Morphological Responses of Vascular Endothelial Cells Induced by Local Stretch Transmitted Through Intercellular Junctions

Abstract

It has been well established that mechanical stimuli including fluid shear stress and cyclic stretch play a key role in endothelial cell (EC) remodeling. However, in contrast to global remodeling to these mechanical stimuli, little is known of how local mechanical forces are transmitted through cells to induce cell remodeling leading to alteration in cell functions. In this study, we demonstrated that EC remodeling can be exerted by local tension generated in a neighboring EC. In this technique, a glass microneedle was used to apply local stretch in an EC in confluent monolayer and the resulting tension is transmitted to a neighboring EC across intercellular junctions. Local stretch induced reorientation and elongation of ECs parallel to the direction of stretch associated with reorganization of stress fibers. In addition, recruitment of Src homology 2-containing tyrosine phosphatase-2, binding to intercellular adhesion molecules platelet-endothelial cellular adhesion molecules-1, was selectively observed at the force-transmitted intercellular junctions after application of local stretch. These findings suggest that intercellular junctions can not only transmit but also sense local forces, and are potentially involved in EC mechanotransduction pathways.

Introduction

Endothelial cells (ECs) lining inner surfaces of blood vessels are exposed to a variety of mechanical stimuli including fluid shear stress due to blood flow, cyclic stretch due to the wall deformation, and hydrostatic pressure due to blood pressure. Since the relationship between EC morphology and mechanical conditions has been implicated in vascular pathology such as atherosclerosis, in vitro model systems have been widely used to investigate the effects of mechanical stimuli on EC remodeling. Most previous studies have exclusively applied global mechanical stimuli including fluid shear stress [1–4], cyclic stretch [5, 6] and hydrostatic pressure [7, 8] to cultured EC monolayer. For example, under exposure to fluid shear stress, ECs exhibit elongation and alignment parallel to the direction of flow concomitant with reorganization of actin stress fibers (SFs) [2, 4]. It is also well known that ECs exposed to cyclic stretch elongate and align perpendicular to the direction of stretch [5]. Furthermore, pressure-imposed ECs elongate with random orientation [7]. Thus, ECs respond very specifically to the type of mechanical stimuli, raising the question of how externally applied mechanical forces transmit in cells and where the transmitted forces are sensed and converted into biochemical signals. It has been suggested that mechanotransduction of ECs may occur at specific sites located in cell membrane (e.g. stretch activated channels (SACs)), attachment points of ECs to extracellular matrix (e.g. integrins), intercellular junctions (e.g. platelet-endothelial cellular adhesion molecules-1, PECAM-1) and more likely a combination of these sites. However, the details of these mechanotransducers in cell signaling pathways remain unclear, because, in part, there is a possibility that these mechanotransducers can be simultaneously activated by the globally applied mechanical stimuli. Therefore, it is necessary to develop an experimental method, with which each mechanotransducer can be selectively stimulated.

Intercellular junctions of ECs involve adherens-junctions, tight-junctions, gap-junctions and PECAM-1 junctions. Particularly, adherens-junctions composed of VE-cadherin which is connected to actin-cytoskeleton via α/β catenins are thought to exclusively mediate forces at intercellular junctions by forming homophilic binding, probably because mechanical coupling by adherens-junctions may be stiffer than the other junctions. Recently, there has been a growing attention on the role of intercellular adhesion molecules with implications of EC mechanotransduction signalling pathways. PECAM-1 is known to form homophilic binding between neighboring ECs [9] in the vicinity of adherens-junctions and be connected to actin-cytoskeletons via β/γ catenins [10]. Previous reports revealed that ECs under sparse conditions do not show morphological changes in response to fluid shear stress [11, 12]. Osawa et al. [13] reported that PECAM-1 is rapidly tyrosine-phosphorylated when ECs are exposed to fluid shear stress or hyper osmotic shock, and bind to a protein tyrosine phosphatase Src homology 2-containing tyrosine phosphatase-2 (SHP-2) to activate intracellular signal cascades. Tzima et al. [14] reported that PECAM-1-mediated mechanotransduction is involved in the upstream of the integrin signaling cascade, leading to cytoskeletal reorganization in flow conditioned ECs. This leads to an idea that transmitted forces at adherens-junctions may stimulate PECAM-1 phosphorylation, leading to SHP-2 activation followed by morphological changes of ECs. Thus, PECAM-1 may be responsible for morphological responses to mechanical stimuli transmitted through intercellular junctions serving as one of principal mechanotransducers.

To test this hypothesis, in this study, local stretch was applied to an EC by using a microneedle to selectively stimulate intercellular junctions between the EC and a neighboring EC. After application of mechanical stimuli, morphological changes of the neighboring EC were fluorescently observed to test the hypothesis that local stimuli are transmitted via intercellular junctions and induce EC remodeling. In addition, localization of SHP-2 was observed to assess whether PECAM-1 is involved in the mechanotransduction pathways.

Materials and Methods

Cell Culture and Transfection

Human umbilical vein endothelial cells (HUVECs) were obtained from umbilical veins with trypsin treatment adapted from Sakamoto et al. [15]. Cells were cultured in Medium 199 (Invitrogen, USA) containing 20% fetal bovine serum (Sigma-Aldrich, USA), 10 ng/ml basic fibroblast growth factor (Austral Biologicals, USA) and penicillin–streptomycin (Invitrogen), and were used from the third to the sixth passage. For experiments, cells were plated on a glass base culture dish (ø = 35 mm, Asahi Techno Glass, Japan) coated with 0.1% gelatin (Sigma-Aldrich). A plasmid encoding enhanced green fluorescent protein (EGFP)-actin (Clontech, USA) was transfected into HUVECs with a liposomal method using Lipofectin (Invitrogen) according to the manufacturer’s protocol for visualization of actin cytoskeleton.

Local Stretch Experiment

Figure 1 shows the experimental setup used to apply local stretch. A glass microneedle with a diameter of ca. 1 μm was made from a glass tube (ø = 1 mm, GD-1, Narishige, Japan) using a pipette puller (PP-83, Narishige, Japan) and manipulated by a 3-D hydraulic micromanipulator. A pair of ECs (cell 1 (C₁) and cell 2 (C₂)) in the figure) expressing EGFP-actin were selected under a confocal laser scanning microscope (CLSM, FV1000, Olympus, Japan), and the nucleus of C₂ was then moved horizontally by 10 μm by manipulating the glass microneedle to locally stretch C₁. This procedure allows us to mimic cell deformation induced by externally applied local tension via intercellular junctions between C₁ and C₂. In a separate study, particle tracer velocimetry (PTV) method was performed to calculate distribution of intracellular deformation induced by local stretch by using a software (Flowvec, Library, Japan) (Fig. 2). The result showed that high displacement was observed specifically at the intercellular junctions between C₁ and C₂, confirming force transmission across the junctions. After application of local stretch, fluorescent images were obtained up to 60 min at intervals of 5 min with the CLSM to assess features of C₁ remodeling process. All procedures were performed in a CO₂ incubator (MI-IBC, Olympus) mounted on the microscope stage to maintain the cell culture environment at 37°C and 5% CO₂.

Fig. 1

Schematic illustrations of experimental system for local stretch application. A microneedle was used to apply local stretch to C₂, possibly inducing remodeling of C₁ via intercellular junctions. The experiments were performed in a CO₂ incubator at 37°C and 5% CO₂

Fig. 2

Distribution of intracellular displacement calculated by using PTV method. Longer arrows represent higher displacement. Applied local stretch in C₂ was found to be transmitted to C₁ at intercellular junctions

Image Analysis

Image analysis was performed to evaluate morphological and cytoskeletal changes of C₁ using the following parameters: angle of cell orientation, cell aspect ratio, angle of SF orientation and uniformity index of SF. Angle of cell orientation and cell aspect ratio were calculated with the public domain program Image J version 1.37v (National Institute of Health, USA). A cell outline was manually extracted from a fluorescence image of GFP-actin by tracing the cell edge. An equivalent ellipse for the cell outline was automatically determined with a function of the software. An angle between the major axis of the ellipse and the direction of stretch was then defined as the angle of cell orientation. A ratio of length of the minor axis to the major axis was defined as the cell aspect ratio, with being 1.0 for circles and approaching zero for highly elongated shape. Analyses were performed on the fluorescent images obtained at every 5 min after application of local stretch. See Appendix for details of the angle of SF orientation and the uniformity index of SF.

SACs Blocking Experiment

In order to inhibit function of SACs, 20 μM GdCl₃ (Wako pure chemical, Japan) was added to cell culture medium at 30 min before application of local stretch.

Immunofluorescence

Localization of SHP-2 was examined with immunofluorescence. ECs were pre-incubated with 1 mM NaVO₃ (Wako Pure Chemical, Japan) for 30 min to enhance tyrosine phosphorylation before application of local stretch. Five minutes after the mechanical loading, ECs were then fixed in 4% paraformaldehyde (Wako Pure Chemical) containing 1 mM NaVO₃ for 15 min at 4°C. Next, ECs were permeabilized with 0.1% Triton X (Wako Pure Chemical), blocked with Block ace (Dainippon Sumitomo Pharma, Japan), and stained with anti-SHP-2 mouse monoclonal antibody (Santa Cruz Biotechnology, USA) and Alexa fluor 594 anti-mouse IgG (Invitrogen).

Statistical Analysis

Statistical comparisons in morphological parameters were performed using repeated measures analysis of variance (ANOVA) followed by Bonferroni’s multiple comparison (0 min vs 30 min, 60 min). A p value less than 0.05 was considered as significant. Data are expressed a mean ± SEM.

Results

Morphological Changes of ECs After Application of Local Stretch

Typical florescent images of EGFP-actin were shown in Fig. 3 at 0 min, 30 min and 60 min after application of local stretch. The observed cell C₁ initially elongated perpendicular to the direction of stretch [Fig. 3(a)] exhibited contraction at 30 min [Fig. 3(b)] followed by elongation parallel to the direction of stretch at 60 min [Fig. 3(c)]. Concurrently with these morphological changes, SFs initially orientated perpendicular to the direction of stretch at 0 min [Fig. 3(a), arrowheads], and disappeared [Fig. 3(b), arrowheads] while formation of new SFs parallel to the direction of stretch was observed [Fig. 3(b), arrow] at 30 min. The new SF formation was continuously observed at 60 min [Fig. 3(c), arrow].

Fig. 3

Typical examples of fluorescent images of EGFP-actin at (a) 0 min, (b) 30 min, and (c) 60 min after application of local stretch. Broken lines indicate the edge of C₁. The horizontally elongated C₁ at 0 min (a) showed contraction at 30 min (b) and re-elongated parallel to the direction of the stretch at 60 min (c). Preexisting SFs at 0 min (a) disappeared at 30 min (b) and new SFs formation was observed at 60 min (c). Bar = 40 μm

Figure 4 represents distribution of the angle of cell orientation and the cell aspect ratio of the observed cell C₁ at 0 min, 30 min and 60 min. From different types of initial cell morphology, the experimented data can be divided into two groups: Group A, ECs with angle of cell orientation >60° and aspect ratio <0.5; Group B, the rest of Group A. Roughly, distribution of the two groups is schematically drawn in the figure by ellipsoids. For both Groups, the angle of cell orientation significantly decreased from 0 min [Fig. 4(a), 71.3 ± 4.4°, 53.3 ± 4.6° for Groups A and B, respectively] to 60 min [Fig. 4(c), 52.1 ± 6.8°, and 29.0 ± 3.7° for Groups A and B respectively, p < 0.05 vs 0 min for both groups]. For Group A, aspect ratio significantly increased from 0 min [Fig. 4(a), 0.42 ± 0.02] to 30 min [Fig. 4(b), 0.57 ± 0.07, p < 0.05 vs 0 min] and then decreased at 60 min [Fig. 4(c), 0.53 ± 0.08, NS vs 0 min]. On the other hand, for Group B, aspect ratio did not show a significant change from 0 min [Fig. 4(a), 0.57 ± 0.05] to 30 min [Fig. 4(b), 0.55 ± 0.07] and significantly decreased at 60 min [Fig. 4(c), 0.47 ± 0.06, p < 0.05 vs 0 min].

Fig. 4

Changes in morphological parameters of ECs at (a) 0 min, (b) 30 min and (c) 60 min after application of local stretch. Group A represents cells with angle of cell orientation >60° and aspect ratio <0.5 and Group B, the rest of Group A. The two groups showed different time course of changes in the morphological parameters

Time course of change in the SF parameters is summarized in Fig. 5. Angle of SF orientation for both groups significantly decreased at 60 min [Fig. 5(a), p < 0.05, vs 0 min]. For Group A, change in the uniformity index was not significant between 0 min and 60 min [Fig. 5(b)] while the uniformity index significantly increased at 60 min for Group B [Fig. 5(b), p < 0.05 vs 0 min].

Fig. 5

Time course of changes in SF parameters. (a) Angle of SF orientation significantly decreased for both groups. (b) Change in uniformity index of SF was not significant for Group A, while the index significantly increased for Group B. See Appendix for details for determination of SF parameters. *p < 0.05 vs 0 min

SACs Blocking Experiment

SACs blocking experiment was performed on Group B with treatment of GdCl₃. Treated cells showed the similar time course of change in morphology, and morphological parameters showed no significant difference compared to those of non-treated control cells (data not shown). This result indicates that SACs should not be involved in the present remodeling.

Change in Distribution of SHP-2 After Application of Local Stretch

Figure 6 represents typical fluorescent images of EGFP-actin and anti-SHP-2 for control [Fig. 6(a,b,c)] and GdCl₃-treated ECs [Fig. 6(d,e,f)] after application of local stretch. For control ECs, recruitment of SHP-2 was clearly observed at the intercellular junctions between C₁ and C₂ [Fig. 6(b,c)]. For GdCl₃ -treated ECs, localization of SHP-2 was still observed at the intercellular junctions, suggesting that applied mechanical loading can be transduced into molecular signals [Fig. 6(e,f)].

Fig. 6

Typical examples of fluorescent images of SHP-2 for control (a, b, c) and GdCl₃ treated ECs (d, e, f). Boxed regions in (b) and (e) are magnified in (c) and (f), respectively. Expression of SHP-2 was found to be localized at intercellular junctions (b, d; arrowheads). Bar = 40 μm for (a, b, d, e), 20 μm for (c, f), respectively

Discussion

In this study, local tension transmitted via intercellular junctions was applied to ECs to induce morphological changes in a neighboring EC. The result showed that ECs elongate and orient parallel to the direction of stretch together with reorganization of stress fibers. Moreover, localization of SHP-2 was selectively observed at intercellular junctions, indicating that mechanotransduction might occur. Most previous studies relating to cell remodeling have utilized global mechanical stimuli including fluid shear stress [1–4] and cyclic stretch [5, 6]. With these conventional techniques, it is difficult to find both mechanotransmission pathways and mechanotransduction sites, probably because these events may occur by a combination of several candidate sensors. To overcome this difficulty, this study proposed a new experimental method to locally apply mechanical stimuli to ECs to identify the role of intercellular junctions involved in EC remodeling.

Interestingly, ECs finally aligned to the direction of stretch at 60 min after application of mechanical stimuli. The mechanism in the present EC remodeling process is unclear. It is well known that ECs exposed to cyclic stretch show alignment perpendicular to the direction of stretch [5]. Sokabe et al. [16] have suggested that when cells are subjected to cyclic stretch, actin cytoskeletons are disassembled during relaxation phase and disruption of actin fibers activates several downstream signals, leading to a cell shape change. On the other hand, ECs under continuous stretch have been shown to align parallel to the direction of stretch [17]. This is consistent with our results in which ECs align to the direction of local continuous stretch, strongly suggesting a critical role of relaxation phase in the EC alignment in response to the two different types of stretch. It is interesting to note that flow-imposed ECs are well known to show alignment to the direction of flow [1, 2, 4]. Although the types of mechanical loadings are different, this tendency is similar to the result of this study in terms of the directional alignment. Several numerical studies have well explained these phenomena of ECs in from the viewpoint of structural optimization. For example, Ohashi et al. [18] reported that ECs exposed to fluid shear stress change their morphology to reduce intracellular stress concentrations. Relating to biomolecular events, Tzima et al. [19] reported that, in ECs exposed to fluid shear stress, Rac1, which is small GTPase controls lamellipodia formation, is locally activated in the downstream region. Likewise, it is worth doing these numerical and extra experimental approaches to elucidate underlying mechanism in the present EC remodeling process. Further investigations may, therefore, include more detailed quantification of intracellular strain field and observations of local activity of Rho small GTPases including Rac1 responsible for reorganization for actin-cytoskeletons.

Time course of change in EC morphology was investigated to characterize EC remodeling process. The remodeling process depended on initial shapes of ECs and thus divided into the two groups: Group A and B. For Group A, ECs initially elongated perpendicular to the direction of stretch exhibited spontaneous contraction prior to elongation/alignment to the direction of stretch. In contrast, for Group B, ECs initially aligned to the direction of stretch immediately exhibited elongation parallel to the direction of stretch. It has been reported that ECs exposed to shear stress show the following time course of change: 1) contraction, 2) reorientation and 3) elongation [7]. Taken these results and our results into consideration, there may exist specific time course of change depending on the type of externally applied mechanical stimuli. It is beyond our findings to see if the present remodeling process might be involved in the EC remodeling exposed to global mechanical stimuli including fluid shear stress. Figure 7 summarizes representative time course of EC remodeling process with GFP-actin together with schematic drawings representing cell morphology and actin organization. This result suggests that cell contraction or low angle of cell orientation should be prerequisite for EC elongation parallel to the direction of stretch. Noria et al. [20] suggested that flow-induced morphological changes of ECs were caused by reorganization of SFs. Moreover, Li et al. [21] reported that ECs transfected with dominant negative RhoA, responsible for SF formation, do not respond to fluid shear stress. Taken together, SF reorganization associated with cell morphological changes may play an important role in EC remodeling process.

Fig. 7

Representative time course of EC remodeling process with fluorescent images and schematic drawings. Although time courses of change for the two groups are different, the observed ECs (C₁) finally elongated and orientated parallel to the direction of stretch

Well-known mechanotransducers in EC remodeling to mechanical stimuli are SACs, integrins and PECAM-1. SACs are known to be activated by tension. Naruse et al. [22] reported that EC remodeling under exposure to cyclic stretch is inhibited by the blockade of SACs with GdCl₃ treatment. In this study, GdCl₃ treatment did not significantly affect either morphological changes or localization of SHP-2 at intercellular junctions, indicating that mechanotransduction of SACs does not have a pivotal role in the present EC remodeling. Osawa et al. [13] showed that when mechanotransduction by PECAM-1 occurred, SHP-2 molecules bound to phosphorylated PECAM-1 and recruited to cell-cell borders. Therefore, at least, mechanotransduction by PECAM-1 might be activated by application of local stretch in this study. A recent study [14] suggested that mechanotransduction by PECAM-1 may trigger flow-induced morphological changes of ECs. It is speculated that spatial gradient of magnitude of mechanical stimuli in cells might induce localized mechanotransduction by PECAM-1, possibly leading to the directional EC remodeling. To study this mechanism, further investigations in terms of the quantitative relationship between intracellular stress transfer and SHP-2/PECAM-1 expression should be included.

Conclusions

In this study, local stretch was applied to an EC to stimulate a neighboring EC via intercellular junctions using micromanipulation technique. ECs reorientated and elongated parallel to the direction of stretch. This remodeling process varied from cell to cell, particularly depending on initial cell shape. In addition, SHP-2 was found to be recruited to the stimulated intercellular junctions. These results strongly suggest that intercellular junctions can transmit mechanical forces between cells and transduce the forces into biochemical signals.

Appendix. Calculation of SF parameters with image processing

The protocol for image processing to determine SF parameters is shown in Fig. 8. Algorithm based on pixel intensity of fluorescence images of GFP-actin was adapted from previous reports [23, 24]. SFs were extracted from original image [Fig. 8(a)] with convolution filter application, as shown in Fig. 8(b). For giving pixels p _{i ,j} in 8-bit gray-scale, the pixels surrounding p _{i ,j} were selected as a sample matrix [equation (1)]

$${\text{Sample matrix (}}S_{i{\text{,}}j} {\text{) = }}\left( {\begin{array}{*{20}c}{p_{i - 2,j - 2} } & {p_{i - 2,j - 1} } & {p_{i - 2,j} } & {p_{i - 2,j + 1} } & {p_{i - 2,j + 2} } \\{p_{i - 1,j - 2} } & {p_{i - 1,j - 1} } & {p_{i - 1,j} } & {p_{i - 1,j + 1} } & {p_{i - 1,j + 2} } \\{p_{i,j - 2} } & {p_{i,j - 1} } & {p_{i,j} } & {p_{i,j + 1} } & {p_{i,j + 2} } \\{p_{i + 1,j - 2} } & {p_{i + 1,j - 1} } & {p_{i + 1,j} } & {p_{i + 1,j + 1} } & {p_{i + 1,j + 2} } \\{p_{i + 2,j - 2} } & {p_{i + 2,j - 1} } & {p_{i + 2,j} } & {p_{i + 2,j + 1} } & {p_{i + 2,j + 2} } \\\end{array} } \right)$$

(1)

Fig. 8

Protocol for determination of SF parameters based on image processing. (a) Manually extracted single cell image, (b) automatically detected SFs with convolution filters, (c) calculated minimal gradient in pixel intensity with Sobel filter, (d) angle of SF orientation in each pixel

Convolution using horizontal and vertical kernels [equations (2) and (3)], which define weighted sum of neighboring pixels, provided magnitude of brightness gradient of SFs in each pixel [equation (4)].

$${\text{Vertical kernel (}}K_{\text{v}} {\text{) = }}\left( {\begin{array}{*{20}c}{ - {\text{1}}}&{\text{0}}&{\text{2}}&{\text{0}}&{ - {\text{1}}} \\{ - {\text{2}}}&{\text{0}}&{\text{4}}&{\text{0}}&{ - {\text{2}}} \\{ - {\text{4}}}&{\text{0}}&{\text{8}}&{\text{0}}&{ - {\text{4}}} \\{ - {\text{2}}}&{\text{0}}&{\text{4}}&{\text{0}}&{ - {\text{2}}} \\{ - {\text{1}}}&{\text{0}}&{\text{2}}&{\text{0}}&{ - {\text{1}}} \\\end{array} } \right)$$

(2)

$${\text{Horizonal kernel (}}K_{\text{h}} {\text{) = }}K_v^T $$

(3)

$$G = S_{i,j} * K_{\text{h}} + S_{i,j} * K_{\text{v}} $$

(4)

The pixel whose G value was bigger than 5-fold of averaged pixel intensity in whole area of the target cell was assigned as a constituent of SFs. In order to calculate the angle of SF orientation, the angle of minimal gradient in pixel intensity in each pixel [Fig. 8(c)] was calculated using Sobel kernels as previously reported [20]. The angle data allocated to only pixels which have been assigned as a constituent of SFs [Fig. 8(d)]. Angle values were summed up across the cell with vectorial summation and were assumed that each pixel has a vector consisted of the calculated angle and unit length. The angle between the direction of the resultant vector and the tensile direction was defined as the angle of SF orientation. The length of the resultant vector divided by the number of pixels was defined as the uniformity index of SFs, which indicate the degree of alignment of SFs. All calculation processes were executed on Excel 2004 for Mac (Microsoft, USA). Analyses were performed on images obtained at every 5 min after application of local stretch.