Shear Stress and Cyclic Circumferential Stretch, But Not Pressure, Alter Connexin43 Expression in Endothelial Cells

Abstract

Hemodynamic forces play a critical role in atherogenesis, as evidenced by the focal pattern of development of atherosclerotic lesions. Whereas disturbed flow in the branches and curved regions of large arteries is proatherogenic, laminar flow in the straight parts of vessels is atheroprotective. In addition, hypertension and age-related changes in arterial stiffness are important risk factors of the disease. Hemodynamic forces induce various changes in the structure and function of vascular endothelium, many of which reflect alterations in gene expression. Endothelial cells are linked by gap junctions, which facilitate the propagation of electrical and chemical signals along the vascular wall. Using an in vitro perfusion system, we investigated the effects of pulsed unidirectional and oscillatory flows in combination with different levels of hydrostatic pressure and circumferential stretch on the expression of Cx43 in endothelial cells. Our results show that shear stress and circumferential stretch, but not pressure, modulate the expression of Cx43. In view of the distribution of this protein along the vascular tree, our findings provide new insights into the role of mechanical forces on gap junctional communication in regions prone to the development of atherosclerosis.

Keywords:

• gap junction
• connexin
• atherosclerosis
• hemodynamics

INTRODUCTION

Atherosclerosis is a focal disease that principally affects the aorta, the carotid, the coronary and the iliac arteries. This progressive disease is thought to be initiated by a dysfunctional state of the vascular endothelium (1, 2). Possible causes of endothelium dysfunction include elevated and modified LDL, free radicals caused by cigarette smoking, hypertension, age, diabetes or infectious microorganisms (3, 4). Regardless of the risk factor profile of individual patients, the early lipid-rich lesions of atherosclerosis show a clearly nonrandom distribution within the arterial vasculature. Atherosclerotic plaques often form at branch points or at curved areas of large arteries, (i.e. in regions which face a blood flow with low mean shear stress values and a cyclic reversal flow direction) (5). Conversely, the endothelium of plaque-free regions is exposed to a relatively high unidirectional shear stress. The suggestion that local fluid dynamic factors may contribute to the focal distribution of lesions is supported by in vitro experiments. Indeed, endothelial cells (ECs) subjected to sustained unidirectional shear stress show elevated expression of proteins associated with an atheroprotective phenotype, such as endothelial nitric oxide synthase and Cap G (6). In contrast, oscillatory shear stress fails to induce the expression of such genes but induces the expression of genes implied in atherosclerotic plaque development, such as preproendothelin-1 gene (5, 7).

Vascular endothelial cells are linked by gap junctions, which facilitate the propagation of electrical and chemical signals along the vessel wall. These communication pathways have been implicated in a variety of vascular functions, such as the coordination of vasomotor responses, the regulation of angiogenesis, the repair of the endothelial lining, and endothelial senescence (8, 9, 10, 11). In situ, the gap junctions of arterial endothelium are known to consist mainly of Cx40 and Cx37 (12, 13, 14, 15, 16). Whereas Cx43 is minimally expressed or absent in quiescent endothelia, a growing number of observations suggest that this protein may be induced under conditions associated with endothelium dysfunction. In a study examining rat aorta and its bifurcation, high levels of Cx43 were exclusively localized in areas of endothelium facing turbulent blood flow (17). During atherogenesis, Cx43 expression is induced in the endothelium at the shoulder region of advanced atherosclerotic plaques, a localization known to experience disturbed hemodynamic forces (18). Furthermore, we recently observed reduced progression of atherosclerotic plaque development in a mouse model of atherosclerosis expressing half the normal levels of Cx43 (19, 20). These studies suggest a link between endothelial Cx43 expression and hemodynamic conditions that may be relevant to focal vulnerability to atherosclerosis.

The hemodynamic regulation of Cx43 has been studied in ECs using different in vitro models. Thus, it has been reported that Cx43 mRNA increased in ECs subjected to sustained levels of high laminar shear stress (21). In contrast, De Paola et al. (22) observed a short-term increase in Cx43 under both laminar and disturbed flow that were sustained in regions of flow disturbance but returned to control levels in areas of undisturbed laminar flow. The cause of these different responses on shear stress is still unknown. In addition to device-induced differences in the experimental protocol, cell origin might play a role. For example, qualitative and quantitative differences in the pattern of connexins may determine different responses in cultured ECs to hemodynamic forces.

In vivo, the arterial endothelium is subjected to a complex mechanical environment. In addition to fluid shear stress, pulsed pressure generates a cyclic circumferential strain on the entire arterial wall. Consequently, the cellular effects observed under two-dimensional culture conditions and in response to single mechanical loads may not necessarily reflect those taking place under the multifactorial, three-dimensional in vivo environment. The perfusion system developed in our laboratory (23) allows for exposure of ECs to a complex mechanical environment which closely mimicks that observed in vivo. Using this device, we have determined the combined effects of unidirectional or oscillatory shear stress and cyclic circumferential stretch on levels of Cx43 in bEnd.3 cells, the only available EC that coexpresses in vitro the three endothelial connexins (Cx37, Cx40, Cx43) detected in the endothelial cells of arteries in situ. In addition, we took advantage of this perfusion system by examining the effect of high pulse pressure on endothelial Cx43 expression in these cells.

MATERIALS AND METHODS

Experimental Flow System

The perfusion system used in this study has been described in detail (23). Briefly, it consists of four custom-manufactured compliant tubes (Sylgard 184, Dow Corning Europe) mounted onto fittings, which are connected to a flow system consisting of a temperature- and gas-controlled reservoir, and a gear pump (MCP-Z, Ismatec) driven by a pulse generator (HP-8116A, Hewlett-Packard). The pump system is capable of generating a constant flow rate onto which any desired waveform may be imposed. For the present study, an offset sinusoidal waveform with a 1 Hz frequency was chosen. The flow rates were monitored downstream of all four Sylgard tubes with transit time ultrasonic flow probes (Transonic Systems, Inc). Shear stress values were calculated using the method outlined by He et al. (24). The diameters of the tubes receiving circumferential strain (set at 4% diameter change) were monitored with a NIUS 02 ultrasonic echo tracking device (Asulab), which records diameter changes with an accuracy of 3 μm (0.05% of the present tube diameter (25)). The tubes in which radial extension was to be prevented were surrounded by rigid plastic shells. Control hydrostatic pressure measured inside the tubes was approximately 100 mmHg. In some experiments, it was elevated to 150 mmHg. When pulse pressure was modified, stretch was avoided with a solid cast over the compliant tubing.

Experimental Flow System

Sylgard tubes (70 mm in length, 6-mm inner diameter, and 0.19- to 0.24-mm wall thickness) were treated for 1 min in 70% sulfuric acid and then rinsed extensively with deionized water. Upon mounting onto the fittings, the tubes were subjected to an axial stretch of 10% to ensure that they would remain straight under pressurization. After sterilization in an autoclave, the inner surface of the tubes was coated with fibronectin (20 μg/ml in PBS, Boehringer Mannheim).

Cell Culture

PymT-transformed mouse endothelial cells, clone bEnd.3 (26), were maintained in DMEM supplemented with 10% fetal calf serum, 50 IU/ml penicillin, and 50 μg/ml streptomycin. Cells from one confluent culture flask (75 cm²) were seeded inside four Sylgard tubes at a final density of 5 × 10⁴ cells/cm² in culture medium on a rotating device that allowed obtaining an homogeneous confluent monolayer after 24 h. At this time, the fittings carrying the confluent cell cultures were connected to the previously assembled flow device and submerged in a temperature-controlled water bath at 37°C. The culture medium in the flow reservoir was supplemented with 2% clinical grade dextran (M_r = 70.000, Sigma) to increase the viscosity to 1.07 × 10⁻³ N m⁻² s⁻¹ and it was constantly gassed by a 5% CO₂/95% air mixture to keep a constant pH of 7.2. Cells were exposed to unidirectional flow (0.3 ± 0.1 or 6 ± 3 dynes/cm²) or to oscillatory flow (0.3 ± 3 dynes/cm²). Cells subjected to any of the experimental conditions remained attached as confluent monolayers over the 4- or 24-h period of exposure. Each experiment had its own internal static control culture, maintained in a 5% CO₂ atmosphere at 37°C during the time of the experiment.

Western Blotting

At the end of the experiment, cells were rinsed with cold PBS, harvested into an ice-cold solubilization buffer consisting of 50 mM Tris-HCl (pH 7.4), 150 mM NaCl, 1% sodium deoxycholate, 1% Nodinet-P40, 0.1% sodium dodecylsulfate, 2.5 mM sodium orthovanadate, 125 mM phenylarsine oxide and 2 mM phenylmethyl sulfonyl fluoride, and stored at -80°C. After thawing, the samples were centrifuged for 30 min at 13,000 × g and 4°C. Supernatants containing solubilized material were recovered, and total amounts of protein were quantified using a bicinchoninic acid quantification assay (Sigma). Ten μg protein were loaded on 12% SDS-polyacrylamide gels, electrophoresed, and electrotransferred onto nitrocellulose membranes (Biorad). Membranes were then soaked overnight at 4°C in a 2% defatted milk saturation buffer consisting of 10 mM Tris-HCl (pH 7.4), 2 mM EDTA, 133 mM NaCl, 0.05% Triton X-100 and 0.2% sodium azide. Blotted proteins were then incubated for 1 h at room temperature with rabbit polyclonal (2 μg/ml; Zymed Laboratories) or mouse monoclonal (1:250; BD Transduction Laboratories) Cx43 antibodies or anti β-actin (1:5000; Pharmingen), which was used to control for loading. After rinsing, the membranes were incubated 1 h with goat anti-rabbit or goat anti-mouse secondary antibodies conjugated to peroxidase (Jackson Laboratories), whichever appropriate. Specificity of the Cx43 labeling was confirmed by preabsorption of the polyclonal antibodies for 15 min at room temperature with 20 μg/ml cognate immunogenic peptide (Zymed Laboratories). Immunoreactivity was detected using the ECL chemiluminescent detection kit (Amersham), according to the manufacturer's instructions. The chemiluminescence reaction was visualized on Biomax ML film (Eastman Kodak) and band intensity was quantified by densitometry. Levels of Cx43 were related to those of β-actin for each sample. Results were normalized to the static controls and presented as mean ± SEM. Statistical analysis on results of 3–5 independent experiments was performed using a two-tailed Student's t test. Data were considered statistically significant at P < 0.05.

RESULTS

Pulsed Unidirectional and Oscillatory Shear Stress Increase Cx43 in ECs

bEnd.3 cells were subjected to unidirectional and oscillatory flow for either 4 or 24 h, and total protein extracts were then analyzed for the expression of Cx43. Cx43 antibodies recognized a triplet of proteins migrating between 41 and 47 kDa in all extracts (Figure 1A). Densitometric analysis of 3–5 independent experiments showed that Cx43 expression was upregulated when bEnd.3 cells were subjected to increased levels of unidirectional or oscillatory shear stress. However, the kinetics of flow-induced Cx43 expression were different for unidirectional and oscillatory shear stress. Indeed, 4 h of oscillatory shear stress already induced a significant increase in Cx43 expression, compared to static control cultures, and the levels of the protein further increased with continued exposure up to 24 h (Figure 1C). In contrast, increased levels of Cx43 in response to unidirectional shear stress were only observed after 24 h (Figure 1B). At that time, the expression of the connexin was dependent on the level of unidirectional shear stress, being significantly higher at 6 dyne/cm² than at 0.3 dyne/cm² (Figure 1B). In addition to the different kinetics, the increase in Cx43 expression reached higher levels (up to 3-fold) under oscillatory shear than under unidirectional shear stress (Figure 1B, Figure 1C).

Figure 1. Effects of fluid shear stress on endothelial Cx43 expression. (A) Western blot for Cx43 (top lanes) and β-actin (bottom lanes) in bEnd.3 cells exposed to static conditions (lanes 1), low (lanes 2) or high (lanes 3) unidirectional shear for 24 h in combination with 100 mmHg mean pressure and no circumferential stretch. β-actin was used as a control for loading. (B) bEnd.3 cells exposed to static conditions, low or high unidirectional shear for 4 or 24 h in combination with 100 mmHg pulse pressure and no circumferential stretch. Quantification of Western blots is expressed as the ratio of Cx43/β-actin for each sample. (C) bEnd.3 cells exposed to static conditions or to oscillatory shear stress for 4 or 24 h in combination with 100 mmHg pulse pressure and no circumferential stretch. Quantification of Western blots is expressed as the ratio of Cx43/β-actin for each sample. Data are mean ± SEM of 3–5 independent experiments. *P < 0.05 compared to static controls; **P < 0.05 compared to 24 h of low unidirectional shear stress; ***P < 0.01 compared to 4 h of oscillatory shear stress.

Hydrostatic Pressure Does Not Affect Cx43 in ECs

To examine whether Cx43 expression in bEnd.3 cells was affected by hydrostatic pressure, cells were subjected to various unidirectional shear stress in combination with mean pulse pressures of 100 or 150 mmHg, with a pulsation of ± 20 mmHg, for 24 h. As depicted in Figure 2, increasing mean pressure level by 50 mmHg did not change the level of endothelial Cx43 at low unidirectional shear stress of 0.3 dyne/cm² in three independent experiments. Increasing the level of shear stress to 6 dyne/cm² caused a significant increase in Cx43 expression. However, at this higher unidirectional shear stress, Cx43 levels were similar whether hydrostatic pressure was 100 or 150 mmHg.

Figure 2. Effects of pulse pressure on endothelial Cx43 expression. bEnd.3 cells were exposed to low or high unidirectional shear in combination with 100 or 150 mmHg mean pressure and no circumferential stretch for 24 h. Quantification of Western blots is expressed as the ratio of Cx43/β-actin for each sample. Data are mean ± SEM of 3 independent experiments. *P < 0.05 compared to low unidirectional shear stress.

Cyclic Circumferential Stretch Increases Cx43 in ECs

In addition to wall shear stress, pulsed pressure generates a cyclic circumferential stretch on the arterial endothelium. To assess whether Cx43 expression in bEnd.3 cells was also affected by circumferential stretch, bEnd.3 cells were subjected to various shear stress in combination with mean pulse pressure of 100 mmHg and either a nil or a 4% circumferential stretch. Cx43 expression in bEnd.3 cells exposed to 4% stretch was upregulated as compared to no-stretch conditions (Figure 3). Interestingly, the kinetics of stretch-induced Cx43 expression were different for unidirectional and oscillatory shear stress. Four hours of unidirectional shear stress in combination with 4% stretch induced a modest increase in Cx43 (Figure 3A) and expression of the protein remained stable with continued exposure up to 24 h (data not shown). Stretch-induced expression of Cx43 was dose-dependent on the level of unidirectional shear stress, being significantly increased only when shear stress was 6 dyne/cm². In contrast, 4% circumferential stretch in combination with oscillatory shear stress, caused a 7-fold increased in Cx43 expression at 4 h, an effect that returned to 3.5-fold of no-stretch controls at 24 h (Figure 3B, Figure 3C).

Figure 3. Effects of cyclic circumferential stretch on endothelial Cx43 expression. (A) bEnd.3 cells exposed to low or high unidirectional shear in combination with 100 mmHg pulse pressure and no or 4% circumferential stretch for 24 h. Quantification of Western blots is expressed as the ratio of Cx43/β-actin for each sample. Data are mean ± SEM of three independent experiments. (B) Western blot for Cx43 (top lanes) and β-actin (bottom lanes) in bEnd.3 cells exposed to oscillatory shear for 4 h in combination with 100 mmHg pulse pressure and no (lanes 1) or 4% (lanes 2) circumferential stretch. β-actin was used as a control for loading. (C) bEnd.3 cells exposed oscillatory shear stress in combination with 100 mmHg pulse pressure and no or 4% circumferential stretch for 4 or 24 h. Quantification of Western blots is expressed as the ratio of Cx43/β-actin for each sample. Data are mean ± SEM of 4 independent experiments. *P < 0.05 compared to no-stretch controls. **P < 0.01 compared to 4 h of cyclic stretch.

DISCUSSION

In this study, we have addressed Cx43 expression in bEnd.3 cells after exposure to different combinations of pulse pressure, shear stress and cyclic circumferential stretch. The main findings are that (1) oscillatory shear stress significantly increased endothelial Cx43 expression within 4 h, reaching a 3-fold increase after 24 h; (2) at the same time point, unidirectional shear stress induced a more modest increase in Cx43 expression; (3) increasing hydrostatic pressure from 100 to 150 mmHg did not affect Cx43 expression levels at various levels of shear stress; and (4) 4% cyclic circumferential stretch combined with oscillatory shear stress induced a 7-fold increase of Cx43 levels within 4 h, decreased by half after 24 h.

The effect of unidirectional shear stress on endothelial Cx43 expression is controversial. Thus, whereas one study documented an increase in Cx43 levels of ECs in response to sustained levels of high laminar shear stress (21), no change was reported in another study (22). In this study, pulsed unidirectional shear stress was associated with a modest increase in Cx43 expression in bEnd.3 cells. This effect was dependent on the level of shear stress and reached significance only after 24 h. In contrast, endothelial Cx43 expression rapidly increased in response to oscillatory shear stress within 4 h, and continued to rise during the next 24 h. This enhanced Cx43 expression in cells exposed to oscillatory flow is consistent with immunohistochemical data showing endothelial Cx43 at flow dividers, upstream of valves and at the shoulder of atherosclerotic plaques (17, 18, 27). In such regions, the endothelium is characterized by an increased permeability, an increased expression of adhesion molecules, and a decreased expression of endothelial nitric oxide synthase (6, 7, 28, 29, 30). It is becoming increasingly clear that the level of gap junctional communication might affect gene expression of other proteins via a regulation of the cell-to-cell transfer of factors important for gene transcription (31). Whether changes in Cx43 expression levels are causally related to the proatherogenic EC phenotype remains to be investigated.

The possible relation between connexin expression and blood pressure has been extensively studied. Mice deficient in Cx40 display a hypertensive phenotype and endothelial-specific deletion of Cx43 leads to hypotension in mice (32, 33, 34). Spontaneous hypertensive rats show increased levels of Cx40 (12). Furthermore, rats made hypertensive by clipping of one renal artery (2K, 1C model) or by administration of deoxycorticosterone and salt (DOCA-salt model) display increased levels of aortic Cx43 (35). In contrast, rats made hypertensive by inhibiting nitric oxide synthase with N G-nitro-L-arginine methyl ester (L-NAME model) showed decreased levels of aortic Cx43 (36). Whereas hypertension in the 2K, 1C and DOCA-salt rats was associated with increased arterial distensibility, this was not observed in L-NAME rats, indicating that the increase in Cx43 in some hypertensive animals may have been due to the chronic stretch of the arterial wall rather than to the increased blood pressure. Moreover, due to the large mass of smooth muscle cells (SMCs) expressing Cx43 in the intact vessel, these previous studies do not provide direct evidence about eventual concurrent changes in endothelial Cx43. Our experimental model allows for the independent variation of hydrostatic pressure and wall stretch over ECs, under conditions mimicking the three-dimensional arrangement these cells have in situ. Using this model, we observed that increased hydrostatic pressure did not alter the endothelial Cx43 response at various levels of shear stress.

Previous studies have shown increased Cx43 expression in response to chronic mechanical stretch in smooth muscle cells of the uterus, vascular wall, and bladder, as well as in osteoblasts and cardiomyocytes (21, 37, 38, 39, 40, 41, 42). At all time points and levels of shear stress studied here, the presence of 4% circumferential stretch did enhance the Cx43 response in ECs. The induction of endothelial Cx43 expression occurred rapidly after imposition of cyclic circumferential stretch, the maximal response being observed already at 4 h. This time course is similar to that observed in other cell types, in which stretch has been associated with increased Cx43 gene transcription (21, 43). The mechanism underlying the stretch-induced increase in endothelial Cx43 expression remains, however, to be elucidated.

Although our experimental model has many advantages over intact vessels for studying endothelial Cx43 expression in response to individual hemodynamic forces, it should be stressed that in vivo ECs interact with other cell types of the vascular wall. Physical contact between ECs and SMCs via myo-endothelial cell junctions has been demonstrated (44). In addition, leukocytes have been reported to couple to ECs, thus affecting their transmigration (45, 46, 47). Thus, it is conceivable that bidirectional crosstalk between different cell types the vascular wall may influence the response of ECs to hemodynamic forces. It should also be recalled that bEnd.3 cells, which we selected for these experiments because of their unique in vitro coexpression of the three connexin isoforms that are observed in intact arteries (11), are of capillary origin (26). Therefore, our data do not exclude that arterial ECs may behave differently when exposed to the mechanical forces explored here.

In conclusion, we have found that shear stress and cyclic circumferential stretch, but not hydrostatic pressure, alter Cx43 expression in bEnd.3 cells. Previous studies have shown that Cx43 is exclusively expressed in endothelium of regions prone to develop atherosclerotic lesions. Knowing that the progression of atherosclerosis is reduced in mice expressing reduced levels of Cx43, these results provide new insights into the possible link between hemodynamics and early atherogenesis.

ACKNOWLEDGEMENTS

We thank Tecla Dudez, Fabienne Burger, and Isabelle Roth for excellent technical assistance as well as Marc Chanson for critical reading of the manuscript. This work was supported by grants from the Swiss National Science Foundation (#PPOOA68883 and #3100-067777 to BRK, #3100A0-103823 to PS, #310000-109402 to PM), the Swiss University Conference (to BRK and DH), the Fondation Leenaards (to BRK and PS), the Juvenile Diabetes Research Foundation (#1-2005-46 to PM) and the National Institute of Health (#DK-63443-01 to PM).

Authors contributed equally to the study.