Spreds Are Essential for Embryonic Lymphangiogenesis by Regulating Vascular Endothelial Growth Factor Receptor 3 Signaling^▿

Animals.

Spred-1^−/− and Spred-2^−/− knockout mice backcrossed into the C57BL/6J background have been previously reported (11, 19). Spred-1/2 DKO embryos (Spred-1^−/−/Spred-2^−/−) and corresponding WT control embryos with the same genetic background (C57BL/6J) as the knockout embryos were used. Timed mating was accomplished in the late afternoon, followed by a plug check the next morning. The point of a detected plug was counted as E0.5. The mice were housed in an environment with a 12-h light/dark cycle and a controlled temperature. Animal care and all experiments were conducted in accordance with the institutional guidelines of Kyushu University, Fukuoka, Japan.

Plasmids and viruses.

Human VEGFR-3 short-form cDNA was cloned from human embryonic kidney (HEK) 293T cells by PCR (25) and cloned into the pMX retroviral vector (kindly provided by Toshio Kitamura, University of Tokyo, Tokyo, Japan). Flag-Spred-1, Flag-Spred-2, Flag-Sprouty4, and VEGFR-2 plasmids, as well as Sendai virus vector carrying Spred-1 or GFP, have been described previously (16, 23).

Cell culture, transfection, and infection.

HEK 293T cells and mouse embryonic fibroblasts (MEFs) were cultured in Dulbecco's modified Eagle's medium supplemented with 10% fetal bovine serum (FBS), penicillin, and streptomycin. OP9 stromal cells were cultured in α-modified Eagle's medium supplemented with 20% FBS, penicillin, and streptomycin. To generate 293T cells expressing VEGFR-3, 293T cells were transfected with pMX-VEGFR-3 using FuGENE 6 (Roche Molecular Biochemicals), and cells were selected with 1 μg/ml puromycin (Sigma). To generate MEFs expressing VEGFR-3, MEFs were infected with the retroviruses produced by PLAT-E transfected with pMX-VEGFR-3. Infected cells were selected with 1 μg/ml puromycin. CD45⁻ CD31⁺ endothelial cells sorted by the magnetic activated cell-sorting (MACS) system (Miltenyi Biotec) were infected with the Sendai virus at a multiplicity of infection of 2 and cultured for 3 to 14 days until they were used for experiments. Green fluorescent protein (GFP) indicated that more than 95% of the cells were infected. VEGFR-3 proteins in 293T cells and MEFs were detected by Western blotting analysis with an anti-VEGFR-3 antibody (Ab).

Abs and reagents.

The anti-mouse lymphatic vessel endothelial hyaluronan receptor 1 (LYVE-1) monoclonal Ab (MAb) (ALY2) and rabbit polyclonal Ab anti-LYVE-1 have been described previously (17, 18). Other Abs used in this experiment were as follows: anti-PECAM-1/CD31 (clone 390) and anti-VEGFR-3 (AFL4) (eBioscience); anti-CD34 (RAM34), anti-PECAM-1/CD31 (MEC13.3), and anti-CD45 (clone 104) (PharMingen); anti-Prox1 and anti-podoplanin (Angio Bio); anti-vWF (DAKO); anti-α-smooth muscle actin (SMA) (Sigma); anti-phospho-ERK1/2 (P-ERK1/2) (no. 9106 and 9101), anti-FLAG (M2), anti-phospho-Akt (no. 4058), and anti-Akt (no. 9272) (Cell Signaling Technology); and anti-ERK2 (C-14), anti-VEGFR-2 (A-3), and anti-Myc (9E10) (Santa Cruz Biotechnology). Human VEGF-C C156S, which acts on VEGFR-3 but not on VEGFR-2, was purchased from R&D Systems. VEGF-C C156S was used at a concentration of 400 ng/ml, according to the manufacturer's recommendation. VEGF-A and PDGF-BB were purchased from Peprotech.

Immunohistochemistry.

E14.5 mouse embryos were stained by whole-mount immunohistochemistry with 1:200-diluted anti-mouse LYVE-1 Ab (ALY2) or anti-PECAM-1/CD31 (MEC13.3) as previously described (17). The skin tissues from mouse embryos were fixed with 4% paraformaldehyde to make paraffin tissue specimens or were embedded in an OCT compound (SAKURA Finetechnical Co., Ltd.) to make frozen tissue specimens, and then they were sectioned at 10 μm. Samples were stained with anti-LYVE-1-biotin, antipodoplanin, anti-vWF, anti-α-SMA, and anti-PECAM-1/CD31-PE (eBioscience) at 4°C overnight. Rabbit immunoglobulin G (heavy plus light chains)-biotin and streptavidin-fluorescein isothiocyanate were used to detect anti-LYVE-1 Ab. Immunofluorescence images were acquired using an Olympus BX51 microscope with a fluorescent attachment (Olympus, Tokyo, Japan). When horseradish peroxidase-conjugated immunoglobulin was used as the secondary Ab, samples were incubated in diaminobenzidine at room temperature to visualize reactive samples.

FACS analysis.

To identify the characterization of lymphatic endothelial cells (LECs), we performed fluorescence-activated cell sorter (FACS) analyses of mouse embryo cells with FACSAria (Becton Dickinson), as previously described (17). After removing the embryonic liver and spleen microscopically, E14.5 mouse embryos were dissected and treated with 2.4 U/ml Dispase (Gibco) and 0.05% collagenase S-1 (Nitta Gelatin Co. Ltd.) at 37°C for 30 min to produce a single-cell suspension. To exclude erythrocytes from this preparation prior to analysis with FACSAria, lymphoprep (Axis-Shield PoCAS) treatment was performed as previously described (17). Before the cells were stained, Fc receptors were blocked with an anti-mouse CD16/CD32 Fc receptor (PharMingen). First, all cells were stained with anti-mouse CD45-PerCP Cy5.5, CD31-fluorescein isothiocyanate, CD34-phycoerythrin, and biotinylated ALY2 at 4°C for 30 min and washed three times. Subsequently, the cells were reacted with allophycocyanin-conjugated streptavidin (PharMingen) to visualize LYVE-1-positive cells at 4°C for 30 min and then were washed three times. The cells were sorted and analyzed by FACSAria.

RT-PCR analysis.

Total RNA was extracted from sorted LECs, blood endothelial cells (BECs), and LYVE-1 and CD34 double-positive cells (DPs) using TRIzol (Invitrogen) as previously reported (17). Total RNA was reverse transcribed using the Reverse Transcription Kit (Roche), and the product was used for further analysis. The sequences of primers for reverse transcription (RT)-PCR were previously described (11, 17, 19). PCR products were separated on a 2.0% agarose gel and stained with ethidium bromide. Quantitative real-time PCR was performed using SYBR Green Real-time PCR Master Mix (Toyobo) and the ABI 7000 sequence detector system (Applied Biosystems). The comparative threshold cycle method and an internal control (glyceraldehyde-3-phosphate dehydrogenase) were used to normalize the expression of the target genes.

Primary LEC coculture on OP9 feeder layers.

To examine the effect of Spred on LEC growth, MACS sorted CD45⁻ CD31⁺ endothelial cells from E14.5 (5.0 × 10⁴/well in six-well plates) (see Fig. 4B). Unsorted cells from E12.5 (1.0 × 10⁶/well in six-well plates) (see Fig. 4C and D) embryos were plated on OP9 stromal cells in RPMI1640 (Gibco) with 10% FBS and 10⁻⁵ mol/liter 2-mercaptoethanol (Gibco) as previously reported (17). Seven or 14 days later, the dishes were immunostained with anti-LYVE-1 MAb (ALY2). To quantify the effect of Spred on the colony formation of LECs, we counted the LYVE-1⁺ endothelial cell colonies, which formed the compact type.

Immunohistochemical detection of P-ERK.

Immunohistochemical detection of P-ERK was performed according to the procedure described previously (26). For section preparation, E14.5 embryos were incubated with or without 400 ng/ml human VEGF-C C156S in 2 ml serum-free Dulbecco's modified Eagle's medium for 30 min at 37°C. Then, samples were washed with phosphate-buffered saline and fixed with 4% paraformaldehyde in 0.1 M phosphate buffer at 4°C overnight. The samples were dehydrated, embedded in paraffin wax, sectioned into specimens 7 μm thick, and mounted onto 3-aminopropyltriethoxysilane-coated slides. The sections were deparaffinized, rinsed in Tris-buffered saline, microwaved for 5 min, and then treated with 3% H₂O₂ for 10 min. The sections were then incubated with rabbit polyclonal anti-P-ERK1/2 (no. 9101; 1:200 dilution; Cell Signaling Technology) in Tris-buffered saline containing 1% bovine serum albumin at 4°C overnight. Detection was done with the ENVISION+ System-HRP (DAKO) according to the manufacturer′s instructions. All sections were counterstained with hematoxylin.

Luciferase assay.

Elk-1 activation was measured as described previously (14). HEK 293T cells (2.0 × 10⁵/well) were seeded into six-well plates. The cells were transfected by the calcium phosphate precipitate method 12 h after being seeded with each expression vector (0.01 to 0.1 μg), as indicated. The total DNA concentration was kept constant by supplementation with empty-vector DNAs. Luciferase activity was determined using the Promega luciferase assay system. The β-galactosidase vector (0.05 μg) was used for normalizing transfection efficiencies. The values shown are the averages of one representative experiment in which each transfection was performed in triplicate.

Capillary formation assay.

Human lymphatic endothelial cells (HULECs), which were isolated from human neonatal dermis, were purchased from AngioBio Co. and cultured in EGM2-MV (from Cambrex [formerly Clonetics]) supplemented with 20% off-clot human serum instead of FBS. The HULECs were infected with the Sendai virus carrying GFP or Spred-1 at a multiplicity of infection of 2 and then were cultured for 3 days in EGM2-MV medium. ERK and Akt activation was measured by Western blotting after stimulation with or without 400 ng/ml VEGF-C C156S for 5 and 15 min. Infected HULECs were seeded on precoated 24-well plates with collagen/laminin (Matrigel; BD Biosciences) and incubated at 37°C for 24 h.

MTT assay and wound-healing assay.

Microtiter tetrazolium (MTT) assays and wound-healing assays in the presence of 400 ng/ml VEGF-C C156S were performed as previously described (16).

Statistical analysis.

Data are expressed as the mean ± standard deviation or the mean ± SEM. Statistical analysis was conducted using Student's t test. Statistical significance was defined as a P value of <0.05.

Spred-1/Spred-2 DKO embryos showed severe subcutaneous hemorrhage and edema.

Spred-1 and Spred-2 have many structural similarities, and we have not found any functional differences between the two proteins in vitro (29). Furthermore, using in situ hybridization and RT-PCR analysis, we and others demonstrated that Spred-1 and Spred-2 were expressed in largely overlapping tissues during embryonic development (8; data not shown). Thus, we suspected a physiological and functional redundancy between them. To define overlapping functions of Spred-1 and Spred-2, we generated Spred-1/Spred-2 DKO mice. We crossed Spred1^−/+ Spred2^−/+ male and female mice. The Spred-1^−/+ Spred-2^−/− mice were smaller than the Spred1^−/+ Spred2^−/+ mice but were healthy and fertile. Spred-1^−/− Spred-2^−/+ mice were rarely born, but several natal mice appeared sick: most died within a few months for unknown reasons. We could not obtain any Spred-1^−/− Spred-2^−/− pups among >200 pups, indicating that the deletion of both Spred-1 and Spred-2 genes resulted in embryonic lethality.

We found that most of the Spred-1/2 DKO embryos died of severe subcutaneous hemorrhage and edema between E12.5 and E15.5 (Fig. 1A to C). Spred-1^−/− Spred-2^−/+ embryos often showed bleeding similar to that of Spred-1/2 DKO embryos (Fig. 1B). A subcutaneous hemorrhagic appearance in Spred-1/2 DKO embryos was the most striking phenotype; it was also observed in mouse embryos lacking Syk, SLP-76, or PLCγ2 (1, 30). Whole-mount immunohistochemical staining with the anti-LYVE-1 MAb confirmed that dilated and irregular vessels in Spred-1/2 DKO embryos were LYVE-1-positive lymphatic vessels (Fig. 1D to H). On the other hand, whole-mount immunohistochemical staining with anti-PECAM-1/CD31 MAb confirmed that blood vessels were almost normal in Spred-1/2 DKO embryos compared with control embryos (Fig. 1I and J). Other organs, such as the liver, heart, and placenta, showed no apparent abnormalities (data not shown). These data strongly suggest that lymphatic vessel development is abnormal in Spred-1/2 DKO embryos.

Abnormal lymphatic vessel development in Spred-1/2 DKO embryos.

Histologically, in Spred-1/2 DKO embryos, a number of nucleated red blood cells were observed in subcutaneous hyperplastic dilated vessels, which were LYVE-1 and podoplanin (lymphatic endothelial markers) positive but were von Willebrand factor (vWF) (a blood endothelial marker) and α-SMA (a pericyte and smooth muscle cell marker) negative, confirming that these dilated vessels were lymphatic vessels (Fig. 2A to F). This hyperplastic lymphatic vessel pattern closely resembled that of VEGF-C transgenic mice (28). In control embryos, red blood cells were observed only in blood vessels, but not in lymphatic vessels (Fig. 2G). Consistent with the whole-mount staining with LYVE-1 and CD31 Abs shown in Fig. 1, although the numbers of vWF-positive blood vessels of control and Spred-1/2 DKO embryos were the same, LYVE-1-positive lymphatic vessels in the skin of Spred-1/2 DKO embryos increased compared with the control (Fig. 2H).

To further characterize LYVE-1-positive cells, we performed double immunofluorescence staining of LECs and BECs in the skin of E12.5 embryos with anti-LYVE-1 Ab and anti-CD31 Ab, respectively. CD31 is a panendothelial marker, but it has been reported that blood vessels were stained strongly with the anti-CD31 Ab while newly formed LYVE-1-positive lymphatic vessels were only weakly stained (3). As shown in Fig. 2I and J, compared with WT embryos, the number of CD31-positive BECs in the skin of Spred-1/2 DKO embryos was not significantly different, but there were more LYVE-1-positive LECs. Surprisingly, LYVE-1 and CD31 double-positive dilated vessels, which were never seen in WT embryos, were occasionally observed in the neck region of E12.5 Spred-1/2 DKO embryos (Fig. 2K and L). The number of LYVE-1 and CD31 double-positive vessels in Spred-1/2 DKO embryos was much higher than that in WT embryos (Fig. 2M). These data suggest that the developing lymphatic circulation in embryos lacking both Spred-1 and Spred-2 might communicate with blood circulation. Thus, LYVE-1 and CD31 double-positive dilated vessels could represent an incomplete separation of lymphatic vessels from blood vessels.

Expression of Spred-1 and Spred-2 in LECs and BECs.

To determine the expression of Spred-1/2 in developing lymphatics, we isolated LECs and BECs by using a cell sorter with anti-LYVE-1 and anti-CD34 Abs from E14.5 WT mouse embryos, as previously described (17) (Fig. 3A), and quantitative real-time PCR analysis was performed. As shown in Fig. 3B, LYVE-1⁺ CD34^low/− cells expressed the LEC markers Prox1 and podoplanin, but not the BEC markers neuropilin 1 (Nrp-1), CD44, and VEGFR-1, while LYVE-1⁻ CD34⁺ cells expressed BEC markers but not LEC markers. Thus, LECs and BECs were well separated by this procedure. Both Spred-1 and Spred-2 were expressed in LECs but not expressed in BECs (Fig. 3B). A small number of LYVE-1⁺ CD34⁺ DPs were observed in E14.5 embryos (Fig. 3B). DPs expressed both LEC and BEC markers, suggesting that this DP fraction could represent a transition state of LECs from BECs. Spred-1 and Spred-2 expression was high in this DP fraction (Fig. 3B). In contrast, Sprouty2 (Spry2) and Sprouty4 (Spry4), which have been implicated in angiogenesis (10, 15), were expressed in both LECs and BECs, but their expression was higher in BECs than in LECs (Fig. 3B).

Inhibition of in vitro LEC colony and capillary formation by Spreds.

Next, we examined the effect of Spred overexpression and loss of expression on the proliferation and differentiation of LECs. MACS-sorted CD45⁻ CD31⁺ endothelial cells from E14.5 embryos were cocultured on OP9 stromal cells in which VEGF-C/D was shown to be expressed (17). First, we confirmed that the efficacy of the infection, which was estimated by GFP, was almost 100%. Sendai virus-mediated Spred-1 gene transfer resulted in several-times-higher Spred-1 expression in endothelial cells than in control GFP-infected cells (Fig. 4A). Overexpression of Spred-1, but not control GFP, by the Sendai virus (16) resulted in a strong reduction of LYVE-1-positive LEC colonies (Fig. 4B). Conversely, LYVE-1-positive colony formation from E12.5 Spred1/2 DKO embryos was two to three times more than that from WT embryos (Fig. 4C). To confirm that the increase in LYVE-1-positive colonies in Spred-1/2 DKO embryos was specifically due to the lack of Spreds, we reduced LYVE-1-positive colonies in Spred-deficient cells by transducing exogenous Spred-1 into Spred-deficient embryonic cells (Fig. 4D). These data suggest that Spred-1 and Spred-2 negatively regulate the proliferation and/or differentiation of LECs.

We also examined the effect of Spred overexpression on VEGF-C signaling and lymphatic capillary formation in HULECs. As shown in Fig. 5, Spred-1 overexpression resulted in the suppression of ERK with or without VEGF-C and Akt phosphorylation with VEGF-C (Fig. 5A), as well as capillary formation of HULECs on Matrigel (Fig. 5B). These data suggest that Spred negatively regulates VEGF-C signaling and lymphatic vessel formation.

The VEGF-C/VEGFR-3 pathway is negatively regulated by Spreds.

It has been shown that the VEGF-C/VEGFR-3 pathway is essential for embryonic lymphatic vessel development (13). To determine whether Spred-1 and Spred-2 affect VEGF-C/VEGFR-3 signaling in vivo, we examined ERK activation by VEGF-C in situ using immunohistochemistry with Abs against P-ERK. Although we could not detect ERK phosphorylation in either E14.5 WT or Spred-1/2 DKO embryos at basal levels, stronger ERK activation by VEGF-C was observed in the LECs of the skin of Spred-1/2 DKO embryos than in those of WT embryos (Fig. 6A). The expression of VEGFR-3 and that of Prox-1 were detected equally in the LECs of Spred-1/2 DKO and WT mouse embryos (Fig. 6B and data not shown). These data suggest that Spred-1 and Spred-2 are physiological negative regulators of VEGF-C/VEGFR-3 signaling.

To further investigate the functions of Spreds on VEGF-C/VEGFR-3 signaling in vitro, we generated 293T cells that stably expressed human VEGFR-3 (293T/hVEGFR-3 cells). The overexpression of Spred-1 efficiently suppressed VEGF-C-induced ERK activation in 293T/hVEGFR-3 cells (Fig. 7A). We could not determine the effect of Spred-1 on Akt activation, since Akt was already highly activated in 293T cells before stimulation (data not shown). An Elk-1 reporter assay indicated that Spred-1 and Spred-2 showed similar suppressive effects on VEGF-C-induced ERK activation (Fig. 7B). Interestingly, Spred-1 and Spred-2 suppressed VEGFR-3-mediated ERK activation more strongly than Sprouty4 did (Fig. 7B), while Sprouty4 suppressed VEGF-A-mediated ERK activation as efficiently as Spreds did (Fig. 7C).

To define the effect of the loss of Spred-1/2 expression on VEGFR-3 signaling, we stably expressed hVEGFR-3 in MEFs derived from WT and Spred-1/2 DKO embryos. The expression levels of hVEGFR-3 were similar in WT and Spred-1/2 DKO MEFs (Fig. 8A). VEGF-C induced much stronger ERK and Akt activation in Spred-1/2 DKO MEFs than in WT MEFs (Fig. 8A). Recently, it has been shown that PDGF-BB potently induces the growth of lymphatic vessels in vivo (4). Thus, we examined PDGF-BB-induced ERK activation using WT and Spred-1/2-deficient MEFs (Fig. 8B). Only a slight enhancement of PDGF-BB-induced ERK activation was observed by Spred-1/2 deletion. Moreover, only a slight enhancement of VEGF-A-induced ERK activation was observed by Spred-1/2 deletion (Fig. 8C). These data suggest that the effect of Spred-1/2 deletion was more specific to VEGF-C than to PDGF-BB and VEGF-A. Consequently, the growth and migration of Spred-1/2 DKO MEFs were strongly enhanced by VEGF-C compared with WT MEFs (Fig. 8D and E). VEGF-C was not a strong mitogen in WT MEFs, while the proliferation of Spred-1/2 DKO MEFs was enhanced with VEGF-C (Fig. 8D). Similarly, the migration of Spred-1/2 DKO MEFs in response to VEGF-C was more evident than that of WT MEFs (Fig. 8E). These data indicate that Spred-1 and Spred-2 are important negative regulators of the VEGF-C/VEGFR-3 pathway.