Endothelial Rictor is crucial for midgestational development and sustained and extensive FGF2-induced neovascularization in the adult

Abstract

To explore the general requirement of endothelial mTORC2 during embryonic and adolescent development, we knocked out the essential mTORC2 component Rictor in the mouse endothelium in the embryo, during adolescence and in endothelial cells in vitro. During embryonic development, Rictor knockout resulted in growth retardation and lethality around embryonic day 12. We detected reduced peripheral vascularization and delayed ossification of developing fingers, toes and vertebrae during this confined midgestational period. Rictor knockout did not affect viability, weight gain, and vascular development during further adolescence. However during this period, Rictor knockout prevented skin capillaries to gain larger and heterogeneously sized diameters and remodeling into tortuous vessels in response to FGF2. Rictor knockout strongly reduced extensive FGF2-induced neovascularization and prevented hemorrhage in FGF2-loaded matrigel plugs. Rictor knockout also disabled the formation of capillary-like networks by FGF2-stimulated mouse aortic endothelial cells in vitro. Low RICTOR expression was detected in quiescent, confluent mouse aortic endothelial cells, whereas high doses of FGF2 induced high RICTOR expression that was associated with strong mTORC2-specific protein kinase Cα and AKT phosphorylation. We demonstrate that the endothelial FGF-RICTOR axis is not required during endothelial quiescence, but crucial for midgestational development and sustained and extensive neovascularization in the adult.

Figure 1. Constitutive homozygous endothelial Rictor knockout during embryonic development is generally lethal.

(A) Rictor^flox/flox females were mated with Rictor^flox/−; VE-Cadherin-Cre^+/−; LacZ reporter^+/+ males to generate homozygous deletion of Rictor in the endothelium. Litter genotypes were determined by qPCR and are displayed as total distribution and average number of pups per genotype (n_{total pups} = 101, ****P < 0.0001, ***P < 0.001, 1-way ANOVA with Bonferroni multiple comparison) (B). The abovementioned breeding scheme was used to isolate embryonic day (E) 10.5 wildtype and endothelial Rictor knockout embryos. Representative β-galactosidase staining (blue) of E10.5 embryos shows the active sites of VE-Cadherin-Cre recombination. Arrows on the right indicate reduced peripheral LacZ staining in Rictor knockout embryos (n = 7 of 11).

Figure 2. Lethality and growth retardation of induced endothelial Rictor knockout mice peaks around E12.

(A) Tamoxifen (Tx) injection scheme: Three doses of Tx were administered to pregnant females every 48 hours, beginning on E6.5, E7.5, E8.5, E10.5, E12.5, and E14.5. Pregnant females were sacrificed, and embryos were harvested on E17.5 for further histological analysis. Grey area symbolically depicts stepwise Rictor knockout. Control females were injected with corn oil. (B) Rictor^iΔec mice exhibit decreased litter size. Statistical analysis of the litter size of individual breedings (n_Eday) after three and two Tx injections at different time points (n_E6.5 = 3, n_E7.5 = 3, n_E8.5 = 4, n_E10.5 = 3, n_E12.4 = 4, n_E14.5 = 4,) compared to controls (n = 7). *P < 0.05, Student’s t-test. (C) Statistical analysis of normal, growth retarded, and absorbed embryos per breeding after Tx-injection at different time points. Total litter size for breeding at each time point was set to 100%, n_breedings = 3. (D) Statistical analysis of the length of surviving embryos per litter at different starting time points of Tx injections. n = 4 (embryos per time point, length was measured in both extremities), *P < 0.01 compared to controls. Student’s t-test. (E) Representative picture of a E17.5 Rictor^iΔec embryo that was induced by Tx on E8.5 in comparison to a wildtype embryo. (F) Rictor^iΔec embryos displayed distinct vascular deficits in the eye when Tx injections started on E7.5 and embryos harvested at E12.5 and stained with the vessel-specific antibody, endomucin. Scale bar = 100 μm. Arrows indicate angiogenic sprouts. (G) Rictor^iΔec embryos display a delay in ossification. Representative pictures of the upper limbs of embryos stained with alizarin red (bone) and alcian blue (cartilage). Arrows: ossification centers. Below, quantification with number of ossification centers in fingers upon knockout of Rictor at indicated starting time points of Tx injections. *P < 0.05, **P < 0.01, compared to E14.5; n_embryo = 4 (number of centers was measured in both extremities and averaged for each embryo), Mann-Whitney Rank Sum Test.

Figure 3. Rictor knockout does not affect weight gain and viability in adolescent mice.

(A) Body weights were followed in male and female mice over a period of 27 weeks after induction of knockout or control on week 4. n = 10 per genotype and gender, n.s., 2-way ANOVA with group-wise comparison. All mice displayed normal health, behavior, and viability. (B) At the end of the experiment, RNA was extracted from endothelial layer for quantitative polymerase chain reaction (PCR) analysis to test for efficient excision of Rictor (n = 3/2, *P < 0.05. 2-tailed T-test), and qualitatively for purity of endothelial tissue (endothelial marker Pecam1, smooth muscle marker αSMA). Total aorta mRNA was used as comparative control. Representative immunostainings for estrogen receptor 2 (red fluorescence) in histological sections of the skinfold from 10-week-old Rictor^iΔec mice demonstrates specific expression of CreER^T2 recombinase associated with capillaries.

Figure 4. Rictor knockout in mouse aortic endothelial cells (MAEC) decreases endothelial network formation.

(A) Representative micrographs show endothelial network formation after 18 hours of seeding. Quantification (total number of master segments that connect to at least two other segments) of endothelial tube formation from three experiments is shown below. (Bars; mean and SEM, n_exp = 3, *P < 0.05 versus control. Paired T-Test). FGF-treated Rictor ko MAEC typically formed star-shaped centers with omni-directional sprouting and no connection to neighboring centers. Efficient Rictor knockout displayed by RICTOR and CRE protein expression from control (AdCre–) and Rictor ko (AdCre+) MAEC (n = 3) is displayed on the lower right.

Figure 5. FGF2 amplifies RICTOR protein and Rictor-dependent phosphorylation of AKT on serine 473 and PKCa on serine 657.

(A) Western blots show RICTOR and downstream targets of mTORC2 after 15 min stimulation of control and Rictor ko MAEC with 5–50 ng/ml of FGF2, 10% FCS or 1 μg/ml insulin (Ins). Significant increase in RICTOR protein at 5 ng/ml of FGF2 stimultion peaking at an 8-fold expression at 50 ng/ml of FGF2 compared to diluent in 3 repeated experiments in a MAEC isolate (control = white bars, n_exp = 3, *P < 0.05, **P < 0.001, 1-way ANOVA with Bonferroni multiple comparison, upper left panels). PKCα protein was nearly absent in Rictor ko MAEC compared to control (middle panels). Densitometric quantification shows dose-dependent AKT ^Ser473 phosphorylation in response to FGF2, FCS and Ins (right panels, n_exp = 3, *P < 0.05, **P < 0.01, ***P < 0.001, repeated measures ANOVA). Lower middle blots show S6K1 phosphorylation on P^Thr389 compared to total S6K1 and phosphorylation of ERK1/2 on P^{Thr202/Tyr204} compared to total ERK1/2 after Rictor knockout. (B) mRNA expression after 1–6 hours stimulation (25 ng/ml FGF2) of control and Rictor ko MAEC of VEGF receptor 1 (mFLT1, sFlt1), VEGF receptor 2 (Kdr), VEGFA (Vegfa), FGF receptor 1 (Fgfr1) and protein kinase Cα (PKCα) detected by quantitative real-time PCR (n = 3, ns knockout versus control, repeated measures ANOVA). (C) Absolute proliferation values (Absorption = A4_50 nm-A_650 nm) in FGF2-stimulated control (open circles) or Rictor ko (filled squares) MAEC are presented. (n_exp = 3, n.s., repeated measures ANOVA). In response to 10% FCS (right), proliferation of Rictor ko MAEC was significantly lower compared to that of control MAEC (n = 3, *P < 0.05, two-tailed T-test). (D) Migration of MAEC was measured after FGF2 (25 ng/ml) or diluent administration for 1, 3, 6 and 9 hours (wound completely closed = 1, n = 3, n.s., repeated measures ANOVA).

Figure 6. Rictor knockout disables a sustained increase in skin capillary diameters and restricts extensive capillary remodeling in response to FGF2.

(A) Stacked frames of representative videos of the capillary vasculature were recorded through the dorsal skinfold chamber by intravital fluorescence microscopy. Baseline capillary bed of the skin muscle was assessed in 10-week-old mice (6 weeks after knockout induction) by intravital microscopy through the unmodified dorsal skinfold chamber. Upper (20 × magnification) and lower micrographs (10×) display representative skin capillary beds of control (left) and Rictor^iΔec (right) mice. Capillary diameters from 10 control and Rictor^iΔec mice were quantified and pooled for statistical analysis (n_capillaries = 253/206, no differences between groups, 2-tailed T-test and Whisker plot, indicating median, 25% percentile and total range to the right). Color scheme applies for whole Fig. 6 and is displayed in upper right corner. B. Wounding response of the capillary bed. Skin muscle capillary structure from day 1, 2, 4 and day 7 in control (left) and Rictor^iΔec mice (right) after wound sealing with heparin-containing matrigel (20 × magnification). Capillary diameters from 4 control and Rictor^iΔec mice were quantified and pooled for statistical analysis by 1-way ANOVA followed by Bonferroni multiple comparison (Scatter plot with medians on the below; number of capillaries (n_capillaries) are indicated below the X-axis, ****P < 0.0001). C. Heterogeneous capillary diameter increase in the FGF2-stimulated capillary bed. Skin muscle capillary structure from day 1, 2, 4 and day 7in control (left) and Rictor^iΔec mice (right) after wound sealing with FGF2 (1.5 μg/ml; heparin-containing matrigel, 20 × magnification). Capillary diameters from 7 control and Rictor^iΔec mice were quantified and pooled for statistical analysis by 1-way ANOVA followed by Bonferroni multiple comparison (Scatter plot and medians below; number of capillaries are indicated below the X-axis, ****P < 0.0001). D. Capillary remodeling in the FGF2-stimulated capillary bed. The line graph displays the normalized distribution of capillaries (n_max = 100) resolved in a 1-μm range after FGF2 stimulation for 7 days and illustrates differences in remodeling between groups. Micrograph to the right shows 10x magnification of the vascular bed of control and Rictor^iΔec mice after 7 days of FGF2 exposure.

Figure 7. Fibroblast growth factor 2(FGF2)-induced angiogenesis in matrigel plugs is reduced in Rictor^iΔec mice.

(A) 1^st row: Representative matrigel plugs containing heparin (diluent) or 1.5 μg/ml FGF2 with heparin (FGF2) from 8-week-old male control (left) and Rictor^iΔec (right) mice removed 7 days after injection (scale bar = 1mm). Estimation of blood content by optical densitometry is shown on the right (n_plugs = 4; **P < 0.01; 1-way ANOVA with Bonferroni multiple comparison). (A) 2nd row: Paraffin sections from corresponding plugs immunostained for CD31 (brown) and hematoxylin (blue/nuclei). Representative micrographs show 10 × magnification and display the depth of newly in grown microvessels from the surface towards the center of the matrigel plugs. Quantification to the right displays the significant reduction in the ingrowth (μm) of neovessels into plugs from Rictor^iΔec mice compared to control mice (n_plugs = 4/7; **P < 0.01, ****P < 0.0001; 1-way ANOVA with Bonferroni multiple comparison). (A) 3^rd row: Representative micrographs of peripheral stroma covering matrigel plugs. Identifiable inner microvessel diameters were measured (graph to the right; n_plugs = 3; *P < 0.05;2-tailed T-test ). No vessels were found in peripheral stroma covering diluent-containing plugs. (A) 4^th row: Representative micrographs of macrophage marker CD68-immunestainings of peripheral stroma and matrigel. Graphs to the right show ratio of CD68⁺/total cell nuclei in the stroma, and average of CD68⁺cells per field counted in the stroma (n_plugs = 4, no significant differences found after 1-way ANOVA/Bonferroni multiple comparison). (B) Representative micrographs of one set of experiments displaying hematoxylin and eosin stained (H&E) matrigel areas showing local leakage and hemorrhagic areas in FGF2 containing plugs from control mice compared to plugs from Rictor^iΔec mice (upper micrographs). Lower micrographs show higher magnification of CD31-stained matrigel areas. Arrowheads point to local spots of leaked erythrocytes in FGF2-containg control plugs. (C) Confluent and starved control and Rictor ko MAEC were stimulated for 1 and 24 hours with 25 ng/ml FGF2 or diluent. mRNA expression of monocyte attracting protein 1 (Mcp1), vascular and inducible cell adhesion molecules 1 (Vcam1, Icam1) was detected by quantitative real-time PCR (n = 3; *P < 0.05; 1-way ANOVA with Bonferroni multiple comparison test).