MIF Increases sFLT1 Expression in Early Uncomplicated Pregnancy and Preeclampsia

Insufficient immune tolerance during pregnancy is associated with pathological conditions such as preeclampsia (PE). Soluble fms-like tyrosine kinase-1 (sFLT1), which exerts a role in the late stage of PE, has shown its beneficial anti-inflammatory effects in inflammation-associated diseases. Macrophage migration inhibitory factor (MIF) was reported to upregulate sFLT1 production in experimental congenital diaphragmatic hernia. However, the placental sFLT1 expression in early uncomplicated pregnancy and whether MIF can regulate sFLT1 expression in uncomplicated and preeclamptic pregnancy are unclear. We collected first-trimester placentas and term placentas from uncomplicated and preeclamptic pregnancies to investigate sFLT1 and MIF expression in vivo. Primary cytotrophoblasts (CTBs) and a human trophoblast cell line (Bewo) were used to study the regulation of MIF on sFLT1 expression in vitro. In placentas from first-trimester pregnancy, we observed a high expression of sFLT1, specifically in extravillous trophoblasts (EVTs) and syncytiotrophoblast (STB) cells. MIF mRNA levels strongly correlated with sFLT1 expression in term placentas from preeclamptic pregnancies. In in vitro experiments, sFLT1 and MIF levels increased significantly in CTBs during their differentiation to EVTs and STBs, and MIF inhibitor (ISO-1) significantly reduced sFLT1 expression in a dose-dependent manner during this process. sFLT1 showed significant upregulation with increasing doses of MIF in Bewo cells. Our results show that sFLT1 is highly expressed at the maternal–fetal interface during early pregnancy and that MIF can increase sFLT1 expression in early uncomplicated pregnancy and PE, which suggests that sFLT1 plays an essential role in the modulation of inflammation in pregnancy.


Introduction
It is well known that appropriate placentation in the first trimester is essential for pregnancy maintenance, and that defective placental development can lead to pregnancy complications such as miscarriage, fetal growth restriction, and preeclampsia (PE) [1,2]. Upon implantation, cytotrophoblasts (CTBs) undergo proliferation and differentiate into either multinucleated syncytiotrophoblast (STB) or invasive extravillous trophoblasts (EVTs). The STB serves as the site of gas and nutrient exchange between the mother and the fetus. The EVTs are composed of two subtypes of cells, one of which invades deeply into the maternal decidua, termed interstitial EVT, and the endovascular EVTs that remodel maternal spiral vessels, establishing maternal blood flow into the intervillous space [3,4].
As the only fetal-derived cells carrying paternal antigens, both EVTs and STB play a core role in mediating maternal tolerance. They come into direct contact with maternal immune cells at the maternal-fetal interface and contribute to the unique immunotolerant microenvironment via cell-cell direct interaction and through actions of chemokines and cytokines [5]. Immune tolerance is necessary to protect the fetal cells against attack from the maternal immune system and ensure a successful pregnancy [6]. Inadequate or insufficient anti-inflammatory responses followed by excessive inflammation is thought to be an important trigger for pathological conditions such as PE [7].
Increased levels of soluble fms-like tyrosine kinase-1 (sFLT1) reflect a key factor driving PE at the late stage as an antagonist of vascular endothelial growth factor (VEGF) and placental growth factor (PIGF) [8][9][10]. Another study suggested that in PE, placental sFLT1 is upregulated as a protective response to an increase in endometrial VEGF levels, which is associated with placental deformities and pregnancy loss [11]. In line with this, several studies have shown that sFLT1 has beneficial anti-inflammatory properties in inflammationassociated diseases, such as psoriasis and type 1 diabetes [12,13].
Despite great interest in the potential beneficial immuno-modulatory properties of sFLT1 in pregnancy and its pathogenic role in the late stage of PE, the expression and regulation of sFLT1 are not fully understood. Hypoxia leads to the upregulation of sFLT1 expression [14]. However, in vivo and in vitro studies also suggest that inflammationassociated factors such as macrophage migration inhibitory factor (MIF) increase sFLT1 expression [15][16][17]. MIF is a versatile cytokine acting as a central regulator of different physiological processes including cell proliferation and differentiation, angiogenic biological activities, and innate immune response [18]. It can be produced by immune cells, endothelial, epithelial, and many other cell types [19]. MIF is abundantly expressed at the maternal-fetal interface and is proposed to have a role in establishing and maintaining a healthy pregnancy [20]. MIF has also been associated with some placental pathological conditions, including infection and PE [20]. Additionally, it has been shown that MIF increases sFLT1 production in experimental congenital diaphragmatic hernia [21]. Therefore, we hypothesized that placental sFLT1 has high expression at the maternal-fetal interface during pregnancy and that MIF is involved in the regulation of increased placental sFLT1 production during normal and preeclamptic pregnancy.

High Expression of sFLT1 in EVTs and STB in Placentas during First Trimester
To investigate the expression of sFLT1 and MIF at the maternal-fetal interface in early pregnancy, we performed immunohistochemistry (IHC) on placentas during the first trimester. HLA-G and HCG-β were stained as markers of EVTs and STB, respectively. Representative images are shown in Figure 1. There was a high expression of sFLT1 in EVTs ( Figure 1B) and STB ( Figure 1D), while MIF staining can be seen in CTBs, EVTs, and STB ( Figure 1E,F).

sFLT1 and MIF Expression Increases upon Human Primary Cytotrophoblast Differentiation into EVTs and STBs
Next, to confirm MIF and sFLT1 expression in vitro, human primary CTBs were isolated from first-trimester placentas and differentiated into EVTs or STBs. Quantitative PCR (qPCR) was used to measure sFLT1 and MIF mRNA levels on day 0 (undifferentiated CTBs) and 3 and 6 days after differentiation. Upon CTB differentiation into EVTs, sFLT1 mRNA expression had a 1.2-fold increase (p = 0.456) on day 3 and increased 21-fold on day 6 (p < 0.001) compared to day 0 (CTBs) (Figure 2A). The mRNA level of MIF was similar on day 3 and rose by 1.4 fold (p < 0.01) on day 6 ( Figure 2C) compared to CTBs. When CTBs were differentiated to STB, sFLT1 mRNA levels increased substantially by 9 fold (p < 0.001) on day 3 and 518 fold (p < 0.001) on day 6 ( Figure 2B). For MIF mRNA expression, it increased 1.7-fold (p < 0.001) on day 3 and 2.3 fold (p < 0.001) on day 6 ( Figure 2D).

sFLT1 and MIF Expression Increases upon Human Primary Cytotrophoblast Differentiation into EVTs and STBs
Next, to confirm MIF and sFLT1 expression in vitro, human primary CTBs were isolated from first-trimester placentas and differentiated into EVTs or STBs. Quantitative PCR (qPCR) was used to measure sFLT1 and MIF mRNA levels on day 0 (undifferentiated CTBs) and 3 and 6 days after differentiation. Upon CTB differentiation into EVTs, sFLT1 mRNA expression had a 1.2-fold increase (p = 0.456) on day 3 and increased 21-fold on day 6 (p < 0.001) compared to day 0 (CTBs) (Figure 2A). The mRNA level of MIF was similar on day 3 and rose by 1.4 fold (p < 0.01) on day 6 ( Figure 2C) compared to CTBs. When CTBs were differentiated to STB, sFLT1 mRNA levels increased substantially by 9 fold (p < 0.001) on day 3 and 518 fold (p < 0.001) on day 6 ( Figure 2B). For MIF mRNA expression, it increased 1.7-fold (p < 0.001) on day 3 and 2.3 fold (p < 0.001) on day 6 ( Figure 2D).  CTBs were incubated with EVT or STB medium for 6 days; qPCR was used to measure sFLT1 and MIF expression on day 0 (CTBs), 3, and 6. Data were expressed relative to CTBs (day 0). Results were corrected for the housekeeping gene GAPDH. Summary data are presented as means ± SD from three independent experiments. The mRNA levels of sFLT1 (A,B) and MIF (C,D) during differentiation of CTBs into EVTs (A,C) and STBs (B,D). One-way ANOVA on log-transformed gene expression was performed, * p < 0.05, ** p < 0.01, *** p < 0.001.

MIF Inhibitor Reduces sFLT1 Expression during Differentiation of CTBs into EVTs and STB
To study whether MIF can regulate sFLT1 expression during differentiation from CTBs to EVTs or STB, we incubated CTBs in EVT or STB medium with or without an increasing dose of MIF inhibitor ISO-1 (12.5 µg/mL, 25 µg/mL, 50 µg/mL, and 100 µg/mL). We found a significant dose-dependent decrease in sFLT1 mRNA expression after incubation with MIF inhibitor ISO-1. During differentiation of CTBs into EVTs with ISO-1, the expression of sFLT1 at mRNA level reduced by 0.4 fold with every 10-fold increase in ISO-1 concentration ( Figure 3A, p < 0.01), and sFLT1 mRNA levels showed a 0.14-fold decrease ( Figure 3B, p < 0.01) when STB was differentiated from CTBs in the presence of ISO-1. CTBs were incubated with EVT or STB medium for 6 days; qPCR was used to measure sFLT1 and MIF expression on day 0 (CTBs), 3, and 6. Data were expressed relative to CTBs (day 0). Results were corrected for the housekeeping gene GAPDH. Summary data are presented as means ± SD from three independent experiments. The mRNA levels of sFLT1 (A,B) and MIF (C,D) during differentiation of CTBs into EVTs (A,C) and STBs (B,D). One-way ANOVA on log-transformed gene expression was performed, * p < 0.05, ** p < 0.01, *** p < 0.001.

MIF Inhibitor Reduces sFLT1 Expression during Differentiation of CTBs into EVTs and STB
To study whether MIF can regulate sFLT1 expression during differentiation from CTBs to EVTs or STB, we incubated CTBs in EVT or STB medium with or without an increasing dose of MIF inhibitor ISO-1 (12.5 µg/mL, 25 µg/mL, 50 µg/mL, and 100 µg/mL). We found a significant dose-dependent decrease in sFLT1 mRNA expression after incubation with MIF inhibitor ISO-1. During differentiation of CTBs into EVTs with ISO-1, the expression of sFLT1 at mRNA level reduced by 0.4 fold with every 10-fold increase in ISO-1 concentration ( Figure 3A, p < 0.01), and sFLT1 mRNA levels showed a 0.14-fold decrease ( Figure 3B, p < 0.01) when STB was differentiated from CTBs in the presence of ISO-1.  CTBs were incubated with EVT or STB medium for 6 days. Different concentrations (12.5 µg/mL, 25 µg/mL, 50 µg/mL, and 100 µg/mL) of MIF inhibitor (ISO-1) were added to the medium from day 3 onwards. After that, sFLT1 mRNA expression was measured on D6 using qPCR. As a control, cells were cultured in differentiation medium without ISO-1. Data were expressed relative to their own control without ISO-1. sFLT1 mRNA levels during differentiation of CTBs into EVTs (A) and STB (B) without or with an increased dose of ISO-1. Linear regression analysis on log-transformed ISO-1 concentration and sFLT1 mRNA expression, taking experiments into account. ** p < 0.01.

MIF Increases sFLT1 Expression in a Human Trophoblast Cell Line
As sFLT1 already increased in expression during differentiation, we wanted to analyze the regulatory properties of MIF on sFLT1 expression in human trophoblasts. Therefore, we cultured Bewo cells, a widely used human trophoblast cell line, and stimulated CTBs were incubated with EVT or STB medium for 6 days. Different concentrations (12.5 µg/mL, 25 µg/mL, 50 µg/mL, and 100 µg/mL) of MIF inhibitor (ISO-1) were added to the medium from day 3 onwards. After that, sFLT1 mRNA expression was measured on D6 using qPCR. As a control, cells were cultured in differentiation medium without ISO-1. Data were expressed relative to their own control without ISO-1. sFLT1 mRNA levels during differentiation of CTBs into EVTs (A) and STB (B) without or with an increased dose of ISO-1. Linear regression analysis on log-transformed ISO-1 concentration and sFLT1 mRNA expression, taking experiments into account. ** p < 0.01.

MIF Increases sFLT1 Expression in a Human Trophoblast Cell Line
As sFLT1 already increased in expression during differentiation, we wanted to analyze the regulatory properties of MIF on sFLT1 expression in human trophoblasts. Therefore, we cultured Bewo cells, a widely used human trophoblast cell line, and stimulated these cells with different concentrations of MIF (0.2 ng/mL, 2 ng/mL, and 20 ng/mL). With each 10-fold increase in MIF concentration, the expression of sFLT1 showed a 2.3-fold elevation ( Figure 4, p < 0.05).
CTBs were incubated with EVT or STB medium for 6 days. Different concentrations (12.5 µg/mL, 25 µg/mL, 50 µg/mL, and 100 µg/mL) of MIF inhibitor (ISO-1) were added to the medium from day 3 onwards. After that, sFLT1 mRNA expression was measured on D6 using qPCR. As a control, cells were cultured in differentiation medium without ISO-1. Data were expressed relative to their own control without ISO-1. sFLT1 mRNA levels during differentiation of CTBs into EVTs (A) and STB (B) without or with an increased dose of ISO-1. Linear regression analysis on log-transformed ISO-1 concentration and sFLT1 mRNA expression, taking experiments into account. ** p < 0.01.

MIF Increases sFLT1 Expression in a Human Trophoblast Cell Line
As sFLT1 already increased in expression during differentiation, we wanted to analyze the regulatory properties of MIF on sFLT1 expression in human trophoblasts. Therefore, we cultured Bewo cells, a widely used human trophoblast cell line, and stimulated these cells with different concentrations of MIF (0.2 ng/mL, 2 ng/mL, and 20 ng/mL). With each 10-fold increase in MIF concentration, the expression of sFLT1 showed a 2.3-fold elevation ( Figure 4, p < 0.05).

MIF Expression Correlates with sFLT1 Expression in Placentas from Preeclamptic Women
We first performed IHC for sFLT1 and MIF on normal control (NC) and preeclamptic (PE) placentas. Representative images of the sFLT1 and MIF staining of an NC and PE are shown in Figure 5A-D. MIF was mainly present in STB, CTBs, and fetal endothelial cells ( Figure 5C,D), while sFLT1 was expressed prominently in STB ( Figure 5A,B). We scored

MIF Expression Correlates with sFLT1 Expression in Placentas from Preeclamptic Women
We first performed IHC for sFLT1 and MIF on normal control (NC) and preeclamptic (PE) placentas. Representative images of the sFLT1 and MIF staining of an NC and PE are shown in Figure 5A-D. MIF was mainly present in STB, CTBs, and fetal endothelial cells ( Figure 5C,D), while sFLT1 was expressed prominently in STB ( Figure 5A,B). We scored MIF expression semi-quantitatively and did not find a significant difference between the NC (N = 19) and PE (n = 14) groups (p = 0.541). MIF expression semi-quantitatively and did not find a significant difference between the NC (N = 19) and PE (n = 14) groups (p = 0.541). Next, we measured sFLT1 and MIF mRNA levels in placentas using qPCR. The expression of sFLT1 mRNA increased significantly in the PE group compared to the NC group ( Figure 5E, p < 0.01). MIF mRNA expression had a mild elevation in PE, but this increase was not significant ( Figure 5F, p = 0.335).
Finally, to investigate the relationship between MIF and sFLT1 expression in PE, we correlated placental sFLT1 with MIF mRNA levels. Within the PE group, there was a significant positive correlation between MIF mRNA and sFLT1 mRNA levels ( Figure 5G, p < 0.01).

Discussion
Here, we showed that in early uncomplicated pregnancy, sFLT1 is highly expressed in EVTs and STB, and that placental sFLT1 increases significantly in PE compared to Next, we measured sFLT1 and MIF mRNA levels in placentas using qPCR. The expression of sFLT1 mRNA increased significantly in the PE group compared to the NC group ( Figure 5E, p < 0.01). MIF mRNA expression had a mild elevation in PE, but this increase was not significant ( Figure 5F, p = 0.335).
Finally, to investigate the relationship between MIF and sFLT1 expression in PE, we correlated placental sFLT1 with MIF mRNA levels. Within the PE group, there was a significant positive correlation between MIF mRNA and sFLT1 mRNA levels ( Figure 5G, p < 0.01).

Discussion
Here, we showed that in early uncomplicated pregnancy, sFLT1 is highly expressed in EVTs and STB, and that placental sFLT1 increases significantly in PE compared to uncomplicated term pregnancy. Moreover, we showed that MIF increases sFLT1 production in cultured STB and that MIF inhibitor ISO-1 reduces sFLT1 expression during differentiation of CTBs into EVTs and STB in vitro. Furthermore, we showed a strong correlation between sFLT1 mRNA and MIF mRNA in term placental tissue of patients with PE. Together, our findings suggest high placental sFLT1 expression at the maternal-fetal interface during early pregnancy and MIF as an important mediator in regulating sFLT1 expression in early uncomplicated pregnancy and PE.
The finding that placental sFLT1 expression has a significant increase in PE is in line with the current literature. Abundant studies have described that placental sFLT1 mRNA and protein levels increase significantly in PE compared to normotensive term pregnancy [8, 22,23]. Additionally, the study by Buurma et al. revealed that preeclamptic placentas have more intense sFLT1 staining in the syncytial knots [24]. In another study, Fan et al. showed an increased sFLT1 expression in EVTs in preeclamptic women [11]. However, there is no literature about the expression and distribution of placental sFLT1 in normal pregnancy, especially during early gestation. We showed for the first time that sFLT1 is highly expressed in EVTs and STB in first-trimester placentas, which suggests a physiological role for sFLT1 during early normal pregnancy.
The importance of sFLT1 in normal pregnancy has been emphasized in other studies. For instance, the study by Richard et al. showed high levels of serum sFLT1 in healthy pregnancy compared to a non-pregnant state [10]. Moreover, Kerry et al. found that lower levels of sFLT1 expression are associated with invasive placentation, indicating a critical functional role for sFLT1 in the regulation of placental invasion [25]. sFLT1 has also been suggested to have protective properties in preeclampsia. The first evidence comes from a study by Perucci et al. They showed that sFLT1 knockdown in VEGF overexpressing animals enhances PE-like symptoms in comparison to conditions of only VEGF overexpression [11]. Additionally, Nakakita et al. reported that the removal of sFLT1 in PE has negative consequences, such as the development of fetal bradycardia [26]. More recently, Parchem et al. revealed that reducing PlGF levels has a protective effect against the development of preeclampsia [27]. Taken together, these findings support our hypothesis that the upregulation of sFLT1 plays a protective role during normal and preeclamptic pregnancies.
One possible explanation why sFLT1 is upregulated in EVTs and STB at the maternal-fetal interface could be that both EVTs and STB are in close contact with maternal immune cells; thus, sFLT1 is secreted as an anti-inflammatory cytokine to contribute to immune tolerance. It has been abundantly described in the literature that EVTs and STB are involved in immune tolerance via chemokines and cytokines. For instance, a study by Veras et al. reported that STB has higher levels of programmed death-ligand 1 (PD-L1), which is a immune-suppressive protein, when compared with CTBs in early and term normal placentas [28]. More recently, Michelle et al. summarized that STB expresses several other immune suppressive factors, such as the metabolizing enzyme heme oxygenase-1 (HMOX1) [29]. Furthermore, HLA-G, which is specifically expressed in EVTs, is widely regarded as the pivotal protective factor for a tolerogenic microenvironment at the maternal-fetal interface [30]. Meanwhile, sFLT1 was shown to have an anti-inflammatory function in various immune-related diseases [31][32][33].
To further explore the regulation of sFLT1 production in uncomplicated and preeclamptic pregnancy, we performed in vitro experiments to investigate whether MIF is involved in increased sFLT1 production during differentiation of CTBs into EVTs and STB and whether MIF upregulates sFLT1 expression in a human trophoblast cell line. Moreover, we correlated placental sFLT1 with MIF expression in patients with PE. We found a regulatory relationship between MIF and sFLT1 expression. It was first reported by Perveen et al. that MIF increases sFLT1 expression in a rat model with congenital diaphragmatic hernia [21]. Our results prove the role of MIF in the regulation of sFLT1 expression and show that MIF increases sFLT1 production in early uncomplicated pregnancy and PE.
The role of MIF in uncomplicated pregnancy and PE has been reported. For example, Krivokuca et al. elaborated that MIF promotes trophoblast migration and invasion [34]. Additionally, the role of MIF in the spiral artery remodeling process was shown by Vilotić et al. [35]. Moreover, a study by Ietta et al. demonstrated that MIF is able to maintain trophoblast homeostasis by preventing abnormal apoptotic death [36]. Furthermore, Hristoskova et al. found that serum MIF concentrations are elevated throughout pregnancy [37], and other studies showed elevated MIF serum levels in PE compared to normal pregnancy [38,39]. Most importantly, Arcuri et al. revealed that MIF can be produced by immune cells and contributes to maternal-fetal immunotolerance via autocrine or paracrine manner [40], which provides a possible inflammation-modulated link between MIF and sFLT1 at the maternal-fetal interface. We confirmed that MIF upregulates sFLT1 expression when CTBs differentiate into EVTs and STB, and that placental MIF positively correlates with sFLT1 expression in PE, which suggests a potential inflammation-associated mechanism between MIF and sFLT1 at the maternal-fetal interface in early uncomplicated pregnancy and PE.
The regulation of inflammatory factors on sFLT1 expression has been abundantly tested in previous studies. In 2004, Eubank et al. revealed that granulocyte-macrophage colony-stimulating factor (GM-CSF) induces sFLT1 production in human monocytes [16]. More recently, a study by Xia et al. showed that both interleukin-4 (IL-4) and GM-CSF mediate the upregulation of sFLT1 in monocytes [17]. Even more important, tumor necrosis factor-α (TNF-α), agonistic autoantibodies to the angiotensin II type I receptor (AT1-AA), and angiotensin II (ANG II) increase sFLT1 expression during pregnancy in vivo or in vitro [15,41,42]. Taken together, these findings suggest that inflammatory factors serve as an important stimulus for sFLT1 production during pregnancy and firmly support our assumption that sFLT1 plays a potential inflammatory role in pregnancy and PE.
A limitation of our study was that we do not have data on sFLT1 expression in firsttrimester placentas or placental explants from women who later go on to develop PE. Currently, there is limited effective prediction for PE, which makes it difficult to obtain placental tissue at early gestation [43].
In conclusion, we reported that sFLT1 has a high expression at the maternal-fetal interface during early pregnancy. Furthermore, we showed that MIF increases sFLT1 expression in early uncomplicated pregnancy and PE, which suggests that sFLT1 is involved in the modulation of inflammation in pregnancy. When it comes to the treatment of PE, negative consequences of sFLT1 removal should be taken into consideration. Further studies should focus on the downstream effect of sFLT1 and its anti-inflammatory mechanisms in uncomplicated pregnancy and PE.

Human Tissue Collection
First-trimester placental tissues were used to investigate sFLT1 and MIF expression. These placentas were also used for isolation of primary cytotrophoblasts as previously described by Eikmans et al. [44]. All samples are from donors (n = 3) who had undergone termination of pregnancy due to social reasons between six and nine weeks gestation.
A term placental cohort, previously documented by van 't Hof et al. [45], was used to measure sFLT1 and MIF expression. The clinical characteristics are summarized in Table 1. In short, placental tissue specimens were obtained from patients with (n = 14) and without (n = 19) PE, as defined according to the definition by the International Society for the Study of Hypertension in Pregnancy (ISSHP). Cases in which the pregnancy had no preeclampsia, HELLP, preterm birth, decreased birth weight, fetal growth restriction, or infection were included as a normal control group. All placental samples were collected within 24 h of delivery and subsequently embedded in paraffin or frozen at −80 • C. Table 1. Clinical characteristics of controls and preeclampsia cases.
Informed consent was obtained from every patient. The study protocol was approved by the ethics committee of the LUMC with protocol number P16.048 (4 May 2016).
To study STB, CTB cultures were harvested and 1 × 10 5 cells were transferred to a fresh collagen IV-coated (2.5 µg/mL) dish and incubated in 2 mL of STB medium. This medium contained DMEM/F-12 supplemented with 0.1 mM 2-mercaptoethanol, 1% p/s, 0.3% BSA, 1% ITS-X, 2.5 µM Y-27632, 2 µM forskolin (FSK), and 4% KSR. After three days, the medium was replaced with the same content. At six days, the cells were harvested for further measurements.
To study the effect of MIF inhibition on sFLT1 expression, after precoating with appropriate concentration of collagen IV, 0.75 × 10 5 CTBs per well with EVT medium or 1 × 10 5 CTBs per well with STB medium were plated in a 6-well plate. After 3 days, medium for EVTs or STBs was changed as described above and 12.5 ug/mL, 25 ug/mL, 50 ug/mL, or 100 ug/mL MIF inhibitor (ISO-1) (Selleckchem) was added. On day 6, all samples were harvested for gene expression analysis using qPCR. As a control, cells were incubated with EVT or STB medium without ISO-1.
To study the effect of MIF on sFLT1, human trophoblast Bewo cells (ATCC) were cultured in DMEM/F12 (Gibco Laboratories, Gaithersburg, MD, USA) supplemented with 10% fetal bovine serum (FBS, Sigma-Aldrich, St. Louis, MO, USA) at 37 • C in 5% CO 2 . To analyze the regulated effect of MIF on sFLT1 production, 3 × 10 5 Bewo cells per well were plated in a 12-well plate. One day after the Bewo cells reached confluence, they were further cultured in serum-free medium for 24 h and subsequently incubated with 0.2 ng/mL, 2 ng/mL, or 20 ng/mL of MIF (Pepro Tech, Rocky Hill, NJ, USA) for 24 h. As a control, cells were incubated in the medium without MIF. RNA isolation, cDNA synthesis, and qPCR were performed as described below.

Scoring of Sections
In the biopsy cohort, the MIF-positive area in the placentas was scored semi-quantitatively by two independent observers on images obtained at 100× magnification using the following four categories: MIF-positive area (MIF) very low, MIF low, MIF moderate, and MIF high. Consensus regarding the score of each section was reached and used in the statistical analysis.

Statistical Analysis
Data are expressed as means ± SD. Multigroup comparison was performed using one-way ANOVA followed by the LSD multiple comparison test for subgroup comparison if the overall F-test was significant. To analyze the effect of MIF or ISO-1 stimulation on cells, a linear regression model was used. Independent t-tests were performed in healthy and preeclamptic pregnancy groups and Pearson correlation coefficient was determined for correlations using SPSS. The MIF-positive area in preeclamptic placentas and normal controls were compared using a Chi-square test. All normalized gene expression data were log-transformed and an alpha level of 0.05 was used to assess statistical significance.