Hemodynamic regulation of directional endothelial cell (EC) migration implies an essential role of shear stress in governing EC polarity. Shear stress induces reorientation of the microtubule organizing center toward the leading edge of migrating cells in a Cdc42-dependent manner. We have characterized the global patterns of EC migration in confluent monolayers as a function of shear stress direction and exogenous pleiotropic factors. Results demonstrate the presence of mitogenic factors significantly affects the flow-induced dynamics of movement by prolonging the onset of monolayer quiescence up to 4 days, but not shear stress-induced morphology. In conjunction with increased motility, exogenous growth factors contributed to the directed migration of ECs in the flow direction. ECs exposed to arterial flow in serum/growth factor-free media and then supplemented with growth factors rapidly increased directional migration to 85% of cells migrating in the direction of flow and induced an increase in the distance traveled with the flow direction. This response was modulated by the directionality of flow and inhibited by the expression of dominant-negative Par6, a major downstream effector of Cdc42-induced polarity. Shear stress-induced directed migratory polarity is modulated by exogenous growth factors and dependent on Par6 activity and shear stress direction.
the endothelium can exquisitely sense and respond to acute and chronic changes in the humoral and hemodynamic environment. Acute and systemic changes in blood flow or tissue demand can lead to endothelial and vascular remodeling (e.g., vasculogenesis, angiogenesis, wound healing, angioplasty). During the process of vascular remodeling, shear stress characteristically regulates the cellular morphology and orientation with the direction of flow (3, 30). In vivo endothelial cell (EC) polarity appears to be dependent on shear stress and varies by age and vessel (20). In vitro studies have demonstrated that distinct hemodynamic waveforms cause similar changes in morphology, but differ in migratory characteristics (1). Of these metrics, migration polarity (i.e., directed migration or planar cell polarity) may play a significant role in maintaining stable cell-cell junctions that regulate vascular permeability and thus serve to protect arteries from the progression of atherosclerosis (20).
Fluid shear stress regulates endothelial migration and polarity through small GTPases and reorientation of the microtubule organizing center (MTOC). At the onset of shear stress, actin stress fibers reorganize to induce EC elongation and alignment in the direction of net flow that is dependent on microtubule activity and the shear-activated GTPases, RhoA, and Rac1 (18, 19, 22, 32, 33). Accordingly, the correct spatial and temporal activation of Cdc42 is required for cellular polarity. Fluid shear stress causes increased activation of Cdc42 at the leading downstream edge of the cell (18) relative to the direction of flow that is dependent on integrin activation to induce a planar cell polarity. Integrin activation by flow leads to Cdc42 activity and induces a Par6-protein kinase C (PKC)-ζ complex to realign the MTOC in the downstream direction relative to the nucleus. Cells expressing either dominant-negative or constitutively active Cdc42, however, do not orient the MTOC (34). Mechanisms of polarization of the MTOC appear to be conserved in a wound injury model as well. Gomes et al. (7) demonstrated that Par6 and myotonic dystrophy protein kinase-related Cdc42-binding kinase, which are downstream of Cdc42, regulate MTOC orientation through divergent pathways following stimulation of serum or lysophosphatidic acid (LPA). In this work, Par6 maintains the MTOC at the centroid of the cell, while myotonic dystrophy protein kinase-related Cdc42-binding kinase induces actin retrograde flow that pulls the nucleus rearward, therefore aligning the MTOC downstream of the migratory direction (7). Last, shear-induced microtubule stabilization downstream of flow is regulated by glycogen synthase kinase-3β, which is spatially inactivated at the leading edge of migrating cells in a Cdc42-dependent manner (6, 20).
Migration and cell polarity can be varied greatly by the effects of the promigratory stimuli of exogenous growth factors and flow. Two angiogenic (mitogenic) factors, basic fibroblast growth factor (bFGF) and vascular endothelial growth factor (VEGF), induce increased EC migration and directional persistence (35). VEGF-activated migration is Rac dependent and unaffected by the activity levels of RhoA or Cdc42 within static conditions (31, 38). Additionally, flow alone can activate tyrosine kinase receptors (e.g., VEGFR2/Flk-1) that share signaling pathways with exogenous growth factors, as well as junction protein complexes (2, 29). EC stimulated with sphingosine 1-phosphate in the presence of shear stress increased migration into a wound at a similar rate to VEGF alone and is dependent on PKC (11). Experiments studying the effects of promigratory stimuli on ECs have primarily focused on wound injury models. To date, the effects of fluid shear stress and growth factors on regulating endothelial morphology, migration, and polarity in a confluent monolayer have not been investigated, and therefore understanding the synergistic or individual contribution on these basic EC responses may provide insight into the link between planar cell polarity and EC phenotypic modulation in intact endothelium in culture and in vivo.
We have developed a new methodology to track bulk statistical migratory and morphological behaviors of confluent endothelium under continuous arterial fluid shear stress with high temporal resolution (1 h) and long exposure times (up to 4 days) (Fig. 1). We show that exogenous growth factors have profound impact on the inability of the endothelium to reach a quiescent motile phase. Furthermore, we illustrate the remodeling capacity of apparently quiescent monolayers to immediately activate and enhance directed migration (or planar cell polarity) on the addition of exogenous pleiotropic factors. This activation demonstrates the endothelium's ability to become sensitized to the local hemodynamic environment and respond with elevated levels of directed polarity. Consequently, we investigated the regulation of Cdc42 by its downstream effector, Par6, in regulating the migratory polarity of ECs under fluid flow and exogenous growth factors.
MATERIALS AND METHODS
Migration and morphological analysis.
Custom image analysis and computational algorithms were developed to track and measure, in an automated fashion, changes in cell shape, speed, and motility of individual cells within a confluent monolayer from real-time video microscopy images (1). Using a microscope mounted version of the hemodynamic flow device (Fig. 1), sequential images of the monolayer were acquired using phase optics from a Nikon TE2000 microscope and an automated digital image acquisition system controlled by MetaMorph software (Molecular Devices). Images were acquired every 2 min for up to 96 h using a ×4 phase objective. With this temporal and spatial resolution, cellular movements can optimally be resolved in ∼1,200 cells per image before processing and cell selection. Images are stored offline for postprocessing analyses.
Time-dependent changes in morphology and migration of individual cells within a monolayer were quantified using methods adapted from those previously described (1). Briefly, in MetaMorph, a series of image filters were applied to the sequential stack of images to enhance the contrast between cell borders. Intensity variations within images were smoothed using a low-pass filter, converted to a binary image, and subjected to a watershed function to skeletonize the image based on its phase contrast. The resultant image revealed traces of individual cell borders for all cells in the monolayer (Fig. 1C). For each image, outlines of each cell were assigned an object number. The x-y centroid, area, and shape factor measurements were recorded for all object numbers throughout the entire movie (2,880 images). The sequence of skeletonized images was overlayed on the original phase image to collect average intensity and integrated intensity values for all cells. Together, these measurements are used to automatically select for and track individual cells in the monolayer. On average, each processed image contained ∼1,200 cells. A secondary set of macros was written for input of the data into Matlab for further analysis.
In Matlab, custom macros were used to 1) track the average and individual morphometric measurements as a function of time, and 2) calculate cell distances between consecutive time frames and track cellular movements based on a minimum distance algorithm. From this data, the intrinsic migratory characteristics, root mean square cell speed (S) and directional persistence time (P), are computed using a well-documented random walk model (5). The equation for the model is as follows: (D2) = S2P2 (T/P − 1 + e−T/P), where D2 is the mean square displacements, and T is the time intervals over which D2 is calculated. A nonlinear least squares curve-fit routine is used to solve for the coefficients S and P. The persistence time is defined as the average time between changes in direction, which quantitatively describes the translation movement of the cell. Of the 1,200 cells captured from the MetaMorph analysis, cells that met the following four selection criteria were not including in the analyses: dividing cells, apoptotic cells, merging objects and incorrectly tracked cells, and migration of cells out of the field of view. This analysis routinely generates positive selection data for ∼100–600 cells per experiment. Furthermore, the computation of S and P is conducted in sequential 1-h time intervals to compute dynamic changes in the migration parameters. Data from individual cells and identical experimental conditions are combined to generate a mean ± SE for each 1-h time block. Motility equals the square of the speed multiplied by directional persistence time and divided by the number of dimensions of motion (in this case two dimensions) and conveys the motile state of the cells. Speed, persistence, and motility calculations represent two-dimensional migration, irrespective of the flow direction. Additionally, distance measurements represent planar migration that is evaluated independently from the direction of shear stress, unless otherwise noted. Directional migration, however, represents locomotion with the direction of flow.
Morphological changes were measured using the shape factor calculation within MetaMorph. A shape factor of “1” represents a circular object, and a shape factor of “0” represents a straight line. Shape data for each object were recorded and averaged for each time frame. A mathematical rolling average was performed on the data set for the entire experimental duration. The mean for each time point was then processed by a median filter to minimize noise and allow for more accurate analysis of general trends in elongation.
Cell culture and flow experiments.
Human umbilical cords were obtained with institutional approval (University of Virginia Health System, Human Investigation Committee no. 10486; Martha Jefferson Hospital, IRB no. 0179). Human umbilical vein ECs (HUVEC), isolated from single, normal term cords, were cultured in normal growth medium (NGM) (M199 with 25 mM HEPES; Biowhitaker), 10% fetal bovine serum (FBS; Gibco and HyClone), and supplemented with 5 μg/ml heparin, 5 μg/ml EC growth supplement (ECGS) (Biomedical Technologies), 2 mM l-glutamine, and 100 U/ml penicillin G + 100 μg/ml streptomycin, and incubated at 37°C in 5% CO2 in air. Experiments were performed on HUVEC from single donors and used at passage 2–3.
ECs were exposed to a human arterial shear stress flow waveform using our laboratory's hemodynamic in vitro model, as previously described (1). For the flow experiments, HUVEC were plated at 80,000 cell/cm2 in NGM (10% FBS in M199, 2 mM l-glutamine, 5 μg/ml ECGS, 5 μg/ml heparin, 100 units penicillin/streptomycin) for 24 h to attain a confluent monolayer. Cells were exposed to a continuous, pulsatile arterial shear stress waveform for a period of 48–96 h in either NGM or growth factor-deprived basal media (BSM; M199, 1% bovine serum albumin, 2 mM l-glutamine, 100 units penicillin/streptomycin) that contains no serum or growth factors. For the growth factor addition experiments, cells were preconditioned with BSM for 31 h of shear and then supplemented with serum plus growth factors (10% FBS, 5 μg/ml ECGS, and 5 μg/ml heparin), growth factors alone [VEGF (20 ng/ml) and heparin (5 μg/ml); bFGF (20 ng/ml) and heparin (5 μg/ml); serum (10% FBS) and heparin (5 μg/ml); or LPA (2 μM) and heparin (5 μg/ml)] for an additional 5 h for a total flow exposure time of 36 h. All media contained 2% (wt/vol) high molecular weight dextran (∼460 kDa; Sigma) to increase media viscosity to 2 cP, as previously described (1).
Dominant-negative Par6 transfection experiments.
HUVEC were transfected with dominant-negative Par6 (DN-Par6), a mouse Par6B isoform lacking the first 100 amino acids, cloned into the pkVENUS vector, a green fluorescent protein variant. The NH2-terminal of Par6 binds atypical PKCs, and the deletion mutant knocks down the activity of the Cdc42 pathway, as previously described (12, 34). Fluorescently labeled cells, expressing DN-Par6, were utilized to demonstrate the effects on directed migration. Both the DN-Par6 and pkVENUS control constructs were generously provided by Dr. Martin Schwartz, University of Virginia. Cells were transfected utilizing the Lipofectamine 2000 (Invitrogen) system and 20 μg of the indicated DNA as per manufacturer's instructions. Briefly, 30 μl of Lipofectamine 2000 and 20 μg of DNA were each incubated in 1.5 ml OptiMEM (Invitrogen) for 5 min and then combined and incubated an additional 20 min together. During the incubation periods, the cells were maintained in 6 ml of OptiMEM. On completion, the 3 ml of OptiMEM containing Lipofectamine 2000 and DNA were added to the cells and incubated for 2 h. Following 2 h, the cells were maintained in NGM without antibiotics for 24 h. Transfection efficiency before the start of shear stress was 33.7 ± 7.8% (mean ± SD) (supplemental Fig. 1; the online version of this article contains supplemental data). Control and DN-Par6 transfected monolayers were preconditioned to arterial shear stress for 24 h before the addition of NGM. The 24-h time point was selected since it coincided with optimal expression, whereas longer time points reduced vector expression (data not shown). Cells expressing average fluorescent intensity >10% of the median expression level above background were identified as positively transfected cells. The average intensity value of the transfected cells was 33 ± 8% (mean ± SD) above background, which corresponds to the same amount of expression of the DN-Par6 protein. Expressed DN-Par6 was fused to GFP, allowing fluorescent intensity to be positively correlated to expression levels. Individually labeled ECs were tracked for net distance traveled over a 3-h period to determine directed migration.
Data analysis and statistics.
Migration results (S, P, motility) are combinations of at least three independent replicate experiments, and sample sizes are presented within the figure text. Shape factor data are a representation of three independent experiments. Differences in migration polarity are replicates of at least three independent experiments and represent the increase in positive polarity normalized to the polarity before growth factor addition. Student's T-test was performed to determine significance in migration, shape factor, and cellular polarity data. One-way ANOVA was performed for migration characteristics and NGM addition experiments. A Rayleigh test for statistical analysis of circular data was conducted to determine nonuniformity of polar migration.
Methodology to evaluate statistical migratory and morphological behavior of confluent ECs.
Custom image processing algorithms were developed to track individual ECs in the continual presence of hemodynamic relevant shear stress forces and assess the bulk migratory and morphological behavior within confluent monolayers (see material and methods for details). This was accomplished using time-lapse microscopy and a previously described cone and plate flow device that mounts onto the microscope stage and permits the continual monitoring of cells under a fluid flow environment (Fig. 1) (1). The tracking of cells resulted in finely capturing individual cell borders and converting them to objects whereby information about a cell's shape, coordinate, and integrated intensity were recorded. The automated method eliminated cells from the analysis that were poorly tracked due to phase variation, moved out of the fixed viewing field, divided, or underwent apoptosis. Positive cell selection identified a mean of 316 ± 156 (SD) cells per experiment over 31 independent experiments.
Shear stress regulates EC morphology independent of exogenous growth factors.
EC morphology in vitro and in vivo is highly dependent on the net direction of shear stress. Under NGM (see materials and methods), our laboratory has previously observed that shear stress causes cells to translate while changing their morphology (i.e., elongating and aligning with the flow direction), and the rate and extent of their morphological change were dependent on the time-average shear stress (1). It is unclear, however, whether exogenous growth factors commonly used in cell culture-based flow experiments influence this response. To assess the influence of exogenous growth factors on the long-term morphological response, HUVECs were exposed to an arterial pulsatile shear stress profile (Fig. 1) for 96 h in the presence of NGM or BSM (without serum, ECGS, and heparin supplements), and morphological changes were quantified as a function of time. Results show the extent of cellular elongation and rate of alignment in the direction of flow were independent of the media conditions (Fig. 2) (P > 0.05). These results demonstrate changes in EC morphology are dependent on shear stress and independent of exogenous mitogenic factors in the media.
Exogenous growth factors stimulate shear stress-induced migration.
To further quantify the effects of the growth factors on shear stress-induced cell migration, the net and total distance traveled per 1-h interval was computed. The average net and total distance traveled over 96 h of flow exposure was significantly reduced in the absence of exogenous growth factors, BSM, compared with NGM and independent of the direction of cell translation relative to the direction of flow (Fig. 3 and supplemental Table 1, respectively, P < 0.05). Furthermore, a greater percentage of cells migrated in the direction of flow, irrespective of the media conditions (supplemental Table 2).
Exogenous serum and growth supplements in NGM regulate endothelial migration behavior.
To assess real-time migratory behavior, cellular migratory characteristics, including the S and P (using the random walk model), were determined. Persistence time is a measure of the path by which the cell moves: a longer P indicates persistent movement in a given direction, whereas a lower P indicates the cell is changing direction more frequently. These metrics were used to establish the dynamic adaptation and potential stabilization (i.e., quiescence) of remodeling endothelium exposed to long-term fluid shear stress. Under NGM, results demonstrated for the first time that ECs have a dynamic migratory pattern under continuous arterial flow that varied consistently and dramatically as a function of time (Fig. 4A). Cells exposed to NGM exhibited greater fluctuations in cell speed, directional persistence, and motility throughout 96 h of flow, and the extent of fluctuations rose at later times. Cells consistently displayed increased cellular motility to a level two and half times the initial basal levels, beginning after 20 h of flow. The increase in motility was dominated by the increase in mean directional persistence and reached a maximum at ∼36 h. Another maxima in motility was consistently observed at 60 h of flow, also coinciding with an increase in persistence time. Although not as dramatic, dynamic changes in cell speed throughout the 4 days were significant and ranged from ±20% of the initial starting speed. In contrast, cells exposed to media without serum or exogenous growth factors (BSM) exhibited a dramatically more quiescent migratory state that persisted throughout the 4 days (Fig. 4B) (statistical analysis, supplemental Table 3). The increase in directional persistence observed in the NGM media was completely abolished in BSM, resulting in motility that was nearly one-third of the initial basal level. In fact, the cells in BSM appeared to move around a tightly confined region bounded by the adjacent cells in the monolayer, which was in stark contrast to the highly motile cells in NGM conditions. A reduction in cellular persistence was the prominent factor in limiting motility. Increased levels of cellular speed were measured, but were attributed to the greater rate of change in movement of the centroid in a confined region.
We hypothesized that exogenous growth factor components in the media were triggering a continually active migratory phenotype in response to the flow environment. To test this hypothesis, growth factor addition experiments were conducted to determine the consequences of serum/growth supplement media (NGM) on cells that were preadapted in BSM. Cells were initially preconditioned with flow in BSM for 31 h, which represents a time point consistent with a steady-state reduction in motility (Fig. 4B). NGM was then added to the cells in the continual presence of flow and monitored for an additional 17 h (up to 48 h) (supplemental Fig. 2A). The addition of the NGM immediately reduced S, increased P and motility, and led to a sustained response similar to experiments in which cells were exposed to NGM for the entire duration (Fig. 5, A, B, and C, supplemental Table 4, statistical analysis). The addition of NGM also resulted in a dramatic enhanced translational movement of the cells in the direction of flow (Fig. 5D) (P < 0.05). These results demonstrate the profound effect growth media has on the dynamics of endothelial migration and directed migration during the continual exposure to shear stress.
Shear stress is a major determinant of EC polarity.
In an attempt to isolate the effects of exogenous growth factors and shear stress on EC-directed migration (i.e., polarity), we conducted an experimental study in which NGM was added 31 h after preconditioning the cells in BSM condition and either 1) continued flow, 2) stopped flow, or 3) reversed the arterial flow field (supplemental Fig. 2). Postprocessing analyses were performed on each condition to assess the cell migratory characteristics (S, P, motility) and relative changes in the total number of cells migrating in the direction of flow. Similar to results in the previous section, addition of NGM in the presence of continued flow decreased speed, enhanced persistence and motility, and caused a twofold increase in directed cell migration with the flow field. In contrast, stopping flow following the addition of NGM resulted in no change in the speed, persistence, or motility, thus maintaining the migratory characteristics that were present before the cessation of flow (Fig. 6B). An approximate 50% increase in directed migration was observed 2 h following flow cessation. Reversing the arterial flow field immediately following the addition of NGM caused the EC monolayer to reverse polarity (>2-fold) and began to migrate in the new direction of the flow field (Fig. 6C, bottom). The change in cellular direction is depicted by the relatively rapid increase in cell speed and concomitant decrease in directional persistence in the hour following this imposed stimuli. The differential response to the addition of exogenous serum/growth factors further corroborates the concept that the cells are reacting to the physical stimuli of shear stress and that the presence of exogenous serum/growth factors only elicits a higher sensitivity to flow.
Remarkably, cell translation and direction were drastically influenced by shear stress in response to the addition of NGM following preconditioning with BSM. Following the addition of growth media, the percentage of cells translating in the direction of flow increased to nearly 90% of the field of view, and the average net distance of these cells increased by two and one-half times the baseline level; whereas the small percentage of cells traveling against flow did not exhibit the dramatic increase in net distance traveled (Fig. 5D).
Exogenous serum/growth factors stimulate shear stress-induced directed migration.
Numerous components present within the serum and ECGS are known to induce a promigratory response. To determine the relative response of the endothelium to some of the major components, growth factor addition experiments were conducted following the preconditioning of EC monolayers for 31 h in the presence of BSM. First, the responses to 10% FBS or ECGS were tested independently. Although a modest transient increase in polarity was observed in the presence of EGCS alone (1.4 ± 0.3-fold), the addition of serum replicated the robust and sustained response seen within the NGM condition (Fig. 7 vs. Fig. 6A, bottom). A major component of serum, LPA has been shown to increase migration in a scratch-wound assay (7), although it has not been tested in intact monolayers. Addition of LPA enhanced the migration polarity (2 ± 0.3-fold) to the levels seen for FBS and NGM (Fig. 7). Next, two prominent components of ECGS were tested: VEGF and bFGF. Addition of VEGF or bFGF also significantly enhanced the directed migration polarity (67 and 86%, P < 0.10 and P < 0.05, respectively; Fig. 7). This transient increase was followed by a sustained increase in polarity above prestimulated levels (Fig. 7). Furthermore, it was determined that the addition of these individual factors resulted in no change in the computed migratory characteristics (cell speed, persistence, motility) (supplemental Fig. 3 and Fig. 1), which is in dramatic contrast to the addition of NGM. Thus EC migration polarity appears to be independently regulated from other migratory characteristics of speed and persistence, and the combination of exogenous factors present in NGM has an effect on both migration and polarity.
Flow-induced polarity is dependent on Par6 activity.
Cdc42 and its downstream effectors are well known to regulate polarity in several cell types. Previous studies showed that shear stress-induced polarity (i.e., reorientation of the MTOC relative to the nucleus) was dependent on Cdc42 activity and the activity of Par6, a downstream effector of Cdc42. Activation of this pathway contributed to the reorientation of the MTOC downstream of the nucleus in the direction of flow. Here we tested whether Par6 regulates the increase in migration polarity observed upon serum/growth factor (NGM) addition in the presence of shear stress. A dominant-negative mutant of Par6 (DN-Par6) or control vector (pkVENUS) was transfected into confluent ECs. Cells were preconditioned for 24 h of flow in BSM, followed by the addition of NGM in the presence of continuous arterial flow. ECs transfected with the DN-Par6 demonstrated no preferential direction of migration relative to the net flow direction (P = 0.57). In contrast, EC transfected with the control vector demonstrated significant directed migration with the net flow direction (P = 1.33 × 10−9), as determined by the Rayleigh test (Fig. 8). Furthermore, DN-Par6 EC exhibited a mean net translation of ∼12 μm, randomly, in both the negative and positive flow directions. In contrast, control vector translated ∼15 μm in the positive flow direction while only traveling 7.25 μm in the negative direction (data not shown), similar to the shear-dependent increase in translation following NGM addition. As a result, shear stress-induced polarity (i.e., directed migration) is dependent on Par6 and implicates Cdc42 as a major signaling axis.
We studied the bulk remodeling of confluent ECs in response to shear stress and exogenous growth factors and assessed the functional relationship to cellular migration, morphology, and polarity (i.e., directed migration). Arterial shear stress induced elongation and alignment of ECs in the direction of net flow, as shown previously. The rate and extent of cellular alignment and elongation, however, were determined to be independent of exogenously added growth factors that are commonly used in cell-based experiments. In the continual presence of arterial flow, exogenous growth factors sustained a promigratory phenotype up to 96 h by eliciting greater motility, persistence, and translational distance. In striking contrast, ECs in media without serum or growth factors (BSM) resulted in a nonmigratory phenotype. This phenotype was reversible by the subsequent addition of serum/growth factors back into the flow media. This also resulted in an increase in the bulk directional migration of ECs, effectively increasing the cellular polarity. Polarity was found to be dependent on both continuous exposure to flow and the net flow direction. Our data indicate that shear stress induces 1) a promigratory phenotype that is dependent on exogenous growth factors, and 2) a polarity that is a function of Par6, a downstream effector of Cdc42.
The local hemodynamic shear stress environment regulates endothelial morphology and orientation in vivo and in vitro (15, 20, 27, 30). This fundamental morphological response is dependent on actin stress fiber reorganization, partial disassembly of adherens junctions, microtubule dynamics, and small GTPases RhoA and Rac1 (16, 18, 19, 21, 22, 32, 33). A previous study demonstrated that, in response to shear stress, ECs simultaneously elongate and exhibit decreased random motility (4). In our model, we were able to decouple these events, such that, under BSM conditions, EC elongate without a change in migratory characteristics. This implies that distinct biological pathways regulate morphological and migratory behavior. The data also support previous findings that EC morphology is dependent solely on the time-average shear stress (1, 9, 13). Collectively, it is clear that the shear stress is the dominating factor regulating the morphological changes of the endothelium.
Shear stress regulates endothelial motility and directed migration that is dependent on exogenous growth factors. Reduced levels of migration and directed polarity were evident in BSM conditions within 24 h. The promigratory phenotype previously observed was completely rescued by the addition of serum/growth factors, and the elevated motility was accompanied by a distinct migration polarity downstream of flow. To decouple the effects of exogenous growth factors and shear stress, the serum/growth factor addition was accompanied by a cessation of flow. Following the cessation of shear stress, the addition of serum/growth factors failed to elicit a remodeling effect, whereby motility and directed migration remained unchanged. Furthermore, when the serum/growth factor addition was accompanied by a reversal of the net flow direction, we did not observe the significant changes in migratory characteristics (S, P, motility) compared with the condition with continual flow. This reduction in EC motility may be the result of the time needed for repolarization of the cellular locomotive machinery. In sparse cultures, the onset of flow recruits focal adhesion kinase to the leading edge of the cell, and ECs exhibit very little motility. Subsequently, existing focal adhesions remodel in response to the new migration direction in a process that can take several hours (17). This remodeling may explain the significant lag period in mass migration observed after NGM addition with a reverse in net flow direction. Furthermore, the NGM addition induces a directed migration and increase in net translation that is dependent on the net flow direction, which supports previous studies exhibiting increased EC motility preferential to the flow direction in confluent monolayers (4). An alternative hypothesis is that extracellular matrix organization by the endothelium may have a “directional-sense,” in and of itself, which may contribute to the suppressed migration upon flow reversal, although this is untested to date. Kiosses et al. and McCue et al. demonstrated within a rabbit model that retrograde flow disrupts the MTOC polarization of local ECs, theoretically disrupting the directed migration toward the heart found in rat aortas (14, 20). This flow pattern preferentially develops atherosclerosis, inferring a possible link between polarity, directed migration, and disease progression. In vivo wound repair models have demonstrated that regeneration of a denuded endothelium is dependent on EC proliferation as well as migration occurring at an increased rate along the blood flow axis (8, 28). Further studies demonstrated that wound repair in a rabbit common carotid was reduced in low-flow conditions, while normal-flow conditions demonstrated elevated wound closure oriented in the flow direction (36). Although the mechanism of action needs to be further studied, the ability of ECs to remodel in response to differential flow demonstrates that shear stress is the major regulator of rate and extent of migration polarity.
The increase in migration characteristics and directionality appear to be regulated by separate pathways that can induce an increase in migratory state, directionality, or both. Increases in migration could be attributable to Rac1 stimulation, whereas directionality may be a result of localized Cdc42 activity (10, 34). In static conditions, VEGF-stimulated motility is dependent on Rac activation (31). The addition of shear stress, however, creates a more complex regulation of endothelial migration that does not elicit a promigratory phenotype on the addition of a single growth stimulus, although it will induce a change in polarity. Major components of both serum and ECGS were tested individually for the capacity to induce shear responsive-directed migration and net translation. bFGF and VEGF both stimulated increased directionality, however, at a reduced rate compared with the full NGM addition. This increase in directionality did not translate into increased levels of basic motility (S, P, motility, and net translation). This phenomenon may be attributable to the tyrosine phosphorylation of the VEGF receptor Flk-1 (2) and the elevated mRNA levels of both VEGF and bFGF that are induced by shear stress (25). The reduced response to VEGF/bFGF compared with NGM may be due to a desensitization of Flk-1 and/or the bFGF-receptor by flow. The shear stress-induced elevation of endogenous levels of both VEGF and bFGF could also be desensitizing the effects of exogenous stimulation of each of these factors. Furthermore, the addition of ECGS alone, which comprises multiple factors, including bFGF and VEGF, did not affect polarity, suggesting that the addition of defined combinatorial factors may serve to suppress the observed increase in polarity in response to the individual factors alone.
A second flow-induced pathway regulating directionality may contribute to increased EC migration. The serum component, LPA, was able to induce significant directional migration in the presence of continuous flow. Previous studies utilizing static models established that LPA treatment of ECs along a wound edge induces a MTOC reorientation toward the wound and rearward movement of the nucleus (7). Similarly, LPA may be signaling the MTOC and nucleus to realign in response to the shear stimulus. Similar to the bFGF and VEGF response, LPA was also unable to induce increased levels of basic migratory characteristics (S, P, motility) or net translation, whereas serum stimulation induced an increase in both migration and polarity. As a result, we hypothesize that the increased migratory state and directionality is a result of the aggregate of several growth stimulants.
The directed migration of ECs appears to be dependent on Cdc42 and its downstream effectors. LPA activates Cdc42 and is necessary to reorient the MTOC in wound models (23). Activation of Cdc42 at the leading edge (18, 34) provides a signal for directed migration absent in cells cultured in growth-depleted media. Activation of the Par6-PKC-ζ complex regulates glycogen synthase kinase-3β, which is locally phosphorylated at the leading edge of the cell and induces migration polarity (6). While conflicting evidence has demonstrated that Rho and Rac regulate cell polarity in both shear and static conditions, flow studies have identified Cdc42 as the regulator of MTOC orientation (24, 34, 37). These studies have included fibroblasts and sparse EC, which exhibit differential responses to growth stimulation from confluent EC cultures (26, 35). In our migration model of confluent EC, inhibition of Par6 activity resulted in a loss of directed migration upon NGM addition and negated the cell polarity regulation by flow. Therefore, it is inferred that exogenous growth factors are sensitizing the monolayer to shear stress and that directed migration is dependent on Par6 signaling and likely its upstream regulator, Cdc42.
We conclude that shear stress in the presence of exogenous growth factors activates the Cdc42-Par6-PKC-ζ signaling axis, leading to directed EC migration and net translation. In conjunction, shear stress regulates EC bulk migratory characteristics as well as morphology. This study also reveals that traditional markers of EC quiescence in confluent monolayers are more complex than contact inhibition of proliferation and that exogenous growth stimulants may still be rendering an active, promigratory phenotype. Furthermore, this model of EC stimulation by both shear stress and exogenous factors may be useful in understanding how growth factor stimulation from adjacent tissue can influence endothelium residing in a quiescent vessel; for example, during the development of the vasculature, angiogenesis, and atherosclerosis.
This work was supported by The Whitaker Foundation Biomedical Research Grant RG-02-0853 to B. R. Blackman. M. B. Simmers is a predoctoral trainee support on National Institutes for Health Basic Cardiovascular Research Training Grant 5T32HL0084.
The authors acknowledge Drs. Martin A. Schwartz and A. Wayne Orr for reagents and thoughtful scientific discussions; Bradley Gelfand, Ryan Feaver, and Nikki Hastings for critical review of the manuscript; and Dr. Jae Lee for guidance in statistical analyses.
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 © 2007 by the American Physiological Society