Regulation of Ca2+-dependent K+ Current by αvβ3 Integrin Engagement in Vascular Endothelium*

Interactions between endothelial cells and extracellular matrix proteins are important determinants of endothelial cell signaling. Endothelial adhesion to fibronectin through αvβ3 integrins or the engagement and aggregation of luminal αvβ3 receptors by vitronectin triggers Ca2+ influx. However, the underlying signaling mechanisms are unknown. The electrophysiological basis of αvβ3 integrin-mediated changes in endothelial cell Ca2+ signaling was studied using whole cell patch clamp and microfluorimetry. The resting membrane potential of bovine pulmonary artery endothelial cells averaged -60 ± 3 mV. In the absence of intracellular Ca2+ buffering, the application of soluble vitronectin (200 μg/ml) resulted in activation of an outwardly rectifying K+ current at holding potentials from -50 to +50 mV. Neither a significant shift in reversal potential (in voltage clamp mode) nor a change in membrane potential (in current clamp mode) occurred in response to vitronectin. Vitronectin-activated current was significantly inhibited by pretreatment with the αvβ3 integrin antibody LM609 by exchanging extracellular K+ with Cs+ or by the application of iberiotoxin, a selective inhibitor of large-conductance, Ca2+-activated K+ channels. With intracellular Ca2+ buffered by EGTA in the recording pipette, vitronectin-activated K+ current was abolished. Fura-2 microfluorimetry revealed that vitronectin induced a significant and sustained increase in intracellular Ca2+ concentration, although vitronectin-induced Ca2+ current could not be detected. This is the first report to show that an endothelial cell ion channel is regulated by integrin activation, and this K+ current likely plays a crucial role in maintaining membrane potential and a Ca2+ driving force during engagement and activation of endothelial cell αvβ3 integrin.

Interactions between endothelial cells and extracellular matrix proteins are important determinants of endothelial cell signaling. Endothelial adhesion to fibronectin through ␣ v ␤ 3 integrins or the engagement and aggregation of luminal ␣ v ␤ 3 receptors by vitronectin triggers Ca 2؉ influx. However, the underlying signaling mechanisms are unknown. The electrophysiological basis of ␣ v ␤ 3 integrin-mediated changes in endothelial cell Ca 2؉ signaling was studied using whole cell patch clamp and microfluorimetry. The resting membrane potential of bovine pulmonary artery endothelial cells averaged ؊60 ؎ 3 mV. In the absence of intracellular Ca 2؉ buffering, the application of soluble vitronectin (200 g/ml) resulted in activation of an outwardly rectifying K ؉ current at holding potentials from ؊50 to ؉50 mV. Neither a significant shift in reversal potential (in voltage clamp mode) nor a change in membrane potential (in current clamp mode) occurred in response to vitronectin. Vitronectin-activated current was significantly inhibited by pretreatment with the ␣ v ␤ 3 integrin antibody LM609 by exchanging extracellular K ؉ with Cs ؉ or by the application of iberiotoxin, a selective inhibitor of large-conductance, Ca 2؉ -activated K ؉ channels. With intracellular Ca 2؉ buffered by EGTA in the recording pipette, vitronectin-activated K ؉ current was abolished. Fura-2 microfluorimetry revealed that vitronectin induced a significant and sustained increase in intracellular Ca 2؉ concentration, although vitronectin-induced Ca 2؉ current could not be detected. This is the first report to show that an endothelial cell ion channel is regulated by integrin activation, and this K ؉ current likely plays a crucial role in maintaining membrane potential and a Ca 2؉ driving force during engagement and activation of endothelial cell ␣ v ␤ 3 integrin.
Integrins are a large family of membrane-spanning, cell adhesion proteins composed of ␣ and ␤ subunits, with over 18 ␣ and 8 ␤ subunits combining to form more than 24 different heterodimers (1,2). In the vascular system, integrins play various roles in coordinating cell function, such as adhesion, spreading, and migration (3,4). The vitronectin (VN) 1 receptor, ␣ v ␤ 3 integrin, is expressed both luminally and abluminally on endothelium (5,6) and is thought to play an important role in several vascular pathologies (4,(7)(8)(9). Soluble ␣ v ␤ 3 integrin ligands are also capable of acutely regulating vascular tone. For example, synthetic peptides containing the arginine-glycine-aspartic acid (RGD) sequence that binds ␣ v ␤ 3 integrin have been shown to block flow-induced, endothelium-dependent vasodilation in coronary arterioles (10). It has therefore been suggested that soluble ␣ v ␤ 3 integrin ligands may acutely modulate blood flow by interacting with endothelial cell (EC) integrins (11,12).
VN is a plasma glycoprotein first identified as a ligand of ␣ v ␤ 3 integrin and circulates at a concentration of 200 -400 g/ml in normal human plasma (13). VN can potentially interact with unbound ␣ v ␤ 3 integrin, and plasma levels of VNcontaining complement complexes increase after complement activation. Adhesion to VN-covered substrates leads to an increase in EC intracellular Ca 2ϩ concentration ([Ca 2ϩ ] i ) (14). Acute application of VN or cross-linking of ␣ v ␤ 3 integrin with the ␣ v ␤ 3 antibody, LM609, also increases EC [Ca 2ϩ ] i (15). The [Ca 2ϩ ] i increase occurs as a consequence of tyrosine phosphorylation of phospholipase C-␥1 following ␣ v ␤ 3 activation. It has been proposed that stimulation of ␣ v ␤ 3 integrin activates an unidentified Ca 2ϩ influx pathway (14,15).
Although integrins have been shown to regulate ion channels, including Ca 2ϩ and K ϩ channels in other tissues (16 -20), it is not known if integrins regulate EC ion channels. The rapid effects of integrin activation on EC [Ca 2ϩ ] i suggest that ion channels are involved in this response, and the purpose of the present study was to directly test this idea.
Electrophysiology-Conventional whole cell current recordings were performed (21) using an EPC-9 patch clamp system (HEKA Elektronik, Lambrecht, Germany). Continuous recordings of membrane potential (E m ), membrane current (I m ), cell capacitance, and/or seal resistance were simultaneously monitored using Pulse and X-Chart software (HEKA). Patch pipettes were pulled from borosilicate glass (catalog number 7040, Sutter Instruments, Novato, CA) with resistances of 3-8 megaohms. To determine current-voltage (I-V) relationships, 1-s voltage ramps from Ϫ120 to ϩ60 mV or voltage steps between Ϫ120 and * This work was supported by National Institutes of Health Grants HL-60180 (to M. J. D.) and HL-59971 to (to G. E. D.). The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
¶ To whom correspondence should be addressed: ϩ20 mV in 20-mV increments (200 ms duration) were applied. Data were fitted using IGOR analysis routines (Wavemetrics, Lake Oswego, OR).
Solutions and Reagents-The compositions of all solutions are listed in Table I. All experiments were performed at 35°C using a bath temperature controller (HCC-100, Dagan Corp., Minneapolis, MN). The calculated equilibrium potentials for the standard experimental conditions, solutions I-A (solution I in the pipette and solution A in the bath), were Ϫ84.1 mV for K ϩ (E K ) and Ϫ32.8 mV for Cl Ϫ (E Cl ) at 35°C.
Bradykinin, thapsigargin, and La 3ϩ were added to the appropriate solutions at the concentrations indicated in the Fig. 6 legend and applied to individual cells from wide-tipped glass micropipettes connected to a Picospritzer (Parker-Hannifin, Fairfield, NJ). The low shear stress associated with local solution exchange did not in itself induce changes in current (n ϭ 4).
Fura-2 Microfluorimetry-Measurements of Fura-2 fluorescence were performed as described previously (22). Briefly, BPAECs on coverslips were loaded with 10 mol/liter Fura-2/AM in solution A for 30 min at 25°C, followed by washing for 10 min at 35°C. Fluorescence emission at 510 nm associated with alternating 340/380-nm excitation (at ϳ15 Hz) was used as an index of [Ca 2ϩ ] i (23). Calibrations were performed using Fura-2 (pentapotassium salt) and the Molecular Probes calibration kit. Solution A was used for all Fura-2 experiments on cells, except for Ca 2ϩ -free bath protocols (solution C).
Data Analysis-Analysis of variance was used to determine statistically significant differences between I-V curves. Scheffe's post hoc tests (24) were used to compare differences between different groups at the same holding potentials. Paired and unpaired t tests were used to test differences between groups for E m or the reversal potential for whole cell current (E rev ). Values of p Ͻ 0.05 were considered statistically significant.

RESULTS
Localization of ␣ v ␤ 3 and ␣ 5 ␤ 1 Integrin on BPAECs-Consistent with previous reports (5, 6), both ␣ v ␤ 3 and ␣ 5 ␤ 1 integrins were expressed on BPAECs (Fig. 1). ␣ v ␤ 3 integrin was preferentially localized to focal adhesions at the tips of actin filaments ( Fig. 1A), whereas ␣ 5 ␤ 1 integrin tended to be distributed around the cell center ( Fig. 1B), possibly at fibrillar adhesions (25). Both types of integrins were abundant on the surface of BPAECs under the conditions used in our experiments.
Vitronectin Activates a Whole Cell Current in BPAECs-The basic characteristics of unstimulated BPAECs in solutions I-A were as follows: cell capacitance ϭ 29.8 Ϯ 1 pF (n ϭ 56); E m ϭ Ϫ60.1 Ϯ 3 mV (n ϭ 42); E rev ϭ Ϫ57.5 Ϯ 3 mV (n ϭ 52). In solutions I-A, BPAECs (n ϭ 52) exhibited an inward rectifying K ϩ current as the predominant resting current with mean ϭ Ϫ12.6 Ϯ 1 pA/pF at holding potential (V H ) ϭ Ϫ120 mV. At V H ϭ Ϫ50 mV, VN (200 g/ml) activated a significant outward current (from 0.2 Ϯ 0.1 to 5.0 Ϯ 1 pA/pF) in 16 of 22 (73%) cells ( Fig. 2A). The I-V relationships for the cells at rest and after activation of current by VN, as determined using voltage ramps, are shown in Fig. 2B. Whole cell currents in the presence of VN were outwardly rectifying and relatively large at voltages from Ϫ50 to ϩ50 mV.
The net current evoked by VN, after subtracting the resting inward rectifying current, is shown in Fig. 2, B and C. VN significantly increased the amplitude of the whole cell difference current from 1.7 Ϯ 1 to 30.1 Ϯ 5 pA/pF at V H ϭ ϩ60 mV (n ϭ 16). This was associated with a slight shift in E rev to more negative potentials (Ϫ66.4 Ϯ 2 to Ϫ70.6 Ϯ 1 mV; n ϭ 16). In a different series of experiments, E m was measured under current clamp conditions (n ϭ 11). There was no significant difference between the average E m at rest, Ϫ64.4 Ϯ 5 mV, and during VN stimulation, Ϫ65.3 Ϯ 5 mV. These data suggest that the resting E m of BPAECs is determined by a K ϩ current and that VN activates a whole cell current whose E rev is close to E K .
Vitronectin-activated Current Is Inhibited by ␣ v ␤ 3 Integrin Ab-To confirm that VN-activated current was mediated by ␣ v ␤ 3 integrin engagement, BPAECs were pretreated with integrin-specific antibodies prior to VN application. In the presence of LM609 (200 g/ml), VN-induced current was completely inhibited ( Fig. 3A; n ϭ 5). However, pretreatment with ␣ 5 ␤ 1 integrin Ab did not have a significant effect on VN-induced current ( Fig. 3B; n ϭ 4). These results suggest that VN-activated current results from the specific activation of ␣ v ␤ 3 integrin.
Acute application of soluble LM609 alone failed to elicit any change in current (n ϭ 5; data not shown). In a previous study of BPAECs, aggregation of ␣ v ␤ 3 integrin was required to initiate Ca 2ϩ influx and [Ca 2ϩ ] i increases (15). To test if aggregation of ␣ v ␤ 3 integrin by LM609 would stimulate current, BPAECs were pretreated with LM609 in solution A for 30 min at 25°C, followed by acute application of secondary Ab (IgG, 200 g/ml). However, this procedure failed to significantly activate current (n ϭ 5, data not shown).
Vitronectin-activated Current Is Carried by K ϩ -The negative E rev of VN-induced current suggested that the current was carried primarily by K ϩ . To test this assumption further, extracellular K ϩ was substituted with solutions I-B containing Cs ϩ , an inhibitor of multiple K ϩ channels (26). Cs ϩ bath solution completely blocked both the inward rectifying current at FIG. 1. Both ␣ v ␤ 3 and ␣ 5 ␤ 1 integrins are present in BPAECs. Immunofluorescence staining revealed the presence of both ␣ v ␤ 3 (A) and ␣ 5 ␤ 1 (B) integrins (both shown in green) on the surface of BPAECs. Actin filaments were stained with phalloidin-rhodamine (red). A control in which no primary Ab was added is shown in panel C. Bar, 5 m. rest and the VN-activated current ( Fig. 4A; n ϭ 8). In addition, VN did not significantly shift E rev from its more depolarized initial value (Ϫ46.3 Ϯ 4 mV in Cs ϩ bath; Ϫ43.3 Ϯ 5 mV during VN). The effect of iberiotoxin (IBTX), a selective inhibitor of large conductance Ca 2ϩ -activated K ϩ (K Ca ) channels, was subsequently tested. In standard bath solution (solution A), IBTX (100 nmol/liter) completely inhibited the VN-activated outward current without significantly affecting the inward rectifying component of current (Fig. 4B).
To further test if the VN-induced current was carried by K ϩ , the net VN-induced current was measured at various extracellular K ϩ concentrations ([K ϩ ] o ) equalling 5.9, 30, or 120 mmol/ liter, and E rev was plotted as a function of [K ϩ ] o (Fig. 4C). As determined from the slope of the graph, E rev changed by 50.5 mV per 10-fold increase in [K ϩ ] o (Fig. 4C; solid line). The predicted slope for a pure K ϩ current at 35°C is 61.1 mV per 10-fold increase in [K ϩ ] o (Fig. 4C; dotted line). In contrast, changing [Cl Ϫ ] i by switching from solutions I-A (E Cl ϭ Ϫ32.8 mV) to III-A (E Cl ϭ Ϫ1.1 mV), with [K ϩ ] o ϭ 5.9 mmol/liter in both cases, did not significantly shift E rev (Fig. 4C). Collectively, these results suggest that VN-evoked current is almost exclusively carried by K ϩ through a large conductance K Ca channel.
Vitronectin-activated Current Requires an Increase in Intracellular Ca 2ϩ -Agonist-induced increases in [Ca 2ϩ ] i are known to activate K ϩ current in ECs (27), so the role of intracellular Ca 2ϩ in mediating the effects of VN was tested. When intra-cellular Ca 2ϩ was strongly buffered with 5 mmol/liter EGTA in the pipette (solutions II-A), VN failed to activate any significant current at V H ϭ Ϫ50 mV (Fig. 5A). Periodic voltage ramps from Ϫ120 mV to ϩ60 mV confirmed that there was no activation of current by VN at other potentials (Fig. 5B). The average I-V relationships with and without EGTA in the pipette solution are shown in Fig. 5C. Thus, VN-evoked current was completely abolished when intracellular Ca 2ϩ was strongly buffered by EGTA. These results suggest that an increase in [Ca 2ϩ ] i is required for activation of K ϩ current by VN.

Mechanism of the VN-induced [Ca 2ϩ ] i Increase-
To test if VN would activate a non-voltage-gated Ca 2ϩ current (I Ca ) that might contribute to secondary activation of a K Ca channel, protocols were performed using solutions previously used to record store-operated cation currents (28). Using solutions IV-D, in which Ca 2ϩ was the primary charge carrier (28,29), I Ca averaged Ϫ0.09 Ϯ 0.01 pA/pF (V H ϭ Ϫ120 mV) and E rev averaged Ϫ86 Ϯ 6 mV in unstimulated cells (Fig. 6, A and F). One minute after application of VN (200 g/ml), I Ca was Ϫ0.07 Ϯ 0.02 pA/pF (V H ϭ Ϫ120 mV; Fig. 6, B and F) and E rev was Ϫ89 Ϯ 11 mV. Neither value was significantly different from its corresponding value in unstimulated cells.
Due to the absence of a clear response to VN, bradykinin (BK), an agonist known to induce Ca 2ϩ release and influx (30,31), and thapsigargin (TG), an inhibitor of the endoplasmic reticulum Ca 2ϩ pump (28, 32), were used as positive controls. One minute after application of BK (1 M), I Ca averaged Ϫ0.13 Ϯ 0.04 pA/pF (V H ϭ Ϫ120 mV; Fig. 6, C and F) and E rev averaged Ϫ76 Ϯ 10 mV. Neither value was significantly different from its corresponding value in unstimulated cells. In contrast, when TG (1 M) was preloaded into the patch pipette, there was a rapid activation of inward current, which then partially inactivated and stabilized after ϳ3 min (Fig. 6E). I Ca peaked at Ϫ0.87 Ϯ 0.08 pA/pF, and E rev shifted to ϩ3 Ϯ 1 mV (V H ϭ Ϫ120 mV; Fig. 6, D and F) within 30 s after cell membrane rupture by pipette suction (Fig. 6E). The addition of 50 M La 3ϩ to the bath solution inhibited current activation (Fig.  6E) from Ϫ0.87 Ϯ 0.08 to Ϫ0.09 Ϯ 0.01 pA/pF (V H ϭ Ϫ120 mV) and shifted E rev from ϩ2.8 Ϯ 1 back to Ϫ86.8 Ϯ 2 mV (n ϭ 5).
To test if VN would elevate [Ca 2ϩ ] i , even though VN-induced I Ca activation could not be detected, VN was applied to nonvoltage clamped BPAECs loaded with Fura-2/AM. Indeed, VN (200 g/ml) induced a significant [Ca 2ϩ ] i increase in BPAECs (Fig. 7A), as reported previously (15). LM609 alone did not significantly alter [Ca 2ϩ ] i , as reported in the same study (15), nor was a significant [Ca 2ϩ ] i increase detected in LM609-pretreated cells when a secondary Ab (anti-mouse IgG, 200 g/ml) was applied to aggregate ␣ v ␤ 3 integrin (although several different durations and temperatures for LM609 pretreatment were tested). Because the magnitude of the VN-induced [Ca 2ϩ ] i increase was rather modest, BK and TG were used as positive controls known to elevate EC [Ca 2ϩ ] i (31,33). Both 1 mol/liter BK (Fig. 7B) and 1 mol/liter TG (Fig. 7C) induced significant and sustained [Ca 2ϩ ] i increases, with the TG response (210 Ϯ 22%, n ϭ 6) being substantially higher than that for VN (118 Ϯ 4%, n ϭ 8) or BK (138 Ϯ 12%, n ϭ 5) (Fig. 7D). These sustained [Ca 2ϩ ] i increases required Ca 2ϩ influx, because they were prevented when [Ca 2ϩ ] o was absent from the bath solution (solution C; Fig. 7C). DISCUSSION A novel finding of this study is that the interaction between ␣ v ␤ 3 integrin and the extracellular matrix protein VN results in rapid activation of at least one type of endothelial cell ion channel. The following observations suggest that the major VN-activated current is a K Ca current. (i) Neither E rev nor E m significantly shifted from E K (Ϫ84.1 mV) during VN application. (ii) E rev for VN-activated current shifted with changes in [K ϩ ] o close to what would be predicted for a pure K ϩ current. (iii) VN-activated K ϩ current was completely inhibited by buffering intracellular Ca 2ϩ . (iv) VN-activated K ϩ current was blocked by Cs ϩ and IBTX. Activation of the K Ca current by VN did not appear to substantially hyperpolarize cultured BPAECs due to their already negative resting membrane potentials (Ϫ64 mV), but it would be predicted to hyperpolarize other ECs with more positive resting potentials (34 -36). Although a Ca 2ϩ current directly activated by VN could not be detected, our collective electrophysiological and microfluorimetric data suggest that a Ca 2ϩ -permeable channel is also activated by VN in these cells and that its activation underlies the plateau phase of the VN-stimulated [Ca 2ϩ ] i increase. These results show, for the first time, that ion channels in ECs are rapidly activated in response to EC integrin interaction with an extracellular matrix protein. This points to a new mechanism for the control of vascular resistance under physiological conditions and during adaptive responses to vascular injury.
Mechanisms of Activation of K ϩ Current by ␣ v ␤ 3 Integrin-The mechanism underlying VN-activated K ϩ current may be similar to those of other K ϩ currents activated by agonists. By measuring whole cell current and [Ca 2ϩ ] i simultaneously in bovine aortic EC (BAEC), Himmel et al. (27) found that several agonists, including BK, induced [Ca 2ϩ ] i increases and subse- quently activated K ϩ current. BK-induced current was completely abolished after intracellular Ca 2ϩ chelation with EGTA, similar to our results for VN. There are some differences, however, between the characteristics of the current activated by BK and VN. In BAEC, BK not only activated a K ϩ current but also Cl Ϫ and non-selective cation currents that were dependent on the [Ca 2ϩ ] i increase. Thus, BK shifted E rev from Ϫ79 to Ϫ63 mV, away from E K (Ϫ83 mV) (27). In our study of BPAECs, VN tended to shift E rev toward a more negative potential (Ϫ66 mV to Ϫ71 mV), which is consistent with more selective activation of a K ϩ current. Indeed, significant activation of another current during K ϩ channel inhibition by Cs ϩ could not be detected, and IBTX, a selective blocker of large conductance K Ca channels, inhibited the VN-activated current.
Recent studies have revealed that integrin ligands acutely regulate several types of ion channels (16 -20). For example, integrins have been found to modulate K Ca current in at least two different types of cells. Attachment of fibronectin-coated beads to leukemia cells led to hyperpolarization (37), an effect most likely mediated by activation of K Ca current. However, this effect was not completely abolished by strong intracellular Ca 2ϩ buffering (38); therefore, the K Ca current was probably not activated by an increase in [Ca 2ϩ ] i but by another mechanism such as channel phosphorylation. Mechanical stimulation of human articular chondrocytes through ␤1 integrins was shown to induce membrane hyperpolarization by autocrine production of interleukin-4 (39,40). The hyperpolarization in that case was due to activation of apamin-sensitive, small conductance K Ca channels and was blocked by pretreatment with ␤1 integrin Ab (41). It is not clear whether an increase in [Ca 2ϩ ] i was critical to that response, because [Ca 2ϩ ] i changes were not measured or prevented in those experiments. Collectively, these studies suggest that integrin engagement and activation can result in K Ca channel activation by several different pathways.
Intracellular signal transduction by integrins usually involves tyrosine kinase activation (1). Indeed, several lines of evidence suggest that non-receptor tyrosine kinases directly regulate K Ca channels. In Chinese hamster ovary cells cotransfected with prolactin receptors and K Ca channels, the addition of prolactin led to channel activation that persisted after patch excision and was inhibited by an antibody to Janus tyrosine kinase 2 (JAK2) (42). These results suggest that the channels were directly regulated by JAK2 or a downstream kinase. In human embryonic kidney 293 cells, co-expression of K Ca channels with c-Src, another non-receptor tyrosine kinase, led to Ca 2ϩ -sensitive enhancement of K ϩ current (43). This enhancement was mediated by phosphorylation of residue Tyr-766 on the C-terminus of the channel's ␣-subunit. Interestingly, both phosphorylation of the channel and potentiation of current were more pronounced at high levels of intracellular Ca 2ϩ (43). It is possible that the K Ca current activated by VN-␣ v ␤ 3 interaction in BPAECs may also involve a Ca 2ϩ -sensitive, tyrosine kinase-dependent pathway. Whether this mechanism involves phosphorylation of a specific C-terminal tyrosine residue on the BK Ca channel remains to be investigated. Our observation that VN-activated K ϩ current is blocked by EGTA does not preclude direct regulation of the channel by tyrosine phosphorylation if that process also requires a relatively high [Ca 2ϩ ] i (43). Integrin-mediated Ca 2ϩ Signaling in ECs-Cheresh and coworkers (14) first reported VN-induced [Ca 2ϩ ] i increases in human umbilical vein ECs as they adhered to a VN-covered substrate. Based on the use of pharmacological inhibitors, the [Ca 2ϩ ] i increase was attributed to activation of a non-voltagedependent Ca 2ϩ channel. Bhattacharya et al. (15) demonstrated that both VN and cross-linking of ␣ v ␤ 3 integrin by LM609 increased [Ca 2ϩ ] i in BPAECs. The [Ca 2ϩ ] i increase occurred as a consequence of tyrosine phosphorylation of phospholipase C-␥1 and an increase in inositol 1,4,5-trisphosphate levels that would induce Ca 2ϩ release from Ca 2ϩ stores. However, Ca 2ϩ influx was required for sustained elevation in [Ca 2ϩ ] i . The mechanism for Ca 2ϩ entry was not shown, because neither study examined changes in the electrophysiological mechanisms underlying the [Ca 2ϩ ] i increase.
Conceptually, integrin activation could increase [Ca 2ϩ ] i by increasing the driving force for passive calcium entry or by activating an ion channel or transporter to alter Ca 2ϩ conductance. In other types of non-excitable cells, multiple lines of evidence suggest that integrin activation increases Ca 2ϩ conductance by activating a Ca 2ϩ -permeable ion channel. For example, thrombin application in human platelets leads to increased Ca 2ϩ channel activity via activation of ␣ IIb ␤ 3 integrin (44). In Madin-Darby canine kidney cells, the application of beads coated with the RGD peptide elicits an [Ca 2ϩ ] i increase that correlates with bead adhesion, suggesting the involvement of integrins in regulating Ca 2ϩ homeostasis (45,46).
Our observation that the E rev of VN-activated current did not shift exactly as predicted for a pure K ϩ current (Fig. 4C) suggests that another ion channel is also activated by VN. In non-excitable cells, including ECs, a store-operated Ca 2ϩ current (I Ca ) is considered to play a predominant role in controlling Ca 2ϩ influx (47)(48)(49). Some laboratories have been able to detect this current (28,50,51), whereas others have not (52,53). In ECs, the amplitude of I Ca is typically very small (ϳ Ϫ0.1 pA/pF at V H ϭ 0 mV), even in response to a non-physiological stimulus such as inositol 1,4,5-trisphosphate inclusion in the pipette (50). To detect I Ca following ␣ v ␤ 3 integrin activation, intracellular solutions containing weakly buffered free Ca 2ϩ and extracellular solutions containing high Ca 2ϩ in the absence of any other membrane-permeable cations (solutions IV-D) were used. Under the same conditions, supra-physiological stimulation with TG activated an easily detectable I Ca of Ϫ0.87 pA/pF (V H ϭ Ϫ120 mV). However, no significant I Ca could be detected in response to VN or BK, even though both agents produced small but sustained [Ca 2ϩ ] increases. The VN and BK responses could reflect intracellular Ca 2ϩ release only and/or weak activation of I Ca below the detection threshold of whole cell current recording.
In conclusion, we report for the first time that engagement of ␣ v ␤ 3 integrin by an extracellular matrix protein leads to activation of an endothelial cell K ϩ current. The current has the characteristics of a K Ca channel that is activated secondary to Ca 2ϩ influx and/or release. Activation of whole cell K Ca current following integrin engagement would hyperpolarize the endothelium, particularly in electrically coupled ECs in vivo, where the resting E m is relatively depolarized (54). This would sustain the plateau phase of the [Ca 2ϩ ] i transient by enhancing the electrochemical driving force for Ca 2ϩ (34 -36). Thus, ␣ v ␤ 3 engagement by extracellular matrix proteins is predicted to regulate production of Ca 2ϩ -dependent EC vasoactive substances such as endothelium-derived hyperpolarizing factor (EDHF), nitric oxide, prostacyclin, and endothelin and thereby acutely modulate vascular tone (10 -12). In this way, VN and other ␣ v ␤ 3 integrin ligands have the potential to acutely regulate the local circulation.