Decreased Circulating Soluble Tie2 Levels in Preeclampsia May Result from Inhibition of Vascular Endothelial Growth Factor (VEGF) Signaling

Abstract

Objectives:

Recent studies have found dysregulation in circulating levels of a number of angiogenic factors and their soluble receptors in preeclampsia. In this study, we examined the mechanism of production of soluble Tie2 (sTie2) and its potential connection to the failure of vascular remodeling in preeclamptic pregnancies.

Design/Setting/Patients:

Serum samples were collected prospectively from 41 pregnant subjects at five different time points throughout pregnancy. Five of these subjects developed preeclampsia. For a second study, serum and placental samples were collected at delivery from preeclamptic and gestational age-matched controls. We examined serum sTie2 levels, and angiopoietin 1, angiopoietin 2, and Tie2 mRNA expression and localization in placental samples from the central basal plate area. We also examined the effects of vascular endothelial growth factor (VEGF) and a matrix metalloproteinase (MMP) inhibitor on proteolytic shedding of Tie2 in uterine microvascular endothelial cells.

Results:

Serum sTie2 levels were significantly lower in preeclamptic subjects starting at 24–28 wk of gestation and continued to be lower through the time of delivery. In culture experiments, VEGF treatment significantly increased sTie2 levels in conditioned media, whereas the MMP inhibitor completely blocked this increase, suggesting that VEGF-induced Tie2 release is MMP dependent.

Conclusions:

Our data suggest, for the first time, an interaction between VEGF and Tie2 in uterine endothelial cells and a potential mechanism for the decrease in circulating sTie2 levels in preeclampsia, likely through inhibition of VEGF signaling. Further studies on VEGF-Tie2 interactions during pregnancy should provide new insights into the mechanisms underlying the failure of vascular remodeling in preeclampsia and other pregnancy complications.

Although preeclampsia (PE) has been recognized as a disease since antiquity, its etiology and pathogenesis are still not fully understood. Pathological observations in PE indicate defective placentation and the failure of normal physiological transformation of spiral arteries, leading to the development of highly muscular, thick-walled arteries and resulting in reduced uteroplacental blood flow and placental insufficiency (1). However, the pathogenetic processes leading to the development of thick-walled arteries in PE remain unknown.

Numerous reports implicate angiopoietin 1 (Ang1), acting through its receptor Tie2, in regulating the recruitment of vascular smooth muscle cell precursors during arterial development and maturation of vessels in different systems (2). Angiopoietin 2 (Ang2) also binds to Tie2 but primarily acts as an antagonist of Ang1 signaling (2). Consistent with these findings, in an earlier study, we discovered a potential role for Ang1 in spiral artery growth in the primate endometrium (3). However, a number of recent studies have found no marked alteration in serum and placental levels of Ang1 in PE (4, 5), suggesting that changes in Ang1 expression may not be associated with the failure of spiral artery remodeling in PE.

Abnormal elevation of circulating levels of the soluble vascular endothelial growth factor (VEGF) receptor 1 [soluble fms-like tyrosine kinase-1 (sFlt1)] has been widely accepted as a biomarker of PE and as a contributor to the systemic endothelial dysfunction and clinical manifestations of the disease through its antagonism of VEGF activity (6–8). Because physiological changes in spiral arteries occur during pregnancy in the presence of invading trophoblasts, it is presumed that inadequate trophoblast invasion of spiral arteries may contribute to the pathology (1, 9). However, spiral arteries in the absence of trophoblasts, such as at nonimplantation sites and in ectopic pregnancies, also display these physiological changes (1), and sFlt1 levels increase in the circulation several weeks before the onset of the clinical symptoms of PE (7, 8). Thus, antagonism of VEGF activity by circulating sFlt1 is believed to underpin the failed arterial remodeling in preeclamptic pregnancies (1). However, there is no clear insight in the literature on how VEGF blockade may lead to inhibition of remodeling of these endometrial arteries because VEGF has not been directly implicated in the process of arterial growth and remodeling (arteriogenesis) (2).

There is emerging evidence that Tie2 can be proteolytically cleaved, resulting in the production of a 75-kDa soluble Tie2 receptor (sTie2) fragment, which binds to both Ang1 and Ang2 and inhibits angiopoietin-mediated Tie2 phosphorylation (10, 11). Furthermore, several matrix metalloproteinases (MMP) are known to be expressed in the endometrium and spiral artery walls at the putative time of spiral artery remodeling, such as menstruation and implantation (12, 13), and VEGF is known to stimulate MMP production in number of cell types, including endothelial cells and other endometrial cells (12–14). In the present study, we have demonstrated a novel MMP-dependent mechanism by which VEGF induces proteolysis of Tie2 in endometrial endothelial cells and may suppress angiopoietin signaling in endometrial vessels through release of sTie2. We have also shown that this signaling is highly relevant to the understanding of the mechanisms of failure of vascular remodeling in PE because circulating sTie2 levels decrease in tandem with increases in sFlt1, a potent inhibitor of VEGF, in preeclamptic pregnancies.

Materials and Methods

The experimental details, including human subjects, have been published previously (3, 6, 15) and are presented in the Supplemental Methods, published on The Endocrine Society's Journals Online web site at http://jcem.endojournals.org.

Human subjects

Forty-one healthy pregnant women receiving prenatal care were enrolled with an approved protocol. Five subjects developed PE, and seven gestational age-matched controls without any other pregnancy complications were selected from the remaining normotensive women who participated through term and provided all five serum samples from the different stages of pregnancy (<13 wk, 15–20 wk, 24–28 wk, 30–34 wk, and delivery). Details of clinical evaluations have been reported previously (6). PE was defined as a systolic blood pressure of 140 mm Hg or greater or diastolic blood pressure of 90 mm Hg or greater more than once along with significant proteinuria (see Supplemental Methods) (6). In a different cohort of preeclamptic patients (n = 19) and gestational age-matched controls (n = 24), serum and placental samples were collected. Placenta and placental bed biopsy samples were processed for real-time PCR, in situ hybridization, and immunohistochemistry studies.

Human uterine microvascular endothelial cell (UtMVEC) culture

UtMVEC were obtained from Lonza (Clonetics, Walkersville, MD) at passage 3 and cultured as described previously (15). Cells were grown in endothelial cell basal medium-2 (EBM-2) supplemented with the BulletKit (Clonetics). Confluent cells at passage 6 were treated with or without VEGF (25 ng/ml; R&D Systems, Minneapolis, MN) and/or GM6001 (an MMP inhibitor, 100 nmol/liter GM; Calbiochem, San Diego, CA) in the basal medium (EBM-2), and conditioned media were collected after 24 h.

Enzyme-linked immunosorbent assays

ELISA was performed on duplicate samples for determination of human sTie2 levels in serum and UtMVEC-conditioned medium using a commercial kit from R&D Systems as previously described (6).

Statistical analysis

The significance of differences between means was analyzed by t tests using the SPSS software (SPSS Inc., Chicago, IL) (6, 15). P < 0.05 was considered significant.

Results and Discussion

Tie2 is an endothelial cell-specific receptor tyrosine kinase and plays a critical role in vessel maturation and remodeling (2). Proteolytic cleavage of Tie2 has been shown to inhibit Tie2-mediated signaling in multiple systems (10, 11). Although their findings are contradictory, several recent studies show changes in circulating levels of Tie2 in preeclamptic pregnancies (16–18). However, the potential mechanisms of sTie2 production and the biological significance of changes in sTie2 levels during normal and abnormal pregnancy have not been described previously. Our results demonstrate that proteolytic shedding of Tie2 is the predominant mechanism of release of sTie2 from UtMVEC and that it is both constitutive and VEGF inducible. Furthermore, we showed significant decreases in circulating sTie2 levels in preeclamptic pregnancies, likely caused by inhibition of VEGF action through the increased sFlt1 levels in PE (7, 8). This is the first prospective, longitudinal study of gestational differences in maternal serum sTie2 levels between women who went on to develop PE and those who did not. Details of these results are presented and discussed below.

sTie2 shedding in UtMVEC is VEGF inducible and MMP dependent

sTie2 was detected in the conditioned medium of UtMVEC grown in EBM-2 (without supplementation with serum and growth factors), and its level increased significantly after VEGF treatment (Fig. 1). The effect of VEGF was specific because cotreatment of cells with both VEGF and a potent inhibitor of VEGF, sFlt1, completely abrogated the VEGF-induced increase in sTie2 in the conditioned medium (Supplemental Fig. 1). Because a number of publications have indicated the involvement of MMP in proteolytic shedding of receptor tyrosine kinases (RTK) (10, 11), including VEGF-induced sTie2 shedding in human umbilical vein endothelial cells (11), we examined the effects of the nonspecific MMP inhibitor GM6001 on sTie2 concentrations in the conditioned media of UtMVEC (Fig. 1). Consistent with earlier studies, our results showed significant decreases in both basal and VEGF-induced sTie2 levels after GM6001 treatment in the conditioned media of UtMVEC (Fig. 1). These findings demonstrate that, as for the shedding of other RTKs, both the constitutive and VEGF-induced shedding of sTie2 in UtMVEC are MMP dependent. Because VEGF and angiopoietins are known to act through distinct RTK on endothelial cells (2) and because several MMPs are known to be expressed in the endometrium and spiral artery walls at the putative time of spiral artery remodeling, such as menstruation and implantation (12, 13), our results, together with other published reports, suggest that VEGF signaling controls Ang1-Tie2 activity by promoting sTie2 shedding and thereby the loss of functional Tie2 receptors. Inhibition of Ang1 signaling, which normally acts to stabilize the vessels and promote arteriogenesis, thereby enables endometrial vascular remodeling during pregnancy (2). These findings, as discussed more fully below, have implications for understanding of the mechanisms underlying the failure of endometrial vascular remodeling in PE, which is associated with the inhibition of VEGF by the increased circulating levels of sFlt1.

Fig. 1.

sTie2 concentrations in the conditioned media of UtMVEC after various treatments. VEGF treatment significantly increases sTie2 levels in conditioned media, and these increases are completely blocked by cotreatment with GM6001, a nonspecific MMP inhibitor. The sTie2 level is also significantly reduced after GM6001 treatment alone, when compared with levels in controls, suggesting that constitutive sTie2 release is MMP dependent. a, b, and c, Significantly different from each other (P < 0.05).

Decreased shedding of sTie2 in preeclamptic pregnancies: potential cause for failure of vascular remodeling in PE

To determine whether circulating sTie2 concentrations are altered in preeclamptic pregnancies and whether they are linked to the failure of endometrial vascular remodeling, we obtained serum samples prospectively from 41 pregnant women and compared sTie2 levels of five subjects who developed PE with the levels in subjects who did not develop PE or any other pregnancy complication. In both the control and PE samples, the highest concentrations of sTie2 were observed in early pregnancy (<13 wk). Concentrations declined with increasing gestational age until 30–34 wk and then again increased to early pregnancy levels at delivery (Fig. 2A). However, compared with the levels in normal pregnancies, serum sTie2 levels in preeclamptic pregnancies were significantly decreased after 24–28 wk of gestation (Fig. 2A). For further confirmation, we compared serum sTie2 levels at delivery from 19 PE and 24 gestational age-matched control samples. Similar to the prospectively collected samples, these samples also had significantly lower levels of sTie2 when compared with controls (Fig. 2B). These results are consistent with Gotsch et al. (16) and Vuorela et al. (18), who reported decreased serum sTie2 levels in patients with preeclampsia. Additionally, consistent with previous reports (reviewed in Ref. 1), evaluation of placental bed biopsy specimens from the control and PE subjects revealed typical changes in the myometrial segment of spiral arteries (Supplemental Fig. 2).

Fig. 2.

Serum sTie2 levels at different stages of pregnancy. A, Tie2 levels in prospectively collected serum samples from subjects who delivered at term with or without preeclampsia. Note that, compared with its levels in controls, sTie2 levels in preeclamptic subjects were significantly (P < 0.05) lower starting at 24–28 wk of gestation and continued to be significantly (P < 0.05) lower through the time of delivery. B, sTie2 levels at delivery in a different cohort of preeclamptic subjects compared with its levels in gestational age-matched controls. Subjects with preeclampsia had significantly lower serum sTie2 levels (P < 0.01).

Our results show no marked changes in Ang1 and Tie2 mRNA levels (both cellular localization and total tissue content, Supplemental Fig. 3) in the placental (basal plate) samples from the control and preeclamptic subjects. There was also no significant change in total Ang2 mRNA levels between preeclamptic and control samples (Supplemental Fig. 3). However, cytotrophoblasts in some focal areas in the junctional zone (basal plate) and a few villi showed moderate increases in Ang2 expression in PE (Supplemental Fig. 3). Although reports on Ang2 levels are highly contradictory (4, 5, 17), consistent with our results, most previous studies indicate no significant changes in Ang1 or Tie2 levels in PE (4, 5). These results, together with our findings from UtMVEC studies, suggest that VEGF may antagonize Ang1-Tie2 signaling, which normally acts to stabilize the vessels, and promote a proangiogenic state in the placental bed during pregnancy through proteolytic shedding of Tie2 (MMP dependent) and not by regulation of the expression of Ang1 or Tie2. In PE, inhibition of VEGF action through increases in sFlt1 levels may lead to a reduction in the proteolytic shedding of Tie2, resulting in enhanced Ang1-Tie2 activity and the failure of normal spiral artery remodeling in the placental bed. In fact, preeclamptic decidual endothelial cells express lower levels of MMP1 than normal (19), and a recent study has found decreases in a number of MMP in the umbilical cord arterial wall in preeclamptic pregnancies and implicating this as a contributing cause for decreases in blood flow to the fetus in PE (20).

Because our sample size is small, further large-scale studies are required to validate our results and to determine the sensitivity and specificity of circulating levels of sTie2 as a biomarker for PE screening. However, our findings suggest that inhibition of VEGF action and the resulting reduction in proteolytic shedding of Tie2 in uterine endothelial cells may be a potential cause of the failure of the vascular remodeling in preeclamptic pregnancies.