Loss of Thrombomodulin in Placental Dysfunction in PreeclampsiaSignificance
Objective—Preeclampsia is a pregnancy-specific syndrome characterized by placental dysfunction and an angiogenic imbalance. Systemically, levels of thrombomodulin, an endothelium- and syncytiotrophoblast-bound protein that regulates coagulation, inflammation, apoptosis, and tissue remodeling, are increased. We aimed to investigate placental thrombomodulin dysregulation and consequent downstream effects in the pathogenesis of preeclampsia.
Approach and Results—Placentas from 28 preeclampsia pregnancies, 30 uncomplicated pregnancies, and 21 pregnancies complicated by growth restriction as extra controls were included. Immunohistochemical staining of thrombomodulin, caspase-3, and fibrin was performed. Placental mRNA expression of thrombomodulin, inflammatory markers, matrix metalloproteinases 2 and 9, and soluble Flt-1 were measured with quantitative polymerase chain reaction. Thrombomodulin mRNA expression was determined in vascular endothelial growth factor–transfected trophoblast cell lines. Thrombomodulin protein and mRNA expression were decreased in preeclampsia as compared with both control groups (P=0.001). Thrombomodulin mRNA expression correlated with maternal body mass index (P<0.01) and diastolic blood pressure (P<0.05) in preeclampsia. An increase in placental apoptotic cells was associated with preeclampsia (P<0.001). Thrombomodulin expression correlated positively with matrix metalloproteinase expression (P<0.01) in preeclampsia, but not with fibrin deposits or inflammatory markers. Placental soluble Flt-1 expression correlated with decreased thrombomodulin expression. Vascular endothelial growth factor induced upregulation of thrombomodulin expression in trophoblast cells.
Conclusions—Decreased thrombomodulin expression in preeclampsia may play a role in placental dysfunction in preeclampsia and is possibly caused by an angiogenic imbalance. Hypertension and obesity are associated with thrombomodulin downregulation. These results set the stage for further basic and clinical research on thrombomodulin in the pathogenesis of preeclampsia and other syndromes characterized by endothelial dysfunction.
Preeclampsia complicates 2% to 8% of all pregnancies and is a leading cause of maternal and fetal morbidity and mortality.1,2 The exact pathogenesis of the syndrome is unknown, but impaired placentation plays a key role in its pathogenesis.3 In preeclampsia, placental production of antiangiogenic factors such as soluble Flt-1 (sFlt-1) and soluble endoglin is increased, which results in deprivation of essential survival factors in the endothelium and syncytiotrophoblast.3 This contributes to the development of endothelial dysfunction, causing increased systemic vascular resistance, high blood pressure, and hypercoagulability.3 Further, a hyper-inflammatory state develops, characterized, for example, by complement deposits in the kidneys and in the placenta.4,5
Thrombomodulin is essential for the maintenance of endothelium; it is a transmembrane glycoprotein found on the endothelium of arteries, venules, and capillaries and on the syncytiotrophoblast in the placenta.6 Inflammation results in the release of thrombomodulin from endothelial cells into the circulation, leading to decreased expression of thrombomodulin on the endothelial surface.7 Whether this cleaved form of thrombomodulin is functional remains unclear, but soluble thrombomodulin concentrations can be used to monitor inflammatory conditions.8 In women with preeclampsia, maternal serum levels of cleaved soluble thrombomodulin in serum are elevated.9–11 The placenta is a possible source of this soluble thrombomodulin, which could result in decreased available thrombomodulin on the syncytiotrophoblast.12 Vascular endothelial growth factor stimulates thrombomodulin expression,8 so the angiogenic imbalance in the placenta in preeclampsia could lead to a decrease in placental thrombomodulin expression.
One of the pathways through which thrombomodulin exerts its protective effects on endothelium is by activating protein C. In mice, loss of thrombomodulin from the endothelium results in disruption of the activated protein C (APC) pathway, which leads to lethal massive thrombosis.13 APC has an anticoagulant function by cleaving activated cofactors FV and FVIII. In addition to its anticoagulant effects, APC has an important cytoprotective function for endothelium through the endothelial protein C receptor: namely, direct anti-inflammatory and antiapoptotic effects on endothelial cells.14 In an experimental mouse model, this antiapoptotic effect protected against diabetic nephropathy by inhibiting endothelial apoptosis.15 Furthermore, APC plays a role in tissue remodeling by activating matrix metalloproteinase (MMP) 2 and 9, which are involved in placental development and are dysregulated in preeclampsia.16–18
We set out to investigate whether placental thrombomodulin dysregulation plays a role in the pathogenesis of preeclampsia. Therefore, we measured thrombomodulin expression in placentas from women with preeclampsia and compared this with thrombomodulin expression in placentas from control subjects with uncomplicated pregnancies and in placentas from control subjects with pregnancies complicated by intrauterine growth restriction (IUGR) to confirm that our findings were preeclampsia-specific rather than a consequence of lower gestational age or birth weight in our case group. Further, we investigated whether thrombomodulin expression in the placenta is correlated with its downstream effects: fibrin depositions, cell survival, inflammation, and MMP expression in the placenta, and whether the angiogenic imbalance of preeclampsia is associated with a decrease in thrombomodulin expression.
Materials and Methods
Materials and Methods are available in the online-only Data Supplement.
Diastolic blood pressure and proteinuria were significantly higher in the preeclampsia group as compared with both control groups. The mean gestational age in healthy control subjects was 39 weeks and 4 days; in IUGR control subjects and in preeclampsia patients, the mean gestational ages were 33 weeks and 2 days and 30 weeks and 4 days, respectively, which were both significantly lower as compared with healthy control subjects. The mean birth weight in the healthy control group was 3609 g. Mean birth weights in the IUGR control group and in the preeclampsia group were 1521 g and 1167 g, respectively, which were both significantly lower compared with the mean birth weight in the healthy control group. The mean placenta weight was significantly lower in preeclampsia patients and in growth-restricted control subjects as compared with healthy control subjects. Patient characteristics are illustrated in the Table.
Immunohistochemical Staining for Thrombomodulin
In 29 of 30 placentas from healthy controls, thrombomodulin was observed in an overall staining pattern, with thrombomo dulin present on the syncytiotrophoblast of >50% of villi, and in 1 control placenta, a focal staining pattern was observed, with thrombomodulin being present on the syncytiotrophoblast of <50% of villi. In the IUGR group, thrombomodulin was present on the syncytiotrophoblast of >50% of villi in all 11 cases. In 12 of 28 placentas from patients with preeclampsia, thrombomo dulin protein expression was found to be decreased. In 10 cases, we found a focal staining pattern, with thrombomodulin present on the syncytiotrophoblast of <50% of villi, and in 2 cases, thrombomodulin was nearly absent (present on the syncytiotrophoblast of <10% of villi). In the remaining 16 placentas from the preeclampsia group, thrombomodulin staining was present in an overall staining pattern. These findings are illustrated in Figure 1. Chi-square analysis indicated a strong association between decreased thrombomodulin expression and preeclampsia (P=0.001). Thrombomodulin staining on fetal vessel endothelium was present in a similar pattern in preeclampsia and control groups. Figure I in the online-only Data Supplement shows thrombomodulin staining on fetal vessel endothelium in placentas from preeclampsia and control cases with most reduced thrombomodulin staining on the syncytiotrophoblast; fetal vessel thrombomodulin staining was present in these samples and consequently serves as an internal control for the adequacy of the staining procedure.
Placental mRNA Expression of Thrombomodulin
Placental mRNA levels of thrombomodulin were on average 3-fold lower in preeclamptic women compared with control subjects (P=0.001), whereas in women with IUGR, the relative mRNA expression of thrombomodulin was not significantly different compared with healthy control placentas. These results are illustrated in Figure 2. In placenta samples where thrombomodulin protein expression was absent, thrombomodulin mRNA levels were 4-fold lower than those in placentas with a focal or overall staining pattern for thrombomodulin at the syncytiotrophoblast in the preeclampsia group (Figure II in the online-only Data Supplement).
Placental mRNA Expression of Thrombomodulin and Clinical Parameters in Preeclampsia
Placental thrombomodulin mRNA levels were significantly lower in patients with mild preeclampsia and in patients with severe preeclampsia compared with healthy control subjects; levels were not significantly different between patients with mild and severe preeclampsia. Placental thrombomodulin mRNA levels were negatively correlated with maternal diastolic blood pressure and maternal body mass index in patients with preeclampsia. In controls, this correlation was not present. There were no correlations between gestational age or placental weight and thrombomodulin mRNA levels. These correlations are illustrated in Figure 2. Thrombomodulin mRNA levels were not associated with parity (P>0.05). When exclusively primiparous cases were analyzed, thrombomodulin mRNA levels were still significantly lower in preeclampsia cases compared with placentas from control cases.
In placentas from healthy control subjects, villous fibrin deposits were present on 13% of the villi, and in IUGR control, placentas on 7%. In pregnancies complicated by preeclampsia, fibrin depositions were present on 17% of villi, on average. One-way analysis of variance analyses indicated an association between increased fibrin deposits and preeclampsia, but post hoc analyses did not reveal any significant differences between preeclampsia cases and healthy or growth-restricted control cases. These data are shown in Figure 3. There was no significant association between fibrin deposits and thrombomodulin mRNA or between fibrin deposits and the thrombomodulin protein staining pattern.
Immunohistochemical Staining of Cleaved Caspase-3
In placentas from healthy subjects and IUGR control subjects, there were 4 and 3 apoptotic cells per mm2, on average. In placentas from preeclamptic patients, there were on average 16 apoptotic cells per mm2, which was significantly different from both control groups. These data are shown in Figure 3. There was no significant association between thrombomodulin mRNA and the amount of apoptotic cells, and a focal or absent thrombomodulin staining pattern was not associated with an increase in apoptotic cells.
Villitis was present in 14.8% of the control placentas and in 42.9% of the preeclampsia cases; this difference was significant (P<0.05). In all villitis cases, the infiltrate was multifocal and composed of mononuclear cells. Intervillositis occurred in 7.4% of control placentas and in 19% of preeclampsia placentas; the infiltrate consisted of histiocytes in the majority of cases. Neither villitis nor intervillositis was associated with a decreased thrombomodulin staining pattern or with decreased thrombomodulin mRNA expression. Placental mRNA levels of tumor necrosis factor-α and intercellular adhesion molecule 1 were lower in placentas from preeclampsia patients as compared with healthy controls (P<0.001 for both; Figure 3). There was no correlation between thrombomodulin mRNA expression and tumor necrosis factor-α or intercellular adhesion molecule 1 expression in placentas from preeclamptic patients (Figure III in the online-only Data Supplement).
Placental mRNA Expression of MMPs
On average, MMP-2 mRNA levels were 2-fold lower in placentas from preeclamptic women compared with placentas from both control groups (P<0.01). MMP-9 mRNA levels were 5-fold lower in placentas from preeclamptic women compared with control subjects (P<0.01). In IUGR control placentas, MMP-2 and MMP-9 mRNA levels were not significantly different compared with control placentas. In preeclamptic patients, MMP-2 expression was positively correlated with thrombomodulin mRNA expression. The correlation between MMP-9 and thrombomodulin did not reach statistical significance (P=0.095). These findings are illustrated in Figure 3.
Placental Expression of sFlt-1 and Thrombomodulin
An inverse correlation between thrombomodulin protein expression and sFlt-1 expression was observed; in placentas from preeclampsia patients with a focal syncytiotrophoblast thrombomodulin staining pattern, mRNA expression of sFlt-1 was 3-fold higher compared with the expression in placentas with an overall thrombomodulin staining pattern at the syncytiotrophoblast (P<0.05; Figure 4). This inverse correlation was not observed between thrombomodulin mRNA and sFlt-1 mRNA.
In Vitro Experiments
In BeWo cells, vascular endothelial growth factor transfection resulted in a 1.5-fold upregulation of thrombomodulin mRNA expression. This increase was significantly (P<0.05) compared with untreated control cells and control cells that had received the transfection reagent without plasmid DNA. In Jeg-3 cells, vascular endothelial growth factor transfection was associated with a 1.3-fold upregulation of thrombomodulin mRNA expression, but this was not significantly different when compared with both control cell groups. These results are illustrated in Figure 5.
Increasing evidence suggests that preeclampsia is caused by deprivation of factors essential for the placenta and endothelium, with consequent placental and endothelial dysfunction. This study demonstrates that thrombomodulin mRNA and protein expression are decreased in placentas of mildly and severely preeclamptic patients, and that this decrease correlates negatively with maternal body mass index and diastolic blood pressure. Additionally, thrombomodulin mRNA expression correlates directly with placental expression of MMPs, and the decrease in placental thrombomodulin is accompanied by impaired villous cell survival.
In the preeclampsia group, parity, birth weight, placenta weight, and duration of pregnancy were significantly different compared with those of the term control group. Therefore, we added an extra control group to our study, consisting of placentas from pregnancies complicated by growth restriction and not by hypertensive disorders. In this group, parity, birth weight, placenta weight, and duration of pregnancy were comparable to those of the preeclampsia group. Thrombomodulin expression in the growth-restriction control group was not different from the expression in the term control group; therefore, we conclude that our findings on placental thrombomodulin loss in the preeclampsia group cannot be explained by, for example, the lower gestational age in the preeclampsia group.
A significant correlation was present between placental thrombomodulin expression and diastolic blood pressure in preeclampsia cases. This indicates that thrombomodulin expression is connected to the degree of endothelial dysfunction in preeclampsia, which could be upstream, in the pathogenesis of endothelial dysfunction, or downstream, as a consequence of endothelial dysfunction. However, thrombomodulin expression was not significantly different between mild and severe preeclampsia cases, so placental loss of thrombomodulin alone does not seem to be a major contributor to the development of end-organ involvement in preeclampsia.
If thrombomodulin is involved in the pathogenesis of preeclampsia, one would expect mutations in the thrombomodulin gene to be associated with the syndrome. For example, hemolytic uremic syndrome, a disease also characterized by endothelial dysfunction and complement activation, is associated with mutations in the thrombomodulin gene.19 However, a recent meta-analysis of genetic variants in preeclampsia revealed that thrombomodulin mutations are not associated with preeclampsia.20 Our study shows that a decreased thrombomodulin protein expression pattern in the placenta is accompanied by an increase in placental expression of sFlt-1, and that vascular endothelial growth factor transfection is associated with thrombomodulin upregulation in trophoblast cells in vitro. These results strengthen the hypothesis that the angiogenic imbalance of preeclampsia could cause the decreased thrombomodulin expression in the placenta that we found in our cohort.
The decreased thrombomodulin expression in the placenta in preeclampsia could have unfavorable downstream effects on the placenta through at least 4 pathways, which are illustrated in a hypothetical scheme in Figure 6.
First, thrombomodulin inhibits coagulation through the APC pathway. Systemically, increased levels of the thrombin–antithrombin complex in the presence of increased thrombomodulin cleavage product have been reported in preeclampsia,21 but we did not find an increase in placental fibrin deposits in preeclampsia or a correlation between thrombomodulin expression and placental villous fibrin deposits in our study; apparently, the placental decrease in thrombomodulin we observed in our cohort did not lead to increased fibrin at the fetomaternal interface. Second, we found that decreased placental thrombomodulin expression was accompanied by impaired placental cell survival in preeclampsia; these changes could also contribute to the impaired placentation seen in preeclampsia. However, a direct correlation between the amount of apoptotic cells detected with immunohistochemical staining and thrombomodulin expression was not present, possibly caused by sampling error because of the heterogeneous nature of placental lesions. In vitro studies investigating the effects of thrombomodulin depletion on trophoblast cells would shed some light on this area. Third, although preeclampsia itself is associated with a hyper-inflammatory state,4,5 we found no increased mRNA expression of intercellular adhesion molecule-1 and tumor necrosis factor-α in placentas from preeclamptic patients, and there was no correlation between thrombomodulin expression and these parameters. Further, thrombomodulin expression was not associated with villous- or intervillous infiltrates. Last, we found a positive correlation between thrombomodulin expression and expression of MMP-2 and MMP-9. These results indicate that thrombomodulin expression could lead to decreased MMP-2 and MMP-9 activity in the placenta, which is associated with impaired trophoblast invasion and preeclampsia.22 Because our cohort contained exclusively third-trimester placentas obtained after delivery, early stages of placental development, and trophoblast invasion, were not studied. Further studies on thrombomodulin expression in the placenta at earlier stages of placental development are warranted.
In our cohort, we found a negative correlation between body mass index and placental thrombomodulin expression. Obesity is associated with endothelial dysfunction and increased levels of circulating thrombomodulin, suggestive for loss of endothelial thrombomodulin.23 Obese pregnant women show increased levels of circulating markers of endothelial dysfunction and have an increased risk of developing preeclampsia.24 Possibly, the highly inflammatory and hypoxic vascular environment of obesity leads to decreased expression of vasculoprotective factors, such as thrombomodulin, both on the fetomaternal interface and in the systemic vasculature and contributes to the development of systemic endothelial dysfunction and preeclampsia. However, obesity in healthy control pregnancies was not associated with thrombomodulin loss. Apparently, additional endothelial- or syncytiotrophoblast damaging factors are needed to cause placental thrombomodulin loss in preeclampsia.
Thrombomodulin has raised interest as a possible target in the treatment of preeclampsia; it could act beneficially by reducing both endothelial and placental dysfunction. Recently, administration of recombinant thrombomodulin was shown to rescue fetal tissue oxygenation and growth in a rat model of preeclampsia.25 However, in rats, thrombomodulin did not improve the maternal systemic symptoms of preeclampsia, such as hypertension. This could indicate that thrombomodulin administration rescues fetal growth by attenuating placental dysfunction. Our findings, which link decreased thrombomodulin expression to impaired placental cell function, support this hypothesis. Further research exploring the safety and efficacy of thrombomodulin as a target for the prevention and treatment of preeclampsia are of interest.
To conclude, preeclampsia is characterized by placental and endothelial dysfunction, but the exact pathogenesis is not completely understood. Our study shows that thrombomodulin is a new candidate role player in the pathogenesis of preeclampsia that is associated with both clinical parameters and dysregulation of factors essential for placental function. These insights set the stage for further basic and clinical research on placental development, the development of placental pathology, and on thrombomodulin as a target for the prevention and treatment of preeclampsia.
We thank Malu Zandbergen, Kimberley Veraar, Carin van der Keur, and Godelieve Swings for their excellent technical assistance and help with data collection. We thank Manon Bos for assisting with the cell culture experiments. This study has not been published previously in any form.
The online-only Data Supplement is available with this article at http://atvb.ahajournals.org/lookup/suppl/doi:10.1161/ATVBAHA.115.306780/-/DC1.
- Nonstandard Abbreviations and Acronyms
- activated protein C
- intrauterine growth restriction
- matrix metalloproteinase
- soluble Flt-1
- Received October 28, 2015.
- Accepted February 9, 2016.
- © 2016 American Heart Association, Inc.
- Penning M,
- Chua JS,
- van Kooten C,
- Zandbergen M,
- Buurma A,
- Schutte J,
- Bruijn JA,
- Khankin EV,
- Bloemenkamp K,
- Karumanchi SA,
- Baelde H
- Maruyama I,
- Bell CE,
- Majerus PW
- Isermann B,
- Hendrickson SB,
- Zogg M,
- Wing M,
- Cummiskey M,
- Kisanuki YY,
- Yanagisawa M,
- Weiler H
- Mosnier LO,
- Zlokovic BV,
- Griffin JH
- Anteby EY,
- Greenfield C,
- Natanson-Yaron S,
- Goldman-Wohl D,
- Hamani Y,
- Khudyak V,
- Ariel I,
- Yagel S
- Buurma AJ,
- Turner RJ,
- Driessen JH,
- Mooyaart AL,
- Schoones JW,
- Bruijn JA,
- Bloemenkamp KW,
- Dekkers OM,
- Baelde HJ
- Li W,
- Mata KM,
- Mazzuca MQ,
- Khalil RA
- Jeyabalan A
The events leading to impaired placentation and systemic endothelial dysfunction in preeclampsia are unclear. This study sheds light on this area by demonstrating loss of placental thrombomodulin mRNA and protein expression in preeclampsia; this is accompanied by decreased placental expression of matrix metalloproteinases, which are required for trophoblast invasion, and impaired syncytiotrophoblast survival. Loss of placental thrombomodulin is associated with placental soluble Flt-1 production, hinting to a link between the angiogenic imbalance of preeclampsia and loss of activated protein C signaling. Further, loss of thrombomodulin is associated with hypertension and obesity in preeclampsia. These findings give rise to the hypothesis that obesity creates a state in the microvasculature and on the placenta where vasculoprotective factors are lost, leading to a favorable environment to develop preeclampsia. These new findings reveal thrombomodulin as a new target of interest in preeclampsia and other syndromes characterized by endothelial dysfunction.