Up-regulation of CD81 inhibits cytotrophoblast invasion and mediates maternal endothelial cell dysfunction in preeclampsia

Significance

Preeclampsia (PE) is a severe trophoblast-related disorder that threatens the health of mother and fetus. Impaired placentation in the early stages of pregnancy and the subsequent systemic endothelial cell activation constitute the principle pathogenic mechanisms of PE. Currently, the molecules involved in the pathogenesis of PE that link the two stages of this disorder remain elusive. Our study demonstrates that CD81 is associated with key pathological changes that occur in both the placenta and maternal endothelial cells of patients with severe PE (sPE). Importantly, overexpression of CD81 induces a PE-like phenotype in pregnant rats. This study provides evidence of the involvement of CD81 in the pathogenesis of PE and supports the use of CD81 as a potential biomarker for PE.

Abstract

Preeclampsia (PE) is initiated by abnormal placentation in the early stages of pregnancy, followed by systemic activation of endothelial cells of the maternal small arterioles in the late second or third trimester (TM) of pregnancy. During normal pregnancy, placental cytotrophoblasts (CTBs) invade the maternal uterine wall and spiral arteries, whereas this process is interrupted in PE. However, it is not known how the malformed placenta triggers maternal endothelial crisis and the associated manifestations. Here, we have focused on the association of CD81 with PE. CD81, a member of the tetraspanin superfamily, plays significant roles in cell growth, adhesion, and motility. The function of CD81 in human placentation and its association with pregnancy complications are currently unknown. In the present study, we have demonstrated that CD81 was preferentially expressed in normal first TM placentas and progressively down-regulated with gestation advance. In patients with early-onset severe PE (sPE), CD81 expression was significantly up-regulated in syncytiotrophoblasts (STBs), CTBs and the cells in the villous core. In addition, high levels of CD81 were observed in the maternal sera of patients with sPE. Overexpressing CD81 in CTBs significantly decreased CTB invasion, and culturing primary human umbilical vein endothelial cells (HUVECs) in the presence of a high dose of exogenous CD81 resulted in interrupted angiogenesis and endothelial cell activation in vitro. Importantly, the phenotype of human PE was mimicked in the CD81-induced rat model.

Preeclampsia (PE) is characterized by a new onset of hypertension and proteinuria after 20 wk of gestation. Early-onset severe PE (sPE, ≤ 34 wk) is associated with a high incidence of eclampsia, cerebrovascular complications, and fetal growth restriction, which severely threaten maternal and fetal health (1). Although the etiology of PE remains elusive, this disease has two known stages: In stage I, inadequate cytotrophoblast (CTB) invasion early in the pregnancy causes abnormal placentation; in stage II, systemic endothelial cell activation and clinical manifestations occur in the second or third trimester (TM), which are associated with the release of molecules and factors from the shallowly implanted placenta (2, 3).

CD81 is a widely expressed tetraspanin that provides a scaffold for signaling molecules and orchestrates interactions between membrane-associated proteins to initiate signaling cascades that regulate cell adhesion, migration, and invasion (4–7). CD81 is also a tumor suppressor that inhibits the migration and invasion of some malignant tumor cells (8, 9). In addition, an increasing number of studies have reported that CD81 is one of the main components of exosomes and can be released into maternal circulation or delivered to certain organs and tissues (10–12). Our previous study showed that CD81 is highly expressed in LPS-treated HTR-8/SV neo cells derived from human first TM extravillous trophoblast cells and induces trophoblast syncytialization (13); however, the role of CD81 in human placentation and PE development remains unknown.

CTBs play a pivotal role in the development and maintenance of a successful pregnancy. During human pregnancy, villous CTB progenitors follow one of two differentiation pathways, becoming either syncytiotrophoblasts (STBs) or extravillous CTBs (14). STBs play roles in maternal–fetal nutrient exchange, immunological defense, and placental endocrine hormone secretion (15). Extravillous CTBs invade the maternal decidua and the adjacent third of the myometrium during interstitial invasion; they also penetrate the walls of the uterine spiral arteries, replace the endothelium, and disrupt the vascular smooth muscle, transforming these vessels into large-diameter, low-resistance conduits that enable the increased maternal blood perfusion that is required to develop the fetoplacental unit (16). Shallow CTB invasion and insufficient spiral artery modification are the hallmarks of PE (17).

CTB invasion is tightly regulated both temporally and spatially in an autocrine or paracrine manner by trophoblastic and uterine factors at the maternal–fetal interface. During pro-CTB invasion, through the regulation of conventional proangiogenic factors, CTBs adopt vascular phenotypes and become more invasive (18). Additionally, CTBs can secrete matrix metalloproteinases to facilitate their invasion into the maternal endomyometrium and uterine spiral arteries (19, 20). At the same time, CTB invasion is negatively regulated by a series of antiangiogenic factors and tumor suppressors, such as sFlt1, sEng, and SEMA3B, which tightly control the depth of CTB invasion (21–23).

In this study, we hypothesized that the autocrine and paracrine regulation of CD81 plays an important role in normal human placentation and that the dysregulation of CD81 signaling contributes to abnormal placentation and the pathogenesis of PE. We demonstrated that CD81 is preferentially expressed in first TM human placentas and progressively down-regulated with gestation advance in normal physiological conditions; CD81 up-regulation is detected in STBs, CTBs, cells in the villous core, and maternal sera of patients with early-onset sPE. The overexpression of CD81 in CTBs inhibits CTB invasion, and exposing primary human umbilical vein endothelial cells (HUVECs) to a high dose of exogenous CD81 induces endothelial cell activation and pathogenic angiogenesis. Furthermore, the overexpression of CD81 in pregnant rats triggers PE-like manifestations in vivo.

Results

CD81 Expression Is Progressively Down-Regulated on CTBs with Gestation Advance.

We first evaluated CD81 expression in human placental villi at different gestational stages by immunostaining. An anti-cytokeratin 7 (CK) antibody was used to identify the trophoblasts. The immunostaining of normal first, second, and third TM samples indicated that CD81 expression was tightly regulated with gestation advance. In the anchoring villi (AV), strong CD81 staining was detected on CTBs in the proximal column (P-col) from first TM placentas, with much less intense staining observed on CTBs in the distal column (d-col). CD81 was also strongly expressed by the initiating layer of the proximal CTB column and was dramatically down-regulated on the following layers of the proximal and distal CTB columns and the interstitial CTBs in second TM placentas. In the third TM placentas, CD81 staining was barely detectable on the invading CTBs (Fig. 1A). In the floating villi (FV), CD81 was localized on the villous CTB layer in the first TM placentas, exhibited discontinuous expression in the second TM placentas, and was barely detected in the third TM placentas. There was no detectable CD81 staining on STBs (Fig. 1B). In the villous core, vimentin-, CD45-, and CD66-positive staining were also observed (Fig. S1), suggesting that CD81 was expressed on multiple types of cells, including Hofbauer, stromal, and vessel cells.

Fig. 1.

CD81 expression is progressively down-regulated on CTBs with gestation advance. Placental tissue sections were double-stained with anti-CK and anti-CD81 by immunofluorescence. (A) CD81 expression on the cell columns of the AV (6 and 15.3 wk) and the basal plate (BP, 38 wk). CD81 immunostaining intensity was down-regulated in D-col of villi compared with the P-col. The data were representative of the analysis of first, second, and third TM placentas; n = 5 in each group. (Scale bar, 25 μm.) (B) CD81 expression in FV of the first, second, and third TM placentas (6, 15.3, and 38 wk; n = 5 in each group). (Scale bar, 25 μm.) (C) CD81 transcriptional levels in CTBs isolated from the first, second, and third TM placentas by qRT-PCR analysis (first TM vs. second TM, P < 0.01). (D and E) CD81 protein levels in CTBs during gestation by Western blotting (first TM vs. second TM, P < 0.01; first TM vs. third TM, P < 0.01). (F) The transcription of CD81 along the CTB differentiation in vitro. CD81 mRNA level was detected at the 12-h time course (0 h vs. 12 h, P < 0.05). (G and H) CD81 protein along CTB differentiation in vitro (0 h vs. 12 h, P < 0.01). The result was from three independent experiments in first TM CTBs. All Western blotting and qRT-PCR data are presented as mean ± SD. The relative intensity of CD81 levels was assessed by Image J. *P < 0.05, **P < 0.01.

Fig. S1.

Vimentin, CD45, and CD66 immunostainings are expressed in the 19-wk villous core of FV. (A) Vimentin and CK coimmunofluorescence staining. (B) CD45 and CK coimmunofluorescence staining. (C) CD66 and CK coimmunofluorescence staining. BV, blood vessel. n = 5. (Scale bar, 25 μm.)

To determine whether CD81 expression is regulated at the transcriptional or translational level in CTBs isolated from first, second, and third TM placentas, quantitative real-time polymerase chain reaction (qRT-PCR) and Western blotting analyses were used to quantify the levels of CD81 mRNA and protein. Compared with the CTBs from the first TM villi, lower CD81 mRNA and protein expression levels were observed in CTBs from the second TM placentas, and CD81 protein expression was significantly down-regulated in CTBs from the third TM placentas (Fig. 1 C–E). In addition, when CTBs isolated from 6- to 8-wk villi were cultured for 12 h to mimic CTB differentiation in vitro, we observed down-regulation of both CD81 mRNA (Fig. 1F) and protein (Fig. 1 G and H).

CD81 Expression Is Up-Regulated in the Placentas and Maternal Sera of Patients with sPE.

During the development of PE, the expression of the molecules that restrict CTB invasion and differentiation is dysregulated. Therefore, we examined CD81 expression in placental tissue sections from patients with early-onset sPE and from gestational age-matched patients who experienced noninfected preterm birth (nPTB). Using immunostaining, we found significant CD81 up-regulation at the maternal–fetal interface. Intense CD81 immunoreactivity was detected on the extravillous CTBs in the placental basal plates of patients with sPE but not on those of patients with nPTB (Fig. 2A and Fig. S2A). We then identified a significant up-regulation of CD81 expression on the majority of cells in the FV in sPE, which included STBs and villous CTBs (Fig. 2B and Fig. S2B), as well as on the cells and blood vessels in the villous core (Fig. 2C). Although syncytial knots were observed in the FV of both sPE and nPTB patients, CD81 up-regulation was only detected on the syncytial knots of sPE patients (Fig. S3). To quantify the changes in CD81 levels in patients with sPE, placental lysates and isolated CTBs were subjected to immunoblotting analysis. Higher levels of CD81 were detected in both the placental lysates and CTBs from patients with sPE (Fig. 2 D–G).

Fig. 2.

CD81 expression is up-regulated in CTBs, STBs, and the cells in villous core and sera of patients with early-onset sPE. (A) CD81 staining on the CTBs in basal plate (BP) of sPE and nPTB. (Scale bar, 25 μm.) (B) CD81 staining in FV of patients with sPE and nPTB by immunofluorescence. (Scale bar, 25 μm.) The figures were representative of the results of 12 individual placentas in each group. (C) CD81 staining in the villous stroma of patients with sPE and nPTB by immunofluorescence. (Scale bar, 25 μm.) BV, blood vessel. (D and E) Immunoblotting of CD81 in the placental lysates from patients with sPE (n = 15) and nPTB (n = 15) (sPE vs. nPTB, P < 0.01). (F and G) Immunoblotting of CD81 in the CTBs isolated from patients with sPE and nPTB (n = 3 in each group; sPE vs. nPTB, P < 0.05). (H and I) Immunoblotting of CD81 in the sera from sPE (n = 12) and gestational age-matched nPTB (n = 12) (sPE vs. nPTB, P < 0.01). (J and K) CD81 expression in exosome-free sera from sPE by Western blotting (sPE vs. nPTB, P < 0.01; n = 12 in each group). (L and M) CD81 and PLAP expression in serum exosomes from sPE by Western blotting (sPE vs. nPTB, P < 0.05; n = 12 in each group). All Western blotting data are presented as mean ± SD. The relative intensity of CD81 levels was assessed by Image J. GA, gestational age. *P < 0.05, **P < 0.01.

Fig. S2.

CD81 expression is up-regulated on CTBs, STBs, and the cells of the villous core from patients with early-onset sPE. (A) CD81 immunofluorescence staining on the CTBs in the basal plate (BP) (sPE, 7; nPTB, 7). (B) CD81 immunofluorescence staining in FV (sPE, 7; nPTB, 7). (Scale bar, 25 μm.)

Fig. S3.

CD81 expression is up-regulated in the syncytial knots of FV from patients with sPE. (A) CD81 immunofluorescence staining on syncytial knots of FV from sPE (n = 5). (B) CD81 immunofluorescence staining on syncytial knots of FV from nPTB (n = 5). (Scale bar, 25 μm.)

Because CD81 expression was significantly up-regulated in the sPE placentas, we examined whether these placentas released increased levels of CD81 protein into maternal circulation. A total of 24 serum samples were collected, including 12 from patients with early-onset sPE and 12 from gestational age-matched nPTB. An immunoblotting analysis showed that serum CD81 levels were significantly increased in the patients with sPE (Fig. 2 H and I).

CD81 is a main component of exosomes in many cell types (10–12, 24). Based on its association with the cellular membrane, CD81 can be found in either a soluble or insoluble form (12). To determine whether CD81 was packaged in exosomes and then released into maternal circulation or directly released into maternal circulation, we separated the serum samples from patients with sPE and gestational age-matched nPTB into exosome-containing and exosome-free fractions by differential centrifugation (25). Compared with the control samples, CD81 levels were significantly up-regulated in the sPE exosome-free samples (Fig. 2 J and K) and only slightly increased in the sPE exosome-containing samples (Fig. 2 L and M). Because placental alkaline phosphatase (PLAP) is considered as a placental origin marker for exosomes (26), we compared PLAP expression levels in the exosome-containing serum samples and found no significant difference between the sPE and nPTB groups (Fig. 2L).

CD81 Plays an Inhibitory Role in CTB Invasion and Disturbs Endothelial Cell Function.

CTB migration and invasion are critical events in human placentation. Therefore, we tested the hypothesis that CD81 up-regulation inhibits CTB invasion in an autocrine manner. To determine how CD81 controls CTB invasion, we took advantage of the fact that isolated CTBs spontaneously invade when plated onto a layer of Matrigel (19). Using this model system, we generated a CD81-positive adenovirus, Ad-CD81, and a CD81-negative adenovirus, Ad-CTL, and then used both viruses to infect CTBs isolated from first TM villi. GFP and anti-CK staining indicated that the infective efficiency in CTBs was ∼95% (Fig. 3A). An immunoblot analysis showed that CD81 expression was increased by 2.72-fold in CTBs infected with Ad-CD81 (Fig. 3B). After a 36-h incubation, the invasive activity of the CTBs was assessed. Invasiveness was significantly decreased by ∼60% in Ad-CD81–infected CTBs compared with Ad-CTL–infected CTBs (Fig. 3 C and D), which suggests that CD81 inhibits CTB invasion during human placentation.

Fig. 3.

CD81 overexpression inhibits CTB invasion and interrupts endothelial cell functions. CTBs isolated from first TM placentas were infected with Ad-CD81 for 36 h. Ad-CTL served as the control. (A) Adenovirus infection efficiency was identified by GFP and anti-CK double immunofluorescence staining. (Scale bar, 25 μm.) (B) The expressional level of CD81 protein on Ad-CD81–infected CTBs was determined by Western blotting analysis. I-CTB, first TM CTB. (C and D) CTB invasion was assessed on Ad-CD81–infected CTBs by transwell invasion assay. After 36-h incubation, CTBs invaded to the bottom side of the insert membrane were stained by anti-CK and counted under a Leica DMR microscope. (Scale bar, 25 μm.) The results were repeated five times (Ad-CD81 vs. Ad-CTL, P < 0.01). (E–G) HUVECs were plated in the Matrigel-coated plates and incubated in cell-conditioned medium from Ad-CD81–infected HTR-8/SV neo cells. Ad-CTL media served as the control. After 6-h incubation, HUVEC tube formation was observed under a Leica DMIL microscope and assessed by the number and area of the formed tubes (Ad-CD81 vs. Ad-CTL, P < 0.05 and P < 0.01). (Scale bar, 500 μm.) (H and I) ICAM-1 and VCAM-1 expression in CD81-overexpressed HUVECs by Western blotting analysis. Ad-CTL served as the control. The result was from three independent experiments in HUVECs isolated from three different umbilical cords of normal fetuses. MOI, multiplicity of infection. The data of Western blotting, CTB invasion, and HUVEC tube formation are presented as mean ± SD. *P < 0.05, **P < 0.01.

The principal pathogenesis of PE is thought to be associated with maternal endothelial cell activation (27). In an attempt to understand the pathogenic role of CD81 in sPE, we tested the hypothesis that increased CD81 levels in maternal circulation cause endothelial lesions. We performed a tube formation assay with HUVECs using recombinant CD81. As shown in Fig. 3 E–G, HUVECs treated with recombinant CD81 exhibited a smaller regular tube area and formed fewer tubes than the negative control cells. The activation of endothelial cells is considered as a key pathological event in the second stage of sPE (27). Therefore, we next examined the expression of molecules involved in endothelial cell activation, including vascular cell adhesion molecule-1 (VCAM-1) and intercellular cell adhesion molecule-1 (ICAM-1), in HUVECs overexpressing CD81. We infected HUVECs with either Ad-CD81 or Ad-CTL and observed that both VCAM-1 and ICAM-1 were up-regulated in the cells overexpressing CD81 compared with the Ad-CTL–infected control cells (Fig. 3 H and I). Furthermore, we found that the expression levels of VCAM-1 and ICAM-1 were increased in HUVECs treated with recombinant CD81 (Fig. S4). These results further support our proposed hypothesis that CD81 is associated with the pathogenesis of PE.

Fig. S4.

The expressions of VCAM-1 and ICAM-1 are up-regulated by recombinant CD81 protein in HUVECs. (A) Flag-tagged CD81 immunoblotting on the conditioned media of Ad-CD81– and Ad-CTL–infected HTR-8/SV neo cells. (B–E) ICAM-1 and VCAM-1 immunoblotting on HUVECs treated with recombinant CD81 for 24 h. Ad-CTL media served as the control. The relative intensity of VCAM-1 and ICAM-1 levels was assessed by Image J. The data are presented as mean ± SD. (Ad-CD81 vs. Ad-CTL, *P < 0.05, **P < 0.01.)

CD81-Induced Rat Model Mimics the Phenotype of Human PE.

In previous studies, a low-dose infusion of LPS into rats at the early stages of pregnancy successfully triggers a PE-like phenotype (28–30). Inspired by this model, we tested the hypothesis that CD81 overproduction participates in preeclamptic placentation and triggers the clinical manifestations of PE.

Pregnant rats were infected with either Ad-CD81 or Ad-CTL on the 5th day of gestation (GD5). As controls, nonpregnant rats were also infected on the same day. The blood pressure and proteinuria of the rats were monitored, and both groups of rats were euthanized either 10 or 14 d after infection (at GD15 or GD19). CD81 expression and trophoblast-directed uterine spiral artery remodeling were also analyzed. Compared with the Ad-CTL–infected rats, the Ad-CD81–infected rats exhibited a significant elevation in systolic blood pressure (SBP; 113.5 ± 1.95 mmHg vs. 108.76 ± 4.62 mmHg) on GD6, and this elevation was maintained until GD14. Interestingly, Ad-CD81–infected nonpregnant rats showed no obvious changes in SBP during the same time period (Fig. 4A). In addition, elevated urinary protein concentrations were observed on GD6 in Ad-CD81–infected pregnant rats (2.52 ± 0.17) compared with Ad-CTL–infected pregnant rats (2.20 ± 0.18) and Ad-CD81–infected nonpregnant rats (2.13 ± 0.25). This trend was also sustained until GD14 (Fig. 4B). Additionally, we analyzed the percentage of fetal resorption, fetal weight, and placental weight in both pregnant groups. The percentage of fetal resorption was 9.88% (8/81) and 3.53% (3/85) in the Ad-CD81–infected and Ad-CTL–infected rats, respectively (P < 0.01). Compared with the Ad-CTL–infected rats, fetal weight was mildly, but not significantly, reduced in the Ad-CD81–infected group (0.279 ± 0.103 vs. 0.281 ± 0.075 g, P > 0.05), which may be due to the early euthanization time point (a full-term pregnancy in a rat, 21 d). In addition, no differences in placental weight were detected between the two groups (Ad-CD81 vs. Ad-CTL, 0.194 ± 0.084 vs. 0.209 ± 0.070 g, P > 0.05). To evaluate the final pregnancy outcomes, we examined SBP, urinary protein concentration, fetal weight, and the percentage of fetal resorption in the adenovirus-infected pregnant rats that were euthanized at GD19. As shown in Fig. 4A, the SBP of Ad-CD81–infected rats was lower on GD18 than on GD14, but it was still higher than that of the Ad-CTL–infected rats on GD18 (106.70 ± 4.58 vs. 100.04 ± 3.07, P < 0.05). The urinary protein concentration was higher in the Ad-CD81–infected rats than in the controls on GD18 (Fig. 4B, P < 0.05). The fetal weight was significantly reduced in the Ad-CD81–infected rats compared with the Ad-CTL–infected rats (2.455 ± 0.256 vs. 2.554 ± 0.238, P < 0.01). Additionally, the percentage of fetal resorption was higher in the Ad-CD81–infected rats than in the Ad-CTL–infected rats [10.67% (16/150) vs. 2.03% (3/148), P < 0.01] (Table 1).

Fig. 4.

Ad-CD81 infection triggers PE-like syndrome in pregnant rats. (A) SBP and (B) urinary protein concentration were increased in Ad-CD81–infected pregnant rats on GD15 and GD19 but not in the adenovirus-infected nonpregnant rats. GD, gestational day; NP, nonpregnant; P, pregnant. Ad-CD81+P-GD15 (n = 6) vs. Ad-CTL+P-GD15 (n = 7), P < 0.05; Ad-CD81+P-GD19 (n = 11) vs. Ad-CTL+P-GD19 (n = 11), P < 0.05; Ad-CD81+NP-D15 (n = 4) vs. Ad-CTL+NP-D15 (n = 5), P > 0.05. (C) CD81 expression in the placentas of rats at GD15 by immunohistochemistry. IHC, immunohistochemistry. (Scale bar, 50 μm.) (D and E) CD81 protein expression in Ad-CD81– and Ad-CTL–infected rat placentas at GD15 by Western blotting (Ad-CD81 vs. Ad-CTL, P < 0.05). The relative intensity of CD81 levels was assessed by Image J. (F) Placenta and the attached mesometrial triangle of the myometrium from Ad-CD81– and Ad-CTL–infected rats at GD15 were immunostained with anti-CK. The data were representative of the analysis of samples from 13 different placentas. IHC, immunohistochemistry. (Scale bar, 100 μm.) (G) The measurement on diameters of all arteries in the sections of uterine mesometrial triangle from rats at GD15 (Ad-CD81 vs. Ad-CTL, P < 0.01). All quantified data are presented as mean ± SD. *P < 0.05, **P < 0.01.

Table 1.

Pregnancy outcomes of rats in different pregnant groups

Analyzed items	Ad-CTL, GD15 n = 82	Ad-CD81, GD15 n = 73	Ad-CTL, GD19 n = 145	Ad-CD81, GD19 n = 134
Fetal weight, g	0.281 ± 0.075	0.279 ± 0.103	2.554 ± 0.238	2.455 ± 0.256**
Placental weight, g	0.209 ± 0.070	0.194 ± 0.084	0.418 ± 0.076	0.419 ± 0.065
Fetal resorption, %	3.53 (3/85)	9.88 (8/81)**	2.03 (3/148)	10.67 (16/150)**

n, the number of fetuses that were weighted in each pregnant group. The data of fetal weight and placental weight are presented as mean ± SD. **P < 0.01.

We next examined CD81 expression in the placentas from Ad-CD81–infected rats. An immunohistochemistry analysis showed that CD81 expression was up-regulated in the labyrinth, spongiotrophoblast, and trophoblast giant cells of the placentas from Ad-CD81–infected rats on GD15 compared with those from Ad-CTL–infected rats (Fig. 4C). An immunoblotting analysis confirmed the up-regulation of CD81 in the placentas from Ad-CD81–infected rats (Fig. 4 D and E, P < 0.05). Meanwhile, we examined the infection efficiency of adenovirus vector by detecting the expression of GFP (Fig. S5) to rule out the effect of adenovirus infection. As shown in Fig. S6, the placentas from the Ad-CTL– and Ad-CD81–infected rats expressed nearly equal levels of GFP.

Fig. S5.

The construction of Ad-CTL and Ad-CD81.

Fig. S6.

The adenovirus vector equally infects the placentas of Ad-CD81– and Ad-CTL–infected pregnant rats. GFP was used to tag Ad-CD81 and Ad-CTL vector. (A) GFP immunohistochemical staining was detected on the placentas of Ad-CTL–infected GD15 rats. (B) GFP immunohistochemical staining was detected on the placentas of Ad-CD81–infected GD15 rats. (C) Rabbit IgG served as the negative control. (Scar bar, 25 μm.)

In addition to the increased blood pressure and proteinuria, the PE model rats also exhibited reduced trophoblast-directed uterine spiral artery remodeling (31). To determine how increased CD81 expression affects trophoblast-directed uterine spiral artery remodeling, an anti-CK antibody was used to probe trophoblasts in the placenta and uterus. The tissue sections containing the spiral arterial channel from the infected rats were identified, and all arteries in these sections were analyzed (32). All of the arteries in each of the sections were examined, and the minimal diameter across the center of each artery was measured. Compared with the diameters of the arteries in the mesometrial triangle of the myometrium from the Ad-CTL–infected rats, the diameters of the arteries in the Ad-CD81–infected rats were significantly smaller (71.53 ± 6.39 vs. 105.05 ± 13.40 μm, P < 0.01; Fig. 4 F and G).

Taken together, our data indicate that Ad-CD81–infected pregnant rats mimic not only the manifestations of PE but also the specific PE-induced pathological changes in the placenta and placental bed biopsies.

Discussion

The development of PE occurs in two stages: It is first initiated by reduced placental perfusion resulting from insufficient spiral artery remodeling, and then impaired placentation causes systemic pathophysiological changes in the maternal circulation. In the present study, we first assessed the expression pattern of CD81 in placentas from normal pregnancies. In normal pregnancy, CD81 is preferentially expressed by trophoblastic progenitors and CTBs in the P-col of AV and is progressively down-regulated with gestation advance when the number of trophoblastic progenitors is reduced. In patients with early-onset sPE, CD81 expression is up-regulated in the extravillous CTBs, STBs, and cells in the villous core and maternal sera. We demonstrated that overexpressing CD81 in CTBs in vitro negatively impacted CTB invasion and that exposing HUVECs to exogenous CD81 led to endothelial cell activation. In addition, pregnant rats overexpressing CD81 displayed a placental phenotype consistent with human PE, specifically that of poor uterine artery modification by trophoblasts, further supporting the conclusion that CD81 is involved in PE.

Physiological CTB invasion must be tightly regulated to ensure that the depth of CTB invasion proceeds to the appropriate extent but no further. Failure or exacerbation of CTB invasion results in pregnancy complications. For example, shallow invasion and minimal vessel remodeling are characteristics of PE (33), whereas aggressive invasion is characteristic of placental site tumors or choriocarcinoma (34, 35). Therefore, pro- and anti-invasive mechanisms function simultaneously and are delicately balanced to keep normal human placentation. The identification of CD81 as a negative mediator of CTB invasion broadens our understanding of the molecular mechanisms of CTB invasion.

As CD81 overexpression-induced repression of CTB invasion appears to be one of the mechanisms mediating abnormal placentation in sPE, the increased levels of serum CD81 in patients with sPE suggest that this molecule may be involved in mediating the symptoms and signs development of this disorder. After separating serum samples into exosome-containing and exosome-free fractions, we observed a dramatic up-regulation of CD81 in the exosome-free fraction of sera from patients with sPE, suggesting that soluble CD81 could be delivered to its targets through maternal circulation. To determine the potential adverse effects of soluble CD81 on the key cells involved in PE onset, maternal endothelial cells, we performed a HUVEC tube formation assay and assessed the endothelial expression of VCAM-1 and ICAM-1 by exposing these endothelial cells with HTR-8/SV neo cell-conditioned medium. The results revealed that a high dose of CD81 elicited poor tube formation corresponding to VCAM-1 and ICAM-1 up-regulation in HUVECs, suggesting cross-talk between trophoblasts and endothelial cells through CD81. Therefore, CD81 shed from the placenta into the maternal circulation may mediate the dysfunction of maternal endothelial cells and drive the development of the maternal manifestations of sPE. To test this hypothesis, it may be important to understand the molecular mechanisms of placental original CD81 webs or tetraspanin-enriched microdomains (36).

Our in vivo rat model provided additional evidence of CD81 involvement in the development of sPE. By injecting pregnant rats with Ad-CD81, we successfully generated a human PE-like syndrome that resulted in high blood pressure, proteinuria, and some adverse pregnancy outcomes. In contrast, injecting Ad-CD81 into nonpregnant rats did not trigger PE-like symptoms, suggesting that the placenta is necessary to induce these changes in rats, which is consistent with a critical feature of human PE.

In conclusion, our present study reports that CD81 levels are increased in the extravillous CTBs, STBs, and maternal circulation of pregnant women with sPE. Our in vitro results confirmed that the overexpression of CD81 inhibits CTB invasion and mediates endothelial cell dysfunction. The overexpression of CD81 in vivo causes PE symptoms and impairs spiral arterial modulation in a rat model. Our findings indicated that CD81 may be a potential biomarker for the early diagnosis of PE and that attenuating CD81 actions might be a more specific and safer therapeutic method for treating patients with PE.

Materials and Methods

Detailed descriptions of all materials and methods used in this study, including the placenta and serum samples (37), serum exosome isolation (25), human CTB isolation (38) and culture (Fig. S7), HTR-8/SV neo cells and HUVECs (39), generation of recombinant adenoviruses and recombinant CD81, the animal model (30, 32, 40), qRT-PCR, Western blotting, immunohistochemistry, CTB invasion assay, HUVEC tube formation, and statistical analysis, are presented in SI Materials and Methods. Demographic information on the patients who provided samples for this study is included in Table 2. This study was approved by the Scientific Research Ethics Committee of the Drum Tower Hospital (2009041), and informed consent was obtained from all participants. All animal protocols were approved by the Experimental Animals Management Committee (SYXK 2014–0052, Jiangsu, China).

Fig. S7.

CD45 and vimentin are almost not detected on those CTBs isolated from 6- to 8-wk placentas. (A) CK and CD45 coimmunofluorescence staining. (B) CK and vimentin coimmunofluorescence staining. (C) CD45 and vimentin coimmunofluorescence staining. (Scar bar, 25 μm.)

Table 2.

The demographics and clinical characteristics^a of patients collected for placentas and sera

Analyzed items	nPTB, n = 18	sPE, n = 18	P value
Maternal age, y	26.91 ± 4.81	29.55 ± 5.32	>0.05
Gestational age, wk	30.15 ± 3.60	30.30 ± 2.81	>0.05
BMI,b kg/m2	27.45 ± 2.80	30.31 ± 3.23	>0.05
Systolic BP, mmHg	113.11 ± 15.96	160.14 ± 14.35	<0.05
Diastolic BP, mmHg	69.56 ± 12.28	107.64 ± 13.52	<0.05
Proteinuria	–	++ to ++++	<0.05
Proteinuria, mg/24 h	–	8,052.9 ± 5,565.0	<0.05
AST, U/L	17.1 ± 8.7	26.8 ± 11.4	>0.05
ALT, U/L	15.3 ± 11.0	20.2 ± 10.5	>0.05
Platelet, ×109/L	169.4 ± 35.4	189.9 ± 91.8	>0.05
Fetal weight, g	1,926.11 ± 834.34	1,511.82 ± 503.25	>0.05
Placental weight, g	413.89 ± 79.13	415.50 ± 74.63	>0.05

ALT, aspartate transaminase; AST, alanine transaminase; BP, blood pressure; nPTB, gestational age-matched noninfected preterm birth.

^aValues are mean ± SD.

^bPrepregnancy.

SI Materials and Methods

Placentas and Serum Samples.

To detect CD81 expression in the placenta with gestation advance, 15 placental samples were used for immunostaining (first TM, 5; second TM, 5; third TM, 5). To study the change of CD81 in early-onset sPE, 36 placentas (sPE, 18; nPTB, 18) were collected (Table 2). Among them, 24 placental samples were used for immunostaining analysis (sPE, 12; nPTB, 12); 30 placental lysates (sPE, 15; nPTB, 15) and 6 CTB lysates (sPE, 3; nPTB, 3) were used for immunoblotting analysis; 24 blood samples (sPE, 12; nPTB, 12) obtained from the same patients’ pool were subject to immunoblotting analysis. In addition, CTBs isolated from 21 normal placental samples were used for qRT-PCR and Western blotting analysis (first TM, 15; CTBs were isolated from five 6–8-wk placentas in one separation prop; second TM, 3; third TM, 3). Also, another 30 placentas from first TM placentas were used for CTB invasion assay, and 15 first TM placentas were used for CTB differentiation in vitro. The procedure of placental sample collection was described previously (37).

Serum Exosome Isolation.

Serum exosome were isolated as follows: The cell debris and large membrane vesicles were removed by sequential centrifugation at 500 × g for 30 min followed by 12,000 × g for an additional 30 min. Then, the cleared supernatants were mixed with 1/10th volume of Na acetate buffer (1.0 M; pH, 4.75) for 30–60 min on ice and an additional 5 min on 37 °C. Finally, the suspension was centrifuged for 10 min at 5,000 × g. The exosome was collected by resuspending the pellets in 200 μL of PBS (25).

Human CTB Isolation and Culture.

Using previously published procedures (38), CTBs were isolated from human chorionic villi at 6–8-wk gestation. Briefly, the major steps of CTB isolation included removing the outer layer of syncytium, releasing the underlying CTBs by sequential enzymatic digestion, and purifying the cells by Percoll density gradient centrifugation. Then the remaining leukocytes and most vimentin-positive cells were removed by the anti-CD45 antibody-conjugated magnetic beads. CTB purity reached up to ∼95% (Fig. S7). The purified CTBs were used immediately or cultured in serum-free DMEM high glucose with 2/100 mL Nutridoma (Boehringer Mannheim Biochemicals) on Matrigel-coated substrates (Life Technologies, Inc.) at 37 °C in a 5% CO₂ incubator for the indicated time. Collagenase, sodium pyruvate, Hepes, glutamine, and gentamycin were purchased from Sigma-Aldrich, and 0.05% trypsin was obtained from Invitrogen.

Cells.

The HTR-8/SV neo cells, derived from human first TM extravillous trophoblast cells (kindly provided by Charles Graham, Queen’s University, Kingston, Ontario, Canada), were maintained in a 5% CO₂ incubator at 37 °C. RPMI medium 1640 (HyClone, Thermo Scientific) was supplemented with 50/500 mL heat-inactivated FBS (HyClone, Thermo Scientific), 100 U/mL penicillin, and 100 μg/mL streptomycin.

Primary HUVECs were isolated from the umbilical vein vascular wall by 0.2% collagenase treatment for 10 min in 37 °C and then seeded on 0.15% gelatin-coated plates and cultured in M199 medium (Gibco BRL/Invitrogen) containing 1/500 mL FBS and antibiotics (100 U/mL penicillin and 100 μg/mL streptomycin) at 37 °C in a 5% CO₂ incubator (39).

Generation of Recombinant Adenovirus and Recombinant CD81 Protein.

A cDNA fragment encoding CD81 (NM_004356.3) was subcloned (BamHI/AgeI digestion) into GV314-CMV-3Flag-GFP (Fig. S5; GENECHEM). The recombinant plasmids pCMV-CD81-3Flag-SV40-EGFP were transfected into HEK293A cells to obtain adenovirus prestocks, and then the recombinant adenovirus Ad-CD81 was purified using CsCl banding followed by dialysis against 10 mM Tris-buffered saline with 5/50 mL glycerol. The viral titer was determined using HEK293A cells and the Adeno-X Rapid Titer Kit (Clontech). CD81 was overexpressed in HTR-8/SV neo cells, CTBs, and HUVECs after adenovirus infection. Adenovirus expressing GV314-CMV-Flag-GFP was used as the control (Ad-CTL) in the CD81-overexpression experiments.

Recombinant CD81 was generated by Ad-CD81–infected HTR-8/SV neo cells. First, HTR-8/SV neo cells had been infected with Ad-CD81 [200 multiplicity of infection (MOI)] or Ad-CTL (200 MOI) for 48 h, and then culture medium that contained recombinant CD81 protein was collected. Recombinant CD81 protein was identified by detecting Flag tag in the cell-conditioned medium by Western blotting analysis (Fig. S4A). HUVECs were treated with the recombinant CD81 for 24 h and used for VCAM-1 and ICAM-1 immunoblotting.

Animal Model.

Sprague–Dawley rats from the Animal Center of Nanjing Medical University, aged 8–12 wk and weighing 200–300 g, were used. Pregnant rats were randomly divided into four groups (GD15, Ad-CD81, n = 6; Ad-CTL, n = 7; GD19, Ad-CD81, n = 11; Ad-CTL, n = 11). Another nine nonpregnant rats at the same age and weight were divided into two groups (Ad-CD81, n = 4; Ad-CTL, n = 5).

Blood Pressure, Proteinuria, and Sample Collection.

As previously described (30, 32), pregnancy was achieved by housing female and male rats together for one night. Day 0 of pregnancy (GD0) was determined by the presence of vaginal spermatozoa. The pregnant rats on GD5 and the nonpregnant rats were injected with Ad-CD81 or Ad-CTL (3 × 10⁹ plaque-forming units), respectively, through the tail vein. The SBP of each rat (08:00 AM–10:00 AM) was monitored by tail-cuff plethysmography every 3 d (BP-2010A; Softron). The urine of each rat (18:00 PM–8:00 AM) was collected every 3 d with dams housed individually in metabolic cages in the absence of food. Urinary protein level was measured using the pyrogallol red method (40). The pregnant rats were euthanized to collect placenta samples at GD15 or GD19 and the nonpregnant rats at the same day. Uterus, placenta, kidney, and liver were collected, and pregnant outcomes, fetal resorption, fetal weight, and placenta weight were examined.

The Assessment of Uterine Blood Vessel Modification.

Placentas with the associated mesometrial triangle were paraffin-fixed, and sections were cut step-serially from each implantation site parallel to the mesometrial–fetal axis. Sections containing a central maternal arterial channel or approaching the central maternal arterial channel were selected, and all of the arteries in each of the chosen sections were analyzed. The lumen size was defined as the minimal diameter that passed through the center of the artery by Image-Pro Plus 6.0 software (Media Cybernetics).

Quantitative Real-Time PCR.

Briefly, total RNA (2 μg), extracted from cells with TRIzol (Invitrogen), was reverse-transcribed with random primer into cDNA in a 20-μL reaction. Then, 2 μL cDNA was applied in each qRT-PCR containing 10 μL SYBR Green PCR Master Mix (Bio-Rad Laboratories) and 500 nM forward and reverse primers on a MyiQ Single Color Real-time PCR Detection System (Bio-Rad Laboratories). The data obtained from three independent experiments were used for analysis of relative gene expression. The primers were as follows: hsa-CD81, forward (F), 5′-AGATGATCCTGAGCATG-′3, and reverse (R), 5′-GGTAACAGGAAAGTTCAGAAC-′3; and hsa-18s, F, 5′-CGGCTACCACATCCAAGGAA-′3, and R, 5′-CTGGAATTACCGCGGCT-′3 (13).

Western Blotting.

Western blotting analysis was performed using standard procedures. Antibodies against CD81 (1:1,000; Santa Cruz Biotechnology), Flag (1:5,000; Sigma), PLAP (1:1,000; sangon), VCAM-1 (1:1,000; Abcam), ICAM-1 (1:1,000; Abcam), albumin (1:1,000; Bioworld Technology), β-actin (1:1,000; Bioworld), and GAPDH (1:1,000; Bioworld) were used. Serum samples from patients with sPE and gestational age-matched nPTB were processed directly with 0.2% SDS before Western blotting analysis. Equal volumes of serum were separated by SDS/PAGE.

Immunolocalization.

Immunofluorescence.

After being fixed in cold 4 g/100 mL paraformaldehyde, the human placenta tissue was transferred to 5/100 mL, 10/100 mL, and 15/100 mL sucrose/PBS, followed by embedding in optimum cutting temperature compound (OCT; Sakura Finetek USA Inc.). The blocks were then frozen in liquid nitrogen. Sections (5 μm) were prepared (Leica 3050) and collected on charged and precleaned microscope slides (Fisher Scientific). After three washes with PBS removing OCT, the slides were incubated in 1.5/50 mL BSA/PBS for 1 h in 37 °C and then incubated in antibodies. The positive signals were visualized by the fluorescence-conjugated secondary antibodies. The slides were mounted with mounting medium with DAPI (Vector Lab) and viewed using a fluorescence microscope (Leica DMR). Antibodies against CK (1:500; ProteinTech), CD81 (1:800; Santa Cruz), CD45 (1:100; Dako, Denmark), CD66 (1:200; Dako), and vimentin (1:200; Abcam) were used for immunofluorescence.

Immunohistochemistry.

Rat placenta tissues for immunoperoxidase staining were fixed in 10/100 mL formalin/PBS. Paraffin sections (5 μm) were prepared, de-waxed, and hydrated, and endogenous peroxides were quenched with 0.3% H₂O₂. After heat-induced antigen retrieval, the sections were incubated with antibodies, and then the sections were incubated with horseradish peroxidase-conjugated secondary antibody. Signals were visualized by incubation with diaminobenzidine. The sections were viewed using a Leica DM2000 microscope. CK (1:500; ProteinTech), CD81 (1:500; Epitomics), and GFP (1:800; abcam) were used for immunohistochemistry.

CTB Invasion Assay.

Briefly, CTBs isolated from first TM placentas were plated in transwell inserts on Matrigel-coated polycarbonate filters (pore size, ϕ = 8.0 μm; Millipore). Ad-CD81 (100 MOI) or Ad-CTL (100 MOI) was added to the top wells of 24-well plates. After 36-h incubation, the uninvaded cells were removed with cotton swabs, and the lower surfaces of the filters were fixed in 4 g/100 mL paraformaldehyde and stained with anti-CK. Three filters were plated per test variable, and the entire experiment was performed five times. For each test, the cells in six randomly selected fields were determined, and the counts were averaged. The data were presented as the number of invaded cells per field viewed using a Leica DMR microscope.

HUVEC Tube Formation.

HUVECs were suspended with recombinant CD81-positive or -negative medium that was collected from HTR-8/SV neo cells. The cell-conditioned medium was collected from Ad-CD81– or Ad-CTL–infected HTR-8/SV neo cells. Then 5 × 10⁴ primary HUVECs were suspended in the medium containing recombinant CD81 (vol/vol, 1:1), plated on the Matrigel-coated wells of 24-well plates, and incubated for 6 h. HUVEC tube formation was evaluated under a Leica DMIL microscope. The area and the number of the formed tubes were analyzed. The experiment was performed three times on HUVECs from three different umbilical cords of normal fetuses.

Statistical Analysis.

Normally distributed data were presented as the mean ± SD. Differences between means were analyzed by two-tailed Student’s t test. The data of rat SBP and urinary protein concentration were analyzed by ANOVA test. P values < 0.05 were considered significantly different.