isolation, anti-CD31 Microbeads (Miltenyi) were added to deplete endothelial cells in addition to the anti-CD45 and anti-CD326 Microbeads.Primary culture of mouse IntSCsFollowing the isolation of IntSCs, cells were cultured with DMEM/F12 containing 10% FBS on a tissue culture plate for 24鈥塰 or 2 days. For drug treatments and immunofluorescence staining, 50,000 cells per well were plated on 4-well Nunc Lab-Tek II chamber slides (Sigma-Aldrich). Drug concentrations are described in the indicated figure legends and treatments lasted 6鈥塰 for immunofluorescence and gene expression analysis. A 35-mm imaging dish with an Elastically Supported Surface (Ibidi) was used for IntSC culture on soft (1.5鈥塳Pa) and stiff (28鈥塳Pa) matrices for 24鈥塰. For induction of cre activity in primary cultured IntSCs derived from WT or Yap/Tazi鈭咶RC mice, cells were treated with 5鈥壜礛 of 4-hydroxy-tamoxifen (4-OHT) in 100% ethanol (EtOH) or 100% EtOH alone for 2鈥塪ays as a control. Culture-expanded monolayer of IntSCs that were validated as PDGFR尾+ were used for experiments.Flow cytometry and cell sortingTo sort PDGFR尾+ IntSCs or intestinal LECs, the enriched fractions went on RBC lysis by suspension in ACK lysis buffer (Gibco) for 5鈥塵in at RT. After blocking Fc纬 receptors with mouse anti-CD16/CD32 (553141, BD Bioscience), cells were incubated for 15鈥塵in with indicated antibodies in FACS buffer (2% FBS in PBS). After several washes, cells were analyzed by FACS Canto II (BD Biosciences) and the acquired data were further evaluated by using FlowJo software (Treestar). Cell sorting was performed with FACS Aria Fusion (Beckton Dickinson). Dead cells were excluded using DAPI (Sigma-Aldrich) staining and cell doublets were systematically excluded. Fluorescence intensity is expressed in arbitrary units on a logarithmic scale, and forward scatter and side scatter are represented on a linear scale. The following antibodies were used for flow cytometry: FITC anti-mouse CD45 (11-0451-85, rat monoclonal, Biolegend); FITC anti-mouse TER-119 (11-5921-85, rat monoclonal, Biolegend); APC anti-mouse CD31 (551262, rat monoclonal, BD Bioscience); and PE/Cy7 anti-mouse Podoplanin (127412, syrian hamster monoclonal, Biolegend).Bulk RNA-sequencingBulk RNA-sequencing of the isolated PDGFR尾+ IntSCs was performed by obtaining the alignment file. Briefly, reads were mapped using TopHat software tool. The alignment file was used to assemble transcripts, estimate their abundances, and detect differential expression of genes or isoforms using cufflinks. The Ingenuity Pathway Analysis tool (QIAGEN) was used to further evaluate the data in the context of canonical signaling and determine whether YAP/TAZ-hyperactivated genes were specifically related to the secretory molecules. The significance of the canonical signaling was tested by the Benjamini鈥揌ochberg procedure, which adjusts the P value to correct for multiple comparisons, and their activation or inhibition was determined with reference to activation z-scores. Cluster analysis and heatmaps were generated using Morpheus (https://software.broadinstitute.org/morpheus/). For GSEA, gene set collections from the Molecular Signatures Database 4.0 (http://www.broadinstitute.org/gsea/msigdb/) were used.MEFs and HDLECsMouse embryonic fibroblasts (MEFs) were isolated from E12.5 embryos as previously described32. Briefly, head, limb and heart tissues were removed and samples were minced with scissors and digested with 0.1% trypsin/EDTA with DNase (1 U/ml, Invitrogen) at 37鈥壜癈 for 20鈥塵in. After digestion and removal of undigested tissues, the cells were spun briefly, plated onto a 10鈥塩m dish, and allowed to grow to subconfluence. MEFs were then maintained in DMEM/F12 containing 10% FBS. Human dermal lymphatic endothelial cells (HDLECs) were purchased from Lonza and cultured in endothelial growth medium (EGM2-MV, Lonza). MEFs and HDLECs were used at passage 2 or 3. Cells were incubated in a humidified atmosphere of 5% CO2 at 37鈥壜癈 and confirmed to be mycoplasma-negative (MycoAlert Detection Kit, Lonza).Spheroid-based sprouting assayLEC spheroids were generated by culturing HDLECs at passage 2 or 3 in culture medium containing 0.25% methylcellulose and incubating overnight as hanging drops36. The spheroids were then collected and embedded in 2鈥塵g/ml collagen type I (Corning), treated with control bovine serum albumin (BSA, Sigma-Aldrich), WISP2 (300鈥塶g/ml, Peprotech), TNFSF15 (50鈥塶g/ml, Peprotech), BDNF (50鈥塶g/ml, Peprotech), EBI3 (100鈥塶g/ml, Peprotech) or ANGPT2 (5鈥壩糶/ml, in-house generated) with or without VEGF-C (200鈥塶g/ml, R D Systems) in Endothelial Cell Basal Medium (PromoCell) containing 1% FBS for 24鈥塰. At the end of the incubation, spheroid was stained with cell tracker (Molecular Probes, 1.5鈥壩糓, 37鈥壜癈, 30鈥塵in). The spheroids were then fixed in 4% PFA for 15鈥塵in at RT and imaged using Cell observer (Carl Zeiss).Quantitative RT-PCRRNA was extracted using RNeasy Micro kit (Qiagen) or Trizol RNA extraction kit (Invitrogen). A total of 1鈥壜礸 of extracted RNA was transcribed into cDNA using GoScript Reverse Transcription Kit (Promega). Quantitative real-time PCR was performed using FastStart SYBR Green Master mix (Roche) and S1000 Thermocycler (Bio-Rad) with the indicated primers. The primers were designed using Primer-BLAST or adopted from previously published studies, which are described in Supplementary Table1. Gapdh was used as a reference gene and the results were presented as relative expressions to control. Primer reaction specificity was confirmed by melting curve analysis. Relative gene expression was analyzed by 螖螖Ct method using the CFX Manager software (Bio-Rad).In vitro stretch and osmotic experimentsMEFs and primary mouse IntSCs were stretched using a mechanical cell stretching instrument (STREX) with a 4% of linear stretch at 10 cycles/min, and osmotic stress was applied with 0.4鈥塎 sorbitol for 3鈥塰 as previously described40,50. Cells were kept in a humidified 5% CO2 incubator at 37鈥壜癈 during all experiments.Ex vivo osmotic experimentsAfter opening the abdominal cavity under anesthesia, jejunum part of small intestine was cut into 2鈥塩m pieces. Gut fragments were opened longitudinally and washed with PBS. Tissues were then cultured in DMEM/F12 including 5% FBS and osmotic stress was applied immediately with 0.4鈥塎 sorbitol in the culture medium for 3鈥塰. Tissues were kept in a humidified 5% CO2 incubator at 37鈥壜癈 during all experiments.Immunofluorescence staining of MEFsMEFs were plated on 8-well Nunc Lab-Tek II chamber slides (Sigma-Aldrich) and fixed with 4% paraformaldehyde for 15鈥塵in at 4鈥壜癈. After several washes with PBS, samples were blocked with 5% goat (or donkey) serum in 0.5% PBST for 30鈥塵in at RT. Cells were incubated with the indicated primary antibodies at 4鈥壜癈 overnight. Bound primary antibodies were detected by incubating with secondary antibodies for 90鈥塵in at RT. The following primary antibodies were used for the cell staining: anti-YAP (rabbit monoclonal, 14074, Cell signaling); anti-TAZ (rabbit polyclonal, HPA007415, Sigma-Aldrich); and Alexa Fluor 488-conjugated anti-phalloidin (A12379, Thermo Fisher) antibodies. Alexa Fluor 594-conjugated secondary antibodies were purchased from Jackson ImmunoResearch. Nuclei were stained with DAPI (Invitrogen) and slides were mounted and imaged as described above.ChIP-qPCRMEFs were fixed with 1% PFA for 10鈥塵in and quenched with glycine. Samples were washed with PBS and lysed with lysis buffer containing 1% SDS. The fixed DNA in the cell lysates was sonicated using Focused-ultrasonicator (Covaris). The cell lysates were centrifuged, and all but 5% (saved for whole cell lysate input control) of the resulting supernatants were diluted with ChIP dilution buffer containing 0.5% Triton X-100. Diluted samples were then incubated overnight at 4鈥壜癈 with anti-TEAD4 antibody (mouse monoclonal, ab58310, Abcam). Then, protein A/G Dynabeads (Thermo Fisher) were added and the samples were incubated for 2鈥塰 at 4鈥壜癈. The beads were isolated with DynaMag-2 (Thermo Fisher), washed with series of ChIP wash buffers: low-salt wash buffer, high-salt wash buffer, LiCl wash buffer, and TE buffer. Samples were eluted in 1% SDS two times for 15鈥塵in each at 65鈥壜癈. The beads were removed and both the eluted material and 5% whole cell lysate input samples were reverse-cross-linked overnight at 65鈥壜癈. After normalizing the pH, samples were incubated for 30鈥塵in with RNase (3鈥壩糶/mL) at 37鈥壜癈 and for 2鈥塰 with proteinase K (20鈥塵g/ml) and glycogen (20鈥塵g/ml) at 55鈥壜癈. DNA was eluted using standard procedures and ChIP-DNA were analyzed by qPCR compared with input samples using the primers listed in Supplementary Table2.ELISASupernatant of primary cultured mouse IntSCs were harvested after stretching, and supernatant was centrifugation for 15鈥塵in. The concentration of VEGF-C in supernatant was measured by enzyme鈥恖inked immunosorbent assay (ELISA) with mouse VEGF-C鈥?/b>specific ELISA kit (CSB-E07361m, Cusabio).Morphometric analysisMorphometric measurements were performed using ImageJ software (NIH), ZEN 2012 software (Carl Zeiss), or Imaris (Bitplane). For lacteal surface area quantification, a threshold was set for the images to be analyzed and absolute villus LYVE-1+ area was measured for each lacteal within the villus. Absolute lacteal length, absolute villi length, and absolute villi width were measured using images of villus E-cadherin+ or CD31+ or LYVE-1+ immunostaining in each random 850鈥壜祄2 fields of the indicated part of the intestine. For lacteal and villi surface area, length, and width quantifications, at least 100 villi were analyzed along the whole small intestine. Quantification for the following parameters in the small intestine were performed in jejunum. Number of lacteal sprouts and number of lacteal branching were measured in 10-20 random LYVE-1+ lacteal. Number of Prox1+ LECs and number of Prox1+EdU+ LECs were counted manually within 100鈥壩糾 length of 10鈥?0 random LYVE-1+ lacteal. Quantification of VE-cadherin+ junctions of lacteal was performed with 脳40 objective lens in lacteals of 5鈥?0 villi per mouse. Briefly, zipper-type junction was defined as continuous junctions at cell鈥揷ell borders with cells that have elongated shape51. Button-like junction was defined as discontinuous junctions that are not parallel with the cell鈥揷ell borders and the oak leaf-like cell shape. Junctions that do not match either pattern were categorized as mixed type. Fluorescence intensity (FI) of BODIPY and its quantification was performed as previously described34. Oil red O area was measured as Oil red O+ area divided by villus area and normalized by the average of those of control mice. Villus smooth muscle cell coverage area was measured as absolute 伪SMA+ coverage area in five 850鈥壜祄2 fields per sample, which was then normalized by the average of those of control mice. To quantify the relative expressions of lacteal VEGFR3, mean FI measured in five 425鈥壜祄2 fields per samples and was presented as relative values divided by its control. To determine blood vessel density, area of VEGFR2 was measured in five 425鈥壜祄2 fields per sample and presented as relative values to its control. The area of CD3+ T cell or F4/80+ macrophage was measured in five 425鈥壜祄2 fields per samples. Lymphatic vessel density was calculated by measuring LYVE-1+ area in five random 425鈥?50鈥壜祄2 fields from whole mount samples and presented as relative values to its control.Droplet-based single-cell RNA sequencingSorted single cells were processed using the 10X Chromium Single cell 3鈥?Reagent Kit v3 (10X genomics) following manufacturer鈥檚 protocols. Briefly, sorted cells were resuspended in PBS with 0.5% BSA and mixed with RT reagent mix and RT primer, which were then added to each channel in 10X chips, targeting 5000 cells. Cells were separated into Gel Beads in Emulsion where RNA transcripts from single cells were barcoded and reverse-transcribed. After cDNA library construction and amplification, cDNA molecules were enzymatically fragmented, end-repaired and A-tailed. After selecting the appropriate size of processed cDNA molecules through double-sized size selection using SPRI beads (Beckman Coulter), they were ligated with an adaptor, and sample-index PCR was performed. After another double-sized size selection using SPRI beads, final library constructs were diluted 10-fold and ran on the Agilent Bioanalyzer High Sensitivity Chip for quality control. Single-cell libraries were then sequenced using Illumina HiSeq-X platform.Pre-processing of single-cell RNA dataRaw sequencing data were first de-multiplexed and mapped to mouse reference genome (mm10) by using Cell Ranger 3.0.2 toolkit from 10X Genomics (http://10xgenomics.com). Using R package Seurat (version 3.0.0)52, raw expression matrices were built using Read10X function. Cells with fewer than 1000 detected genes or higher than 6000 detected genes (considered as potential doublets) were excluded (Supplementary Fig.17a). In addition, cells with high mitochondrial gene associated unique molecular identifier (UMI) counts ( 7% of total) were considered as dead cells and thus excluded. For genes, those expressed by fewer than 3 cells were removed. In addition, a small number of contaminating Ptprc+ immune cells and Pecam1+ Cdh5+ endothelial cells were excluded after initial round of clustering. After removing unwanted cells and genes, UMI counts for each gene for a cell were divided by the total expression of a given cell and multiplied by 10,000 and log-transformed. Then, genes were scaled and centered by regressing out variables such as mitochondrial gene percentage and number of UMIs.Identifying variable genes and dimensionality reductionR package Seurat was used for clustering analysis. First, highly variable genes across dataset were identified using FindVariableFeatures function in Seurat with option: selection.method鈥?鈥夆€渧st鈥? Top 2000 genes with highest variability were selected for downstream analysis. Using the identified variable genes, initial dimensionality reductions were performed using principal component analysis (PCA). JackStraw function was used to determine statistical significance of PCA scores. The most significant 30 principal components (PCs) were used as input for Uniform Manifold Approximation and Projection (UMAP) to reduce the data into two-dimensional space.Cluster analysisIn order to cluster cells by their transcriptional profiles, we performed an unsupervised clustering on top 2000 highly variable genes for all four samples. We first built shared nearest neighbors(SNN) graph on the principal component space by using FindNeighbors function. Then applied louvain algorithm for modularity optimization-based clustering by using FindClusters function. Clustering resolution parameters were set at the point where clusters showed highly distinct transcriptional profiles. Clusterings were independent to the number of genes detected per cell (Supplementary Fig.17b). Cluster-specific Markers were differentially expressed genes for each cluster identified through FindMarkers function in Seurat with the following settings: min.pct鈥?鈥?.3, logfc.threshold鈥?鈥?.25, min.diff.pct鈥?鈥?.15,test.use鈥?鈥夆€滿AST鈥? Identified markers with adjusted P鈥?lt;鈥?.05 were used for further analysis. Since some cell types such as smooth muscle cells and mural cells had well-known markers, their appearance at the top of the marker list for each cluster validated our marker selection processes. To remove unwanted source of variations such as sex, cells were split by presence female specific transcript Xist and integrated. To perform integration of different datasets, we first log normalized each raw expression matrices and identified top 2000 highly variable genes for each dataset. Then, using FindIntegrationAnchors function in Seurat, we identified anchors between datasets. Next, datasets were integrated based on the identified anchors via IntegrateData function in Seurat. Integrated datasets were then scaled and dimensions were reduced for further cluster analysis. Clusters in the integrated datasets were annotated by their expression profiles of genes in the identified maker lists.Gene ontology enrichment analysisGene ontology enrichment analysis on the cluster markers were performed using R package goseq (version 1.34.0)53. Depending on the gene鈥檚 length, weightings for each gene were obtained and the Wallenius approximation was used to identify associated gene ontology terms. Gene ontology terms that have adjusted P鈥?lt;鈥?.05 and biological process categories were selected.StatisticsNo statistical methods were used to predetermine sample size. The experiments were randomized and investigators were blinded to allocation during experiments and outcome analyses. All values are presented as mean 卤 standard deviation (SD). Statistical significance was determined by the two-sided Mann鈥揥hitney U test between two groups or the two-way ANOVA with Holm鈥揝idak鈥檚 multiple comparisons test for multiple-group comparison. Statistical analyses were performed using GraphPad Prism 8.0 (GraphPad Software). Statistical significance was set at P鈥?lt;鈥?.05.Reporting summaryFurther information on research design is available in theNature Research Reporting Summary linked to this article. The source data underlying all main and Supplementary figures are provided as a Source Data file. RNA sequencing data including bulk and single-cell data are available in the National Center for Biotechnology Information鈥檚 Gene Expression Omnibus under accession number GSE124488. All the other materials are available from the corresponding author upon reasonable request. References1.Bernier-Latmani, J. Petrova, T. V. Intestinal lymphatic vasculature: structure, mechanisms and functions. Nature reviews. Gastroenterol. Hepatol. 14, 510鈥?26 (2017).CAS聽Google Scholar聽 2.Aspelund, A., Robciuc, M. R., Karaman, S., Makinen, T. Alitalo, K. Lymphatic system in cardiovascular medicine. Circ. Res. 118, 515鈥?30 (2016).CAS聽 PubMed聽Google Scholar聽 3.Petrova, T. V. Koh, G. Y. Organ-specific lymphatic vasculature: from development to pathophysiology. J. Exp. Med. 215, 35鈥?9 (2018).CAS聽 PubMed聽 PubMed Central聽Google Scholar聽 4.Bernier-Latmani, J. et al. DLL4 promotes continuous adult intestinal lacteal regeneration and dietary fat transport. J. Clin. Investig. 125, 4572鈥?586 (2015).PubMed聽Google Scholar聽 5.Nurmi, H. et al. VEGF-C is required for intestinal lymphatic vessel maintenance and lipid absorption. EMBO Mol. Med. 7, 1418鈥?425 (2015).CAS聽 PubMed聽 PubMed Central聽Google Scholar聽 6.Davis, R. B., Kechele, D. O., Blakeney, E. S., Pawlak, J. B. Caron, K. M. Lymphatic deletion of calcitonin receptor-like receptor exacerbates intestinal inflammation. JCI Insight 2, e92465 (2017).PubMed聽 PubMed Central聽Google Scholar聽 7.Zhang, F. et al. Lacteal junction zippering protects against diet-induced obesity. Science 361, 599鈥?03 (2018).ADS聽 CAS聽 PubMed聽 PubMed Central聽Google Scholar聽 8.Hagerling, R. et al. Distinct roles of VE-cadherin for development and maintenance of specific lymph vessel beds. EMBO J. 37, https://doi.org/10.15252/embj.201798271 (2018).9.Stzepourginski, I. et al. CD34+ mesenchymal cells are a major component of the intestinal stem cells niche at homeostasis and after injury. Proc. Natl Acad. Sci. USA 114, E506鈥揈513 (2017).CAS聽 PubMed聽Google Scholar聽 10.Greicius, G. et al. PDGFRalpha(+) pericryptal stromal cells are the critical source of Wnts and RSPO3 for murine intestinal stem cells in vivo. Proc. Natl Acad. Sci. USA 115, E3173鈥揈3181 (2018).CAS聽 PubMed聽Google Scholar聽 11.Degirmenci, B., Valenta, T., Dimitrieva, S., Hausmann, G. Basler, K. GLI1-expressing mesenchymal cells form the essential Wnt-secreting niche for colon stem cells. Nature 558, 449鈥?53 (2018).ADS聽 CAS聽 PubMed聽 PubMed Central聽Google Scholar聽 12.Shoshkes-Carmel, M. et al. Subepithelial telocytes are an important source of Wnts that supports intestinal crypts. Nature 557, 242鈥?46 (2018).ADS聽 CAS聽 PubMed聽 PubMed Central聽Google Scholar聽 13.Cao, Z. et al. Targeting of the pulmonary capillary vascular niche promotes lung alveolar repair and ameliorates fibrosis. Nat. Med. 22, 154鈥?62 (2016).ADS聽 CAS聽 PubMed聽 PubMed Central聽Google Scholar聽 14.Tallquist, M. D. Molkentin, J. D. Redefining the identity of cardiac fibroblasts. Nat. Rev. Cardiol. 14, 484鈥?91 (2017).PubMed聽 PubMed Central聽Google Scholar聽 15.Shook, B. A. et al. Myofibroblast proliferation and heterogeneity are supported by macrophages during skin repair. Science https://doi.org/10.1126/science.aar2971 (2018).16.Di Carlo, S. E. Peduto, L. The perivascular origin of pathological fibroblasts. J. Clin. Investig. 128, 54鈥?3 (2018).PubMed聽Google Scholar聽 17.Hong, A. W., Meng, Z. Guan, K. L. The Hippo pathway in intestinal regeneration and disease. Nature reviews. Gastroenterol. Hepatol. 13, 324鈥?37 (2016).CAS聽Google Scholar聽 18.Totaro, A., Panciera, T. Piccolo, S. YAP/TAZ upstream signals and downstream responses. Nat. Cell Biol. 20, 888鈥?99 (2018).CAS聽 PubMed聽 PubMed Central聽Google Scholar聽 19.Moya, I. M. Halder, G. Hippo-YAP/TAZ signalling in organ regeneration and regenerative medicine. Nature reviews. Mol. Cell Biol. https://doi.org/10.1038/s41580-018-0086-y (2018).20.Wang, Y. et al. Deletion of yes-associated protein (YAP) specifically in cardiac and vascular smooth muscle cells reveals a crucial role for YAP in mouse cardiovascular development. Circ. Res. 114, 957鈥?65 (2014).CAS聽 PubMed聽 PubMed Central聽Google Scholar聽 21.Kim, J. et al. YAP/TAZ regulates sprouting angiogenesis and vascular barrier maturation. J. Clin. Investig. 127, 3441鈥?461 (2017).PubMed聽Google Scholar聽 22.Neto, F. et al. YAP and TAZ regulate adherens junction dynamics and endothelial cell distribution during vascular development. eLife 7, e31037 (2018).PubMed聽 PubMed Central聽Google Scholar聽 23.Kato, K. et al. Pulmonary pericytes regulate lung morphogenesis. Nat. Commun. 9, 2448 (2018).ADS聽 PubMed聽 PubMed Central聽Google Scholar聽 24.Cho, H. et al. YAP and TAZ negatively regulate Prox1 during developmental and pathologic lymphangiogenesis. Circ. Res. 124, 225鈥?42 (2019).CAS聽 PubMed聽Google Scholar聽 25.Meng, Z., Moroishi, T. Guan, K. L. Mechanisms of Hippo pathway regulation. Genes Dev. 30, 1鈥?7 (2016).CAS聽 PubMed聽 PubMed Central聽Google Scholar聽 26.Halder, G., Dupont, S. Piccolo, S. Transduction of mechanical and cytoskeletal cues by YAP and TAZ. Nat. Rev. Mol. cell Biol. 13, 591鈥?00 (2012).CAS聽 PubMed聽Google Scholar聽 27.Aragona, M. et al. A mechanical checkpoint controls multicellular growth through YAP/TAZ regulation by actin-processing factors. Cell 154, 1047鈥?059 (2013).CAS聽 PubMed聽Google Scholar聽 28.Wada, K., Itoga, K., Okano, T., Yonemura, S. Sasaki, H. Hippo pathway regulation by cell morphology and stress fibers. Development 138, 3907鈥?914 (2011).CAS聽 PubMed聽Google Scholar聽 29.Dupont, S. et al. Role of YAP/TAZ in mechanotransduction. Nature 474, 179鈥?83 (2011).CAS聽 PubMed聽Google Scholar聽 30.Park, D. Y. et al. Plastic roles of pericytes in the blood-retinal barrier. Nat. Commun. 8, 15296 (2017).ADS聽 CAS聽 PubMed聽 PubMed Central聽Google Scholar聽 31.Heallen, T. et al. Hippo pathway inhibits Wnt signaling to restrain cardiomyocyte proliferation and heart size. Science 332, 458鈥?61 (2011).ADS聽 CAS聽 PubMed聽 PubMed Central聽Google Scholar聽 32.Kim, M. et al. cAMP/PKA signalling reinforces the LATS-YAP pathway to fully suppress YAP in response to actin cytoskeletal changes. EMBO J. 32, 1543鈥?555 (2013).CAS聽 PubMed聽 PubMed Central聽Google Scholar聽 33.Baluk, P. et al. Functionally specialized junctions between endothelial cells of lymphatic vessels. J. Exp. Med. 204, 2349鈥?362 (2007).CAS聽 PubMed聽 PubMed Central聽Google Scholar聽 34.Choe, K. et al. Intravital imaging of intestinal lacteals unveils lipid drainage through contractility. J. Clin. Investig. 125, 4042鈥?052 (2015).PubMed聽Google Scholar聽 35.Suh, S. H. et al. Gut microbiota regulates lacteal integrity by inducing VEGF-C in intestinal villus macrophages. EMBO Rep. https://doi.org/10.15252/embr.201846927 (2019).36.Choi, D. et al. Laminar flow downregulates Notch activity to promote lymphatic sprouting. The. J. Clin. Investig. 127, 1225鈥?240 (2017).PubMed聽Google Scholar聽 37.Antila, S. et al. Development and plasticity of meningeal lymphatic vessels. J. Exp. Med. 214, 3645鈥?667 (2017).CAS聽 PubMed聽 PubMed Central聽Google Scholar聽 38.Xin, M. et al. Regulation of insulin-like growth factor signaling by Yap governs cardiomyocyte proliferation and embryonic heart size. Sci. Signal 4, ra70 (2011).PubMed聽 PubMed Central聽Google Scholar聽 39.Humphrey, J. D., Dufresne, E. R. Schwartz, M. A. Mechanotransduction and extracellular matrix homeostasis. Nat. Rev. Mol. cell Biol. 15, 802鈥?12 (2014).CAS聽 PubMed聽 PubMed Central聽Google Scholar聽 40.Lin, K. C. et al. Regulation of Hippo pathway transcription factor TEAD by p38 MAPK-induced cytoplasmic translocation. Nat. Cell Biol. 19, 996鈥?002 (2017).CAS聽 PubMed聽 PubMed Central聽Google Scholar聽 41.Fu, J. et al. Mechanical regulation of cell function with geometrically modulated elastomeric substrates. Nat. Methods 7, 733鈥?36 (2010).CAS聽 PubMed聽 PubMed Central聽Google Scholar聽 42.Tropini, C. et al. Transient osmotic perturbation causes long-term alteration to the gut microbiota. Cell 173, 1742鈥?754 (2018).CAS聽 PubMed聽 PubMed Central聽Google Scholar聽 43.Thomson, C. A. et al. Expression of the atypical chemokine receptor ACKR4 identifies a novel population of intestinal submucosal fibroblasts that preferentially expresses endothelial cell regulators. J. Immunol. 201, 215鈥?29 (2018).CAS聽 PubMed聽 PubMed Central聽Google Scholar聽 44.Walton, K. D. et al. Villification in the mouse: Bmp signals control intestinal villus patterning. Development 143, 427鈥?36 (2016).CAS聽 PubMed聽 PubMed Central聽Google Scholar聽 45.Augustin, H. G. Koh, G. Y. Organotypic vasculature: From descriptive heterogeneity to functional pathophysiology. Science https://doi.org/10.1126/science.aal2379 (2017).46.Vanlandewijck, M. et al. A molecular atlas of cell types and zonation in the brain vasculature. Nature 554, 475鈥?80 (2018).ADS聽 CAS聽 PubMed聽Google Scholar聽 47.Li, Y., Lui, K. O. Zhou, B. Reassessing endothelial-to-mesenchymal transition in cardiovascular diseases. Nat. Rev. Cardiol. 15, 445鈥?56 (2018).PubMed聽Google Scholar聽 48.Owens, B. M. Simmons, A. Intestinal stromal cells in mucosal immunity and homeostasis. Mucosal Immunol. 6, 224鈥?34 (2013).CAS聽 PubMed聽Google Scholar聽 49.Kinchen, J. et al. Structural remodeling of the human colonic mesenchyme in inflammatory bowel disease. Cell 175, 372鈥?86 (2018).CAS聽 PubMed聽 PubMed Central聽Google Scholar聽 50.Lorenz, L. et al. Mechanosensing by beta1 integrin induces angiocrine signals for liver growth and survival. Nature 562, 128鈥?32 (2018).ADS聽 CAS聽 PubMed聽Google Scholar聽 51.Ahn, J. H. et al. Meningeal lymphatic vessels at the skull base drain cerebrospinal fluid. Nature https://doi.org/10.1038/s41586-019-1419-5 (2019).52.Butler, A., Hoffman, P., Smibert, P., Papalexi, E. Satija, R. Integrating single-cell transcriptomic data across different conditions, technologies, and species. Nat. Biotechnol. 36, 411鈥?20 (2018).CAS聽 PubMed聽 PubMed Central聽Google Scholar聽 53.Young, M. D., Wakefield, M. J., Smyth, G. K. Oshlack, A. Gene ontology analysis for RNA-seq: accounting for selection bias. Genome Biol. 11, R14 (2010).PubMed聽 PubMed Central聽Google Scholar聽 Download referencesAcknowledgementsThis study was supported by the Institute for Basic Science funded by the Ministry of Science, ICT and Future Planning, Korea (IBS-R025-D1-2015 to G.Y.K.). We thank Pilhan Kim and Young Seok Ju (KAIST) for helping intravital imaging and bioinformatical analysis; Sungyong Park, Sujin Seo, Jeomil Bae, Duri Choi and Hyun-Tae Kim (IBS) for technical assistances.Author informationAuthor notesThese authors contributed equally: Seon Pyo Hong, Myung Jin Yang, Hyunsoo Cho.AffiliationsCenter for Vascular Research, Institute for Basic Science (IBS), Daejeon, 34141, Republic of KoreaSeon Pyo Hong,聽Hyunsoo Cho,聽Intae Park,聽Hosung Bae,聽Sang Heon Suh聽 聽Gou Young KohBiomedical Science and Engineering Interdisciplinary Program, Korea Advanced Institute of Science and Technology (KAIST), Daejeon, 34141, Republic of KoreaMyung Jin Yang聽 聽Gou Young KohGraduate School of Medical Science and Engineering, KAIST, Daejeon, 34141, Republic of KoreaHyunsoo Cho,聽Sang Heon Suh聽 聽Gou Young KohGraduate School of Nanoscience and Technology, KAIST, Daejeon, 34141, Republic of KoreaKibaek ChoeDepartment of Tissue Morphogenesis, Max Planck Institute for Molecular Biomedicine, and University of M眉nster, D-48149, M眉nster, GermanyRalf H. AdamsTranslational Cancer Biology Program and Wihuri Research Institute, Biomedicum Helsinki, University of Helsinki, Helsinki, 00290, FinlandKari AlitaloDepartment of Biological Sciences, KAIST, Daejeon, 34141, Republic of KoreaDaesik LimAuthorsSeon Pyo HongView author publicationsYou can also search for this author in PubMed聽Google ScholarMyung Jin YangView author publicationsYou can also search for this author in PubMed聽Google ScholarHyunsoo ChoView author publicationsYou can also search for this author in PubMed聽Google ScholarIntae ParkView author publicationsYou can also search for this author in PubMed聽Google ScholarHosung BaeView author publicationsYou can also search for this author in PubMed聽Google ScholarKibaek ChoeView author publicationsYou can also search for this author in PubMed聽Google ScholarSang Heon SuhView author publicationsYou can also search for this author in PubMed聽Google ScholarRalf H. AdamsView author publicationsYou can also search for this author in PubMed聽Google ScholarKari AlitaloView author publicationsYou can also search for this author in PubMed聽Google ScholarDaesik LimView author publicationsYou can also search for this author in PubMed聽Google ScholarGou Young KohView author publicationsYou can also search for this author in PubMed聽Google ScholarContributionsS.P.H. and G.Y.K. conceived and designed the study. S.P.H. performed all mouse work and most of the experiments with contributions from M.J.Y., H.C., H.B., K.C., S.H.S., and I.P. M.J.Y. performed single-cell RNA sequencing, and analyzed single-cell RNA-seq datasets. H.C. performed in vitro ChIP-qPCR experiment. K.C. performed intravital imaging. H.B. analyzed bulk RNA-seq datasets. S.H.S. helped with mouse work and in vivo data analysis. K.A., R.H.A., and D.L. provided mice and reagents, and reviewed the manuscript. H.C. wrote the manuscript with contributions from S.P.H., M.J.Y., I.P. under the supervision of G.Y.K. I.P. proofread and edited the manuscript. S.P.H., M.J.Y., H.C., and G.Y.K. analyzed and interpreted data, and generated figures. S.P.H., M.J.Y., H.C. actively discussed the experiments and the results under the supervision of G.Y.K. All authors discussed the results and commented on the manuscript. G.Y.K. supervised and directed the project.Corresponding authorCorrespondence to Gou Young Koh.Ethics declarations Competing interests The authors declare no competing interests. Additional informationPeer review information Nature Communications thanks the anonymous reviewer(s) for their contribution to the peer review of this work. Peer reviewer reports are available.Publisher鈥檚 note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.Supplementary information CommentsBy submitting a comment you agree to abide by our Terms and Community Guidelines. If you find something abusive or that does not comply with our terms or guidelines please flag it as inappropriate. Sign up for the Nature Briefing newsletter 鈥?what matters in science, free to your inbox daily.

>>> 更多资讯详情请访问蚂蚁淘商城