|Home | About | Journals | Submit | Contact Us | Français|
The inflammatory mediator thrombin proteolytically activates protease-activated receptor (PAR1) eliciting a transient, but reversible increase in vascular permeability. PAR1-induced dissociation of Gα subunit from heterotrimeric Gq and G12/G13 proteins is known to signal the increase in endothelial permeability. However, the role of released Gβγ is unknown. We now show that impairment of Gβγ function does not affect the permeability increase induced by PAR1, but prevents reannealing of adherens junctions (AJ), thereby persistently elevating endothelial permeability. We observed that in the naive endothelium Gβ1, the predominant Gβ isoform is sequestered by receptor for activated C kinase 1 (RACK1). Thrombin induced dissociation of Gβ1 from RACK1, resulting in Gβ1 interaction with Fyn and focal adhesion kinase (FAK) required for FAK activation. RACK1 depletion triggered Gβ1 activation of FAK and endothelial barrier recovery, whereas Fyn knockdown interrupted with Gβ1-induced barrier recovery indicating RACK1 negatively regulates Gβ1-Fyn signaling. Activated FAK associated with AJ and stimulated AJ reassembly in a Fyn-dependent manner. Fyn deletion prevented FAK activation and augmented lung vascular permeability increase induced by PAR1 agonist. Rescuing FAK activation in fyn−/− mice attenuated the rise in lung vascular permeability. Our results demonstrate that Gβ1-mediated Fyn activation integrates FAK with AJ, preventing persistent endothelial barrier leakiness.
A persistent increase in endothelial permeability during inflammatory conditions such as pneumonia, trauma, and burn leads to the life-threatening illness acute respiratory distress syndrome (Mehta and Malik, 2006; Liu and Matthay, 2008). Increased endothelial permeability occurs because of loss of cell–cell contacts and disruption of cell–extracellular matrix (ECM) adhesions (Yuan, 2002; Mehta and Malik, 2006). Focal adhesion kinase (FAK) and VE-cadherin play a fundamental role in establishing the endothelial barrier to macromolecules and liquid by maintaining intercellular adhesion and cell–ECM adhesivity (Nelson et al., 2004; van Buul et al., 2005; Wu, 2005; Mehta and Malik, 2006; Dejana et al., 2008; Rudini and Dejana, 2008). We have shown that thrombin, a serine protease generated early on during acute respiratory distress syndrome, plays a critical role in increasing endothelial permeability by inducing the loss of VE-cadherin homotypic adhesion and redistribution of focal adhesions dependent on FAK (Mehta et al., 2002; Kouklis et al., 2004; Holinstat et al., 2006). Interestingly, the thrombin-induced increase in endothelial permeability is reversed within 2–3 h, indicating activation of endogenous pathways that limit the persistent increase in endothelial permeability produced by thrombin (Kouklis et al., 2004; Holinstat et al., 2006; Kaneider et al., 2007).
Thrombin binds to endothelial cell surface receptor, protease-activating receptor 1 (PAR1) and PAR4 (Coughlin, 2000, 2005; Kataoka et al., 2003). We have shown that the permeability increasing effects of thrombin in lung endothelium are predominantly mediated through PAR1 because thrombin and selective PAR1 peptide agonists failed to induce endothelial contraction and lung microvascular permeability increase in mice lacking PAR1 (Vogel et al. 2000). PAR1 is a seven-transmembrane domain receptor that couples to heterotrimeric G proteins of the Gq and G12/13 families (Hung et al., 1992; Coughlin 1999). Upon ligation by thrombin, PAR1 signals the dissociation of the α-subunits of Gq and G12/13 from the Gβγ dimer. Gαq and Gα12/13 activate myosin light chain kinase and RhoA pathways, which by inducing endothelial cell contraction increase permeability (Goeckeler and Wysolmerski, 1995; Dudek and Garcia, 2001; Holinstat et al., 2003; McLaughlin et al., 2005; Knezevic et al., 2007; Singh et al., 2007; Gavard and Gutkind, 2008; Korhonen et al., 2009). However, the role of Gβγ after its dissociation from these heterotrimeric G proteins in the mechanism of PAR1-induced alteration in endothelial barrier function is unknown.
The Gβγ pathway has progressively emerged as a critical element of GPCR signaling (Clapham and Neer, 1997; Cabrera-Vera et al., 2003; Oldham and Hamm, 2008). Gβγ is known to induce cyclic AMP generation (Tang and Gilman, 1992; Taurin et al., 2007), Ca2+ signaling (Herlitze et al., 1996; Blackmer et al., 2001), oxidant generation (Niu et al., 2003), neurotransmission (Blackmer et al., 2005), chemotaxis (Neptune and Bourne, 1997; Jin et al., 2000), and caveolae-mediated transcytosis (Shajahan et al., 2004). The β subunit of Gβγ contains WD 40 repeats that are thought to mediate protein–protein interactions (Neer et al., 1994; Chen et al., 2004b). Studies show that Gβ interacts with receptor for activated C kinase 1 (RACK1; Dell et al., 2002; Chen et al., 2004a), p60cSrc (Luttrell et al., 1996; McGarrigle and Huang, 2007), and Fyn (Yaka et al., 2002; Thornton et al., 2004). Fyn, p60cSrc, and RACK1 are known to influence adherens junctions (AJ) and focal adhesions (Xing et al., 1994; Bockholt and Burridge, 1995; Thomas and Brugge, 1997; Roura et al., 1999; Owens et al., 2000; Mourton et al., 2001; Schaller, 2001; Piedra et al., 2003). We tested the hypothesis that, besides restraining Gα subunits, Gβγ orchestrates signaling to terminate endothelial permeability increase through its ability to coordinate intercellular and cell–matrix interactions.
To explore the function of Gβγ in regulating endothelial permeability, we interfered with expression of Gβγ, RACK1, or Fyn using small interfering RNA (siRNA) or Fyn knockout mice. We show that Gβγ plays a fundamental role in signaling endothelial barrier recovery. Moreover, we identified Fyn and FAK as the key downstream effectors of Gβγ. Fyn-mediated activation of FAK facilitated the association of FAK with p120-catenin and reannealing of AJ, which reversed the increased endothelial permeability responses produced by G protein–coupled receptor agonists.
To address the contribution of Gβγ to the mechanism of increased endothelial permeability, we transduced the 194-aa C terminus of β-adrenergic receptor kinase 1 (CT-βARK-1; Drazner et al. 1997) in endothelial cells to inhibit Gβγ (Fig. 1 A, inset). Changes in endothelial permeability were determined by quantifying interendothelial gap area in endothelial monolayers immunostained with anti–VE-cadherin antibody (Ab) and dynamically by determining the decrease in transendothelial electrical resistance (TER) after thrombin challenge. Consistent with our previous study (Holinstat et al., 2006), thrombin rapidly disrupted AJ, induced gap formation (Fig. 1 A), and decreased TER (Fig. 1 B) in cells receiving control vector. AJ subsequently reannealed, as interendothelial gaps closed and TER returned to basal levels within 2 h (Fig. 1, A and B). Inhibition of Gβγ did not affect the acute disruption of AJ, interendothelial gap formation, or TER decrease induced by thrombin, but in contrast to control cells, these thrombin responses persisted without recovery to basal level (Fig. 1, A and B).
The γ subunit must dimerize with the β subunit in order for Gβγ to function normally (Jones et al., 2004). Approximately 12 γ isoforms and 5 β isoforms are reported in various cell types (Hurowitz et al., 2000; Robishaw and Berlot, 2004). Our semiquantitative RT-PCR analysis of human pulmonary arterial endothelial (HPAE) cells demonstrated that two β isoforms are expressed in endothelial cells, of which the β1 is the predominant isoform (Fig. 1 C, top left). Therefore, we targeted the β1 isoform to achieve the maximal disruption of the Gβγ dimer. Using siRNA, we depleted Gβ1 (Fig. 1 C, top right) and determined the effect of Gβ1 depletion on the endothelial barrier recovery response. We observed that in Gβ1 knockdown cells, thrombin induced an irreversible disruption of AJ (Fig. 1 C) resulting in a persistent increase in the duration of endothelial permeability responses (Fig. 1 D), confirming our findings that Gβγ is required to restore basal endothelial barrier function.
Based on our previous study showing that FAK, upon being phosphorylated at tyrosine residues Y397 and Y576, suppresses RhoA activation and thereby promotes restoration of basal endothelial barrier function (Holinstat et al., 2006), we addressed the possible role of Gβγ in regulating FAK activation. Endothelial cells transfected with CT-βARK-1 or Gβ1 siRNA were stimulated with thrombin for the indicated times and FAK activation was determined by using anti-Y397 or anti-Y576 phospho-FAK antibodies. We found that Gβ1γ was necessary for FAK activation because thrombin failed to increase FAK phosphorylation at 397 and 576 Tyr residues in endothelial cells expressing CT-βARK-1 (Fig. 2 A) or transfected with Gβ1 siRNA (Fig. 2 B). Inhibition of Gβγ increased basal RhoA activity and potentiated the increase in RhoA activation in response to thrombin (Fig. 2 C), in agreement with our previous study (Holinstat et al. 2006).
We next used the cell-permeant Gβγ-activating peptide mSIRK (Goubaeva et al. 2003) to further substantiate the role of Gβγ in regulating FAK activation. Stimulation of HPAE cells with mSIRK induced FAK phosphorylation at both sites without thrombin stimulation; control peptide, i.e., L9A mSIRK, was ineffective (Fig. 2 D). mSIRK also counteracted the disruptive effect of thrombin on AJ (Fig. 2 E), and thereby markedly suppressed thrombin-induced increase in endothelial permeability (Fig. 2 F). These findings support the conclusion that up-regulation of FAK activity by Gβγ induces barrier repair by facilitating reannealing of AJ, which thus prevents persistent increase in endothelial permeability.
Gβ1 belongs to a large family of WD 40 proteins and shares 57% amino acid homology with its binding partner RACK1 (Chen et al., 2004b). RACK1 also binds Fyn, a Src family nonreceptor tyrosine kinase that phosphorylates FAK (Yaka et al., 2002). To address the mechanism of Gβ1 up-regulation of FAK activity, we determined whether thrombin induces formation of a signaling complex consisting of Gβ1, RACK1, and Fyn. To do this, we cotransduced endothelial cells with GFP-tagged Gβ1 and GFP-tagged Gγ2 and used coimmunoprecipitation assay to detect complex formation. We noted that in unstimulated endothelial cells, Gβ1γ2-associated with RACK1 and Fyn (Fig. 3). However, under these conditions, an association between Gβ1γ2 and FAK was not found. Thrombin potentiated the interaction of Gβ1γ2 with Fyn and FAK within 10–30 min, but induced a transient dissociation of RACK1 from Gβ1γ2 (Fig. 3) that correlated with FAK activation, initiation of AJ reannealing, and recovery of endothelial barrier function, as shown in Figs. 1 and and22.
Next, we suppressed endogenous expression of RACK1 to assess the causal role of RACK1 in regulating Gβ1–Fyn interaction, FAK activation, and endothelial barrier permeability. We observed that knockdown of RACK1-potentiated basal interaction of Gβ1 with Fyn and FAK, which did not increase further after thrombin stimulation (Fig. 4 A). Suppression of endogenous RACK1 basally activated FAK (Fig. 4 B), and significantly accelerated reannealing of AJ within 10 min (Fig. 4 C), thereby speeding recovery of endothelial barrier function (Fig. 4 D). We noted a marked suppression of thrombin-induced disruption of endothelial barrier function in RACK1-deficient endothelial cells (Fig. 4 D).
We addressed the possibility that a physical interaction between Gβ1 and Fyn contributes to Fyn activation of FAK. We first determined whether thrombin phosphorylates Fyn in a Gβ1-dependent manner. HPAE cells transfected with Gβ1 siRNA or scrambled siRNA were stimulated with thrombin and lysed. Lysates were immunoprecipitated with anti-Fyn Ab, and tyrosine phosphorylation was assessed using phosphotyrosine Abs. Under control conditions (scrambled siRNA-transfected cells), we observed that thrombin challenge increased tyrosine phosphorylation of Fyn above basal (Fig. 5 A). However, basal Fyn phosphorylation was reduced to undetectable levels in Gβ1-depleted cells and did not increase after thrombin stimulation (Fig. 5 A).
To determine the effect of Gβ1 knockdown on the activity of Fyn kinase (which phosphorylates FAK), we performed an in vitro kinase assay using recombinant FAK as a substrate. Fyn immunocomplexes obtained from cells transfected with Gβ1 siRNA were incubated with recombinant FAK and tyrosine phosphorylation of recombinant FAK was determined using phosphotyrosine Abs. In parallel, Fyn kinase activity was determined in cells transfected with RACK1 siRNA. We observed that Fyn immunoprecipitated from endothelial cells transfected with control siRNA or RACK1 siRNA phosphorylated recombinant FAK (Fig. 5 B). However, Fyn immunoprecipitated from Gβ1-siRNA–transfected cells failed to phosphorylate FAK, demonstrating that Gβ1 is required to induce Fyn activation (Fig. 5 B).
We suppressed endogenous expression of Fyn in endothelial cells to assess the specific role of Fyn in mediating Gβ1-induced FAK activation and endothelial barrier repair. Western blot analysis showed that suppression of Fyn expression did not alter the expression of Src family kinase p60cSrc (Fig. 6 A, inset). Depletion of Fyn did not alter the basal or thrombin-induced association of Gβ1 with RACK1. However, Gβ1 failed to interact with FAK in Fyn-depleted cells (Fig. 6 A), and under these conditions FAK phosphorylation on 397 or 576 tyrosine residues was not found (Fig. 6 B). Fyn knockdown did not alter transfer of Evans blue–conjugated albumin across the endothelial monolayer (0.11 ± 0.01 vs. 0.10 ± 0.01 µl/min) or basal TER (Fig. 6 C). However, suppression of endogenous Fyn expression resulted in irreversible increase in endothelial permeability after thrombin challenge (Fig. 6 C). In contrast, knockdown of p60cSrc significantly inhibited thrombin-induced increase in endothelial permeability (Fig. 6 C). These findings demonstrate that downstream of Gβ1, Fyn plays a key role in mediating FAK activation and endothelial barrier repair. Our data also suggest that Fyn and p60cSrc play contrasting roles in PAR1 regulation of endothelial permeability.
In another series of experiments, we investigated the role of Fyn in regulating endothelial barrier integrity of pulmonary microvessels using Fyn-deficient mice. First we determined whether activation of PAR1 by the specific activating peptide TFLLRN alters FAK activity in lungs. WT or fyn−/− mice received i.v. injection of either control peptide or PAR1-activating peptide (1 mg/kg). Lungs were homogenized to assess FAK phosphorylation. We observed that PAR1 agonist peptide increased FAK phosphorylation at tyrosine 397 and 576 residues in WT lungs (Fig. 7 A). However, in lungs of fyn−/− mice, FAK phosphorylation at tyrosine 397 and 576 was barely detectable under basal conditions and did not increase after PAR1 activation (Fig. 7 A), which is consistent with our findings in Fyn-depleted endothelial cells. Deletion of Fyn had no effect on either cSrc expression or cSrc phosphorylation induced by PAR1 (Fig. 7 B), implicating Fyn as the key enzyme regulating FAK activation.
Next, we determined the microvessel filtration coefficient (Kf) in lungs isolated from WT and fyn−/− mice. We showed that basal Kf was significantly higher in fyn−/− lungs than WT lungs (Fig. 7 C). To confirm that deletion of Fyn in mouse lungs leads to pulmonary edema, we determined Evans blue albumin extravasation (EBAE) and lung wet-to-dry weight ratio after PAR1 activation. Lung vascular albumin permeability was the same in WT and fyn−/− mice receiving a scrambled PAR1 peptide (Fig. 7, D and E). However, injection of PAR1 peptide (i.v.) produced a significantly greater increase in EBAE (Fig. 7 D) and edema formation (Fig. 7 E) in the Fyn knockout, indicating that Fyn mitigates PAR1-agonist–induced pulmonary edema.
To assess whether Fyn-dependent activation of FAK induces the association of FAK with endothelial junctions, we obtained lysates from WT and fyn−/− mice stimulated with PAR1-activating peptide, immunoprecipitated FAK with anti-FAK Ab, and immunoblotted with anti–p120-catenin Ab. PAR1 activation markedly increased the association between FAK and p120-catenin (a marker of endothelial junctions) between 15 and 30 min in WT but not fyn−/− lungs (Fig. 8 A). Fyn deletion had no effect on p120-catenin expression (Fig. 8 A). The results indicate that Fyn-dependent phosphorylation of FAK induces association between FAK and AJ.
To demonstrate that FAK and AJ associate in endothelial cells, we coimmunostained HPAE cells with anti-FAK and anti–p120-catenin antibodies. As expected, FAK localized at focal adhesions under basal conditions and to a greater extent after thrombin challenge (Fig. 8 B). FAK also colocalized with p120-catenin under basal conditions, and the association was significantly increased by thrombin (Fig. 8 B). No interaction was found between FAK and isotype-matched control IgG Ab (unpublished data).
Because Fyn deletion prevented FAK phosphorylation at tyrosine residues 397 and 576 (Fig. 7 A), we generated a phosphorylation-mimicking mutant of FAK to test whether restoring FAK phosphorylation in the fyn−/− mouse lung could restore lung vascular permeability responses to PAR1 agonist to the level seen in WT mice. We replaced Tyr 397 and Tyr 576 with Asp and expressed the mutant construct in cultured endothelial cells. First, we confirmed that expression of FAK mutant Y397D/Y576D increased FAK phosphorylation. As shown in Fig. 9 A, transduction of the mutant construct increased basal FAK phosphorylation at Y397 and Y576 to levels seen in vector control cells after thrombin stimulation. Thrombin could not increase FAK phosphorylation further in cells transducing FAK phosphorylation mimicking mutant (Fig. 9 A).
Next we transduced the same mutant construct into WT and fyn−/− murine lung microvessels using cationic liposomes encapsulating GFP-tagged FAK-Y397D/-Y576D cDNA. We observed increased FAK phosphorylation in lungs of fyn−/− (Fig. 9, B and C) and WT mice (Fig. 9 C). Immunostaining of lung sections from WT or Fyn-null mice expressing the FAK-Y397D/-Y576D mutant with anti–VE cadherin and phosphospecific FAK antibodies showed that activated FAK was localized in pulmonary vascular endothelial cells (Fig. 9 C). We next determined transvascular albumin permeability and lung edema formation in WT or fyn−/− mice transduced with either control vector (GFP) or GFP-tagged FAK-Y397D/-Y576D cDNA. As shown in Fig. 9 (D and E), transduction of FAK-Y397D/Y576D in the lungs significantly decreased albumin leakage and edema formation in response to PAR1 activation in WT lungs. We also observed that restoring FAK phosphorylation reversed PAR-1–induced lung vascular permeability in fyn−/− lungs to the level observed in WT mice lungs (Fig. 9, D and E).
Thrombin, which is generated during vascular injury, acutely increases endothelial permeability by activating PAR1 on endothelial cells (Coughlin, 2000; Landis, 2007; McLaughlin et al., 2007; Suzuki et al., 2009). PAR1 produces these effects by stimulating dissociation of the heterotrimeric G protein complexes into functional Gα and Gβγ subunits (Swift et al., 2000; Rahman et al., 2002; McLaughlin et al., 2005). Several lines of evidence favor the conclusion that the dissociated Gαq and Gα12/Gα13 subunit increases endothelial permeability by signaling the activation of myosin light chain kinase and RhoA pathways, which in turn leads to endothelial cell contraction and disruption of cell–cell adhesion (Dudek and Garcia, 2001; Mehta et al., 2001; Holinstat et al., 2003; Singh et al., 2007; Knezevic et al., 2007; Singh et al., 2007; Gavard and Gutkind, 2008; Korhonen et al., 2009). However, such studies have not defined the role of the released Gβγ unit under these conditions. Impairing Gβγ function using CT-βARK1 or Gβ1 siRNA allowed us to address the functional role of Gβγ in the mechanism of PAR1-induced increase in endothelial permeability. These results showed that inhibition of Gβγ did not augment thrombin-induced increase in endothelial permeability because of dysregulation of α subunit activity, and hence exacerbated RhoA signaling (Holinstat et al., 2003; Holinstat et al., 2006; Mehta and Malik, 2006), but prevented recovery of endothelial barrier function. We also showed that direct stimulation of Gβγ by mSIRK, bypassing the α-subunit activation, accelerated the recovery of endothelial barrier function after thrombin challenge. Thus, Gβγ functions to promote recovery of barrier function through orchestration of signaling involving Fyn activation of FAK.
Studies show that five Gβ isoforms exist in various cell types (Schwindinger and Robishaw, 2001; Robishaw and Berlot, 2004; Thompson et al., 2008). However, our data show that Gβ1 is the predominant isoform expressed in human endothelial cells. Because Gβ dimerization with Gγ is required for the dimer to function (Jones et al. 2004), we showed that targeted suppression of endogenous Gβ1 (~80%) prolonged the increase in permeability caused by thrombin, as contrasted with full recovery normally seen within 2 h in control cells. Together, these findings further support the conclusion that the primary function of Gβγ is to restore endothelial barrier function.
Gβ1 belongs to a large family of WD40 repeat proteins and interacts with several proteins, including RACK1 (Chen et al., 2004b). RACK1 also binds Fyn (Yaka et al., 2002). Fyn induces FAK activity by phosphorylating Y397 and Y576 residues in FAK. We have shown that thrombin induces a sustained phosphorylation of FAK at Y397 and Y576 (Holinstat et al., 2006). We also showed that activation of FAK is required to reverse the increase in endothelial permeability (Holinstat et al., 2006). Thus, we considered a model in which Gβ1 interacts with Fyn, inducing Fyn activation, which in turn up-regulates FAK activity, subsequently restoring endothelial barrier function. We noted that under basal conditions Gβ1 coimmunoprecipitated with RACK1 and Fyn, but not FAK. The results of this study further showed that thrombin induced the interaction between Gβ1, Fyn, and FAK, coinciding with Fyn and FAK activation and restoration of endothelial barrier function. However, the formation of this ternary complex required dissociation of Gβ1 from RACK1 because knockdown of RACK1 enhanced interaction between Gβ1, Fyn, and FAK in the absence of thrombin stimulation and augmented FAK activity. Also, Gβ1, but not RACK1, was required for Fyn activation, which in turn induced FAK activity. We also showed that Gβ1 failed to interact with FAK in Fyn-depleted cells. Our data show that there is a basal interaction between Gβγ, Fyn, and RACK1. We interpret these findings to mean that Gβγ exists in a dynamic equilibrium with RACK1, Fyn, and PAR1 receptor. Thrombin stimulation results in an increased association of Gβγ with Fyn and FAK at the expense of release of Gβγ from RACK1, and presumably from the PAR1 receptor, as GPCR activation leads to dissociation of Gα subunit from Gβγ (Cabrera-Vera et al., 2003; Oldham and Hamm, 2008). Additionally, our data show that upon stimulation the equilibrium is shifted away from the receptor, and definitely from RACK1, favoring Gβγ association with Fyn and FAK. However, our studies cannot delineate whether Gβγ forms a single complex with RACK1 and Fyn/FAK or if these two form separate complexes. Thus, our findings suggest that RACK1 negatively regulates Gβ1 restoration of barrier function by preventing Gβ1–Fyn interaction and restricting Fyn up-regulation of FAK activity. Supporting this conclusion are our findings that depletion of RACK1 enhanced Gβ1 interaction with Fyn, thereby stimulating Fyn activation of FAK activity, with the predictable consequences of counteracting PAR1-induced barrier dysfunction and of accelerating the recovery of barrier function to basal levels. The mechanism by which thrombin facilitates the dissociation of RACK1 from Gβ1 and Fyn is not clear. Evidence indicates that RACK1 acts as a negative regulator of Gβ1 and Fyn (Dell et al., 2002; Thornton et al., 2004). Gβ1 and Fyn share binding sites on RACK1 (Yaka et al., 2002; Thornton et al., 2004), and in some studies RACK1 also heteromerizes with Gβ1 (Dell et al. 2002). PKC and phosphatases such as SHIP were shown to modify RACK1 function by inducing its posttranslational modification (Ron et al., 1994; Nishio et al., 2007). Because thrombin activates PKC and SHIP (Mehta et al., 2001; Tiruppathi et al., 2002; Dyson et al., 2003), a possible scenario is that posttranslational modification of RACK1 decreases the affinity of RACK1 for Fyn and Gβ1, thus enabling them to interact with FAK to signal barrier restoration.
Our data also showed that the permeability increasing effects of thrombin partially overlapped with the process of barrier recovery such that the 10-min time point reflects both processes, whereas the 30-min time points predominantly reflect recovery. It is therefore possible that Gβγ-Fyn signaling may serve to protect against the “acute” phase of thrombin-induced permeability. However, we showed that FAK activity increased at 10 min, reached a maximum at 30 min, and persisted for 2 h. Interfering with Gβγ or Fyn function inhibited FAK phosphorylation as early as 10 min, and therefore at all succeeding time points, and also prevented barrier recovery to basal levels. These results indicate the importance of Gβγ-Fyn signaling in FAK-dependent reannealing of endothelial junctions.
We found that activated PAR1 failed to induce phosphorylation of FAK at tyrosine residues 397 and 576 in Fyn knockdown cells and in fyn−/− lung preparations. However, Fyn deletion did not alter thrombin-induced p60cSrc phosphorylation. These findings implied that Fyn is required for FAK activation in endothelial cells. Studies showed that several members of the Src family of tyrosine kinases, including Fyn, can activate FAK (Xing et al., 1994; Bockholt and Burridge, 1995; Thomas and Brugge, 1997; Schaller, 2001; Brunton et al., 2005; Mitra and Schlaepfer, 2006). In this regard, our experimental observations that expression of p60cSrc, as well as p60Src phosphorylation, was normal in Fyn-depleted endothelial cells, and Fyn-null lungs showed that p60cSrc could not substitute for Fyn in inducing FAK activation.
Because FAK organizes focal adhesions, a likely scenario could be that FAK restores endothelial barrier function by its effect on cell–ECM interaction. Intriguingly, our data show that a pool of FAK colocalized with the AJ marker p120-catenin under basal conditions, and colocalization increased further when the endothelial barrier function was restored. FAK activation or junctional reassembly was not evident when endogenous expression of Fyn was suppressed. Moreover, PAR1 stimulation in lungs from Fyn-deficient mice did not induce FAK activation, or FAK interaction with p120-catenin, and thus induced an exaggerated lung vascular permeability increase. Studies show that p120-catenin is required for maintaining AJ integrity and for repairing the endothelial barrier (Iyer et al., 2004; Mehta and Malik, 2006; Dejana et al., 2008; Rudini and Dejana, 2008). Our findings that interaction of activated FAK with p120-catenins subsequent to Fyn activation of FAK facilitates reannealing of AJ leading to reestablishment of the basal endothelial barrier suggest that the Fyn–FAK pathway plays a critical role in restoring endothelial barrier function. We have demonstrated that tyrosine phosphorylation of p190RhoGAP by FAK is crucial in down-regulating RhoA activity, which in turn facilitates the restoration of barrier function after thrombin challenge (Holinstat et al., 2006). The findings of this study indicate that FAK can also interact with AJ components and reanneal AJ.
Our findings also demonstrate the contrasting roles of p60cSrc and Fyn in regulating endothelial permeability. We showed that knockdown of p60cSrc suppressed the increase in endothelial permeability induced by PAR1. These findings are consistent with the observations in p60cSrc-null mice that were protected from VEGF-induced vascular leak (Eliceiri et al., 1999). However, Fyn knockdown resulted in a persistent increase in endothelial permeability. Because we showed that basal pulmonary microvascular permeability is markedly increased in fyn−/− mouse lungs, it is indicated that activation of FAK regulates basal endothelial barrier function in vivo, as previously described (Holinstat et al., 2006). We also showed that Fyn was required for suppressing the lung vascular permeability increase after PAR1 activation because fyn−/− lungs failed to recover and therefore developed persistent edema. We conclude that the absence of FAK activation in the endothelium of fyn−/− lungs impaired endothelial barrier function. This scenario is likely because the transduction of a phosphomimicking FAK mutant rescued lung vascular function in fyn−/− lungs to near normal values. Thus, these findings demonstrate that Gβγ-Fyn dependent pathway for activation of FAK stimulates reannealing of junctions. Additionally, studies have shown that activation of SPHK1 is required for restoring endothelial barrier function (Li et al., 2008; Tauseef et al., 2008). Whether these two pathways act in concert remains to be investigated, but it is likely that multiple mechanisms are engaged to correct endothelial barrier defects.
In conclusion, our results indicate that Gβγ plays an essential role in preventing the PAR1-induced long-lived increase in endothelial permeability. We propose a model whereby thrombin-mediated dissociation of RACK1 from Gβγ results in Gβγ and Fyn complex formation that leads to Fyn activation and phosphorylation of FAK (Fig. 10). A pool of activated FAK interacts with p120 catenin, where it induces reannealing of AJ downstream of Fyn. FAK also associates with p190RhoGAP (Holinstat et al., 2006) which suppresses endothelial contraction by down-regulating RhoA activity. Together, these events promote stabilization of cell–ECM attachment and intercellular adhesion, resulting in the restoration of endothelial permeability. Development of targets that specifically modify this pathway could therefore lead to therapeutic strategies to protect against persistent lung vascular barrier leakiness.
HPAE cells and endothelial growth medium (EBM-2) were obtained from Lonza. Human α-thrombin was obtained from Enzyme Research Laboratories. Nucleofector kit and electroporation system were obtained from Lonza. Superfect transfection reagent was purchased from QIAGEN. Alexa Fluor–phallodin, anti–Alexa Fluor–568 and –488 antibodies, DAPI, and ProLong Gold antifade were obtained from Invitrogen. Electrodes for transendothelial resistance measurements were obtained from Applied Biosciences. 12-mm-diam transwell plates and 0.4-µm polyester membranes were purchased from Corning. Primers for Gβ 1–5 subunits, siRNA for Gβ1, Fyn, cSrc, and control scrambled siRNA, as well as transfection reagent for siRNA, were purchased from Santa Cruz Biotechnology, Inc. RACK1 siRNA (5′-CCAUCAAGCUAUGGAAUACTT-3′ sense and 5′-GUAUUCCAUAGCUUGAUGGTT-3′ antisense) were purchased from Sigma-Aldrich. Anti-FAK, anti–phospho FAK PY576, anti-RhoA, anti-RACK1, anti-Gβ1, anti-Fyn, anti-cSrc, anti-VE-cadherin, anti-GRK2, anti-GFP, anti-p120, anti-phosphotyrosine Abs (PY20, PY99, and PY350), and protein A/G agarose beads were purchased from Santa Cruz Biotechnology, Inc. Anti–phospho FAK PY397 was purchased from Invitrogen, anti–β catenin Ab was obtained from Abcam. mSIRK and mSIRK-L9A were purchased from Calbiochem. GST-rhotekin-Rho binding domain beads were purchased from Cytoskeleton. Ad-CT-βARK (C terminus of β-adrenoreceptor kinase) was a gift from R. Minshall (University of Illinois, Chicago, IL, and GFP-tagged Gβ1 and Gγ2 were gifts from J. McLaughlin (University of Illinois, Chicago, IL).
All animal studies were approved by the Institutional Animal Care and Use Committee of the University of Illinois. fyn−/− male mice obtained from The Jackson Laboratory were backcrossed to C57BLk/6J background. C57BLk/6J mice were used as WT controls. All experiments were performed on 6–8-wk-old male mice.
siRNA or cDNA were transduced into cells by electroporation or using transfection reagents, as previously described (Singh et al., 2007). For adenoviral vector-mediated gene transduction, cells grown to confluence were infected with Ad-control vector or Ad-CT-βARK in serum-free medium. 6 h after infection, the medium was replaced with fresh complete medium without virus. In all experiments, cells were used between 32 and 36 h after virus infection. Control vector (Ad-control vector) or Ad-CT-βARK was generated as previously described (Drazner et al., 1997).
Total RNA was isolated from HPAE cells with Trizol reagent (Invitrogen) according to the manufacturer's instructions. RNA (1 µg) was used for semiquantitative RT-PCR to assess the expression of Gβ1, Gβ2, Gβ3, Gβ4, and Gβ5 subunits using primers designed for the human mRNA sequences (Santa Cruz Biotechnology, Inc.) with the aid of the Invitrogen SuperScript RT-PCR kit. We used the following conditions: an initial 94°C denaturing for 2 min, followed by 25 cycles at 95°C for 15 s each, annealing at 55°C for 30 s, and extension at 72°C for 30 s. The last cycle was followed by a 7-min reaction at 72°C. RT-PCR products were separated by electrophoresis on a 2% agarose gel.
HPAE cells grown in 100-mm dishes were lysed and equal amounts of protein were immunoprecipitated with appropriate Abs overnight at 4°C, followed by the addition of protein A/G agarose beads for 4 h at 4°C, as described previously (Holinstat et al., 2006).
HPAE cells transduced with appropriate siRNA were stimulated with thrombin for the indicated times. Lysates were and immunoprecipitated with anti-Fyn Ab, followed by incubation with protein A/G agarose beads to form immunocomplexes. Immunocomplexes were used for determining Fyn in vitro kinase activity, using purified FAK as the substrate as previously described (Holinstat et al., 2006).
HPAE cells transducing indicated cDNA were stimulated with 50 nM thrombin, and RhoA activity was measured using the GST-rhotekin-Rho binding domain, as previously described (Holinstat et al., 2006).
Cells transduced with cDNA or siRNA were stimulated with 50 nM thrombin for the indicated times, fixed, and stained with appropriate Abs, as previously described (Knezevic et al., 2007; Singh et al., 2007). Cells were visualized with 63× 1.2 NA objective on a LSM 510 confocal microscope (Carl Zeiss, Inc.). Interendothelial gap area was quantified using a MetaMorph software (Molecular Devices).
For TER measurements, HPAE cells were seeded on 8-well gold-coated electrodes, where they formed confluent monolayers. TER is expressed as specific electrical resistance (Ω cm2; Mehta et al., 2002). Albumin permeability of HPAE monolayers grown on transwell plates was measured by quantifying the flux of Evans blue–labeled albumin (EBA) from luminal to abluminal chamber (Tauseef et al., 2008).
Cationic liposomes were made using a mixture of dimethyldioctadecyl-ammonium bromide and cholesterol in chloroform, as described previously (Holinstat et al., 2006). GFP or GFP-FAK-Y397D/Y576D (50 µg) were mixed with 100 µl of liposomes. The mixture of liposomes and cDNA were injected intravenously (via retroorbital injection) into WT or fyn−/− mice. After 48 h, mouse lungs were used for determining lung microvascular permeability or protein expression.
Formalin fixed 4-µm-thick lung sections were immunostained with appropriate Abs using manufacture protocol (Bethyl Laboratories). Lung sections were visualized with a LSM510 confocal microscope (Carl Zeiss, Inc.).
EBA (20 mg/kg) was injected retroorbitally 30 min before sacrifice to assess vascular leak, as previously described (Peng et al., 2004; Tauseef et al., 2008). Blood was collected from the right ventricle into heparinized syringes and plasma was separated by centrifugation at 1,300 g for 10 min. Lungs homogenates were prepared as previously described (Tauseef et al., 2008). Lung homogenates and plasma were incubated with 2 volumes of formamide (18 h, 60°C), centrifuged at 5,000 g for 30 min, and the optical density of the supernatant was determined spectrophotometrically at 620 and 740 nm (to correct for hemoglobin). EBA extravasation is given as lung/plasma ratio.
Fyn−/− and WT mice were anesthetized with an i.p. injection of ketamine (100 mg/kg) and xylazine (2.5 g/kg). Microvessel permeability in the lung was measured by determining microvascular filtration coefficient (Kf) and isogravimetric lung water determinations as previously described (Vogel et al. 2000; Holinstat et al. 2006).
Left lungs from the same mice used for Evans blue albumin extravasation were excised and completely dried in the oven at 60°C overnight for calculation of lung wet/dry ratio (Barnard et al. 1995).
Comparisons between experimental groups were made by one-way ANOVA and post-hoc test. Differences in mean values were considered significant at P < 0.05.
We acknowledge Drs. Asrar B. Malik and Stephen M. Vogel for their constructive criticism during the preparation of this manuscript; Yulia Komarova for her advice in analysis of interendothelial gap formation; Tiffany Sharma and Debra Salvi for constructing FAKY387D/Y576D mutant; and Ms. Vidisha Kini for technical help. We also thank Dr. Xiaoping Dufor his gift of fyn−/− mice.
This work was supported by National Institutes of Health grants HL71794 and HL84153.
The authors have no conflicting financial interests.