Vascular Endothelial Growth Factor Receptor-2 Promotes the Development of the Lymphatic Vasculature

Vascular endothelial growth factor receptor 2 (VEGFR2) is highly expressed by lymphatic endothelial cells and has been shown to stimulate lymphangiogenesis in adult mice. However, the role VEGFR2 serves in the development of the lymphatic vascular system has not been defined. Here we use the Cre-lox system to show that the proper development of the lymphatic vasculature requires VEGFR2 expression by lymphatic endothelium. We show that Lyve-1wt/Cre;Vegfr2flox/flox mice possess significantly fewer dermal lymphatic vessels than Vegfr2flox/flox mice. Although Lyve-1wt/Cre;Vegfr2flox/flox mice exhibit lymphatic hypoplasia, the lymphatic network is functional and contains all of the key features of a normal lymphatic network (initial lymphatic vessels and valved collecting vessels surrounded by smooth muscle cells (SMCs)). We also show that Lyve-1Cre mice display robust Cre activity in macrophages and in blood vessels in the yolk sac, liver and lung. This activity dramatically impairs the development of blood vessels in these tissues in Lyve-1wt/Cre;Vegfr2flox/flox embryos, most of which die after embryonic day14.5. Lastly, we show that inactivation of Vegfr2 in the myeloid lineage does not affect the development of the lymphatic vasculature. Therefore, the abnormal lymphatic phenotype of Lyve-1wt/Cre;Vegfr2flox/flox mice is due to the deletion of Vegfr2 in the lymphatic vasculature not macrophages. Together, this work demonstrates that VEGFR2 directly promotes the expansion of the lymphatic network and further defines the molecular mechanisms controlling the development of the lymphatic vascular system.


Introduction
The lymphatic vasculature transports immune cells, absorbs dietary fats, and regulates tissue fluid homeostasis by returning fluid and macromolecules to the blood vascular system [1]. Insufficiency of the lymphatic vascular system leads to the formation of lymphedema, a condition characterized by massive swelling of affected limbs, fibrosis, and impaired immunity [1]. In humans, mutations in VEGF-C, VEGFR3, GJC2, GJA1, KIF11, FOXC2, CCBE1, SOX18, PTPN14, and GATA2 have been found in families with inherited forms of lymphedema and cause striking defects in the development of the lymphatic vasculature [2][3][4][5][6][7][8][9][10][11][12]. These important clinical observations have fueled efforts to further identify the molecular mechanisms governing the formation of the lymphatic vasculature.
The current model for the development of the mammalian lymphatic vascular system conforms to the centrifugal theory originally proposed by Florence Sabin over 100 years ago [13]. According to this model, lymphatic endothelial cells (LECs) differentiate from blood endothelial cells (BECs) and migrate from veins to form lymph sacs during embryogenesis. Sprouting from these sacs gives rise to an immature lymphatic network which remodels into a hierarchal pattern of capillaries and valved collecting vessels [14]. The centrifugal theory for mammals has been supported by the expression pattern of molecular markers of LECs [15], lineage tracing experiments [16], and by the unique mutant phenotypes of genetically modified mice (reviewed in [17]). Despite these recent advances, the molecular mechanisms driving the expansion of the lymphatic network remain largely unknown.
There is growing evidence that VEGFR2, a receptor tyrosine kinase activated by VEGF-A, -C and -E, stimulates lymphangiogenesis. Overexpression of VEGFR2 ligands in the skin of adult mice and in tumors induces the growth of lymphatic vessels [18][19][20][21]. Furthermore, VEGF-A has been shown to promote lymphangiogenesis in the corneal micropocket assay in a VEGF-C/-D/-R3 independent manner [21]. To gain a better understanding of how VEGFR2 stimulates lymphangiogenesis, numerous in vitro experiments have been performed with LECs. These reports have shown that VEGF-A/VEGFR2 signaling promotes LEC proliferation, migration, and tube formation as well as the permeability of LEC monolayers [22][23][24][25][26][27]. We recently reported that VEGF-A stimulation of LECs leads to the phosphorylation of VEGFR2 on several tyrosine residues (Tyr 951, Tyr 1054, Tyr 1059, Tyr 1175 and Tyr 1214), and promotes protein kinase C dependent phosphorylation of ERK1/2 and PI3-K dependent phosphorylation of Akt [28]. The activation of both of these pathways was required for VEGF-A/VEGFR2-induced proliferation and migration of LECs [28]. These studies have begun to shed light on the molecular pathways and cellular processes activated by VEGFR2 in LECs. However, the role VEGFR2 serves in the development of the lymphatic vasculature has not been explored, in part, because Vegfr2 knockout mice die before the lymphatic vascular system forms [29]. In the present study, we use the Cre-lox system to overcome this obstacle and conditionally inactivate Vegfr2 in LECs to characterize its function in the development of the lymphatic vascular system.

Results
Adult Lyve-1 wt/Cre ;Vegfr2 flox/flox mice display lymphatic hypoplasia Lyve-1 is a hyaluronan receptor highly expressed by lymphatic capillaries but not by collecting lymphatic vessels or valves [30,31]. Although collecting lymphatic vessels and valves do not express Lyve-1, they are thought to arise from Lyve-1 positive LECs [30,32]. Lyve-1 Cre mice were recently developed to conditionally delete floxed (flanking loxP) DNA sequences in LECs [33]. However, the removal of floxed DNA sequences in collecting lymphatic vessels and valves was not previously analyzed. To further characterize the pattern of Cre-mediated recombination in Lyve-1 Cre mice, we crossed this strain with the mT/mG reporter strain. mT/mG mice possess a GFP reporter whose expression is induced by Cre-mediated recombination [34]. Lymphatic vessels in ear skin whole-mounts from Lyve-1 wt/Cre or mT/mG mice did not express GFP. However, Lyve-1 wt/Cre ;mT/mG mice displayed strong GFP expression in macrophages as well as in lymphatic capillaries, collecting lymphatic vessels and valves ( Figure 1A, 1B). This result confirms previous reports documenting that Lyve-1-negative collecting vessels arise from Lyve-1-positive vessels [30,32] and indicates that the Lyve-1 Cre mouse can be used to excise floxed DNA sequences in LECs that give rise to lymphatic capillaries, collecting lymphatic vessels and valves.

Vegfr2 is not required for the maturation of collecting lymphatic vessels
Lymphatic capillaries and collecting lymphatic vessels are differentially covered by SMCs. Lyve-1-positive lymphatic capillaries are free of SMCs whereas Lyve-1-negative collecting vessels are covered by SMCs. Interestingly, defects in the patterning of the lymphatic vasculature have been associated with the mislocalization of SMCs on Lyve-1-positive lymphatic vessels [30,31,35]. To determine whether the hypoplastic lymphatic network in Lyve-1 wt/Cre ;Vegfr2 flox/flox mice was aberrantly covered by SMCs, we stained ear skin from adult mice for Lyve-1 and smooth muscle actin. This revealed that the Lyve-1-positive lymphatic networks in Vegfr2 flox/flox and Lyve-1 wt/Cre ;Vegfr2 flox/flox mice were not  However, SMCs were properly associated with Lyve-1-negative vessels in both strains of mice (data not shown).
The CD31 staining also revealed that the patterning of the blood vasculature was normal in the ear skin of adult Lyve-1 wt/ Cre ;Vegfr2 flox/flox mice ( Figure 3E, 3F). The number of blood vessel branch points was not significantly different between Vegfr2 flox/flox (94.3861.679; n = 4) and Lyve-1 wt/Cre ;Vegfr2 flox/flox mice (97.4160.8079; n = 4). Therefore, Lyve-1 wt/Cre ;Vegfr2 flox/flox mice display a dermal lymphatic vascular phenotype in the absence of a dermal blood vascular phenotype. Together, these results suggest that the lymphatic defect in the skin of Lyve-1 wt/Cre ;Vegfr2 flox/flox mice is not secondary to a blood vessel defect in the skin.
Vegfr2 is not expressed by macrophages and inactivation of Vegfr2 in the myeloid lineage does not affect lymphatic development Crosses with mT/mG reporter mice revealed that Lyve-1 Cre mice exhibit Cre recombinase activity in macrophages as well as lymphatic vessels ( Figure 1A, 1B). Therefore, several experiments were performed to rule out the possibility that the lymphatic phenotype of Lyve-1 wt/Cre ;Vegfr2 flox/flox mice was due to the inactivation of Vegfr2 in macrophages. First, we explored the expression of Vegfr2 by macrophages. Whole-mount immunofluorescence staining showed that Vegfr2 was not expressed by macrophages in the ear skin of Vegfr2 wt/GFP mice ( Figure S4). Next, the LysM Cre strain was used to conditionally delete target sequences in the myeloid lineage. LysM Cre mice were bred with mT/mG reporter mice to characterize the expression pattern of Cre recombinase and to trace the fate of cells of the myeloid lineage. All LysM wt/Cre ;mT/mG mice displayed strong GFP expression by macrophages in the ear skin ( Figure S5). GFP did not co-localize with Vegfr3 in the ear skin of LysM wt/Cre ;mT/mG mice ( Figure S5;  . This finding indicates that cells genetically marked by LysM Cre do not differentiate into LECs during normal murine development and is in agreement with another report using the LysM Cre line [38]. LysM Cre mice were then crossed with Vegfr2 flox mice to determine whether deleting Vegfr2 in myeloid cells affects the development of the lymphatic vasculature. Whole-mount immunofluorescence staining of ear skin for Lyve-1 revealed that the density of lymphatic vessels was not significantly different between Vegfr2 flox/flox (8.55060.278, n = 5 mice) and LysM wt/ Cre ;Vegfr2 flox/flox (9.15060.5895, n = 5 mice) littermates ( Figure  7A-D). Furthermore, the diameter of lymphatic vessels was not significantly different between Vegfr2 flox/flox (52.98 mm61.328, n = 5 mice) and LysM wt/Cre ;Vegfr2 flox/flox (50.05 mm62.031, n = 4 mice) littermates ( Figure 7A-D). EBD was also effectively transported from injected hind paws to popliteal and iliac lymph nodes in all Vegfr2 flox/flox and LysM wt/Cre ;Vegfr2 flox/flox mice (data not shown). Together, these data reveal that Vegfr2 is not required in the myeloid lineage for the proper development of the lymphatic system. This demonstrates that the lymphatic phenotype of Lyve-1 wt/Cre ;Vegfr2 flox/flox mice is due to the ablation of Vegfr2 in LECs, not macrophages.

Discussion
VEGFR2 is widely recognized as an essential gene driving the formation of the blood vasculature during embryogenesis. The present study demonstrates that VEGFR2 also directly promotes the development of the lymphatic vasculature. We show that the density, but not diameter, of lymphatic vessels is dramatically reduced in Lyve-1 wt/Cre ;Vegfr2 flox/flox mice. Additionally, we demonstrate that lymphatic vessels in Lyve-1 wt/Cre ;Vegfr2 flox/flox mice properly mature into collecting vessels. These findings indicate that VEGFR2 is required for the expansion, but not the specification or maturation, of the lymphatic vasculature.
The lymphatic vasculature of mammals can grow by undergoing sprouting lymphangiogenesis, in which new lymphatic vessels emerge from pre-existing vessels, or circumferential lymphangiogenesis, which is characterized by an increase in the diameter of lymphatic vessels [19]. Sprouting lymphangiogenesis increases the number of lymphatic branch points and complexity of the lymphatic network. Therefore, if VEGFR2 stimulates sprouting lymphangiogenesis, there should be fewer lymphatic branch points in Lyve-1 wt/Cre ;Vegfr2 flox/flox mice than in Vegfr2 flox/flox mice. On the other hand, if VEGFR2 promotes circumferential lymphangiogenesis, lymphatic vessels should have a smaller diameter in Lyve-1 wt/Cre ;Vegfr2 flox/flox mice than in Vegfr2 flox/flox mice. Importantly, we found that the number of lymphatic branch points was greatly reduced in Lyve-1 wt/Cre ;Vegfr2 flox/flox mice and that the width of lymphatic vessels was not significantly different between wildtype and mutant mice. In contrast to a previous report [19], our results suggest that VEGFR2 signaling promotes sprouting lymphangiogenesis rather than circumferential lymphangiogenesis. The earlier model proposing that VEGFR2 stimulates circumferential rather than sprouting lymphangiogenesis was based on experiments showing that adenoviral and transgenic overexpression of VEGF-E (VEGFR2-specific ligand) preferentially induces the enlargement of lymphatic vessels instead of the formation of new lymphatic vessels [19]. The discrepancy between our findings and this previous report could be due to the fact that overexpression of VEGF-E increases blood vessel permeability [19], an effect which may contribute to the enlargement of lymphatic vessels. Alternatively, the effect of VEGF-E on lymphatics may be unique to this factor, as adenoviral expression of VEGF-A 164 induces lymphatic vessel sprouting in vivo, albeit to a lesser extent than VEGF-C [19]. Furthermore, VEGF-A has also been shown to induce sprouting lymphangiogenesis in mouse corneas and in vitro by LEC spheroids [21,39]. Together, these observations suggest that the function of VEGFR2 in lymphatic vessels is similar to its function in blood vessels, where it also stimulates the budding of new vessels from pre-existing vessels [40].
Although it is well established that lymphatic vessels can grow by sprouting, the underlying mechanism of sprouting lymphangiogenesis is not well defined. In contrast, the process of sprouting hemangiogenesis is well characterized. Growing blood vessels are led by specialized tip cells which migrate and extend numerous filopodia to probe the microenvironment for directional cues [40]. These cells are followed by stalk cells that proliferate in response to growth factors and thereby promote vessel extension [40]. VEGF-A/VEGFR2 signaling plays a critical role in regulating tip cell activity (migration/filopodia extension) and stalk cell proliferation [41]. VEGFR2 may also control sprouting lymphangiogenesis in a similar fashion. In vivo, VEGFR2 is expressed by growing lymphatic vessels and their filopodia [42]. Whether the LECs extending filopodia are true tip cells has yet to be determined. Nevertheless, VEGFR2 at the end of a growing lymphatic vessel could be involved in sensing directional signals and migration. Additionally, VEGFR2 could promote vessel extension by stimulating the proliferation of LECs. We found that LEC proliferation is reduced in Lyve-1 wt/Cre ;Vegfr2 flox/flox embryos. Furthermore, we and others have previously shown that VEGF-A/VEGFR2 signaling promotes LEC migration and proliferation in vitro [28]. Therefore, the loss of VEGFR2 in LECs in Lyve-1 wt/ Cre ;Vegfr2 flox/flox mice may impair lymphangiogenesis by affecting the migration and proliferation of LECs. During development specific lymphatic vessels acquire a collecting vessel phenotype, a process involving the recruitment of mural cells and formation of intraluminal valves. Recent studies of genetically modified mice have identified several genes that participate in collecting vessel maturation, such as Ang2, Ephrinb2, and NFATC1 [30][31][32]43,44]. Importantly, VEGFR2 signaling induces the expression of Ang2, nuclear translocation of NFATC1, and is regulated by Ephrinb2-mediated internalization and trafficking [24,32,45]. These observations, as well as the high expression of VEGFR2 by collecting vessels and valves ( [19] and data not shown), led us to test the hypothesis that VEGFR2 participates in the maturation of lymphatics into collecting vessels. To our surprise we found that Lyve-1 wt/Cre ;Vegfr2 flox/flox mice develop normal collecting lymphatic vessels. Mural cells were properly associated with lymphatic vessels and intraluminal valves were present in Lyve-1 wt/Cre ;Vegfr2 flox/flox mice. The lack of an abnormal collecting vessel phenotype may be due to compensation by VEGFR3, which is also expressed by collecting vessels and valves [19]. Future work with VEGFR3 and VEGFR2 mutant mice will help elucidate the role these receptors serve in the remodeling of the lymphatic system.
During the course of our study we discovered that most Lyve-1 wt/ Cre ;Vegfr2 flox/flox mice die during embryonic development. Previously described mutant embryos that display fatal lymphatic defects die from edema [15,46]. However, Lyve-1 wt/Cre ;Vegfr2 flox/flox embryos were not edematous at any of the time points analyzed. This suggests that Lyve-1 wt/Cre ;Vegfr2 flox/flox embryos do not die from a lymphatic defect. Therefore, the Lyve-1 Cre allele must induce the loss of Vegfr2 in a different cell type that is required for survival. In agreement with a previous study that documented Lyve-1 expression in blood vessels of the yolk sac, liver and lung [36], we found that Lyve-1 Cre mice expressed Cre recombinase in blood vessels of the yolk sac, liver and lung. We also found that the density of blood vessels was dramatically reduced in these tissues in Lyve-1 wt/Cre ;Vegfr2 flox/flox embryos. Proper development of the yolk sac vasculature is required for mice to survive to birth [37]. Therefore, Lyve-1 wt/Cre ;Vegfr2 flox/flox embryos may die of a yolk sac vascular defect. Additionally, vascular defects in the liver and lung may also contribute the lethal phenotype of Lyve-1 wt/Cre ;Vegfr2 flox/flox mice. Select Lyve-1 wt/Cre ;Vegfr2 flox/flox mice may survive because they display low Cre recombinase activity in BECs in the yolk sac, liver and lung. Alternatively, differences in genetic background may contribute to the survival of a subset of Lyve-1 wt/Cre ;Vegfr2 flox/flox mice.
In conclusion, we show that VEGFR2 directly promotes the expansion of the lymphatic vessel network. This newly identified function of VEGFR2 further defines the molecular pathways controlling the development of the lymphatic vasculature and sheds light on the mechanisms by which therapeutic agents targeting VEGFR2 inhibit lymphangiogenesis.

Ethics Statement
Experiments performed with mice were carried out in accordance with an animal protocol approved by the IACUC of the University of Texas Southwestern Medical Center (APN 0974-07-05-1).

Quantitative analysis of lymphatic branch points and diameter
For adult mice, the number of branches was counted in 4 images from each mouse at 10X magnification. For embryos, the number of branches was counted in 1-6 images from each mouse at 10X magnification. The same images were used to assess lymphatic vessel diameter. A grid with lines spaced 75mm from one another was placed over each 10X image. Vessel diameter was measured with the NIS-Elements imaging software at locations where two perpendicular grid lines intersected a lymphatic vessel.

Immunofluorescence and immunohistochemical staining of tissue sections
Embryos were fixed overnight at 4uC in 4% PFA, washed with 50% EtOH, processed, and then sectioned at 5 mm for staining. Slides were deparaffinized with xylene and rehydrated through a descending EtOH series. Antigen retrieval was performed with 0.01 M citric acid (pH = 6.0) in a pressure cooker. Slides were then washed with PBS and blocked for 1 hour with TBST + 20% Aquablock. Primary antibodies diluted in TBST + 5% BSA were then added and allowed to incubate overnight at 4uC. Slides were washed with TBST then secondary antibodies diluted in TBST + 5% BSA were added and allowed to incubate for 1 hour at room temperature. Slides were then washed again with TBST and coverslips were mounted with ProLong Gold plus DAPI. Immunohistochemistry was performed with using a similar protocol except endogenous peroxidase activity was blocked by incubating slides with hydrogen peroxide diluted in MeOH and signal was detected via the DAB chromogen system (Dako, cat no. K3468).

Evans blue dye lymphangiography
The popliteal and iliac regions were examined for lymph transport following the injection of Evans blue dye (1% w/v) into the hind paws of mice anesthetized with isoflurane and kept warm with a heating pad.

Assessment of Vegfr2 expression by lymphatic vessels
Frozen sections of adult ear skin were stained with antibodies against Lyve-1 and Vegfr2. Four images at 20X magnification were taken of each section of ear skin. The number of Vegfr2positive and Vegfr2-negative LECs in each image was manually counted. The number of Vegfr2-positive LECs was divided by the total number of LECs and then multiplied by 100 to determine the percent of Vegfr2-positive LECs.

Assessment of LEC proliferation
Tissue sections of E14.5 and E16.5 embryos were stained with antibodies against podoplanin and phospho-histone H3. Six images at 40X magnification were taken of each tissue section. The number of phospho-histone H3-positive and phospho-histone H3-negative LECs in each image was manually counted. The number of phospho-histone H3-positive LECs was divided by the total number of LECs and then multiplied by 100 to determine the percent of LECs proliferating.

Statistical analysis
Data were analyzed using GraphPad Prism statistical analysis software (Version 5.0). All results are expressed as mean 6 SEM. Unpaired student's T-tests were performed to test means for significance. Data were considered significant at P,0.05.