Real-time analysis of very late antigen-4 affinity modulation by shear.

Shear promotes endothelial recruitment of leukocytes, cell activation, and transmigration. Mechanical stress on cells caused by shear can induce a rapid integrin conformational change and activation, followed by an increase in binding to the extracellular matrix. The molecular mechanism of increased avidity is unknown. We have shown previously that the affinity of the alpha(4)beta(1) integrin, very late antigen-4 (VLA-4), measured with an LDV-containing small molecule, varies with cellular avidity, measured from cell disaggregation rates. In this study, we measured in real time affinity changes of VLA-4 in response to shear. The resulting affinity was comparable with the state mediated by receptor signaling and corresponded in time with intracellular Ca(2+) responses. Ca(2+) ionophores and N,N'-[1,2-ethanediyl-bis(oxy-2,1-phenylene)]bis[N-[2-[(acetyloxy)methoxy]-2-oxoethyl]]-, bis[(acetyloxy)methyl]ester demonstrate that the affinity regulation of VLA-4 in the presence of shear was related to Ca(2+) signaling. Pertussis toxin treatment implicates G(i) in an unknown pathway that connects shear, Ca(2+) elevation, VLA-4 affinity, and cell avidity.

Shear promotes endothelial recruitment of leukocytes, cell activation, and transmigration. Mechanical stress on cells caused by shear can induce a rapid integrin conformational change and activation, followed by an increase in binding to the extracellular matrix. The molecular mechanism of increased avidity is unknown. We have shown previously that the affinity of the ␣ 4 ␤ 1 integrin, very late antigen-4 (VLA-4), measured with an LDV-containing small molecule, varies with cellular avidity, measured from cell disaggregation rates. In this study, we measured in real time affinity changes of VLA-4 in response to shear. The resulting affinity was comparable with the state mediated by receptor signaling and corresponded in time with intracellular Ca 2؉ responses. Ca 2؉
Leukocytes are recruited to endothelial cells in a multistep process using selectin and integrin adhesion molecules (1,2). These molecules allow a cell to tether, roll, adhere, and transmigrate along and across an endothelial layer. Selectin and some integrin molecules and their associated ligands mediate tethering and rolling interactions. Firm adhesion is mediated by vascular ligands of the immunoglobulin superfamily such as vascular cell adhesion molecule 1 (VCAM-1) 1 and their associ-ated integrins (1,2). The adhesive strength or avidity (3) of cells expressing integrins can be rapidly modulated by chemokines and chemoattractants, which also regulate leukocyte recruitment and migration across vascular endothelium. The rapid changes in avidity have been attributed to changes in the number of interacting molecules or valency due to molecular redistribution or clustering and to changes in the affinity of the individual receptor-ligand bonds (3)(4)(5)(6)(7)(8)(9)(10).
Physiological shear can also regulate leukocyte traffic by stimulating mechanosensors on neutrophils, monocytes, lymphocytes, erythrocytes, and platelets (see Ref. 11 and references therein). Shear arises from bifurcating blood vessels or rapid changes in blood vessel diameters. Shear acting on leukocytes, bound to endothelial cells, produces mechanical stress on the cells or their receptors, regulating cell growth and proliferation, protein synthesis, gene expression, and blood cell recruitment (12,13). Integrins (such as ␣ v ␤ 3 , ␣ 5 ␤ 1 , ␣ 4 ␤ 1 , and ␣ 2 ␤ 1 ) on endothelial cells can act as mechanosensors to changes in blood flow (13,14) and trigger an intracellular signaling pathway involving focal adhesion kinase and mitogen-activated protein kinase cascades. How shear specifically induces blood cell adhesiveness or recruitment through mechanosensors is unknown. Indirect evidence shows that increased integrin binding to the extracellular matrix occurs when shear acts on cells or their mechanosensors to induce intracellular signaling. For example, intracellular signaling leads to conformational changes and activation of ␣ v ␤ 3 on endothelial cells and ␣ 4 ␤ 1 and ␣ 5 ␤ 1 integrins on monocytic cells (15,16). Shear acting on endothelial cells affects the GTPase Rho signaling pathway and in monocytic cells induces inositol 1,4,5-trisphosphate-sensitive Ca 2ϩ release that affects cell adhesion avidity.
We have used an LDV-containing small molecule fluorescent probe to determine whether mechanical stress generated by shear can affect the affinity of VLA-4 by monitoring in real time the changes in VLA-4 affinity on live cells (17). We examined the contribution of intracellular signaling mechanisms to VLA-4 activation by shear. We found that VLA-4 affinity induced by shear was intermediate in affinity between the resting state and the Mn 2ϩ -activated affinity state and similar to the physiologically activated receptor state generated using "inside-out" signaling (17). We found a temporal correlation between the intracellular Ca 2ϩ response and the higher VLA-4 affinity. We used Ca 2ϩ ionophores (A23187 and ionomycin) and BAPTA-AM to show that VLA-4 affinity regulation in response to shear was related to intracellular Ca 2ϩ signaling. Finally, we pretreated cells with pertussis toxin (PTX) to block G i signaling) and observed that VLA-4 activation was inhibited in the presence of shear. Our data suggest that shear regulates cell adhesion avidity by changing VLA-4 affinity and involves an incompletely characterized inside-out signaling pathway.
Cell Lines and Transfectant Construct-Human monoblastoid U937 cells were purchased from ATCC (Manassas, VA). Cells were grown in RPMI 1640 (Invitrogen) supplemented with 10% heat-inactivated fetal bovine serum, 100 units/ml penicillin, 100 g/ml streptomycin, 10 mM HEPES, pH 7.4, 100 g/ml ciprofloxacin, 2 mM L-glutamine, at 37°C in a humidified atmosphere of 5% CO 2 and 95% air. Site-directed mutants of formyl peptide receptor in the human monoblastoid line U937 constitutively expressing human VLA-4 integrin were prepared as described (19). High expressors were selected using the MoFlo Flow Cytometer (Cytomation, Inc., Fort Collins, CO). VLA-4 expression was measured with FITC-44H6 and quantified by comparison with a standard curve generated with Quantum Simply Cellular microspheres (Flow Cytometry Standards, San Juan, Puerto Rico) stained in parallel with the same monoclonal antibody. This produces an estimate of the total monoclonal antibody-binding sites/cell. Typically, we find 40,000 -60,000 VLA-4 sites/U937 cell.
LDV-FITC Probe-The VLA-4 probe (20 -22) was initially optimized from the ILDV binding sequence of the alternatively spliced connecting segment 1 of fibronectin. This sequence is homologous and isosteric with the QIDS peptide found in the VCAM-1-binding site (23). The peptide sequence (Leu-Asp-Val-Pro-Ala-Ala-Lys-FITC) of the probe was based on structure-activity relationships of a potent VLA-4 binding inhibitor (compound 13 in Ref. 22). The specificity of the molecule for the VCAM-1/VLA-4 interaction was examined previously in cell adhesion and ligand binding assays (17). The binding characterization showed that molecular dissociation rates of the LDV-FITC probe from VLA-4 on U937 were homogenous (i.e. single exponential) regardless of the cation present (18).
Cell Preparation-U937 cells (10 ϫ 10 6 cells/ml) for shear experiments were loaded with 6 M Fura Red or 200 nM Fluo-4, for 30 -60 min at 37°C and gently mixed every 10 min. Then the cells were washed with complete RPMI and resuspended in phenol red-deficient RPMI (supplemented with 0.1% human serum albumin (Bayer Corp., Elkhart, IN) and 1.5 mM Ca 2ϩ ). Cells were kept on ice after staining and washing. Typically, 5 min prior to each experiment, 4 nM LDV-FITC probe was added as a ligand to 1 ϫ 10 6 cells/ml, and the sample was incubated in a 37°C water bath. Cells were illuminated with the 488-nm argon laser from a Becton-Dickinson FACScan flow cytometer (BD Immunocytometry Systems, San Jose, CA). Emission fluorescence was detected using a 585-nm band pass filter for Fura Red (FL2) and 530-nm band pass filter for Fluo-4 (FL1). Fura Red fluorescence decreased when the indicator bound to free Ca 2ϩ . Changes in the affinity state of VLA-4 were monitored using the LDV-FITC probe. The probe was added 5 min prior to each experiment usually at 4 nM and incubated in a 37°C water bath. Detailed analysis of real time binding and dissociation of the LDV-FITC probe was previously described in Refs. 17 and 18. In several experiments (where the extracellular Ca 2ϩ concentration varied), Hepes buffer (110 mM NaCl, 10 mM KCl, 10 mM glucose, and 30 mM HEPES, pH 7.4) supplemented with 0.1% human serum albumin was used. Cell density was determined using a Z2-Coulter counter (Coulter Corp., Miami, FL).
Intracellular Calcium Calibration-Molecular Probes calcium calibration kit 1 was used to generate a series of free calcium buffers that were used to obtain an intracellular cellular calcium calibration curve (Fig. 1). The kits contain two 50-ml solutions, one solution containing 10 mM K 2 EGTA and the other 10 mM CaEGTA. Both solutions contained 100 mM KCl, 30 mM MOPS, pH 7.2. Intermediate free calcium concentrations between 0 and 39 M were obtained by cross-diluting the two buffers. Before adding U937 to each of the prepared buffers, the cells were stained with the intracellular calcium indicator Fura Red. Prior to each experiment, 1 ϫ 10 6 U937 cells were added to 1 ml of a specific free calcium buffer. Then the solution was incubated for 5 min in a 37°C water bath. A base line was established during the first 2 min of sampling with a FACScan to measure the resting state of the cells. Then 10 ng/ml of a calcium ionophore (A23187) was added and mixed gently, and sampling was resumed. Measurements of intracellular calcium were obtained when the Fura Red signal equilibrated. Fig. 1 shows that changes in the mean channel fluorescence (MCF) corresponded to logarithmic changes in the intracellular calcium levels. The intracellular Ca 2ϩ calibration curve depended on Fura Red staining efficiency, viability of U937 cells, sensitivity of cells to external activation, and flow cytometer voltage and gain settings. The Fura Red MCF values for cellular resting states between 550 and 650 correspond to intracellular calcium concentrations between 100 and 10 nM. MCF values of ϳ400 after stimulation indicate an intracellular calcium concentration of Ն1000 nM.
Creating Fluid Shear-Fluid shear was initially generated using a Fischer Scientific minivortexer (Fischer Scientific, Hampton, NH) set to 3200 rpm. The shear rate was estimated to be ϳ200 -12,000 s Ϫ1 , comparing the vortexed fluid motion inside a 12 (outer diameter) ϫ 75-mm tube with the fluid motion inside a Couette viscometer. The maximum (S max ) and minimum (S min ) wall shear rate for a given rotational velocity was approximated (24) as follows, where R I (ranging from ϳ0.535 to 0.25 cm) and R O (0.55 cm) represent radii of the inner fluid and outer fluid surfaces, and ⍀ is the angular speed of the inner cylinder. Before being subjected to shear, U937 cells were incubated for 5 min in a 37°C water bath. Each sample was gently mixed to resuspend cells, and a tube was attached to a flow cytometer. Data were acquired for 1-3 min to establish a base line for resting cells, and then each sample was removed from the flow cytometer to be exposed to shear for 5-30 s using a minivortexer. Samples were reattached to the flow cytometer, and data sampling was resumed.
where Q represents the flow rate, and r is the tube radius (the corresponding wall shear rates were calculated to be 750, 2300, 4600, and 9200 s Ϫ1 , respectively), which was the maximal shear rate cells would experience (24). For deformable particulates, such as cells, there is a net radial hydrodynamic force moving particulates toward the flow axis (25), even at a low Reynolds number. Thus, not all cells flow along a capillary wall (maximal shear rate) or capillary axis (zero shear rate). Consequently, there was a range of shear experienced by flowing cells, and the maximal shear rate does not represent the shear experienced by all cells. For a simple approximation, we have assumed an average shear rate (between the maximum and minimum shear rate experienced by cells) for the four flow rates to be 375, 1150, 2300, and 4600 s Ϫ1 .  Fig. 2 shows a schematic of capillary shear. Typically, a 1-ml sample containing 1 ϫ 10 6 U937 cells was aspirated into a 1-ml computerdriven syringe. After the sample was loaded into a syringe, the sample was pushed into a FEP tube at one of the four flow rates to generate shear. When that cycle was completed, the sample was aspirated into the same syringe. This cycle was repeated five times. After the fifth cycle, a computer-operated solenoid valve (NResearch, Caldwell, NJ), used to separate the shear FEP line from an FEP line leading to a FACScan, was switched to allow samples to be pushed toward a FAC-Scan at 1 l/s. Flow Cytometry and Data Analysis-Flow cytometric analysis was done on a Becton-Dickinson FACScan flow cytometer (BD Biosciences). Data acquisition was performed using CellQuest (BD Biosciences). Data were analyzed offline using the Windows Multiple Document Interface Flow Cytometry Interface (Scripps, La Jolla, CA). Time and fluorescence information were extracted from the data using FacsQuery software, developed by Bruce Edwards. Peak analysis and data fitting were done using PeakFit version 4.11 (Systat, Point Richmond, CA) and GraphPad Prism 4 (GraphPad, San Diego, CA), respectively.

VLA-4 Affinity Modulation by Shear
A ligand dissociation analysis would not readily distinguish heterogeneity in the affinity of resting and activated receptors on a given cell as compared with heterogeneity in the distribution of receptors on activated and resting cells. However, the distribution of the amount of ligand bound would distinguish cells that had activated receptors from cells that did not. Thus, we have analyzed cell distributions before and after activation as shown in Fig. 3, regions A and B. The same principles were used for the analysis of ligand binding and Ca 2ϩ response. For this analysis, a Gaussian curve was fitted to the mean channel fluorescence distribution obtained from region A, the resting state of cells. Region B was fitted with two Gaussian curves. One fit used the peak centroid and the full-width half maximum of Region A. The peak height was allowed to vary. This component represents resting cells. A second Gaussian curve was fitted to the remainder of the distribution in which the centroid and peak height were allowed to vary but full-width halfmaximum was fixed using the fit values obtained from Region A. The second curve represented activated cells. A simultaneous two-Gaussian fit to the mean channel fluorescence distribution obtained from Region B was done. The ratio of the total events under the two histograms was taken to estimate the fraction of cells activated under shear ((resting Ϫ activated)/activated).

Fluid Forces Increase the Affinity of the ␣ 4 ␤ 1 Integrin in Real
Time-Studies were conducted in a turbulent fluid flow environment using a Fischer Scientific minivortexer. To determine whether shear can affect the affinity of VLA-4, we used the LDV-FITC probe (17). Prior to applying shear, U937 cells (1 ϫ 10 6 cells/ml) were equilibrated with 4 nM probe. The concentration chosen for the experiments was below the dissociation constant (K d of ϳ12 nM) for probe binding to resting VLA-4 and above the K d for the physiologically activated receptor (K d of ϳ1-2 nM) (17). Therefore, the transition from the low affinity to the high affinity receptor leads to an increased binding of the probe (from ϳ25 to ϳ75% of receptor occupancy). Fig. 4 shows the rapid and transient increase in probe binding to sheared cells. The binding of the probe was detected after data acquisition was re-established, indicating that seconds were needed  (18). Fig. 4 shows that Mn 2ϩ increased probe binding above the level detected for shear.

Affinity Changes in a Controlled Fluid Force Environment-
The range of shear rates was narrowed with computer-driven syringes and capillary tubes (see "Experimental Procedures"). Fig. 3, A and B, shows the kinetics of intracellular Ca 2ϩ (determined using Fluo-4) and VLA-4 probe binding in response to the different levels of shear rates. The percentage of activated cells was calculated from Regions A and B in Fig. 3, A and B, as described under "Experimental Procedures." The resting state and shear histograms were normalized to the largest value in each of their distributions. Fig. 3, C and D, show how shear affected the number of activated cells. Those results are quantified in Fig. 3E, where the fraction of activated cells versus shear stress, fit to a hyperbolic equation, were comparable for both the LDV-FITC probe and intracellular Ca 2ϩ responses.
Simultaneous Observation of Integrin Activation and Intracellular Ca 2ϩ Elevation in Response to Fluid Forces-To follow VLA-4 affinity changes simultaneously with intracellular Ca 2ϩ responses, the cells were stained with both Fura Red and the LDV-FITC probe. In several experiments, we used Fluo-4 to detect intracellular Ca 2ϩ and VLA-4 activity in parallel. Fig.  5A shows Ca 2ϩ and LDV-FITC binding responses after U937 cells were vortexed at 3200 rpm for 5, 15, and 30 s. The Fura Red fluorescent signal decreased as the intracellular Ca 2ϩ concentration increased (the Fura Red axis in Fig. 5A is inverted). A transient and dose-dependent increase in intracellular Ca 2ϩ was accompanied by an increase in the binding of the LDV-FITC probe. The kinetics of probe binding was similar, but the amplitude of signal was dependent on the duration of shear, reflecting differences in the number of activated cells (see "Affinity Changes in a Controlled Fluid Force Environment" and Fig. 3). Intracellular Ca 2ϩ and Integrin Affinity Changes-To show the effect of intracellular Ca 2ϩ on VLA-4 affinity, we activated cells through their G-protein-coupled receptors (GPCR), added Ca 2ϩ ionophores (ionomycin and A23187), and chelated intracellular Ca 2ϩ with BAPTA. It is known that VLA-4 can be activated through formyl peptide, CXCR2, CXCR4, and CCR3 receptors (17). Here, we took advantage of nucleotide receptors constitutively expressed on U937 cells (P 2Y2 and P 2Y6 ) (26 -28) TABLE I LDV-FITC probe dissociation rates for U937 cells treated with different concentrations of divalent cation, N-formyl-Met-Leu-Phe-Phe, or exposed to shear Experiments were performed with U937 cells expressing a nondesensitizing/noninternalizing formyl peptide receptor mutant. 10 6 cells/ml were equilibrated for 5 min in a 37°C water bath with 12 or 16 nM LDV-FITC probe in HEPES buffer (supplemented with 0.1% human serum albumin and 10 mM Ca 2ϩ ϩ 1 mM Mg 2ϩ ). After obtaining a base line, the sample was removed from the flow cytometer and vortexed or not for 10 s, and the LDV-FITC probe was added. Data were averaged for two experiments. The numbers in parenthesis represent the fraction of slow (76%) and fast (24%) dissociation in response to shear. Fractions were obtained from fitting dissociation data to a two-exponential fit, with the fast rate fixed to the unvortexed resting state.

VLA-4 Affinity Modulation by Shear
that bind ATP to mediate a rapid and transient increase in intracellular Ca 2ϩ (28). Fig. 6 shows that the addition of 1 M ATP results in rapid increases in the Ca 2ϩ signal with slower LDV-FITC probe binding of amplitude similar to 30 s of vortexing. The binding of the LDV-FITC probe to the cells was limited by the rate of probe binding (k on ϳ3-5 ϫ 10 6 M Ϫ1 s Ϫ1 ) and was somewhat slower than the actual VLA-4 activation rate (for comparison, see the probe binding kinetics in response to Mn 2ϩ ) (Fig. 4) (17,18).
The dissociation of the LDV-FITC probe followed the slow decrease in the intracellular Ca 2ϩ measured using Fura Red. This slow decay (ϳ50 s) reflected the kinetics of restoration of VLA-4 basal activity and was slower than probe dissociation from the resting state ϳ0.06 s Ϫ1 (half-life of ϳ11 s) (17). Thus, the kinetics of VLA-4 activation on U937 cells coincides with the kinetics of intracellular Ca 2ϩ signaling when the cell was activated through GPCR. The data were consistent with a resting Ca 2ϩ concentration between 10 and 100 nM with elevation to ϳ1000 nM following activation.
We used the Ca 2ϩ ionophore ionomycin to increase the intracellular Ca 2ϩ concentration. Ionomycin acts as a mobile ion carrier across membranes and was used as a Ca 2ϩmobilizing agent (29). After establishing a sample base line for 1 min, ionophores (1 M ionomycin in Fig. 7A and 10 g/ml A23187 in Fig. 7, B and C) were added. Cell activation was prevented during mixing by gently inverting the sample. Fig.  7A shows that ionomycin activated VLA-4 in the presence of 1 and 10 mM extracellular Ca 2ϩ , and the time course of the Ca 2ϩ elevation was similar to the time course of VLA-4 activation. An increase in the extracellular Ca 2ϩ concentration alone did not change the total binding of the LDV-FITC probe. Since both intracellular Ca 2ϩ conditions led to similar total probe binding, it was likely that the two conditions had the same affinity state. However, the decay phase of the integrin activation was ϳ3 times longer in 10 mM Ca 2ϩ , suggesting that VLA-4 activation was strongly intracellular Ca 2ϩ -dependent.
Intracellular Ca 2ϩ was chelated by incubating cells with BAPTA. Then A23187 was added to elevate intracellular free Ca 2ϩ (Fig. 7, B and C) and detected as a decrease in Fura Red fluorescence corresponding to an alteration from resting to elevated (ϳ1000 nM) Ca 2ϩ levels. The binding of the LDV-FITC probe increased at the same time (Fig. 7B). Buffering intracellular Ca 2ϩ with BAPTA allowed A23187 to induce a slow increase in the intracellular Ca 2ϩ and LDV-FITC probe binding. Thus, the amount of the BAPTA (100 M) loaded inside the cells was nearly sufficient to completely buffer Ca 2ϩ influx. The slow increase in the binding of the LDV-FITC probe coincides with a slow increase in the intracellular free Ca 2ϩ .
Effect of Fluid Forces on the LDV-FITC Probe Dissociation Rate-We measured LDV-FITC dissociation rates of vortexed cells to characterize VLA-4 affinity under conditions where the duration of VLA-4 activation corresponds to the duration of the intracellular Ca 2ϩ response (ϳ100 s). Cells were incubated in 10 mM Ca 2ϩ , where the calcium signal lasts long enough to measure the LDV-FITC probe dissociation rate under shear (see Fig. 7A). The results are summarized in Table I and compared with a range of values found for other modes of VLA-4 activation. We found that the dissociation behavior after 10 s of vortexing required two exponential curves (fast and slow components) to fit the data. The fraction of the sites that appeared to remain in the resting state (fast component) was 24%, whereas the remaining sites exhibited a dissociation rate 4 times slower (activated state). The rate was comparable with the physiological GPCR activation pathway or divalent cation conditions (10 mM Ca 2ϩ and 1 mM Mn 2ϩ ). Intracellular pathways activated through extracellular stimuli (N-formyl-Met-Leu-Phe-Phe, interleukin-5, or IgE) all lead to VLA-4 of a similar affinity (17) and presumably in an extended conformational state of higher avidity. Our data suggest that intracellular signaling also occurs when cells are subjected to shear. Consequently, VLA-4 is activated to a similar affinity state as those generated from physiological stimuli. Fig. 8 shows the simultaneous LDV-FITC probe and intracellular Ca 2ϩ response to shear in cells incubated with and without BAPTA. Cells that were treated with BAPTA do not respond to shear, whereas

VLA-4 Affinity Modulation by Shear
untreated cells do. Our results indicate that VLA-4 activation in response to shear was downstream of Ca 2ϩ signaling and that an increase in intracellular Ca 2ϩ was associated with activation of VLA-4.
Pertussis Toxin Effect on Ca 2ϩ Signaling and Integrin Affinity in Response to Fluid Force-Heterotrimeric G-proteins are part of a pathway that activates integrins (30). To determine whether heterotrimeric G-proteins were involved in the VLA-4 response to shear, U937 cells were treated with PTX. After establishing a base line of LDV-FITC probe binding, the sample was vortexed for 10 s, and sampling resumed (Fig. 9A). Treatment of the cells with PTX nearly abrogated the activation of VLA-4 by shear, suggesting that G␣ i -related signaling can be an intermediate step in a mechanosensing pathway for VLA-4 activation. To test this hypothesis, we activated the same PTXtreated cells using P 2Y receptors, constitutively expressed on U937 cells (26,27). These receptors are coupled to the G␣ q subunit/phospholipase C␤2 pathway (31)(32)(33), which is PTXresistant (34). To promote Ca 2ϩ signaling and VLA-4 activation, 1 M ATP was applied to PTX-treated and -untreated U937 cells. Whereas VLA-4 activation was reduced (Fig. 9B), the intracellular Ca 2ϩ response was retained. Thus, a functional G␣ i subunit was required for VLA-4 activation in response to shear but was not required for the intracellular Ca 2ϩ response.
To determine whether PTX-treated cells lose viability as represented by their capacity to respond through the G␣ q pathway, we examined the Ca 2ϩ dose-response curve for ATP (Fig.  10). A quantitative analysis was obtained by measuring the peak height of the Ca 2ϩ response (measured with respect to a base line defined to be the time course before the addition of ATP) after the addition of ATP. The time courses of the ATP dose curve for cells treated with and without PTX were the same. Thus, the data indicate that cells treated with PTX were not adversely affected when compared with untreated cells.

DISCUSSION
Fluid Forces, Intracellular Ca 2ϩ , and VLA-4 Affinity-We have previously detected the real time regulation of VLA-4 affinity by divalent cations, physiological signaling, and reducing agents. Here we have shown that VLA-4 affinity was elevated in the presence of shear and that the effect was rapid and transient (Figs. 3-6, 8, and 9). A significant fraction of the cells, correlating with the receptors on them, responded to shear. The affinity of VLA-4 produced by this pathway was indistinguishable from the affinity produced by GPCR signaling.
The kinetics of intracellular Ca 2ϩ signaling also corresponded to the time course of LDV-FITC binding to VLA-4 (Figs. 5 and 6) in the presence of shear. It was conceivable that the shorter vortex duration-induced responses (Fig. 4) as compared with the response to capillary fluid flow was due to shear produced during delivery through 0.03-inch internal diameter tubing that may preserve cells in an activated state for a longer period of time. In the absence of shear, Ca 2ϩ ionophores (ionomycin and A23187) regulated VLA-4 affinity. Moreover, increased intracellular Ca 2ϩ was always associated in time with increased LDV-FITC probe binding. The relevance of intracellular Ca 2ϩ response in the presence of shear was further demonstrated with BAPTA-AM, which abolished the VLA-4 response to shear (Fig. 8).
Response Pathways for VLA-4 Activation-GPCR stimulation affects cell adhesive avidity through a G␣ i dependent process (10). Since VLA-4 activation can involve G␣ i pathways (35)(36)(37)(38)  The bent and extended conformation of VLA-4 is shown; a more extended conformation has higher affinity for ligand (76). I, mechanical extension of VLA-4 can be accomplished by pulling a counterstructure ligand (VCAM-1). II, VLA-4 is extended directly by the flow of fluid. III, an unknown "shear sensor" transducing a signal through the G␣ i subunit and/or generating intracellular free Ca 2ϩ results in inside-out signaling and VLA-4 conformational change. IV, signaling leads to the activation of other integrin molecules ("inside-out" signaling). V, elevation of intracellular free Ca 2ϩ and VLA-4 activation through the G␣ q pathway. Crossed out arrows, integrin activation via the G␣ i pathway was blocked using PTX.

VLA-4 Affinity Modulation by Shear
examined the role of G␣ i response to shear (Fig. 9A) by pretreating cells with PTX. Because VLA-4 activation was associated with intracellular Ca 2ϩ signaling, we used ATP to initiate a Ca 2ϩ response for cells treated with PTX. Those cells were activated through P 2Y receptors (Fig. 11 (V)), which were coupled to G␣ q (PTX-resistant (31)(32)(33)(34)). Based on experiments with PTX, we observed that a functional G␣ i subunit was required for VLA-4 activation by shear but not for ATP (Fig. 10) or intracellular Ca 2ϩ signaling induced by shear (data not shown).
Integrins are one of four classes of mechanosensors (43, 44) that include ion channels (45), G-protein receptors (46), and tyrosine kinase receptors (47). Each can be associated with intracellular Ca 2ϩ signaling pathways (48 -50). Connections among the classes are illustrated by G i and G 12 /G 13 signaling pathways that are sufficient to activate ␣ IIb ␤ 3 receptors on platelets (30). Whereas G␣ q -mediated signaling is not essential for ␣ IIb ␤ 3 activation but is for Ca 2ϩ mobilization (Fig. 11 (V)), the overall mechanism connecting G protein receptors to integrin activation in platelets is unknown.
Fluid flow generated by a vortexer can affect suspended cells in several ways. Turbulent fluid motion produced stress on a cell membrane as a result of differential fluid velocities that can activate mechanosensors. In principle, fluid vortex motion can cause cells to collide in a nonbinding manner and activate receptors or the cell membrane. Alternatively, colliding U937 cells, potentially forming homotypic aggregates (doublets or triplets) through engagement of integrins and their ligands, would be subject to mechanical stress that would pull the aggregates apart and could initiate a cell signaling sequence and/or molecular extension. Cellular aggregates between VLA-4 and a U937 cellular ligand would be inhibited by the presence of LDV peptides binding specifically to VLA-4 (17). Whereas U937 homotypic aggregation involves ␤ 2 integrin (51, 52), our previous data (53) have shown that U937 at 3 ϫ 10 6 cells/ml in a shear environment exhibited no homotypic aggregates even in the absence of anti-␤ 2 antibodies. We have examined whether blocking CD18 binding using anti-bodies (TS1/18; Endogen, Woburn, MA) to block ␤ 2 integrin-dependent adhesion would affect intracellular Ca 2ϩ signaling and VLA-4. No signal reduction was observed (data not shown). Thus, formation of cellular aggregates and engagement of integrins was unlikely to account for significant outside-in signaling in our study.
A schematic diagram of potential mechanisms that may induce a higher integrin affinity state in the presence of shear is shown in Fig. 11. An integrin can be stretched under force by its counterstructure (Fig. 11 (I)) or directly by fluid flow (Fig. 11 (II)) and may increase its bond adhesion strength in a catch bond mechanism (54). The latter remains a viable option, since integrins are known to be flexible (55), and shear may lead to an extension similar to the extended chain conformation observed for von Willebrand factor (56). It is worth noting that the integrin binding partners talin and paxillin that regulate cell adhesion, migration, and integrin conformation (57)(58)(59)(60)(61)(62)(63) could provide a means of mechanotransduction. That signaling has been documented with a magnetic drag force (64) to extend integrin molecules, generating an intracellular calcium response, gene transcription (65) and tyrosine phosphorylation (66 -68).
Another mechanism could involve an outside-in signaling pathway and a mechanoreceptor (Fig. 11 (III)), such as an ion channel, tyrosine kinase, or G-protein-coupled receptors. Two lines of research support the existence of integrin activation through shear signaling (Fig. 11 (II and III)). First, shear rapidly stimulated ␣ v ␤ 3 via a small GTPase Rho signaling pathway (16) and caused an increase in the avidity of ␣ v ␤ 3 and ␣ 5 ␤ 1 integrin bearing cells to the extracellular matrix (69). Further, shear promoted lymphocyte migration across vascular endothelium in an ␣ 4 ␤ 1 -and ␣ L ␤ 2 -dependent manner, and shear-induced signal was coupled to G␣ i (70). The participation of intracellular Ca 2ϩ pathways in integrin activation (Fig. 11 (IV)) was shown for ␤ 1 integrin activation (71), for intracellular Ca 2ϩ elevation induced by shear (72), and through identification of a novel calcium and integrin binding in ␤ 3 integrin activation (73).
Catch Bond: A Cellular Braking Mechanism-Our results were consistent with shear-induced mechanotransduction resulting in intracellular Ca 2ϩ signaling and VLA-4 activation. The new VLA-4 affinity state observed under fluid flow was the same one induced by GPCR signaling, which was shown previously to increase the length of the VLA-4 molecule, to decrease the cellular avidity, and to decrease the ligand dissociation rate (17,18). These VLA-4 structural and functional changes appear to parallel the global conformational rearrangement of the extracellular domains induced by ligands and divalent cation (74) and the switchblade model for the ␣ v ␤ 3 integrin based on electron microscopy, NMR, and epitope exposure data (75). Using fluorescence resonance energy transfer and the LDV-FITC probe (21), we found a striking correlation between the degree of VLA-4 extension and its affinity (76). The prediction that force could increase adhesion bond strengths, catch bond (54), was verified by atomic force microscopy of P-selectin binding to P-selectin glycoprotein ligand-1 (77). We hypothesize that extension of an integrin could also be part of a braking system in leukocyte rolling (78) and that shear could play a role in the pathways shown in Fig. 11 (I, II, and III). We have obtained direct evidence for the first of these. 2