A Membrane-Bound Biosensor Visualizes Shear Stress-Induced Inhomogeneous Alteration of Cell Membrane Tension.

Cell membrane is the first medium from where a cell senses and responds to external stress stimuli. Exploring the tension changes in cell membrane will help us to understand intracellular force transmission. Here, a biosensor (named MSS) based on fluorescence resonance energy transfer is developed to visualize cell membrane tension. Validity of the biosensor is first verified for the detection of cell membrane tension. Results show a shear stress-induced heterogeneous distribution of membrane tension with the biosensor, which is strengthened by the disruption of microfilaments or enhancement of membrane fluidity, but weakened by the reduction of membrane fluidity or disruption of microtubules. These findings suggest that the MSS biosensor is a beneficial tool to visualize the changes and distribution of cell membrane tension. Besides, cell membrane tension does not display obvious polar distribution, indicating that cellular polarity changes do not first occur on the cell membrane during mechanical transmission.

Introduction

It is well observed that external mechanical stimulus can be transmitted or transduced into intracellular biological signal to regulate cell alignment (Goldfinger et al., 2008), deformation (Pfafferott et al., 1985), differentiation (Altman et al., 2002), and some other cellular functions (Liu et al., 2010). However, the information about detailed signal pathway is limited. At present, it is extensively believed that mechanical stress signal acts on the cell membrane, which activates membrane proteins such as sensitive ion channels, G-protein-coupled receptors, and focal adhesions kinases (Boycott et al., 2013, Chachisvilis et al., 2006, Liu et al., 2014). Consequently, they further activate intracellular signaling chain and regulate gene expression as well as protein synthesis. This could explain the major phenomena of mechanotransduction. However, it has been reported that a rapid (<300 ms) activation of Rac was induced by the stress at cell periphery but there was no Rac activation within 30 s when it was stimulated by platelet-derived growth factor (Poh et al., 2009). Besides, cells display a lot of polar changes upon shear stress stimuli including Cdc42-dependent polarization of microtubule system (microtubule-organizing center) (Tzima et al., 2003, McCue et al., 2006) and different activation levels of Src or focal adhesion kinase at upstream and downstream sites (Liu et al., 2014). These cellular polarized changes may result from the asymmetric distribution of stress during transmission via cell membrane, cytoskeleton, or other signal proteins.

As a barrier of the whole cell, cell membrane is the first medium of mechanotransduction to sense extracellular stress. The induced alteration of membrane tension is likely to be an association between cellular polarized changes and stress distribution. In some neural cells, membrane tension also takes part in regulating vesicle trafficking and driving fusion pore expansion (Apodaca, 2002, Kozlov and Chernomordik, 2015). Studies have reported that high membrane tension activates exocytosis, whereas low membrane tension activates endocytosis (Gauthier et al., 2009). In addition, tension of cell membrane has been shown to regulate many cell behaviors, including cell motility (Gauthier et al., 2011), polarization (Houk et al., 2012), spreading (Raucher and Sheetz, 2000), and membrane repair (Togo et al., 2000). However, how the membrane tension changes in response to external stress remains unclear due to the lack of appropriate tension sensors. Membrane tension is composed of in-plane tension in the lipid bilayer and tension from membrane-cytoskeleton adhesion (Keren, 2011). In recent years, some methods have been developed gradually to measure membrane-associated tension. One way is by using micropipette aspiration, which sucks the cellular surface into a micropipette to form a hemispheric protrusion and thus calculates the tension by taking advantage of Laplace's law (Hochmuth, 2000). This technique supplies information about the overall tension of the cell, so it is not appropriate for measuring membrane tension only (Tinevez et al., 2009). Tether pulling is another way to measure membrane tension by pulling membrane tethers from the plasma membrane with the help of optical or magnetic tweezers (Diz-Munoz et al., 2013). The tether force measured by this approach is contributed by in-plane tension, a component of membrane tension, and cytoskeleton attachment to some extent (Lafaurie-Janvore et al., 2013, Dai and Sheetz, 1999). Recently, a new technology based on fluorescence resonance energy transfer (FRET) has been developed to evaluate tension across proteins in focal adhesions that gives a complex distribution of tension within individual focal adhesion (Grashoff et al., 2010, Morimatsu et al., 2013). Such a method could provide the spatial and temporal dynamics of tension visually in living cells in an invasive way.

In the current project, a membrane-bound FRET-based tension sensor (named MSS) is constructed to visualize the dynamics of membrane tension in HeLa cells under applied laminar shear stress of 0.5 (low shear stress), 2 (intermediate shear stress), and 4 Pa (high shear stress) (Mitchell and King, 2013). Cell membrane tension is found to be positively related to cell membrane fluidity but not to the magnitude of shear stress. In addition, shear stress-induced membrane tension depends on the cooperation between cytoskeleton components, although it is mainly maintained by microtubules.

Results

MSS Biosensor Can Visualize Membrane Tension

MSS is a membrane-bound tension sensor, including a tension sensor module and two anchoring proteins, which are linked with lipid molecules in lipid raft and non-lipid raft regions through Lyn and K-Ras kinases, respectively. The tension sensor module comprises an elastic spider silk protein inserted between two fluorescence proteins, enhanced cyan fluorescence protein (ECFP) and yellow fluorescent protein for energy transfer (YPet) (Figure 1A). With this sensor, tension changes can be transformed into FRET efficiency changes of these two fluorescence proteins. The control probe, KMSS, is a head-less mutant whose amino terminal disassociates in the cytoplasm, whereas only the carboxyl terminal anchors on the membrane (Figure 1B). Therefore, no tension could be applied to the KMSS mutant. The FRET ratio of HeLa cells expressing MSS shows no changes over time under static condition (Video S1), demonstrating the stability of the sensor in a living cell. To further check these tension sensors' effects, the osmotic pressure of culture medium is changed artificially. HeLa cells expressing KMSS are exposed to hypertonic solution (0.05 g/mL, sucrose) or hypotonic solution (adding 3 times volume of H2O into culture medium) separately, but the FRET efficiency remains almost unchanged from the baseline (p > 0.05; Figures 1C and 1D), indicating that KMSS is inert to the changes of membrane tension. In contrast, the FRET ratio of HeLa cells expressing MSS achieves a clear higher level in hypertonic solution, whereas it decreases significantly from the baseline in hypotonic solution after 15 min (p < 0.05; Figure 1E), demonstrating that MSS is significantly affected and that it is sensitive to the changes of membrane tension induced by osmotic pressure.

Figure 1

MSS Plasmid Could Visualize the Changes of Membrane Tension

(A) Tension sensor module contains a linker sequence (GPGGA)8 inserted into two fluorescent proteins. When force extends the elastic linker, FRET efficiency decreases, otherwise it increases.

(B) MSS sensor consists of tension sensor module and membrane-bound sequences Lyn and K-Ras. KMSS is a Head-less control sensor of MSS.

(C and D) (C) The representative YPet/ECFP emission ratio images and (D) their average time courses of FRET biosensors in HeLa cells after exposure to hypertonic (MSS, n = 10; KMSS, n = 11) and hypotonic solutions (MSS, n = 7; KMSS, n = 12), respectively.

(E) Average normalized YPet/ECFP emission ratio of KMSS at 0 min (baseline), MSS at 15 min in hypertonic solution (sucrose) and MSS at 15 min in hypotonic solution (H2O). *p < 0.05 significantly different from baseline. All error bars represent SEM.

Scale bar, 20 μm. See also Figure S1.

Furthermore, when the cells expressing MSS are exposed to hypertonic solution (0.025 g/mL, sucrose) after a hypotonic stimulus (adding 2 times volume of H2O into culture medium) for 10 min, the FRET ratio reverses immediately, even though it decreases slowly in the beginning (Figure S1A; Video S2). On the contrary, the FRET ratio increases from the start but later decreases slowly when the cells expressing MSS are exposed to hypertonic solution (0.05 g/mL, sucrose) for 10 min first and then to hypotonic stimulus (adding the same volume of H2O into the culture medium) (Figure S1B). These results suggest that the response of this biosensor to the changes in membrane tension is reversible, whereby it will react quickly if the membrane tension changes in the opposite direction.

In addition, the FRET ratio of MSS increases linearly when cells are exposed to 0.025, 0.5, and 0.075 g/mL sucrose solution (Figure S1C). On the contrary, the ratio decreases linearly after reducing the osmotic pressure of culture medium by adding 1, 2, 3, or 4 times volume of H2O separately (Figure S1D). Results indicate a linear relationship between the response of MSS biosensor and the changes in membrane tension induced by osmotic pressure in current experiments. Therefore, MSS could be applied to visualize the changes and distribution of membrane tension with good sensitivity, reversibility, and linearity.

Shear Stress Changes Membrane Tension

To check the effects of different shear stresses on cell membrane tension, HeLa cells expressing MSS are exposed to shear stress of 0.5 (MSS-0.5 group), 2 (MSS-2 group), and 4 Pa (MSS-4 group) for 15 min. The group of HeLa cells transfected with KMSS, with shear stress of 2 Pa, serves as control. Results show that the FRET ratio of MSS decreases obviously in response to shear tress, whereas that of the control group remains unchanged basically (Figures 2A and 2B). In addition, when the cells are exposed to shear stress, the FRET ratio of MSS decreases sharply within 30 s and then decreases slowly with time whatever be the magnitude of shear stress. Strangely, these are coincident changes of FRET ratio under shear stress of 0.5 and 4 Pa, both of which are lower than the change of FRET ratio under stress of 2 Pa (Figures 2B and 2C), suggesting that membrane tension is not positively related to the magnitude of shear stress.

Figure 2

The Overall Changes of Membrane Tension under Different Shear Stresses

(A and B) (A) The representative YPet/ECFP emission ratio images and (B) their average time courses of MSS in HeLa cells after exposure to the shear stress of 0.5 (n = 13), 2 (n = 14), and 4 Pa (n = 11) and KMSS in HeLa cells (served as control) after exposure to the shear stress of 2 Pa (n = 9).

(C) Average normalized YPet/ECFP emission ratio of FRET biosensors at 15 min under different shear stresses. *p < 0.05 significantly different from control. All error bars represent SEM.

Scale bar, 10 μm.

Further analyses show that cells display a heterogeneous distribution of FRET ratio with the lowest in the middle part, whereas the highest in upstream and downstream parts initially. This distribution manner remains unchanged over time and is not affected by low or high shear stress (Figures 3A and 3B; Videos S3, S4, and S5). By fitting the FRET ratio in different regions of a cell at 0 and 15 min with Extreme equation (Tables S1 and S2), it is found that after its exposure to shear stress, the minimal FRET ratio at 15 min (named min value) decreases, and that 2 Pa induces a significantly larger min value than the other two groups (p < 0.05; Figure 3C). However, the position of min value shows a tendency of moving toward downstream as the shear stress ranges from 0.5 to 4 Pa (Figure S2). Furthermore, upstream variation range (UVR) and downstream variation range (DVR) become smaller under shear stress application, especially for 0.5 and 4 Pa, although there is no significant difference between UVR and DVR in spite of the magnitude of shear stress (Figure 3D). These results suggest that shear stress can enlarge membrane tension but plays only a gentle role in its uneven distribution.

Figure 3

Shear Stress Changes Membrane Tension

(A and B) (A) YPet/ECFP emission ratio of different cellular regions in timescale and (B) the fitting curve (points for sample data and solid lines for fitting) at 0 and 15 min under shear stress of 0.5, 2, or 4 Pa.

(C and D) (C) Comparisons of normalized minimum YPet/ECFP emission ratio and (D) normalized upstream variation range and downstream variation range between different shear stress groups at 15 min *p < 0.05 compared with 0.5 Pa group. #p > 0.25 between groups of UVR and DVR. All error bars represent SEM.

See also Figures S2 and S3 and Tables S1 and S2.

Cell Membrane Fluidity Regulates Membrane Tension

Cell membrane fluidity is reported to display polarity upon shear stress application (Butler et al., 2002). To check whether the cell membrane fluidity participates in the regulation of membrane tension induced by shear stress, HeLa cells expressing MSS are pretreated with 0.1 mmol/L of cholesterol (Cho) for 3 hr or 45 mmol/L benzyl alcohol (BA) for 15 min to reduce or enhance the fluidity of cell membrane, respectively (Butler et al., 2002). Cells transfected with MSS plasmid under shear stress of 2 Pa, namely, the MSS-2 group, serves as a control group. Upon the stimulus of 2 Pa, BA treatment decreases the FRET ratio sharply, whereas Cho decreases it slowly, although both treatments show a decreasing trend in the FRET ratio (Figures 4A and 4B; Videos S6 and S7). At 15 min, BA treatment leads to a significant lower FRET ratio but Cho leads to a significant higher FRET ratio compared with the control group (p < 0.05; Figure 4C). These results demonstrate that cell membrane fluidity affects the alteration of membrane tension induced by shear stress, and higher membrane tension is associated with higher cell membrane fluidity.

Figure 4

The Overall Changes of Membrane Tension under Different Cell Membrane Fluidities

(A and B) (A) The representative YPet/ECFP emission ratio images and (B) their average time courses of MSS in HeLa cells pretreated for 15 min with 45 mmol/L BA (n = 10) or 3 hr with 0.1 mmol/L of Cho (n = 11) under shear stress of 2 Pa.

(C) Average normalized YPet/ECFP emission ratio of FRET biosensors at 15 min under different cell membrane fluidities. *p < 0.05 significantly different from 2 Pa group. All error bars represent SEM.

Scale bar, 10 μm.

Initially, FRET ratio is low in the whole cell after BA or Cho applications, especially under Cho treatment. After exposure to shear stress, the ratio decreases in all regions of the cell (Figure 5A). However, the smallest ratio (at 15 min) is still located in the middle part of the cell (Figure 5B). From the fitting curve, it can be obtained that BA treatment significantly decreases the min value, whereas Cho plays an opposite role (p < 0.05; Figure 5C; Tables S1 and S2), but neither of the above treatments changes the position of the min value (Figure S2). Besides, either UVR or DVR is drastically decreased by BA treatment but not by Cho. UVR has no significant difference with DVR whether under BA or Cho application (Figure 5D). These results indicate that cell membrane fluidity plays a significant role in adjusting the distribution of membrane tension induced by shear stress. They also suggest that enhanced cell membrane fluidity can help to homogenize cellular membrane tension.

Figure 5

High Cell Membrane Fluidity Enlarges Membrane Tension

(A and B) (A) YPet/ECFP emission ratio of different cellular regions in timescale and (B) the fitting curve (points for sample data and solid lines for fitting) at 0 and 15 min for cells pretreated for 15 min with 45 mmol/L BA or 3 hr with 0.1 mmol/L of Cho under shear stress of 2 Pa.

(C and D) (C) Comparisons of normalized minimum YPet/ECFP emission ratio and (D) normalized upstream variation range and downstream variation range between different membrane fluidities group at 15 min *p < 0.05 compared with 2 Pa group. #p > 0.3 between groups of UVR and DVR. All error bars represent SEM.

See also Figures S2 and S3 and Tables S1 and S2.

Cytoskeleton Regulates Membrane Tension

Studies show that cytoskeleton binds with cell membrane directly and takes part in the process of sensing external stress and mediating mechanical force transduction (Choi and Helmke, 2008, Davies, 2009, Alenghat and Ingber, 2002, Raucher et al., 2000). Therefore, it is tested whether cytoskeleton plays any role in the regulation of membrane tension induced by shear stress. HeLa cells expressing MSS are pretreated with 1 μg/mL of cytochalasin D (CytoD) for 30 min or 5 μg/mL of nocodazole (Noco) for 15 min to destroy actin filaments or microtubules, respectively (Kim et al., 2012, Zaal et al., 2011). It is observed that CytoD decreases FRET ratio significantly, whereas Noco strangely increases the ratio to a higher level compared with the control group (MSS-2 group) (p < 0.05; Figure 6; Videos S8 and S9). When HeLa cells expressing MSS are pretreated with 1 μg/mL of myosin light-chain kinase (MLCK) inhibitor, ML7, for 1 hr to suppress the contraction but keep the structure of microfilament intact (Liu et al., 2011), the FRET ratio decreases obviously but is slightly higher than that obtained from CytoD treatment, although the tendency is similar to that of CytoD (p < 0.05; Figure 6; Video S10). These results indicate that both microfilaments and microtubules are necessary for maintaining cellular membrane tension.

Figure 6

The Overall Changes of Membrane Tension under Different Disruptions of Cytoskeleton

(A and B) (A) The representative YPet/ECFP emission ratio images and (B) their average time courses of MSS in HeLa cells pretreated for 30 min with 1 μg/mL of CytoD (n = 13), 15 min with 5 μg/mL Noco (n = 18), or 1 hr with 1 μg/ml ML-7 (n = 10) under shear stress of 2 Pa.

(C) Average normalized YPet/ECFP emission ratio of MSS at 15 min with different pretreatments. *p < 0.05 significantly different from control. All error bars represent SEM.

Scale bar, 10 μm.

Furthermore, cells pretreated with CytoD or ML7, but not with Noco, show an obviously decreased ratio in all cellular regions upon shear stress application, and all of them have similar distributions of membrane tension (Figure 7A). It can be seen clearly from the fitting curve that shear stress decreases the min value after the cytoskeleton is destroyed, but not obviously under microtubule disruption (Figures 7B and 7C; Tables S1 and S2). Moreover, the position of min value remains almost unchanged under any component disruption of cytoskeleton (Figure S2). Further comparison reveals that UVR has no significant difference compared with the control group after the cytoskeleton disruption, whereas DVR is found different under Noco treatment (p < 0.05; Figure 7D). These results suggest that the maintenance of membrane tension depends on the structure and contractility of cytoskeleton, and that the cytoskeleton is supposed to have little effect on the heterogeneous distribution of membrane tension. In addition, there is still no significant difference between UVR and DVR under any kind of treatment (Figure 7D).

Figure 7

The Membrane Tension Depends on Not Only Microtubules but Also Actin Filaments

(A and B) (A) YPet/ECFP emission ratio of different cellular regions in timescale and (B) the fitting curve (points for sample data and solid lines for fitting) at 0 and 15 min for cells pretreated for 30 min with 1 μg/mL of CytoD, 15 min with 5 μg/mL Noco, or 1 hr with 1 μg/ml ML-7 under shear stress of 2 Pa.

(C and D) (C) Comparisons of normalized minimum YPet/ECFP emission ratio and (D) normalized upstream variation range and downstream variation range at 15 min with different pretreatments. *p < 0.05 compared with 2 Pa group. #p > 0.11 between groups of UVR and DVR. All error bars represent SEM.

See also Figures S2 and S3 and Tables S1 and S2.

Discussion

Many techniques have been used for measuring cell mechanics, including elastic substrates, bendable micropost array, atomic force microscopy, and optical and magnetic tweezers (Xie et al., 2017, Guo et al., 2014). However, these methods have difficulties in detecting intracellular mechanical force. Recently, FRET-based biosensors have been widely used in life sciences field owing to their high temporal and spatial resolution, although only a few studies have applied FRET technology to the research of stress transfer within cells. A tension sensor was reported to be inserted into vinculin protein to study the dynamics of focal adhesion with piconewton sensitivity (Grashoff et al., 2010, Freikamp et al., 2017). This genetically encoded biosensor provides methods to measure intracellular stress, but it is constructed by insertion of the tension sensor into a specific protein, which may affect the function of endogenous protein. Later, a FRET biosensor was developed to describe the tension across membrane receptor, epidermal growth factor receptor (Stabley et al., 2011). Similarly, another group has applied their biosensors to integrin proteins, which has allowed us to visualize the distribution of tensions within individual focal adhesions (Morimatsu et al., 2013). These non-genetically encoded biosensors are developed by linking the tension sensor to the membrane proteins that are usually not part of the endogenous proteins (Liu et al., 2017). They are located outside of the cells so that these biosensors can describe the stress transmission toward intracellular region in a non-invasive manner. In the current project, this membrane-bound biosensor is constructed by extending the two terminals of tension sensor with two different anchor proteins, Lyn and K-Ras, and the tension changes on cell membrane, not a protein, was measured with this biosensor since it was supposed to be the first step of intracellular stress transmission. FRET efficiency decreases under membrane tension because of the larger distances between two fluorophores induced by tension. Results confirm that the MSS biosensor is sensitive to the changes of membrane tension and that this response is reversible as well as linear with the changes in the membrane tension. Furthermore, after in-depth analysis of the relationship between the FRET-proportional value and the actual stress magnitude by theoretical calculation in current research (Figure S3), it is revealed by the MSS biosensor that membrane tension ranges from 0 to 2 pN under shear stress application, which has the same order of magnitude as in previous reports (Lieber et al., 2013). Therefore, this genetically designed tension sensor can be applied to visualize membrane tension and study the detailed changes of tension on cell membrane with piconewton sensitivity, which provides a powerful tool to explore cellular mechanical transmission.

Current results show that shear stress causes an obvious heterogeneous distribution of membrane tension, which is the highest in the midstream but lower at upstream and downstream. Either cell membrane fluidity or cytoskeleton has slight effects on this uneven distribution. However, shear stress can strengthen membrane tension, although it is not directly proportional to the amplitude of shear stress. Strangely, low shear stress induces higher membrane tension than that under intermediate shear stress. It has been reported that cortical actin cytoskeleton forms a meshwork at the cell periphery just beneath the plasma membrane while some actin filaments and microtubules are connected directly with lipid rafts in cell membrane (Allen et al., 2007, Hall, 1998, Viola and Gupta, 2007). In response to low shear stress, a quick reduction of RhoA activity is observed corresponding to a decrease in the formation of actin stress fibers in C28/I2 chondrocytes (Wan et al., 2013). Therefore, membrane tension is increased under low shear stress as less stress fibers reduce restriction on the mobility of lipid molecules, while lipid rafts are still confined by microtubules. Consistently, cell membrane also exhibits higher tension compared with that under intermediate shear stress after destroying the structure or contractility of microfilaments artificially. Different from the nonlinear association between membrane tension and shear stress, the position of maximum membrane tension moves toward downstream with increasing shear stress. It has been reported with computational finite element modeling techniques that cell deformation induced by fluid shear stress (0.6 Pa) is the largest in the midstream part (McGarry et al., 2005). Higher shear stress may push the position of maximum membrane tension to the downstream since it has larger tangential effects on the middle part of cell membrane.

It has been reported that the overall cell membrane fluidity increases upon higher laminar shear stress application (Butler et al., 2001). Present results indicate that high shear stress can increase membrane tension significantly and quickly, which may be due to the increased cell membrane fluidity since enhancement of cell membrane fluidity can accelerate this process. This swift change of membrane tension is consistent with an immediate (5 s) significant increase in membrane fluidity under shear stress application (Butler et al., 2001). It seems that higher membrane fluidity promotes the relative movement of phospholipid molecules and thus stretches cell membrane and leads to higher membrane tension. This has been verified in the current experiments by enhancing or reducing the cell membrane fluidity. The results also show that higher membrane fluidity helps to homogenize membrane tension. Consistently, high shear stress has also reduced the variation ranges of membrane tension in both upstream and downstream significantly.

The generation of membrane tension has been reported to have a close association with integrated actin cytoskeleton (Cramer, 1999). Recently, an investigation about ezrin, a major cross-linker of the membrane-cytoskeleton interface, has shown that membrane tension is decreased after reducing ezrin activity (Rouven Bruckner et al., 2015). Membrane mechanics could also be altered by the presence of actin shell, which seems to lead to a larger friction coefficient and restrict lipid mobility (Guevorkian et al., 2015). Current experiments find that disruption of microtubules decreases membrane tension, whereas actin disruption increases the tension. Microtubules may enhance the membrane tension by anchoring lipid rafts in the membrane, but if microtubules are destroyed, there will be little limitations on lipid rafts, which lead to a reduction of membrane tension. According to the above discussion, actin disassembly and the inhibition of its contractility enable more freedom for lipid molecules' mobility but less for lipid raft mobility, which results in increased membrane tension. These findings demonstrate that any of the cytoskeleton components as well as their contractility contributes to membrane tension. However, the local distribution of membrane tension is nearly not affected by cytoskeleton, since UVR and DVR remain almost unchanged after being pretreated with cytoskeletal-disrupting agents. Some interfaces have been reported to be formed between membrane and cytoskeletal proteins, known as plasma membrane skeleton (Sechi and Wehland, 2000, Smith et al., 2017). Some other membrane-associated proteins like supervillin, myosin, talin, and G-protein-coupled receptor also connect cell membrane to actin filaments or microtubules (Allen et al., 2007, Chichili and Rodgers, 2009). These interactions may also contribute to maintaining membrane tension along with its local distribution. Moreover, microtubule disruption seems to eliminate the tension changes in all cellular regions under shear stress application, suggesting that microtubules play a main role in maintaining membrane tension. This is supported by a previous report that cellular morphology usually depends on the interaction of microtubules with plasma membrane (Kuchnir Fygenson et al., 1997).

Shear stress can induce a lot of cellular polarity changes, like polarized Rac activation at cellular downstream (Shao et al., 2017). During this process, external stress signal is transmitted to target molecules step by step. However, which step first reveals polar changes remains unclear. The present results observe that membrane tension does not display obvious polar distribution, suggesting that cell membrane only plays a role in transmitting stress intracellularly. It has been reported that local stress could activate Src in 0.3 s at remote cytoplasmic sites, which depends on the tension across cytoskeleton (Na et al., 2008). Microtubules are reorganized and stress fibers become thicker and longer after shear stress stimuli (Galbraith et al., 1998). Therefore, cytoskeleton, which is considered as the subsequent pathway of stress transmission after plasma membrane, will display polar distribution of tension probably. This is supported by a previous report that actin filaments could enable the establishment of a polarized distribution by breaking symmetry condition, whereas microtubules maintain the stability of the polarized organization (Li and Gundersen, 2008). Whether and how the cytoskeleton determines the polar reaction of cells upon directional shear stress application still needs to be elucidated.

Limitations of the Study

In current system, a classic cell line, HeLa cell, is chosen to observe heterogeneous distribution of membrane tension in response to shear stress. It is reported that the direction of cell migration in tumor cells is opposite to that in endothelial cells, indicating that there may be entirely different mechanotransduction pathways in these two kinds of cell lines. Therefore, some different results probably can be achieved by using vascular endothelial cells instead of HeLa cells.

Methods

All methods can be found in the accompanying Transparent Methods supplemental file.