HLA class I-mediated stress fiber formation requires ERK1/2 activation in the absence of an increase in intracellular Ca2+ in human aortic endothelial cells

Mary E. Ziegler, Yi-Ping Jin, Steven H. Young, Enrique Rozengurt, Elaine F. Reed

Abstract

Following transplantation, HLA class I antibodies targeting donor endothelium stimulate cell proliferation and migration, which contribute to the development of transplant vasculopathy and chronic allograft rejection. Dynamic remodeling of the actin cytoskeleton regulates cell proliferation and migration in endothelial cells (ECs), but the mechanism(s) involved remain incompletely understood. We explored anti-HLA class I antibody-mediated alterations of the cytoskeleton in human aortic ECs (HAECs) and contrasted these findings to thrombin-induced cytoskeleton remodeling. Our results identify two different signaling pathways leading to myosin light chain (MLC) phosphorylation in HAECs. Stimulation of HAECs with thrombin at 1 U/ml induced a robust elevation of intracellular Ca2+ concentration, increased MLC phosphorylation, and promoted stress fiber formation via MLC kinase (MLCK) and Rho kinase (ROK) in an ERK-independent manner. In contrast, HAECs stimulated with HLA class I antibodies did not promote any detectable change in intracellular Ca2+ concentration but instead induced MLC phosphorylation and stress fiber assembly via MLCK and ROK in an ERK1/2-dependent manner. Stimulation of HAECs with low-dose thrombin (1 mU/ml) induced signaling cascades that were similar to stimulation with HLA class I antibodies. HLA class I antibodies also stimulated the translocation of mammalian target of rapamycin complex 2 (mTORC2) and ERK1/2 from the cytoplasm to the plasma membrane independently of stress fiber assembly. These findings identify novel roles for HLA class I signaling in ECs and provide new insights into the role of ERK1/2 and mTORC2 in cytoskeleton regulation, which may be important in promoting transplant vasculopathy, tumor angiogenesis, and atherosclerosis.

  • HLA class I
  • endothelial cell
  • transplant vasculopathy
  • cytoskeleton
  • thrombin

the endothelial cell (EC) cytoskeleton functions to maintain an intact endothelial monolayer, and its remodeling plays a critical role in the restoration of the endothelium when it is disrupted (34). A major form of cytoskeleton reorganization involves the formation of stress fibers, caused by rearrangement of the cytoskeleton resulting in actin filament bundles and attachment to focal adhesions (8). Stress fibers function in EC adhesion, migration, and permeability (20, 46). It is well established that a stress fiber is a contractile actomyosin bundle, but the signals regulating its assembly remain incompletely understood (10, 20).

Central to the formation of the stress fiber is the phosphorylation of myosin light chain (MLC) at Thr18/Ser19 (28). The regulation of MLC phosphorylation at these sites is thought to be a consequence of at least two upstream signaling pathways. One mechanism of MLC phosphorylation involves the activation of Rho kinase (ROK), which is downstream of Ras homolog gene family member A (RhoA), and modulates the organization of stress fibers by stimulating the phosphorylation of MLC in a Ca2+-independent manner (29). ROK promotes MLC activation indirectly by phosphorylating and inactivating its inhibitor, myosin phosphatase targeting subunit 1 (MYPT1), an MLC phosphatase (MLCP) subunit, leading to increased phosphorylation of MLC. MLCP dephosphorylates MLC and consequently acts as a negative regulator of actomyosin contraction (6, 30, 39, 50, 62). The activation of ERK1/2 has been linked to Rho activation through phosphorylation and inhibition of RhoGAP p190A, which increases Rho-GTP leading to MLC phosphorylation via ROK (44). A second mechanism of MLC regulation utilizes the Ca2+/calmodulin-dependent MLC kinase (MLCK) to induce stress fibers and depends on several factors, which include elevated intracellular Ca2+ and protein kinase C activation. Additionally, the MEK1/2 and ERK1/2 pathway can activate cGMP-dependent protein kinase and subsequently activate MLCK (47). There is also evidence that ERK1/2 can directly phosphorylate MLCK thereby increasing its sensitivity to Ca2/calmodulin and its capacity to phosphorylate MLC (7, 16, 31). These signaling pathways can act in collaboration or separately depending on the cell type and upstream stimuli (32, 38, 40).

Thrombin induces MLC phosphorylation and activates stress fiber formation in ECs (3). Thrombin intracellular signaling is mediated by G-protein-coupled receptors (GPCRs) called proteinase-activated receptors (PARs). PAR1 is the major mediator of EC signaling following thrombin stimulation (11). PAR1 couples to the heterotrimeric GTP binding (G) proteins G12/13 and Gαq. G12/13 activate Rho GTPases, which regulate ROK and MLC phosphorylation. Gαq activates phospholipase C (PLC) to induce two second messengers: IP3 and DAG. IP3-mediated Ca2+ signaling leads to MLCK phosphorylation and the activation of MLC (11, 37). Ca2+ signaling, MLCK, ROK, and ERK1/2 have all been implicated in thrombin-induced stress fiber formation, but the precise contribution of these pathways remains unclear (3, 4). The physiologic response to thrombin can vary depending on the nature of the EC. For example, in pulmonary ECs, thrombin induces a very strong increase in intracellular Ca2+ concentration ([Ca2+]i) that declines rapidly to baseline levels. In contrast, in umbilical vein ECs, thrombin stimulates a persistent increase in [Ca2+]i (3). Additionally, the dose of thrombin can affect the downstream signaling cascades. For example, in tumor cells, thrombin has a bimodal effect on PAR1. At low concentrations growth signaling cascades are enhanced, while at higher concentrations thrombin impairs growth and induces apoptosis (60). The mechanism by which ERK1/2 mediates thrombin-induced stress fiber formation and MLC activation also remains controversial. In human vascular ECs, MEK inhibition had no effect on thrombin-induced stress fiber formation (56). In contrast, in bovine pulmonary artery ECs, while MEK inhibition did not affect MLC phosphorylation, it partially attenuated stress fiber formation. This mechanism of ERK1/2-mediated stress fibers was postulated to occur via calmodulin (CaM) kinase II in an MLCK-independent manner (4). The signaling pathways by which thrombin promotes stress fiber formation and MLC activation require clarification.

In addition to understanding thrombin-induced EC responses, deciphering the mechanism(s) of HLA class I-mediated EC signaling remains an important topic of investigation. Transplant vasculopathy is a key feature of chronic rejection that is manifested by diffuse, concentric intimal thickening culminating in vessel occlusion within the transplanted organ (49). Following transplantation, HLA class I antibodies targeting donor endothelium stimulate cell proliferation and migration, which contribute to the development of transplant vasculopathy and chronic allograft rejection. The ligation of HLA class I molecules by HLA class I antibodies on the surface of ECs induces signaling cascades, which regulate the cytoskeleton and activate the focal adhesion signaling complex via phosphorylation of focal adhesion kinase (FAK) and paxillin. HLA class I ligation on the surface of HAECs induces stress fiber formation (24, 35), and an intact cytoskeleton is required for HLA class I-mediated phosphorylation of the focal adhesion proteins (25). Class I ligation induces a rapid and prominent increase in RhoA activation, and inhibition of RhoGTPase or ROK prevents class I-mediated phosphorylation of FAK and paxillin and suppresses stress fiber formation (35). Additionally, RhoA is rapidly translocated to the cell membrane following HLA class I ligation in a manner that is associated with stress fiber formation and cytoskeleton reorganization (12). These signaling cascades, which alter the cytoskeleton following HLA class I ligation, are linked to cell proliferation pathways, which involve mammalian target of rapamycin (mTOR) activation (24, 27). In particular, mTORC2, which is composed of Rictor, Sin1, and GβL, has been shown to modulate the cytoskeleton through its stimulation of F-actin stress fibers, paxillin, RhoA, Rac1, Cdc42, and PKCα (23, 45). Despite its potential importance in the development of transplant vasculopathy, the signaling mechanism(s) underlying HLA class I-induced actin stress fiber formation have not been elucidated.

In this study, we examined the role of ERK1/2 and Ca2+ signaling in HLA class I- and thrombin-induced signaling pathways in HAECs leading to the formation of actin stress fiber formation. HLA class I ligation and thrombin stimulation at a low dose promoted actin stress fiber assembly that depended on ERK1/2 activation for MLC phosphorylation in the absence of a detectable increase in Ca2+ signaling. In contrast, thrombin, at a higher dose, stimulated a robust increase in [Ca2+]i that bypassed any requirement of ERK1/2 activation to induce stress fiber formation. In addition, we demonstrated that stimulation of HAEC with anti-HLA class I antibodies prompted the subcellular translocation of mTOR, Rictor, and inactive ERK1/2 to the plasma membrane. These findings implicate mTORC2 in the regulation of ERK1/2 signaling.

MATERIALS AND METHODS

Antibodies and chemicals.

The hybridoma (HB-95) W6/32, recognizing a monomorphic epitope on HLA class I, was purchased from the American Type Culture Collection. W6/32 was added to ECs for a final concentration of 0.1 μg/ml, which was previously determined to be optimal for examining ERK1/2 phosphorylation (26). The mouse IgG monoclonal antibody isotype control was purchased from Sigma-Aldrich. Human α-thrombin was from Enzyme Research Laboratories. Rabbit polyclonal antibodies against phospho-ERK1/2 (Thr202/Tyr204), phospho-MLC (Thr18/Ser19), mTOR, ERK, and raptor were purchased from Cell Signaling Technology and used for Western blot and/or immunoprecipitation. Goat polyclonal antibodies against mTOR and rabbit polyclonal phospho-MLC (Thr18/Ser19), Rictor, and ERK1/2 were purchased from Santa Cruz Biotechnology and used for confocal microscopy. Rabbit polyclonal anti-Rictor was purchased from Novus Biologicals and used for Western blot and immunoprecipitation. The Alexafluor 488 goat anti-rabbit and donkey anti-goat and the rhodamine red X goat anti-mouse antibodies and Texas red-phalloidin were from Invitrogen. The inhibitors ML-7, Y-27632, PD98059, and U0124 were from Calbiochem. U0126 was purchased from Promega. A23187 was from Sigma.

Cell culture.

HAECs were isolated from the aortic rings of explanted donor hearts, as described previously (59). Cells were cultured in M199 complete medium containing sodium pyruvate (1 mM; Irvine Scientific), penicillin, and streptomycin at 100 U/ml and 100 μg/ml, respectively (both from Invitrogen), 20% (vol/vol) heat-inactivated FBS (HyClone), heparin (90 μg/ml; Sigma-Aldrich), and endothelial cell growth supplement (20 μg/ml; Fisher Scientific).

Confocal microscopy.

HAECs were grown on 35-mm glass bottom dishes to 80–90% confluence and starved for 2 h in basal medium containing 2% FBS followed by treatment with the specified agonists or inhibitors. Treated HAECs were fixed with 4% paraformaldehyde and permeabilized with 0.2% Triton-X-100. Cells were incubated with primary antibody in 5% BSA overnight at 4°C, washed, and incubated with a fluorescent conjugated secondary antibody for 30 min at room temperature. The presence of F-actin was visualized by direct staining with Texas Red-phalloidin (Molecular Probes). Cell images were captured using a Zeiss LSM 510 confocal microscope at ×63 magnification using the Zeiss LSM 5 PASCAL software (Carl Zeiss MicroImaging). Colocalization was determined by the ImageJ plugin Colocalization Finder (http://rsbweb.nih.gov/ij/plugins/colocalization-finder.html).

Stress fiber quantitation.

To analyze stress fibers, ×63 magnification images were used. We utilized the ImageJ threshold tool for each image to determine an optical density limit that encapsulated most of the visible F-actin bundles, eliminating background fluorescence. The ROI manager was used to analyze the stress fiber intensities. Four independent experiments were performed and in each experiment three fields were measured. The ratio between the mean fluorescence intensity and the cell area was calculated to account for changes in cell shape due to contraction. The data were normalized to the control group, and the others groups were calculated to reflect the change compared with the control group. A Student's t-test was performed to determine significant changes between the groups, P < 0.05.

Western blot analysis.

Serum-starved cultures of HAECs were stimulated, washed with ice-cold PBS, and lysed in buffer (containing 20 mM Tris pH 7.9, 137 mM NaCl, 5 mM EDTA, 1 mM EGTA, 10% glycerol, 1% Triton X-100, 10 mM NaF, 1 mM PMSF, 1 mM Na3VO4, 10 μg/ml aprotinin, and 10 μg/ml leupeptin) for 10 min on ice. Protein concentration was determined using the BCA protein assay kit (Pierce). Lysates were mixed with 2× SDS loading buffer, boiled, and run on a SDS-PAGE gel, and proteins were transferred overnight to Immobilon-P membranes (Millipore). Membranes were blocked using 5% BSA or 5% nonfat milk in TBS-Tween for 15 min and incubated overnight at 4°C with the appropriate antibodies. Primary antibodies to immunoreactive bands were visualized using horseradish peroxidase-conjugated anti-rabbit, anti-mouse, or anti-goat antibodies (Santa Cruz Biotechnology). The Western blot quantification was performed using ImageJ densitometry software. Each band was normalized to the loading control and the intensity was calculated relative to control of each experiment.

Fluorimetry.

HAECs were plated onto rectangular glass coverslips, grown to ∼80% confluence, and serum starved for 2 h before measurement. Cells were then incubated in HBSS containing 1.8 mM Ca2+ and the Ca2+ indicator fura-2-AM at 5 mM (Molecular Probes, Invitrogen) for 20 min, then washed once with HBSS. Coverslips were mounted in a standard 1-cm path length cuvette filled with saline (37°C) using a special holder (ANO-2100; Hitachi Instruments). The cuvette was placed in a fluorimeter (F-2000, Hitachi Instruments) with a heated jacket (37°C), and the solution was continuously stirred using a small magnetic stir bar. Small volumes of 200× concentrated agonist solutions were introduced into the bottom one-third of the cuvette, with a Hamilton syringe. All concentrations reported are the final steady-state mixed value. Injection was completed within 1 s. Measurements of mixing kinetics showed that introduced test solutions were completely mixed (at the level of the detection window-about the middle 1/3 of the cuvette) within 2 s and with no sizable overshoot. The size of the detection window allowed measurement on the order of 105 cells. Excitation was set to 340 and 380 nm, and emission signal was collected at 380 nm, all with a 10-nm bandwidth. Samples were taken every 0.5 s using associated software (F-2000 Intracellular Cation Measurement System; Hitachi Instruments). The software created the 340/380 nm ratios, which are proportional to intracellular Ca2+ concentration.

Immunoprecipitation.

HAECs were grown in 100-mm dishes to 80–90% confluence and serum starved before immunoprecipitation. They were rinsed once with ice-cold PBS and lysed in ice-cold lysis buffer (40 mM HEPES pH 7.5, 120 mM NaCl, 1 mM EDTA, 10 mM β-glycerophosphate, 50 mM NaF, 1.5 mM Na3VO4, 0.3% CHAPS, 10 μg/ml aprotinin, and 10 μg/ml leupeptin). Samples were sonicated for 5 min and placed on a rotator at 4°C for 20 min. After lysis, cell debris was removed by centrifugation at 14,000 rpm for 10 min. Then, 4–8 μg of the appropriate antibody were added to the cleared supernatant, and the samples were placed on a rotator overnight. Protein A/G beads (60 μl) were added to pull down immune complexes. Immunoprecipitates were washed four times in wash buffer (10 mM HEPES pH 7.5, 50 mM β-glycerophosphate, and 5 mM NaCl), 30 μl of 2× SDS loading buffer were added, and samples were boiled 5 min and loaded onto a 6 or 12% SDS-PAGE gel.

RESULTS

Characterization of thrombin-induced cytoskeleton regulation in HAECs.

Initially, we examined the effect of either 1 U/ml or 1 mU/ml of thrombin on actin stress fiber formation, MLC phosphorylation (Thr18/Ser19), and [Ca2+]i. Stimulation with thrombin at 1 U/ml induced a striking increase in [Ca2+]i that peaked at 60 sec and declined to a plateau level that was maintained for the duration of the experiment (Fig. 1A). In striking contrast to the results obtained using thrombin at 1 U/ml, stimulation of HAECs with thrombin at 1 mU/ml showed no detectable increase in [Ca2+]i (Fig. 1A).

Fig. 1.

Characterization of thrombin-induced cytoskeleton regulation in human aortic endothelial cells (HAECs). A: HAECs were stimulated with thrombin at 1 U/ml (indicated by black line) or thrombin at 1 mU/ml (indicated by the grey line) and intracellular Ca2+ concentration ([Ca2+]i) was measured by fluorimetry. Arrow at 30 s indicates the time of thrombin injection. Ratio of emission at 340 nm/380 nm is proportional to the [Ca2+]i. HAECs were untreated (control) or were stimulated with thrombin (1 U/ml, 10 min or 1 mU/ml, 10 min) alone or in combination with a pretreatment of the MEK inhibitor U0126 (10 μM, 30 min) or the inactive analog U0124 (10 μM, 30 min). B: F-actin and phosphorylated myosin light chain (MLC-p; Thr18/Ser19) were visualized by confocal microscopy. Images are representative of 4 independent experiments. C: stress fibers were quantitated as described in the methods and normalized to the control group. *P < 0.0001. D: HAEC lysates were analyzed by Western blot to determine the phosphorylation of ERK1/2 (Thr202/Tyr204) and MLC (Thr18/Ser19). Vinculin was used as a loading control. This Western blot is representative of 3 independent experiments. E: HAEC lysates were analyzed by Western blot to determine the phosphorylation of ERK1/2 (Thr202/Tyr204) and MLC (Thr18/Ser19). Vinculin was used as a loading control. This Western blot is representative of 3 independent experiments.

To characterize signaling pathways leading to actin remodeling in HAECs, cells were stimulated with thrombin at 1 U/ml or at 1 mU/ml alone or in combination with the MEK inhibitor U0126, and the cytoskeleton and MLC phosphorylation were imaged by confocal microscopy (Fig. 1B). Treatment of HAECs with thrombin at 1 U/ml for 10 min stimulated a 60% increase (P = 0.0004) in stress fiber formation compared with control and concomitant MLC phosphorylation. At 1 U/ml, thrombin-induced stress fiber formation and MLC phosphorylation was not affected by suppression of MEK/ERK1/2 signaling (Fig. 1, B and C). At 1 mU/ml, thrombin stimulated a 53% increase in stress fiber formation (P < 0.0001), which was accompanied by MLC phosphorylation, and both were prevented by ERK inhibition with U0126 (Fig. 1, B and C).

Next, by Western blot, we determined the activation of ERK1/2 and MLC phosphorylation following thrombin stimulation. Thrombin (1 U/ml) induced robust phosphorylation of ERK1/2 at Thr202/Tyr204 and MLC at Thr18/Ser19 (Fig. 1D). Pretreatment of HAECs with the MEK inhibitor U0126 prevented ERK1/2 activation induced by thrombin (1 U/ml) but did not block MLC phosphorylation (Fig. 1D). U0124, an inactive analog of the MEK1/2 inhibitor U0126, did not block ERK1/2 or MLC phosphorylation. In contrast, thrombin-induced MLC phosphorylation was reduced by inhibitors of ROK and MLCK (data not shown).

The treatment of HAECs with thrombin at 1 mU/ml stimulated the phosphorylation of ERK1/2 (Thr202/Tyr204) and MLC (Thr18/Ser19). At this dose, exposure to U0126 virtually abolished MLC phosphorylation whereas the inactive analog U0124 did not affect ERK1/2 or MLC phosphorylation (Fig. 1E). ROK and MLCK inhibition decreased thrombin (1 mU/ml)-induced MLC phosphorylation (data not shown).

Collectively, the data demonstrated that in the same cell type, thrombin activated two different signaling pathways depending on its dose. One pathway, elicited by a high dose of thrombin (1 U/ml), induced an elevation of [Ca2+]i, activated MLC, and caused stress fiber formation via ROK and MLCK. This pathway did not require functional ERK1/2. The alternative pathway, activated by a low dose of thrombin (1 mU/ml), did not increase [Ca2+]i, but increased the phosphorylation of MLC and induced stress fiber formation. In this case, MLC activation and stress fiber formation not only depended on ROK and MLCK but also on ERK1/2.

Characterization of HLA class I-induced cytoskeleton regulation in HAECs.

Based on the data described above and our previous studies (26) demonstrating that engagement of HLA class I molecules on HAECs led to ERK activation, we postulated that, similar to thrombin at 1 mU/ml, ligation of HAECs by HLA class I antibodies would also regulate the cytoskeleton in an ERK-dependent manner. Ligation of class I molecules with antibodies (0.1 ug/ml) did not produce a detectable increase in [Ca2+]i (Fig. 2A). HAECs were stimulated with an anti-HLA class I antibody alone or in combination with MEK (U0126), ROK (Y-27632), or MLCK (ML-7) inhibitors and then imaged by confocal microscopy. HLA class I ligation stimulated a significant increase in actin stress fiber formation (56%, P < 0.0001; Fig. 2, B and C) (24, 35). Pretreatment of HAEC with U0126, Y-27632, or ML-7 substantially inhibited HLA class I-induced actin stress fiber formation and MLC phosphorylation (Fig. 2, B and C). These data established a role for ERK1/2 and MLC in HLA class I-induced stress fiber formation.

Fig. 2.

Characterization of HLA class I-induced cytoskeleton regulation in HAECs. A: HAECs were stimulated with the HLA class I antibody W6/32 (0.1 μg/ml, 10 min) and [Ca2+]i was measured by fluorimetry. Arrow at 30 s indicates the time of agonist injection. Ratio of emission at 340/380 nm is proportional to the [Ca2+]i. HAECs were treated with an isotype control mouse IgG antibody (control; 0.1 μg/ml) or the HLA class I antibody (0.1 μg/ml, 10 min) alone or in combination with a pretreatment of the MEK inhibitors U0126 or PD98059 (both at 10 μM, 30 min) or the MLCK inhibitor ML-7 (25 μM, 30 min) or the Rho kinase (ROK) inhibitor Y-27632 (10 μM, 30 min). B: F-actin and MLC-p (Thr18/Ser19) were visualized by confocal microscopy. Images are representative of 4 independent experiments. C: stress fibers were quantitated as described in materials and methods and normalized to the control group. *P < 0.0001. D–F: Western blot bands were measured as described in materials and methods and the columns below each band correspond to the phosphorylation of ERK1/2 (Thr202/Tyr204; black) and MLC (Thr18/Ser19; grey). **P < 0.001 and * P < 0.05. D: vinculin was used as a loading control. Data are representative of 5 independent experiments. E: vinculin was used as a loading control. Data are representative of 4 independent experiments. F: total MLC was used as a loading control. Data are representative of 3 independent experiments.

To further examine this signaling pathway, HLA class I-induced ERK1/2 and MLC phosphorylation were determined by Western blot analysis. HLA class I ligation stimulated phosphorylation of ERK1/2 at Thr202/Tyr204 and MLC at Thr18/Ser19. MEK1/2 inhibition by either PD98059 or U0126 blocked class I-induced ERK1/2 and MLC phosphorylation (Fig. 2D). In addition, inhibition of both MLCK and ROK reduced MLC phosphorylation but did not alter class I-induced ERK1/2 phosphorylation (Fig. 2, E and F, respectively). Taken together, these data demonstrate that HLA class I ligation did not alter [Ca2+]i but instead induced MLC phosphorylation and stress fiber formation in a manner that depended on ROK, MLCK, and ERK1/2.

Increased [Ca2+]i levels bypass the requirement of ERK1/2 activation for stress fiber formation.

In view of these results, we hypothesized that an elevation of [Ca2+]i would bypass the requirement of ERK1/2 activation for inducing MLC phosphorylation and stress fiber formation in response to HLA class I antibodies or thrombin at 1 mU/ml. To test this hypothesis, we treated HAECs with the Ca2+ ionophore A23187 (41) and F-actin and MLC phosphorylation were assessed by confocal microscopy. Treatment of ECs with A23187 alone induced both stress fibers (58%, P < 0.0001) and MLC phosphorylation (Figs. 3, A and B). Stimulation of HAECs with HLA class I antibodies plus Ca2+ ionophore (Figs. 3, A, c and B) or thrombin (1 mU/ml) with Ca2+ ionophore (Fig. 3, A, d and B) mediated a significant increase in stress fiber formation (55 and 52%, respectively; P < 0.0001) that was accompanied by phosphorylation of MLC. Pretreatment of HAECs with U0126 inhibited HLA class I and thrombin (1 mU/ml)-mediated stress fiber formation (Fig. 2B and 1B, respectively). The addition of Ca2+ ionophore to class I- and thrombin-treated HAECs canceled the effect of U0126 and stress fibers were formed [Fig. 3A, d (76% stress fiber increase) and Fig. 3A, f (58% stress fiber increase)]. These data illustrated that when [Ca2+]i levels were elevated, MLC phosphorylation and stress fiber formation bypass the requirement of functional ERK1/2.

Fig. 3.

Altering the cytoskeleton response by elevating [Ca2+]i. HAECs were prepared for staining and F-actin and MLC-p (Thr18/Ser19) were visualized by confocal microscopy. A, a: HAECs were treated with an isotype control mouse IgG antibody (0.1 μg/ml; control). A, b: HAECs were stimulated with calcium ionophore A23187 (1 μM, 10 min) alone. A, c: HAECs were treated with the HLA class I antibody in combination with A23187. A, d: HAECs pretreated with U0126 (10 μM, 30 min) and then stimulated with HLA class I and A23187. A, e: HAECs were stimulated with thrombin (1 mU/ml, 10 min) and A23187 (1 μM, 10 min) alone. A, f: HAECs were stimulated in combination with a pretreatment of U0126 (10 μM, 30 min). These images are representative of 3 independent experiments. B: stress fibers were quantitated as described in materials and methods and normalized to the control group. *P < 0.0001.

Characterization of the subcellular localization of Rictor and ERK1/2.

Previous studies from our laboratory (26), indicated that HLA class I ligation activated ERK1/2 (Thr202/Tyr204) in a Rictor/mTORC2-dependent manner. We therefore investigated the possibility that Rictor and ERK1/2 function together in HLA class I signaling in HAECs to regulate the cytoskeleton. HAECs were stimulated with HLA class I antibody or control IgG, immunoprecipitated with antibodies to Rictor, and immunoblotted with antibodies to total and phosphorylated forms of ERK1/2. Immunoblotting with an antibody to ERK1/2 showed that ERK1/2 was present in the Rictor complex in both treatment groups (Fig. 4A). ERK1/2 was also detected in the complex after immunoprecipitation of mTOR (Fig. 4B), but ERK was not coprecipitated when lysates were immunoprecipitated with antibodies to Raptor (Fig. 4C). Stimulation with HLA class I antibody increased complex formation between Rictor and phosphorylated ERK1/2 (Fig. 4A). Similarly, HLA class I antibody stimulation increased complex formation between mTOR and phospho-ERK1/2 (Fig. 4B). Reverse immunoprecipitation with Rictor confirmed that both mTOR and ERK1/2 were in the Rictor complex. Reciprocal ERK1/2 pull down showed both mTOR and Rictor, but not Raptor, complexed with ERK1/2 (Fig. 4D).

Fig. 4.

Rictor and ERK1/2 are in an immune-complex together. A–C: HAECs were treated with an isotype control mouse IgG antibody (−) (0.1 ug/ml) or were stimulated with the HLA class I antibody (+) (0.1 ug/ml, 10 min). Cell lysates were immunoprecipitated with antibodies to Rictor (A), mammalian target of rapamycin (mTOR; B) and Raptor (C) and subjected to SDS-PAGE and the presence of Rictor, mTOR, ERK1/2, and phospho-ERK1/2 (Thr202/Tyr204) were determined by Western blot as indicated. D: HAEC lysates were immunoprecipitated with an anti-Rictor antibody or an anti-ERK1/2 antibody. Complexes and the total cell lysates were subjected to SDS-PAGE and the presence of mTOR, Raptor, Rictor, and ERK were determined by Western blot. All immunoprecipitations (A–D) are representative of at least 5 independent experiments.

To further understand the molecular interactions between ERK1/2 and Rictor, the localization of these proteins was determined in HAECs by confocal microscopy. Rictor was distributed throughout the cytoplasm and concentrated in the nucleus in unstimulated HAECs, whereas nonphosphorylated ERK1/2 was primarily concentrated in the nucleus and showed some cytoplasmic distribution. Areas of colocalization (white) were identified in both the cytoplasm and the nucleus (Fig. 5A, top overlay). Treatment of HAEC with class I antibodies triggered increased Rictor and ERK1/2 colocalization (white; Fig. 5A, bottom). This observation was consistent with the Rictor immunoprecipitation results presented in Fig. 4A, which showed increased complex formation between Rictor and ERK1/2 after HLA class I ligation. Colocalization of Rictor and ERK1/2 in class I-treated HAECs was prominent in the cytoplasm and at the plasma membrane (Fig. 5A, bottom overlay).

Fig. 5.

Rictor and a nonphosphorylated form of ERK1/2 colocalize at the EC membrane. HAECs were treated with an isotype control mouse IgG antibody (0.1 μg/ml; control; top) or were stimulated with the HLA class I antibody (0.1 μg/ml, 10 min; bottom). A: Rictor and nonphosphorylated ERK1/2 were visualized by confocal microscopy. Colocalization is shown in the last column where white indicates areas of colocalization (green represents Rictor and red represents ERK1/2). Manders' Overlay coefficients were 0.914 (control) and 0.928 (HLA class I). These images are representative of 4 independent experiments. B: phospho-ERK1/2 (Thr202/Tyr204) (p-ERK) and Rictor were visualized by confocal microscopy. Colocalization is shown at right where white indicates areas of colocalization (green represents p-ERK1/2 and red represents Rictor). Manders' Overlay coefficients were 0.939 (control) and 0.892 (HLA class I). These images are representative of 3 independent experiments. C: mTOR and Rictor were visualized by confocal microscopy. Overlay is shown at right where white indicates areas of colocalization (green represents mTOR and red represents Rictor). Manders' Overlay coefficients were 0.940 (control) and 0.969 (HLA class I). Images are representative of 3 independent experiments. D: HAECs were pretreated with the MEK inhibitor U0126 (10 μM, 30 min) and then stimulated with the HLA class I antibody (0.1 μg/ml, 10 min). HAECs were prepared for staining and F-actin and Rictor were visualized by confocal microscopy. These images are representative of 3 independent experiments.

To further determine the activation status of ERK complexed with Rictor, the localization of Rictor and phospho-ERK1/2 was determined in unstimulated and HLA class I stimulated HAECs. Rictor and phospho-ERK1/2 colocalized (white) in the nucleus of unstimulated cells (Fig. 5B, top). Treatment with class I antibodies significantly increased localization of Rictor at the plasma membrane (Fig. 5B, bottom middle). Colocalization between Rictor and p-ERK1/2 was also increased following class I ligation (Fig. 5B, bottom). Interestingly, total ERK1/2 (Fig. 5A), but not phosphorylated ERK1/2, was found colocalized at the plasma membrane with Rictor (Fig. 5B). Thus the Rictor and ERK1/2 complex that translocated to the plasma membrane following class I ligation contained a nonphosphorylated form of ERK1/2. To confirm the immunoprecipitation data, we determined whether the complex of Rictor and ERK1/2 at the membrane was part of mTORC2 by observing the colocalization of Rictor and mTOR. In unstimulated HAECs, Rictor and mTOR colocalized (white) throughout the cell, with evidence of mTOR localized without Rictor (Fig. 5C). After HLA class I stimulation, Rictor translocated with mTOR and colocalized at the plasma membrane (white) (Fig. 5C).

Finally, to determine whether class I-mediated translocation of Rictor and nonphosphorylated ERK1/2 to the plasma membrane in HAECs required ERK1/2 phosphorylation and/or stress fiber formation, HAECs were stimulated alone or in combination with U0126. The cells were analyzed by confocal microscopy to characterize Rictor localization and F-actin stress fiber formation. Treatment with the ERK inhibitor U0126 reduced HLA class I-induced stress fiber formation. However, inhibition of ERK1/2 by U0126 did not alter class I-induced Rictor membrane translocation (Fig. 5D). These results indicate that HLA class I ligation on HAEC triggers translocation of Rictor to the plasma membrane in an ERK-independent manner.

DISCUSSION

The phosphorylation of MLC is the central event leading to actin reorganization and stress fiber formation. We report for the first time that ligation of HLA class I molecules on the surface of HAECs triggers the phosphorylation of MLC at Thr18/Ser19 and remodeling of the actin cytoskeleton in a manner that employed ROK, MLCK, and ERK1/2 but in the absence of any detectable change in [Ca2+]i. Notably, HLA class I signaling was similar to the signaling pathway triggered by the treatment of HAECs with a low dose of thrombin (1 mU/ml) in that it did not increase [Ca2+]i but instead induced MLC phosphorylation and actin stress fiber formation in an ERK1/2-dependent manner. These findings contrast with the results obtained when HAECs were stimulated with a high dose of thrombin (1 U/ml), which significantly increased [Ca2+]i and caused MLC phosphorylation and actin remodeling in an ERK1/2-independent manner.

Our data revealed that the two fundamental regulators of MLC activation, ROK and MLCK, are required for MLC phosphorylation after stimulation with thrombin at 1 U/ml and at 1 mU/ml or HLA class I ligation in HAECs. Differential regulation of MLC activation was revealed when we examined [Ca2+]i and ERK1/2 activation. We propose that thrombin at 1 U/ml utilizes a pathway that is distinct from the pathway employed by lower concentrations of thrombin and HLA class I ligation to activate MLCK and ROK to induce stress fibers. The key element mediating MLCK and ROK-induced stress fiber formation involved their capacity to stimulate elevations in [Ca2+]i. In the absence of an [Ca2+]i increase, ERK1/2 was the major regulator of stress fiber formation. The notion that thrombin can utilize different pathways to trigger MLC activation and stress fiber formation is in agreement with previously reported work in bovine pulmonary artery ECs. A thrombin-induced MLCK-dependent pathway was shown to regulate cytoskeleton rearrangement, and, in a parallel pathway, ERK1/2 acting via CaM kinase II functioned to elicit stress fibers in an MLCK-independent manner (4, 17). In the clinical setting, understanding the subtleties of thrombin signaling cascades may contribute to the development of better anticoagulants that would benefit patients with coronary syndromes (14).

Inhibition of the ERK1/2 pathway impairs MLCK and MLC phosphorylation and hinders cell migration in ECs, fibroblasts, tumor cells, and osteoblasts in response to cell matrix proteins and growth factors (1, 7, 15, 22, 31, 33, 48, 58). The activation of MLCK via ERK1/2 may be important in focal adhesion turnover, which is essential for cell migration (54). Migration signaling occurs via integrin β4-induced ERK1/2 activation (42). We (61) recently demonstrated that HLA class I molecules partner with integrin β4 to transduce signals resulting in HAEC migration and proliferation. This molecular interaction is required for HLA class I-induced ERK1/2 activation and EC migration, as knockdown of integrin β4 abrogates HLA I Ab-induced signaling (61). The current findings suggest that ERK1/2 may be a crucial regulator of HLA class I and integrin β4-induced migration via MLCK-induced MLC phosphorylation. Our findings are consistent with signaling cascades in kidney tubular cells following depolarizing stimuli, where the phosphorylation of MLC does not require an increase in [Ca2+]i but instead relies on ERK to activate Rho and ROK (53, 57). Furthermore, it was recently demonstrated that when ECs are activated by monocyte adherence the HRas/Raf/MEK/ERK signaling pathway leads to MLC phosphorylation. This activation results in the recruitment of Src to VE-cadherin and phosphorylation of VE-cadherin, which ultimately results in EC gap formation and enhanced transendothelial migration of monocytes (18). A hallmark of antibody-mediated rejection is the accumulation of intravascular leukocytes and platelet in the capillaries of the graft (19). Our current findings, in conjuction with the aforementioned study, suggest that the phosphorylation of MLC may be central in the regulation of leukocyte recruitment within the graft and should be further explored.

Another key finding in our study is ligation of HLA class I molecules with antibodies stimulates mTORC2 complex formation with ERK1/2 (26). We show, for the first time, that Rictor, mTOR, and ERK1/2 are in a molecular complex. Notably, we provide evidence that following HLA class I ligation, mTORC2 and a nonphosphorylated form of ERK1/2 translocate and colocalize at the plasma membrane. The subcellular localization of ERK is essential for proper signal transduction and is regulated by protein-protein interactions (2). One study found that ERK2 interacts with many proteins in its nonphosphorylated state, whereas phospho-ERK2 associates with fewer proteins. The dynamic interaction between phospho-ERK2 and non-phospo-ERK2 with other proteins was reported to be modulated by Ca2+ concentration (9). Our data suggest that mTORC2 transports inactive-ERK1/2 to the plasma membrane and this localization might be necessary to permit HLA class I-mediated ERK1/2 activation and dissociation of phosphorylated ERK1/2 from the memebrane. This implies that mTORC2 localization at the plasma membrane serves as a scaffold for ERK1/2 to facilitate complex assembly with proteins such as Rac, p21-activated kinase (PAK), and MEK1/2 that enable its release from the membrane. Indeed, recent studies underscore the importance of subcellular localization for understanding the mTORC2/Akt signaling axis in ECs. The localization of mTORC2 in lipid rafts has been shown to be important for PKC-α-dependent activation of Akt (43). In addition, Rictor was shown to functionally regulate the localization of F-actin enrichment. The spatial regulation of F-actin assembly is PKC-dependent, and it was suggested that mTORC2 must be translocated to the plasma membrane in order for this to happen (36). Localization of mTORC2 to the plasma membrane potentially could occur through the plekstrin homology domain of the Sin1 component (5).

Based on published data and our current findings, we propose the model illustrated in Fig. 6 to explain how varying concentrations of thrombin and HLA class I molecules differentially regulate signaling pathways to promote MLC phosphorylation (Thr18/Ser19) and actin remodeling. Thrombin intracellular signaling is mediated by PARs that couple with distinct G protein subunits to differentially activate downstream signaling pathways (11). As a consequence of thrombin stimulation at a high dose (1 U/ml), we propose that PAR1 couples with G12/13 to activate RhoGTP and ROK and phosphorylate MLCP (11). PAR1, activated by thrombin at 1 U/ml, also couples with Gαq to increase [Ca2+]i leading to MLCK phosphorylation and its inactivation. Coordinated activation of both of these pathways by thrombin (1 U/ml) culminates in MLC phosphorylation and stress fiber formation. In contrast, we propose that when PAR1 is activated by thrombin at lower dose (1 mU/ml) it primarily couples with G12/13 to activate ROK. Activation of PAR1 by low-dose thrombin also couples with the Gβγ subunit to activate Rac, PAK, MEK, and ERK1/2 phosphorylation which promotes MLCK activation. Thrombin (1 mU/ml)-mediated activation of these two G-protein pathways culminate in the phosphorylation of MLC and stress fiber formation. Evidence supporting this model comes from recent studies showing PAK and MLC phosphorylation in ECs downstream of Rac activation in response to serum, VEGF, bFGF, TNFα, histamine, and thrombin. Furthermore, induction of MLC phosphorylation and stress fiber formation by PAK is mediated by mitogen-activated protein kinase (51, 52, 55).

Fig. 6.

Proposed model describing the signaling cascades leading to HAECs cytoskeleton remodeling following either thrombin stimulation or HLA class I ligation. Thrombin stimulation at 1 U/ml signals through PAR1 and activates G12/13 to induce RhoGTP and ROK. ROK activation leads to MLCP phosphorylation, which results in its inactivation. In addition, thrombin at 1 U/ml utilizes Gαq to induce an [Ca2+]i increase, which promotes MLCK phosphorylation. Both of these pathways, activated by thrombin (1 U/ml), culminate in MLC phosphorylation (Thr18/Ser19) and stress fiber formation. When the dose is lowered, thrombin (1 mU/ml) utilizes PAR1 and activates, primarily, G12/13 to induce ROK. Low-dose thrombin depends on ERK1/2 activation and we forecast that this may occur via the Gβγ subunit, which activates Rac to induce PAK, MEK, and ERK1/2 phosphorylation (Thr202/Tyr204) and then promotes MLCK activation. Thrombin (1 mU/ml) utilizes these two G-protein pathways to activate MLC and induce stress fiber formation. HLA class I ligation mimics a pattern comparable to thrombin at 1 mU/ml, but does not occur via PAR1. Thus we postulate that HLA class I ligation induces the translocation of mTORC2 and inactive ERK1/2 to the plasma membrane. Rac is then activated to induce PAK and MEK activation induces ERK1/2 (Thr202/Tyr204) where it is released from mTORC2 at the membrane. Phospho-ERK1/2 activates one pathway to induce RhoGTP and ROK signaling and another pathway, which promotes the induction of MLCK. Together, these two ERK1/2-mediated signaling cascades elicit stress fiber formation following class I ligation.

Similar to stimulation with a low dose of thrombin, HLA class I signaling also requires phospho-ERK1/2 in lieu of increased [Ca2+]i to phosphorylate MLC and induce actin stress fibers. Our results are consistent with a model (Fig. 6) whereby antibody-mediated crosslinking of HLA class I molecules causes the translocation of mTORC2 and inactive ERK1/2 to the plasma membrane. Rac signaling initiates PAK and MEK activation and the phosphorylation of ERK1/2 (51). Once phosphorylated, ERK1/2 dissociates from mTORC2 at the memebrane it can induce RhoGTP and ROK signaling (44). Activated ERK1/2 also functions to induce MLCK (31, 51). Together, these two ERK1/2-mediated pathways trigger stress fiber formation following HLA class I ligation.

In conclusion, this study presents evidence that HLA class I-induced cytoskeleton remodeling utilizes both ROK and MLCK, the two fundamental pathways required for MLC activation. Importantly, we found that MLCK-induced MLC phosphorylation occurs in the absence of an [Ca2+]i increase. Thus, following HLA class I ligation, ERK1/2 activation bypasses the requirement for an [Ca2+]i increase to induce cytoskeleton remodeling in HAECs. The drug captopril was shown to suppress transplant coronary artery disease and its mechanism of action in an anti-hypertension tension model was correlated with the inhibition of ERK activation and phospho-MLC expression (13, 21). Our data suggest that HLA class I-mediated ERK1/2 and MLC activation may influence the development of acute and chronic antibody-mediated rejection and transplant vasculopathy by controlling stress fiber and focal adhesion complex formation, both of which are involved in cell migration and proliferation. Thus efforts to inhibit EC activation by blocking HLA class I antibody-induced signal transduction may be of potential therapeutic value to prevent chronic antibody-mediated transplant rejection. The ERK pathway plays a pivotal role in cell signaling and finding differential affects with HLA class I-induced cytoskeleton changes compared with thrombin gave insight into the subtleties of cytoskeleton remodeling that should be further explored.

GRANTS

Funding for this work was supported by the National Science Foundation Graduate Research Fellowship (to M. E. Ziegler) and National Institutes of Health Grants U01-AI-077821, RO1-AI-042819, and RO1-HL-090995 (to E. F. Reed) and P30-DK-41301, RO1-055003, and RO1-056930 (to E. Rozengurt).

DISCLOSURES

No conflicts of interest, financial or otherwise, are declared by the author(s).

AUTHOR CONTRIBUTIONS

Author contributions: M.E.Z., E.R., and E.F.R. conception and design of research; M.E.Z., Y.-P.J., and S.H.Y. performed experiments; M.E.Z. and S.H.Y. analyzed data; M.E.Z., E.R., and E.F.R. interpreted results of experiments; M.E.Z. prepared figures; M.E.Z. drafted manuscript; M.E.Z., E.R., and E.F.R. edited and revised manuscript; M.E.Z., Y.-P.J., S.H.Y., E.R., and E.F.R. approved final version of manuscript.

REFERENCES

  1. 1.
  2. 2.
  3. 3.
  4. 4.
  5. 5.
  6. 6.
  7. 7.
  8. 8.
  9. 9.
  10. 10.
  11. 11.
  12. 12.
  13. 13.
  14. 14.
  15. 15.
  16. 16.
  17. 17.
  18. 18.
  19. 19.
  20. 20.
  21. 21.
  22. 22.
  23. 23.
  24. 24.
  25. 25.
  26. 26.
  27. 27.
  28. 28.
  29. 29.
  30. 30.
  31. 31.
  32. 32.
  33. 33.
  34. 34.
  35. 35.
  36. 36.
  37. 37.
  38. 38.
  39. 39.
  40. 40.
  41. 41.
  42. 42.
  43. 43.
  44. 44.
  45. 45.
  46. 46.
  47. 47.
  48. 48.
  49. 49.
  50. 50.
  51. 51.
  52. 52.
  53. 53.
  54. 54.
  55. 55.
  56. 56.
  57. 57.
  58. 58.
  59. 59.
  60. 60.
  61. 61.
  62. 62.
View Abstract