Fluid shear stress stimulation induces endothelial cells to elongate and align in the direction of applied flow. Using the complementary techniques of photoactivation of fluorescence and fluorescence recovery after photobleaching, we have characterized endothelial actin cytoskeleton dynamics during the alignment process in response to steady laminar fluid flow and have correlated these results to motility. Alignment requires 24 h of exposure to fluid flow, but the cells respond within minutes to flow and diminish their movement by 50%. Although movement slows, the actin filament turnover rate increases threefold and the percentage of total actin in the polymerized state decreases by 34%, accelerating actin filament remodeling in individual cells within a confluent endothelial monolayer subjected to flow to levels used by dispersed nonconfluent cells under static conditions for rapid movement. Temporally, the rapid decrease in filamentous actin shortly after flow stimulation is preceded by an increase in actin filament turnover, revealing that the earliest phase of the actin cytoskeletal response to shear stress is net cytoskeletal depolymerization. However, unlike static cells, in which cell motility correlates positively with the rate of filament turnover and negatively with the amount polymerized actin, the decoupling of enhanced motility from enhanced actin dynamics after shear stress stimulation supports the notion that actin remodeling under these conditions favors cytoskeletal remodeling for shape change over locomotion. Hours later, motility returned to pre-shear stress levels but actin remodeling remained highly dynamic in many cells after alignment, suggesting continual cell shape optimization. We conclude that shear stress initiates a cytoplasmic actin-remodeling response that is used for endothelial cell shape change instead of bulk cell translocation.
- cytoskeletal dynamics
- endothelial cells
atherosclerotic lesions develop at predictable locations in the arterial vascular network, forming primarily where vessels branch and bend while largely sparing nearby straight vessel segments (8). Vessel bending and branching disturbs normal blood flow patterns, generating complex local regions of flow separation and recirculation that alter the magnitude, direction, and frequency of forces applied to the vessel wall (23, 41). Endothelial cells, which line the vessel wall and regulate vascular homeostasis, are exquisitely sensitive to the forces produced by flowing blood (10, 11, 14). The blood flow profiles present at atherogenic regions cause endothelial injury and chronic dysfunction, leading to the accumulation of lipids in the vessel wall that eventually form atherosclerotic plaques (18, 25).
One marker of endothelial dysfunction in vivo is a disorganization of cell shapes. Endothelial cells populating atheroprotective regions are elongated into torpedo shapes that point into the direction of flow (9), while those located at atherogenic regions are polygonal (19). These observations are reproduced in vitro by exposing primary endothelial cell cultures to fluid shear stress using different flow profiles (10). In combination with computational modeling efforts, such studies have confirmed that it is the arterial regions with the lowest shear stress that correlate with aberrant endothelial shapes and that are at greatest risk of atherosclerotic lesion development (25).
Although the mechanosensitive molecular mechanisms that determine shear stress-mediated endothelial shape change remain controversial, a growing body of evidence supports a decentralized, integrated signaling network in which cytoskeletal polymers transmit apical shear forces to membrane attachment sites at which conformational changes in associated proteins initiate signaling (21). Changes in the number, type, and structure of cytoskeletal connections alter the properties of transmitted forces and may modify the specific endothelial phenotype, depending on the spatial and temporal microstimuli that each endothelial cell senses (12).
Of the three main cytoskeletal polymers that determine endothelial cell shape, actin filaments (F-actin) are the most abundant and locate in close proximity to the cell membrane. Confluent endothelial cells assemble ∼70% of their 100 μmol/l total actin into a rich meshwork of just over 50,000 filaments that, on average, are ∼3 μm long (26). Cross-linking proteins organize F-actin into a viscoelastic gel that interconnects transmembrane proteins and signaling complexes located at adhesion sites. Direct connections of F-actin to β-integrin tails by talin (4) and filamin (36) and to cadherins by vinculin and catenins (40) have been described well in the literature.
During cell locomotion and shape change events, the actin cytoskeleton is remodeled extensively (26, 39), primarily by adding and subtracting subunits at free filament ends. Cytoskeletal remodeling also occurs in response to fluid shear stress in endothelial cells. Microvilli present on the apical endothelial surface under static conditions disappear, leaving a smooth, glassy contour (32). Basally located dense peripheral bundles of F-actin (stress fibers) dissolve shortly after shear stress exposure only to reform hours later just under the apical membrane and aligned with the long axis of the cell (16, 17, 32). These structural alterations reduce the peak shear stresses imposed on individual cells (1, 33) and cause elongated endothelial cells to become more resistant to micropipette surface deformations (34).
Herein we report our measurement of the temporal remodeling of actin in living endothelial cells stimulated with steady, laminar fluid flow using the complementary techniques of photoactivation of fluorescence (PAF) and fluorescence recovery after photobleaching (FRAP), which allow simultaneous measurement of the amount of polymerized actin and the rate of F-actin subunit turnover in individual endothelial cells. We show for the first time that fluid shear stress rapidly enhances endothelial actin remodeling to levels measured in the most dynamic endothelial cells under static conditions. Increased F-actin turnover precedes a drop in the amount of polymerized actin, revealing that net cytoskeletal depolymerization characterizes the earliest phase of the endothelial shear stress response. In contrast to endothelial cells in static culture that enhance actin remodeling in proportion to their rate of migration across a substrate (26), we have found that the shear stress-induced stimulation of actin remodeling is initially decoupled from increased endothelial motility. Although migration rates eventually recovered to pre-shear stress levels, actin cytoskeletal remodeling remained active in many endothelial cells after shear stress accommodation.
MATERIALS AND METHODS
Cell culture, motility, microinjection, and transfection.
Bovine aortic endothelial cells (BAECs; VEC Technologies) were cultured in DMEM containing 10% FCS and 0.1 U/l penicillin and streptomycin. For experiments, BAECs subcultured at passages 2-10 were plated onto 0.1% gelatin-coated glass coverslips. Cell motility was measured using quantitative time-lapse video microscopy (26).
Caged-resorufin iodoacetamide-labeled rabbit muscle actin (CR-actin) (39) was diluted to 20 μmol/l in injection buffer (in mmol/l: 1 HEPES, 0.2 MgCl2, and 0.5 ATP, pH 7.4) containing 5 μmol/l Alexa Fluor 488 dextran (10 kDa; Molecular Probes), microinjected into BAECs, and allowed to recover for 1 h.
cDNA encoding a human β-actin enhanced green fluorescent protein fusion protein (EGFP-actin; Clontech) was transfected into 70% confluent BAECs using a ratio of 2 μg/ml cDNA to 20 μl/ml LipofectAMINE (Invitrogen) diluted in serum-free OptiMEM I and incubated for 5 h at 37°C and 5% CO2, after which the transfection solution was replaced with normal growth medium. EGFP-actin was detectable by 48 h, and transfected cells were used at between 48 and 72 h. The transfection efficiency was 13 ± 4% (n = 8).
BAECs were fixed with 4% paraformaldehyde for 20 min at room temperature, permeabilized with 0.2% Triton X-100 for 5 min, incubated with 1 μmol/l Alexa Fluor 546 phalloidin (Molecular Probes) for 10 min to stain F-actin, washed, and mounted onto microscope slides. Images were digitized with a cooled charge-coupled device camera (OrcaII-ER; Hamamatsu) on a Nikon Eclipse TE2000 microscope using a ×63 magnification lens objective and MetaMorph software (Universal Imaging).
SDS-PAGE and immunoblotting.
Petri dishes (60 mm) with confluent monolayers were extracted from (in mM) 60 PIPES-25 HEPES-10 EGTA-2 MgCl2 buffer (26) with 1% Triton X-100 at 4°C and collected using a rubber scraper. Samples were solubilized with SDS-PAGE sample buffer in total for whole cell lysates or after separation into Triton X-100-soluble and -insoluble fractions using centrifugation for 30 min at 215,000 g at 4°C, matched for total protein, and separated using SDS-PAGE. Proteins were electrophoretically transferred to PVDF membranes (Millipore) for 60 min, blocked with 0.02% Tween-20 and 5% Carnation nonfat dry milk (pH 7.4), incubated with a 1:1,000 dilution of mouse anti-actin MAb or anti-GFP PAb for 1 h at room temperature, washed, and probed with a 1:5,000 dilution of horseradish peroxidase-labeled goat anti-mouse antibody (BioRad) for 1 h. Washed membranes were developed using SuperSignal West Pico chemiluminescent substrate (Pierce), digitized, and densitometrically analyzed. For GFP quantitation, purified recombinant GFP protein (Clontech) standards were used.
Steady laminar shear stress was applied at 12 dyn/cm2 to the surface of confluent BAECs using a parallel plate flow chamber system (35). The chamber consisted of two stainless steel plates maintained at 37°C using a copper heating block that were separated by a 0.5-mm silicone sheet (Allied Biomedical) with a 5 × 50-mm rectangular section removed to create a flow channel. Confluent BAECs grown overnight on a 25-mm-diameter coverslip coated with 0.1% gelatin were placed into a recess milled into the surface of the bottom stainless steel plate, and the flow chamber was assembled. A Masterflex peristaltic pump (Cole-Parmer) was used to propel DMEM containing 10% FCS at 37°C and 5% CO2 from a reservoir through the flow chamber. A second sealed reservoir was placed between the pump and the flow chamber to eliminate pulsations. Because the flow through the channel can be approximated as two-dimensional fully developed laminar flow with a parabolic velocity profile, the wall shear stress was determined as a linear function of the volume flow rate pumped through the chamber.
PAF and FRAP.
PAF and FRAP experiments were performed on individual BAECs forming confluent monolayers contained within a parallel plate flow chamber using a Zeiss Axiovert 405M microscope with a ×100 magnification oil-immersion lens objective. For PAF, BAECs microinjected with CR-actin were located by a coinjected Alexa Fluor 488 dextran volumetric tracer, and the CR-actin was photoactivated in a 7- to 10-μm-wide strip spanning the cell width using the 365-nm line of a mercury arc lamp (27). For FRAP, a 1-to 2-μm-wide band was photobleached across one dimension of cells expressing EGFP-actin using a 100-mW argon ion laser (Melles Griot) at 488 nm. The photoactivation or photobleaching time was <1 s. Control studies were performed under static conditions, after which fluid flow was initiated and experiments were performed during shear stress stimulation for up to 24 h. In certain static controls, BAECs were incubated with 1 μmol/l jasplakinolide for 30 min at 37°C before photobleaching to induce actin polymerization.
The evolution of fluorescence at the center of the photoactivated or photobleached region was measured over time and analyzed using a continuum model of steady state actin cytoskeleton dynamics to calculate the F-actin turnover time and the fraction of total actin polymerized (27, 38). This model is based on prior work in endothelial cells that showed that the fluorescence evolution is biphasic and can be characterized by an initial rapid change that occurs within seconds due to G-actin diffusion, followed by a slower decay that occurs over minutes proportional to the rate of F-actin turnover as fluorescent subunits replace nonfluorescent ones in bleached filaments (26, 27, 38). Filament diffusion is discounted from the slow component because F-actin is predominately cross linked (20) and endothelial actin filaments are long (26); therefore, they diffuse on a time scale of hours, far slower than filament turnover. The polymerized fraction was calculated from the distribution between G- and F-actin fluorescence populations validated biochemically in BAECs using SDS-PAGE and densitometry (26).
Fluorescent actin analogs function similarly to native actin in endothelial cells.
Fluorescently tagged actin monomers introduced into individual BAECs to form confluent monolayers were used as tracers of actin cytoskeletal dynamics in PAF and FRAP experiments. For PAF, BAECs were microinjected with CR-actin at a final concentration estimated to account for <2% of total actin and then photoactivated. CR-actin has been shown previously to incorporate into the cytoskeleton of various cell types and to function as a reliable marker during actin remodeling (26, 39). FRAP studies were performed by photobleaching BAECs transiently transfected with EGFP-actin. EGFP-actin colocalizes with native F-actin by staining with Alexa Fluor 546 phalloidin in fixed and permeabilized BAECs (Fig. 1A), where it both concentrates at the cell cortex and incorporates into F-actin bundles spanning the cell. In living BAECs, EGFP-actin is diffusely present throughout the cell cytoplasm and becomes particularly enriched in regions of active membrane ruffling and lamellipodial extension at the cell edge during cell crawling. Please refer to the Supplementary Material1 for this article (published online at the American Journal of Physiology-Cell Physiology web site) to view movies 1 and 2. EGFP-actin localization within these actin-based structures required for cell crawling behavior supports the hypothesis that EGFP-actin expressed in BAECs functions similarly to endogenous actin.
Detergent permeabilization of confluent BAECs with Triton X-100 followed by high-speed centrifugation (>100,000 g) separated actin between soluble and insoluble fractions that represented bulk measurements of monomer (G-actin) and filament populations, respectively. Immunoblots of these fractions with an anti-GFP antibody revealed that EGFP-actin was detectable only in transfected cells and, because of its low level of expression, was only weakly identified compared with endogenous actin using an anti-actin antibody (Fig. 1B). Densitometric quantitation with calibrated protein standards of EGFP-actin and unlabeled actin revealed that EGFP-actin partitioned into the G- and F-actin fractions at the same ratio as endogenous actin; 51 ± 2% of EGFP-actin is Triton insoluble (n = 4) compared with 58 ± 6% of Triton-insoluble total actin (24). With the use of the fraction of total cells positive for EGFP-actin expression (13 ± 4%, n = 8), the EGFP-actin mass per cell (0.64 ± 0.05 pg, n = 3) was determined and compared with the previously published 6.6 ± 1.4 pg of endogenous actin per BAEC (24) to calculate an average increase in the total actin content of 10 ± 1% per expressing cell. This level of overexpression is not likely to have a significant effect on actin organization or dynamics, given the excess of actin monomer sequestering proteins present in nonmuscle cells. Taken together, the present evidence shows that the EGFP-actin fusion protein and CR-actin probes function similarly to endogenous actin in endothelial cells.
Shear stress alters the dynamics of actin remodeling in endothelial cells.
We measured the temporal shear stress response of actin remodeling in BAECs with PAF and FRAP. Photobleached or photoactivated regions marked within BAECs changed more rapidly after fluid shear stress mechanostimulation than did those not exposed to fluid flow (Fig. 2, A and B). Quantitation of the fluorescence recovery in FRAP experiments showed that mechanostimulation increased both the extent and rate of EGFP-actin fluorescence recovery (Fig. 2C). Because EGFP-actin is a tracer of actin cytoskeleton remodeling, enhanced fluorescence photobleaching recovery kinetics implies that remodeling increases in BAECs after a shear stress challenge. The biphasic shape of the FRAP recovery curves indicates that the evolution of EGFP-actin fluorescence can be decomposed into fast and slow dynamic components. Previous work showed that the fast actin dynamics represent primarily G-actin diffusion into the bleached region, whereas the slow dynamics depend on the rate of F-actin turnover as fluorescent subunits replace nonfluorescent ones in bleached filaments (26, 27, 38). A two-compartment model describing this biphasic behavior was used to interpret the fluorescence recovery into the photobleached or photoactivated region in FRAP and PAF experiments, respectively, allowing determination of the fraction of total actin that was polymerized and the average F-actin turnover time (38). PAF fluorescence decay measurements, the inverse of FRAP, provided quantitatively equivalent relationships between shear stress stimulation and enhanced fluorescence dynamics.
Endothelial cells respond rapidly to shear stress by net depolymerization of their cytoskeletons.
We have found that individual BAECs forming confluent monolayers polymerized 68 ± 12% of their total actin and had average filament lifetimes of 35.2 ± 7.1 min (n = 10) before shear stress exposure using PAF and FRAP with CR-actin and EGFP-actin, respectively, confirming our previously published values of 73 ± 11% as polymer and 38.9 ± 11.0-min lifetimes for F-actin turnover (26). Treatment with jasplakinolide, a potent cell membrane-permeant actin-polymerizing agent, increased the percentage of actin polymerized to 95 ± 4% (n = 10) and halted F-actin turnover, indicating that the expressed EGFP-actin was functional.
Individual BAECs in a confluent monolayer responded rapidly to 12 dyn/cm2 fluid shear stress by increasing their rate of F-actin subunit turnover (Fig. 3A), although there was great cell-to-cell variability. Despite the individuality of the specific response, as a population, the average filament turnover time in confluent BAECs decreased to a minimum of 12.2 ± 10.0 min (n = 20) between 30 and 60 min after shear stress application. This level of F-actin turnover approaches the most rapid turnover rates measured in fast-moving endothelial cells (26). Filament turnover times remained low for ≥6 h after the onset of fluid shear mechanostimulation (Fig. 3B). Our measurements were performed with bleaching/photoactivation lines perpendicular to the flow direction. We observed no significant difference in actin dynamics with respect to the cell major axis or after comparing measurements upstream or downstream from the nucleus (data not shown); however, this was not the primary end point of our study.
The initial decrease in the average lifetime of cytoplasmic F-actin after shear stress stimulation was followed by a net loss of cellular F-actin. Despite large individual variability in the enhancement of filament turnover, all BAECs showed net depolymerization of cytoskeletal F-actin in response to a shear stress challenge (Fig. 3C); the percentage of total actin polymerized decreased to an average value of 43 ± 10% (n = 20) per endothelial cell at 30–60 min after shear stress application when F-actin turnover times were at a minimum. The filament turnover time stabilized at this lower value, but the amount of actin polymer continued to fall, reaching a minimum amount of 34 ± 4% after ∼3–3.5 h of shear stress stimulation (Fig. 3D) (n = 14; P < 0.01 vs. 30–60 min). At 5 h, the F-actin content tended to recover slightly to 55 ± 19% (n = 6) but remained below pre-shear stress levels.
Actin remodeling is transiently decoupled from motility during shear stress exposure.
Because endothelial cell motility correlates positively with enhanced filament turnover and decreased polymer content under static conditions (26), we examined the movement of endothelial cells to determine whether this relationship would hold during shear stress-induced actin remodeling. In the absence of fluid shear stress, BAECs in monolayers crawled at the relatively slow rate of 0.18 ± 0.04 μm/min (n = 35). Instead of accelerating motility in response to accelerated actin dynamics, BAECs exposed to 1–2 h of shear stress decreased their rate of translocational movement to a minimum speed of 0.09 ± 0.04 μm/min (n = 35), which remained decreased for an additional 3 h (Fig. 4). This observation is in stark contrast to actin remodeling measurements in sparse endothelial cells under no-flow conditions, in which nonconfluent endothelial cells move about threefold faster with depleted F-actin content and accelerated F-actin turnover rates (26). Experiments with human umbilical vein and mouse cardiac endothelial cells handled under identical conditions showed a similar shear stress-dependent decrease in cell speed on the same time scale (data not shown), implying that these observations are not species or cell line specific.
Actin remodeling remains enhanced in many endothelial cells after shear stress-induced endothelial shape change.
In the presence of continual shear stress stimulation, BAECs elongated and aligned in the direction of fluid flow in an accommodation response to the applied force. In individual accommodated BAECs, we observed that the shear stress-induced actin remodeling response was highly variable. Although the polymerized fraction recovered to near pre-shear stress levels in most cells, some BAECs continued to have enhanced rates of F-actin subunit turnover, whereas others were more dynamically stable (Fig. 5, A and B). Therefore, unlike the relationship observed in unstressed cells, there was less correlation between polymer content and filament turnover in shear stress-accommodated BAECs.
Despite the nature of the shear stress response in individual aligned cells, as a population, shear stress-accommodated BAECs maintained relatively enhanced levels of actin remodeling. The polymerized fraction fell slightly short of returning to pre-shear stress levels, averaging 57 ± 13% between 22–24 h (Fig. 5C), or ∼10% less F-actin than was present under static conditions (n = 23; P ≤ 0.05 vs. static control). The filament turnover time during the same period recovered to a lesser extent at 26.0 ± 19.7 min (Fig. 5D), with the large variability representing differences in the activation of individual cells.
The factors contributing to the migrational properties of individual endothelial cells are complicated, depending on the origin of the cells, the type of substrate, and importantly, the density of cells on the surface. Endothelial cells are most motile in sparse culture in which they establish few contacts with their neighbors. As cells incorporate into a confluent monolayer, their movement diminishes and their cytoplasmic F-actin turnover rate slows (26). Application of shear stress rapidly increases cytoplasmic actin turnover but slows the rate of endothelial migration in confluent monolayers.
Subconfluent endothelial cells have an upper limit on their rate of movement (∼0.6 μm/min) and F-actin turnover lifetimes (∼5 min), in which movement and dynamics are coupled (26). After fluid shear stress stimulation of endothelial monolayers, filament turnover increased but motility slowed and remained depressed for hours before recovering. Hence, actin remodeling decouples from motility in shear stress-stimulated endothelium. Increased substrate adhesion (Trap mechanism) is one possible means by which to achieve this decoupling because adhesion is an important determinant of cell speed (29). Alternatively, the connections between F-actin and transmembrane adhesion molecules (Clutch mechanism) may change, such as that shown for increased filamin binding to β-integrin tails, which negatively regulates cell movement (5).
The actin remodeling response of endothelial cells is different, depending on the length of time subjected to fluid flow. In our measurements, actin cytoskeleton remodeling became enhanced 5 min after shear stress stimulation and reached maximal levels with minimal F-actin content at 30–60 min after shear stress stimulation, confirming the actin depolymerization response observed by Morita et al. (28). The rapid phase of cytoskeletal depolymerization is followed by a steady state in which the actin polymer fraction remains low, the actin filament turnover time is enhanced, and cell motility is reduced. Finally, shear stress-accommodated endothelial cells recover their motility, and their actin dynamics return to near prestressed values. The end result is a new cell shape that is retained during chronic shear stress by periods of enhanced actin remodeling.
Deciphering the molecular control of the shear stress-induced actin remodeling response requires analysis of the different time-dependent phases of endothelial actin dynamics under shear stress. F-actin turnover depends on the distribution of filament lengths, the number of free filament ends available for subunit exchange, and the rate of subunit addition and loss from these ends. Recent comprehensive efforts to model the kinetics of the actin cycle revealed that as actin cycling becomes more dynamic, polymer content decreases (2). In cells, actin-binding proteins modulate these parameters and provide multiple access points at which the cell can influence filament turnover.
To depolymerize filaments, loss of subunits from the pointed filament end is the rate-limiting factor (7). Mechanisms that influence pointed end kinetics and exposure, without affecting barbed end capping, affect depolymerization of the actin cytoskeleton. Actin-depolymerizing factor (ADF)/cofilins increase the rate of subunit dissociation from the pointed ends of filaments in vitro up to ∼25-fold (7). ADF/cofilin activity is regulated by phosphorylation of serine-3. Phosphorylation of this residue by the LIM or TES kinases inactivates ADF/cofilin, and this phosphate has been shown to be removed by the Slingshot phosphatases (31).
Although cofilin dephosphorylation activity may explain the initial fluid flow-sensitive actin-remodeling events in endothelial cells, evidence for the role of ADF/cofilins in the shear stress response is inconclusive. The Rho-GTPases Rac, Rho, and Cdc42, which activate LIM kinase, are differentially activated within the first 15 min of shear stress (38, 39, 41), and there is indirect evidence suggesting progressive cofilin phosphorylation and inactivation beginning ∼30 min after shear stress initiation (22). Hence, the burst of shear stress-induced filament turnover and depolymerization occurs before cofilin inactivation. However, results in endothelial cells under static conditions suggest that there may not be a simple one-to-one correspondence between cofilin activation and increased filament turnover. Endothelial cells in a confluent monolayer under static conditions have ample amounts of F-actin-associated cofilin but maintain high polymer fractions with slow rates of filament turnover (24). Furthermore, cofilin association with the actin cytoskeleton actually decreases in moving endothelial cells having the most dynamic actin cytoskeletons (24). Hence, additional investigations are required to uncover the importance of ADF/cofilins in the initial shear stress response.
Increasing the number of free pointed filament ends available for disassembly is another way to accelerate F-actin turnover and cause net cytoskeletal depolymerization, which can be accomplished by filament severing, filament nucleation and release, and/or pointed end uncapping. Of these mechanisms, fragmentation and severing have an advantage over uncapping or nucleation and release for mediating rapid depolymerization events in that they increase the number of free filament ends while decreasing filament lengths. ADF/cofilin and the gelsolin family of proteins are widely studied filament-severing proteins, and of the two, gelsolin is the more potent (31, 37). Unlike ADF/cofilin, the severing activity of gelsolin is Ca2+ regulated (37). Because one of the earliest responses to fluid shear stress is a transient oscillatory rise in free cytoplasmic Ca2+ concentration (3, 35) and cytoplasmic Ca2+ chelation blocks shear stress-induced depolymerization in bulk endothelial lysates (28), gelsolin-mediated F-actin severing may also be important during the initial shear stress response. Under static conditions, however, gelsolin and cofilin appear to work synergistically to regulate endothelial cell actin dynamics and movement (26), and therefore both activities may be important during the shear stress response.
After the initial depolymerization event, an extended cytoskeleton remodeling phase commences that requires a mechanism promoting enhanced filament turnover without changing F-actin content. Because most studies have focused on early (<1 h) and late (>24 h) end points, data supporting a mechanism during this intermediate phase of the actin response are limited. To sustain the new equilibrium of diminished polymer and a high level of filament recycling after turnover-mediated F-actin depolymerization, either cofilin must remain active (24) or other mechanisms must be brought into play. CapG, a member of the gelsolin superfamily that caps barbed ends but does not sever filaments, becomes increasingly associated with the cytoskeleton after shear stress exposure, beginning at 2 h and remaining elevated over static controls at 24 h (30). Because cells vigorously protect the number of barbed ends exposed, this work implies that shear stress-exposed endothelial cells have increased numbers of shorter actin filaments, a finding consistent with observations that average actin filament lengths decrease in endothelial cells with fast actin dynamics under static conditions (26). In addition, because CapG capping events are Ca2+ dependent, basal Ca2+ levels may be elevated compared with no-flow conditions. Studies examining the long-term Ca2+ response after shear stress exposure remain to be performed. Additional investigations must also be performed at the intermediate stages of the shear stress-induced endothelial shape change response to address the regulatory events controlling actin remodeling during this phase.
Unexpectedly, the amount of polymerized actin in shear stress-oriented cells (24 h) is decreased compared with unstressed cells, revealing that increased F-actin content is not necessary to resist shear stress forces in chronically stimulated endothelium; rather, it is the architectural organization of the cytoskeleton that is optimized. The large variability in the rate of filament turnover, even in cells with similar elongated shapes and polymer fractions, suggests that cytoskeletal rearrangement is a continuous, heterogeneous response of the endothelial cell to comply with the force pattern imposed on its surface and to maintain its optimized morphology (1, 33).
F-actin-bundling proteins (e.g., α-actinin) and orthogonal cross-linking proteins (e.g., filamins) may also influence the stability of the actin cytoskeleton at all phases during architectural remodeling, through either a direct effect on F-actin stability or an indirect effect through binding partners. F-actin bundles are more dynamically stable than other cytoplasmic regions (22) and α-actinin and other bundling proteins modestly slow F-actin depolymerization in vitro (6). Although the ability of filamins to affect F-actin dynamics directly are not currently known, filamins have been proposed as organizing centers for actin remodeling, given the large number of binding partners that have been described, including the Rho-GTPases (36); therefore, they may be involved in coordinating the activity of other actin-binding proteins that are potent modulators of actin dynamics through severing, nucleation, capping, or accelerating F-actin subunit loss. In addition, bundling proteins and orthogonal cross linkers may play a direct role in the slow accommodation phase or the maintenance of cell shape after alignment, although additional studies using reconstituted systems with purified proteins must be conducted to understand their influence on actin remodeling more fully.
The observation that the endothelial actin response to shear stress is variable between individual cells is consistent with subcellular spatial differences in fluid shear stress demonstrated by individual endothelial cells on their apical surface (12). Because the surface of an individual endothelial cell undulates, topographical differences determined by the submembrane F-actin network architecture create microvariations in the fluid profile (and hence the fluid shear stress) that locally change the distribution of shear stress (1, 33). Remodeling of actin is an important mechanism for the cell to distribute the stress it experiences by minimizing the load and/or by optimizing its structure to resist the surface mechanical load. This phenomenon may take the form of a pure mechanical optimization or may also include remodeling and alteration of mechanotranducing signaling complexes at the actin membrane surface. Depending on surface topography, neighboring endothelial cells in a confluent monolayer can have radically different phenotypes, given their specific force stimulation. Indeed, this phenomenon is observed in vivo where endothelial cell shape and arterial atherogenicity are markedly different between cells located just a few cell diameters away from each other (12, 13, 15). Therefore, changes in the actin network structure and connections mediated by remodeling events may have a significant impact on cellular mechanotransduction and the atherogenic process.
Support for this research was provided by National Heart, Lung, and Blood Institute Grant HL-54145. E. A. Osborn was a Whitaker Foundation Graduate Fellow.
We thank M. Gimbrone and the Vascular Research Laboratory at Brigham and Women's Hospital (Boston, MA) for generous help with the cell culture and shear stress experiments.
↵1 Supplementary Material to this article (movies 1 and 2) is available online at http://ajpcell.physiology.org/cgi/content/full/00218.2005/DC1.
The costs of publication of this article were defrayed in part by the payment of page charges. The article must therefore be hereby marked “advertisement” in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
- Copyright © 2006 the American Physiological Society