The proteins analyzed in this study are described below. The integrin heterodimer α_vβ₃, obtained from R&D Systems (Minneapolis, MN, United States; Catalog #3050-AV-050), consists of an α_v extracellular subunit (110.5 kDa) and a β₃ extracellular subunit (80.8 kDa) linked by covalent bonding of disulfide bonds. The N-terminal head region is the ligand binding site of FN, formed by combination of the β-propeller structure of α_v subunit and vWF-A domain of β₃ subunit. The FN protein (R&D Systems, Minneapolis, MN, United States; Catalog #1918-FN-02M) was derived from human plasma. The FN used in this study was made up of three types of homologous structural repeats, termed FN type I, type II, and type III repeats (EDA and EDB). A truncated fibronectin 1.3 (FN1.3; R&D Systems, Minneapolis, MN, United States; Catalog #3938-FN-050), which includes the EDB plus type III domains #8–13 and the initial region of domain 14, was also used; these domains facilitate association with α_vβ₃. Furthermore, the RGD motif, located between type III domains #9 and #10, determines the affinity of FN to integrin α_vβ₃.

To couple integrin α_vβ₃ to flow chambers, a square region (25 mm²) coated with proteins was labeled in the center of each 35 mm dish well (Corning, United States) and was constrained with a clean silicon rubber. Then, 40 μL of integrin α_vβ₃ was added onto the square region in the center of the Petri dish, and directly adsorbed onto the bottom via incubating it at 4°C refrigerator for 16 h (Li et al., 2012). Here, different concentrations of integrin α_vβ₃ were chosen for specific experimental purposes. Based on the results of a preliminary experiment with various concentration gradients, 20 ng/mL and 40 ng/mL of integrin α_vβ₃ were used to investigate transient tethers and rolling adhesion, respectively. After overnight incubation, the coated square region was washed three times with phosphate-buffered saline (PBS) to remove the unadhered proteins, and then the 35 mm dish bottom was incubated with 800 μL PBS containing 2% bovine serum albumin (BSA) at room temperature for 30 min to prevent non-specific adhesion.

To coat microspheres with FN, glass beads of 3 μm radius (BD Warrington, PA) were washed three times with 1 mL PBS, followed by functionalization of the microspheres by mixing the beads with 40 μL of FN (200 μg/mL) or FN1.3 (100 μg/mL). To mix the microspheres and proteins well, they were incubated in a falcon rotating every 10 min for 2 h in a 4°C refrigerator overnight. Afterwards, the beads with protein were washed once and blocked for 2 h with PBS containing 2% BSA. Thereafter, the beads were stored at 4°C in PBS containing 0.1% sodium azide for up to 5 days (Yago et al., 2007).

FN- or FN1.3-expressing microspheres (0.5 × 10⁶ beads/mL in HBSS, containing 1% BSA, with calcium and magnesium ions at a concentration of 1 mM) were perfused under various flow shear stress τ_w (0.1–0.7 dyn/cm²) over circular flow chamber (GlycoTech, Gaithersburg, Maryland) with low or high density of integrin α_vβ₃. A silicon rubber gasket was used to create a working flow space with length × width × thickness = 2 × 0.5 × 0.0254 cm³ (Li et al., 2016; Yago et al., 2004). The flow shear stresses were converted into flow shear rates and controlled using a Harvard pump. The cell adhesion process was observed using a Zeiss inverted microscope and adhesion images were recorded using a CMOS camera at a speed of 100 frames per second (fps). After transferring the imaging data into video, the rolling trajectory of the cells was tracked using IPP software (Image Pro Plus). To change the viscosity in specific experiments, 3% (w/v) ficoll (3–5 × 10⁵ Mr; Amersham Biosciences) was added to the medium to increase the viscosity by 1.69-fold, as determined using a cone plate viscometer (Yago et al., 2007). To certify all tethering and rolling events were mediated by specific interaction of integrin α_vβ₃ with FN, HBSS with 2% BSA were added to the incubated substrate to exclude non-specific adhesion of microspheres.

Transient tether lifetimes of one stop 3 μm-radius beads bearing FN or FN1.3 were measured on low densities of integrin α_vβ₃ (coated at 20 μg/mL) that did not support rolling or skipping at any wall shear stresses (Li et al., 2012; Yago et al., 2004). Images of the transient tethers were captured using a high-speed complementary metal-oxide speed CMOS acquisition system (Mikrotron GmbH MC 1310; Norpix Ins.) at 100 fps (Li et al., 2012). The replayed images were captured at a slow speed, and the duration of the transient tethers were recorded through frame-by-frame analysis with IPP. Three sets of lifetimes (approximately 60 tethering events per set) were measured for each wall shear stress. Data are presented as mean ± standard deviation of average lifetimes. When the tether lifetimes were produced by one bond interaction of integrin α_vβ₃ with FN/FN1.3, first-order dissociation kinetics model can be applied to perform linear fitting for off rates (Coburn et al., 2011; Li et al., 2016). Therefore, off-rates can be derived from negative slope by analyzing the plot of ln (number of events with a lifetime ≥ t) vs. t, which was fitted using a straight line. The correlation coefficient R ² was >0.9 for all fits. The means and standard deviations of the off rates at each wall shear stress were calculated (Li et al., 2012; Sarangapani et al., 2004).

To determine the rolling behavior, we first observed the cells rolling with a ×10 objective on a Zeiss microscope and captured images with a high-speed camera at 100 fps. The Image Pro Plus software (IPP, Media Cybernetics, v6.0) was used to track the rolling cells and obtain the coordinates of the centroids. The instant velocity was calculated by dividing the distance between two continuous centroids by their elapsed times (10 ms). The continuous sliding average of every 10 frames for raw instant velocity data was obtained to reduce the noise levels. Custom-designed Excel macros were used to analyze the rolling step for the beads. Microsoft Excel macros were written based on a minimal model that describes a continuous stop-and-go motion mediated by only one and two bonds (Yago et al., 2004). As soon as a bond dissociated in the rear region of the bead, the bead entered the acceleration phase. If a new bond was formed in the front region of the bead, the bead entered the deceleration phase. Here, a step was defined as a cycle of acceleration and deceleration, which may or may not include a stop phase. Based on this model, several researches have conducted rolling step analysis for rolling cells or microspheres mediated by L-selectin-PSGL-1 interaction (Yago et al., 2004), E-selectin-ligand interaction (Li et al., 2012; Li et al., 2016), and vWF-GPIbα interaction (Coburn et al., 2011). In this study, we analyzed the rolling behavior of FN/FN1.3-coated bead on integrin α_vβ₃ substrate. A stop phase was defined for rolling bead when the bead’s instantaneous velocity was lower than the system noise level (<20 μm/s) measured from settled beads in a control preliminary experiment. An acceleration and deceleration threshold of ±600 μm/s² was selected based on specific formation and breakage of bonds measured for free flowing cells caused by Brownian motion. More than 800 stop-and-go events were recorded from approximately 45 beads for each shear stress. Finally, several parameters were extracted from these stop-go events, including the mean stop times and stop frequencies.

Statistical significance was evaluated using GraphPad Prism (version 9.5.0) and Origin (2017). All data are expressed as mean ± standard deviation (SD) in the figures. Each group underwent three independent replicate experiments, with a minimum of 20 microspheres collected per experiment. Statistical analyses were performed using Student’s t-test and one-way ANOVA with multiple comparisons to assess the significance between groups (ns for p > 0.05, * for p < 0.05, ** for p < 0.01, *** for p < 0.001, **** for p < 0.0001). Prior to applying these parametric tests, we conducted the Shapiro-Wilk test to verify the normality of the data and Levene’s test to check for homogeneity of variance.

After setting up the flow chamber as shown in Figure 1A, binding experiments were performed to test the adhesion specificity of the integrin α_vβ₃-FN interaction. We perfused the FN-expressing beads into the flow chamber to give the beads the opportunity to adhere to the integrin α_vβ₃ substrate at flow shear stress 0.3 dyn/cm². The rolling, tethering, and firm adhesion events were determined based on distance-time curves (Figures 1B, C). Tethering refers to the transient stop event mediated by a single molecular interaction (Figures 1D, E). Rolling refers to the occurrence of more than two consecutive molecular interactions following the “stop-go-stop” pattern (Figures 1F, G) (Kong et al., 2009), and the firm adhesion refers to the stable adhesion occurring after the interaction of multiple pairs of molecules, with durations exceeding 3 min (Figure 1B). The percentage adhesion rate was calculated by dividing the number of the three types of adhered cells (Figure 1C) by the number of all flowing cells in the same window and time. The results showed that the adhesion rate was very low with approximately 5%–10% of FN or FN1.3-captured beads adhering to PBS or 2% BSA-coated substrates. However, the percentage of adhesion rate was increased to approximately 30% and 40% for FN1.3- and FN-coated beads adhered on integrin α_vβ₃ substrates, respectively. Therefore, the tethering and rolling and firm adhesion events in our experiment system were specific for FN- or FN1.3-coated beads because they were eliminated by blocking with PBS containing 2% BSA (Figure 1C).

To investigate the influence of shear force on the interaction of integrin α_vβ₃ with FN, we measured transient tether lifetimes on bottom of the chamber with low-density of integrin α_vβ₃ (coated at 20 μg/mL). These densities did not allow the FN-bearing beads to roll or skip on the substrate. Only a few sliding cells had the opportunity to transiently and briefly stop on the substrate. We used a high-speed camera to record these events at 100 fps, and the transient tether lifetimes were obtained by analyzing the duration of the stop events by offline tracking. The results indicated that the transient tether lifetime first shortened, then prolonged with increasing wall shear stress, exhibiting catch-slip bonds in the interaction of integrin α_vβ₃ with FN (Figure 2A) and FN1.3 (Figure 2E). This biphasic lifetime curve has previously been reported for selectin family member E/P/L-selectin interacting with PSGL-1 (Li et al., 2012; Marshall et al., 2003; Yago et al., 2004), as well as integrin family member α_Lβ₂ interacting with ICAM-1 (Chen et al., 2010), and immunoglobulin superfamily member TCR interacting with pMHC (Depoil and Dustin, 2014). Furthermore, we used the transient tether lifetimes to derive the dissociation rate constant off-rates (k _off ), which can be reported as the dissociation ability of single bonds. By linear fitting curve of ln (number of events with a lifetime ≥ t) vs. t for low shear stress 0.1–0.3 dyn/cm² (Figures 2C, G) and high shear stress 0.3–0.7 dyn/cm² (Figures 2D, H), the off-rates were extracted from the negative slope of the fitting plots. As shown in the figure, the off-rates decreased initially, then increased finally with wall shear stress in the tethers mediated by both integrin α_vβ₃-FN (Figure 2B) and integrin α_vβ₃-FN1.3 (Figure 2F). The biophasic transient tether lifetime curve was inversely related to the dissociation rate constant off-rate, indicating that the force-enhanced bond lifetime was due to force-reduced bond dissociation. The lifetime and off-rate curves had the same tendency for both FN and FN1.3, indicating that FN1.3 containing #8–14 domains was sufficient for the interaction with integrin α_vβ₃.

First, we measured mean rolling velocity of beads rolling on immobilized integrin α_vβ₃ for 10 s and found it to change with increasing flow from the threshold to and above the optimal value. The mean velocity, a global parameter of rolling adhesion, increased monotonously with the wall shear stress from 0.1 to 0.7 dyn/cm². At first glance, there was no catch bond phenomenon in this observation (Figures 3A, B). However, when we changed the vertical coordinate from rolling velocity to a reduced percentage of velocity, defined as the ratio of the decreased velocity (from free flowing to rolling) to the free-flowing velocity, the reduced percentage of the velocity first increased and then decreased as the wall shear stress increased (Figures 3C, D), similar to previous catch bond observations. This suggests that the interaction of integrin α_vβ₃-FN increased the reduced percent of velocity, that is decreasing value of rolling velocity, to help stabilize rolling on the substrate in the catch bond region (0.1–0.3 dyn/cm²). Beyond the stable rolling optimum shear stress (0.3 dyn/cm²), the reduced percent of velocity decreased with shear stress in the slide bond region (0.3–0.7 dyn/cm²), where the beads rolled quickly.

The biphasic lifetime likely governs the biphasic mean rolling velocity by controlling the stop times between the two go phases. To test these hypotheses, we analyzed the rolling behaviors of FN-bearing microspheres on integrin α_vβ₃-coated substrates recorded using high speed camera at 100 fps. The centroid coordinates of the microspheres were obtained by outline tracking frame-by-frame with IPP. Instantaneous velocity profiles were plotted by dividing the distance between two continuous centroid coordinates by the corresponding elapsed time. As shown in Figure 4, the irregular stop-and-go motions (Figures 4B–F) owning to specific FN-integrin α_vβ₃ interaction were different from free-flowing microsphere motions (Figure 4A). As shear stress increased from 0.1 dyn/cm² to 0.3 dyn/cm², the flow force promoted the frequency of irregular motions with longer stop times. As shear stress increased beyond the optimal value of 0.3 dyn/cm², the trend was reversed. With increasing shear stress from 0.3 dyn/cm² to 0.7 dyn/cm², the flow force reduced the frequency of irregular motions with shorter stop times. Similar rolling behaviors were observed in the FN1.3-bearing microspheres (Figures 4H–L). These rolling phenomena are possibly correlated with the lifetimes of integrin α_vβ₃-FN bonds, and governed by catch bond mechanism. That is, a weak short-lived bond will become stronger and longer-lived as the flow force increases from a low level to an optimal value (a catch bond-dominating region) and then switch to a weak shorter-lived bond as the force increases across the optimum (a slip bond-dominating region). The transitions between catch and slip bonds govern the stop-and-go instantaneous velocities below and above the optimal force.

To quantify the rolling behaviors shown in Figure 4, a homemade Excel macro was used to describe a minimum stop-and-go rolling model. A rolling step cycle includes a stop-and-go phase with or without a stop phase in the middle. We separated the whole rolling process into numerous stop and go phases from the instantaneous velocity curve when the acceleration and deceleration thresholds were set as 600 μm/s². Thousands of events were gathered for approximately 15 microspheres during 10 s of rolling at each shear stress. Subsequently, we extracted the mean stop time and stop frequency as parameters to describe the rolling characteristics. As shown in Figure 5, the mean stop time and stop frequency first increased (from 0.1 to 0.3 dyn/cm²) and then decreased (from 0.3 to 0.7 dyn/cm²) when flow shear stress increased across the optimal value (0.3 dyn/cm²). This biphasic mean stop time derived from rolling of full/truncated FN-bearing beads on high density integrin α_vβ₃-coated substrate is consistent with the biphasic transient tether lifetime measured at very low integrin α_vβ₃ density. In addition, the curve of the reduced percentage of velocity (Figures 3C, D) matched the transient tether lifetime (Figures 2A, B) and mean stop time (Figures 5A, B), with similar biphasic curves and the same flow force optimum. This suggests that transitions between catch and slip bonds govern the lifetimes of transient tethers by single bonds and rolling tethers via a small number of bonds, and that they regulate rolling behaviors. In the catch bond regime, the increasing of force strengthens integrin α_vβ₃-FN bonds to promote beads to stop, which increases the rolling regularity and reduces rolling velocity. Conversely, the force weakens integrin α_vβ₃-FN bonds in the slip bond regime. In summary, the transitions between catch and slip bonds regulate rolling behavior.

In a study on flow-enhanced adhesion mechanisms, it was suggested that the increase in molecular bond formation with increasing fluid shear force is due to the transport effect of the fluid, which increases the binding rate of the molecular bond (Yago et al., 2007). This results in an increased number of bonds being formed during cell or microsphere rolling. There are two possible explanations for this phenomenon. First, fluid transport can carry the ligands to the receptors, increasing the binding probability of the receptor and ligand and facilitating rapid bond formation and breakage. Second, the fluid shear force can flatten elastic cells and expand their contact area, leading to increased bond formation. To investigate whether the transport mechanism is involved in the adhesion of microspheres mediated by the α_vβ₃-FN1.3 molecular bond, we added 3% (w/v) ficoll to the solution to change the viscosity of the fluid and repeated the rolling experiment under the same conditions. As shown in Figure 6, by comparing the rolling parameters of the microspheres in different adhesive fluids, it was found that the rolling parameters were not aligned under the shear rate coordinates, but a single curve was synthesized under the shear stress coordinates. This scaling relationship precludes the possibility that shear rate transport is a flow-enhanced rolling mechanism, suggesting that shear stress may be involved in the regulation of cell rolling behavior. These findings suggested that the flow-enhanced rolling mechanism is influenced by shear stress rather than the transport effect (Yago et al., 2007; Zhu et al., 2008).