Vitamin E Isoforms Differentially Regulate Intercellular Adhesion Molecule-1 Activation of PKCα in Human Microvascular Endothelial Cells

Aims ICAM-1-dependent leukocyte recruitment in vivo is inhibited by the vitamin E isoform d-α-tocopherol and elevated by d-γ-tocopherol. ICAM-1 is reported to activate endothelial cell signals including protein kinase C (PKC), but the PKC isoform and the mechanism for ICAM-1 activation of PKC are not known. It is also not known whether ICAM-1 signaling in endothelial cells is regulated by tocopherol isoforms. We hypothesized that d-α-tocopherol and d-γ-tocopherol differentially regulate ICAM-1 activation of endothelial cell PKC. Results ICAM-1 crosslinking activated the PKC isoform PKCα but not PKCβ in TNFα-pretreated human microvascular endothelial cells. ICAM-1 activation of PKCα was blocked by the PLC inhibitor U73122, ERK1/2 inhibitor PD98059, and xanthine oxidase inhibitor allopurinol. ERK1/2 activation was blocked by inhibition of XO and PLC but not by inhibition of PKCα, indicating that ERK1/2 is downstream of XO and upstream of PKCα during ICAM-1 signaling. During ICAM-1 activation of PKCα, the XO-generated ROS did not oxidize PKCα. Interestingly, d-α-tocopherol inhibited ICAM-1 activation of PKCα but not the upstream signal ERK1/2. The d-α-tocopherol inhibition of PKCα was ablated by the addition of d-γ-tocopherol. Conclusions Crosslinking ICAM-1 stimulated XO/ROS which activated ERK1/2 that then activated PKCα. ICAM-1 activation of PKCα was inhibited by d-α-tocopherol and this inhibition was ablated by the addition of d-γ-tocopherol. These tocopherols regulated ICAM-1 activation of PKCα without altering the upstream signal ERK1/2. Thus, we identified a mechanism for ICAM-1 activation of PKC and determined that d-α-tocopherol and d-γ-tocopherol have opposing regulatory functions for ICAM-1-activated PKCα in endothelial cells.


Introduction
Leukocytes bind to endothelial cell adhesion molecules such as intercellular adhesion molecule-1 (ICAM-1) and vascular adhesion molecule-1 (VCAM-1) during their migration across endothelial barriers at sites of inflammation. In allergic inflammation, recruitment of lymphocytes and eosinophils is dependent on binding to ICAM-1 and VCAM-1 as demonstrated by in vivo administration of blocking antibodies for these adhesion molecules or ICAM-1 knockout mice [1,2,3,4,5]. We have reported that, in allergic lung inflammation in mice, the recruitment of lymphocytes and eosinophils is inhibited by the vitamin E isoform d-atocopherol and elevated by the vitamin E isoform d-c-tocopherol [6]. These two isoforms of tocopherols differ by one methyl group [7,8]. The tocopherol isoform-specific regulation of leukocyte recruitment to the lung in allergic responses occurs without altering expression of several mediators of inflammation including cytokines, chemokines and VCAM-1 [6]. Interestingly, leukocyte migration across endothelium expressing VCAM-1 is regulated by d-a-tocopherol and d-c-tocopherol when endothelial cells are pretreated with tocopherols but not when leukocytes are pretreated with tocopherols [6]. Pretreatment of endothelial cells with d-a-tocopherol inhibits leukocyte transendothelial migration whereas pretreatment of endothelial cells with d-c-tocopherol elevates migration [6]. Furthermore, d-a-tocopherol inhibits and d-c-tocopherol elevates VCAM-1 signaling in endothelial cells. However, it is not known whether these tocopherol isoforms, which regulate ICAM-1-dependent leukocyte recruitment in vivo, also differentially regulate ICAM-1 signaling in endothelial cells.
We recently reported that purified recombinant PKCa directly binds a-tocopherol and c-tocopherol [28]. We also reported that a-tocopherol and c-tocopherol function as a PKCa antagonist and agonist, respectively, during cofactor-dependent activation of purified recombinant PKCa or during oxidative activation of purified recombinant PKCa [28]. Furthermore, we reported that d-a-tocopherol inhibits VCAM-1-dependent oxidative activation of PKCa in mouse endothelial cells [6]. Whether ICAM-1 activation of PKC is regulated by tocopherols is not known, the isoform of PKC in ICAM-1 signaling in endothelial cells is not known, and whether ICAM-1 induces oxidative activation of PKC is not known. Therefore, to determine whether tocopherols regulate ICAM-1 signaling, we first determined determined the isoform of PKC activated by ICAM-1, we determined the mechanism for ICAM-1 activation of PKC and determined whether ICAM-1 stimulates oxidative activation of PKC. Then, we addressed our hypothesis that d-a-tocopherol and d-ctocopherol regulate ICAM-1 activation of PKC in human microvascular endothelial cells.
In this study, we demonstrated that ICAM-1 activated the PKC isoform PKCa and that this activation was not by direct oxidation of PKCa, although it was dependent on ICAM-1-induced XOgenerated ROS in endothelial cells. ICAM-1-stimulated XO induced the activation of ERK1/2 that then activated PKCa. The ICAM-1-dependent activation of PKCa was inhibited by d-atocopherol but there was no effect of tocopherols on the upstream ERK1/2 activity. Furthermore, d-c-tocopherol ablated d-atocopherol's inhibition of ICAM-1-activated PKCa. This study demonstrates that d-a-tocopherol and d-c-tocopherol have opposing regulatory functions for ICAM-1-activated PKCa in endothelial cells.

Cytotoxicity Assay
The Vybrant Cell Metabolic Assay Kit (V-23110, Invitrogen, MA, USA) was used to detect the metabolic activity of the HMVECLs treated with d-a-tocopherol, d-c-tocopherol or pharmacological inhibitors at the doses indicated. In this assay, damaged release the cytosolic enzyme G6PD from damaged cells into the medium. G6PD reduces non-fluorescent resazurin (R-12204) to red-fluorescent resorufin. The resulting fluorescence is proportional to the amount of G6PD released into the medium, and this release correlates with the number of dead cells in the sample (Assay kit V-23111). Fluorescence was measured in a microplate reader (excitation/emission ,530/590 nm). The % relative cytotoxicity was calculated as follows: 100 x (fluorescence of the experimental cells-background of untreated control cells)/ (fluorescence of the fully lysed cells-background of the untreated control cells).

Tocopherol Loading of Endothelial Cells
80% confluent monolayers of HMVECLs were incubated with 10 ng/ml TNFa for 6 hr. Then, d-a-tocopherol, d-c-tocopherol, or the solvent control 0.01% dimethyl sulfoxide (DMSO) were added overnight. To determine the endothelial cell tocopherol levels, HMVECL cells were suspended and homogenized in absolute ethanol with 5% ascorbic acid on ice. Homogenates were extracted with an equal volume of 0.1% butylated hydroxytoluene in hexane to prevent oxidation and increase recovery of tocopherol. The samples were vortexed and then centrifuged for 5 minutes at 2000 rpm at room temperature. The hexane layer was removed to a separate vial and the hexane extraction step was repeated two more times for a total of three hexane extractions per sample. The hexane layer was dried under nitrogen and stored at 220uC. The samples were reconstituted in methanol and then tocopherols were separated using a reverse phase C18 HPLC column and HPLC (Waters Company, Milford, MA) with 99% methanol-1% water and 1% lithium perchlorate as a mobile phase. Tocol was used as internal standard. Tocopherols were detected with an electrochemical detector (ECD) (potential 0.7V) (Waters Company).

Antibody-coated Beads
Streptavidin-coated 9.9 mm diameter beads (80 ml) (Bangs Laboratories) were labeled with 24 mg of biotin-conjugated goat anti-mouse Ig in 375 ml of PBS with gentle rocking for 1 hr at 4uC and then washed three times [29]. These beads were incubated with 16 mg of rat anti-human ICAM-1 or rat anti-human PECAM-1 in 375 ml of PBS with gentle rocking for 1 hr at 4uC and then washed.

Western Blotting
Cell lysates were analyzed by 10% SDS-PAGE electrophoresis and transferred to nitrocellulose membranes using the semi-dry method according to manufacturer's instructions (Bio-Rad). The membranes were blocked in 5% non-fat dried milk in Trisbuffered saline plus 0.1% Tween 20 (TBS-T) for 1 hour at room temperature and washed 3 times for 5 minutes in TBS-T. Membranes were incubated with primary antibodies in TBS-T plus 5% milk overnight, washed 3 times for 5 minutes in TBS-T, incubated with secondary antibodies in TBS-T plus 5% milk for 1 hour, washed 3 times for 10 minutes in TBS-T, and examined for detection with enhanced chemiluminescence (Amersham) and autoradiography. Densitometry was performed using Image J software (NIH). The data were presented as the fold increase in the ratio of relative intensity of the band/the relative intensity of band for the loading control (total PKCa, total PKCbII or total ERK1/2).

Assessment of Cysteine Oxidation
Cysteine oxidation was assessed using N-(biotinoyl)-N'-(iodoacetyl)ethylenediamine (BIAM) according to the protocol previously described [30]. This is a sensitive method that detects proteins that contain H 2 O 2 -sensitive Cys residues [30]. Endothelial cells were stimulated with anti-ICAM-1-coated beads or anti-PECAM-1coated beads. These cells were lysed in a buffer containing 50 mM 4-morpholine ethane sulfonic acid (MES), 100 mM NaCl, 50 mM phenylmethylsulfonyl fluoride, 1.0 mg/ml leupeptin, 1.0 mg/ml aprotinin, and 0.5% Triton X-100, 100 mM BIAM (pH 6.5) for 30 min [30]. The buffer was made fresh, rendered free of O 2 by bubbling with N 2 at a low flow rate for 1 hr. The positive control include lysates oxidized with 200 mM H 2 O 2 for 20 min before addition of BIAM. The BIAM reaction was terminated by addition of 20 mM b-mercaptoethanol. PKCa was immunoprecipitated and separated by SDS-PAGE. Proteins labeled with BIAM were detected with HRP-conjugated streptavidin and ECL. BIAM reacts only with nonoxidized cysteines. Thus, loss of BIAM labeling indicates oxidation.

Statistics
Data were analyzed by a one way ANOVA followed by Tukey's multiple comparisons test (SigmaStat, Jandel Scientific, San Ramon, CA).

ICAM-1 Crosslinking Activates PKCa but not in PKCbII in HMVECLs
It is reported that ICAM-1 activates endothelial cell PKC [24] but the PKC isoform is not known. Therefore, we determined whether ICAM-1 activates PKCa or PKCb. For these studies, primary cultures of human microvascular endothelial cells from the lung (HMVECLs) were stimulated overnight with 10 ng/ml TNFa to induce ICAM-1 expression. ICAM-1 was expressed as determined by immunofluorescence labeling and flow cytometry ( Figure 1A). Then, the TNFa-pretreated HMVECLs were activated with anti-human ICAM-1-coated beads for 10-60 min and examined for autophosphorylation at PKCaThr 638 . Phosphorylated PKCaThr 638 is the active form of PKCa and is required for enzyme activity [31,32]. At 20 min, ICAM-1 crosslinking induced an increase in phosphorylation of PKCaThr 638 (Fig. 1C). In contrast to ICAM-1 activation of PKCa, PKCbII was not activated by anti-ICAM-1-coated beads as indicated by no increase in phosphorylation of PKCbII Thr 641 (Fig. 1B). Anti-PECAM-1-coated control beads did not activate phosphorylation of PKCaThr 638 or PKCbII Thr 641 (Fig. 1B,D). These data indicate that ICAM-1 activates the PKCa isoform.

ICAM-1 Activation of PKCa Phosphorylation in HMVECLs is Mediated by a Xanthine Oxidase Pathway
In endothelial cells, ICAM-1 ligation activates xanthine oxidase (but not nitric oxide synthase) for the production of ROS [18,19,21,33,34]. ICAM-1 binding also induces a calcium/ PLCc 1 /PKC pathway for the activation of Src phosphorylation of cytoskeletal proteins during leukocyte diapedesis [24]. It is also reported that, in other signaling pathways, PI3 kinase can function in the activation of PKCa [35,36,37]. However, whether XO, PLC/PKC, Src, and PI3 kinase function in a single pathway during ICAM-1 activation of PKCa is not known. Therefore, we first determined whether XO, PLC, ERK, PI3 kinase or Src function in ICAM-1 activation of PKCa. For these studies, TNFastimulated HMVECLs were treated for 1 hr with inhibitors of XO, PLC, ERK, PI3 kinase, Src, or the solvent control 0.01% DMSO and then examined for ICAM-1 activation of PKCa. Inhibitor treatment of the HMVECLs was not cytotoxic as compared to the positive control 200 mM H 2 O 2 ( Figure 2D) and the inhibitors were used at doses that we and others have reported for studies with endothelial cells [29,38,39,40,41,42,43,44]. The solvent control 0.01% DMSO did not affect ICAM-1 activation of PKCa ( Figure 2A). The PKCa inhibitor Go6876, which was used as a control for inhibition of PKCa autophosphorylation, inhibited ICAM-1-stimulated phosphorylation of PKCaThr 638 (Figure 2A). The Src kinase inhibitor PP2 (10 mM) and the PI3 kinase inhibitor Ly294002 (10 mM) did not block ICAM-1induced activation of PKCa Thr 638 (Fig. 2B). In contrast, ICAM-1-induced activation of PKCa Thr 638 was blocked by the PLC inhibitor U73122 (10 mM), ERK1/2 inhibitor PD98059 (10 mM), and xanthine oxidase inhibitor allopurinol (0.3 mg/ml) ( Fig. 2A), suggesting that PLC, ERK1/2 and XO-generated ROS are required for ICAM-1 activation of PKCa.
PKCa is not Oxidized during its Activation by ICAM-1 XO-generated ROS were required for ICAM-1 activation of PKCa in endothelial cells (Figure 2A), but it is not known whether PKCa is activated by oxidation. Oxidative activation of PKCa has been reported for VCAM-1 signaling through NADPH oxidase-generated ROS [44]. Therefore, we determined whether ICAM-1 stimulates oxidation of PKCa and whether NADPH oxidase has a role in ICAM-1 activation of PKCa. Treatment of TNFa-stimulated HMVECLs for 1 hr with the NADPH oxidase inhibitor apocynin (4 mM) did not block ICAM-1 activation of PKCa ( Figure 2B). To examine oxidation of PKCa, TNFastimulated HMVECLs were stimulated with anti-ICAM-1 and then lysed in the presence of BIAM which binds to non-oxidized cysteines. ICAM-1 was immunoprecipitated and BIAM labeling was detected with HRP-conjugated streptavidin in a western blot. Loss of BIAM labeling indicates cysteine oxidation of PKCa. Anti-ICAM-1-coated beads did not induce oxidation of PKCa cysteines, compared to the positive control, lysates treated with 200 mM H 2 O 2 (Fig. 2C). Therefore, the ROS that are generated by xanthine oxidase during ICAM-1 activation of PKCa do not directly oxidize and activate PKCa.

Treatment of HMVECLs with Isoforms of the Antioxidant Vitamin E
We previously reported a) that ICAM-1-dependent and VCAM-1-dependent recruitment of leukocytes in vivo is regulated by tocopherols, b) that in microvascular endothelial cells, d-atocopherol inhibits VCAM-1 oxidative activation of PKCa, and c) that d-c-tocopherol ablates this inhibition by d-a-tocopherol [6]. This tocopherol regulation of VCAM-1 signaling occurred without altering VCAM-1 expression [6]. We have also reported that these tocopherol isoforms can directly bind to and regulate purified recombinant PKCa activity [28]. However, it is not known whether tocopherols regulate ICAM-1 activation of PKC. To examine tocopherol function during ICAM-1 signaling, HMVECLs were pretreated with TNFa for 6 hours to induce ICAM-1 expression and then the cells were incubated overnight with d-a-tocopherol or d-ctocopherol to generate tocopherol levels in the HMVECLs equivalent to the levels of tocopherol in lung tissue. Tocopherol levels were determined by HPLC with ECD [6]. Treatment with 40-80 mM d-a-tocopherol or 1-4 mM d-c-tocopherol was not toxic to the cells as determined by the Vybrant Cell Metabolic Assay ( Figure 4A). The tocopherols did not affect ICAM-1 expression by the HMVECL's as determined by immunolabeling and flow cytometry ( Figure 4B,C). Treatment of HMVECLs with 60 mM a-tocopherol or 2 mM c-tocopherol resulted in 10 mg a-tocopherol/ g cells and 2.8 mg c-tocopherol/g cells ( Table 1). This is consistent with reports that human and mouse lung tissue levels of a-tocopherol are 9-10 mg/g of tissue [45] and that c-tocopherol tissue levels are 5-10 fold lower than a-tocopherol [6]. In vivo, c-tocopherol tissue levels are lower because of the preferential transfer in the liver by atocopherol transfer protein [46]. In human plasma, the normal range of d-a-tocopherol is 20-30 mM, whereas 60 mM d-atocopherol is achieved in plasma by oral supplementation with 200-800 IU of d-a-tocopherol per day [47]. Therefore, HMVECLs were loaded in vitro with tocopherols at the levels of tocopherols reported for lung tissue.

D-a-tocopherol Inhibits ICAM-1-stimulated Activation of PKCa but not ERK1/2 in HMVECLs; the Inhibition by d-atocopherol is Abrogated by d-c-tocopherol
We determined whether ICAM-1 activation of PKCa was inhibited by preloading tocopherols in HMVECLs as in TABLE 1. Treatment with the tocopherols did not alter total PKCa expression ( Figure 5A-D). ICAM-1 activation of PKCa Thr 638 phosphorylation was inhibited by d-a-tocopherol ( Figure 5A) but not by d-c-tocopherol ( Figure 5B). Interestingly, d-c-tocopherol ablated d-a-tocopherol's inhibition of ICAM-1-activated PKCa ( Figure 5C). Basal levels of PKCa Thr 638 phosphorylation were not affected by d-a-tocopherol ( Figure 5D) or d-c-tocopherol (data not shown). In contrast to the tocopherol regulation of ICAM-1 activation of PKCa, these tocopherols did not affect ICAM-1 activation of ERK1/2 ( Figure 5E,F). These data suggest that during ICAM-1 signaling, the tocopherols do not function to block xanthine oxidase-mediated activation of ERK1/2. The tocopherols function, at least, downstream of ERK1/2 to regulate PKCa activity.

Discussion
In this report, we identify a mechanism for ICAM-1 activation of PKC in endothelial cells and determined that this activation of PKC is regulated by tocopherols. We determined that stimulation of ICAM-1 activated PKCa but not PKCb. Antibody crosslinking of ICAM-1 activated XO, PLC, and ERK1/2 which then activated PKCa. ICAM-1 activation of PKCa was blocked by the PLC inhibitor U73122, ERK1/2 inhibitor PD98059, and xanthine oxidase inhibitor allopurinol. ERK1/2 activation was blocked by inhibition of XO and PLC but not by inhibition of PKCa, indicating that ERK1/2 is downstream of XO and upstream of PKCa during ICAM-1 signaling. ICAM-1 activation of PKCa was not blocked by inhibitors of Src or PI3 kinase. During ICAM-1 activation of PKCa, the XO-generated ROS did not oxidize PKCa. Instead, crosslinking ICAM-1 stimulated XO/ ROS which activated ERK1/2 that then activated PKCa. The ICAM-1 activation of PKCa was inhibited by a-tocopherol treatment of the endothelial cells and this inhibition by atocopherol was abrogated by c-tocopherol. In contrast, the tocopherols did not alter the ICAM-1 activation of PKCa's upstream signal, ERK1/2. This suggests that since ERK1/2 activation was not affected by tocopherols but ERK1/2 activation was dependent on XO, the antioxidant properties of the tocopherols did not function to block the XO-induced ERK1/2 that leads to activation of PKCa.
Separate reports indicate that ICAM-1 can activate PKC, XO, PLC or ERK1/2 [20,26,27], but the mechanisms for activation of these signals in endothelial cells was not known. In brain endothelial cell lines, ICAM-1 binding induces a calcium/ PLCc 1 /PKC pathway [24]. They also report that pharmacological inhibition of PKC blocks transendothelial migration [24]. Chelating calcium or inhibitors of PKC or Src block leukocyte migration across endothelial cells [12,24,25]. However, the PKC isoform activated by ICAM-1 signaling was not known. Furthermore, the mechanism for ICAM-1 activation of PKC was not known. We report here that crosslinking ICAM-1 on human microvascular lung endothelial cells activates endothelial cell PKCa but not PKCb. In other signaling pathways, PKC can function upstream of ROS production in that PKC can activate NADPH oxidase for the release of ROS in neutrophils and monocytes [48,49]. During VCAM-1 signaling, NADPH oxidasegenerated ROS directly oxidize and activate PKCa followed by downstream activation of ERK1/2 [50]. In contrast to VCAM-1induced oxidation of PKCa [50], we demonstrated that ICAM-1 signals do not induce oxidation of PKCa. Instead, ICAM-1stimulated XO/ROS induce activation of ERK1/2 which then activates PKCa. Therefore, the mechanism for ICAM-1 activation of PKCa is different than the mechanism for VCAM-1 activation of PKCa.
We have previously reported that the in vivo ICAM-1-dependent and VCAM-1-dependent recruitment of lymphocytes and eosinophils in allergic lung inflammation is inhibited by d-a-tocopherol and elevated by d-c-tocopherol [6]. Furthermore, VCAM-1 induces oxidative activation of endothelial cell PKCa which is blocked by treatment with d-a-tocopherol [44]. This d-atocopherol inhibition of VCAM-1-stimulated oxidative activation of PKCa is ablated by d-c-tocopherol [44]. Although the mechanism for VCAM-1 activation of PKCa is through oxidation of PKCa whereas we report here that ICAM-1 activation of PKCa is not through oxidation of PKCa, we determined that treatment of endothelial cells with d-a-tocopherol inhibits ICAM-1 activation of PKCa and d-c-tocopherol ablates the inhibition by da-tocopherol.
We have recently reported that d-a-tocopherol and d-ctocopherol can directly bind recombinant PKCa and regulate recombinant PKCa activity [28]. D-a-tocopherol and d-ctocopherol bind to the regulatory C1A domain of recombinant PKCa. Recombinant PKCa activity is decreased by d-atocopherol and increased by d-c-tocopherol [28]. Furthermore, d-a-tocopherol inhibition of recombinant PKCa is blocked by d-ctocopherol [28]. However, these studies were with recombinant PKCa. It was not known whether tocopherols regulate ICAM-1 activation of PKC in endothelial cells. In the present study, ICAM-1-stimulated PKCa was inhibited by d-a-tocopherol and this inhibition by d-a-tocopherol was blocked by d-c-tocopherol. This tocopherol regulation of PKCa occurred without tocopherol regulation of the upstream XO/ROS activation of ERK1/2. Thus, the tocopherol regulation of ICAM-1 activation of PKCa in There are conflicting reports with regards to whether tocopherols alter adhesion molecule expression. It is reported that in vitro a-tocopherol inhibits cytokine or oxidized LDL-induced expression of ICAM-1, VCAM-1, or E-selectin [51,52,53,54], but whether in these studies the in vitro doses of tocopherols in cells were at concentrations of tocopherols found in tissues is not known. In contrast, in vivo d-a-tocopherol and d-c-tocopherol do not alter VCAM-1 expression and, when tocopherols are loaded in endothelial cells in vitro at concentrations found in vivo, tocopherols do not alter VCAM-1 expression or lymphocyte binding to endothelial cells in vitro [6]. In our current report, HMVECL's were stimulated with TNFa for 6 hours to induce ICAM-1 expression and then supplemented overnight with doses of tocopherols to generate cellular concentrations of tocopherols equivalent to that reported for lung tissue levels of tocopherols [6,45]; there were no effects of tocopherols on ICAM-1 expression by the HMVECL's in vitro. Therefore, without altering ICAM-1 expression, tocopherols regulated ICAM-1 activation of PKCa.
In reports of receptors other than ICAM-1, activation of PKC is inhibited by d-a-tocopherol treatment of muscle cells [55,56,57,58], monocytes [59,60], epithelial cells [61], endothelial cells [6,48,59,62], and platelets [63]; in these studies, regulation by c-tocopherol was not examined. In addition, although these reports indicate that a-tocopherol inhibits activation of PKC in cells, the mechanisms for the a-tocopherol inhibition of PKC activation are often not described. It is reported that in smooth muscle cells, d-a-tocopherol inhibits PKCa without inhibition of PKC expression [64]. Consistent with this, we report that d-atocopherol and d-c-tocopherol did not alter total PKCa expression in endothelial cells. It is also reported that d-a-tocopherol activates smooth muscle PP2 and that purified PP2 can dephosphorylate both PKC and ERK1/2 [65,66]. In our studies, d-a-tocopherol regulated PKCa without altering ERK1/2 activation during ICAM-1 signaling.