Secretion of VEGF-165 has unique characteristics, including shedding from the plasma membrane

Abstract

Vascular endothelial growth factor (VEGF) is a critical regulator of endothelial cell differentiation and vasculogenesis during both development and tumor vascularization. VEGF-165 is a major form that is secreted from the cells via a poorly characterized pathway. Here we use green fluorescent protein- and epitope-tagged VEGF-165 and find that its early trafficking between the endoplasmic reticulum and the Golgi requires the small GTP-binding proteins Sar1 and Arf1 and that its glycosylation in the Golgi compartment is necessary for efficient post-Golgi transport and secretion from the cells. The relative temperature insensitivity of VEGF secretion and its Sar1 and Arf1 inhibitory profiles distinguish it from other cargoes using the "constitutive" secretory pathway. Prominent features of VEGF secretion are the retention of the protein on the outer surface of the plasma membrane and the stimulation of its secretion by Ca(2+) and protein kinase C. Of importance, shedding of VEGF-165 from the cell surface together with other membrane components appears to be a unique feature by which some VEGF is delivered to the surroundings to exert its known biological actions. Understanding VEGF trafficking can reveal additional means by which tumor vascularization can be inhibited by pharmacological interventions.

FIGURE 1:

Dimerization, glycosylation, and secretion of biologically active VEGF165-GFP. Secreted and cell-associated VEGF165-GFP was analyzed by western blotting from transfected COS-7 cells using an anti-GFP antibody. (A) Under reducing conditions, VEGF165-GFP associated with the cells (“expressed”) shows both nonglycosylated and glycosylated forms, the latter being dominant. (B) Both cell-associated and secreted-fraction VEGF165-GFP form dimers under nonreducing conditions. The asterisk labels a degradation product that still shows dimerization. (C) Mutation of the single N-linked glycosylation site of VEGF165-GFP (GlycM) prevents glycosylation and greatly reduces secretion into the medium. Similarly, treatment of cells with 5 μg/ml Tn prevents glycosylation and also blocks secretion. (D) Secretion of VEGF165-GFP takes place at 20°C and is stimulated by 1 μM Iono. Cells were transfected with VEGF165-GFP for 24 h and incubated in fresh medium for the indicated times with or without Iono (1 μM) The medium was TCA precipitated and analyzed by Western blotting. Secretion of endogenous cyclophilin B is also stimulated by ionomycin. (E) Comparison of secreted and expressed VEGG165 between GFP-tagged and untagged proteins. COS-7 cells were transfected with the indicated constructs, and the secreted and cell-associated VEGF was analyzed as described in Materials and Methods. Note that a larger fraction of the untagged proteins is secreted compared with the GFP-tagged form.

FIGURE 2:

Cellular and in vivo actions of VEGF165-GFP. (A) The effect of conditioned medium (CM) on Ca²⁺ responses of HUVECs. Medium collected at 20°C (4 h) from COS-7 cells transfected for 24 h with empty vector, VEGF165-GFP, untagged VEGF165, or the GlycM form was added to HUVECs preloaded with Fura-2 for single-cell [Ca²⁺]_i measurements. Cells were also treated with the EGF receptor tyrosine kinase inhibitor AG1478 (1 μM) for 30 min to eliminate the effect of any EGF potentially secreted. Although VEGF165-GFP shows a smaller response than untagged VEGF165, it is still active, whereas the medium collected from GlycM cells is less effective. (B) VEGF165-GFP or control GFP sequences were cloned into a lentiviral backbone, and the resulting lentiviruses were injected into the left and right cerebral cortices of P0 rat pups, respectively. Pups were killed at P10, and fixed brain slices were immunostained with antibodies specific for GFP and RECA-1 to visualize virus-infected cells and cortical microvasculature, respectively. Low-magnification epifluorescence images demonstrate an intense, focal vasculogenic reaction around VEGF-GFP injection sites in contrast to control GFP injections. Scale bars, 100 μm. Note that the cells expressing VEGF165-GFP are not the endothelial cells that proliferate in response to VEGF. (B) Quantifications of microvessel density based on high-magnification confocal images randomly captured at VEGF165-GFP and GFP injection sites show that both the number of vessels and the average RECA-1–positive area/image is significantly higher after VEGF165-GFP injections than after control GFP injections, indicating strong biological activity of the fusion protein (30 pairs of images from three independent brains, ****p < 0.0001).

FIGURE 3:

Cellular distribution of VEGF165-GFP. (A) COS-7 or HUVECs were transfected with VEGF165-GFP and examined live by confocal microscopy. In both cell types a clear signal is visible over the Golgi and also associated with the plasma membrane. The enlarged image (bottom) shows a punctuate pattern at the plasma membrane. (B) Comparison of VEGF165-GFP signals inside and outside the cells. Cells were transfected with VEGF165-GFP and fixed with 3% paraformaldehyde. Immunostaining was then performed on nonpermeabilized cells with anti-GFP primary and Alexa Fluor 568–coupled secondary antibodies. Cells were analyzed by confocal microscopy, where the green channel shows the distribution of VEGF165-GFP and the red channel shows what is seen by the anti-GFP antibody. These studies suggest that VEGF165-GFP is associated with the outer surface of the plasma membrane. (C) Similar findings with a VEGF165-HA epitope–tagged protein fixed and stained under either permeabilized (left) or nonpermeabilized (right) conditions.

FIGURE 4:

Secretion of VEGF165-GFP requires Sar1 and Arf1 GTPases. COS-7 cells were transfected with VEGF165-GFP and the indicated GTPases for 24 h. Secreted VEGF165-GFP was then analyzed from fresh medium collected after 4-h incubation at 20°C. VEGF165-GFP was TCA precipitated and analyzed by Western blotting. (A) Top, the GTP-locked forms of Sar1 and Arf1 (Sar1-GTP, H79G; Sar1-GDP, T39N; Arf1-GTP, Q71L; Arf1-GDP, T31N) completely blocked VEGF secretion. Of note, the GDP-locked form of Arf1 was strongly inhibitory, but that of Sar1 was without effect. Arf6-GDP (T27N) had no effect, whereas Arf6-GTP (Q67L) exerted partial inhibition. Remarkably, endogenous cyclophilin B secretion was influenced quite differently: the GDP-locked forms of Arf6, Arf1, or Sar1 had little or no effect, and the GTP-locked forms of both Arf1 and Arf6 showed enhancement. The graph shows the results of quantification, where the secreted/cell-associated ratios were found and normalized to control values. Data were pooled from three experiments with VEGF165-GFP and three experiments with VEGF165-GFP since they showed identical results (mean ± SEM, n = 6; **p < 0.005, ***p < 0.001, assessed by paired t test). (B) Distribution of VEGF165-GFP cotransfected with GDP- and GTP-locked forms of Arf1 and Sar1. The GTP-locked forms of either Arf1 or Sar1 caused accumulation of VEGF165-GFP in the ER. The GDP-locked form of Arf1 increased the intracellular fraction of VEGF165-GFP, but patches still can be observed in the plasma membrane. The GDP-locked form of Sar1 showed VEGF165-GFP distribution characteristic of control cells.

FIGURE 5:

The post-Golgi trafficking of VEGF165-GFP. COS-7 cells were transfected with VEGF165-GFP for 24 h. After change of medium, cells were treated with the indicated drugs for 4 h at 20°C. The medium was collected, and VEGF165-GFP was TCA precipitated and analyzed by Western blotting. (A) Top, release of VEGF165-GFP was increased in cells treated with either 1 μM Iono or 100 nM PMA. Treatment with Tn or BFA (both at 5 μg/ml) blocked VEGF165-GFP secretion. The secretion of endogenous cyclophilin B increases with Iono treatment and is only slightly inhibited by BFA but not Tn treatment at 20°C. The columns show mean ± SEM from 7–10 experiments for VEGF168-GFP (*p < 0.05, **p < 0.005 vs. control assessed by paired t test). (B) Distribution of VEGF165-GFP after treatment with selected chemicals. COS-7 cells were transfected on coverslips with VEGF165-GFP for 24 h, and its distribution was observed live by confocal microscopy after treatment with the indicated drugs at 37°C for 5 h. Treatment with 1 μM (Iono) reduced VEGF165-GFP association with the plasma membrane. Tunicamycin treatment causes accumulation of VEGF165-GFP in the Golgi, with no signal at the plasma membrane. In cells treated with 100 nM PMA, VEGF165-GFP is cleared from the Golgi, and its presence in the periphery is also reduced. Note the Golgi localization and strong presence on punctuate structures on the plasma membrane in control cells.

FIGURE 6:

Effect of PtdIns(4,5)P₂ depletion and rapamycin on VEGF165-GFP secretion. The amount of PtdIns(4,5)P₂ in the plasma membrane was acutely decreased by the use of a recruitment system based on the heterodimerization of the FRB domains of mTOR and the FKBP12 protein as described previously (Varnai et al., 2006; Hammond et al., 2012). Here rapamycin was used to recruit the 5-phosphatase domain of INPP5E to the plasma membrane. (A) COS-7 cells were transfected with VEGF165-GFP, a PM-targeted FRB, and either the FKBP12-fused 5-phosphatase domain or FKBP-12 for 24 h. After change of medium, cells were treated with 1, 10, and 30 nM rapamycin for 4 h at 20°C. The medium was collected, and VEGF165-GFP was TCA precipitated and analyzed by Western blotting. The graph shows quantification (mean ± SEM) from three independent experiments. Rapamycin treatment strongly inhibited secretion, but only in the presence of the 5-phosphatase (**p < 0.05 vs. control assessed by paired t test). Paradoxically, rapamycin increased secretion when the FKPB-only control construct was used. (B) Secretion of VEGF165-GFP also showed small but consistent increases after rapamycin treatment without expression of other constructs, although this reached significance only at 10 nM concentration (mean ± SEM, n = 4).

FIGURE 7:

Shedding of VEGF165-GFP from PM. (A) COS-7 cells were transfected with VEGF165-GFP and a PM-targeted mRFP construct and analyzed live by confocal microscopy. VEGF165-GFP shows localization in Golgi and punctate structures on the cell surface. The enlarged image (right) shows that VEGF165-GFP is enriched in contact points where the cell attaches to the matrix and even at puncta outside the cell. (B) Electron microscopy from COS-7 cells or glia cells shows that VEGF165-GFP–positive membranes bud off and shed from PM (scale bars, 200 nm). (C) Isolation of membrane vesicles by ultracentrifugation from conditioned medium obtained from COS-7 cells transfected with VEGF165-GFP and stimulated with PMA. After removal of cell debris and detached cells with low-spin centrifugation, the medium was subjected to ultracentrifugation (see Materials and Methods for details). The supernatant (soluble fraction) was TCA precipitated, and the pellet (shed particles) was resuspended in Laemmli sample buffer. The samples were analyzed by Western blotting, and the graph shows the summary of quantification from three independent experiments. Note the comparable increase in both the particulate and soluble fractions after stimulation. The arrows point to shorter and longer exposure of the left and right GFP blots, respectively. (D) Fibronectin (FN) was knocked down by treatment of COS-7 cells with a specific siRNA for 72 h, followed by transfection with VEGF165-GFP for 24 h. After change of the medium, cells were incubated for 4 h at 20°C and the medium collected for TCA precipitation. Both VEGF165-GFP and FN were analyzed by Western blotting. Note that VEGF165-GFP was still secreted even though FN expression and secretion were substantially reduced.