VE-Cadherin Is Required for Cardiac Lymphatic Maintenance and Signaling

Abstract

Background:

The adherens protein VE-cadherin has diverse roles in organ-specific lymphatic vessels. However, its physiological role in cardiac lymphatics and its interaction with lymphangiogenic factors, has not been fully explored. We sought to determine the spatio-temporal functions of VE-cadherin in cardiac lymphatics and mechanistically elucidate how VE-cadherin loss influences pro-lymphangiogenic signaling pathways, such as adrenomedullin (AM) and VEGF-C/VEGFR3 signaling.

Methods:

Cdh5^flox/flox;Prox1CreER^T2 mice were used to delete VE-cadherin in lymphatic endothelial cells (LECs) across life stages, including embryonic, postnatal and adult. Lymphatic architecture and function was characterized utilizing immunostaining and functional lymphangiography. To evaluate the impact of temporal and functional regression of cardiac lymphatics in Cdh5^flox/flox;Prox1CreER^T2 mice, left anterior descending artery ligation was performed and cardiac function and repair after myocardial infarction was evaluated by echocardiography and histology. Cellular effects of VE-cadherin deletion on lymphatic signaling pathways were assessed by knock-down of VE-cadherin in cultured LECs.

Results:

Embryonic deletion of VE-cadherin produced edematous embryos with dilated cardiac lymphatics with significantly altered vessel tip morphology. Postnatal deletion of VE-cadherin caused complete disassembly of cardiac lymphatics. Adult deletion caused a temporal regression of the quiescent epicardial lymphatic network which correlated with significant dermal and cardiac lymphatic dysfunction, as measured by fluorescent and quantum dot lymphangiography, respectively. Surprisingly, despite regression of cardiac lymphatics, Cdh5^flox/flox;Prox1CreER^T2 mice exhibited preserved cardiac function, both at baseline and following myocardial infarction, compared to control mice. Mechanistically, loss of VE-cadherin leads to aberrant cellular internalization of VEGFR3, precluding the ability of VEGFR3 to be either canonically activated by VEGF-C or non-canonically transactivated by AM signaling, impairing downstream processes such as cellular proliferation.

Conclusions:

VE-cadherin is an essential scaffolding protein to maintain pro-lymphangiogenic signaling nodes at the plasma membrane, which are required for the development and adult maintenance of cardiac lymphatics, but not for cardiac function basally or after injury.

Graphical Abstract

INTRODUCTION

VE-cadherin is a transmembrane adherens protein located at endothelial cell junctions and is central to numerous cellular processes including regulation of endothelial permeability, transduction of mechanical cues, and vasculogenesis and remodeling^1–3. Most widely studied in blood endothelial cells (BECs), the roles of VE-cadherin in lymphatic endothelial cell (LEC) junctional integrity and signaling are being rapidly uncovered. In LECs, VE-cadherin regulates endothelial permeability by either forming permeable button-like junctions or largely impermeable zipper-like junctions, which are selectively positioned throughout the lymphatic vascular system to regulate fluid entry and trafficking within lymphatic vessels⁴. VE-cadherin also exerts a wide variety of indirect effects on LECs by serving as a central component of the endothelial mechanosensory complex, controlling β-catenin and LEC transcription factor expression, and eliciting VEGFR2/VEGFR3 AKT signaling⁵. In BECs, the VE-cadherin transmembrane domain can bind directly to VEGFR2, and loss of a stable VE-cadherin/β-catenin/PI3K/VEGFR2 complex abrogates VEGF-A mediated survival signaling⁶. Although prior studies have also shown that VE-cadherin directly binds VEGFR3⁶, it remains unknown whether VE-cadherin is a requisite factor to support a stabilized lymphangiogenic signaling node in LECs.

Prior work using embryonic deletion of the VE-cadherin (gene=Cdh5) in murine LECs established its essential role in mesenteric valve development and dermal lymphatic maturation^{5, 7}. Consistent with the genetic and physiological heterogeneity of the lymphatic vascular system^{8 9}, loss of lymphatic VE-cadherin exerted a range of tissue-specific effects, with robust phenotypes in intestinal and mesenteric lymphatics, but only minor dilation and fragmentation of dermal ear lymphatics⁷. However, the effects of lymphatic VE-cadherin deletion in cardiac lymphatics were not reported in these studies. The functions of cardiac lymphatics in normal and cardiovascular disease conditions are just beginning to be appreciated, along with their mutli-progenitor developmental origins^10–12. Yet, there remain substantial gaps in our knowledge of cardiac lymphatic signaling pathways that influence their postnatal growth, maintenance, and regeneration. Thus, in this study we aim to characterize the effect of lymphatic VE-cadherin loss on cardiac lymphatic development and maintenance and elucidate the mechanisms by which VE-cadherin maintains essential cardiac lymphangiogenic signaling nodes.

METHODS

Data Availability.

Full methods and all supporting data are available within the Supplemental Material.

Mice.

Cdh5^fl/fl;Prox1CreER^T2;Prox1-GFP mice were gifted from Dr. Joshua Scallan⁵, Univ. South Florida. Both sexes were used in all experiments. Tamoxifen, dissolved in filtered corn oil, (Sigma-Aldrich, Cat. No. T5648–1G) was used to delete VE-cadherin at specified timepoints. Pregnant dams were monitored for the presence of a vaginal mucus plug (E0.5) following mating, and 50 ug/g tamoxifen was administered by oral gavage at the indicated developmental stage. Postnatal deletion of VE-cadherin was achieved by subcutaneously injecting pups at postnatal day 1 (P1) and P3 with 5 μl of 10 mg/ml tamoxifen. Adult deletion of VE-cadherin was achieved by i.p. injection of 50 μg tamoxifen per gram of bodyweight for days 1–3, 5, and 8 in mice that were between 2–4 months of age.

All animal studies were approved by the Office of Animal Care and Use at the University of North Carolina Chapel Hill. All animal experiments were completed under conditions of MTA University of South Florida Agreement Number 18–1285, prior to the expiration date.

Cell Culture.

Primary human dermal lymphatic endothelial cells, isolated from juvenile human foreskin (LECs, PromoCell, Cat. No. C-12216) were used within 5 passages and cultured in Endothelial Cell Growth Medium (EGM-MV2) bullet kit medium (PromoCell) 37˚C under 5% CO₂.

Statistical Analyses.

Statistical analysis was performed with the GraphPad Prism version 8. All data are presented as means ± standard deviation (SD). Detailed description of group sizes, normality tests, (non)parametric testing, statistical tests and post-tests, as well as exact p values for each table, figure and supplemental figure can be found in the statistical supplement. Normality of distribution was assessed by using Shapiro-Wilk test (for n < 8) or D’Agostino-Pearson tests (for n >= 8). Student’s t-test or Mann-Whitney test was used to analyze the differences between two groups when appropriate. In PLA assays, quantification of the number of foci per cell were determined by BlobFinder v3.2. One-way ANOVA with Bonferonni post-test was used to determine statistical significance for multiple comparisons for adrenomedullin transactivation of VEGFR3, or a Two-way ANOVA with Tukey’s multiple comparisons was used to investigate transactivation under shRNA treated cells. Two-way ANOVA with Tukey’s multiple comparisons test was performed to determine statistical significance between genotypes and over time for percent change in paw width for Paw Edema assays. Significance was considered when p<0.05. Two-way ANOVA with a Tukey’s multiple comparison test was used for comparison of p-ERK/t-ERK and p-AKT/t-AKT time course on shScramble or shCDH5 treated LECs. Survival curves were established using the KaplanMeier approach by the GraphPad Prism 8, with log-rank tests applied to appraise the differences between the groups. All tests were two-tailed, and p-values below 0.05 were considered significant.

RESULTS

Embryonic Lymphatic-Specific Deletion of VE-cadherin Results in Hydrops Fetalis with Altered Cardiac Lymphatic Vessel Morphology.

Cardiac lymphatics emerge at E12.5 near the outflow tract and populate the dorsal ventricle by E14.5¹³. The nascent lymphatics sprout from the sinus venosus region, as well as isolated Prox1+ LECs and LEC clusters, during a critical period in the establishment of the cardiac lymphatic network¹⁴. By E18.5, the ventricular cardiac lymphatic network is established¹³. To elucidate the role of VE-Cadherin in developing cardiac lymphatics, Cdh5^flox/flox mice were bred to Prox1CreER^T2 mice, and tamoxifen-induced deletion of VE-cadherin was performed by oral gavage at either E14.5 and E15.5 (Figure 1A) or E13.5 (Supplemental Figure 1A-D) or E10.5 and E13.5 (Supplemental Figure 1 E-N) for analysis of cardiac lymphatics at E18.5. To confirm efficient Cdh5 deletion from cardiac lymphatics, we stained E18.5 thoracic cavity sections for LYVE1 and VE-cadherin and examined both peri-epicardial vessels and diaphragmatic vessels (Supplemental Figure 1L). This analysis revealed LYVE1 positive epicardial lymphatics which are negative for VE-cadherin staining in Cdh5^LEC-KO hearts (Supplemental Figure 1M). In contrast, Cdh5^flox/flox controls showed LYVE1+/VE-cadherin+ peri-epicardial lymphatic vessels. We further noted LYVE1+/VE-cadherin negative lymphatic vessels in the diaphragm of in Cdh5^LEC-KO thoracic cavities (Supplemental Figure 1N). Together this indicates efficient deletion of VE-cadherin, specific to the lymphatic vasculature, in both the heart and diaphragm.

Figure 1:

Embryonic Deletion of VE-cadherin Results in Edematous Embryos with Pleural Fluid and Altered Lymphatic Vessel Tip Morphology A) Tamoxifen injection scheme to delete VE-Cadherin in lymphatics during embryogenesis at E14.5 and E15.5. TM, tamoxifen. B) Representative images of Cdh5^flox/flox and Prox1CreER^T2;Cdh5^flox/flox (abbreviated subsequently as Cdh5^LEC-KO) embryos at E18.5. Scale bar 0.5 mm. TM administered E13.5 via IP. Scale bar 100 μm. Morphometric and gravimetric analysis of E18.5 embryos: C) body weight, D) crown-rump length, E) heart weight, F) ratio of heart weight:crown rump length. Data displayed as mean ± SD (n = 9 for Cdh5^flox/flox, n = 8 for Cdh5^LEC-KO). G) Cartoon depicting embryonic cardiac lymphatic development E13.5-E18.5. H) Embryonic hearts at E18.5 stained with LYVE1 (white). Scale bar 100μm. Dashed white box represents inset location. LYVE1 vessel tips in “Inset” with blunt morphology marked by blue arrows and tips with spiky morphology marked by red arrows. Scale bar 20μm. Quantification of LYVE1 vessel tip morphology: I) number of spiky tips, J) Number of blunt tips, K) total number of tips. Data displayed mean ± SD (n = 4 for Cdh5^flox/flox, n = 4 for Cdh5^LEC-KO). L) Pie chart depicting average percentage of blunt vs. spiky tips for Cdh5^flox/flox and Cdh5^LEC-KO. M) Ratio of spiky tips to blunt tips Data displayed mean ± SD (n = 4 for Cdh5^flox/flox, n = 4 for Cdh5^LEC-KO). An unpaired Student’s t-test with Welch’s correction was used to compare between genotypes (C-F). Mann-Whitney Test was used to compare between genotypes (I-K, and M). *p<0.05 and **p<0.01. Detailed description of group size, normality testing, statistical tests and post-tests, and exact p-values can be found in the statistical supplement.

Expectedly, Cdh5^flox/flox;Prox1CreER^T2 mice (henceforth referred to as Cdh5^LEC-KO) displayed marked whole-body edema (Figure 1B, Supplemental Figure 1B and 1F) or hydrops fetalis, especially nuchal edema, which was confirmed by hematoxylin and eosin (H&E) staining (Supplemental Figure 1C). The edema resulted in significantly greater body weights and crown-rump lengths for Cdh5^LEC-KO embryos compared to controls (Figure 1C, D and Supplemental Figure 1D) or a trend for increased body weights and crown-rump lengths (Supplemental Figure 1G-H). Raw and normalized measurement of heart weights revealed no significant changes in organ size (Figure 1E-F and Supplemental Figure 1J-K).

Whole-mount LYVE-1 staining at E18.5 revealed a largely intact cardiac lymphatic network in Cdh5^LEC-KO hearts compared to controls (Figure 1G, H and Supplemental Figure 1I). However, high-resolution examination of the actively growing lymphatic network revealed a significant change in vessel tip morphology and number of tips (Figure 1I-M). Control Cdh5^flox/flox cardiac lymphatics exhibited an unequal distribution of two distinct vessel tip shapes (Figure 1H, inset): a narrowed, spiky morphology (Figure 1I) or a rounded, blunt morphology (Figure 1J). Cdh5^flox/flox hearts had a significantly higher number of tips in comparison to Cdh5^LEC-KO hearts (Figure 1K). Further these tips were predominantly of spiky morphology (77.27%) rather than blunt (22.73%) (Figure 1L) and significantly outnumbered blunt tips approximately 5:1 when examining the ratio of spiky to blunt tips (Figure 1M). In contrast, Cdh5^LEC-KO hearts displayed a significant reduction in the total number tips (Figure 1K) and a complete reversal in the percentage of spiky (22.73%) and blunt tips (77.27%) and spiky tips, (Figure 1L). Furthermore, the diameter of lymphatic vessels in the Cdh5^LEC-KO hearts appeared thicker than Cdh5^flox/flox controls (Figure 1H and Supplemental Figure 1I). Collectively, these data establish a requirement for lymphatic VE-cadherin expression for normal fluid homeostasis during embryogenesis and for normal branching morphogenesis of the developing cardiac lymphatic network.

Postnatal Deletion of Lymphatic VE-cadherin Causes Regression of the Cardiac Lymphatic Network, with Sustained Cardiac Function.

Cardiac lymphatics continue to grow and mature until age P15¹³, so we used Cdh5^flox/flox;Prox1CreER^T2 mice crossed to a Prox1-GFP reporter strain¹⁵ to delete VE-Cadherin at P1 and P3, with analysis of cardiac lymphatics at various postnatal stages (Figure 2A,B).

Figure 2: Postnatal Deletion of VE-cadherin Leads to Loss of Cardiac Lymphatic Network.

A) Tamoxifen injection scheme to delete VE-Cadherin in lymphatics postnatally at P1 and P3, with analysis at indicated time points. B) Cartoon depicting cardiac lymphatic development E18.5-P15. C) P31 Hearts whole-mount stained for SMA imaged for SMA and Prox1-GFP. Scale bar 200 μm. Arrows point to isolated LECs in the Cdh5^LEC-KO hearts. D) Adult (3 months) hearts whole mount stained for podoplanin to confirm absence of cardiac lymphatic is maintained into adulthood. Scale bar 500 μm. E) Quantification of vessel percentage area in both Cdh5^flox/flox and Cdh5^LEC-KO hearts. Data displayed at Individual points are the averaged vessel percentage area of both the dorsal and ventral sides of each heart. (n=6 Cdh5^flox/flox, n=3 Cdh5^LEC-KO from two litters). F) Echocardiography measurements displaying no statistical difference in function between genotypes, and a trend towards smaller left ventricular (LV) mass, p = 0.0635. Data displayed as mean ± SD (n=4 Cdh5^flox/flox, n=5 Cdh5^LEC-KO from two litters). A Mann Whitney test was used to compare between genotypes (E-F). *p<0.05. Detailed description of group size, normality testing, statistical tests and post-tests, and exact p-values can be found in the statistical supplement.

Consistent with prior studies, we found that Cdh5^LEC-KO mice develop chylous ascites as early as P14 (data not shown), as a result of mesenteric valve deterioration and chyle-enriched lymph leakage into the abdominal cavity^{5, 7}. Predictably, due to distended intestinal lacteals and hindered nutrient absorption⁷, we found that Cdh5^LEC-KO mice were smaller than their Cdh5^flox/flox littermates, as evidenced by smaller body weights, tibia lengths and organ weights (heart, lung, liver and kidney, normalized to either body weight or tibia length, Supplemental Figure 2A,B,D,E,F). Thus, the reduced size of Cdh5^LEC-KO mice is influenced by several factors, including systemic growth restriction due to nutrient malabsorption and previously-reported, organ-specific effects driven by differences in the Hippo/YAP pathway¹⁶. Moreover, early postnatal deletion of VE-cadherin also caused significant lethality in Cdh5^LEC-KO mice, leaving only a small percentage of survivors for further experimentation of cardiac phenotypes (Supplemental Figure 2C).

In these surviving animals, analysis of cardiac lymphatics at one month of age, following P1 and P3 tamoxifen injection, revealed a striking regression of the epicardial lymphatic network, as evidenced by lack of Prox1-GFP+ lymphatic vessels in Cdh5^LEC-KO mice compared to control littermates (Figure 2C). The few remaining cardiac lymphatic vessels were discontinuous and instead, individual Prox1-GFP+ lymphatic endothelial cells could be observed on the epicardial surface (arrows, Figure 2C). Cardiac lymphatic vessel regression with fragmented individual lymphatic endothelial cells was validated on both the dorsal and ventral surfaces of the heart by podoplanin staining (Figure 2D). Similar structural fragmentation of the tail dermal lymphatic capillaries could be appreciated by high-resolution, light sheet imaging accompanied by functional deficits in tail microlymphography analysis (Supplemental Figure 3A,B), further validating systemic regression of lymphatic structure and function with postnatal deletion of VE-cadherin.

Most interestingly, despite the lack of an intact cardiac lymphatic network, we found no evidence of myocardial edema, assessed by gravimetric analysis (Cdh5^flox/flox=0.8801±0.0630, n =8, and Cdh5^LEC-KO=0.8115±0.0597, n = 3). Consistent with the growth restriction described above, M-mode echocardiography measurements indicated that Cdh5^LEC-KO mice show a trend for decreased intraventricular septal distance and left ventricular mass compared to control littermates (Figure 2F, Table 1). Yet surprisingly, the Cdh5^LEC-KO mice displayed normal cardiac function, with no differences fractional shortening or ejection fraction compared to controls (Figure 2E, Table 1). Collectively, these data indicate that that postnatal growth and maintenance of the cardiac lymphatic network is dependent on the junctional integrity offered by VE-cadherin. In addition, an unexpected finding is that loss of the cardiac lymphatic network does not appear to impair overall cardiac contractility and function.

Table 1:

Conscious echocardiography measurements of littermate Cdh5^fl/fl or Cdh5^LEC-KO mice.

	IVS;d mm	IVS;s mm	LVID:d mm	LVID;s mm	LVPW;d mm	LVPW;s mm	LV Mass mg	LV Mass (Corrected) mg	LV Vol;d μl	LV Vol;s μl	EF %	FS %	Heart Rate BPM
P30 after postnatal injection
Cdh5fl/fl	1.003±0.049**	1.637±0.101*	2.752±0.161	1.154±0.102	0.835±0.179	1.543±0.241	80.33±14.70*	64.26±11.76*	28.44±4.048	3.063±0.748	88.97±3.498	57.91±5.119	636.8±17.94
Cdh5LEC-KO	0.746±0.101**	1.434±0.075*	2.796±0.287	1.170±0.194	0.699±0.116	1.465±0.101	57.62±12.17*	46.09±9.74*	29.85±7.430	3.287±1.331	89.00±3.795	58.14±5.664	664.2±65.06
Day 22 after adult injection
Cdh5fl/fl	1.128±0.136	1.831±0.208	2.553±0.205	1.009±0.270	1.107±0.184	1.877±0.270	99.70±25.97	79.76±20.77	23.75±4.977	2.418±1.499	89.92±5.927	60.56±9.887	670.7±30.37
Cdh5LEC-KO	1.115±0.148	1.822±0.148	2.546±0.269	1.025±0.266	1.145±0.202	1.853±0.269	100.6±23.04	80.46±18.43	23.78±6.483	2.504±1.519	89.90±5.371	60.07±8.456	665.9±31.37
Day 43 after adult injection
Cdh5fl/fl	1.144±0.151	1.899±0.204	2.607±0.210	1.103±0.163	1.092±0.125	1.743±0.237	102.2±20.77	81.73±16.62	25.04±4.910	2.802±1.165	88.92±3.399	57.72±4.803	673.1±38.78
Cdh5LEC-KO	1.169±0.133	1.882±0.142	2.594±0.168	1.165±0.163	1.153±0.168	1.866±0.187	107.5±19.85	86.03±15.88	24.63±4.112	3.227±1.262	87.18±3.231	55.21±4.188	656.6±38.99

BPM indicates beats per minutes; d, diastolic; EF, ejection fraction; FS, fractional shortening; IVS, interventricular septal; LV, left ventricle; LVID, left ventricle internal diameter; LVPW, left ventricle posterior wall; and s, systole. Cdh5^fl/fl n=4 and Cdh5^LEC-KO n=5 for P30 timepoint; Cdh5^fl/fl n=20 and Cdh5^LEC-KO n=22 for Day 22 timepoint; and Cdh5^fl/fl n=14 and Cdh5^LEC-KO n=15 for Day 43 timepoint. Data represented as an average ±SD.

* P≤0.05,

**P≤0.01, by unpaired student’s two-tailed t test. Detailed description of group size, normality testing, statistical tests and post-tests, and exact p-values can be found in the statistical supplement.

Adult Deletion of Lymphatic VE-cadherin Causes Regression of the Cardiac Lymphatic Network, with Sustained Cardiac Function.

Since early postnatal deletion of VE-Cadherin results in few surviving animals that are frail and growth-restricted, we chose to administer tamoxifen to adult animals and then evaluate the effects on cardiac lymphatics at 2-weeks and 5-weeks post-injection (Figure 3A,D). First, we noticed that this injection scheme improved overall animal survival. We also confirmed efficient knockdown of Cdh5 mRNA levels in LYVE1+ isolated cardiac lymphatic endothelial cells (Supplemental Figure 4A-B). Podoplanin staining demonstrated that cardiac lymphatics of Cdh5^LEC-KO mice begin to deteriorate 2-weeks (22 days) after deletion of VE-cadherin (Figure 3B), with islands of Prox1+ cells appearing near anatomic locations that commonly have intact lymphatic vessels (Figure 3C, yellow braces in inset), suggestive of a progressive dissolution of the lymphatic network. Indeed, a progressive deterioration of the network was confirmed at 5- weeks (43 days) following the last tamoxifen injection (Figure 3D-E), when genetic Cdh5 knockdown is sustained (Supplemental Figure 4C-D). Interestingly, while islands of LECs were still noted by Prox1-GFP imaging at 5-weeks post tamoxifen, the number of islands was visibly reduced compared to control animals and 2-week Cdh5^LEC-KO mice (Figure 3F).

Figure 3: Loss of VE-cadherin in Adults Leads to Progressive Regression of Cardiac Lymphatics, Compatible with Long-Term Survival.

A) Tamoxifen injection scheme to delete VE-cadherin in lymphatics in adult mice, with analysis at day 22. B) Hearts whole-mount stained with podoplanin. Scale bar 200 μm. C) Hearts fluorescence imaged for Prox1-GFP. Yellow brackets on inset indicate discontinuous Prox1-GFP positive vessels and punctate islands of Prox1-GFP positive LECs. Scale bar 100 μm. D) Tamoxifen injection scheme to delete VE-cadherin in lymphatics in adult mice, with analysis at day 43. E) Hearts whole-mount stained with podoplanin. Scale bar = 200 μm, Scale bar = 80 μm for inset. F) Hearts fluorescence imaged for Prox1-GFP. Yellow brackets on inset indicate islands of Prox1-GFP positive LEC islands, which are reduced in number in comparison to D22. Scale bar = 100 μm, Scale bar = 60 μm for inset. G) Angiotool quantification of vessel percentage area in both Cdh5^flox/flox and Cdh5^LEC-KO at both the D22 and D43 analysis timepoints (Cdh5^flox/flox: D22|n=8, D43|n=8 ; Cdh5^LEC-KO: D22|n=10, D43|n=11). All values are mean ± SD. H) Characterization of the Prox1-GFP+ population of non-myocyte cells by flow cytometry. Hearts were collected from tamoxifen treated Cdh5^flox/flox and Cdh5^LEC-KO at D43, mechanically and enzymatically digested (n=1 per genotype). Cell suspension was run through a 40-um filter to remove cardiomyocytes and remaining cell suspension was labeled with Hoechst, to mark live cells, and analyzed by FACS. Flow cytometry dot plots for GFP displayed; cell populations expressed as percentages (%) of scatter/singlets from digested cardiac tissue. SCC, side scatter; BL1-A::Prox1-GFP-GFP; indicates filter selected for GFP signal. A two-way ANOVA with a Tukey’s multiple comparisons post-test was used to compare vessel percentage area between genotypes and timepoints. **p<0.01, ***p<0.001, and ****p<0.0001. Detailed description of group size, normality testing, statistical tests and post-tests, and exact p-values can be found in the statistical supplement.

Using AngioTool software, we quantified a statistically significantly reduction in cardiac lymphatic coverage in Cdh5^LEC-KO mice when compared to Cdh5^flox/flox mice at both timepoints, and between Cdh5^LEC-KO mice, 2-weeks vs. 5-weeks following tamoxifen injection (Figure 3G). In addition, flow cytometry of cardiomyocyte-depleted cell isolates revealed an approximately 70% decrease in the number of Prox1-GFP+ cells in the non-CM cell population of Cdh5^LEC-KO hearts compared to littermate Cdh5^flox/flox controls (Figure 3H, Supplemental Figure 5).

To determine whether adult Cdh5^LEC-KO mice exhibited functional lymphatic deficits, we first confirmed systemic lymphatic insufficiency using a paw edema assay in response to an inflammatory stimulus and found that Cdh5^LEC-KO mice exhibited significant and long-lasting impairments in edema resolution compared to Cdh5^flox/flox controls (Supplemental Figure 6A,B). Next, to evaluate the localized effects of the regressed cardiac lymphatic network on basal cardiac lymphatic transport at the 2-week time point, we used cardiac lymphangiography¹⁷ (Figure 4A). Briefly, fluorescent Quantum Dots (Qdots) were injected intramyocardially in the apex of Cdh5^flox/flox and Cdh5^LEC-KO hearts, where it is selectively up taken by Prox1-GFP+ lymphatic vessels and shuttled towards draining cardiac lymph nodes in as little as 5 minutes (Figure 4B). Cdh5^flox/flox hearts exhibited robust and continuous uptake of Qdots in multiple cardiac lymphatic vessels, while uptake was largely restricted to a single vessel in Cdh5^LEC-KO hearts. Interestingly, but predictably, the Qdot uptake in the VE-cadherin deficient Cdh5^LEC-KO vessels was discontinuous, often resulting in substantial lymphatic leakage of Qdots (Figure 4B, white bars on inset). Further evidence of basal cardiac lymphatic dysfunction was demonstrated by gravimetric analysis of cardiac water content on apex tissue harvested from Cdh5^flox/flox and Cdh5^LEC-KO hearts, revealing a trend for increased water content in Cdh5^LEC-KO hearts (Figure 4C).

Figure 4: Loss of VE-cadherin leads to basal cardiac lymphatic dysfunction, results in increased infarct size and blunts cardiac lymphangiogenesis after myocardial infarction (MI).

A) Tamoxifen injection scheme to delete VE-cadherin in lymphatics in adult mice, indicating assessment of basal cardiac lymphatic function analysis at day 22, Pre-MI assessment of basal cardiac function by conscious echocardiography and induction of MI by permanent ligation of the left anterior descending artery (LAD) performed at D43. Assessment of Post-MI cardiac function performed routinely between D43 and D63 (Table 2). Hearts collected 20 days-post MI (D63) and histological analysis of the heart and evaluation of lymphatic coverage was performed. B) Cardiac lymphangiography of Cdh5^flox/flox and Cdh5^LEC-KO. Fluorescent quantum dots (Qdot 605, red) are injected intramyocardially in the apex of the heart and selectively taken up by Prox1-GFP (green) superficial cardiac lymphatics and transported towards draining lymph nodes of the heart. Yellow signal indicated Prox1-GFP+ vessels containing Qdot605. Leakage of Qdot from Prox1-GFP+ lymphatics denoted by white bars. Scale bar 100 μm, Scale bar = 60 μm for inset. C) Gravimetric analysis of cardiac water content in uninjured Cdh5^flox/flox and Cdh5^LEC-KO at D117. Portion of heart collected, weighed and dried for 6 days at 65°C. Ratio wet weight to dry weight displayed (n = 4 for Cdh5^flox/flox and n = 4 for Cdh5^LEC-KO). D) Histological Analysis of cardiac sections 20 days post-MI. Left: Hematoxylin-eosin (H&E), Middle: Masson’s Trichrome, Right: Sirus Red. Scale bar = 500 μm for all three staining sets. E) Quantification of Infarct Area using Sirius Red Staining Intensity (n = 6 for Cdh5^flox/flox and n = 5 for Cdh5^LEC-KO). All values are mean ± SD. F) Immunohistochemistry of cardiac sections in infarcted Cdh5^flox/flox and Cdh5^LEC-KO 20 days post-MI (yellow=LYVE1, Magenta=Podoplanin, green=Prox1-GFP, blue=DAPI. Podoplanin positive epicardium denoted with white arrow. Scale bar 1000 μm. A Mann Whitney test was used to compare between genotypes. **p<0.01. Detailed description of group size, normality testing, statistical tests and post-tests, and exact p-values can be found in the statistical supplement.

Echocardiography was performed in Cdh5^LEC-KO and Cdh5^flox/flox adults following both 2-week and 5-week time points. Once again, despite the significant regression and dysfunction of cardiac lymphatics, we observed no significant ventricular dysfunction in Cdh5^LEC-KO mice compared to controls (Table 1). Therefore, these data underscore the importance of lymphatic VE-cadherin expression for the prolonged structural integrity, maintenance, and function of the cardiac lymphatic network. However unexpectedly, these data capitalize on a unique animal model to establish that maintenance of an intact, quiescent adult cardiac lymphatic network is not required for normal cardiac contraction and cardiac function.

Adult Deletion of Lymphatic VE-Cadherin Blunts Lymphangiogenic Response, with Exacerbated Injury and Preserved Cardiac Function after Myocardial Infarction.

Alterations to cardiac lymphatics have been shown to play important roles in diseases such as atherosclerosis and myocardial infarction (MI), where lymphatic dysfunction and alterations to the endothelial barrier can modulate outcomes such as inflammation, fibrosis, edema, and hemorrhage^{11, 17–20}. A growing body of work indicates that lymphatic vessels undergo a reactivation following myocardial infarction where increased vessel coverage and secretion of lymphangiocrine signals (like VEGFC, adrenomedullin and reelin) stimulate resolution of inflammation and edema^{17, 19–21}. The post-MI lymphangiogenic response has also been found to be enhanced by VEGF-C stimulation, resulting in an improvement in cardiac function¹⁹. Thus, we sought to exploit the unique Cdh5^LEC-KO genetic model of regressed cardiac lymphatics to query whether dysfunctional lymphatic vasculature could impact the lymphangiogenic and cardiac repair response following MI, induced by permanent ligation of the left anterior descending (LAD) coronary artery (Figure 4A).

Consistent with their poor basal health, it is worth noting that a larger, but not significant, number of Cdh5^LEC-KO mice died less than 10 days after MI in comparison to Cdh5^flox/flox controls. Histological analysis revealed a significant increase in the size of the infarct and in the extent of fibrosis in Cdh5^flox/flox compared Cdh5^LEC-KO 20 days after LAD ligation (Figure 4D-E). In control Cdh5^flox/flox hearts, we observed robust LYVE1+ cardiac lymphatics in the infarct and border zone, indicative of injury-induced lymphangiogenesis (Figure 4F). In contrast, very few LYVE1+ lymphatics were observed in or around the infarct in Cdh5^LEC-KO hearts. Moreover, these vessels consistently appeared thinner than the cardiac lymphatic vessels of control Cdh5^flox/flox hearts.

Despite the significant increase in infarct area and fibrosis in injured Cdh5^LEC-KO hearts, echocardiography failed to reveal any significant changes in cardiac function compared to injured Cdh5^flox/flox mice. Indeed, no differences in ejection fraction or fractional shortening were found pre-MI or post-MI, measuring by conscious echocardiography over a 20-day period (Table 2). Together these data indicates that although VE-cadherin deficient cardiac lymphatics are unable to mount a normal lymphangiogenic response after MI, even a reduced number of lymphatics (Figure 3D-G) with some modest level of functionality (Figure 4B) may shield against a decrease in cardiac function following injury.

Table 2:

Conscious echocardiography measurements of Cdh5^fl/fl or Cdh5^LEC-KO mice before and after MI.

	Baseline		Day 5		Day 10		Day 15		Day 20
	Cdh5 flox/flox	Cdh5 LEC-KO	Cdh5 flox/flox	Cdh5 LEC-KO	Cdh5 flox/flox	Cdh5 LEC-KO	Cdh5 flox/flox	Cdh5 LEC-KO	Cdh5 flox/flox	Cdh5 LEC-KO
IVS;d, mm	1.029 ± 0.139	1.127 ± 0.106	0.8329 ± 0.2468	1.018 ± 0.3062	0.4325 ± 0.1036	0.6709 ± 0.3096	0.4075 ± 0.09025	0.4922 ± 0.1881	0.3717 ± 0.1031	0.6721 ± 0.3817
IVS;s, mm	1.772 ± 0.1595	1.836 ± 0.1531	0.9571 ± 0.2993	1.258 ± 0.4531	0.5179 ± 0.1490	0.8561 ± 0.4097	0.4326 ± 0.241	0.8064 ± 0.5282	0.4282 ± 0.1237	0.9282 ± 0.6187
LVID;d, mm	2.661 ± 0.2021	2.568 ± 0.3109	3.689 ± 0.7316	3.753 ± 0.7586	4.299 ± 0.9802	4.131 ± 1.054	4.389 ± 1.03	4.46 ± 1.16	4.211 ± 1.075	4.54 ± 1.189
LVID;s, mm	1.149 ± 0.163	1.217 ± 0.3032	2.961 ± 0.8119	2.87 ± 0.8926	3.677 ± 1.117	3.457 ± 1.182	3.668 ± 1.2	3.681 ± 1.178	3.509 ± 1.303	3.722 ± 1.314
LVPW;d, mm	1.004 ± 0.1651	1.147 ± 0.171	1.404 ± 1.152	1.152 ± 0.2337	1.109 ± 0.4686	1.241 ± 0.4554	1.109 ± 0.4399	0.9928 ± 0.3717	1.288 ± 0.498	1.112 ± 0.4776
LVPW;s, mm	1.648 ± 0.05064	1.75 ± 0.1623	1.661 ± 0.4229	1.529 ± 0.2121	1.38 ± 0.616	1.484 ± 0.5702	1.528 ± 0.6904	1.227 ± 0.5895	1.81 ± 0.6659	1.351 ± 0.6944
LV mass, mg	89.54 ± 16.9	101.6 ± 12.5	164.3 ± 37.62	167 ± 71.04	121.8 ± 31.45	152.7 ± 42.65	143.7 ± 21.19	16.1 ± 39.48	170.2 ± 59.42	161.8 ± 42.25
LV mass (corrected), mg	71.63 ± 13.52	81.31 ± 9.997	131.4 ± 30.09	133.6 ± 56.83	97.42 ± 25.16	122.1 ± 34.12	115 ± 16395	100.9 ± 31.58	136.1 ± 47.54	129.4 ± 33.8
LV Vol;d, uL	26.28 ± 5.039	24.39 ± 7.538	60.73 ± 29.31	63.58 ± 31.94	88.53 ± 46.16	82.24 ± 50.62	93.27 ± 51.16	98.65 ± 58.85	85.49 ± 54.4	102.9 ± 59.99
LV Vol;s, uL	3.104 ± 1.059	3.595 ± 2.33	37.53 ± 25.57	35.97 ± 27.05	64.29 ± 43.86	57.59 ± 46.29	65.06 ± 50.01	65.63 ± 48.21	60.52 ± 55.14	69.2 ± 54.55
EF, %	87.95 ± 4.561	85.04 ± 6.764	42.58 ± 11.41	48.87 ± 15.80	32.86 ± 15.29	36.94 ± 15.11	36.45 ± 16.83	37.85 ± 12.60	37.68 ± 17.63	39.06 ± 17.49
FS, %	56.67 ± 6.512	53.06 ± 7.415	20.67 ± 6.098	24.83 ± 9.547	15.71 ± 7.669	17.89 ± 8.107	17.78 ± 8.863	16.48 ± 3.959	18.41 ± 9.303	19.51 ± 10.34
Heart rate, beats/min	674.8 ± 29	659.2 ± 29.72	705.7 ± 36.4	665 ± 45.44	689 ± 60.83	675.4 ± 38.33	702.8 ± 17.77	683.9 ± 34.19	702.4 ± 26.78	657.5 ± 37.17
Cardiac Output (approximate), mL/min	34.42 ± 7.412	31.12 ± 8.815	32.75 ± 6.323	41.57 ± 10.95	35.43 ± 13.19	36.69 ± 10.17	19.71 ± 9.227	22.57 ± 10.12	35.5 ± 4.874	51.24 ± 19.29
n	13	15	7	11	7	10	7	10	7	10

BPM indicates beats per minutes; d, diastolic; EF, ejection fraction; FS, fractional shortening; IVS, interventricular septal; LV, left ventricle; LVID, left ventricle internal diameter; LVPW, left ventricle posterior wall; and s, systole. Data represented as an average ±SD. Detailed description of group size, normality testing, statistical tests and post-tests, and exact p-values can be found in the statistical supplement

VE-cadherin is necessary for VEGFR3 localization and signaling in LECs.

Numerous studies have established that sustained signaling of key lymphangiogenic signaling pathways, such as VEGFC/VEGFR3 and AM/CLR/RAMP2, is required for maintaining lymphatic structural and functional integrity in adults^22–25. Further both VEGF-C and AM are both lymphangiogenic and cardioprotective peptides^{19, 20}. Thus, we hypothesized that the regression of cardiac lymphatics and dermal lymphatic dysfunction in Cdh5^LEC-KO mice could be attributed to the destabilization or internalization of these key lymphangiogenic signaling nodes at the plasma membrane. To test this, we first evaluated the cellular localization of VEGFR3 expression in vivo in both control and Cdh5^LEC-KO hearts. In control Cdh5^flox/flox hearts, we consistently observed strong co-localization of LYVE1 staining with VEGFR3 within lymphatic cells of the cardiac lymphatic network (Figure 5A). On the other hand, within the deteriorated cardiac lymphatic network of Cdh5^LEC-KO mice, we noted markedly reduced lymphatic network density and discontinuous LYVE1+ cells. Moreover, a subset of the LYVE1+ cells also lacked VEGFR3 staining, suggesting a loss of VEGFR3 expression or localization within these cells (arrows, Figure 5A).

Figure 5: Loss of VE-cadherin Results in Internalization of VEGFR3 and Attenuated VEGFR3 Signaling.

A) Hearts whole-mount stained with LYVE1 and VEGFR3. Yellow boxed areas indicate the zoomed in area displayed in the inset panels. Yellow arrows indicate cardiac lymphatics in which VEGFR3 signal is decreased or absent from the indicated lymphatic vessel in the Cdh5^LEC-KO heart. Scale bar = 500 μm, Scale bar = 150 μm for inset. B) Representative western blot analysis of CDH5 and VEGFR3 levels after CDH5 knockdown. C) Quantification of CDH5 knockdown normalized to CDH5 levels in shScramble treated control group from n=3 independent experiments. D) Representative images of VE-cadherin and VEGFR3 cellular localization in confluent LECs which were treated with shScramble or shCDH5. Scale bar 20 μm. E) Representative images of VEGFR3/p-Tyr PLA generated foci in shScramble or shCDH5 treated lymphatic endothelial cells (LECs) which were treated with DMSO or VEGF-C to induce VEGF-C-mediated VEGFR3 phosphorylation. Scale bar 20 μm. F) Quantification of the number of foci per cell. Each point represents average of an experiment, n = 4 independent experiments. G) Proliferation in response to VEGFC stimulation after CDH5 knockdown. Data shown as percentage of Edu positive cells (n = 6 independent experiments for shScramble, n = 5 independent experiments for shCDH5). Each point represents average of an individual experiment. All values are mean ± SD. An unpaired Student’s t-test with Welch’s correction was used to compare between treatment groups. Significance for multiple comparisons was determined by two-way ANOVA with Sidak multiple comparisons test. *p<0.05, **p<0.01, ****p<0.0001. Detailed description of group size, normality testing, statistical tests and post-tests, and exact p-values can be found in the statistical supplement.

To further confirm that loss of VE-cadherin affects the localization of VEGFR3 in lymphatic endothelial cells, we cultured human dermal LECs and treated them with either a lentiviral packaged shRNA directed against VE-cadherin (pLV[shRNA]-EGFP-CDH5-shRNA to be referred to as shCDH5) or a shRNA scramble control (shScramble) and then stained for VE-cadherin and VEGFR3. Western blot analysis revealed that shCDH5 was shown to efficiently knockdown CDH5 by approximately 82% in comparison to shScramble control (Figure 5B-C). Under both treatment conditions, LECs were able to form confluent monolayers. shScramble LECs showed distinct VE-cadherin staining at cell borders, with areas of continuous and discontinuous staining characteristic of lymphatic button and zipper junctions, while shCDH5 LECs predictably showed no VE-cadherin staining at cell borders (Figure 5D). VEGFR3 staining in the confluent monolayer was faint, and mostly localized in the cytoplasm. Despite this low signal intensity, there was a further reduction in VEGFR3 signal intensity in shCDH5 LECs compared to shScramble LECs (Figure 5D). This reduction in signal intensity is likely due to an absence of VEGFR3 at the membrane available for staining, as western blot analysis reveals nearly identical levels of VEGFR3 in shScramble and shCDH5 treated LECS (Figure 5B), demonstrating that expression of VE-cadherin is required for the localization of VEGFR3 at the plasma membrane of LECs.

To determine whether cellular signaling of VEGFR3 was diminished by the loss of VE-cadherin, we performed a proximity ligation assay (PLA) in LECs using VEGFR3 and pan-phosphotyrosine antibodies to detect the trans-phosphorylation and activation of VEGFR3 following VEGF-C stimulation, in vitro. First, we confirmed the specificity of the assay by observing PLA signal localized only in HEK293T cells transfected with a GFP-tagged VEGFR3 expression plasmid, but not control cells (data not shown). To explore if destabilization of VEGFR3 at the membrane affected its ability to signal via ligand-mediated autophosphorylation, LECs were transduced with either shScramble or shCDH5 prior to treatment with VEGF-C or DMSO as a vehicle. As predicted, there was a robust increase in p-VEGFR3 PLA-foci in the VEGF-C treated shScramble LECs in comparison to DMSO treated shScramble control (Figure 5E-F), with many of the foci concentrated around cell-cell junctions. Remarkably, shCDH5 LECs showed no increase in p-VEGFR3 PLA-foci between the DMSO or VEGF-C treatment, indicating that loss of VE-cadherin significantly attenuated the ability of VEGFR3 to respond to VEGF-C (Figure 5F).

To ascertain whether the decreased VEGF-C stimulated phosphorylation of VEGFR3 signaling in CDH5 knock-down LECs impacted downstream VEGFR3 signaling, we first profiled ERK and AKT signaling in shScramble or shCDH5 over a 60-minute time-course following VEGF-C stimulation (Supplemental Figure 7). Prior to the experiment, cells were cultured overnight in serum-free media to synchronize cell cycle and reduce basal ERK and AKT signaling prior to stimulation with 75 ng/mL VEGF-C. In shScramble LECs VEGF-C stimulation resulted in a strong and rapid (within 10 minutes) activation of both ERK and phosphoinositide 3-kinase signaling, as determined by AKT phosphorylation (Supplemental Figure 7B-C). Activation of both pathways was largely turned off by 60 minutes. Interestingly, shCDH5 LECs did not display as robust p-ERK or p-AKT signal at 10 minutes and showed highly comparable levels by 20 minutes and all following timepoints. These data suggest that knockdown of VE-cadherin does not significantly alter downstream p-ERK or p-AKT signaling, despite a significant change in VEGFR3 phosphorylation levels. These results are not entirely unsurprising considering there exists significant crosstalk between the VEGFR family members. Further, VEGF-dependent and VEGF-independent intracellular pathways have been shown to create complex positive and negative feedback loops regarding downstream VEGFR signaling²⁶. For example, VEGFR3 is known to participate in expression feedback loops with VEGFR2, specifically loss of VEGFR3 increases VEGFR2 levels²⁷. Ligand stimulated VEGFR2 activities ERK1/2 signaling in a manner similar to VEGFR3. It is possible that VEGFR2 can compensate for a potential loss of VEGFR3 signaling, as indicated by a decrease in VEGFR3 phosphorylation in shCDH5 treated LECs.

Next, to determine whether the altered VEGFR3/VEGF-C signaling dynamics in VE-cadherin knockdown LECs had a consequential effect on broad cellular activity, we performed an Edu incorporation proliferation assay (Figure 5G). Both shScramble and shCDH5 demonstrated a dose-dependent increase in Edu+ cells following VEGF-C stimulation, however shCDH5 treated LECs exhibited significantly less proliferation in response to VEGF-C, adding further evidence to a destabilized VEGFR3 signaling node in the absence of VE-cadherin.

Recently, VEGFR3 has also been shown to be transactivated through ligand-independent mechanisms. For example, VEGFR3 can be transactivated through the G-protein-coupled receptor (GPCR) Endothelin-1 in a c-Src dependent manner^28–30. Another GPCR pathway, adrenomedullin (AM), has been shown to transactivate VEGFR2 in blood endothelial cells, inducing proangiogenic responses²⁹. In addition, AM potently stabilizes the lymphatic endothelial barrier by linearizing VE-cadherin through G-protein mediated signaling events acting on downstream effectors such the small GTP-ase Rap1 (Ras-related protein)^31–33. Thus, we hypothesized that VE-cadherin may also sustain an AM-VEGFR3 transactivation node required for lymphatic maintenance. We first probed this hypothesis by testing whether AM could transactivate VEGFR3 in human neonatal dermal LECs. To detect only transactivation events, LECs were pretreated with sunitinib, a pan-receptor tyrosine kinase inhibitor, to block canonical VEGF-C/D ligand-mediated autophosphorylation and detect only trans-phosphorylation events. Sunitinib pretreatment predictably reduced the baseline level of p-VEGFR3 observed in the vehicle- and VEGF-C-treated cells, confirming the inhibitory action of sunitinib on ligand-mediated receptor activation (Figure 6A and 6B). To determine whether AM signaling could lead to transactivation of VEGFR3, LECs were treated with AM or vehicle in the presence of sunitinib, and a significant increase in p-VEGFR3 was observed with AM treatment (Figure 6A and 6B). Transactivation was confirmed by immunoprecipitating VEGFR3 from the treated cells and blotting for tyrosine phosphorylation (Figure 6C). Simultaneously treating LECs with AM and an excess of its competitive inhibitor, AM 22–52, blocked the increase in p-VEGFR3, supporting the finding that AM signaling directly promotes the transactivation of VEGFR3 (Figure 6B). Finally, to determine whether these transactivation events were mediated by c-Src, LECs were also treated with the c-Src kinase inhibitor PP2. Under these inhibitory conditions, p-VEGFR3 level was no longer elevated in response to AM, confirming that c-Src is a necessary mediator for AM-induced VEGFR3 transactivation in LECs (Figure 6A and 6B). Taken together, these results establish a novel mechanism for AM to transactivate VEGFR3 through c-Src in lymphatic endothelial cells.

Figure 6: Adrenomedullin signaling transactivates VEGFR3 in Lymphatic Endothelial Cells.

A) Representative images of VEGFR3/p-Tyr PLA generated foci in treated lymphatic endothelial cells (LECs). Scale bar 20 μm. (B) Quantification of the number of foci per cell (n=5 independent experiments). In all graphs the red horizontal line represents the mean. Significance for multiple comparisons was determined by one-way ANOVA with Bonferroni posttests. (C) VEGFR3 immunoprecipitated from treated LECs was blotted for VEGFR3 and tyrosine phosphorylation. D) Representative images of VEGFR3/p-Tyr PLA generated foci in shScramble or shCDH5 treated LECs then subject to a pretreatment of sunitinib before administration of DMSO, AM, or AM-22–52, used to assess VEGFR3 transactivation (AM 22–52 alone and VEGF-C images not shown). E) Quantification of number of foci per cell (n=4 independent experiments). Significance for multiple comparisons was determined by two-way ANOVA with Tukey’s multiple comparisons test. All values are mean ± SD.*p<0.05 and **p<0.01. Detailed description of group size, normality testing, statistical tests and post-tests, and exact p-values can be found in the statistical supplement.

We next explored whether loss of VE-cadherin could disrupt this VEGFR3 AM-mediated transactivation using the previously described PLA assay for pan-phosphotyrosine residues in either shScramble or shCDH5 LECs, pre-treated with sunitinib and then vehicle (DMSO), AM, AM + AM 22–52, or VEGF-C (Figure 6D and 6E). AM produced an expected robust increase in p-VEGFR3 foci/cell in the shScramble LECs in comparison to vehicle. A significant decrease in AM generated p-VEGFR3 foci/cell was noted when shScramble LECs were co-treated with AM + AM 22–52, demonstrating that shScramble virus did not impair the ability of AM to transactivate VEGFR3 in these LECs, while no significant change in number of foci for shCDH5 LECs was observed. Importantly, no significant change in number of p-VEGFR3 PLA-foci/cell was noted for shCDH5 LECs treated with AM or AM 22–52 in comparison to vehicle, indicating that VE-cadherin is required to maintain an active AM-mediated transactivation of VEGFR3.

DISCUSSION

Here, using genetic mouse models that enable the temporal deletion of VE-cadherin in lymphatics, we reveal the dispensability of VE-cadherin for embryonic cardiac lymphatic development, as well as its requirement in postnatal cardiac lymphatic growth and maintenance of quiescent adult cardiac lymphatics (Figure 7A). We mechanistically demonstrate that loss of VE-cadherin in LECs results in the instability of VEGFR3 at the membrane, abrogating both VEGF-C ligand-mediated activation and a newly identified AM-mediated transactivation of VEGFR-3, which attenuates downstream signaling hindering LEC proliferation and survival (Figure 7B). This work highlights the complexity of lymphatic vessel-bed-specific differences in development and adulthood and underscores the importance of VE-cadherin in establishing lymphangiogenic signaling nodes that ensure lymphatic maintenance and function.

Figure 7: VE-cadherin is a Critical Component of the VEGFR3 Signaling Node within LECs responsible for Cardiac Lymphatic Maintenance and Function.

A) Schematic depicting the role of VE-cadherin in stabilizing VEGFR3 at the membrane in LECs allowing for VEGFC ligand-mediated activation of VEGFR3 or adrenomedullin mediated transactivation of VEGFR3 through binding of AM to CLR/RAMP complex triggering c-Src mediated phosphorylation of VEGFR3. This stabilized VEGFR3 signaling node in the LECs of Cdh5^flox/flox mice promotes LEC proliferation and survival, allowing for normal maintenance and function of the cardiac lymphatic network. B) Loss of VE-cadherin leads to a destabilized VEGFR3 signaling node, attenuating downstream signaling in LECs, which in Cdh5^EC-KO mice ultimately leads to a failure to form a proper cardiac lymphatic network when VE-cadherin is lost postnatally. When VE-cadherin is lost in adulthood, the fully developed cardiac lymphatic network regresses and leads to basal lymphatic dysfunction.

The temporal dichotomy between the dispensability or requirement of VE-cadherin for cardiac lymphatic growth during development or adulthood, respectively, is of great interest. Evidence from analysis of the plasticity of button-like junctions in in airway lymphatic endothelium reveals a dynamic change in the proportion of VE-cadherin-marked zipper and button lymphatic junctions³⁴. Zipper junctions predominate at E12.5 then transition to primarily button junctions starting at E17.5, rapidly increasing at birth and completing within the first month. This junctional shift appears to correlate with the severity of the effect of VE-cadherin loss. Early embryonic loss, when zippers predominate, cardiac lymphatic formation is spared, while early post-natal loss during the peak of the junctional transition results in loss of the network. Perhaps loss of VE-cadherin during high-zipper junction intervals can be compensated by upregulation of other tight junction molecules. Such compensation would not occur when button-junctions need to predominate, likely to promote remodeling of the cardiac lymphatic plexus. Multiple studies have reported that non-venous cardiac lymphatic progenitors can arise from hemogenic endothelium, secondary heart field, paraxial mesoderm, and muscle progenitor populations^{13, 14, 35–37}. This study shows that embryonic deletion of VE-cadherin is dispensable for the initial development of cardiac lymphatic development, however the loss of this scaffolding protein results in altered morphology of the network. It remains possible that there exist Prox1-negative progenitor populations that evade cre-mediated deletion of VE-cadherin and consequently contribute to proper cardiac lymphatic development. Future studies utilizing lineage tracing techniques or multiple Cre-driver mice could be valuable in identifying the requirement of VE-cadherin within the diverse cardiac lymphatic progenitor populations.

Significantly, loss of VE-cadherin in the growing cardiac lymphatic network resulted in a reduction in the number of sprouting vessel tips and shifted the morphology of these tips from a more classic, sharp spiky shape to a more rounded blunt shape. This finding correlates with recent work establishing that heterogeneous VEGFR3 expression levels in the tips of E17.5 dermal lymphatics drives alterations in tip morphology, with VEGFR3-deficient tips appearing blunted and individual VEGFR3-deficient cells displaying diminished capacity to contribute to vessel tips, compromising the ability of these vessels to extend³⁸. Similar phenotypes were described in Tbx1-Vegfr3 compound heterozygotes, which revealed sensitivity of cardiac lymphatics to VEGFR3 gene dosage³⁹. Furthermore, our data indicate that knockdown of VE-cadherin, in cultured cells, is associated with temporal shifts in ERK and AKT signaling (Supplemental Figure 7), without decreasing expression of VEGFR3 (Figure 5B). Collectively these data support the notion that VE-cadherin stabilizes VEGFR3 at the plasma membrane, but that its loss does not fully abrogate all VEGF-C signaling– which may occur through other VEGF receptors.

In contrast to the developmental deletion of VE-cadherin, we demonstrate that postnatal and adult deletion of VE-cadherin causes a significant loss and fragmentation of the epicardial lymphatic network. This effect is accompanied by significant systemic dermal lymphatic dysfunction as well as deficient cardiac lymphatic transport and mild myocardial edema. Surprisingly, there is no significant cardiovascular dysfunction noted in these unchallenged mice, indicating that cardiac lymphatics are dispensable for normal cardiac function. Moreover, even under conditions of MI, we find that an intact cardiac lymphatic network is dispensable for preserving cardiac function after injury, a finding which is consistent with recent studies^{40, 41}. In larger species, including humans, it is well-established that even small increases in water content in the heart significantly decrease cardiac output^42–46. Thus, the retention of normal ejection fraction and overall cardiac function despite the absence of cardiac lymphatics in the Cdh5^LEC-KO mice may be attributed to a combination of rapid heartbeat and physiologic size of mouse hearts, where the left ventricular systolic wall thickness is ~1 mm, compared to ~3 mm in rats and ~10 mm in humans⁴⁷. This combination, along with extrinsic forces of the beating heart, could permit effective myocardial fluid extravasation into the interstitium, with minimal myocardial edema, in an un-injured or injured heart.

VE-cadherin is a cornerstone junctional molecule through which other lymphangiogenic signaling cascades are assembled. Here we show that loss of VE-cadherin results in instability of VEGFR3 at the plasma membrane of LECs, diminishing its capacity to signal, either through canonical VEGF-C ligand-mediated activation or noncanonical transactivation through AM signaling. Additionally, we demonstrate that AM transactivation of VEGFR3 is mediated through c-Src. VE-cadherin is also phosphorylated by c-Src in response to VEGF⁴⁸. Thus, these data identify VE-cadherin as a scaffolding protein node to maintain VEGFR3 at the plasma membrane, beneficially and proximally to c-Src and AM transactivation of VEGFR3. This highlights a dynamic cellular convergence of critical signaling pathways that are required for the maintenance of quiescent lymphatic vessels. These signaling nodes are also likely to be essential in the context of acute injury, such as MI, when plasma levels of AM and other lymphangiocrine factors surge. In our current study, we find that Cdh5^LEC-KO mice exhibit higher mortality, which is fully consistent with prior studies that deplete VEGRF3 signaling in both murine and zebrafish MI models^{41, 49}.

While we and others find that loss of cardiac lymphatics is not detrimental to heart function after MI, the benefits of cardiac lymphangiogenesis after MI are evident, as both genetic overexpression of AM and intramyocardial-targeted delivery of VEGFC have been shown to drive cardiac lymphangiogenesis and improve cardiovascular outcomes in surgical models of MI^{17, 20}. Therefore, it will be of interest to consider whether VE-cadherin, as a scaffolding protein maintaining signaling nodes, can be exploited for sustaining maximal lymphangiogenesis during cardiac repair. Thus, in the setting of acute heart failure, when VEGF and AM peptide levels soar^{17, 50, 51}, VE-cadherin likely plays a critical role in driving robust cardiac lymphangiogenesis through both canonical AM-mediated GPCR pathways as well as the AM-mediated VEGFR3 transactivation.

NOVELTY AND SIGNFICANCE.

What Is Known?

• VE-cadherin is an endothelial adhesion molecule located between cell junctions contributing to the integrity, permeability, and organization of intercellular junctions.

• VE-cadherin also regulates numerous cellular processes such as proliferation, apoptosis, mechanotransduction and modulates endothelial growth factor receptor functions.

• VE-cadherin is indispensable for blood vascular development, and more recently, its loss in lymphatic vessels has been associated with temporal and organ-specific effects.

What New Information Does This Article Contribute?

• This study demonstrates that VE-cadherin expression in lymphatics is required to develop and maintain the cardiac lymphatic network.

• Loss of VE-cadherin significantly impacts cardiac lymphatic structure and function, but surprisingly, an intact cardiac lymphatic network is dispensable for preserving basal cardiac function before and after injury.

Lymphatic vessels mediate organ-specific effects under both physiologic and pathophysiologic conditions, demonstrating functional specialization and molecular responses to the local environment. Cardiac lymphatics have many unique features amongst lymphatic vessels: i) their lymphatic transport is intrinsically linked to cardiac contractions, ii) they undergo substantial remodeling during cardiovascular diseases, and iii) their therapeutic stimulation of lymphangiogenesis after myocardial infarction improves cardiac recovery. Recent studies demonstrated that VE-cadherin is required for the development and maintenance of some lymphatic vessel beds and valves, yet the role of VE-cadherin in cardiac lymphatics is unexplored. This study is first to demonstrate the requirement of lymphatic VE-cadherin expression for the development of embryonic and postnatal cardiac lymphatics and their maintenance in adulthood. Loss of VE-cadherin and cardiac lymphatics causes significant lymphatic dysfunction and blunted lymphangiogenesis after surgically-induced myocardial infarction. Surprisingly, lack of cardiac lymphatics does not significantly impair cardiac function before or after injury. VE-cadherin regulates junctional integrity and helps regulate numerous cellular processes, including those of endothelial growth factor pathways. Notably, VE-cadherin stabilizes VEGFR3 at the plasma membrane, poising the receptor for canonical activation by VEGF-C and for newly-identified transactivation by adrenomedullin, thereby establishing a functional lymphatic signaling node required for cardiac lymphatic maintenance.

Nonstandard Abbreviations and Acronyms:

VE-cadherin

Vascular Endothelial Cadherin

Cdh5

Cadherin 5; gene encoding VE-cadherin

VEGF

Vascular Endothelial Growth Factor

VEGF-C

Vascular Endothelial Growth Factor C

VEGFR2

Vascular Endothelial Growth Factor Receptor 2

VEGFR3

Vascular Endothelial Growth Factor Receptor 3

Prox1

Prospero Homeobox 1; gene

LEC

Lymphatic Endothelial Cell

BEC

Blood Endothelial Cell

PLA

Proximity Ligation Assay

LYVE1

Lymphatic Vessel Endothelial Hyaluronan Receptor 1

GFP

Green Fluorescent Protein

CM

Cardiomyocyte

Qdots

Fluorescent Quantum Dots

MI

Myocardial Infarction

LAD

Left Anterior Descending (Artery)

AM

Adrenomedullin

AM 22–52

Adrenomedullin Amino Acids 22–52 (truncated version of Adrenomedullin)

CLR

Calcitonin-Receptor Like-Receptor

RAMP2

Receptor Activity Modifying Protein 2

Rap1

Ras-Related Protein 1

GPCR

G-Protein Coupled Receptor

Edu

5-ethynyl-2’-deoxyuridine

DMSO

Dimethylsulfoxide

BPM

Beats Per Minute

D

Diastolic

S

Systolic

EF

Ejection Fraction

FS

Fractional Shortening

IVS

Intraventricular Septal (Distance)

LV

Left Ventricle

LVID

Left Ventricular Posterior Wall (Thickness)

TM

Tamoxifen

IP

Intraperitoneal

SMA

Smooth Muscle Actin

SCC

Side Scatter