Pathogenic Hantaviruses Andes Virus and Hantaan Virus Induce Adherens Junction Disassembly by Directing Vascular Endothelial Cadherin Internalization in Human Endothelial Cells

Abstract

Hantaviruses infect endothelial cells and cause 2 vascular permeability-based diseases. Pathogenic hantaviruses enhance the permeability of endothelial cells in response to vascular endothelial growth factor (VEGF). However, the mechanism by which hantaviruses hyperpermeabilize endothelial cells has not been defined. The paracellular permeability of endothelial cells is uniquely determined by the homophilic assembly of vascular endothelial cadherin (VE-cadherin) within adherens junctions, which is regulated by VEGF receptor-2 (VEGFR2) responses. Here, we investigated VEGFR2 phosphorylation and the internalization of VE-cadherin within endothelial cells infected by pathogenic Andes virus (ANDV) and Hantaan virus (HTNV) and nonpathogenic Tula virus (TULV) hantaviruses. We found that VEGF addition to ANDV- and HTNV-infected endothelial cells results in the hyperphosphorylation of VEGFR2, while TULV infection failed to increase VEGFR2 phosphorylation. Concomitant with the VEGFR2 hyperphosphorylation, VE-cadherin was internalized to intracellular vesicles within ANDV- or HTNV-, but not TULV-, infected endothelial cells. Addition of angiopoietin-1 (Ang-1) or sphingosine-1-phosphate (S1P) to ANDV- or HTNV-infected cells blocked VE-cadherin internalization in response to VEGF. These findings are consistent with the ability of Ang-1 and S1P to inhibit hantavirus-induced endothelial cell permeability. Our results suggest that pathogenic hantaviruses disrupt fluid barrier properties of endothelial cell adherens junctions by enhancing VEGFR2-VE-cadherin pathway responses which increase paracellular permeability. These results provide a pathway-specific mechanism for the enhanced permeability of hantavirus-infected endothelial cells and suggest that stabilizing VE-cadherin within adherens junctions is a primary target for regulating endothelial cell permeability during pathogenic hantavirus infection.

Hantaviruses cause 2 human diseases: hemorrhagic fever with renal syndrome (HFRS) and hantavirus pulmonary syndrome (HPS) (50). HPS and HFRS are multifactorial in nature and cause thrombocytopenia, immune and endothelial cell responses, and hypoxia, which contribute to disease (7, 11, 31, 42, 62). Although these syndromes sound quite different, they share common components which involve the ability of hantaviruses to infect endothelial cells and induce capillary permeability. Edema, which results from capillary leakage of fluid into tissues and organs, is a common finding in both HPS and HFRS patients (4, 7, 11, 31, 42, 62). In fact, both diseases can present with renal or pulmonary sequelae, and the renal or pulmonary focus of hantavirus diseases is likely to result from hantavirus infection of endothelial cells within vast glomerular and pulmonary capillary beds (4, 7, 11, 31, 42, 62). All hantaviruses predominantly infect endothelial cells which line capillaries (31, 42, 44, 61, 62), and endothelial cells have a primary role in maintaining fluid barrier functions of the vasculature (1, 12, 55). Although hantaviruses do not lyse endothelial cells (44, 61), this primary cellular target underlies hantavirus-induced changes in capillary integrity. As a result, understanding altered endothelial cell responses following hantavirus infection is fundamental to defining the mechanism of permeability induced by pathogenic hantaviruses (1, 12, 55).

Pathogenic, but not nonpathogenic, hantaviruses use β₃ integrins on the surface of endothelial cells and platelets for attachment (19, 21, 23, 39, 46), and β₃ integrins play prominent roles in regulating vascular integrity (3, 6, 8, 24, 48). Pathogenic hantaviruses bind to basal, inactive conformations of β₃ integrins (35, 46, 53) and days after infection inhibit β₃ integrin-directed endothelial cell migration (20, 46). This may be the result of cell-associated virus (19, 20, 22) which keeps β₃ in an inactive state but could also occur through additional regulatory processes that have yet to be defined. Interestingly, the nonpathogenic hantaviruses Prospect Hill virus (PHV) and Tula virus (TULV) fail to alter β₃ integrin functions, and their entry is consistent with the use of discrete α₅β₁ integrins (21, 23, 36).

On endothelial cells, α_vβ₃ integrins normally regulate permeabilizing effects of vascular endothelial growth factor receptor-2 (VEGFR2) (3, 24, 48, 51). VEGF was initially identified as an edema-causing vascular permeability factor (VPF) that is 50,000 times more potent than histamine in directing fluid across capillaries (12, 14). VEGF is responsible for disassembling adherens junctions between endothelial cells to permit cellular movement, wound repair, and angiogenesis (8, 10, 12, 13, 17, 26, 57). Extracellular domains of β₃ integrins and VEGFR2 reportedly form a coprecipitable complex (3), and knocking out β₃ causes capillary permeability that is augmented by VEGF addition (24, 47, 48). Pathogenic hantaviruses inhibit β₃ integrin functions days after infection and similarly enhance the permeability of endothelial cells in response to VEGF (22).

Adherens junctions form the primary fluid barrier of endothelial cells, and VEGFR2 responses control adherens junction disassembly (10, 17, 34, 57, 63). Vascular endothelial cadherin (VE-cadherin) is an endothelial cell-specific adherens junction protein and the primary determinant of paracellular permeability within the vascular endothelium (30, 33, 34). Activation of VEGFR2, another endothelial cell-specific protein, triggers signaling responses resulting in VE-cadherin disassembly and endocytosis, which increases the permeability of endothelial cell junctions (10, 12, 17, 34). VEGF is induced by hypoxic conditions and released by endothelial cells, platelets, and immune cells (2, 15, 38, 52). VEGF acts locally on endothelial cells through the autocrine or paracrine activation of VEGFR2, and the disassembly of endothelial cell adherens junctions increases the availability of nutrients to tissues and facilitates leukocyte trafficking and diapedesis (10, 12, 17, 55). The importance of endothelial cell barrier integrity is often in conflict with requirements for endothelial cells to move in order to permit angiogenesis and repair or cell and fluid egress, and as a result, VEGF-induced VE-cadherin responses are tightly controlled (10, 17, 18, 32, 33, 59). This limits capillary permeability while dynamically responding to a variety of endothelial cell-specific factors and conditions. However, if unregulated, this process can result in localized capillary permeability and edema (2, 9, 10, 12, 14, 17, 29, 60).

Interestingly, tissue edema and hypoxia are common findings in both HPS and HFRS patients (11, 31, 62), and the ability of pathogenic hantaviruses to infect human endothelial cells provides a means for hantaviruses to directly alter normal VEGF-VE-cadherin regulation. In fact, the permeability of endothelial cells infected by pathogenic Andes virus (ANDV) or Hantaan virus (HTNV) is dramatically enhanced in response to VEGF addition (22). This response is absent from endothelial cells comparably infected with the nonpathogenic TULV and suggests that enhanced VEGF-induced endothelial cell permeability is a common underlying response of both HPS- and HFRS-causing hantaviruses (22). In these studies, we comparatively investigate responses of human endothelial cells infected with pathogenic ANDV and HTNV, as well as nonpathogenic TULV.

Pathogenic hantaviruses enhance VEGFR2 phosphorylation.

Pathogenic hantaviruses enhance the permeability of endothelial cells in response to VEGF days after infection (22), suggesting that hantavirus infection impacts both VEGFR2 and adherens junction disassembly. Here, we investigated whether pathogenic hantavirus infection of endothelial cells resulted in an increase in the phosphorylation of VEGFR2. Human umbilical vein endothelial cells (HUVECs; Cambrex) were mock infected or infected with pathogenic ANDV or HTNV or nonpathogenic TULV at a multiplicity of infection (MOI) of 0.5. Three days postinfection, cells were serum starved and then stimulated with VEGF (100 ng/ml) for 30 min or left unstimulated (control). Endothelial cells were lysed (37), and cells were comparably infected with ANDV, HTNV, and TULV by immunoblot analysis of the hantavirus nucleocapsid protein (Fig. 1A). Following VEGFR2 immunoprecipitation, the amounts of total and phosphorylated VEGFR2 were comparatively analyzed (Fig. 1A and B). Interestingly, VEGF addition to endothelial cells infected with HTNV or ANDV resulted in the hyperphosphorylation of VEGFR2. We quantified VEGFR2 phosphorylation using NIH Image, and Fig. 1B indicates that there is a 3- to 4-fold increase in the level of VEGFR2 phosphorylation following infection by ANDV or HTNV over that of mock- or TULV-infected controls. These data indicate that VEGF addition to pathogenic hantavirus-infected endothelial cells results in the hyperphosphorylation of VEGFR2. As a result, our findings suggest that pathogenic hantaviruses increase pathway-specific responses to VEGF by enhancing VEGFR2 activation.

FIG. 1.

Pathogenic hantaviruses enhance VEGFR2 phosphorylation responses. HUVECs were mock infected or infected with pathogenic HTNV and ANDV or nonpathogenic TULV at a multiplicity of infection (MOI) of 0.5. Three days postinfection, cells were starved overnight in endothelial basal medium 2 (EBM-2; 0.5% bovine serum albumin [BSA] without growth factors) and subsequently treated with VEGF (100 ng/ml) for 30 min or left untreated (control). Endothelial cells were harvested in lysis buffer (25 mM Tris [pH 7.8], 150 mM NaCl, 2 mM EDTA, 1% NP-40, 10 mM NaF, 2 mM Na₃V0₄, and protease inhibitor cocktail [Sigma]). Cell lysates were clarified by centrifugation and immunoprecipitated (IP) with anti-VEGFR2 rabbit polyclonal antibody (C-1158; Santa Cruz Biotechnology) and protein A/G plus agarose beads (Santa Cruz Biotechnology). Precipitated proteins were fractionated by 7% SDS-PAGE. (A) Immunoblotting (IB) was performed with antibodies against phosphotyrosine (4G10) (Upstate Biotechnology) or anti-VEGFR2 monoclonal antibody (sc-6251; Santa Cruz Biotechnology) and detected using horseradish peroxidase-conjugated anti-mouse and anti-rabbit secondary antibodies (Amersham) and fluorography with enhanced chemiluminescence (Amersham). Control cells were similarly lysed and fractionated by 7% SDS-PAGE, and immunoblotting was performed with antibodies against anti-β-tubulin (Santa Cruz Biotechnology) or antinucleocapsid protein (N protein) as previously described (23). (B) An NIH Image analysis of immunoblotted bands was used to quantitate phosphorylated VEGFR2, and the results are levels normalized to constant amounts of tubulin present on blots.

VE-cadherin internalization is enhanced in pathogenic hantavirus-infected endothelial cells.

VEGFR2 activation controls the internalization of VE-cadherin and thereby regulates barrier functions of endothelial cell adherens junctions. Here, we determined whether pathogenic hantavirus infection resulted in the enhanced internalization of VE-cadherin in response to VEGF addition. Endothelial cells were grown on gelatin-coated coverslips and infected with HTNV, ANDV, and TULV at an MOI of 0.5 or mock infected. Three days postinfection, the expression of hantavirus nucleocapsid protein was evaluated by Western blot analysis to demonstrate that endothelial cells were comparably infected by the three viruses (Fig. 2A). Identically infected endothelial cells were starved and subsequently treated with VEGF (100 ng/ml for 30 min at 37°C) or left untreated. Cells were incubated with antibody to the extracellular domain of VE-cadherin (BV9; Clonetics) at 4°C for 1 h in order to tag extracellular VE-cadherin and block internalization (17). After antibody removal and washing, cells were incubated at 37°C (1 h) to permit cellular trafficking of VE-cadherin. Cells were subsequently washed with a mild acid solution (2 mM phosphate-buffered saline [PBS]-glycine, pH 2.0; three times for 5 min) as described by Gavard and Gutkind (17) in order to remove VE-cadherin antibody that was not internalized; alternatively, cells were left untreated. Cells were paraformaldehyde fixed and Triton X-100 permeabilized prior to incubation with a fluorescein isothiocyanate (FITC)-tagged anti-mouse secondary antibody and examined on an Olympus IX51 microscope.

FIG. 2.

Pathogenic hantaviruses promote VE-cadherin internalization. HUVECs were grown on gelatin-coated coverslips and mock infected or infected with HTNV, ANDV, and TULV at an MOI of 0.5. (A) Three days postinfection, cells were lysed with radioimmunoprecipitation assay buffer (0.1% SDS) and anti-β-tubulin (Santa Cruz Biotechnology), and viral nucleocapsid levels (antinucleocapsid rabbit polyclonal serum [23]) were evaluated by Western blot analysis, as described in the legend to Fig. 1 (46). (B) Three days postinfection, cells were starved overnight in EBM-2 (0.5% BSA without growth factors) and subsequently treated with VEGF (100 ng/ml for 30 min) or mock treated (control). Endothelial cells were subsequently incubated with anti-VE-cadherin antibody, clone BV9 (specific for the extracellular domain of human VE-cadherin; Santa Cruz Biotechnology), in ice-cold EBM-2 (0.5% BSA) at 4°C for 1 h. Unbound antibody was removed by washing coverslips with ice-cold PBS, and cells were shifted to a 37°C CO₂ incubator for 1 h to permit VE-cadherin internalization (17, 18). Cells were left untreated or washed with a mild acid solution (2 mM PBS-glycine, pH 2.0; three times for 5 min) to remove VE-cadherin antibody that was not internalized (17, 18). Cells were washed with PBS, fixed in 4% paraformaldehyde (10 min), permeabilized with 0.25% Triton X-100 (5 min), and incubated with secondary anti-mouse FITC-labeled antibody (Jackson Labs) for 1 h (17, 18). Coverslips were mounted using the SlowFade kit (Molecular Probes) and examined using an Olympus IX51 fluorescence microscope. Results are representative of 3 to 5 independent experiments.

In the absence of VEGF, VE-cadherin is present on the surface of infected and uninfected endothelial cells (Fig. 2B, −acid wash, −VEGF), since acid washing completely removed VE-cadherin from the surface of cells (Fig. 2B, +acid wash, −VEGF). However, following VEGF treatment and subsequent acid washing, there is a dramatic increase in the level of intracellular VE-cadherin (acid washed) within endothelial cells infected by pathogenic ANDV and HTNV (Fig. 2B, +VEGF, +acid wash). This reflects the internalization of VE-cadherin into compartments inaccessible by acid. In contrast, VEGF addition to mock- or TULV-infected cells demonstrates that VE-cadherin is nearly entirely present on the cell surface and sensitive to acid treatment (Fig. 2B, +VEGF, +acid wash). In Fig. 3A, the number of cells with internalized VE-cadherin following acid treatment was quantitated. In the absence of VEGF treatment, the number of intracellular vesicles containing VE-cadherin in mock-infected or hantavirus-infected cells was low and did not exceed 5% (Fig. 3A). However, VEGF stimulation led to the intracellular accumulation of VE-cadherin (acid resistant) in >70% of cells infected with pathogenic ANDV and HTNV but in only 10 to 15% of VE-cadherin-positive mock- or TULV-infected cells.

FIG. 3.

Quantification of VE-cadherin internalization and endothelial cell permeability. (A) Endothelial cells were infected and treated as described in the legend to Fig. 2B. The internalization of endogenous VE-cadherin was quantified using NIH Image to detect VE-cadherin fluorescence (BV9 and FITC, as described for Fig. 2B). The percentage of cells with at least one group of five or more acid-resistant, VE-cadherin-positive vesicle-like structures (internalized VE-cadherin) was quantitated and compared to that of mock, VEGF-untreated, acid-washed controls (17, 18). Findings are from three independent experiments (n = 500 cells) which were analyzed by analysis of variance, and asterisks indicate a significant difference at a P value of <0.001 with respect to unstimulated cells. Results are expressed as the means ± standard errors of the mean (SEM) of three independent experiments. (B) HUVECs were seeded onto vitronectin-coated (10 μg/ml) Costar Transwell inserts (6.5-mm diameter and 3-μm pore size; Corning) at a cell density of 2 × 10⁴ cells/well and grown in EBM-2 (10% of fetal calf serum). Confluent HUVECs were infected in triplicate with pathogenic HTNV or ANDV or nonpathogenic TULV at an MOI of 0.5 or mock infected. Three days postinfection, cells were starved overnight in EBM-2 (0.5% bovine serum albumin without growth factors). FITC-dextran (0.5 mg/ml) was added to the upper chamber in the presence or absence of VEGF (100 ng/ml, 30 min at 37°C) as previously described (22). The level of FITC-dextran in the lower chamber was quantitated using a Perkin-Elmer fluorimeter (490-nm excitation, 530-nm emission).

VE-cadherin internalization experiments were performed in parallel with a previously described endothelial cell permeability assay (22). Briefly, endothelial cells were seeded onto vitronectin-coated Costar Transwell inserts, and confluent cells were infected with HTNV, ANDV, or TULV at an MOI of 0.5. Three days postinfection, FITC-dextran was added to the upper chamber in the presence or absence of VEGF. The amount of FITC-dextran in the lower chamber was assayed using a BioTek 8100 fluorimeter. Figure 3B indicates that VEGF addition to pathogenic ANDV- and HTNV-infected cells results in a dramatic increase in the permeability of endothelial cells compared to the results with mock- or TULV-infected cells. These findings support a prior demonstration that only pathogenic hantavirus-infected endothelial cells are hyperpermeabilized by VEGF addition. In addition, these results demonstrate that the enhanced permeability of ANDV- and HTNV-infected endothelial cells in response to VEGF occurs concomitantly with an increase in VE-cadherin internalization.

Hantavirus-directed VE-cadherin internalization is inhibited by Ang-1 and S1P.

Angiopoietin-1 (Ang-1) and the platelet-derived lipid mediator sphingosine-1-phosphate (S1P) were previously shown to inhibit the ANDV- and HTNV-induced hyperpermeability of endothelial cells in response to VEGF (22). Here, we analyzed the effect of Ang-1 and S1P on VE-cadherin internalization within ANDV- and HTNV-infected cells. Endothelial cells were infected with HTNV, ANDV, or TULV as described above, and 3 days postinfection cells were pretreated with Ang-1 (50 ng/ml) or S1P (0.01 μM) for 5 min before VEGF addition (Fig. 4 A and B). After 3 h, VE-cadherin internalization was assayed and quantitated as described in the legends to Fig. 2B and 3A. Figure 4 indicates that Ang-1 and S1P reduced internalization of VE-cadherin in cells infected by pathogenic HTNV and ANDV to the levels of internalized VE-cadherin present in mock- or TULV-infected cells. These data suggest that Ang-1 and S1P inhibit the permeability of hantavirus-infected endothelial cells by blocking signaling responses that induce VE-cadherin internalization. These results further suggest that adherens junction disassembly may be therapeutically targeted in order to inhibit hantavirus-induced endothelial cell permeability.

FIG. 4.

Ang-1 and S1P prevent VE-cadherin internalization in hantavirus-infected HUVECS. HUVECs were mock infected or infected with pathogenic HTNV and ANDV or nonpathogenic TULV at an MOI of 0.5. Three days postinfection, cells were starved overnight in EBM-2 (0.5% bovine serum albumin without growth factors) and pretreated with Ang-1 (50 ng/ml) 5 min prior to VEGF addition (100 ng/ml, 30 min) or treated with S1P (1 mM) and VEGF (100 ng/ml, 30 min) (22). Cells were incubated with anti-VE-cadherin antibody and acid washed, fixed, and stained with secondary FITC-labeled antibody as described in the legend to Fig. 2 (17, 18). The percentage of total acid-washed cells exhibiting VE-cadherin fluorescence is presented (±SEM; n = 500; P < 0.001 versus mock-treated, acid-washed controls).

Pathogenic hantaviruses have in common the ability to infect endothelial cells and increase capillary permeability (31, 42, 44, 61, 62). Our findings tie the enhanced permeability of endothelial cells infected by pathogenic hantaviruses (22) to increases in VEGFR2 activation and downstream signaling responses which direct VE-cadherin internalization. This defines a pathway that contributes to hantavirus-induced endothelial cell permeability. However, these findings do not resolve the mechanism by which pathogenic hantaviruses enhance VEGFR2 activation, since a multitude of factors regulate VEGFR2 activation, VE-cadherin internalization, and endothelial cell permeability.

VEGFR2 activation is responsible for activating a number of pathways with individual and overlapping effects on cell proliferation, cell migration, actin remodeling, focal adhesion formation, cell survival, and vascular permeability (5, 8, 10, 12, 33, 37, 43, 48, 51, 56, 60). As a result, VEGFR2 activation and signaling responses are highly complex and further integrated with pathways activated by diverse intracellular and extracellular signals, which positively or negatively regulate discrete VEGFR2 responses (17, 18, 58, 59, 63). The VEGFR2 receptor itself contains at least 5 primary tyrosine residues which are phosphorylated under different circumstances and linked alone or in tandem with specific signaling pathway activation responses (17, 25). Although we have demonstrated an increased VEGFR2 phosphorylation response in hantavirus-infected endothelial cells, it remains to be determined whether individual phosphorylation sites specify this response.

VEGFR2-directed endothelial cell permeability results from the activation of a signaling cascade which is linked to VE-cadherin phosphorylation and changes in the actin cytoskeleton, which forms an intracellular anchor for VE-cadherin and stabilizes adherens junctions (17, 18). This pathway is tightly regulated in order to prevent vascular leakage (17, 18, 58, 59, 63) and further complicated by nine potential phosphorylated tyrosine sites on VE-cadherin and roles for VE-phosphatase (VE-PTP) in regulating the phosphorylation state of both VE-cadherin and VEGFR2 (41). Thus, hantaviruses could increase VEGFR2 phosphorylation by recruiting factors which enhance VEGFR2 phosphorylation, preventing VE-PTP association with VEGFR2, preventing VEGFR2 degradation, or altering VEGFR2 binding to additional endothelial cell receptors (5, 8, 10, 12, 33, 37, 43, 48, 51, 56, 60).

VEGFR2 reportedly forms a complex with activated β₃ integrins, and knocking out β₃ integrin functions results in a 2- to 3-fold increase in endothelial cell permeability in response to VEGF addition (3, 24, 48). Pathogenic hantaviruses bind inactive forms of β₃ and block endothelial cell migration on β₃ integrin ligands, suggesting that at least part of the enhanced endothelial cell permeability caused by pathogenic hantaviruses may be due to β₃ integrin dysregulation (21, 23, 39, 46). However, hantaviruses cause a much greater change in endothelial cell permeability than can be ascribed solely to β₃ integrin dysfunction (48), and additional regulation of VEGFR2 responses by viral proteins remain to be investigated.

Ang-1 and S1P are additional inhibitors of VEGFR2-directed permeability (16, 18, 27, 28, 40, 45, 49, 54, 59). We have shown that these factors block hantavirus-directed endothelial cell permeability and in these studies demonstrated that they accomplish this by ultimately inhibiting VE-cadherin internalization (22). These factors bind unique endothelial cell receptors and block VEGF permeability downstream of VEGFR2 activation (16, 18, 27, 28, 40, 45, 49, 54, 59). Although the mechanism by which these factors impact VEGFR2-VE-cadherin signaling pathways remain to be defined, understanding how these factors block hantavirus-induced endothelial cell permeability are likely to provide additional mechanisms for therapeutic intervention in hantavirus-induced permeability. These findings rationalize studies of hantavirus interactions with cellular pathways that regulate VEGFR2-VE-cadherin signaling responses in order to define the mechanism of hantavirus-induced permeability and are fundamental to understanding capillary permeability and edema induced by hantaviruses.