Shear Stress Gradients Remodel Endothelial Monolayers in Vitro via a Cell Proliferation-Migration-Loss Cycle
Abstract Wall shear stress has been implicated in the genesis of atherosclerosis because a strong correlation exists between the location of developing arterial lesions and regions where particular gradients in stress occur. Studying the behavior of endothelial cells in such regions may contribute to our understanding of the disease etiology. We report the detailed migratory history of endothelial cells subjected to large shear stress gradients caused by a surface protuberance in an in vitro model system. The history of cell migration, cell division, and cell loss from the surface was continuously monitored in confluent human umbilical vein endothelial cell monolayers for 48 hours after the onset of flow. Individual cells were tracked using time-lapse video microscopy. In contrast to a uniform laminar flow field in which cells were observed to continually rearrange their relative position with no net migration, in a disturbed flow field there was a net migration directed away from the region of high shear gradient. This organized migration pattern under disturbed flow conditions was accompanied by more than a twofold increase in cell motility. In addition, cell division increased in the vicinity of the flow separation (maximum shear stress gradient of 34 dyne/cm2 per mm) whereas cell loss was increased upstream and downstream in the regions where the shear gradient diminishes. These data suggest a steady cell proliferation-migration-loss cycle and indicate that local shear stress gradients may play a key role in the morphological remodeling of the vascular endothelium in vivo.
- Received December 20, 1996.
- Accepted February 27, 1997.
The localized nature of atherosclerotic lesions along the arterial bed suggests that hemodynamic factors play a key role in the genesis of the disease. This has led many researchers to investigate the response of vascular endothelium to physical factors.1 2 3 Several hypotheses have been suggested to explain the relationship between hemodynamics and arterial disease.4 5 Regions of disturbed flow, where low and oscillatory shear stress conditions prevail, have been shown to correlate well with the localization of atherosclerotic lesions.6 7 These conditions are found near arterial branches, where flow separation and recirculation create large gradients in fluid shear stress on the lumenal surface of arteries.7 8 9
In contrast to the numerous studies involving uniform shear stress,2 3 little is known about the effects of large shear stress gradients on the structure and function of vascular endothelium. In vivo, artificially created constrictions have been used to study the morphology10 11 and repair capability9 of the endothelium as well as lipid deposition12 in the different regions of the stenosis. In an attempt to better characterize the cell biological effects of such disturbances, Depaola et al13 developed an in vitro model that recreates in a cone-and-plate apparatus14 the large gradients in shear found near arterial bifurcations. Cell density measurements and cell proliferation assays in these experiments,13 suggested that cell migration is an important cellular response to disturbed flow. Using a modified parallel plate system that allows for continuous time-lapse video microscopy, it was possible in the current study to explore this hypothesis by analyzing the detailed migratory history of endothelial cells subjected to well-defined large shear stress gradients.
Materials and Methods
Second passage human umbilical vein endothelial cells, established from normal term umbilical cords as described previously,15 were plated on tissue-culture-grade polystyrene (Costar Corp, Cambridge, Mass, and Modern Plastics, Peabody, Mass) coated with 0.1% gelatin (Difco Laboratories Inc, Detroit, Mich) and grown to confluency in medium 199 (with 25 mmol/L HEPES; Gibco BRL, Gaithersburg, Md) supplemented with 10% fetal bovine serum (Gibco BRL), 2 mmol/L of glutamine, 100 μg/mL of penicillin, 100 μg/mL of streptomycin, 25 μg/mL of endothelial cell growth supplement (Collaborative Research Inc., Bedford, Mass), 50 μg/mL of heparin (Sigma Chemical Co, St Louis, Mo) and 250 ng/mL of amphotericin B (Fungizone; Gibco BRL).
Shear Stress Apparatus and Flow Conditions
A parallel plate flow chamber16 was modified by placement of a rectangular obstacle perpendicular to the flow field (Fig 1⇓). The flow-disturbing bar was 1.0 mm wide and 0.4 mm high, and it fully crossed the 5-mm-wide flow channel. The details of the flow separation, reattachment, and recovery to predisturbance conditions were calculated by solving the 2D Navier-Stokes equations using the computational fluid dynamics code NEKTON.17 For our specific experimental conditions (14 dyne/cm2, Reynolds number 20), the simulation predicted a negative shear stress region extending from the obstacle to the reattachment point with a maximum shear stress gradient of 34 dyne/cm2 per mm located 0.6 mm downstream from the obstacle. The reattachment point could be visualized by the trajectories of small debris flowing in the medium and corresponded well to the simulation (0.7 mm downstream from the bar). The shear stress recovered its nondisturbed value 1.5 mm from the bar. As a comparison, 14 dyne/cm2 uniform shear stress experiments, ie, without the flow-disturbing bar, were also performed and are referred to as the “nondisturbed flow condition” in this paper.
Note that the maximum shear stress gradient of 34 dyne/cm2 per mm produced by the disturbed flow conditions selected in this study is comparable with what can be expected in vivo. For example, gradients greater than 10 dyne/cm2 per mm can be calculated from the spatial wall shear stress distribution measured by Lutz et al8 in a model canine aorta during steady flow.
The flow chamber was placed on the stage of a Nikon Diaphot inverted microscope perfused with culture medium. The pH of the circulating medium was maintained at 7.4 by bubbling 5% CO2/95% air in the reservoir. Both the flow chamber and the medium reservoir were maintained at 37°C. Dextran (476 000 mol wt, 2.3% wt/vol; Sigma) was used to control the viscosity of the medium. Phase-contrast images of the cell monolayer were recorded every 4 seconds using a time-lapse video recorder (Panasonic, AG 6730) during a typical 48-hour experiment.
Cell Spatial Density Distribution
At the conclusion of the experiment, cells were fixed in 2% paraformaldehyde at 4°C and stained with Wright’s Giemsa solution (Fig 2⇓). In a typical sample (Fig 3b⇓), the cell density was measured in the disturbed flow region by counting the number of cells in 24 rectangular strips (0.1 mm wide by 1.0 mm length) oriented perpendicular to the flow.
Cell Tracking and Motility Parameters
Software was developed to control the video recorder through the RS232 port of an IBM compatible computer. This allowed the tapes to be replayed and paused on any specified frame. The system would automatically digitize and store one image every 10 minutes on the hard disk. A total of 20 individual cells were then tracked semiautomatically by pointing at their nuclei and creating a database of their position during the course of the experiment.
To characterize cell motility, two parameters were calculated: cumulative distance and net displacement. The cumulative distance is the sum of the distances traveled by the cell integrated along its trajectory. The net displacement is the distance between the locations of the cell at the beginning and at the end of the experiment. To analyze the influence of flow, both parameters were decomposed into their respective projections perpendicular and parallel to the direction of flow.
Cell Loss and Division
When replayed at higher speed (1 minute of replay=2 hours of real time), cell division and detachment of individual cell could clearly be seen and counted. In three experiments, the rate of cell loss and division was measured in three specific regions; directly behind the obstacle, in the flow reattachment region, and downstream from the flow recovery region (the exact location and dimensions of the three regions are specified in Fig 7⇓).
Affected Cell Spatial Distribution
Disturbed flow conditions affect cell spatial distribution. Fig 2b⇑ shows a human umbilical vein endothelial cell monolayer in a region of disturbed flow after 48 hours, as compared with a monolayer maintained in parallel under static (no flow) conditions (Fig 2a⇑). A dramatic remodeling of the cell monolayer, including a marked decrease in cell density (50% compared with static control) in the region just upstream from the reattachment line was observed. The comparison between the cell density distribution for a typical sample (Fig 2b⇑) and the calculated shear stress distribution (Fig 2a⇑) indicates that the depletion zone corresponds approximately to the maximum shear stress gradient region. It appears that the integrity of the monolayer cannot be maintained in this region.
Similar morphological observations were reported by DePaola et al13 with bovine aortic endothelial cells, with the exception of the increased cell density immediately downstream from the flow-disturbing bar that is apparent in the current data from human endothelial cells in Fig 2b⇑. In the earlier experiments, the comparison between the spatial distributions of cell density and percentage of cells committed to division suggested that cell migration was being activated in the region of high shear stress gradient.
Increased Cell Motility and Production of Net Migration
Disturbed flow conditions increase cell motility and produce a net migration away from the region of high shear gradient. A total of 20 randomly selected cells that were initially located along the reattachment line have been tracked during the course of the experiment. Individual cell tracks shown in Fig 4⇓ for disturbed and nondisturbed flow conditions confirm that the shear stress gradients can have a strong effect on cellular movement in vitro. Cells originally located around the reattachment line can migrate up to 10 cell diameters either upstream or downstream (Fig 4b⇓). In contrast, cells subjected to nondisturbed flow conditions slide along each other and continuously rearrange their relative position as they change their shape. The direction of these movements was more or less random, resulting in negligible net cell migration; these cells were found predominantly within one cell diameter (≈40 μm) from their original position (Fig 4a⇓).
Besides having a more directional motility pattern, as measured by an increased net displacement, cells subjected to disturbed flow conditions demonstrated a higher intrinsic motility. The average rate of movement, as measured by the cumulative distance traveled during 48 hours, was several times that of cells subjected to uniform shear stress (Fig 5b⇓).
Finally, it was noted that both disturbed and nondisturbed flow conditions biased cell motility in the flow direction. Both cell populations in Fig 5b⇑ are located in the lower right sector of the figure, signifying a preferential motion in the axial direction.
Taken together, these data indicate that endothelial cells respond to shear stress gradients in two ways: first, by increasing their net motility (increased cumulative distance); and second, by exhibiting a more organized motile activity resulting in a net migration away from the region of maximum shear stress gradient.
The migration rate as a function of shear stress gradient cannot, however, be rigorously measured from these data since an individual cell experiences different flow conditions as it moves in the shear stress gradient. To account for that, groups of 20 cells have been tracked during a limited period of time (1 hour) in three specific regions (flow reversal, flow reattachment, and flow recovery regions). The group velocities obtained in these regions are shown as a function of time in Fig 6⇓. In the flow reversal region, the average group velocity is −9 μm/hour, indicating a net migration toward the obstacle whereas in the flow recovery region, the migration velocity is in the downstream direction (+9 μm/hour). In the reattachment region, the group velocity oscillates around zero, indicating no net migration but a rather disperse distribution of cell velocities. Therefore, the shear stress gradient separates the cell population around the reattachment line, part of it being dragged upstream along with the negative shear stress and the rest of it being pushed downstream across the flow recovery region.
The migration rates are almost constant during the course of the experiment. As discussed below, this indicates that cells have to proliferate near the reattachment line to reach the rather steady spatial density distribution depicted in Figs 2⇑ and 3⇑.
Proliferation, Migration, Accumulation, and Detachment
Cells appear to continuously proliferate in the high shear stress gradient region, then migrate upstream and downstream where they accumulate and eventually detach. Cell proliferation and detachment rates were estimated in the reattachment region and upstream and downstream from the flow reversal and recovery region, respectively (Fig 7⇓). As compared with the nondisturbed flow region, cell proliferation is clearly greater around the reattachment line. Simultaneously, detachment rates are higher than nondisturbed values in the two other regions. Note that these regions of increased cell loss correspond to the regions where cells accumulate, as was observed in the density distribution of Fig 3b⇑.
Combined with the migration pattern observed above, these findings suggest the existence of a chronic cell proliferation-migration-loss cycle in the disturbed flow region. The shear stress gradient stimulates cell motility and separates the cell population that migrates in the flow direction either upstream or downstream from the reattachment line. In this region, the density drop and/or the shear stress gradient stimulates cell proliferation, which is in turn balanced by cell loss in upstream and downstream regions where cells accumulate.
We have observed that steady separated flow conditions stimulate cell motility in confluent human umbilical vein endothelial cell monolayers and result in a net migration away from the region of large shear stress gradients. These observations thus provide an experimental validation of the hypothesis of DePaola et al13 that cell migration plays a key role in the dramatic remodeling that occurs in an endothelial monolayer when it is subjected to disturbed flow conditions in vitro. In addition, the present study suggests that a chronic cell proliferation-migration-loss cycle is occurring, as a localized phenomenon, in the disturbed flow region. Although disturbed flow conditions have been demonstrated to stimulate cell proliferation,13 18 the fact that a shear stress gradient can increase endothelial cell motility has not been demonstrated previously.
The ability to directly visualize the behavior of individual cells and groups of cells by live-time videomicroscopy, as they were subjected to disturbed flow in a specially designed parallel plate flow cuvette, facilitated our experimental analysis of the patterns of cellular responses elicited. In the vicinity of the reattachment line, cells were observed to proliferate and were much more motile than their counterparts in confluent monolayers maintained under static (no flow) conditions. With time, cell migration was seen to occur in the flow direction, away from the reattachment line. This directional movement was not typically observed under nondisturbed flow and may reflect a combination of factors such as the intrinsic greater motility of individual cells stimulated by the disturbed flow and the directional shear stress profile (negative shear stress upstream from the attachment line and positive shear stress on the downstream side). Finally, cells accumulated and then detached in the upstream and downstream regions where the shear stress gradients became negligible. A detailed analysis of the dynamic balance of cell loss and migratory rates would require more sophisticated experiments. However, the observed migration velocities and proliferation rates are self-consistent. In the present work, we had to find a compromise between using a lower magnification lens that provides a larger field of view and therefore more cell counts and a higher magnification that is more appropriate for cell tracking but requires a larger number of experiments.
The mechanism(s) by which endothelial cells can sense externally applied fluid shear stresses have not yet been fully defined, although activation of potassium channels and localized changes in cytosolic ionized calcium have been reported.16 19 20 Recent studies by Davies et al21 suggest that significant shear gradients generated over the irregular surface contour of individual endothelial cells, within a confluent monolayer, can be transduced into intracellular biochemical responses that are important in cell shape change and alignment. Conceivably, opposing tractive forces generated along the line of flow reattachment, acting at the level of the intercellular junctions between adjacent cells, might also contribute to the activation of cell motility observed in the current study.
In vivo, due to the pulsatile nature of blood flow, regions of flow separation are not as stable or localized as they were in our in vitro model. However, if such a cell proliferation-migration-loss cycle exists in vivo, endothelial cells located in regions of high shear stress gradients, eg, near arterial bifurcations, would be predicted to be undergoing more rapid turnover than elsewhere in the arterial system, resulting in episodic and localized changes in endothelial metabolic activity and intimal integrity. This is, in fact, consistent with in vivo findings of focally increased intimal permeability and increased cell cycle activity at these sites in experimental animal models of atherogenesis,22 23 as well as the occurrence of large, senescent-appearing intimal surface cells overlying atherosclerotic lesions in humans.24 Our in vitro findings thus suggest that a chronic endothelial cell proliferation-migration-loss cycle, stimulated by local shear stress gradients, may be a key factor in arterial wall remodeling under pathophysiologic conditions.7 9
We thank K. Case and M.W. Atkinson for valuable assistance in cell culture. This work was partially supported by the Swiss National Science Foundation and grant R37-HL51150 from the National Institutes of Health.
Fry DL. Acute vascular endothelial changes associated with increased blood velocity gradients. Circ Res.. 1968;22:165-197.
Davies PF. How do vascular endothelial cells respond to flow? NIPS.. 1989;4:22-25.
Fox JA, Hugh AE. Localization of atheroma: a theory based on boundary layer separation. Br Heart J.. 1966;26:388-399.
Caro CG, Fitz-Gerald JM, Schroter RC. Atheroma and arterial wall shear: observation, correlation and proposal of a shear dependent mass transfer mechanism for atherogenesis. Proc R Soc Lond.. 1969;177:109-159.
Zarins CK, Giddens DP, Bharadvaj BK, Sottiurai VS, Mabon RF, Glagov S. Carotid bifurcation atherosclerosis: quantitative correlation of plaque localization with flow velocity profiles and wall shear stress. Circ Res.. 1983;53:502-514.
Ku DN, Giddens DP, Zarins CK, Glagov S. Pulsatile flow and atherosclerosis in the human carotid bifurcation. Arteriosclerosis.. 1985;5:293-302.
Lutz RJ, Cannon JN, Bischoff KB, Dedrick RL, Stiles RK. Wall shear stress distribution in a model canine aorta during steady flow. Circ Res.. 1977;41:391-399.
Langille BL, Reidy MA, Kline L. Injury and repair of endothelium at sites of flow disturbances near abdominal aortic coarctations in rabbits. Arteriosclerosis.. 1986;6:146-154.
Levesque MJ, Liepsch D, Moravec S, Nerem RM. Correlation of endothelial cell shape and wall shear stress in a stenosed dog aorta. Arteriosclerosis.. 1986;6:220-229.
Kim DW, Gotlieb AI, Langille BL. In vivo modulation of endothelial F-actin microfilaments by experimental alterations in shear stress. Arteriosclerosis.. 1989;9:439-445.
DePaola N, Gimbrone MA Jr, Davies PF, Forbes CF Jr. Vascular endothelium responds to fluid shear stress gradients. Arterio Thromb Vasc Biol.. 1992;12:1254-1257.
Gimbrone MA Jr. Culture of vascular endothelium. In: Spaet T, ed. Progress in Hemostasis and Thrombosis. New York, NY: Grune & Stratton Inc, 1976:1-28.
Shen J, Luscinskas FW, Connolly A, Dewey CF Jr, Gimbrone MA Jr. Fluid shear stress modulates cytosolic free calcium in vascular endothelial cells. Am J Physiol.. 1992;262:C384–C390.
Maday Y, Patera AT. Spectral element methods for the Navier-Stokes equations. In: Noor AK, Oden JT, eds. State-of-the-Art Surveys on Computational Mechanics. New York, NY: American Society of Mechanical Engineering, 1989:71-143.
Davies PF, Remuzzi A, Gordon EJ, Dewey CF Jr, Gimbrone MA Jr. Turbulent fluid shear stress induces vascular endothelial cell turnover in vitro. Proc Natl Acad Sci U S A.. 1986;83:2114-2117.
Davies PF. Flow-mediated endothelial mechanotransduction. Physiol Rev.. 1995;75:519-560.