Dynamic Adhesion of Umbilical Cord Blood Endothelial Progenitor Cells under Laminar Shear Stress

Abstract

Late outgrowth endothelial progenitor cells (EPCs) represent a promising cell source for rapid reendothelialization of damaged vasculature after expansion ex vivo and injection into the bloodstream. We characterized the dynamic adhesion of umbilical-cord-blood-derived EPCs (CB-EPCs) to surfaces coated with fibronectin. CB-EPC solution density affected the number of adherent cells and larger cells preferentially adhered at lower cell densities. The number of adherent cells varied with shear stress, with the maximum number of adherent cells and the shear stress at maximum adhesion depending upon fluid viscosity. CB-EPCs underwent limited rolling, transiently tethering for short distances before firm arrest. Immediately before arrest, the instantaneous velocity decreased independent of shear stress. A dimensional analysis indicated that adhesion was a function of the net force on the cells, the ratio of cell diffusion to sliding speed, and molecular diffusivity. Adhesion was not limited by the settling rate and was highly specific to α₅β₁ integrin. Total internal reflection fluorescence microscopy showed that CB-EPCs produced multiple contacts of α₅β₁ with the surface and the contact area grew during the first 20 min of attachment. These results demonstrate that CB-EPC adhesion from blood can occur under physiological levels of shear stress.

Introduction

Although a number of therapeutic procedures, such as bypass grafts or balloon angioplasty, limit the clinical consequences of atherosclerosis, the endothelium is often damaged (1–4). Endothelial cell (EC) regrowth often occurs too slowly to prevent platelet adhesion and smooth muscle cell proliferation, resulting in intimal thickening and reduced blood flow (5). Drug eluting stents reduce neointimal hyperplasia by blocking smooth muscle cell proliferation; however, EC proliferation is also blocked, slowing the endogenous repair process and increasing the risk of late stage thrombosis (6,7).

Endothelial progenitor cells (EPCs) represent a promising approach to promote vessel reendothelialization and limit thrombosis and neointimal hyperplasia (8–11). Because EPC concentration in adult blood is extremely low (<10 cells/mL), EPCs may need to be grown ex vivo and high concentrations of cells may need to be injected locally into the artery to increase the probability of successful adhesion and subsequent coverage of the damaged region (12). Human leukocyte antigen-matched late outgrowth umbilical cord blood derived EPCs (CB-EPCs) have the potential to be used for regeneration of damaged arteries or vein grafts because they are readily obtainable (13,14), easily transplantable (15,16), possess a high proliferative potential (13,14), have strong adhesion to underlying surfaces (17,18), and are similar to endogenous ECs in their antithrombotic gene expression (17).

Although CB-EPCs respond similarly to native ECs when plated at confluence under static conditions and introduced to high shear stresses (17), the initial cell capture of CB-EPCs from a flowing solution to the underlying surface in a direct-injection process has not yet been characterized. Studies with white blood cells have shown that adhesion involves a process of capture, rolling, and arrest involving selectins and integrins that is dependent upon both the fluid dynamics (19,20) and the kinetics of receptor-ligand binding (21–23).

For direct injection of CB-EPCs to reendothelialize damaged vasculature, CB-EPCs must:

• 1.Transport through the fluid to extracellular matrix receptor ligands.
• 2.Locally orient adhesion receptors while convection causes sliding of the receptors past the ligands.
• 3.Adhere by cell arrest with or without prior rolling to reduce velocity.
• 4.Cover sufficient surface of the damaged artery to limit the time for cell proliferation to result in total surface coverage.
• 5.Not detach from the arterial surface at high, physiological shear stresses shortly after initial adhesion.

Although several studies establish that ECs adhere from flowing solutions onto fibronectin (FN) and other ligands (23–25), each study only examined a single shear stress and did not establish the mechanism of adhesion. In this study, we used a parallel plate flow chamber to examine dynamic adhesion of CB-EPCs to FN, an adhesion protein adsorbed in high quantities to the exposed intima after arterial injury. We determined the relative contribution of transport and binding events to adhesion, the integrins involved in adhesion, and whether CB-EPCs underwent rolling before arrest. Total internal reflection fluorescence microscopy was used to determine the separation distance of integrins from the surface. The results show that FN can support the capture and arrest of CB-EPCs adhering from a flowing solution and that CB-EPCs make multiple contacts with the surface shortly after attachment.

Materials and Methods

Expanded methods are provided in the Supporting Material.

CB-EPC isolation and cell culture

CB-EPCs were isolated from human umbilical cord blood as previously described (13). Umbilical cord blood samples were obtained from the Carolina Cord Blood Bank at Duke University per protocols approved by the Duke University Institutional Review Board. Before receipt, all patient identifiers were removed. The use of these blood samples are exempt from Human Subjects approval as defined by 45 CFR 46.102(f) and are not subject to the Privacy Rule (45 CFR 164.500(a)). CB-EPCs were characterized and cultured as previously described (17).

Before all flow experiments, cells were incubated with either Cell Tracker Orange or Green (2 μM; Invitrogen, Grand Island, NY) in serum-free quiescent media (DMEM/F12, 1× Insulin-Transferrin-Selenium, 1× Antibiotic/Antimycotic Solution; Gibco, Carlsbad, CA) for 20 min at 37°C. Cells were removed from the tissue culture plastic with 0.025% trypsin-ethylenediaminetetraacetic acid (Gibco) for 5 min at 37°C, neutralized with trypsin neutralizing solution (Gibco), and resuspended with Dulbecco's phosphate buffered saline (DPBS; Sigma, St. Louis, MO). CB-EPCs were used at passages 4–8 for all experiments and there was no effect of CB-EPC passage number on adhesion.

Initial cell capture experiments

Based upon preliminary experiments to identify the concentration at which FN adsorption was highest (Fig. S1 in the Supporting Material), FN (Sigma) was added at 20 μg/mL to DPBS and incubated on a polystyrene SlideFlask (area = 9 cm²; NUNC, Rochester, NY) for 1 h at 37°C. After incubation, the slide was rinsed twice with DPBS and inserted into the glass recess of a parallel plate flow chamber, as previously described (26). A precut silicone rubber gasket served to establish the flow path and set the chamber height, h, and the width, w (0.5 cm). Chamber heights were either 254 ± 2 μm or 508 ± 3 μm. Assembled flow chambers were incubated at 37°C before use.

Flow media consisted of DPBS containing high-molecular mass dextran (2 × 10⁶ kDa; Sigma) to vary the viscosity (1.0–3.0 cP). Dextran does not affect adhesion molecule surface expression (21). Viscosity was measured with a falling ball viscometer (Gilmont, St. Louis, MO) at 37°C. Dextran solutions below 10% (w/v) behaved as Newtonian fluids at 37°C. For a rectangular parallel plate flow chamber, the shear rate is (27)

$$\stackrel{·}{\gamma }=\frac{6Q}{w{h}^2},$$ (1)

and the wall shear stress, τ, is related to the dynamic viscosity, μ, by

$$\tau =\stackrel{·}{\gamma }\mu .$$ (2)

Shear rate was altered by varying the flow rate, Q, and the wall shear stress was modified by changing Q, μ, or h.

The flow media was contained in a reservoir held at 37°C by heat lamps and mixed with a stir bar to ensure adequate distribution of cells in solution. A syringe mounted on a syringe pump (Harvard Apparatus, Holliston, MA) was used to draw the flow media across the chamber. The chamber was placed on an inverted epi-fluorescence microscope (Zeiss Axiovert S100; Carl Zeiss, Thornwood, NY) and images of the adherent cells were taken every 2.5 min for up to 20 min at the same four locations near the channel centerline, 11 cm from the chamber inlet point. All images were analyzed using ImageJ software (Ver. 1.6, National Institutes of Health, Bethesda, MD).

Cell velocity

CB-EPC motion was recorded on videotape at a frame rate of 1/30 s using a video camera with a time-date generator (Panasonic, Secaucus, NJ). To determine instantaneous velocity, a single cell in the focal plane of the FN substrate was tracked across the field of view. The position of the cell centroid was measured at every frame over a linear distance of the viewing field, 1306 μm. Instantaneous velocity was determined by dividing the change in cell displacement between each frame by the time interval between frames (0.033 s). The translational velocity was determined as the time for the cell centroid of 50–100 randomly chosen cells to translocate across the field of view. Cells that collided with adherent cells or were tethered to each other during settling were omitted from the analysis. Measurements were taken during the first 1–2.5 min after onset of laminar flow. A cell was considered firmly arrested if it did not move >3 μm from its fixed position in 1 s.

Integrin blocking and flow cytometry

To identify integrins involved in dynamic adhesion of CB-EPCs to FN, cells were incubated with either mouse-anti-α₅β₁ (1:1000, Sigma), mouse-anti-α_Vβ₃ (1:500, Sigma), both, or neither (positive control) for 1 h before injection over the FN ligand. A cell density of 50,000 cells/mL was used and flow at 1.0 dyn/cm², 1 cP viscosity was conducted for 10 min to minimize use of antibody. Static controls of similar treatment were also tested, in which 200,000 cells/mL were plated for 10 min at 37°C. The number of α₅β₁ and α_Vβ₃ integrins present on ECs was quantified using flow cytometry and Quantum Simply Cellular site density calibration standards (Bangs Laboratory, Fishers, IN) as previously described (28).

Statistical analysis

The software package JMP 8 (SAS, Cary, NC) was used to calculate p-values. ANOVA and Tukey-Kramer post hoc tests were implemented to assess significance in all experiments where appropriate (p < 0.05) and data are reported as mean ± standard error, unless otherwise noted.

Results

CB-EPC solution density significantly affects quantity and size of adherent cells

The concentration of CB-EPCs in media with 1.0 cP viscosity was varied from 0.5 × 10⁵ to 10 × 10⁵ cells/mL and adhesion was measured for 20 min at a wall shear stress of 1.0 dyn/cm² (Fig. 1 A). Cell adhesion was significantly lower for all cell solution densities relative to 10 × 10⁵ cells/mL (p < 0.05); cells at 0.5 × 10⁵ cells/mL, and 1.0 × 10⁵ cells/mL experienced significantly less adhesion compared to 5.0 × 10⁵ cells/mL (p < 0.05). For all cell concentrations, the greatest rate of increase in cellular adhesion to FN occurred during the first 10 min after onset of laminar flow. Between 10 and 20 min after the onset of flow, the adhesion rate per unit area increased more slowly with increasing cell density (Fig. 1 B).

Figure 1.

Effect of cell solution density on CB-EPC adhesion to FN-coated polystyrene slides during dynamic adhesion at 1.0 dyn/cm² shear stress. (A) CB-EPC adhesion at 10 × 10⁵ cells/mL is significantly higher than at all other cell densities (^#p < 0.05, repeated measures ANOVA). At lower cell densities, CB-EPC adhesion increases linearly with time and is significantly reduced relative to 5.0 × 10⁵ cells/mL (^∗p < 0.05, repeated measures ANOVA). (B) CB-EPC adhesion nonlinearly increases after 20 min of exposure over FN as a function of cell density (^∗p < 0.05, relative to 0.5 × 10⁵ cells/mL). (C) Percent coverage of FN-coated surface exposed to CB-EPCs after 20 min of flow increases nonlinearly with cell solution density, while, the projected area of adherent cells and fluid force acting on adherent cells (D) decreases nonlinearly (^∗∗p < 0.01 relative to 0.5 × 10⁵ cells/mL, ^∗p < 0.05 relative to 0.5 × 10⁵ cells/mL, and ^#p < 0.05 relative to 1.0 × 10⁵ cells/mL, n = 3–7, mean ± SE).

The percent cell coverage of the surface (Fig. 1 C) was calculated based on the adherent cell area 20 min after the onset of flow. Coverage increased nonlinearly with cell density and was <10% at the highest cell density analyzed. The size of adherent cells decreased significantly with increasing cell density, which was associated with a corresponding decrease in the force on the adherent cells (Fig. 1 D, p < 0.05). Even at the highest concentration, the cell area of attached cells is >1 SD than the cell area, 222 ± 84 μm², based on the radius of suspended cells, 8.4 ± 1.6 μm (mean ± SD, n = 381). The cell radius distribution was heterogeneous with radii ranging from 5.6 μm to 11.8 μm (Fig. S2). The cell areas of adherent cells are greater than the cross-sectional area of suspended cells, suggesting that the cell size arises from preferential attachment of larger cells at lower cell densities and cell spreading. Based on this data, a cell solution density of 5.0 × 10⁵ cells/mL was used in all subsequent experiments.

CB-EPC adhesion does not simply depend on shear stress or shear rate

To determine the effect of shear stress and shear rate upon cell adhesion to adsorbed FN, CB-EPCs were infused through the flow chamber for 20 min at different flow rates and viscosities and the number of adherent cells per unit area was plotted versus shear rate (Fig. 2 A) or shear stress (Fig. 2 B). For either case, the number of adherent CB-EPCs per cm² exhibited a biphasic response that is similar to the trend observed for rolling fluxes and tethering rates of both neutrophils and monocytes to the endothelium or L-selectin (19,29,30). At higher shear stresses (≥1.5 dyn/cm²), the adhesion curves at different fluid viscosities did collapse onto a single curve, suggesting that the force on cells affects net adhesion at higher shear stresses. When the number of adherent cells was plotted as a function of calculated net force (Fig. 2 B), the trend was similar to that observed with shear stress. Because the adhesion data did not fall onto a single curve for shear stress, shear rate, or force, adhesion is likely influenced by several variables.

Figure 2.

Effect of shear rate (A) and shear stress (B) on CB-EPC adhesion. CB-EPCs exhibit a maximum in adhesion at 100 s⁻¹, 50 s⁻¹, and 33 s⁻¹ for 1 cP, 2 cP, and 3 cP, respectively (^∗p < 0.05, relative to maxima at 50 and 33 s⁻¹). When adhesion is plotted versus shear stress, the maximum is near 1.0 dyn/cm² for 1 and 2 cP fluid, but decreases to 0.75 dyn/cm² for 3 cP (^∗p < 0.05, relative maxima at 2 and 3 cP). A similar trend is observed when CB-EPC adhesion is plotted versus net hydrodynamic cell force. The declining portions of each viscosity curve align above 1.5 dyn/cm² (n = 3–7, mean ± SE).

CB-EPC binding rates suggest adhesion is not limited by the rate of cell settling

There are two distinct mechanisms of cell capture under flow conditions—transport and reaction. According to Yago et al. (19), transport-limiting mechanisms predominate when reaction is very fast relative to the timescale for the receptor on the cell surface to reach the reactive site of the ligand and increased cell flux to the surface causes an increase in cell tethering probability. Reaction-limiting mechanisms dominate when transport occurs rapidly, such that an adhesive receptor on the cell membrane quickly docks with its extracellular matrix ligand. Transport processes can be divided into global and local processes. The global processes are those that bring the cell close to the reactive surface and include settling and convective transport. Local transport processes are those that affect alignment of the cell receptor with its binding site, and include cell diffusion, sliding of the cell, and molecular diffusivity, as shown schematically in Fig. 3 A (19).

Figure 3.

Effect of global transport processes on CB-EPC adhesion. (A) Schematic of various transport processes affecting cell adhesion. A cell of radius R settles under gravity with a velocity V_s, given by Eq. S4 in the Supporting Material. Near the surface, the fluid velocity is proportional to the shear rate multiplied by the height above the surface and the cell translates with a velocity u (Eq. S7 in the Supporting Material). Due to rotation, there is a sliding velocity (u − ΩR), which scales as $R\stackrel{·}{\gamma }$ (19). The cell undergoes random rotational motion with a diffusion coefficient D_s (Eq. S8 in the Supporting Material) and the portion of the integrin receptor protruding into the fluid undergoes random fluctuations with a diffusion coefficient D_m. (Bolded, global; nonbolded, local). (B) The rates of cell binding (R_B) were normalized by the relative rate of settling (ΨV_s) for each shear stress and viscosity combination. A ratio close to 1 represents transport-limited adhesion, whereas a ratio closer to 0 represents reaction-limited adhesion. All ratios are <0.5 and, aside from the outlier, at 2.5 dyn/cm², 3.0 cp, are localized near 0.25, which indicates that maximal adhesion is not transport-limited. (C) CB-EPC adhesion was normalized by the dimensionless cell concentration (Ψ/C) and plotted as a function of shear stress for each viscosity (^∗p < 0.05, relative to 2.0 and 3.0 cP at 1.0 dyn/cm²). Although there was some alignment of the data, there was no direct correlation of normalized adhesion data when maximal adhesion values were plotted against ($R\stackrel{·}{\gamma }$)/ΨV_s (D), which represents the ratio of sliding velocity to settling (n = 3–7, mean ± SE).

To assess the relative contribution of global transport processes, the apparent cell-binding rate was calculated by a linear regression of the number of adherent cells between 0 and 10 min (the linear portion of Fig. 1 A for 5.0 × 10⁵ cell/mL) for each shear stress and viscosity condition and compared to the computed settling transport rate, ΨV_S (see Eq. S4 and Eq. S6 in the Supporting Material). A ratio <<1 indicates that the adhesion rate is dependent solely on local transport processes and reaction, whereas a ratio close to 1 indicate that the rate of settling limited the adhesion. The magnitude of the calculated ratio of the cell binding/transport rate (Fig. 3 B) was always <0.5 and different trends were observed with each viscosity, suggesting that the rate of settling is not responsible for the local adhesion maxima seen in Fig. 2 B.

To further assess the influence of settling upon the number of adherent cells, raw adhesion values for each shear stress and viscosity condition tested in Fig. 2 B were normalized by the dimensionless cell flux (see Eq. S6 in the Supporting Material and Fig. 3 C). Normalizing by cell flux aligned both the rising portion of the CB-EPC adhesion curve and the local maximum for each media viscosity. For each viscosity, normalized adhesion reached its maximum value at 1.0 dyn/cm², with the 1 cP condition yielding significantly higher adhesion at 2.0 and 3.0 cP (p < 0.05). Despite alignment of the three maxima to a single shear stress, cell flux alone does not appear to be the only parameter regulating flow-enhanced adhesion because the three curves did not align at each shear stress. This was verified by plotting the raw number of adherent cells as a function of the ratio $R\stackrel{·}{\gamma }$/(ΨV_s) (Fig. 3 D), where $R\stackrel{·}{\gamma }$ represents the mean sliding velocity (19). A lack of linear correlation suggests that cell settling is not a dominant mechanism regulating CB-EPC adhesion at higher shear rates/stresses. Further, the measured rate of cell binding is slow relative to the convective flux of cells at the substrate surface (Fig. S3).

CB-EPC adhesion to FN ligand is regulated by cell sliding, cell diffusion, and molecular diffusion

The shear rate at which adhesion reached a maximum declined with increasing viscosity (Fig. 2 A), further supporting the hypothesis that the net force on the cells was partly responsible for the maximum and the declining portion of the adhesion curves at higher shear stresses. The decrease in the maximum number of adherent cells per unit area with increasing viscosity suggested that the local transport processes also affect adhesion. Because D_s, the cell diffusivity (see Eq. S8 in the Supporting Material), is inversely related to the media viscosity, higher viscosities would result in decreased cell-substrate collisions and, consequently, a decrease in the probability of cell tethering.

Although there was a good correlation between the maximum number of adherent cells and D_S (R² = 0.966), the correlation was improved further when the maximum number of adherent cells was plotted versus the ratio D_s/($R\stackrel{·}{\gamma }$) (R² = 0.997, Fig. 4 A), which is inversely proportional to the net force on the cell (19).

Figure 4.

Dimensional analysis of CB-EPC adhesion. (A) Adhesion was limited by net force, since maximum adhesion scales linearly with D_s/($R\stackrel{·}{\gamma }$_max) (R² = 0.997). (B) The shear rate at maximum adhesion is inversely correlated with the fluid viscosity (R² = 0.9996). (C) D_m/($R\stackrel{·}{\gamma }$_max) for the shear rate at maximum adhesion is linearly related to D_m, the molecular diffusion coefficient (R² = 0.9986). (D) Total normalization of the adhesion data demonstrates that maximum adhesion is dependent on competitive transport processes, since all adhesion values collapsed onto a single, parabolic curve (R² = 0.770; n = 3–7, mean ± SE).

Next, we examined the transport processes that affected the value of the shear rate at maximum adhesion. To account for the combined effect of force and diffusion, we first plotted D_s/($R\stackrel{·}{\gamma }$) using the shear rate at maximum adhesion versus D_s or 1/D_s (data not shown). These correlations were poor (R² = 0.334 for D_s and R² = 0.144 for 1/D_s), suggesting the independent variable did not scale with D_s alone and that normalization of the number of adherent cells by D_s/($R\stackrel{·}{\gamma }$) accounted for the effect of force on adhesion.

The molecular diffusivity, D_m, represents random fluctuations of the receptor and the ligand-binding regions arising from the collisions with molecules in the fluid and equals k_BT/(6πμl), where l is the characteristic binding distance, 100 nm (19). Given that D_m scales with the reciprocal of the viscosity and that the shear rate at which adhesion is a maximum is linearly related to the reciprocal of viscosity (R² = 0.999; Fig. 4 B), we plotted D_m/($R\stackrel{·}{\gamma }$), again using the shear rate at maximum adhesion, versus D_m and found an excellent correlation (R² = 0.9986; Fig. 4 C). This correlation suggests that the shear rate at which adhesion is a maximum depends upon the balance between local rotation of the receptor in solution and the sliding velocity.

Collectively, the results suggest that the competitive transport processes of cell sliding, molecular diffusion, and force on the cells are responsible for the shape of the CB-EPC adhesion curve at different values of viscosity. To verify that these correlations include the key variables, we plotted the number of adherent cells for each viscosity condition normalized by

$$a_1\left(D_S/R\stackrel{·}{\gamma }\right)+a_2$$

versus the corresponding shear rates normalized by

$$b_{1}R\stackrel{·}{\gamma }\left(1+\frac{b_2}{D_m}\right),$$

where a₁ and a₂ are the coefficients derived from the linear regression of Fig. 4 A and b₁ and b₂ are the coefficients derived from Fig. 4 C (Fig. 4 D). The data aligned onto a single parabolic curve (R² = 0.770), with a clear maximum localized at

$$b_{1}R\stackrel{·}{\gamma }\left(1+\frac{b_2}{D_m}\right)=\mathrm{2.136.}$$

We examined a number of other variables affecting the maximum number of adherent cells and their corresponding shear rates. None produced as good a correlation as that found in Fig. 4 D.

CB-EPCs transiently interact with FN in a shear stress-independent manner immediately before firm arrest

Neutrophils undergo four distinct phases during cell arrest:

• 1.Tethering or bond formation between ligand and receptor (19,31,32).
• 2.Tumbling, which refers to initial cell tethering interrupted by breaks in bond formation, during which the cell travels a small distance with no adhesive interactions before another set of tethers (33).
• 3.Rolling, the erratic motion which balances bond formation and bond breakage for long periods of time (21).
• 4.Firm arrest mediated by integrins.

We measured CB-EPC velocities on bovine serum albumin (BSA) and FN surfaces to assess the types of transient adhesive events before firm arrest.

For cells flowing over BSA and those cells that did not appear to interact with FN, the mean translational velocity of CB-EPCs was linearly dependent on the applied shear rate for all shear rates up to 150 s⁻¹ (R² = 0.973; Fig. 5 A). Significant deviation from linearity occurred at shear rates >150 s⁻¹ (p < 0.001), which suggests that high shear rates may deform CB-EPCs (data not shown). All translational velocities over FN substrates were collectively compared to previously published data from Tempelman et al. (34) that analyzed hydrodynamic velocities of leukocytes, of similar size to CB-EPCs, in close proximity to gel-coated glass substrates. There was no significant deviation between the data of Tempelman et al. (34) and the translational velocities of CB-EPCs on FN or BSA (p > 0.05), as shown by the solid line in Fig. 5 A.

Figure 5.

CB-EPC translational and instantaneous velocity on FN. (A) Translational velocities at shear stresses <250 s⁻¹ were linearly related (R² = 0.973) and did not significantly deviate from the literature values presented by Tempelman et al. (34) up to 150 s⁻¹ (bold line) for similarly sized leukocytes (p > 0.05, n = 3, ± SD). (B) Instantaneous velocities between 1.0 and 4.5 s (expansion of inset) before firm arrest are independent of the applied shear stress, signified by the overlapping profiles of representative CB-EPCs. All representative results are indicative of triplicate measurements and a cell density of 5.0 × 10⁵ cells/mL at 1.0 cP viscosity.

Instantaneous velocities of representative cells were determined for the duration of cell motion across the field of view at long times before arrest (t > 10 s; Fig. S4). Cells chosen were already known to be interacting in some way with the FN substrate, as their mean velocity for the duration of data collection was <85% of both the calculated and experimentally determined hydrodynamic velocity. For 50 s⁻¹, 75 s⁻¹, and 100 s⁻¹, there is a sharp reduction of instantaneous velocity down to zero. The instantaneous velocity plots suggest that cell tumbling occurred because the velocity profiles were not erratic and lacked multiple, long-lived arrest plateaus, characteristic of rolling cells.

In contrast, all CB-EPCs that became adherent abruptly decreased their cellular velocity before firmly adhering, regardless of the applied shear stress (Fig. 5 B). Immediately before firm arrest, there were few transient arrests longer than 0.033 s as the cell decelerated. Thus, the decline in instantaneous velocity was due to short transient arrests (t < 0.033 s). Together, these results indicate that nonadherent CB-EPCs did not interact with the surface and that adherent cells interacted briefly before firm arrest.

α₅β₁ is the primary integrin responsible for initial cell capture to FN

To establish that cells adhered specifically to FN, static and dynamic adhesion experiments were conducted with monoclonal antibodies that blocked binding to either or both α₅β₁ and α_Vβ₃ integrins (Fig. 6). CB-EPC adhesion was significantly reduced relative to both static and flow controls when α₅β₁ integrins were blocked (p < 0.05) and when both α₅β₁ and α_Vβ₃ were blocked (p < 0.05). There were no significant differences between controls and cells blocked with anti-α_Vβ₃ for either static or flow conditions. Cell velocities of CB-EPCs in which α₅β₁ was blocked were similar to that of the control cells (data not shown). Thus, initial capture of CB-EPCs to FN is directly dependent on α₅β₁ and not α_Vβ₃ integrin.

Figure 6.

Integrin blocking of adhesion of flowing CB-EPCs in suspension to FN adsorbed to polystyrene slides. Number of adherent cells after blocking the α₅β₁ integrin decreased significantly for both static and dynamic adhesion to FN + BSA coated substrate (τ = 1.0 dyn/cm², 10 min, saturated FN; ^∗p < 0.05 relative to unblocked cells, ^#p < 0.05 relative to α_Vβ₃ blocking in equivalent conditions, n = 3, mean ± SE).

We also measured the number of α₅β₁ and α_Vβ₃ integrin receptors on the cell surface quantitatively by flow cytometry and a set of calibration beads as described in the Supporting Material. There were 269,100 ± 135,600 α₅β₁ integrin receptors per cell and 84,400 ± 27,600 α_Vβ₃ integrin receptors per cell (n = 2). Thus, the greater role of α₅β₁ in adhesion may, in part, reflect the greater number of these molecules per cell.

CB-EPCs possess a high strength of adhesion after initial settling

To assess cellular retention, the number of adherent CB-EPCs after 20 min of flow over saturating amounts of FN was counted after increasing the shear stress. Upon increasing the fluid shear stress to 20 dyn/cm² for 5 min, which is comparable to physiological shear stresses, the mean percent retention of adherent cells was consistently >90% for each initial shear stress condition analyzed (Fig. S5). This strength of adherence is directly in accord with our previous study with monolayered CB-EPCs exposed to supraphysiological levels of shear stress (17).

CB-EPCs form multiple contacts after initial attachment

We used total internal reflection fluorescence (TIRF) microscopy to examine the number of discrete contacts and contact area during attachment. As described in the Supporting Material, TIRF was performed using either a membrane dye or labeled antibody to α₅β₁. We examined cells attached under static conditions to simplify data collection because the field examined was at 100×, limiting the number of cells that we could examine at any one time. Although the contact areas may be smaller for cells attaching under flowing conditions, we expect the trends to be similar. Representative images (Fig. S6) compare differential interference contrast images with TIRF images for labeled antibody bound to α₅β₁. Several discrete regions of contact are apparent and the size and frequency of these events increases from 5 to 20 min of attachment. The TIRF patterns are similar to those observed with reflection interference contrast microscopy (35,36). These regions are not all within the portion of the cell directly beneath the cell centroid, suggesting that a number of contacts arise from pseudopods projected by the cell after attachment.

We define the contact area as the cumulative cell area at separation distances of 50 nm or less. The contact area of the cell membrane was 6.5 ± 4.0 μm² (n = 13 cells) at 5 min and increased to 20.6 ± 10.0 μm² (n = 4 cells) at 20 min (mean ± SD). The contact area occupied by the α₅β₁ integrin is less than the membrane contact area, with a value of 1.8 ± 2.4 μm² (n = 14 cells) after 5 min of attachment and 2.3 ± 3.5 μm² (n = 25 cells) for 20 min of attachment.

Discussion

Using a parallel plate flow chamber, we have made what we believe are novel observations about the dynamic adhesion of CB-EPC from a flowing solution onto FN surfaces. Adhesion exhibited a maximum at ∼1.0 dyn/cm² and was specific to α₅β₁. Cells made multiple small attachments via α₅β₁ integrin and the integrin contact area was less than the area of the cell membrane within 50 nm of separation from the surface. The apparent binding rate was not limited by settling to the surface and was slow relative to axial convective transport. The maximum number of adherent cells was affected by the net force on the cells, and the shear rate that produced the maximum number of adherent cells was a function of the sliding velocity and reciprocal of the molecular diffusivity. FN supported cell arrest and limited rolling. Once attached, cells adhered firmly, with little subsequent detachment when exposed to 20 dyn/cm² shear stress.

Cell solution density significantly affected number and size of CB-EPCs that adhered to a surface saturated with FN (Fig. 1). The data in Fig. 1 indicate that the cross-sectional cell area of attached cells is greater at the lower cell densities. Even at the highest cell densities, the cross-sectional area is 48% larger than the mean area of cells in suspension. At low cell densities, larger cells preferentially adhere due to their larger size. Additionally, larger cells settle faster, enriching their concentration near the surface, thereby leading to their preferential adhesion. Because the collision frequency increases 100-fold for a 10-fold increase in cell concentration and collision frequency is somewhat less for smaller particles (37), the decline in the size of adherent cells with increasing solution density may result from the increased likelihood of collisions of smaller cells. Further, the net force on smaller cells is less than the force on larger cells and larger cells may be more likely to detach at the higher shear stresses, unless the contact area is greater for larger cells. At the higher cell densities, due to the relative abundance of smaller cells, depletion of larger cells from solution, and increased frequency of collisions between cells, the smaller cells represent a greater fraction of the attached cells. Although the coverage remained <10%, the nonlinear relation between cell concentration and adhesion may be due to increased mixing and interruption of the tethering process when an EPC comes into contact with an adherent EPC.

Our findings show the total cell coverage would require longer times or higher cell concentrations. Extrapolating the data from Fig. 1 C at a shear stress of 1.0 dyn/cm², roughly 2.0 × 10⁷ cells/mL are necessary for complete coverage of a denuded surface in 20 min. For in vivo seeding applications, cell densities of 2.0 × 10⁷ cells/mL are feasible, but the total number of cells may be limiting such that complete coverage is unlikely. CB-EPCs would need to proliferate to fill in gaps along the surface to generate a confluent monolayer similar to the autologous vessel.

The highly significant reduction in adherent cells with increased shear stress and the role of net force on adhesion at higher shear stresses suggests that shear stress plays a more significant role in CB-EPC adhesion relative to hydrodynamic shear rate. Consequently, the effective force to which the cell is exposed and resulting contact area between the cell and ligand interfaces may moderate adhesion rather than contact time. Additionally, CB-EPCs exhibited a biphasic response in adhesion quantity at different viscosities, which is similar to the response observed for neutrophil tethering (19). Thus, as shear stress initially increases, there is an increase in tethering frequency due to an increase in the delivery of cells to the FN surface and contact between receptor and ligand. At higher shear stresses, adhesion decreases as the net force on cells causes binding receptors on the cell surface to be separated from their substrate ligands. Additionally, higher shear stress shortens the contact time between the receptor and substrate, further decreasing the chances of adhesion. This mechanism is consistent with a decrease in the adhesion curves of Fig. 2 after the maximum value is reached and signifies that reaction-limited adhesion occurs.

Interestingly, the biphasic shear response exhibited in Fig. 2 occurs with different media viscosity, though maximal CB-EPC adhesion decreases with increasing viscosity. One hypothesis is that with increasing fluid viscosity, the rotational diffusion coefficient of the cell decreases. Rotational diffusion is known to orient cell binding sites, resulting in successful molecular docking (19). The good correlation presented in Fig. 4 suggests that local rotation and translation of the cell are critical for adhesion. Other factors, such as cell deformation, play a secondary role.

Another criterion for the successful use of CB-EPCs in therapeutic applications is a significant reduction in cell velocity to mediate firm arrest. This statement was motivated by previous studies indicating that due to interactions with selectins, white blood cells undergo substantial reduction of translational velocity before cell arrest (21,29,38,39). The instantaneous cellular velocity sharply decayed up to the point of firm arrest, was independent of the applied shear stress, and included brief transient arrests with lifetimes <0.033 s. This behavior is reminiscent of leukocyte adhesion simulations by Caputo and Hammer (40). They proposed that the behavior of a cell introduced to a shear stress environment may undergo a landing state, defined as the time interval where a cell suddenly switches from traveling near the hydrodynamic velocity to being firmly bound. Thus, CB-EPCs may adhere to FN through a landing mechanism, which would explain the drastic velocity reduction from the hydrodynamic velocity at the short times just before arrest.

Alternatively, fluid shear stress may rapidly activate integrin proteins into high affinity states for FN ligand. The α₅β₁ integrin possesses two active binding motifs in both the presence and absence of cell tension (41). Shear stress promotes the activation of FN-specific α₅β₁ and α_Vβ₃ integrins (42), which facilitate development of focal adhesion complexes (43–46). Our blocking studies suggested that α₅β₁, but not α_Vβ₃, is the primary integrin involved in the initial cell arrest. Single molecule studies show that α₅β₁ can form catch bonds with FN (47) with a maximum bond lifetime at 40 pN per bond. Although α₅β₁ can resist forces applied to its bond with FN, α_Vβ₃ engagement with FN is essential for the engagement with the cytoskeleton (48). Further studies are needed to evaluate the role of catch-bond formation upon CB-EPC adhesion.

We conclude that CB-EPCs are an effective cell source for direct injection applications to rapidly reendothelialize a denuded or damaged vasculature possibly due to atherosclerosis. The cells are able to undergo transient tethering events and firm arrest through linkages between FN and α₅β₁. Our studies provide important proof in concept that EPC adhesion can occur under physiological levels of shear stress. The shear stresses for which the CB-EPCs adhered in vitro are on the low end of shear stresses present in vivo. However, EPC adhesion is enhanced by mixing induced by red blood cells, the complex nature of the flow field on the transport and adhesion of EPCs to sites of vessel wall injury, and the extracellular matrix proteins present on the injured subendothelium. Future studies need to examine how these factors affect the amount of adhesion and the relative importance of transport and molecular binding processes.