OUP user menu

Thrombomodulin domains attenuate atherosclerosis by inhibiting thrombin-induced endothelial cell activation

Hsi-Ju Wei, Yi-Heng Li, Guey-Yueh Shi, Shu-Lin Liu, Po-Chiao Chang, Cheng-Hsiang Kuo, Hua-Lin Wu
DOI: http://dx.doi.org/10.1093/cvr/cvr220 317-327 First published online: 12 August 2011

Abstract

Aims Thrombin modulates the formation of atherosclerotic lesions by stimulating a variety of cellular effects through protease-activated receptor-1 (PAR-1) activation. Thrombomodulin (TM) inhibits thrombin effects by binding thrombin through its domains 2 and 3 (TMD23). We investigated whether recombinant TMD23 (rTMD23) could inhibit atherosclerosis via its thrombin-binding ability.

Methods and results Wild-type mouse rTMD23 and three mutants with altered thrombin-binding sites, rTMD23 (I425A), rTMD23 (D424A/D426A), and rTMD23 (D424A/I425A/D426A), were expressed and purified in the Pichia pastoris expression system. Wild-type rTMD23 and rTMD23 (D424A/D426A) could effectively bind thrombin, activate protein C, and prolong thrombin clotting time, whereas rTMD23 (I425A) and rTMD23 (D424A/I425A/D426A) lost these functions. Wild-type rTMD23, but not rTMD23 (I425A), decreased both the thrombin-induced surface PAR-1 internalization and the increase in cytoplasmic Ca2+ concentrations in endothelial cells (ECs). Wild-type rTMD23 and rTMD23 (D424A/D426A) also inhibited thrombin-induced adhesion molecules and monocyte chemoattractant protein-1 expression and increased permeability in ECs, whereas rTMD23 (I425A) and rTMD23 (D424A/I425A/D426A) had no such effects. Furthermore, wild-type rTMD23 and rTMD23 (D424A/D426A) were effective in reducing carotid ligation-induced neointima formation in C57BL/6 mice and atherosclerotic lesion formation in apolipoprotein E-deficient (ApoE−/−) mice, whereas rTMD23 with the I425A mutation showed impairment of this function. Wild-type rTMD23, but not rTMD23 (I425A), also markedly suppressed the PAR-1, the adhesion molecules expression, and the macrophage content in the carotid ligation model and ApoE−/− mice.

Conclusion rTMD23 protein significantly reduces atherosclerosis and neointima formation through its thrombin-binding ability.

  • Atherosclerosis
  • Protease-activated receptor-1
  • Thrombin
  • Thrombomodulin

1. Introduction

Atherosclerosis is an inflammatory disease that develops in response to multiple injuries to the vascular endothelium.1 Endothelial dysfunction is closely associated with and is usually a harbinger of atherosclerosis formation. Abundant evidence suggests the key effector of the coagulation cascade, thrombin, has multiple pro-inflammatory effects on the endothelium, causes endothelial dysfunction, and promotes atherogenesis.2 Thrombin facilitates the recruitment of circulating monocytes by increasing endothelial expression of monocyte chemoattractant protein-1 (MCP-1), intercellular adhesion molecule-1 (ICAM-1), and vascular cell adhesion molecule-1 (VCAM-1).3,4 All these proatherogenic effects of thrombin are activated via protease-activated receptors (PARs), a subfamily of cell surface G protein-linked receptors.5 Previous studies have demonstrated that there is significant PAR expression in atherosclerotic lesions, and neointima formation was attenuated after arterial injury in PAR-1-deficient mice.6,7 Although several other serine proteases could also activate PARs by cleaving their N-terminal extracellular domains, mounting experimental data suggest that thrombin is the key regulator of coagulation, inflammation, and atherosclerosis formation.

Thrombomodulin (TM) is an endothelial cell (EC) membrane-bound glycoprotein that functions as a thrombin receptor on the endothelial surface. TM has five domains: an N-terminal lectin-like domain (D1), six epidermal growth factor (EGF)-like repeats (D2), a serine/threonine-rich region (D3), a transmembrane domain (D4), and a short cytoplasmic tail (D5). The fifth and sixth EGF repeats (EGF5–6) of TM domain 2 (TMD2) form the binding site of thrombin and are responsible for decreasing thrombin activity and activating protein C.8 Recombinant TM domains 1, 2, and 3 (rTMD123) could inhibit neointima formation in the mouse carotid ligation model.9 Previous report8 and our recent finding10 demonstrated that TM domain 1 (TMD1) has direct anti-inflammatory function independent of activated protein C (APC). Therefore, it is not clear whether rTMD123 inhibited neointima formation is through the direct anti-inflammatory function of TMD1 or through domains 2 and 3 (TMD23). A previous study using rapid alanine-scanning mutagenesis indicated that Ile-424 in the fourth to sixth EGF repeats (EGF4–6) region of human TM is the most important amino acid residue in influencing thrombin-binding and protein C activation.11 Therefore, in the present study, we constructed and purified the mouse recombinant TMD23 (rTMD23) protein and three mutants with altered thrombin-binding activity. We tested the hypothesis that rTMD23 may decrease activation of PAR-1 by binding thrombin and inhibiting thrombin-induced cellular effects. Administration of rTMD23 reduced atherosclerosis and neointima formation in mice.

2. Methods

2.1 Materials

M199 medium, Dulbecco's modified Eagle's medium (DMEM), foetal bovine serum (FBS), and the pPICZαA vector were purchased from Invitrogen (Carlsbad, CA, USA). S-2238 and S-2366 chromogenic substrates were purchased from DiaPharma Group Inc. (West Chester, OH, USA). Thrombin, antithrombin III, heparin, fura-2/AM, l-glutamine, and anti-α-smooth muscle actin (α-SMA) were purchased from Sigma-Aldrich (St Louis, MO, USA). Anti-PAR-1 (H-111), anti-c-Myc, anti-ICAM-1 (G-5), and anti-VCAM-1 (H-276) antibodies were purchased from Santa Cruz Biotechnology, Inc. (Santa Cruz, CA, USA). Transwell cell culture chamber (0.4 μm pore size) was purchased from Corning Costar Corp. (Cambridge, MA, USA). Anti-monocyte and macrophage (MOMA-2) antibody (ab33451) was purchased from Abcam (Cambridge, UK).

2.2 Purification of mouse recombinant TMD2 (EGF4–6) (rTMD2 (EGF4–6)) and rTMD23

Mouse rTMD2 (EGF4–6), wild-type rTMD23 and three mutants with altered thrombin-binding sites: rTMD23 (I425A), rTMD23 (D424A/D426A), and rTMD23 (D424A/I425A/D426A) were constructed in the pPICZαA vector and obtained by using Quick Change Site-directed Mutagenesis kit (Stratagene, Boston, MA, USA). The following primers were used (only mutagenic forward primers are shown): rTMD23 (I425A): 5′-GGTTCCGTATGCACGGACGCTGATGAGTGCAGTC-3′; rTMD23 (D424A/D426A): 5′-GGTTCCGTATGCACGGCCATTGCTGAGTGCAGTC-3′; and rTMD23 (D424A/I425A/D426A): 5′-CGAGGGTTCCGTATGCACGGCCGCTGCTGAGTGCAGTC-3′. Recombinant proteins were expressed in the Pichia pastoris expression system, purified by Ni+-chelating affinity column, examined by sodium dodecyl sulfate–polyacrylamide gel electrophoresis (SDS–PAGE), and blotted by anti-c-Myc antibody.

2.3 APC activity assay

rTMD23 (0.2 μM) was incubated in a 96-well plate. Samples and blank well were incubated with thrombin (0.2 μM) and protein C (5 μg/mL, Calbiochem, San Diego, CA, USA) at 37°C for 30 min. Reactions were stopped by incubation with antithrombin III (50 μg/mL) and heparin (1 μM) for 30 min. APC production was detected by addition of S-2366 chromogenic substrate (1.5 mM). The change in absorbance was measured at 405 nm every 3 min.

2.4 Thrombin clotting time assay

rTMD23 proteins (1.2 μM) were incubated with thrombin (0.6 μM) in a 96-well plate at 37°C for 5 min. The change in absorbance was measured at 450 nm every 10 s after adding 50 μL of human plasma to each well.

2.5 Cell culture

Human umbilical vein ECs (HUVECs) isolated from umbilical cords and human aortic ECs (HAECs) purchased from Cascade Biologics (Portland, OR, USA) were cultured in M199 medium containing 20% FBS, 1% EC growth supplement, and 1% heparin. Mouse monocyte cell line RAW264.7 purchased from Bioresource Collection and Research Center (Taiwan) and human aortic smooth muscle cells (HASMCs) purchased from Clonetics (San Diego, CA, USA) were cultured in DMEM containing 10% FBS and 2 mM l-glutamine. Cells were studied following approval by the National Cheng Kung University Ethics Committee. The investigation conforms with the principles outlined in the Declaration of Helsinki.12

2.6 PAR-1 internalization assay

Wild-type rTMD23 and rTMD23 (I425A) (0.1 μM) were incubated with thrombin (25 nM) at 37°C for 20 min and used to treat the HUVECs, HAECs, and HASMCs for 3 min. After blocking, cells were incubated with anti-PAR-1 and Alexa Fluor 488 goat anti-rabbit IgG antibodies. The fluorescence profile of the cells was analysed by a fluorescence-activated cell sorter (FACSort, BD Bioscience, Franklin Lakes, NJ, USA).

2.7 Cytoplasmic Ca2+ measurement

[Ca2+]i levels of HUVECs, HAECs, and HASMCs were analysed as described previously.13 Wild-type rTMD23 and TMD23 (I425A) (0.1 μM) were incubated with thrombin (25 nM) in Kreb's–Ringer buffer at 25°C for 10 min and applied to each well. [Ca2+]i levels were calculated according to the general formula: [Ca2+]i= Kd× [(RRmin)/(RmaxR)] × β, where Kd equals 224 nM under our experimental condition.

2.8 Transendothelial permeability assay

HUVECs and HAECs were grown in the upper compartment of Transwell cell culture chambers. rTMD23 proteins (0.1 μM) were incubated with thrombin (60 nM) at 37°C for 20 min and used to treat cells for 15 min. Serum-free medium containing 10 nM horseradish peroxidase (HRP) was added for 5 min. HRP concentration was detected via the absorbance at 450 nm by loading tetramethylbenzidine.

2.9 MCP-1 and tumour necrosis factor-α production assay

HUVECs, HAECs, and RAW264.7 cells were starved in serum-free medium for 24 h. rTMD23 proteins (0.1 μM) were incubated with thrombin (60 nM) at 37°C for 20 min and used to treat cells for 6 h (HUVECs and HAECs) and 24 h (RAW264.7 cells). Supernatants were harvested. MCP-1 and tumour necrosis factor-α (TNF-α) were measured with the monoclonal antibody-based enzyme-linked immunosorbent assay (ELISA) kits (R&D Systems).

2.10 Mouse carotid ligation model

The left common carotid artery of adult C57BL/6 mice (age, 8–10 weeks) was ligated as described previously.14 After the surgery, the mice immediately received 145 μg/kg/day of rTMD23 proteins or phosphate-buffered saline (PBS) via subcutaneously implanted osmotic pumps (model 2004, Durect Corporation, Cupertino, CA, USA) for 4 weeks. The mice were sacrificed and perfused with PBS and 4% paraformaldehyde in PBS by placement of a 22 G needle in the left ventricle. Carotid artery was excised, fixed in 4% paraformaldehyde, and embedded in the OCT compound. The carotid artery was first cut off 200 μm from the proximal to the ligature, and five transverse sections (400–800 μm) of each carotid artery were cut at 100 μm intervals, stained with haematoxylin–eosin, and measured as described previously.14 Furthermore, five transverse sections of each carotid artery were stained with antibodies against PAR-1, ICAM-1, and VCAM-1 for 7 days and α-SMA for 4 weeks after carotid ligation, respectively. The percentage of the lesion area with red (AEC) staining against the total area was recorded for each section.

2.11 Apolipoprotein E-deficient (ApoE−/−) mice

Apolipoprotein E-deficient (ApoE−/−) mice (age, 8 weeks; Jackson Laboratory, Bar Harbor, ME, USA) were fed a high-cholesterol diet containing 21% fat and 0.15% cholesterol (PMI LabDiet 40097; Richmond, IN, USA) for 20 weeks. During this period, mice received 145 μg/kg/day of rTMD23 proteins or PBS via osmotic pumps (model 2004, Durect Corporation) for the first 4 weeks. Serum levels of total cholesterol, LDL, HDL, and triglycerides were measured by enzymatic methods with an automatic analyzer (Model 747, Hitachi Ltd Co., Japan). The atherosclerotic lesions in the whole aorta were identified and measured as described previously.14 The composition of atherosclerotic lesions at the level of the aortic arch was specifically analysed. A series of 15–20 sections (thickness, 5 μm), which represented the central area of the arch with an intact morphology of the entire arch, were used for analysis.15 Five consecutive sections of each aortic arch were, respectively, stained for macrophages, smooth muscle cells (SMCs), PAR-1 expression, elastin fibre breaks, and collagen content; for the staining of each marker, three sections were used. Lesion size (intima and media), percentage of the total area of the aortic arch stained for macrophages, SMCs, PAR-1, and collagen content were analysed using a previously reported method.16 For each mouse, the percentage of the positively stained area for each marker was presented as the mean of three sections. Macrophages, SMCs, and PAR-1 expression were stained with anti-MOMA2, anti-α-SMA, and anti-PAR-1 antibodies (green-fluorescent signal). Elastin fibre breaks and collagen were detected with Verhoeff-van Gieson's stain and Masson's trichrome stain. Fibrous cap thickness was measured according to the α-SMA staining.

2.12 Animal care

Animal experiments were approved by Institutional Animal Care and Use Committee of National Cheng Kung University, Taiwan; the investigation conforms with the Guide for the Care and Use of Laboratory Animals published by US National Institutes of Health (NIH publication no. 85-23, revised 1996).

2.13 Statistical analyses

Data are expressed as mean ± SD. Statistical significance was analysed by unpaired Student's t-test or one-way ANOVA with the Bonferroni corrections. A value of P< 0.05 was considered statistically significant.

3. Results

3.1 Purification of mouse rTM domain proteins

Alignment of the amino acid sequence of thrombin-binding regions of human and mouse TM reveals that Ile-424 of human TM is equivalent to Ile-425 of mouse TM (Figure 1A). Purified rTMD23 proteins had one major band with a molecular mass of 50 kDa and a smear band (due to various degree of glycosylation), and rTMD2 (EGF4–6) had molecular masses of 28 and 32 kDa (Figure 1B, left). Proteins could be recognized by anti-c-Myc antibody (Figure 1B, right). All rTMD2 (EGF4–6) and rTMD23 proteins with different molecular masses had expected sequences that started with the fusion peptide, Glu-Phe, followed by the sequences of TM Pro-347 and TM Gly-223, respectively.

Figure 1

Purification and functional assays of mouse rTM domain proteins. (A) Alignment of the amino acid sequences of the thrombin-binding regions of human and mouse TM. Alignment was made by Vector NTI (Invitrogen). Identical amino acids between the two species were marked in red with a yellow background, and similar amino acids were marked with a green background. Alignment revealed a highly conserved region for thrombin binding of TM between human and mouse sequences. (B) Analysis of rTMD23 proteins by gel electrophoresis. SDS–PAGE was performed under reducing conditions, and proteins were detected by Coomassie Blue staining (left) and western blotting (right). (C) rTMD23 was incubated with the reaction mixture (thrombin, protein C, antithrombin III, and heparin). APC activity was measured with the S-2366 chromogenic substrate and the absorbance at 405 nm was detected. (D) Thrombin or rTMD23/thrombin complex was incubated in a 96-well plate at 37°C for 5 min. After human plasma was added, the change in absorbance was measured at 450 nm. Similar results were obtained in three independent experiments.

3.2 Functional assays of rTMD23

When we analysed the ability of rTMD23 to activate protein C using a thrombin-dependent protein C activation assay, wild-type rTMD23 and rTMD23 (D424A/D426A) could activate protein C, whereas rTMD23 (I425A) and rTMD23 (D424A/I425A/D426A) had no effect (Figure 1C). A thrombin-induced plasma clotting assay was used to check the anticoagulant function of rTMD23. Treatment with thrombin induced plasma clotting after 40 s. Plasma clotting after incubation with thrombin combined with rTMD23 (I425A) or rTMD23 (D424A/I425A/D426A) was also detected at 40 s, whereas thrombin combined with wild-type rTMD23 or rTMD23 (D424A/D426A) had a prolonged plasma clotting time of 120 s (Figure 1D). These results indicated that rTMD23 with Ile-425 mutation could completely inhibit thrombin-binding ability and thrombin-related coagulation function.

3.3 rTMD23 effect on endothelial PAR-1 activation

The primary mechanism for downregulation of the PAR-1 signalling cascade is PAR-1 internalization and degradation.17 Thrombin induction of the PAR-1 signalling cascade also results in increased endothelial intracellular [Ca2+]i.18 As shown in Figure 2A–F, thrombin induced PAR-1 activation and internalization. Wild-type rTMD23, but not rTMD23 (I425A), could decrease the activation of endothelial surface PAR-1 in HUVECs (Figure 2A and B), HAECs (Figure 2C and D), and HASMCs (Figure 2E and F). We further investigated endothelial intracellular [Ca2+]i differences after thrombin treatment by using fura-2/AM-loaded cells. Wild-type rTMD23, but not rTMD23 (I425A), could decrease the thrombin-induced [Ca2+]i increase in HUVECs (Figure 2G), HAECs (Figure 2H), and HASMCs (Figure 2I). The PAR-1 internalization and [Ca2+]i measurement assays proved that rTMD23 could directly decrease the thrombin-induced endothelial PAR-1 activation through the thrombin-binding site (Ile-425).

Figure 2

Wild-type rTMD23 treatment inhibited thrombin-induced PAR-1 internalization and Ca2+ release in HUVECs, HAECs, and HASMCs. Thrombin was incubated with wild-type rTMD23 or rTMD23 (I425A) for 20 min before incubation in HUVECs (A and B), HAECs (C and D), or HASMCs (E and F) for 3 min. Surface PAR-1 was detected by flow cytometry. In the [Ca2+]i measurement assay, thrombin was incubated with rTMD23 wild-type or rTMD23 (I425A) in Kreb's–Ringer buffer for 10 min before incubation in HUVECs (G), HAECs (H), or HASMCs (I) at 500 s. Cell fluorescence was measured. TH, thrombin. Similar results were obtained in three independent experiments.

3.4 rTMD23 effect on endothelial permeability and adhesion molecule expression

Thrombin treatment increases endothelial permeability through PAR-1 activation.18 As shown in Figure 3A and B, endothelial permeability of HUVECs and HAECs was increased after thrombin treatment. Preincubation of wild-type rTMD23 or rTMD23 (D424A/D426A) with thrombin significantly reduced the thrombin-induced increase in endothelial permeability. However, rTMD23 (I425A) and rTMD23 (D424A/I425A/D426A) had no such effect. As shown in Figure 3C and D, thrombin treatment induced ICAM-1 and VCAM-1 expression in HUVECs and HAECs. Pre-treatment with wild-type rTMD23 or rTMD23 (D424A/D426A) decreased thrombin-induced ICAM-1 and VCAM-1 expression, whereas rTMD23 (I425A) had no such effect.

Figure 3

Wild-type rTMD23 and rTMD23 (D424A/D426A) reduced thrombin-induced increases in endothelial permeability and adhesion molecules expression. Thrombin was incubated with wild-type rTMD23 or mutants for 20 min before incubation in HUVECs (A) or HAECs (B) for 15 min in the upper compartment. The medium was replaced by serum-free medium containing HRP for 5 min. The HRP concentration at the lower compartment was detected at 450 nm. The results were expressed as mean values ± SD of three experiments. *P< 0.05 compared with the thrombin-treated cultures. Thrombin was incubated with wild-type rTMD23 or mutants for 20 min before incubation in HUVECs (C) or HAECs (D) for another 6 h. The lysates were analysed for ICAM-1 and VCAM-1 expressions by western blotting. Similar results were obtained in three independent experiments.

3.5 rTMD23 effect on MCP-1 and TNF-α production

In ECs, thrombin induces the production of MCP-1 to facilitate recruitment of circulating monocytes into the arterial wall and causes early plaque formation. In addition, monocytes are also sensitive to thrombin activation by their surface PAR-1 expression and consequently release the inflammatory cytokine TNF-α. Thrombin treatment induced MCP-1 and TNF-α release in the media of HUVECs, HAECs, and RAW264.7 cells when compared with those induced by PBS treatment. Preincubation of wild-type rTMD23 or rTMD23 (D424A/D426A) with thrombin significantly decreased MCP-1 and TNF-α release in HUVECs, HAECs, and RAW264.7 cells. However, preincubation of rTMD23 (I425A) or rTMD23 (D424A/I425A/D426A) with thrombin had no such effect (Figure 4A–C).

Figure 4

Wild-type rTMD23 and rTMD23 (D424A/D426A) decreased thrombin-induced MCP-1 and TNF-α release in HUVECs, HAECs, and RAW264.7 cells. Thrombin was incubated with wild-type rTMD23 or mutants for 20 min before incubation in HUVECs (A), HAECs (B), or RAW264.7 cells (C). Supernatants were harvested. MCP-1 (A and B) and TNF-α (C) were measured by ELISA. The results are expressed as mean values ± SD of three experiments. *P< 0.05 compared with the thrombin-treated cultures.

3.6 TM effect on neointima and atherosclerosis formation

The carotid ligation model and ApoE−/− mice were used to investigate the effect of TM domains on neointima and atherosclerosis formation. In the carotid ligation model, there was a progressively decreased lumen area and increased neointima formation in C57BL/6 mice after carotid artery ligation for 4 weeks. The neointima to media ratio (N/M ratio) was significantly decreased in rTMD2 (EGF4–6), wild-type rTMD23, and rTMD23 (D424A/D426A) treatment groups (Figure 5A). Furthermore, rTMD23 (I425A) and rTMD23 (D424A/I425A/D426A) treatment had no such effect on reducing carotid neointima formation (Figure 5B and C). In addition, early in the atherosclerotic process, leucocytes adhere to the endothelium, where adhesion molecules are expressed.19 Increasing growth factor and cytokine secretions induced by the interaction between leucocytes and the endothelium result in proliferation of SMCs and neointima formation.20 The results shown in Figure 5D–G demonstrated that wild-type rTMD23, but not rTMD23 (I425A), treatment could significantly reduce PAR-1, ICAM-1, and VCAM-1 expression in the intima after carotid ligation for 7 days. However, there was no difference among PBS, wild-type rTMD23, and rTMD23 (I425A) treatment in α-SMA content in the neointima after carotid ligation for 4 weeks (Figure 5D and H). In ApoE−/− mice, there were no significant differences in the total cholesterol, triglyceride, LDL, and HDL levels between PBS treatment and TM domains treatment groups (data not shown). Wild-type rTMD23, rTMD23 (D424A/D426A), or rTMD2 (EGF4-6) treatment reduced the atherosclerotic lesion formation. Otherwise, rTMD23 (I425A) or rTMD23 (D424A/I425A/D426A) treatment had no effect in decreasing the atherosclerotic lesion formation (Figure 6A and B). Endothelial dysfunction is considered the earliest pathological signal of atherosclerosis; it results in macrophage invasion of atherosclerotic lesions. The predominant receptor for thrombin on ECs appears to be PAR-1, which can be characterized as a contributing factor to endothelial dysfunction.2 Wild-type rTMD23, but not rTMD23 (I425A), treatment could markedly decrease the expression of PAR-1 and MOMA-2 in the aortic arch lesion area (Figure 6C–E). Apart from the endothelial dysfunction, collagen contents, elastin fibre disruption, and fibrous cap thickness reflect SMC migration from the media to the intima.1 rTMD23 proteins had no significant effect on α-SMA, collagen contents, elastin fibre disruptions, and fibrous cap thickness (Figure 6C, F, and G).

Figure 5

Influence of treatment with rTMD2 (EGF4-6) and rTMD23 proteins on neointima formation in a mouse model of carotid intima thickening. (A) Representative photomicrographs of haematoxylin–eosin staining of arterial sections (600 μm) after ligation for 4 weeks in mice receiving PBS (n= 7), rTMD2 (EGF4-6) (n= 10), wild-type rTMD23 (n= 8), rTMD23 (I425A) (n= 12), rTMD23 (D424A/D426A) (n= 10), or rTMD23 (D424A/I425A/D426A) (n= 14) injection. Arrowheads indicate borders of the neointima and media. Original magnification ×200 (bar = 100 μm). (B) Morphometric analysis of the lumen, neointima, media, and total vascular areas. Values are mean ± SD of five sections in each mouse. **P< 0.01 compared with the mice receiving PBS injection. (C) Degree of neointima formation after carotid ligation for 4 weeks was calculated according to the neointima/media area (N/M) ratio. ***P< 0.001 compared with the mice receiving PBS injection and ###P< 0.001 compared with the wild-type rTMD23 treatment group. (D) Representative photomicrographs of PAR-1, ICAM-1, VCAM-1 and α-SMA staining of carotid arterial sections after carotid ligation in mice receiving wild-type rTMD23, rTMD23 (I425A), or PBS. Original magnification ×400 (bar = 100 μm). Quantification of PAR-1 (E)-, ICAM-1 (F)-, VCAM-1 (G)-, and α-SMA (H)-positive staining areas in arterial sections. Data are presented as mean ± SD (n= 7–9). *P< 0.05 compared with the mice receiving PBS treatment.

Figure 6

Influence of treatment with rTMD2 (EGF4-6) and rTMD23 proteins on the formation of atherosclerotic lesions in ApoE−/− mice. (A) ApoE−/− mice received PBS (n= 12), rTMD2 (EGF4-6) (n= 12), wild-type rTMD23 (n= 12), rTMD23 (I425A) (n= 12), rTMD23 (D424A/D426A) (n= 12), or rTMD23 (D424A/I425A/D426A) (n= 11). The lipid-rich atherosclerotic lesions were identified with Oil-Red-O staining. (B) Lesion area (%) was expressed as a percentage of the total area of the aorta. The results were expressed as mean values ± SD. **P< 0.01 and ***P< 0.001 compared with the PBS treatment group and #P< 0.05 and ##P< 0.01 compared with the wild-type rTMD23 treatment group. (C) Representative photomicrographs of PAR-1, MOMA-2, α-SMA, Masson's trichrome, and Verhoeff-van Gieson's staining of arterial sections with ApoE−/− mice receiving wild-type rTMD23, rTMD23 (I425A), or PBS. Original magnification ×200, ×400 for Verhoeff-van Gieson's stain (bar = 100 μm). Quantification of PAR-1 (D)-, MOMA-2 (E)-, and α-SMA and collagen (F)-positive staining areas in aortic arches sections. (G) Verhoeff-van Gieson's stain for elastin fibre disruption showed a number of discontinuities in the elastic lamina in aortic arch sections. Fibrous cap thickness of each mouse was measured using sections stained with α-SMA. Data are presented as mean ± SD (n= 7–9). **P< 0.01 compared with the mice receiving PBS treatment.

4. Discussion

The biological function of thrombin received much attention recently due to its proatherogenic cellular effects through PAR-1 activation. The active amount of thrombin is the net product of synthesis from the coagulation cascade and inhibition from antithrombin and heparin cofactor II. However, the major physiological effect of thrombin is determined by its binding to the endothelial receptors including PAR-1 and TM. Previous studies have demonstrated that progressive atherosclerosis and neointima formation are associated with a diminution of endothelial TM and increased PAR-1 expression levels.7,9 It appears that the imbalance between PAR-1 and TM expression with augmented thrombin effects plays an important role in atherosclerosis formation and progression. Some of the effects of PAR-1-activated downstream G protein-coupled signal transduction pathways are the generation of inositol-1,4,5-triphosphate and increased intracellular Ca2+.21 A previous study demonstrated that a soluble TM protein containing the TMD2 domain is a potent inhibitor of thrombin-induced inositol-1,4,5-triphosphate production in ECs.22 In this study, we further substantiated that treatment with wild-type rTMD23, but not mutant protein rTMD23 (I425A), decreased thrombin-induced internalization of PAR-1 and elevation of intracellular Ca2+ in ECs. These results proved that rTMD23 protein could effectively decrease PAR-1 activation through binding thrombin and downregulate PAR-1 downstream signalling activation. PAR-1 activation-induced cellular effects, such as endothelial permeability, adhesion molecules expression, and cytokine production, were also decreased after rTMD23 treatment. Because these effects are all involved in the pathogenesis of atherosclerosis, we suggested that supplementation with only rTMD23 to reduce PAR-1 activation may decrease atherosclerosis formation.

Thrombin has three regions, the active site, exosite I, and exosite II, which are functionally inaccessible in prothrombin and become exposed upon activation. Exosite I, located 20–25 Å away from the active site moiety, is the major binding epitope for fibrinogen, TM, and PAR-1. Soluble TM can bind to thrombin at exosite I with very high affinity through its EGF5–6 in the D2 domain.8 In our study, the mutant experiment demonstrated that rTMD23 decreased PAR-1 activation through thrombin binding by EGF5–6. The glycosylation of serine and threonine residues is also involved in thrombin–TM complex formation by promoting binding to the exosite II of thrombin. Adequate TM glycosylation enhances its thrombin binding, but the influence of glycosylation on the physiological effect of TM needs further investigation.

The findings that rTMD23 could suppress the proatherogenic cellular effects of thrombin prompted us to look at its effect on atherosclerosis formation. Previous investigation has shown that treatment rTMD123 decreased neointima formation after vascular injury.9 In this study, we demonstrated that rTM domains that can effectively bind and decrease thrombin activity, including rTMD2 (EGF4–6), wild-type rTMD23, and rTMD23 (D424A/D426A), could reduce atherosclerosis and neointima formation. In the carotid ligation model, the altered flow condition after ligation causes blood stasis, coagulation activation, and inflammation, all of which increase the expression of endothelial adhesion molecules and the subsequent influx of monocytes into the vessel wall.23 Blocking the upstream thrombin effect with rTMD23, such as PAR-1, ICAM-1, and VCAM-1 expression, attenuated neointima formation after ligation. Thrombin also plays a critical role in spontaneous atherogenesis. Repeated endothelial disruption and microthrombus formation are among the major causes of atheroma progression. Persistent thrombin activity can be detected in atherosclerotic plaques.24 ApoE−/− mice develop lesions that have many histological features of human atherosclerosis.25 In our study, rTMD2 (EGF4–6), wild-type rTMD23, and rTMD23 (D424A/D426A) treatment were effective in attenuating the atherosclerotic lesion formation. Wild-type rTMD23, but not rTMD23 (I425A), reduced PAR-1, adhesion molecules, and macrophage expression, which are indicators of endothelial dysfunction in the atherosclerotic process. rTMD23 treatment had no significant effects on the SMC and collagen contents, extent of elastin fibre disruption, and fibrous cap thickness, which were related to SMC migration and proliferation. In summary, our results demonstrate that rTMD23 could inhibit thrombin-induced PAR-1 activation and decrease the atherosclerotic lesion formation.

Another possible mechanism of rTMD23 anti-inflammatory effect is through APC. Thrombin binds to TM on the endothelial surface resulting in APC formation that blocks the procoagulant activity of thrombin.26 The protein C pathway contributes not only anti-coagulant activity but also anti-inflammatory functions. Structural similarity of the EC protein C receptor to the MHC class I/CD1 family of proteins, most of which are involved in inflammatory processes, suggests that the function of the EC protein C receptor may not be limited to its ability to localize APC or protein C to the endothelial membrane.27 APC can also downregulate pro-inflammatory cytokine production and favourably alter tissue factor expression or blood pressure.2830 The anti-inflammatory effects of APC can be divided into its effects on ECs and its effects on leucocytes. APC diminishes cytokine release from leucocytes and thereby may attenuate systemic inflammatory responses. It was thought that APC suppressed inflammation-activated transcription factors of the activator protein-1 (AP-1) family, c-Fos and FosB, which directly induce ICAM-1 and MCP-1 in ECs.31 In this study, we found that thrombin-induced release of MCP-1 and TNF-α was reduced by wild-type rTMD23 and rTMD23 (D424A/D426A), but not by rTMD23 with the I425A mutation, in HUVECs, HAECs, and RAW264.7 cells, suggesting that APC may mediate the anti-inflammatory activity of rTMD23.

Recent studies have implicated high-mobility group box chromosomal protein 1 (HMGB1) as a possible contributor to atherogenesis.32,33 The effects of HMGB1 on macrophages include elevations in the secretion of TNF-α, interleukin (IL)-1α, IL-1β, IL-6, and macrophage inflammatory proteins. HMGB1 also increases pro-inflammatory responses in ECs and induces chemotaxis in SMCs. Transgenic expression of human TM in mice reduced HMGB1 levels after injection of lipopolysaccharide.34 HMGB1 would be degraded by the TM–thrombin complex to a less pro-inflammatory form.35 We found that rTMD23 significantly reduced neointima formation and atherosclerosis due to its thrombin-binding ability. rTMD23 might play a critical role in reducing atherosclerosis through promoting proteolytic cleavage of HMGB1 by thrombin.

In conclusion, the present study demonstrates that the thrombin-binding ability of TMD23 proteins plays an important role in reducing atherosclerosis and neointima formation. The therapeutic potential of TM in patients with atherosclerotic diseases is worthy of further investigation.

Conflict of interest: none declared.

Funding

This work was supported by National Science Council, Taiwan (Grants NSC 97-2752-B-006-003-PAE, NSC 97-2752-B-006-004-PAE, NSC 97-2752-B-006-005-PAE, and DOH100-TD-B-111-002).

References

View Abstract