Atherogenesis involves activation of NF-κB in endothelial cells by fluid shear stress. Because this pathway involves integrins, we investigated the involvement of focal adhesion kinase (FAK). We found that FAK was not required for flow-stimulated translocation of the p65 NF-κB subunit to the nucleus but was essential for phosphorylation of p65 on serine 536 and induction of ICAM-1, an NF-κB-dependent gene. NF-κB activation by TNF-α or hydrogen peroxide was FAK independent. Events upstream of NF-κB, including integrin activation, Rac activation, reactive oxygen production, and degradation of IκB, were FAK independent. FAK therefore regulates NF-κB phosphorylation and transcriptional activity in response to flow by a novel mechanism.
- fluid shear stress
- integrin signaling
current models for atherosclerosis suggest that local endothelial dysfunction results in monocyte recruitment and lipid deposition in the vessel intima (29). In addition to systemic risk factors, fluid shear stress from flowing blood plays a crucial role. Atherosclerotic lesions develop preferentially in regions of oscillatory or disturbed flow that have lower mean shear stress, multidirectionality, and flow separation (33).
The proinflammatory transcription factor NF-κB is thought to be a key determinant of atherogenesis (8). NF-κB is activated in atheroprone regions in vivo concomitant with expression of its target genes including ICAM-1, VCAM-1 (20), monocyte chemoattractant protein 1, tissue factor, and PDGF (8). NF-κB is a dimer, usually consisting of a p65 and a p50 subunit, though other combinations also exist. Inactive NF-κB is sequestered in the cytoplasm by IκB proteins (8, 20). The IκB kinase (IKK) complex phosphorylates IκB, leading to its ubiquitination and proteosomal degradation. Free NF-κB dimer then translocates into the nucleus and binds to κB enhancer sites. Phosphorylation, acetylation, and sumoylation regulates the subsequent recruitment of cofactors and other members of the transcriptional machinery required to drive target gene expression.
Integrins have been implicated in activation of NF-κB and several other proinflammatory pathways in response to flow (19). Shear stress induces conversion of integrins to a high-affinity state (i.e., integrin activation), which is followed by binding of these high-affinity integrins to the extracellular matrix beneath the cells to form new integrin-matrix contacts (53). The newly ligated integrins trigger downstream signals, including RhoA, Rac, and Cdc42 activation (53–55). These GTPases mediate cell alignment, sterol regulatory element-binding protein (SREBP) cleavage (30), and NF-κB activation (54). Consistent with this model, flow-induced activation of NF-κB is integrin and matrix dependent, occurring in cells on fibronectin (FN) or fibrinogen but not on collagen or laminin (38).
Focal adhesion kinase (FAK), a nonreceptor tyrosine kinase involved in integrin signaling (43, 46), is present in endothelial cells and has been implicated in responses of endothelial cells to fluid shear stress. Complete deletion of FAK in mice is lethal at embryonic (E) day 8.5, and endothelium-specific deletion is lethal after day E11.5. In endothelial cells, flow stimulates FAK recruitment into focal adhesions and increased tyrosine phosphorylation (27). Furthermore, dominant-negative FAK inhibited activation of the sterol regulatory element-binding protein 1 (SREBP-1) by flow (31), and introduction of an antibody against FAK phosphorylated on Y397 decreased endothelial nitric oxide synthetase (eNOS) phosphorylation and vasodilation in vivo (25). Thus, several reports show that FAK is involved in endothelial responses to flow, most likely through the integrin activation pathway described above.
These considerations led us to examine the contribution of FAK to flow-induced inflammatory signaling. Many studies have shown that onset of laminar shear induces transient activation of the same pathways via the same mechanisms as disturbed flow does in a sustained fashion (14, 34, 38, 40, 41). We therefore used onset of shear as a simple model system for understanding the basic mechanisms of cell response to flow. Using endothelial cells isolated from the conditional FAK-knockout mouse, our results reveal a novel and unexpected role for FAK in NF-κB activation.
MATERIAL AND METHODS
Antibodies used were rabbit anti-phospho-p65 (Ser536), rabbit anti-ERK, rabbit anti-IkBα, rabbit anti-phospho eNOS (Ser1179) and rabbit anti-phospho-ERK (Thr202/Tyr204), and rabbit phospho-JNK (Thr183/Tyr185) (Cell Signaling Technology, Beverly, MA); mouse-anti-Pyk2 (BD Bioscience, San Diego, CA); mouse anti-FAK (NH2-terminal; Upstate Biotechnology, Lake Placid, NY) and mouse anti-Rac (Upstate Biotechnology); rabbit anti-p65, rabbit anti-SREBP, rabbit anti-eNOS, goat anti-ICAM-1, mouse anti-β-catenin, mouse anti-glutathione S-transferase (GST), and secondary horseradish peroxidase (HRP) donkey anti-goat (Santa Cruz Biotechnology, Santa Cruz, CA); mouse anti-paxillin (Zymed); Alexa568-goat anti-mouse, Alexa488-goat anti-mouse- and Alexa488-conjugated anti-rabbit (Molecular Probes); HRP-goat anti-rabbit and goat anti-mouse (Jackson ImmunoResearch Laboratories, West Grove, PA); and mouse anti-tubulin (Developmental Studies Hybridoma Bank, Iowa city, IA).
Mouse aortic endothelial cell isolation.
All animal procedures were approved by the Institutional Animal Care and Use Committee of the University of Virginia. Mouse aortic endothelial cells (MAECs) were isolated from 8-wk-old FAK conditional knockout mice as described previously (49). Cells were initially grown on 0.2% gelatin-coated dishes with l-valine DMEM growth medium (17) containing 15% fetal bovine serum (FBS), 30 μg/ml endothelial cell growth supplement, and 50 μg/ml heparin. Cells expressed VE-cadherin, and >95% of cells showed uptake of 1,1'-dioctadecyl-3,3,3',3'-tetramethyl-indocarbocyanine perchlorate-labeled acetylated LDL (Dil-Ac-LDL).
Following passage 2, cells were maintained in Cambrex EBM-2 endothelial cell medium containing 10% FBS, 10 U/ml penicillin and 10 μg/ml streptomycin (Invitrogen), and 2 mM l- glutamine (Invitrogen). Cells were immortalized with polyoma middle T retrovirus (produced in the BOSC23 packaging cell line) (4). At passage 7, cells were infected with adenovirus carrying Cre-recombinase-green fluorescent protein (GFP) to delete the FAK gene (2). Transient transfections used Amaxa nucleofection (Amaxa Biotechnology, according to manufacturer's instructions) with 4 μg of DNA encoding wild-type FAK (FAK-WT) or green fluorescent protein pmaxGFP (Amaxa) 48 h before use.
Small interfering RNA experiments.
Bovine aortic endothelial cells (BAECs) were cultured in DMEM supplemented with 10% fetal bovine serum, 10 U/ml penicillin and 10 μg/ml streptomycin, and 2 mM l-glutamine (Invitrogen). Cells were transfected with small interfering RNA (siRNA) oligonucleotides against FAK using Lipofectamine 2000 (Invitrogen) according to the manufacturer's instructions. The sequence was 5′-GCUAGUGACGUAUGGAUGU-3′ as described previously (52).
MAECs were starved for 2 h in EBM-2 medium containing 0.2% FBS before use. BAECS were starved in DMEM with 0.2% serum before use. The parallel plate laminar flow chamber system (15) was maintained at 37°C and perfused with 7% CO2-93% air. Glass slides (38 × 75 mm, Corning) were coated with 10 μg/ml FN in PBS (for 2 h) and blocked with 0.2% BSA in PBS for 30 min. Cells were plated overnight to reach confluence the next day. Glass slides were kept under static conditions or exposed to 24 dyn/cm2 in the same starvation medium.
Cells on coverslips were washed with cold PBS, fixed for 30 min in 2% formaldehyde in PBS, permeabilized with 0.1% Triton-PBS, and blocked with Tris-buffered saline (TBS) containing 0.05% Tween 20 (TBST) and 10% goat serum. Primary antibodies were incubated overnight at 4°C in TBST. Antibodies were anti-FAK (1:500), anti-p65 (1:200), anti-paxillin (1:500), anti-β-catenin (1:500), or anti-phospho-S536 p65 (1:100). Cells were then washed, incubated with secondary Alexa-conjugated antibody (1:500) for 1 h, washed again, and mounted in Fluorochrome G (Southern Biotech). Confocal images were acquired using a Zeiss LSM 510 with ×63 oil or ×40 lenses. Images were processed with Adobe Photoshop 7 and Image J software.
Cells were lysed in 120 μl 5× Laemmli's buffer, sonicated, and boiled for 5 min. Proteins were separated by SDS-PAGE gel electrophoresis and transferred to polyvinylidene difluoride membranes. Membranes were blocked with 5% nonfat milk powder in TBST for 1 h and incubated with primary antibody at 4°C overnight. Blots were washed with TBST and incubated with HRP-conjugated secondary antibodies (Jackson ImmunoResearch) for 1 h at room temperature. Membranes were then treated with ECL reagents (Amersham Biosciences) and exposed to film. Resultant protein bands were scanned and quantified using Image J software. Membranes were reprobed after stripping with ReBlot Strong solution (Chemicon).
Integrin activation assay.
To measure the level of high-affinity, unoccupied integrins on the cell surface, cells were sheared for 5 min and incubated with 10 μg/ml GST-FNIII9–11 in PBS containing 1 mM Mg2+ for 30 min at 37°C as described previously (39). Cells were washed three times with PBS + 1 mM Mg2+ to remove unbound protein, lysed, and analyzed by immunoblotting for GST. Values were normalized for total protein loading (determined by tubulin Western blot).
RNA extraction and quantitative RT-PCR.
Total RNA was isolated using TRIzol reagent (Invitrogen) according to the manufacturer's instructions. cDNA was synthesized by reverse transcription (RT-PCR) using the iScript cDNA Synthesis Kit (Bio-Rad). To quantify mRNA expression, cDNA was analyzed by real-time RT-PCR (iCycler, Bio-Rad) using SYBR green. The following primers were used: ICAM-1 primer 1, 5′-CTGGCTGTCACAGAACAGGA-3′; ICAM-1 primer 2, 5′-AAAGTAGGTGGGGAGGTGCT-3′; 18S primer 1, 5′-CGGCTACCACATCCAAGGAA-3′; and 18S primer 2, 5′-AGCTGGAATTACCGCGGC-3′. ICAM-1 mRNA levels were quantified using a standard curve and were normalized to 18S.
Rac activity assays.
Cells were sheared for the indicated times, then extracted in lysis buffer [50 mM Tris, 500 mM NaCl, 10 mM MgCl2, 0.1% SDS, 1% Triton, 0.5% deoxycholate, and protease inhibitor mix (Sigma)]. Lysates from four slides were combined for each point. GST-p21 binding domain (20 μg) (10) and glutathione-Sepharose beads (20 μl, Amersham Biosciences) were added and incubated for 30 min at 4°C under rotation. Beads were washed with 50 mM Tris, 150 mM NaCl, 10 mM MgCl2, 1% Triton, and protease inhibitor mix. Bound Rac was eluted by adding 5× Laemmli's buffer and boiling for 5 min. Samples were analyzed by immunoblotting, and bound Rac was normalized for total Rac in the whole cell lysates.
Reactive oxygen species assay using dichlorofluorescein-diacetate.
Cells were incubated in media containing 10 μg/ml dichlorofluorescein-diacetate (DCFH-DA) (57) (Molecular Probes) for 30 min, then exposed to flow at 24 dyn/cm2 for 30 min in the continuous presence of DCFH-DA. Cells were rinsed with PBS and scraped into 1% Triton in PBS containing 3 μl/ml butylated hydroxyl toluene to prevent further oxidation. Samples were analyzed in a 96-well plate reader with excitation at 488 nm and emission at 528 nm. Values were normalized to total protein levels measured by Lowry protein assay (32).
FAK−/− endothelial cells.
Aortic endothelial cells were derived from 8-wk-old mice in which the second exon of the kinase domain of the FAK gene was flanked by LoxP sites (FAKfl/fl mice), immortalized with Polyoma middle t-antigen, and infected with Cre-recombinase adenovirus to yield FAK−/− mouse aortic endothelial cells. These FAK−/− endothelial cells contained no detectable FAK by Western blotting (Fig. 1A). The related kinase Pyk2 and the p65 NF-κB subunit were unchanged. Both focal adhesions and cell-cell junctions, structures implicated in shear stress sensing, were similar in FAK−/− cells (Fig. 1B). We conclude that infection with Cre deletes the FAK gene without perturbing Pyk2, NF-κB, or adhesive structures.
SREBP cleavage, ERK, JNK, and eNOS activation by flow.
eNOS (12, 51) and the MAP kinases ERK and JNK (42, 51) are activated by fluid shear stress. We found that phosphorylation of these enzymes in FAKfl/fl (control) and FAK−/− (knockout) cells was similar (supplemental Fig. 1, A–C; supplemental data for this article can be found online at the American Journal of Physiology-Cell Physiology website). Thus, deletion of FAK does not affect a number of previously characterized cellular responses to shear.
Onset of flow also triggers activation of SREBP, which occurs through cleavage of its precursor protein into a 68-kDa mature form that drives target gene expression (30, 31). Activation of this pathway by flow is suppressed by dominant-negative Y397F FAK (31). Accumulation of mature SREBP protein after shear stimulation was abolished in FAK−/− cells (supplemental Fig. 1D), confirming that this process is FAK dependent.
NF-κB activation in FAK−/− endothelial cells.
NF-κB is activated by flow (8). In addition to phosphorylation and degradation of the inhibitory IkBα subunit, which allows NF-κB to translocate into the nucleus, both subunits of NF-κB undergo posttranslational modification (6, 44), including phosphorylation of p65 Ser536 in the carboxyl-terminal transactivation domain, which is important for expression of target genes (6). FAK expression did not affect nuclear translocation of p65 after flow, although FAK−/− cells showed somewhat less nuclear p65 in the absence of flow (Fig. 2A). By contrast, loss of FAK slightly increased the baseline Ser536 phosphorylation but abolished the response to flow (Fig. 2B).
To test the dependence on FAK, FAK−/− cells were transfected with a vector for WT FAK or for GFP. Cells were then sheared for 30 min. Although expression of the exogenous FAK was substantially lower than endogenous protein in control FAKfl/fl cells (Fig. 3A), there was a significant rescue of Ser536 phosphorylation after flow (Fig. 3B). Although the rescue is incomplete, this result is very likely due to the low expression or transfection efficiency. These data indicate that FAK is dispensable for NF-κB nuclear translocation but is required for Ser536 phosphorylation.
As a second way to confirm these results, we asked whether loss of FAK affects NF-κB in BAECs, a commonly used model for shear studies. Transfection of siRNA into BAECs resulted in depletion of FAK by >90% when compared with the luciferase (GL2) controls (Fig. 3C). Flow was found to stimulate p65 phosphorylation on Ser536 FAK in control knockdown cells as in control murine cells. By contrast, FAK siRNA-treated cells again showed an increase in the baseline Ser536 phosphorylation, but the response to flow was completely inhibited. Thus, effects of FAK deletion in MAECs and knockdown in BAECs are similar.
We next examined expression of ICAM-1, a gene that is induced by flow through activation of NF-κB (11, 23, 38). ICAM-1 is a transmembrane receptor that mediates leukocyte adhesion to the endothelium (20) and is upregulated in areas of atheroprone flow in vivo (36). Induction of both ICAM-1 mRNA (Fig. 4A) and protein (Fig. 4B) were strongly dependent on FAK. Thus, flow-induced expression of a well-characterized NF-κB target gene also requires FAK.
The role of integrins.
Previous studies showed that shear stimulates NF-κB by triggering integrin activation followed by binding to matrix proteins (54). To test whether integrin activation was sufficient to trigger the NF-κB pathway, we treated endothelial cells with the activating β1-integrin antibody TS2/16, which induces conversion of integrins to the high-affinity conformation (5, 53). TS2/16 triggered nuclear translocation of p65 equally well in FAKfl/fl and FAK−/− endothelial cells (Fig. 5A); however, p65 Ser536 phosphorylation strictly required FAK (Fig. 5B). FAK is therefore required for NF-κB activation downstream of integrin activation and ligation.
To confirm the requirement for integrins in the flow system, we used a blocking antibody to FN that inhibits binding of integrins α5β1 and αvβ3 to the central cell-binding domain (35). Previous studies showed that short-term treatment of endothelial cells on FN with anti-FN antibody blocks free FN molecules and prevents new integrin binding, however, on this time scale does not disrupt existing adhesions (53, 54). Pretreatment of cells plated on FN with the blocking antibody 16G3 abolished NF-κB activation, whereas the nonblocking antibody 11E5 that binds a different region of FN had no effect (Fig. 5C). Taken together, these data demonstrate that activation of NF-κB by flow occurs though integrins, which requires FAK for induction of NF-κB p536 phosphorylation, whereas nuclear translocation is FAK independent.
Activation of NF-κB by TNF-α.
To determine whether FAK is required for NF-κB activation in response to all stimuli, cells were treated with TNF-α. Binding of TNF-α to its receptor 1 (TNFR-1) recruits a signaling complex consisting of TRADD, TRAF2, and RIP to the cytoplasmic domain of TNFR-1, which triggers the downstream activation of the IKK complex. A recent study using FAK−/− fibroblasts suggested a role for FAK in NF-κB activation by TNF-α (16). However, FAKfl/fl and FAK−/− MAECs treated with 10 ng/ml TNF-α showed no difference in phosphorylation of Ser536 (Fig. 6A) or ICAM-1 expression (Fig. 6B). These results argue that the requirement for FAK in NF-κB function in endothelial cells is not a general requirement.
Analysis of the upstream pathway.
Onset of flow activates NF-κB via a pathway that involves integrin activation, followed by their binding to matrix proteins, followed by activation of Rac (54), which leads to production of reactive oxygen species (ROS), activation of IKK, and degradation of IκB (3, 26, 47, 50). To determine at what point FAK is required, we first measured integrin activation using a FN fragment that binds α5β1- and αvβ3-integrins in an activation-dependent manner (13, 39). When cells were sheared for 5 min (the time at which integrin activation is maximal), the increase was independent of FAK expression (Fig. 7A). Increased Rac activity after flow was also independent of FAK expression (Fig. 7B). When cellular ROS was assayed, shear stimulated ROS in FAK−/− cells slightly better than controls (Fig. 8A). Though the reason for the increase is unclear, FAK is clearly not required for ROS production. To test whether FAK determines the sensitivity of NF-κB phosphorylation to ROS, cells were treated with 100 μM H2O2 for 30 min. Both cell types phosphorylated p65 to a similar extent (Fig. 8B). Thus, the ability of ROS to activate NF-κB is also independent of FAK. Degradation of IκBα in response to flow occurred similarly in both cell lines (Fig. 8C). This result suggests that IKK activation is independent of FAK, consistent with the movement of p65 to the nucleus (Fig. 2).
This study used onset of fluid shear stress as a simple model system to explore the role of FAK in NF-κB activation by flow. In vivo, various types of disturbed shear are associated with NF-κB activation and atherogenesis (9, 18). In vitro, NF-κB as well as other events associated with atherosclerosis are stimulated transiently by the onset of shear but are downregulated at later times (34, 40). By contrast, the same events are stimulated by oscillatory or other disturbed shear profiles in a sustained manner. We and others have proposed that disturbed shear induces continual stimulation of the integrin pathway, whereas cells in laminar flow adapt and downregulate these signals (7, 19). Thus, onset of shear provides a convenient means to investigate biologically relevant mechanisms.
Our studies used endothelial cells from mice with a floxed FAK gene that, on treatment with Cre recombinase, resulted in loss of detectable FAK. No change in the FAK homolog Pyk2 was observed, which removes a potential complicating factor. Two previous studies where FAK was deleted in endothelial cells obtained opposite results on this point, with one study reporting no change in Pyk2 levels (48) and a second reporting increased Pyk2 expression after Cre-mediated recombination (58). The reasons for these differences are unknown, but our results clearly demonstrate no change in Pyk2 under our conditions.
Previous studies showed that FAK phosphorylation increased in response to flow and that dominant-negative FAK inhibited activation of MAP kinases and SREBP (28, 31). Our data confirmed the role of FAK in SREBP cleavage, although we did not observe attenuation of ERK activation. However, Fujiwara and coworkers (42) showed that ERK activation by shear is quite rapid, similar to our data, and is a direct consequence of platelet endothelial cell adhesion molecule signaling. The rapidity of ERK activation is not consistent with the integrin pathway, which requires integrin activation and binding such that signals peak at 15–30 min (53, 56). However, it would not be surprising if dominant-negative FAK constructs have effects on MAP kinases that are not entirely specific.
Our major finding is that although FAK is not required for IκB degradation and nuclear translocation of NF-κB, it is essential for p65 phosphorylation and expression of ICAM-1. FAK is not required for any component of the known upstream events including integrin activation, Rac activation, or production of ROS. Thus, flow-induced NF-κB liberation and transactivation are at least partially regulated by distinct signaling pathways. By contrast, responses to TNF-α or hydrogen peroxide were FAK-independent.
IKK kinases are believed to phosphorylate p65 Ser536 (45), although IKK family members other than the classic IKKα or β may carry out this phosphorylation (1). Ser536 is within the transactivation domain, and its phosphorylation has been linked to transcriptional activity in several (22, 24, 59) but not all systems (37). How FAK controls p65 phosphorylation is currently unknown, and at present it is difficult to even propose viable hypotheses. However, it seems noteworthy that activated IKK localizes to focal adhesions in endothelial cells exposed to shear (38). This result supports the integrin dependence of this pathway and further suggests that localized FAK-dependent signaling events within the adhesion sites may explain the distinct effects of FAK on activation of NF-κB by shear or integrin vs. TNF-α or H2O2. Taken together, these data indicate an unexpected and potentially important role for FAK in the activation of endothelium in response to flow. This pathway may therefore contribute to the initiation of atherosclerosis in regions of disturbed flow.
This work was supported by United States Public Health Service Grant RO1 HL75092 to M. A. Schwartz. A. W. Orr and C. Hahn were supported by National Institutes of Health Training Grant 5T32 HL-7284. T. Petzold was supported by a Fulbright Foundation scholarship and American Heart student fellowship AHA 0525532U.
We thank David Bolick for help with endothelial cell isolation and Konstandinos Moissoglu for helpful discussions.
- Copyright © 2009 the American Physiological Society