INTRODUCTION

Top Abstract Introduction Materials & Methods Results Discussion References

THE ADHESION OF MONOCYTES to arterial endothelium at lesion-prone sites and subsequent transmigration and formation of lipid-filled macrophages are early events in atherosclerosis (8, 15, 23, 36). The localization of monocyte binding may, in part, depend on the local fluid dynamics. Hemodynamics may affect the transport of monocytes directly to the endothelium and the subsequent adherence of monocytes to the endothelium. Hemodynamic behavior at arterial bifurcations may result in monocytes impinging against the endothelium, leading to increased monocyte adhesion at lesion-prone sites. Furthermore, the local fluid dynamics may produce focal upregulation of adhesion proteins (37).

In vitro investigations of the behavior of blood cells and microspheres in recirculating flow have demonstrated that fluid dynamics can affect cell adhesion to biologically inert surfaces and collagen-coated surfaces (16, 17, 29). For example, adhesion of platelets to collagen-coated glass in an annular vortex was highest within the vortex and downstream of the point of flow reattachment and minimal at the reattachment site (16, 17). U-937 cell rolling velocities on a silicone surface in recirculating flow in a sudden expansion varied linearly with wall shear stress, whereas particle residence times and cell adhesion varied inversely with wall shear stress (29).

Because monocytes play a major role in atherogenesis and the localization of atherosclerosis is linked to hemodynamic effects, investigations of monocytes interacting with endothelium in recirculating flow may provide insight into the factors that influence monocyte adhesion in vivo. The purpose of this investigation was to examine the effects of steady recirculating flow on the frequency of U-937 cell arrests to human umbilical vein endothelial cells (HUVEC) in a sudden-expansion flow chamber. Experimental variables included flow rate, concentration of U-937 cells, and HUVEC activation by tumor necrosis factor (TNF)-. Predictions of trajectories of spherical particles in recirculating flow under the experimental conditions were obtained using a computational model, and the simulations were compared with the experimental results.

	MATERIALS AND METHODS

Top Abstract Introduction Materials & Methods Results Discussion References

Endothelial cell culture. HUVEC were isolated by collagenase treatment of human umbilical cord veins (9, 14) and characterized as endothelial cells by acetylated low-density lipoprotein uptake, factor VIII expression, and cobblestone morphology. Cell cultures were maintained in medium 199 (M199) with Earle's salts (Sigma Chemical, St. Louis, MO) supplemented with 10% heat-inactivated fetal bovine serum (FBS) (Sigma), 1% antibiotic-antimycotic solution (100× stock) (Sigma), 2 mM L-glutamine, 120 µg/ml heparin (Sigma), and 100 µg/ml endothelial cell growth supplement (Collaborative Biomedical, Bedford, MA). Cells were grown in tissue culture flasks (Corning, Corning, NY) coated with 0.1% porcine gelatin (Sigma) in M199, and confluent monolayers were split 1:4 using 0.05% trypsin-EDTA (Sigma). For adhesion assays, HUVEC (passage 2-5) were plated on gelatin-coated glass microscope slides (Fisher Scientific, Medford, MA) and grown to confluency. After splitting, HUVEC reached confluency in 2-4 days.

(1)

View larger version (27K): [in this window] [in a new window]   Fig. 1.   Schematic of sudden-expansion flow chamber (not drawn to scale). H, chamber height downstream of expansion; h, chamber height upstream of expansion; S, gasket thickness; x, coordinate parallel to glass slide; y, coordinate normal to glass slide.

2

1

(2)

(3)

(4)

(5)

6

	RESULTS

Top Abstract Introduction Materials & Methods Results Discussion References

Characterization of flow through sudden expansion. Numerical simulations of two-dimensional steady flow through the sudden expansion with H/h = 11.4 were performed for flow rates of 12, 20, and 30 ml/min (Re = 24, 40, and 60, respectively). The sudden expansion produced recirculation regions with reattachment distances that varied, as a function of flow rate, between approximately 1,000 and 2,500 µm. Figure 2A shows predictions for the distributions of _w along the lower wall. Wall shear stresses ranged from 3 to 6 dyn/cm² within the recirculation region, with downstream wall shear stresses from 0.28 to 0.69 dyn/cm². In the aorta and large arteries, physiological mean wall shear stress magnitudes range from approximately 0.5 to 10 dyn/cm², and instantaneous values can vary from zero to as high as 200 dyn/cm² (27, 35). Mean physiological values of Re range from approximately 200 to 6,000 (27, 35). For the in vitro assays, we could not match exactly the physiological range of the wall shear stress and Re due to the excessive flow rates required. Figure 2B shows an expanded view of the distributions of wall shear stresses along the lower wall for the region in the vicinity of reattachment that was analyzed to quantify transient arrests exhibited by U-937 cells.

View larger version (26K): [in this window] [in a new window]   Fig. 2.   A: wall shear stress vs. distance from sudden expansion. B: wall shear stress vs. distance from reattachment. Re, Reynolds number.

U-937 cell behavior in recirculating flow. Upstream of the reattachment, U-937 cells became entrained in the recirculating flow. At the reattachment location, the paths of U-937 cells in the fluid flow diverged, with some U-937 cells traveling downstream toward the fully developed flow region and others traveling upstream into the recirculating flow region. U-937 cells exhibited transient arrests to the HUVEC monolayer immediately upstream and downstream of the reattachment. These transient arrests occurred as individual events, with no leukocyte-leukocyte interactions. U-937 cell rolling was not observed.

To demonstrate the observed transient arrest behavior, Fig. 3 shows three images, 0.4 s apart, in the reattachment region for TNF--activated HUVEC at a flow rate of 20 ml/min (Re = 40). The vertical dashed line in each image indicates the line of reattachment. The black arrow in each image indicates a U-937 cell that translated downstream from the reattachment line (Fig. 3A), transiently arrested ~48 µm downstream from the reattachment for 0.63 s (Fig. 3B), and then detached and accelerated downstream (Fig. 3C). The white arrow in each image indicates a U-937 cell that accelerated downstream away from the reattachment region without transiently arresting.

View larger version (96K): [in this window] [in a new window]   Fig. 3.   Three images, 0.4 s apart, in reattachment region for tumor necrosis factor (TNF)--activated human umbilical vein endothelial cells (HUVEC) at flow rate of 20 ml/min (Re = 40). Vertical dashed line in each image indicates line of reattachment. Black arrow indicates a U-937 cell that translated downstream from reattachment line (A), transiently arrested ~48 µm downstream from reattachment for 0.63 s (B), and then detached and accelerated downstream (C). White arrow in A-C indicates a U-937 cell that accelerated downstream away from reattachment region without transiently arresting.

From a sample of 145 transient arrests at all of the experimental conditions studied, 126 (87%) occurred within 150 µm of the reattachment line. Of these 145 arrests, 120 (83%) occurred at least 30 µm (approximately 1 or 2 endothelial cell widths) away from the measured reattachment line. Thus few arrests occurred directly on the line of reattachment, where the wall shear stress is zero. Of this sample of 145 arrests, 74 occurred upstream of the reattachment line and 71 occurred downstream of the reattachment line. Therefore, the cells did not exhibit a preference for arresting upstream or downstream of the reattachment line.

Within the recirculation zone upstream of the reattachment region, U-937 cells exhibited very few transient arrests and no rolling behavior. Instead, U-937 cells translated across the endothelial monolayer, accelerating as they moved upstream from the reattachment toward the recirculation zone where the wall shear stress was a maximum, until they attained relatively high translational velocities. As the U-937 cells approached the expansion site, they moved away from the endothelial monolayer and became entrained in the recirculating flow.

Frequencies of cell arrests in the vicinity of reattachment. Results from in vitro recirculating flow experiments indicated that U-937 cells exhibited significantly higher arrest frequencies in the vicinity of reattachment to TNF--activated HUVEC compared with unactivated HUVEC (Fig. 4). These higher arrest frequencies occurred at both U-937 cell concentrations and all three flow rates except for the condition of 10⁵ U-937 cells/ml at a flow rate of 20 ml/min. Increasing the U-937 cell concentration from 10⁵ to 10⁶ cells/ml resulted in a roughly two- to threefold increase in arrest frequencies to TNF--activated HUVEC, but no increase occurred for unactivated HUVEC. For TNF--activated HUVEC at a concentration of 10⁶ U-937 cells/ml, the cell arrest frequency at a flow rate of 20 ml/min was significantly higher than at 12 or 30 ml/min. However, for TNF--activated HUVEC at a concentration of 10⁵ U-937 cells/ml, flow rate did not significantly affect arrest frequencies. For unactivated HUVEC, flow rate did not affect arrest frequencies.

View larger version (20K): [in this window] [in a new window]   Fig. 4.   Arrest frequencies vs. flow rate for unactivated and TNF--activated HUVEC at U-937 cell concentrations of 105 and 106 cells/ml. * P < 0.05 vs. unactivated HUVEC at same cell density and flow rate. ** P < 0.05 vs. TNF--activated HUVEC at same flow rate and 105 U-937 cells/ml. *** P < 0.05 vs. TNF--activated HUVEC at flow rate = 12 and 30 ml/min and 106 U-937 cells/ml.

Predicted U-937 cell trajectories in recirculating flow. Using results from the numerical simulations for the flow field, U-937 cell trajectories were calculated at each experimental flow rate by assuming that the U-937 cell is a spherical particle. Figure 5A shows predicted trajectories at 30 ml/min for particles seeded 0.1 cm upstream of the sudden expansion at distances of 7.75-16.75 µm above the lower surface of the inlet. At each flow rate of 12, 20, and 30 ml/min, the simulations predicted that none of the particles became entrained in the recirculating flow or encountered the lower wall of the flow chamber. Instead, all of the particles traveled downstream beyond the reattachment point.

View larger version (22K): [in this window] [in a new window]   Fig. 5.   Predicted particle trajectories in sudden expansion for a flow rate of 30 ml/min. A: particles seeded 0.1 cm upstream of sudden expansion. B: particles seeded within recirculation zone upstream and downstream of reattachment point. C: particles seeded near reattachment point on adjacent streamlines.

Experimental results clearly showed that U-937 cells transiently arrested to the endothelial monolayer in the vicinity of reattachment and also became entrained in the recirculating flow. Therefore, in the numerical simulations, particles were also seeded within the recirculation zone near the reattachment point on adjacent streamlines that diverged at the reattachment site, to determine how closely the particles approached the lower surface. Figure 5, B and C, shows examples of the resulting particle trajectories at 30 ml/min. Particles seeded on streamlines that curved back upstream became entrained in the recirculating flow. These particles gradually spiraled outward and left the recirculation zone (Fig. 5B). This outward migration is similar to experimental observations by Karino and Goldsmith (17) for the motions of red blood cells, platelets, and latex spheres in an axisymmetric annular expansion. For each flow rate, particles approached no closer than ~10 µm to the lower wall (Fig. 5C). This was true for all flow rates studied. Because typical bond lengths are several orders of magnitude smaller than this approach distance (1), none of these particles in the simulations came close enough to the lower wall to allow bond formation. Thus the numerical simulations of U-937 cell trajectories in flow through the sudden expansion predicted that U-937 cells treated as spherical particles do not become entrained in the recirculating flow when seeded upstream of the expansion. Spherical particles seeded near the reattachment on diverging streamlines did not contact the lower wall in the flow chamber.

Median arrest duration. As expected, increasing the U-937 cell density from 10⁵ to 10⁶ cells/ml did not cause a significant change in arrest duration. Despite a significant increase in U-937 cell attachment (Fig. 4), TNF- activation did not cause significant changes in median arrest duration for U-937 cell transient arrests in the vicinity of reattachment (Fig. 6A). For TNF--activated HUVEC at both 10⁵ and 10⁶ cells/ml, the median arrest duration at a flow rate of 30 ml/min was significantly less than the median arrest duration at 12 ml/min (P = 0.013 for 10⁵ cells/ml and P = 0.037 for 10⁶ cells/ml). Flow rate did not affect the median arrest durations for unactivated HUVEC.

View larger version (29K): [in this window] [in a new window]   Fig. 6.   A: median arrest duration vs. flow rate for unactivated and TNF--activated HUVEC at U-937 cell concentrations of 105 and 106 cells/ml. B: shear flow median arrest durations for TNF--activated HUVEC plotted vs. wall shear stress (w) for shear flow compared with recirculating flow median arrest durations for TNF--activated HUVEC plotted vs. root mean square value of wall shear stress (w-rms).

For TNF--treated HUVEC in shear flow, U-937 cells exhibited a shear stress-dependent tethering behavior (3). At shear stresses of 0.043-0.172 dyn/cm², U-937 cells exhibited transient cell arrests. The fractions of U-937 cells that arrest and the frequencies of U-937 cell arrests decreased with increasing shear stresses. However, at shear stresses > 0.172 dyn/cm², U-937 cells exhibited rolling across the endothelium in an erratic fashion at nonuniform velocities that were much lower than the hydrodynamic velocity, with very few transient arrests.

Although the wall shear stress does not vary across the field of view in shear flow, large shear stress gradients occur in the reattachment region in recirculating flow, with shear stress changing from positive to negative at the reattachment location. To compare the median arrest durations in shear flow with the median arrest durations for the transient arrests in recirculating flow, a root-mean-square (rms) value of the wall shear stress, _w-rms, was calculated for each recirculating flow condition from the wall shear stress distributions in the vicinity of reattachment (shown in Fig. 2B)

(6)

The value of _w-rms accounts for differences in the sign of the shear stress and represents the average magnitude of the shear stress exerted on U-937 cells near flow reattachment. As Fig. 6B shows for TNF--treated HUVEC, the recirculating flow median arrest durations at _w-rms = 0.162 dyn/cm² were not significantly different from the shear flow median arrest durations at shear stresses of 0.129-0.172 dyn/cm². However, the recirculating flow median arrest durations decreased by <50% as _w-rms increased by over a factor of 10. Furthermore, under shear flow very few transient arrests were seen at shear stresses >0.172 dyn/cm², indicated by the dashed line in Fig. 6B.

Although the recirculating flow median arrest duration at _w-rms = 0.162 dyn/cm² was not significantly different from the shear flow median arrest duration at a shear stress of 0.172 dyn/cm², the fractions of cells remaining bound vs. time after initiation of arrest, calculated from measured arrest durations, are very different (Fig. 7). In shear flow, the cell arrest durations are best fit by a biexponential model (indicated by the curved line in Fig. 7), which suggests that cells bind by a single bond to two classes of receptors with very different dissociation constants (3). However, in recirculating flow, the cell arrest durations are best fit by a single exponential model (indicated by the straight lines in Fig. 7), which suggests that the cells bind to a single receptor (1). This difference in bond lifetimes for cells adherent in shear flow and near flow reattachment suggests that receptors that detach slowly cannot form near flow reattachment.

View larger version (18K): [in this window] [in a new window]   Fig. 7.   Fraction of U-937 cells remaining bound as a function of time after initiation of arrest, calculated from measured arrest durations for unactivated HUVEC in recirculating flow at w-rms = 0.162 dyn/cm2, TNF--activated HUVEC in recirculating flow at w-rms = 0.162 dyn/cm2, and TNF--activated HUVEC in shear flow at w = 0.172 dyn/cm2. The 2 straight lines represent fits of single exponential model to cell arrest durations for each recirculating flow case, and curved line represents fit of biexponential model to cell arrest durations for shear flow case.

	DISCUSSION

Top Abstract Introduction Materials & Methods Results Discussion References

This study represents the first report of monocytic cell adhesion to activated endothelium in recirculating flow conditions similar to mean flow conditions in arteries. The factors that govern U-937 cell adhesion to HUVEC in recirculating flow include transport effects such as diffusion and convection, the intrinsic forward and reverse rate coefficients for bond formation and dissociation, the numbers and types of receptors expressed by the HUVEC and U-937 cells, and the local wall shear rate .

Unactivated HUVEC constitutively express low levels of intercellular adhesion molecule-1 (ICAM-1) (4, 18, 39). TNF- treatment induces HUVEC to upregulate expression of vascular cell adhesion molecule-1 (VCAM-1), ICAM-1, E-selectin, and P-selectin (20, 21, 31). U-937 cells express the 1 and 2 integrins very late antigen-4 (VLA-4) and lymphocyte function-associated antigen-1 (LFA-1), as well as sialyl Lewis^x counterreceptors to E-selectin and P-selectin (28). However, U-937 cells lack L-selectin (22) and may not express P-selectin glycoprotein ligand-1 (PSGL-1) as a counterreceptor to P-selectin. Unfortunately, we could not use isolated blood monocytes in our cell adhesion assays. Each experiment required between 12 million and 120 million suspended cells because of the high flow rates necessary to create the recirculating flow. We selected the U-937 cell as the most appropriate monocytic cell line to use in our experiments. Due to differences in types and numbers of receptors expressed between U-937 cells and monocytes, isolated monocytes may yield results different from those we obtained using U-937 cells.

In shear flow, the predominant pattern of monocyte adhesion is immediate arrest on initial contact with the endothelium, with a few monocytes rolling before and after firm arrest (19, 21). Previous studies of monocyte adhesion to TNF- or IL-4-activated HUVEC have implicated L-selectin as the major molecule that mediates initial arrest and rolling under flow (21, 22). However, a more recent study of monocyte adhesion to IL-1-activated HUVEC showed that monocytes may adhere via three independent pathways: 1) L-selectin, 2) VLA-4/VCAM-1, and 3) a sialyl-Lewis^x pathway that may involve P-selectin, E-selectin, or some unidentified receptor (19). Thus U-937 cells may bind to HUVEC via several different adhesion molecules under recirculating flow conditions. In support of this, our recent shear flow assays indicate that U-937 cell transient arrests to unactivated HUVEC can be blocked by monoclonal antibodies (MAb) to ICAM-1, whereas U-937 cell transient arrests to TNF--activated HUVEC can be blocked by MAb to VCAM-1, ICAM-1 and E-selectin (3). Flow cytometry also revealed the presence of these adhesion molecules after TNF- treatment of the endothelium. The absence of L-selectin may not be critical for U-937 cell arrests to occur, but it may affect the numbers of cells that arrest.

The complex effects of fluid transport and receptor-ligand binding on U-937 cell adhesion can be understood in terms of a dimensionless forward reaction rate (12, 13, 33)

(7)

where N_R is the endothelial surface receptor density and k_f is the overall forward reaction rate coefficient, assuming that the number of counterreceptors on the U-937 cell is very large. The coefficient k_f depends on fluid transport effects and the intrinsic rate constants for bond formation. The parameter  is a measure of the ratio of the contact time to the time for binding. If  >> 1, then the probability of binding is very high, whereas if  << 1, then the binding probability is very small. In shear flow, the adhesion of rat basophilic leukemia cells to antigen-coated substrates increased as the shear rate decreased, and this increased adhesion scales with  in the simple diffusion-limited case (33). Expressions for diffusion-limited values of k_f exist for the shear flow case; however, recirculating flow is more complicated than shear flow because of the functional dependence of k_f on the flow field (33).

The parameter  is sensitive to adhesion receptor levels on endothelial cells, flow rate, and location of flow reattachment. TNF- treatment induces HUVEC to upregulate expression of adhesion receptors to which U-937 cells may bind. This results in an increase in N_R, due to the presence of more receptors on the HUVEC monolayer, increasing the value of  and causing a higher frequency of arrests compared with unactivated HUVEC.

The frequencies of U-937 cell arrests (Fig. 4) show the effect of increasing flow rate on the magnitude of . We speculate that increasing the flow rate causes more rapid delivery of U-937 cells to the endothelial monolayer. As a result, k_f increases as the adhesion shifts from transport limited to reaction limited, and  increases. At the highest flow rate, k_f reaches a maximum while continues to increase, so that  decreases. The overall effect is an increase and then a decrease in  as the flow rate increases, resulting in the variation in the frequency of U-937 cell arrests to TNF--activated HUVEC at 10⁶ U-937 cells/ml (Fig. 4).

In addition to the numbers and types of receptors expressed by the endothelium, U-937 cell adhesion depends on the nature of the flow itself. U-937 cells transiently arrested within 150 µm upstream and downstream of the reattachment line, but few U-937 cells arrested directly at the reattachment line. This is similar to previous studies of platelet adhesion to collagen-coated glass in an annular vortex (16) and U-937 cell adhesion to a silicone wall in a sudden expansion (29). At the reattachment location, the shear rate is zero and the particle residence time is high. However, few cells may arrest due to a very low flux of U-937 cells along fluid streamlines with low velocities that reach the reattachment location (16, 17), resulting in a low overall forward reaction rate k_f and a low value of . Immediately upstream and downstream of the reattachment site, increases but the flux of cells to the endothelium also increases so that  reaches a maximum, resulting in cell arrests. Farther away from the reattachment region within the recirculation zone,  is small because of higher values of ; thus very little binding occurs.

At shear rates >0.172 dyn/cm², few measurable transient arrests of U-937 cells occur in shear flow (3). Nevertheless, U-937 cells adhere near the reattachment point in recirculating flow. Unlike those in shear flow, U-937 cells in recirculating flow are exposed to fluid velocity components normal to the endothelium. As a result, a U-937 cell may impact the endothelium with a sufficiently greater force that results in compression of microvilli with a larger contact area between the U-937 cell and the endothelium. This larger contact area may increase the chance of forming a bond before the cell moves away. Such interactions may be brief, further influencing the types of adhesion molecules that bind, as well as the kinetics of binding (Fig. 7). Although multiple receptors are present on HUVEC after TNF- activation, the forces and contact duration affect which actually form bonds near flow reattachment.

Computer simulations of spherical particle trajectories within the recirculating flow through the sudden expansion predicted that rigid spheres do not enter the recirculation region or contact the lower wall of the sudden expansion. However, experimental results demonstrated that U-937 cells transiently adhere to the endothelium in the vicinity of the reattachment site. Thus the numerical model cannot accurately simulate this result from the experiments. However, the numerical model did simulate trajectories of particles that gradually spiraled outward and left the recirculation zone (Fig. 5B), which agrees with experimental observations for the motions of particles in an axisymmetric annular expansion (17).

Major assumptions of the computational model include treating the U-937 cells as rigid spherical particles with no microvilli, neglecting cell-cell collisions, and neglecting any disturbances in the local flow field that the cells may cause. These assumptions may not be valid. Cell deformability, microvilli, and aspherical cell shape may cause the observed deviations between the computational predictions of cell trajectories and the experimental results. Inclusion of these effects will require solving the coupled motion of the particle and fluid, a more complex problem than the one addressed in this study. An additional experimental condition not included in the model is the wavy surface of the endothelial monolayer, where variations in height as much as 4-5 µm exist (2). Variations in the local flow over this wavy surface, compared with a flat surface modeled in the simulations, may also account for the experimental deviations from the model predictions.

In vivo, convective transport of monocytes to the vessel walls along curved streamlines that have large radial velocity components may be partly responsible for the flux of monocytes to the walls. In addition, monocyte heterogeneities and variations in cell size and shape, cell-cell collisions, and pulsatile flow can affect transport of monocytes to the wall. Collisions between monocytes and red blood cells are an important aspect due to the high concentrations of red blood cells that are present at physiological hematocrit levels in whole blood. As demonstrated in vitro, leukocyte collisions with red blood cells may enhance transport of leukocytes to the endothelium, due to radial dispersion of leukocytes (24-26, 35). This effect can increase the frequency of encounters between monocytes and vessel walls.

The mechanisms outlined here that affect the localization of U-937 cell transient arrests in these experiments may also play a role in monocyte adhesion in arteries. Our results suggest that, in addition to shear stress-induced alterations in endothelial cell function, fluid dynamics may also affect atherosclerotic lesion localization by influencing monocyte transport and adhesion to the vessel wall. Monocytes that encounter the vessel wall near a stagnation or reattachment point where the wall shear stress is low may exhibit transient arrests, which could lead to long-term arrest and subsequent transmigration to the intima.