E-Selectin mediates pathogenic effects of antiphospholipid antibodies


S. S. Pierangeli, Morehouse School of Medicine, Department of Microbiology, Biochemistry and Immunology, BSMB 332, 720 Westview Dr SW Atlanta, GA 30310-1495, USA.
Tel.: +404 7521882; fax: +404 7521644; e-mail: pierans@msm.edu


Summary.  Antiphospholipid (aPL) antibodies, detected in patients with antiphospholipid syndrome (APS) are associated with thrombosis, pregnancy loss and thrombocytopenia. Studies have shown that aPL are thrombogenic in vivo, but the mechanism(s) involved are not completely understood. Several studies have demonstrated that aPL antibodies activate endothelial cells (ECs) in vitro, as determined by up-regulation of adhesion molecules: E-selectin (E-sel); intercellular adhesion molecule-1 (ICAM-1) and vascular cell adhesion molecule-1 (VCAM-1), and in vivo. The objectives of these study were to determine the effects of aPL antibodies on the expression of E-selectin on ECs, on the adhesion of monocytes to ECs and to study the role of E-selectin on aPL antibodies enhanced thrombus formation and activation of ECs in vivo. We demonstrated that the surface expression of E-selectin on HUVEC by ELISA was increased 400-fold when treated with tumor necrosis factor-alpha (TNF-α) and 421-fold when treated with aPL antibodies during 4 h. APL antibodies also induced activation of the nuclear factor-kappa B (NF-κB). APL antibodies increased significantly the number of adhering leukocytes to ECs in vivo in C57BL/6 J mice when compared to IgG-NHS treated mice. This effect was abrogated in E-selectin-deficient mice. The thrombus size was significantly increased in C57BL/6 J mice treated with aPL antibodies when compared to mice treated with IgG-NHS. This enhancement in thrombus size by aPL antibodies was abrogated in E-selectin-deficient mice treated with aPL antibodies.

Antiphospholipid (aPL) antibodies are associated with aPL syndrome (APS), which is a syndrome of thrombosis, pregnancy loss, and thrombocytopenia [1]. aPL antibodies have been shown to enhance thrombus formation, but the mechanisms involved are not completely understood [2–4].

Several studies have shown that aPL antibodies activate endothelial cells (ECs) in vitro, as demonstrated by enhanced expression of adhesion molecules (intercellular cell adhesion molecule-1 (ICAM-1), vascular cell adhesion molecule-1 VCAM-1, and E-selectin, on human umbilical vein endothelial cells (HUVECs) and enhanced monocyte adherence to ECs in vitro[2–6] and in vivo[7,8].

Our group recently showed that the thrombogenic effects of aPL antibodies are mediated by ICAM-1, VCAM-1 and P-selectin [9]. As additional support for the hypothesis that aPL antibodies activate ECs and may create an hypercoagulable state in APS patients, two recent studies indicated that the levels of soluble ICAM-1 and VCAM-1 were significantly increased in the plasma of patients with APS and recurrent thrombosis [10,11].

E-Selectin is an adhesion molecule, expressed on activated endothelium and has a terminal lectin domain which binds carbohydrate ligands expressed on leukocytes serving to slow leukocytes in the first phase of migration [12]. Neutrophils appear early at sites of acute inflammation and this is in part controlled by the cytokine induction of E-selectin on the surface of endothelium in these areas. Stimulation in vitro of endothelium with tumor necrosis factor-alpha (TNF-α), induces expression of E-selectin over a period of 4–12 h, and this molecule appears early during inflammatory reactions in vivo, but this expression wanes by 24 h [13]. While rapidly induced by various proinflammatory signals, E-selectin may mediate initial attachment and rolling of leukocytes before firm adhesion through interaction with carbohydrate ligands. TNF-α, LPS and other cytokines have been shown to up-regulate expression of VCAM-1, ICAM-1 and E-selectin [14,15], by increasing the transcription of specific mRNA [15] and by inducing the translocation of nuclear factor NF-κB to the nuclei of treated cells [16].

Although there is a wealth of data suggesting that aPL antibodies activate ECs in vitro and in vivo and they enhance thrombosis, and these effects seem to be mediated by ICAM-1, VCAM-1 and P-selectin [6–9], it is unclear what is the relative role of E-selectin in this process. To examine this question several experiments were performed both in vitro and in vivo. E-selectin-deficient [E-sel (−/−)] mice were utilized to determine the role of E-selectin on aPL-mediated enhancement of thrombosis and increased leukocyte adhesion to ECs in the microcirculation of mouse cremaster muscle. Thrombogenic effects were examined by using a previously described mouse model of induced thrombosis. We also examined whether the effects observed in vivo correlated with effects in vitro. Utilizing human umbilical endothelial cells (HUVECs) we found that aPL increased expression of E-selectin to HUVEC in vitro. The effects of aPL antibodies on activation of NF-κB were also examined.

Our findings indicate that E-selectin plays a role in aPL-induced EC activation and thrombus formation in vivo. APL antibodies also enhanced expression of E-selectin and significantly activated NF-κB.

Materials and methods

C57BL/6 J-E-sel-deficient mice [E-sel (−/−)] male weighing approximately 30 g and wild-type C57 BL/6J mice of the same weight and genders were purchased from Jackson Laboratories (Bar Harbor, MA, USA).

Affinity purification aPL antibodies and control IgG

APL antibodies (IgG-APS) were obtained from the sera of one APS patient by ion exchange chromatography, as previously described [3,9]. The protein concentration of the purified antibodies was determined by the method described by Lowry [17] and was adjusted to the required concentration using sterile phosphate-buffered saline. Control IgG (IgG-NHS) was isolated in a similar fashion from a pool of 10 sera from normal healthy controls. preparations were checked for absence/presence of endotoxin (LPS) by the limulus lysate assay (ameolysate; ICN Biochemical, Costa Mesa, CA, USA).

Anticardiolipin activity

ACL activity was determined in the affinity purified aPL and control IgG samples, as well as in the sera of mice treated with IgG-APS or with IgG-NHS by ELISA as described [18,19]. ACL activity was expressed in GPL units.

Cell culture

The established cell line of human umbilical vein endothelial cells (HUVECs; ATCC) was maintained in MCDB110 medium (Sigma Chemical Co., St. Louis, MO, USA) supplemented with 10% FBS, 100 U mL−1 penicillin, 100 mg mL−1 streptomycin, 2 mm l-glutamine, 100 mg mL−1 heparin, and 30 mg mL−1 H-Neurext (Upstate Biotechnology, Inc., Lake Placid, NY, USA) on collagen-coated tissue culture dishes. All experimental data was obtained using HUVECs in their 17th to 19th passage and at 1–2 days postconfluency.

In vitro detection of surface E-selectin expression on ECs

The ability of aPL antibodies from one APS patient to modulate the expression of the adhesion molecule, E-selectin, was assessed as previously described with a colorimetric ELISA [20,21]. HUVECs (104 cells well−1) were seeded in collagen-coated 96-well Plate 24 h before confluent monolayers were incubated with complete MCDB110 medium, normal IgG (IgG-NHS; 100 µg mL−1 in RPMI culture medium), or IgG-APS (100 µg mL−1 in RPMI) for 4 h at 37 °C in a 5% CO2-humidified incubator. As a positive control, some HUVEC monolayers were activated by pretreating with lipopolysaccharide (LPS, 3 µg mL−1, Sigma) or tumor necrosis factor-a (TNFα, 40 ng mL−1, Sigma) for 4 h at 37 °C in a 5% CO2-humidified incubator to increase the surface antigen expression of E-selectin. HUVEC monolayers were 2% paraformaldehyde-fixed and then incubated with phosphate-buffered saline/1% bovine serum albumin for 30 min at room temperature to block non-specific binding. Each subsequent step of the ELISA was conducted at room temperature with three washes of phosphate-buffered saline/1% bovine serum albumin between steps. The fixed monolayers were incubated in turn with a saturating concentration of anti-E-selectin (BD PharMingen, San Diego, CA, USA) and a peroxidase-conjugated goat antimouse IgG (Sigma) for 1 h. O-Phenylenediamine (0.4 mg mL−1)/0.012% hydrogen peroxide in a 0.05/0.025 mol L−1 phosphate/citrate buffer, pH 5.0, was added to each well. Color development was stopped at 3 mol L−1 H2SO4 at 20 min, and the OD was read at 492 nm wavelength on a SpectraMax 250 ELISA plate reader (Molecular Devices, Sunnyvale, CA, USA). The degree of specific antigen expression was calculated by subtracting non-specific binding of the secondary antibody from all test values.

Measurement of activation of NF-κB

NF-κB transcriptional activitation was determined by transiently transfecting cardio-pulmonary aortic endothelial (CPAE) cells with the pNF-κB-luciferase vector driven by a herpes simplex virus thymidine kinase promoter or the thymidine kinase-luciferase vector (negative control) along with pEGFP-C1 [22]. The pNF-κB-luciferase vector contains four copies of kappa enhancer element within the promoter region. Once activated, endogenous NF-κB binds to the kappa enhancer element and activates luciferase expression. Twenty-four hours post-transfection, CPAE cells were exposed to IgG-APS (100 µg mL−1) or with 100 µg mL−1 control IgG (IgG-NHS) for 4 h. Treatment with human TNF-α (40 ng mL−1) for 4 h was used as a positive control. Luciferase activity was measured with a luminometer (Victor II, Wallac) according to the manufacturer's directions (Promega) and normalized by assessing EGFP fluorescence (Victor II, Wallac). All data is expressed as fold activation of luciferase activity/EGFP fluorescence over control transfected cells under baseline conditions.

In vivo experiments

Pre-surgical procedures

 E-Selectin-deficient mice [E-sel (−/−)] or the wild-type mice (C57BL/6 J) in groups of nine were injected intraperitoneally (i.p.) twice with 500 µg IgG-APS antibodies or IgG-NHS control, as previously described. Two surgical procedures were done in the same animal 72 h after the first injection.

Surgical procedures

 (i) Analysis of EC activation in the microcirculation of the exposed cremaster muscle in mice: Activation of ECs in the pretreated mice was assessed by direct visualization and quantitation of leukocytes (WBC) adhering to ECs in the microcirculation of the exposed cremaster muscle in mice as described elsewhere [7–9,23,24]. After a stabilization period of 30 min, the number of adhering WBCs that remained stationary for a period of 30 s (sticking) within five different venules (diameter, 25 to 35 µm) was determined. The means were calculated and compared between treated and control groups.

(ii) Analysis of thrombus dynamics: The analysis of thrombus dynamics in a mouse model has been described previously [2–4,7–9]. In brief, mice were anesthetized 72 h after the first injection with IgG-APS or IgG-NHS, and the right femoral vein was exposed. The vein was pinched with a standardized pressure to introduce an injury and induce a clot. Clot formation and dissolution in the transilluminated vein were visualized with a microscope equipped with a closed-circuit video system (including a color monitor and a recorder). Thrombus size (in square micrometers) was measured when the thrombus reached the maximum size by digitizing the image and tracing the outer margin of the thrombus. Three to five thrombi were successfully induced in each animal, and mean times for formation, disappearance, and total times were then computed for each group of injected animals. The person performing the surgery and measurements (XL) was blinded to what treatment had been given to each animal.

Statistical analysis

In the in vitro experiments, values are expressed as means ± SEM. Monocyte adherence and E-selectin expression was compared between among the various test groups using the Student's t-test. The symbol *denotes a statistical difference (P = 0.05) between test and corresponding control groups.

For the in vivo experiments, the number of animal needed per group was determined by power analysis. It was established that 9–10 animals would provide a statistical power (1-β) for detection of test-group differences in both thrombus size and activation of ECs in vivo. The unpaired Student's t-test was used to compare the means of thrombus sizes and adhering WBC numbers between treated and control groups. Statistical significance was considered to be achieved when P < 0.05.


In vitro effects of aPL antibodies on expression of E-selectin on EC monolayers

Surface antigen expression of E-selectin was determined following exposure to aPL antibodies (IgG-APS). There was extremely low levels of constitutive expression of E-selectin on the surface of unstimulated HUVEC monolayers (Fig. 1). However, incubation of HUVEC with IgG-APS augmented E-selectin expression by 421-fold (Fig. 1, 3rd hatched bar). A similar, robust increase in E-selectin expression was observed on HUVEC monolayers activated with LPS or TNF-α. Both of these inflammatory stimuli induced E-selectin expression by about 400-fold.

Figure 1.

Effect of aPL antibodies (IgG-APS) on surface antigen expression of E-selectin on HUVEC monolayers. Each bar represents the mean ± SEM of quadriplicate wells in a representative experiment of three separate experiments. *Significantly different from control or IgG-NHS control group.

Effects of aPL on activation of NF-κB on endothelial cells

As shown in Fig. 2, treatment of endothelial cells with IgG-APS (100 µg/mL) for 4 h significantly increased transcriptional activation of NF-κB (135% increase = mean 2.3-fold increase over baseline P < 0.0001). TNF-α (40 ng mL−1) was used as a positive control in these experiments. TNF-α significantly activated NF-κB (103% increase = mean 2.0-fold increase over baseline, P= 0.0019). On the other hand, ECs treated with IgG-NHS did not activate NF-κB significantly (14% increase over baseline).

Figure 2.

Effects of aPL antibodies (IgG-APS) on activation of NF-κB on endothelial cells. Luciferase activity is expressed as fold increase activity over the baseline (cells treated with medium). Experiments were run three times and the results are expressed as mean ± SD. *denotes statistically significant different from IgG-NHS control group.

In vivo experiments

APL EC activation is impaired in E-sel (−/−)-deficient mice 

IgG-APS increased significantly the number of adhering leukocytes to EC in vivo in C57BL/6 J mice when compared to IgG-NHS treated mice 35.0 ± 12 vs. 14.0 ± 5 (P = 0.002). Mice treated with IgG-NHS were negative for aCL antibodies. In comparison with wild-type mice, adhesion of leukocytes to endothelium was significantly less in E-sel (−/−) mice whether treated with IgG-NHS or aPL (P-values = 0.003 and 0.0001, respectively). For animals treated with IgG-NHS, adhesion of leukocytes was 2.3 ± 0.5 (Table 1). For animals injected with IgG-APS, the number of adhering leukocytes was 5.3 ± 4.3 for E-sel (−/−) compared with 35 ± 12 for wild type, as noted above (P = 0.0001). There was no statistical difference in leukocyte adhesion in E-sel (−/−) mice infused with aPL (5.3 ± 4.3) compared with those infused with IgG-NHS (2.3 ± 0.5) (P = NS).

Table 1.  Effects of aPL antibodies on adhesion of leukocytes to endothelium in E-selectin-deficient mice
MiceTreatmentaCl titer
(GPL units)
No. of adhering leukocytes
(mean ± SD)
  • *

    Statistically significant different from wild-type mice treated with IgG-NHS(P = 0.002).

  • **

    Statistically significant different from wild-type mice treated with IgG-APS (P = 0.0001).

  • aCL titer at the time of surgery.

  • statistically significant different from wild-type mice treated with IgG-NHS (P = 0.003).

C57BL/6 J (wild type)IgG-NHSNegative14.0 ± 5
C57 BL/6 J (wild type)IgG-APS89.5 ± 35.635.0 ± 12*
E-Selectin (−/−)IgG-NHSNegative2.3 ± 0.5
E-Selectin (−/−)IgG-APS69.4 ± 11.55.3 ± 4.3**

C57 BL/6 J mice and E sel (−/−)-deficient mice treated with IgG-APS had medium to high titers of aCL antibodies at the time of the surgical procedures (89.5 ± 35.6 and 69.5 ± 11.5 GPL units, respectively).

APL-enhanced thrombosis is abrogated in E-sel (−/−) mice 

C57BL/6 J (wild type) mice treated with IgG-APS, as indicated in materials and methods produced significantly larger thrombi (2475.2 ± 856.0 µm2) compared with mice treated with IgG-NHS (mean thrombus size, 896.5 ± 254.0 µm2) (P = 0.002). The ability of IgG-APS to enhance thrombus formation was significantly reduced in E-sel (−/−) mice (mean thrombus size: 642.6 ± 392.0 µm2; (P = 0.002) (Table 2). These values were not statistically significant different from the corresponding control group treated with IgG-NHS, where thrombus size was 575.0 ± 426.0 µm [2] (Table 2) (P = NS). There was no difference in thrombus size between E sel (−/−) mice treated with IgG-NHS and thrombus size in wild-type mice treated with IgG-NHS [642.6 ± 392.0 µm2 vs. 896.5 ± 254.0 µm2 (NS)].

Table 2.  Effects of aPL antibodies on thrombus formation on deficient E-selectin-deficient mice
MiceTreatmentThrombus size (µm2)
  • *

    Statistically significant different from wild-type mice treated with IgG-NHS (P = 0.002).

  • **

    Statistically significant different from wild-type mice treated with IgG-APS (P = 0.002).

C57BL/6 J (wild type)IgG-NHS896.5 ± 254.0
C57BL/6 J (wild type)
 C57 BL/6 J (wild type)IgG-APS2475.2 ± 856.0*
 E-Selectin (−/−)IgG-NHS575.0 ± 426.0
 E-Selectin (−/−)IgG-APS642.6 ± 392.0**


Although there is a wealth of data suggesting that aPL antibodies activate ECs in vitro and in vivo and enhance thrombosis, it was unclear until recently what the relative roles of the individual adhesion molecules in these processes, are [5–8,25]. In a recent study, we reported [9] that the thrombogenic effects of aPL antibodies are mediated by ICAM-1, VCAM-1 and P-selectin. In this study, the data indicate that aPL antibodies up-regulate the expression of E-selectin, causing increased adhesion of leukocytes to ECs and activation NF-κB in vitro. aPL enhanced leukocyte adhesion and enhanced thrombus formation in wild type (C57BL/6 J mice), as has been observed previously [7–9]. In the absence of E-selectin expression, there is no aPL-induced enhancement of WBC adhesion to ECs nor is thrombus size enhanced. The latter observation suggests a role of E-selectin in aPL-induced thrombosis.

Under normal conditions, vascular endothelium maintains an anticoagulant surface of blood vessels [26]. The influence of ECs on the thombosis pathway is complex and involves multiple cell-surface and secreted compounds. Key elements in this process are the expression of tissue factor, tissue factor-pathway inhibitor, and thrombomodulin and cell adhesion molecules on the surface of ECs. The up-regulation of adhesion molecules on ECs induces increased monocyte adherence to endothelium with increased production of tissue factor and the generation of a hypercoagulable state in ECs as a consequence [27,28]. Hence, our mouse model of microcirculation, the adhesion of leukocytes to endothelium, can be used as an indication of EC activation in vivo.

It is known that the rolling and adhesion of leukocytes to the endothelium involves several sequential steps. Initially the interaction of selectins (E- and P-selectin) allows leukocytes to adhere reversibly to the vessel wall, so that circulating leukocytes can be seen to roll along the endothelium [12,14,15]. The first adhesive interaction permits the stronger interaction mediated by ICAM-1 and VCAM-1. In this study, leukocytes that remain stationary for 30 s or more in the ECs of the cremaster muscle are counted as adhering leukocytes. Thus, early disruption of the interaction of leukocytes to ECs in E-sel (−/−) mice may lead to a complete abrogation of the adhesion of leukocytes to the vessel wall, as observed in our studies. What is relevant in this study is that significant increase of adhesion of leukocytes to endothelium induced by aPL was abrogated in knockout mice. Most importantly, this study shows that the enhancement of thrombus size mediated by aPL was abrogated in E-sel (−/−) indicating that E-selectin expression may be at least one of the important factors in aPL-mediated thrombosis. This observation regarding E-selectin in aPL-mediated thrombosis have not previously been reported in vivo.

We cannot exclude the possibility that mechanisms other than EC activation may be involved in the pathogenesis of the thrombotic events associated with aPL antibodies. Studies clearly indicate that abnormalities in platelet function or aPL interfere with phospholipid–protein complexes may play a critical role in regulation of coagulation [29–31]. Such molecules as protein C [32], thrombomodulin [33], or tissue factor [34], are undoubtedly important.

The adhesion molecule E-selectin belongs to a family of adhesive receptors found in leukocytes. Stimuli commonly found in inflammatory processes, such as the cytokines TNF-α or IL-1, concurrently induce the expression of E-selectin and also VCAM-1 and ICAM-1 on ECs in a concentration and time-dependent fashion, by increasing the transcription of specific mRNA [13–16]. The intracellular signals activated during this process are now known. For example it is known that those cytokines induce the translocation of NF-κB to the nuclei of treated cells [16]. Inactive NF-κB is localized in the cytoplasm of most unstimulated cells in complex with its inhibitor, I kappa B. Phosphorylation of I kappa B, followed by its degradation, allows the translocation of NF-κB into the nucleus to activate NF-κB-dependent gene transcription. In our study, IgG-APS activated NF-κB on ECs to a comparable degree as TNF-α (used as a positive control). This was concomitant with an increased expression of E-selectin on HUVECs and increased adhesion of leukocytes to HUVECs.

The conversion of the normal antithrombotic endothelial phenotype to a prothrombotic surface might be one of the pathological events that causes the hypercoagulable state of the antiphospholipid syndrome [35]. Studies by Del Papa and others showed that aPL antibodies bind to ECs through interaction with β2GPI and induce activation of ECs as measured by increased adhesion molecule expression and up-regulation of cytokine secretion and arachidonic acid metabolism [36–38]. Simantov et al. showed that the up-regulation of adhesion molecules (ICAM-1, VCAM-1 and E-Selectin) on HUVECs when cells were treated with aPL antibodies in the presence of β2GPI in vitro, were mediated through the F(ab)′2 fragment of the antibody [5]. In another study by George et al. up-regulation of expression of ICAM-1, VCAM-1 and E-selectin by some murine monoclonal antibodies in vitro on HUVECs, correlated with increased fetal resorption in mice in vivo[25]. Our group showed that affinity purified aPL antibodies from patients with APS and murine monoclonal aPL antibodies up-regulated the expression of ICAM-1, VCAM-1 and E-selectin in vitro, and these effects correlated with enhanced thrombosis and leukocyte adhesion in vivo[7,8]. In another recent study by our group, we demonstrated that those effects were abrogated in ICAM-1 knockout, ICAM-1/P-selectin double knockout mice and by infusions with monoclonal anti-VCAM-1 antibodies [9]. Furthermore recent studies suggested increased levels of soluble ICAM-1 and VCAM-1 are observed in patients with APS and this correlated with recurrences in thrombosis [10,11]).

In summary, this study provides strong evidence that E-selectin expression on activated ECs is important in mediating thrombus formation in patient with APS and based on this study and our previous results [9], we propose an integrated model of aPL-induced endothelial cell activation in which its thrombogenic effects are mediated by adhesion molecules. These new findings show that not only ICAM-1, VCAM-1 and P-selectin but also E-selectin is involved. This study and data from our previous publication [9] did not enable conclusions to be drawn about the relative importance of any one adhesion molecule compared with the other three. Abrogation of any one molecule negates the aPL effect, suggesting that either all four adhesion molecules are required or that aPL stimulation of the expression of any of three molecules does not compensate for the abrogation of a fourth molecule. We do not discard the possibility that other molecules such as tissue factor may also play an important role in aPL-mediated thrombogenesis.

The observations made in this study may have important implications in determining new approaches to the treatment and prevention of thrombosis in APS.