Human CMV infection of endothelial cells induces an angiogenic response through viral binding to EGF receptor and β₁ and β₃ integrins

HCMV Infection Promotes an Angiogenic Response.

We documented that HCMV infection of ECs increased the expression of cell adhesion molecules, naïve monocyte recruitment, and EC permeability while concomitantly degrading lateral junction proteins (12). These HCMV-mediated EC changes typically are observed during angiogenesis (2, 17). Therefore, we asked the question: Does HCMV infection directly induce an angiogenic response? During angiogenesis, ECs proliferate, migrate to the site of damage and/or inflammation, and morph to form new capillary tubes (8, 17, 18).

We first investigated whether HCMV promoted EC proliferation using two distinct assays: (i) a MTS assay (Figs. 1A) and (ii) a total cellular number count (Fig. 1B). Viral-enhanced proliferation was observed as early as 24 h postinfection (hpi) and was significantly increased (P < 0.05) by 48 hpi (Fig. 1). By 96 h, mock-infected human microvascular ECs (HMECs) went through an average of two population doublings, whereas HCMV-infected HMECs went through an average of more than three population doublings, diverging from studies performed in fibroblasts, where HCMV has been documented to prevent cellular proliferation (22). HMECs treated with UV-irradiated HCMV also proliferated at a faster rate than mock-infected cells. Infection of HMECs with a range of multiplicities of infection (MOIs; 0.01–20) demonstrated that HCMV-induced EC proliferation occurred at a slower rate when lower MOIs were used to infect cells [supporting information (SI) Fig. 8]. Thus, viral-induced EC proliferation occurs at multiple MOIs and is not specific to only a high MOI. Furthermore, the results hint that different threshold levels of signaling in ECs may control different rates of proliferation. We also observed a significant increase in umbilical vein ECs in the S and G₂/M phases of the cell cycle by 48 and 72 hpi when compared with mock-infected cells (data not shown), providing further support for the enhanced proliferation we observed. Phorbol 12-myristate 13-acetate (PMA), a known angiogenic promoter (23), did not stimulate enhanced HMEC proliferation in our system.

In addition to proliferating, individual ECs also need to migrate to the site of vascular damage and/or to where new capillary tubes are to be formed. To investigate whether HCMV infection promoted EC motility, a scratch assay was performed on a confluent monolayer of HMECs (Fig. 2A; see SI Fig. 9A for images). The percentage scratch recovery was calculated at 12 hpi, a time when no significant differences in infected versus mock-infected HMEC cell numbers were observed (Fig. 1). All initial scratches were similar. By 12 hpi, HCMV-infected and UV-irradiated HCMV-treated HMECs showed a significant recovery (P < 0.05) of the initial scratch (≈80%), whereas mock-infected or PMA-stimulated cells recovered only 50% of the initial scratch. Second, phagokinetic random motility assays were performed to examine individual EC motility (Fig. 2B; see SI Fig. 9B for images). HCMV-infected and UV-irradiated HCMV-treated HMECs showed a significant (P < 0.05) 3- to 5-fold increase in motility compared with their mock counterparts (Fig. 2B). Last, ECs must be able to form new capillary tubes and blood vessels. Tubular morphogenesis assays revealed a significant (P < 0.05) 2- to 2.5-fold increase in the number of branches/capillary tubes occurred after HCMV infection, UV-irradiated HCMV treatment, or PMA treatment of HMECs (Fig. 3; see SI Fig. 10 for images).

ECs from different vascular beds could have different responses after infection (12); thus, we investigated whether HCMV infection of human macrovascular aortic ECs (HAEC) induced an angiogenic response. HCMV infection of HAECs increased cellular proliferation, motility, and morphogenesis similar to that seen with HMECs (data not shown). Together, these findings provide strong evidence that HCMV infection of ECs promotes the three hallmarks of angiogenesis. Because these changes occurred in the absence of new viral gene expression, the findings identify that receptor ligand-mediated signaling may be the key determinant triggering these EC changes.

PI3K and MAPK Signaling Mediates the Observed Angiogenic Response.

The regulation of EC function is controlled by outside-in signals (17, 18). The PI3K and MAPK pathways are two pathways involved in angiogenesis (17, 18) that also are activated by HCMV [in fibroblasts and monocytes (14, 15, 24, 25)]. To examine whether HCMV infection stimulated these pathways in ECs, Western blot analyses to examine levels of phosphorylated AKT (pAKT) and ERK (pERK) were performed on confluent, serum-starved mock-infected or HCMV-infected HMECs (Fig. 4). Between 10 and 60 min postinfection (min pi), levels of pAKT and pERK1/2 increased in infected cells when compared with levels observed in mock-infected ECs. By 120 min pi, levels of pAKT and pERK1/2 in HCMV-infected HMECs returned to that seen in mock-infected cells. No dramatic changes were seen in pan AKT or ERK levels. UV-irradiated HCMV-treated HMECs showed similar kinetics of PI3K activation to that of wild-type HCMV (data not shown).

To examine the role of HCMV-induced PI3K and MAPK activation on the observed angiogenic response, we used LY294002, to inhibit PI3K activity, and U0126, to inhibit the MAPK-ERK kinase (MEK1/2). Both drugs are specific for their respective pathways and nontoxic at the concentrations used (14, 24, 26). As described above, HCMV infection significantly (P < 0.05) increased HMEC proliferation, motility, and morphogenesis (Fig. 5 A–C), a response that was significantly (P < 0.05) abrogated when cells were treated with LY294002 or U0126 before infection. The solvent control (DMSO) had no effect on the HMEC response when added before infection (Fig. 5 A–C). The role these pathways played in the regulation of basal levels of EC function varied depending on the EC function examined. Regardless of the varying role these pathways play in basal levels of EC function, these data show HCMV-induced PI3K and MAPK activation is required for viral-mediated angiogenesis.

EGFR and Src Family Tyrosine Kinases Regulate the Induced Angiogenic Response.

Multiple receptors, including growth factor receptors (EGFR) and integrins (α_vβ₃ and α₂β₁), have been implicated in promoting angiogenesis (17, 18). These same receptors are reported to be required for HCMV entry, infection, and receptor-mediated signaling events in fibroblasts (14–16), suggesting a possible link between infection and the pathogenic consequences of infection of ECs. To decipher whether these cellular receptors, upstream of the PI3K and MAPK pathways, are responsible for triggering the switch to an angiogenic phenotype after infection, we performed Western blot analyses to examine pan and phosphorylated EGFR (pEGFR) and Src (pSrc) levels in confluent, serum-starved mock-infected or HCMV-infected HMECs (Fig. 4). The levels of pEGFR and pSrc increased in infected cells between 5 and 30 min pi. EGFR tyrosine kinase activation remained throughout the experiment, whereas Src activation was transient, returning to undetectable levels by 120 min pi. No dramatic changes were seen in pan EGFR or Src levels.

Next we used inhibitors of the EGFR tyrosine kinase (AG1478) and the Src family of tyrosine kinases (PP2) to show that these kinases functionally regulated the viral-mediated EC response. Both compounds are potent, specific inhibitors (14, 15) that are nontoxic at the concentrations used. As described above, infected HMECs and DMSO-treated, HCMV-infected HMECs proliferated (Fig. 6A), migrated (Fig. 6B), and formed capillary tubes (Fig. 6C) at significantly (P < 0.05) higher levels than their uninfected counterparts. Pretreatment with AG1478 or PP2 before infection significantly (P < 0.05) abrogated HCMV-induced HMEC proliferation and motility (Fig. 6 A and B). In contrast, pretreatment of HMECs with only AG1478 before infection significantly inhibited (P < 0.05) HCMV-mediated morphogenesis (Fig. 6C). Mock-infected HMECs pretreated with AG1478 or PP2 exhibited similar biological changes as untreated or DMSO-pretreated HMECs (Fig. 6 A–C). Together, these results suggest HCMV-induced activation of the EGFR tyrosine kinase was required for each of the viral-mediated angiogenic steps, whereas the activation of Src family tyrosine kinases was required only for HCMV-induced EC proliferation and motility.

Viral Binding to EGFR and the β₁ and β₃ Integrins Was Required for the Induced Angiogenic Response.

The data described above represent the initial finding that EGFR-mediated and integrin-mediated signaling triggers HCMV-induced EC changes. The direct role these receptors play in potential pathogenic cellular changes after direct binding of HCMV to ECs, however, has not been addressed. Previous work showed that HCMV entry into fibroblasts required viral binding to EGFR, α_vβ₃ integrin, α₂β₁ integrin, and α6β₁ integrin (14–16). Of these cellular receptors, only EGFR, α_vβ₃ integrin, and α₂β₁ integrin have been shown to be important in regulating angiogenesis (17, 18), suggesting their engagement during viral binding may be a key determinant in viral-mediated changes in ECs. It should be noted that the role EGFR plays in HCMV infection is controversial as a recent study reported that EGFR is not required for HCMV infection of fibroblasts and a single EC line (27), although the authors used a laboratory strain that lacks an intact U_LB′ region like the clinical/clinical-like isolates used here. Regardless, we find that EGFR-mediated and integrin-mediated signaling (via Src) is rapidly activated in ECs after infection (Fig. 4), suggesting their engagement after viral binding is intimately associated with functional changes in ECs. Therefore, we focused on the role viral binding to EGFR and the β-subunits of various integrins (β₁–β₄) played in HCMV-induced angiogenesis because of their role in regulating angiogenesis and with the exception of the β₂ integrins, are expressed on ECs (17, 18).

Using function-blocking antibodies to inhibit viral binding to specific cellular receptor (EGFR or the β₁–β₄ integrins) or an isotype-matched IgG₁ control antibody, we found that viral binding to EGFR and the β₁ and β₃ integrins was required for HCMV-induced angiogenesis. HCMV infection significantly (P < 0.05) increased HMEC proliferation (Fig. 7A), motility (Fig. 7B), and morphogenesis (Fig. 7C), which also occurred when infected HMECs were pretreated with function-blocking antibodies to the β₂ and β₄ integrins (SI Fig. 11A) or the isotype control (Fig. 7 A–C). When HMECs were pretreated with function-blocking antibodies specific for EGFR or the β₁ or β₃ integrins before infection, HMEC proliferation (Fig. 7A), motility (Fig. 7B), and morphogenesis (Fig. 7C) were significantly inhibited (P < 0.05) and returned to that observed in mock-infected cells, showing that viral binding to these receptors was required for the HCMV-induced angiogenic response. HMECs also were pretreated with function-blocking antibodies to vascular EC growth factor receptor 1 (VEGFR-1) and VEGFR-2 before infection. The HCMV-induced angiogenic response was not altered by the presence of the VEGFR-1 and VEGFR-2 neutralizing antibodies (SI Fig. 11B). When HMECs were pretreated with any of these antibodies and mock-infected, there was no significant change in EC function (Fig. 7 A–C). Together, these findings identify that HCMV binding to specific cognate receptors directly triggers an angiogenic response.

Cell and Virus Cultures.

HMECs and HAECs were cultured in endothelial growth medium (EGM; Clonetics). HCMV TB40/E-UL32 and Towne/E were cultured in fibroblasts and gradient-purified (12). UV-inactivated virus was prepared from gradient-purified virus (28) and did not express detectable IE gene products in fibroblasts (data not shown). ECs were infected at MOIs ranging from 0.01 to 20 based on infectivity of fibroblasts or mock-infected with media alone. One hundred percent of ECs were infected with both viruses using a MOI of 10–20 (12). Both viruses were used throughout the study with similar results; results from experiments with the TB40/E-UL32 virus are shown.

Proliferation Assays.

MTS assays (Promega) were performed by using the manufacturer's protocol. HMECs were grown to confluence, serum-starved for 24 h, and treated with preconditioned EGM for 24 h. Cells were harvested in preconditioned EGM, and 1 × 10³ cells were replated at subconfluent levels before mock infection, HCMV infection (MOI 0.01–20), or UV-irradiated HCMV treatment (equivalent MOI 20) (12). Total cell count assays were performed initially with 2.5 × 10⁴ HMECs treated as described above. Assays were performed at the beginning of the experiment out through 96 hpi.

In some experiments, HMECs were mock-infected or infected with HCMV (MOI 20) after a 45-min incubation with LY294002 (50 μM; Promega), U0126 (10 μM; Calbiochem), PP2 (1 μM; Calbiochem), AG1478 (1 μM; Calbiochem), or DMSO (Sigma). In other experiments, ECs were incubated with function-blocking antibodies (from Chemicon and R&D Systems specific for EGFR (1 μg/ml), the β₁–β 4 integrins (20 μg/ml), VEGFR-1 and VEGFR-2 (20 μg/ml), or the isotype-matched control antibody (20 μg/ml) for 1 h at 4°C before being plated and mock-infected or infected with HCMV (MOI 20).

Motility Assays.

HMECs were grown to confluence, serum-starved for 24 h, and treated with preconditioned EGM for 24 h. For the in vitro scratch assay (33), cells were injured with a sterile pipet tip, washed, and treated as described above. Immediately after infections and at 12 hpi, images of the scratches were captured for five random fields of view (FOV) along the scratch. By using ImageJ software, EC migration from the edge of the injured monolayer was quantified and the average scratch width was determined. Motility assays also were performed using the inhibitory drugs and function-blocking antibodies described above. For the phagokinetic motility assay (24), cells as treated above were harvested in preconditioned EGM, and 1 × 10³ cells were plated on colloidal gold-covered coverslips. At 12 hpi, cells were fixed and mounted in glycerol on slides. Track images of individual cells were captured at ×100 magnification, and the average area cleared per cell was determined by using Scion Image software.

Capillary Tube Formation Assay (34).

Twenty-four-well plates were coated with growth factor-reduced Matrigel (BD Biosciences). HMECs were grown to confluence, serum-starved for 24 h, and treated with preconditioned EGM for 24 h. ECs were harvested in preconditioned EGM, and 5 × 10⁴ HMECs were added to each well and treated as described above. At 12 hpi, images of capillary tube formation were captured at ×100 magnification. Fifteen random FOV were captured per treatment, and the average number of branches per FOV was determined. Morphogenesis assays were performed by using the inhibitory drugs and function-blocking antibodies described above. Apoptotic bodies were never observed in any of the different arms of the experiment.

Western Blot Analysis.

Cell lysates were harvested in Laemmli buffer (Bio-Rad). Equal sample amounts were separated by SDS/PAGE and transferred to PVDF membranes (Bio-Rad). After incubation with primary antibodies, blots were washed, incubated with horseradish peroxidase-conjugated secondary antibodies, and washed again. Protein was detected by Enhanced Chemiluminescence Plus (Amersham) following the manufacturer's protocol. Primary and secondary antibodies were obtained from Santa Cruz Biotechnology and Cell Signaling.

Statistical Analysis.

All experiments were performed in triplicate, and results are plotted as mean ± SD. The statistical significance between experimental means was determined with Student's t test, and differences of P < 0.05 were considered significant.