The effect of anandamide on endothelial Ca²⁺ signaling inversely depended on external Ca²⁺ and was due to atypical endothelial anandamide receptors

In the presence of 2 mM external Ca²⁺, addition of 10 μM anandamide yielded very small cytosolic Ca²⁺ elevation in only 10% of the cells, which was far less than that obtained with 100 μM histamine under identical conditions (Fig. 1A). Similarly, the hydrolytically more stable analog methanandamide failed to initiate Ca²⁺ elevation in the presence of external Ca²⁺, thus excluding Ca²⁺-dependent hydrolysis as being responsible for the lack of Ca²⁺ signaling in response to anandamide in the presence of external Ca²⁺.

However, in the nominal absence of extracellular Ca²⁺, both agonists increased cytosolic Ca²⁺ levels in endothelial cells (Fig. 1B) and, similar to histamine, over 95% of the cells responded to the endocannabinoid while there was no effect upon stimulation with 10 μM arachidonic acid (supplementary material Fig. S1). Moreover, consequent to a strong endoplasmic reticulum (ER) depletion by either anandamide or histamine (Fig. 1C), capacitative Ca²⁺ entry was detected upon re-addition of extracellular Ca²⁺ (Fig. 1B). The EC₅₀ for anandamide to initiate Ca²⁺ signaling in nominally Ca²⁺-free solution was 7.3 (4.5-11.7) μM (n=10), which was comparable with that of the metabolically stable analog methanandamide [EC₅₀=10.6 (9.9-11.3) μM, n=10] (Fig. 1D), thus indicating that anandamide-induced Ca²⁺ signaling is directly evoked by the compound itself and not by a metabolite.

To define the receptor being involved in anandamide-triggered endothelial Ca²⁺ signaling, the expression of CB₁R but not CB₂R was confirmed by RT-PCR in the human endothelial cells used in this study (Fig. 2A). Involvement of CB₂R was further excluded by the lack of the CB₂R antagonist SR144528 [1 μM (Rinaldi-Carmona et al., 1998)] (Table 1) to inhibit anandamide-induced Ca²⁺ signaling in the nominal absence of extracellular Ca²⁺ (supplementary material Fig. S2A). Similarly, inhibition of TRPV1 (VR1) receptors by SB366791 [4′-chloro-3-methoxycinnamanilide (Fowler et al., 2003)] (10 μM) failed to affect anandamide-induced Ca²⁺ signaling under the same conditions (supplementary material Fig. S2B). By contrast, rimonabant (1 μM), an inhibitor of CB₁R and e-aR (Jarai et al., 1999; Wagner et al., 1999), inhibited anandamide-induced Ca²⁺ signaling in nominally Ca²⁺-free solution (Fig. 2B).

To further differentiate whether CB₁R or e-aR is involved in anandamide-induced Ca²⁺ signaling, the effects of specific receptor agonists and antagonists were tested. Inhibition of CB₁R by 10 μM of the potent CB₁R antagonist AM251 [N-(Piperidin-1-yl)-5-(4-iodophenyl)-1-(2,4-dichlorophenyl)-4-methyl-1H-pyrazole-3-carboxamide] (Table 1) (Pertwee, 2005) unmasked anandamide (10 μM)-induced Ca²⁺ signaling even in the presence of extracellular Ca²⁺ (Fig. 2C). In addition, the synthetic CB₁R agonist HU-210 [10 μM (Felder et al., 1995)], which has been shown to couple with intracellular Ca²⁺ signaling in HEK 293 cells, cultured hippocampal neurons (Lauckner et al., 2005) and rat insulinoma beta cells (De Petrocellis et al., 2007), failed to initiate endothelial Ca²⁺ signaling in the presence and nominal absence of external Ca²⁺ (supplementary material Fig. S2C). By contract, 10 μM of O1602, an agonist of the e-aR (Table 1) (Bukoski et al., 2002; Jarai et al., 1999) evoked a very strong cytosolic Ca²⁺ elevation even in the presence of extracellular Ca²⁺ (Fig. 2D), which was not affected by the TRPV1 receptor inhibitor SB366791 (10 μM) (supplementary material Fig. S2D). In line with these findings, 10 μM of O1918 [1,3-dimethoxy-5-methyl-2-[(1R,6R)-3-methyl-6-(1-methylethenyl)-2-cyclohexen-1-yl]-benzene] (Table 1), a cannabidiol analog that acts as a selective antagonist of `atypical' cannabidiol at the non-CB₁/CB₂ endothelial anandamide receptor (Jarai et al., 1999; Mo et al., 2004; Offertaler et al., 2003), prevented anandamide-induced Ca²⁺ signaling in the nominal absence of extracellular Ca²⁺ (Fig. 2E). Moreover, in the presence of the CB₁R-specific agonist HU-210 (10 μM), activation of the e-aR with 10 μM O1602 failed to evoke Ca²⁺ signaling, while, after washout of HU-210, O1602 initiated Ca²⁺ signaling in the same cells (Fig. 2F). These data point to involvement of both anandamide receptors in endothelial cells in which CB₁R inhibits the signal cascade beyond e-aR, which yields intracellular Ca²⁺ mobilization and becomes overt in the nominal absence of external Ca²⁺.

• Download figure
• Open in new tab
• Download powerpoint

Fig. 1.

Endocannabinoids trigger Ca²⁺ signaling and ER depletion in human endothelial cells. (A) Comparison of the effects of 100 μM histamine (n=18) and 10 μM anandamide (n=31) on cytosolic free-Ca²⁺ concentration in the presence of 2 mM extracellular Ca²⁺. (B) Cytoplasmic free-Ca²⁺ concentration upon stimulation with 100 μM histamine (n=19) and 10 μM anandamide (n=21) in the absence of extracellular Ca²⁺, followed by re-addition of 2 mM extracellular Ca²⁺ to reveal store-operated Ca²⁺ entry due to previous ER depletion. (C) Comparison of the effects of 100 μM histamine (n=4) and 10 μM anandamide (n=3) on free ER Ca²⁺ concentration in the absence of extracellular Ca²⁺, followed by re-addition of 2 mM extracellular Ca²⁺ to visualize ER Ca²⁺ refilling efficiency. (D) Concentration-response curves for anandamide (n=10) and metanandamide (n=10) on peak responses of cytosolic Ca²⁺ elevation in the nominal absence of extracellular Ca²⁺. Cytosolic (A,B,D) and ER (C) Ca²⁺ concentrations were recorded using fura-2 and YC4er, respectively. ^*P<0.01 versus histamine.

The `atypical' non-CB₁ and CB₂ endothelial anandamide receptor (e-aR) was identified as G-protein-coupled receptor 55 (GPR55)

To identify the e-aR molecularly, the endothelial cells used for this study were tested for GPR55, a putative anandamide-binding protein (Baker et al., 2006; Ryberg et al., 2007). Using the respective primers, mRNA of GPR55 was found in endothelial cells (Fig. 2G). Hence, treatment with siRNA against GPR55, which was proved to reduce the expression of GPR55 by ∼56% (supplementary material Fig. S2E), strongly diminished anandamide-triggered Ca²⁺ signaling in Ca²⁺-free solution, as well as the effect of O1602 in the presence of extracellular Ca²⁺ (Fig. 2H,I, respectively), thus pointing to the involvement of GPR55 in anandamide-induced Ca²⁺ signaling in endothelial cells. Overexpression of GPR55 (supplementary material Fig. S2F) strongly elevated anandamide-triggered Ca²⁺ signaling in the nominal absence of extracellular Ca²⁺ (Fig. 2J). Consistent with this, the effect of O1602 (10 μM) on endothelial cytosolic free-Ca²⁺ was largely elevated in GPR55-overexpressing cells (Fig. 2K, Table 1).

Recently, GPR55 was described as a receptor of lysophosphatidylinositol [LPI (Oka et al., 2007)]. These data could be confirmed in endothelial cells in which the effect of 10 μM LPI was strongly reduced in cells treated with siRNA against GPR55 (supplementary material Fig. S2G), whereas overexpression of GPR55 yield enhanced Ca²⁺ signaling by LPI (supplementary material Fig. S2H).

The effect of anandamide on endothelial Ca²⁺ signaling was sensitive to divalent cations acting on extracellular bindings sites

Based on the unusual sensitivity of anandamide-triggered Ca²⁺ signaling to external Ca²⁺, we tested whether or not the inhibitory effect of extracellular Ca²⁺ was due to Ca²⁺-dependent chelating of anandamide in the medium. Using NMR spectroscopy, no evidence for a Ca²⁺-anadamide complex was found (supplementary material Fig. S3A,B). Moreover, anandamide uptake and binding did not differ in the nominal absence or presence of extracellular Ca²⁺ (supplementary material Fig. S3C). The concentration-response curve of the inhibitory action of extracellular Ca²⁺ on anandamide-triggered intracellular Ca²⁺ signaling revealed an IC₅₀ of 7.7 (6.7-8.8) μM (Fig. 3A). Similarly, Sr²⁺ and Ba²⁺ inhibited anandamide-triggered Ca²⁺ signaling in endothelial cells in a concentration-dependent manner (Fig. 3B). These results point to an extracellular Ca²⁺-binding site being responsible for its inhibitory effect on anandamide-induced Ca²⁺ signaling.

Clustered αvβ3 and α5β1 integrins were involved in anandamide-induced Ca²⁺ signaling in human endothelial cells

Remarkably, the IC₅₀ of extracellular Ca²⁺ to prevent anandamide-induced Ca²⁺ signaling was strikingly similar to that found for Ca²⁺-mediated inhibition of integrin clustering (Leitinger et al., 2000). Therefore, involvement of clustered integrins in the transition of anandamide to evoke Ca²⁺ signaling was postulated. To challenge this hypothesis, clustering of integrins was initiated by Mn²⁺, which increases the affinity of integrins for their ligand and can stimulate integrin clustering (Mould et al., 1998; Smith and Cheresh, 1990), prior to measurement of the effect of anandamide on endothelial Ca²⁺ concentration. Notably, in the presence of 70 μM Mn²⁺, anandamide yielded strong Ca²⁺ elevation even in the presence of extracellular Ca²⁺ (Fig. 3C). In line with these results, inhibition of RhoA-associated kinase 1 and RhoA-associated kinase 2 (ROCK1 and ROCK2) that are essentially involved in the clustering of integrins (Rodriguez-Fernandez et al., 2001), by Y27632 [(R)-(+)-trans-N-(4-Pyridyl)-4-(1-aminoethyl)-cyclohexanecarboxamide] [10 μM (Uehata et al., 1997)] strongly diminished anandamide-induced Ca²⁺ signaling in the absence of extracellular Ca²⁺ (Fig. 3D). Involvement of integrins was further confirmed by the inhibitory activities of fibronectin (20 μg/ml, supplementary material Fig. S3D), a large extracellular matrix protein ligand for endothelial integrins, and RGD, Arg-Gly-Asp peptide resembling the integrin-binding motif of fibronectin (Ruoslahti and Pierschbacher, 1987) (20 μg/ml; supplementary material Fig. S3E).

To further challenge the concept that integrins are involved in the process of anandamide-triggered Ca²⁺ signaling, the effects of the crosslinking antibody against αvβ3 integrin [LM609, 2 μM (Charo et al., 1990; Cheresh, 1987)] and two function-blocking antibodies against the β1 [JB1A, 2 μM (Mohri et al., 1996)] and the β3 subunits [B3A, 2 μM (Vignali et al., 1990)]. During preincubation for 15 minutes in Ca²⁺-containing medium, none of the integrin-targeting antibodies affected basal endothelial Ca²⁺ levels, while all prevented the effect of anandamide in the absence of extracellular Ca²⁺ (Fig. 3E-G). In line with these findings, B3A also prevented O1602-induced Ca²⁺ signaling in Ca²⁺-containing buffer (Fig. 3H). These data point to the involvement of αvβ3 and α5β1 integrins in GPR55-mediated Ca²⁺ signaling by anandamide and O1602 in endothelial cells.

• Download figure
• Open in new tab
• Download powerpoint

Fig. 2.

`Atypical' endothelial anandamide receptors are identified as GPR55 that mediates anandamide-initiated Ca²⁺ signaling, while activation of CB₁ receptors prevent anandamide-induced Ca²⁺ signaling. (A) Representative RT-PCR on CB₁R in a human endothelial cell line (lane 2) in comparison with human brain as positive control (lane 3). (B) In Ca²⁺-free solution, the effect of 10 μM anandamide on cytosolic free-Ca²⁺ concentration was analyzed in the absence (n=19) or presence of the CB₁R antagonist rimonabant (1 μM) (n=28). (C) In the presence of the CB₁R antagonist AM251 (10 μM), 10 μM anandamide triggered Ca²⁺ signaling even in the presence of extracellular Ca²⁺ (n=59). (D) In the presence of 2 mM extracellular Ca²⁺, the effect of the e-aR agonist O1602 (10 μM) on cytosolic free-Ca²⁺ concentration was tested (n=17). (E) In the absence of extracellular Ca²⁺, the effect of O1918 (10 μM), an inhibitor of the e-aR, on anandamide (10 μM, AEA) -induced Ca²⁺ signaling was assessed (n=8). (F) The effect of the CB₁R agonist HU-210 (10 μM) on O1602 (e-aR agonist, 10 μM) -induced Ca²⁺ signaling was tested in the presence of 2 mM extracellular Ca²⁺ (n=10). (G) Representative RT-PCR on GPR55 in the human endothelial cell line using various annealing temperatures. (H) Pretreatment with siRNA against GPR55 (n=49; control, n=48) on anandamide (10 μM) -induced cytosolic Ca²⁺ signaling was monitored in nominally Ca²⁺-free solution. (I) Pretreatment with siRNA against GPR55 (n=62; control, n=18) on O1602 (10 μM) -induced cytosolic Ca²⁺ signaling in Ca²⁺-containing solution. (J) Consequences of GPR55 overexpression on anandamide (AEA, 10 μM) -induced Ca²⁺ signaling in the nominal absence of extracellular Ca²⁺ (n=43; control, n=42). (K) Consequences of GPR55 overexpression on O1602 (10 μM) -induced Ca²⁺ signaling in the nominal absence of extracellular Ca²⁺ (n=21; control, n=25). Cytosolic free-Ca²⁺ concentrations were recorded using fura-2. ^*P<0.005 versus the absence of the respective receptor blocker/activator.

Anandamide induced diverse tyrosine phosphorylation patterns depending on the presence of extracellular Ca²⁺

As tyrosine kinases are known to represent early downstream targets of integrin-mediated signaling, the effect of anandamide on the activity of endothelial tyrosine kinases [using the FRET-based sensor Picchu 936X (Kurokawa et al., 2001; Schaeffer et al., 2003)] and the patterns of tyrosine phosphorylation were assessed in the presence and absence of extracellular Ca²⁺. Notably, anandamide yielded fast activation of PP1-sensitive tyrosine kinases under any conditions indicated by increase in the FRET signal of Picchu 936X (Fig. 4A,B) but not the inactive mutant Picchu 938X. Although this is a rather general sensor for several tyrosine kinases and, thus, no conclusion can be drawn on the actual kinase(s) involved, it uniquely reveals the kinetics of tyrosine phosphorylation upon anandamide. However, corresponding western blot analyses revealed markedly different effects of anandamide on protein tyrosine phosphorylation, depending on extracellular Ca²⁺. In the presence of extracellular Ca²⁺ a fast, strong and long lasting phosphorylation of a ∼60 kDa band occurred in response to anandamide stimulation, while phosphorylation of a 70-80 kDa band reached its maximum 5 minutes after agonist stimulation (supplementary material Fig. S4A). By contrast, in the absence of external Ca²⁺ (i.e. under conditions of anandamide-induced Ca²⁺ signaling) the 70-80 kDa band appeared much earlier, while the ∼60 kDa band was delayed compared with the presence of extracellular Ca²⁺ (supplementary material Fig. S4A). Identical findings were obtained in the presence of 70 μM Mn²⁺ in Ca²⁺-containing buffer.

Tyrosine phosphorylation upon anandamide in presence and absence of extracellular Ca²⁺ was prevented by 10 μM PP1 [1-(1,1-Dimethylethyl)-1-(4-methylphenyl)-1H-pyrazolo[3,4-d]pyrimidin-4-amine; a potent inhibitor of Src-family tyrosine kinases (Hanke et al., 1996)] (supplementary material Fig. S4B), which also abolished anandamide-triggered Ca²⁺ signaling in Ca²⁺-free solution (Fig. 4C), pointing to the activation of Src-family kinase(s) under all conditions. This assumption was further supported by the inhibitory effect of 10 μM PP2 [3-(4-chlorophenyl) 1-(1,1-dimethylethyl)-1H-pyrazolo[3,4-d]pyrimidin-4-amine; a potent inhibitor of Src-family tyrosine kinases (Hanke et al., 1996)] on O1602-induced Ca²⁺ signaling in Ca²⁺-containing solution (Fig. 4D, Table 2).

• Download figure
• Open in new tab
• Download powerpoint

Fig. 3.

Anandamide/GPR55-induced Ca²⁺ signaling is linked to integrins. (A,B) The sensitivity of anandamide (10 μM) -triggered cytosolic Ca²⁺ signaling to various concentrations of extracellular Ca²⁺ (n=10), Ba²⁺ (n=24) and Sr²⁺ (n=31) was tested. (C) Comparison between the effect of 10 μM anandamide in Ca²⁺-containing buffer in the absence (n=20) or presence of 70 μM Mn²⁺ (n=40). (D) The effect of 10 μM anandamide on cytosolic free-Ca²⁺ concentration was monitored in the nominal Ca²⁺-free solution in the absence (n=25) or presence of the ROCK inhibitor Y27632 (10 μM, n=29). (E-G) The effects of antibodies (2 μM) against αvβ3 integrin (LM609) (control, n=34; LM609, n=30) and the β1 (JB1A) (control, n=28; JB1A, n=33) and β3 subunits (B3A) (control, n=38; B3A, n=30) on cytosolic Ca²⁺ signaling induced by 10 μM anandamide was assessed in nominally Ca²⁺-free solution. (H) The effect of function-blocking β3-integrin antibody B3A on O1602 (10 μM) -initiated Ca²⁺ signaling in the presence of extracellular Ca²⁺ (control, n=65; B3A, n=84). Cytosolic free-Ca²⁺ concentrations were recorded using fura-2. ^*P<0.0001 versus the absence of the respective compound.

• Download figure
• Open in new tab
• Download powerpoint

Fig. 4.

Anandamide initiates tyrosine phosphorylation. (A,B) The effect of anandamide (10 μM) -induced activation of tyrosine kinases was visualized in single endothelial cells in the presence (A) and absence (B) of extracellular Ca²⁺. As indicated, 10 μM PP1 was added to the media (n=4 under each condition). Tyrosine kinase activity was monitored by measuring the excitation ratio of F₅₃₅/F₄₈₀ nm (FRET) at 440 nm excitation. (C) The effect of 10 μM anandamide on cytosolic free-Ca²⁺ concentration was tested in nominally Ca²⁺-free solution in the absence (n=24) or presence of 10 μM PP1 (n=25). (D) In Ca²⁺-containing solution, Ca²⁺ intracellular Ca²⁺ signaling was initiated by 10 μM O1602 in the presence of 10 μM PP2, followed by an additional stimulation with 10 μM O1602 after washout of PP2 as indicated (n=14). For visualization of tyrosine kinase activity in single endothelial cells, cells were transiently transfected with Picchu 936X. Cytosolic free-Ca²⁺ concentration was recorded using fura-2. ^*P<0.001 versus the absence of PP1.

PI3K, Bmx/Etk and PLC represented downstream targets of integrin-dependent anandamide/GPR55-induced signal transduction in endothelial cells

At 0.1 μM, the selective inhibitor of phosphoinositide 3 kinase (PI3K) wortmannin (Table 2) diminished tyrosine phosphorylation of the 70-80 kDa band (supplementary material Fig. S5) and abolished anandamide-triggered Ca²⁺ signaling in nominally Ca²⁺-free solution (Fig. 5A) and in the presence of 70 μM Mn²⁺ in Ca²⁺ containing buffer (data not shown). Notably, wortmannin was less active in reducing the 70-80 kDa band in the presence of extracellular Ca²⁺, suggesting two different proteins between 70-80 kDa: a wortmannin-sensitive one becoming phosphorylated in the absence of extracellular Ca²⁺ (or the presence of 70 μM Mn²⁺); and a wortmannin-insensitive protein being phosphorylated in the presence of extracellular Ca²⁺. In line with the inhibitory potential of wortmannin on anandamide-induced Ca²⁺ signaling in nominal Ca²⁺-free/Mn²⁺-containing solutions, this PI3K inhibitor prevented Ca²⁺ signaling in response to O1602 in the presence of extracellular Ca²⁺ (Fig. 5B, Table 2).

• Download figure
• Open in new tab
• Download powerpoint

Fig. 5.

PI3K is involved in anandamide/GPR55-evoked Ca²⁺ signaling. (A) The effect of 10 μM anandamide on cytosolic free-Ca²⁺ concentration was tested in the nominal Ca²⁺-free solution in the absence (n=33) or presence (n=38) of 0.1 μM wortmannin. (B) In the presence of extracellular Ca²⁺, intracellular Ca²⁺ signaling was initiated by 10 μM O1602 in the presence of 0.1 μM wortmannin followed by an additional stimulation with 10 μM O1602 after washout of the PI3K inhibitor as indicated (n=14). (C) The effect of the PLC inhibitor U73122 (2 μM, n=26; control, n=30) on anandamide (10 μM) -trigged cytosolic Ca²⁺ signaling was tested in Ca²⁺-free buffer. (D) In Ca²⁺-containing buffer, intracellular Ca²⁺ signaling was initiated by 10 μM O1602 in the presence of 1 μM U73122 followed by an additional stimulation with 10 μM O1602 after washout of the PLC inhibitor as indicated (n=14). (E) Concentration-response relationship of 2APB on cytosolic Ca²⁺ signaling induced by either 100 μM histamine (n=4) or 10 μM anandamide (n=6) in nominally Ca²⁺-free solution. Cytosolic free-Ca²⁺ concentrations were recorded in single cells using fura-2. ^*P<0.001 versus the absence of the inhibitor.

To test the involvement of phospholipase C (PLC) activation in the integrin/PI3K-dependent Ca²⁺ recruitment by anandamide, U73122 [(1-[6-[[(17β)-3-Methoxyestra-1,3,5(10)-trien-17-yl]amino]hexyl]-1H-pyrrole-2,5dione (Bleasdale et al., 1990), Table 2], an inhibitor of PLC (Aschner et al., 1997) was used. Cytosolic Ca²⁺ elevations induced by anandamide in the absence of external Ca²⁺ (Fig. 5C) or O1602 in Ca²⁺-containing buffer (Fig. 5D) were prevented by 2 μM U73122. Moreover, the rather unspecific and membrane permeable Ins(1,4,5)P₃-receptor antagonist 2APB [2-aminoethoxydiphenylborate (Maruyama et al., 1997; Koganezawa and Shimada, 2002)] prevented endothelial Ca²⁺ elevation in response to 10 μM anandamide and 10 μM histamine at comparable potency with IC₅₀ values of 43 (30-60) and 87 (50-152) μM, respectively (Fig. 5E). These data suggest that anandamide-triggered Ca²⁺ signaling is due to PLC-mediated formation of Ins(1,4,5)P₃ that, in turn, elevates cytosolic Ca²⁺ concentration by intracellular mobilization from the ER.

One potential candidate for bridging PI3K activation to PLC stimulation is bone marrow kinase, X-linked/epithelial and endothelial tyrosine kinase (Bmx/Etk) (Qiu and Kung, 2000), a member of the Tec family (Takesono et al., 2002). Immunoprecipitation revealed a 3.2±0.5-fold (n=4) increase in phosphorylation of Bmx/Etk upon anandamide in the absence but not presence of extracellular Ca²⁺ (Fig. 6A). Accordingly, siRNA against Bmx/Etk (efficiency shown in supplementary material Fig. S6A) diminished anandamide-induced Ca²⁺ signaling in the absence of external Ca²⁺ or in the presence of 70 μM Mn²⁺ in Ca²⁺-containing solution (Fig. 6B). Moreover, 10 μM of the selective Tec family inhibitor LFM-A13 [2-cyano-N-(2,5-dibromophenyl)-3-hydroxy-2-butenamide, Table 2] (Chau et al., 2002) prevented anandamide-triggered Ca²⁺ signaling in the absence of external Ca²⁺ or (Fig. 6C) in the presence of Mn²⁺ in Ca²⁺-containing solution. These data are in line with phosphorylation of the 70-80 kDa band upon anandamide stimulation (supplementary material Fig. S5) representing Bmx/Etk that, in turn, activates PLCγ leading to intracellular Ca²⁺ mobilization via the generation of Ins(1,4,5)P₃. Consistent with this, inhibition of Bmx/Etk by 10 μM LFM-A13 also prevented O1602 (10 μM)-triggered Ca²⁺ signaling in Ca²⁺ containing solution (Fig. 6D).

In the presence of external Ca²⁺, activation of Syk via Gα_i/o exhibited inhibitory action on PI3K-Bmx-PLCγ pathway and prevented anandamide/GPR55-induced Ca²⁺ signaling

Based on the western blot (supplementary material Fig. S5) that revealed a wortmannin-insensitive tyrosine phosphorylation at 70-80 kDa in the presence but not absence of extracellular Ca²⁺, a negative-feedback mechanism against the PI3K-Bmx-PLCγ signaling via tyrosine phosphorylation [possibly via spleen tyrosine kinase (Syk)] was hypothesized. RT-PCR of the endothelial cell line used proved the expression of Syk in its long form (Wang et al., 2003) (supplementary material Fig. 6B), which matches the 70-80 kDa band in western blots (supplementary material Fig. S5). To further challenge the concept of inhibitory action of Syk on PI3K-Bmx-PLCγ signaling, the effect of the Syk inhibitor piceatannol [3,4,3′,5′-tetrahydroxy-trans-stilbene (Geahlen and McLaughlin, 1989; Ashikawa et al., 2002)] was investigated. In cells that only weakly responded to anandamide in the presence of external Ca²⁺, this agonist yielded strong Ca²⁺ signaling in the presence of 5 μM piceatannol (Fig. 7A, Table 2). These findings were further supported by the induction of Ca²⁺ signaling by anandamide despite the presence of extracellular Ca²⁺ (Fig. 7B) in cells, which were transiently transfected with siRNA against Syk (supplementary material Fig. S6A). To verify the signaling mechanisms by which endothelial CB₁R triggers activation of Syk, Gα_i/o protein, which has been reported to be coupled to this cannabinoid receptor (Mukhopadhyay et al., 2002), was inhibited by ADP ribosylation with pertussis toxin. Inhibition of Gα_i/o protein by pertussis toxin (Table 2) (400 ng/ml, 3 hours) yielded anandamide-induced Ca²⁺ signaling even in the presence of extracellular Ca²⁺ (Fig. 7C), thus pointing to Gα_i/o-mediated activation of Syk that, in turn, inhibits PI3K-Bmx-PLCγ-Ca²⁺ signaling (see Fig. 10).

• Download figure
• Open in new tab
• Download powerpoint

Fig. 6.

Bmx/Etk is the downstream target for PI3K upon GPR55 stimulation by anandamide. (A) Phosphorylation of Bmx/Etk upon stimulation with 10 μM anandamide in the nominal absence of extracellular Ca²⁺ was tested by immunoprecipitation using anti-Bmx/Etk antibody (upper blot) and subsequent staining with the anti-phosphotyrosine antibody 4G10 (lower blot). (B,C) The effects of the Tec-family (Bmx/Etk) kinase inhibitor LFM-A13 (10 μM, n=46; control, n=30) or pretreatment with siRNA against Bmx/Etk (n=6; control, n=27) on anandamide (10 μM) -induced cytosolic Ca²⁺ signaling were monitored in nominally Ca²⁺-free solution. (D) The effect of the Tec-family (Bmx/Etk) kinase inhibitor LFM-A13 (10 μM, n=75; control, n=71) on O1602 (10 μM) -trigged cytosolic Ca²⁺ signaling in Ca²⁺-containing buffer was tested. Standard immunoprecipitation was applied. In experiments with siRNA, cells were transiently transfected with a vector encoding approved siRNA against Bmx/Etk. Experiments were performed 48 hours after transfection. Cytosolic free-Ca²⁺ concentrations were recorded using fura-2. ^*P<0.0001 versus the absence of the inhibitor.

Clustered β1 integrin released binding to CB₁ receptor upon anandamide stimulation

To explore how integrin-clustering shields anandamide-induced PI3K-Bmx-PLCγ-Ca²⁺ signaling from the inhibitory effect of CB₁R-Gα_i/o-Syk signaling, interaction of integrins with CB₁R was investigated by immunoprecipitation. Under conditions of unclustered integrins (i.e. in the presence of external Ca²⁺), the binding of CB₁R to β1 integrin was not affected by anandamide (normalized CB₁R ratio without versus with anandamide 1: 1.03±0.04, n=4, n.s.; Fig. 7D). By contrast, if integrin clustering was facilitated by Mn²⁺, anandamide yielded detachment of the CB₁R from β1 integrin (CB1R was normalized to β1 integrin; CB₁R ratio without anandamide versus with anandamide 1: 0.31±0.13, n=4, P<0.05; Fig. 7D). Identical results were obtained in Ca²⁺-free solution (CB1R was normalized to β1 integrin; CB₁R ratio without anandamide versus with anandamide 1: 0.30±0.21, n=3, P<0.05; Fig. 7E). Although under conditions of clustered integrins the binding of the CB₁R to β1 integrin was affected by anandamide, it did not alter the localization/distribution of CB₁R (supplementary material Fig. S6C).

• Download figure
• Open in new tab
• Download powerpoint

Fig. 7.

Syk is the mediator of CB₁R-originated repression of anandamide-induced Ca²⁺ signaling. (A,B) The inhibition of Syk by either its inhibitor piceatannol (5 μM, n=24) or siRNA against Syk (n=7; control, n=4) allowed strong cytosolic Ca²⁺ elevation in response to 10 μM anandamide in the presence of 2 mM extracellular Ca²⁺. (C) The effect of the G_i protein inhibitor pertussis toxin (400 ng/ml for 3 hours, n=8; control, n=8) on anandamide (10 μM) -induced Ca²⁺ signaling in the presence of 2 mM extracellular Ca²⁺ was verified in human endothelial cells. (D) Immunoprecipitation using anti-β1-integrin antibody and subsequent staining with either anti-β1-integrin (upper blot) or anti-CB₁R (lower blot). Endothelial extracts were harvested from cells under basal conditions (-) and after stimulation with 10 μM anandamide (+) in the presence of 2 mM extracellular Ca²⁺ (Ca²⁺) or 2 mM extracellular Ca²⁺ plus 70 μM Mn²⁺ (Mn²⁺). Right panels shows precipitation with beads that were not preloaded with anti-β1-integrin were used. (E) Immunoprecipitation using anti-β1-integrin antibody and subsequent staining with either anti-β1-integrin (upper blot), anti-αV-integrin (middle blot) or anti-CB₁R (lower blot). Endothelial extracts were harvested from cells under basal conditions (-) and after stimulation with 10 μM anandamide (+) in the and absence (EGTA) of extracellular Ca²⁺. Right panels shows precipitation with beads that were not preloaded with anti-β1-integrin were used. Standard immunoprecipitation was applied. In experiments with siRNA, cells were transiently transfected with a vector encoding approved siRNA against Syk 48 hours prior to the experiments. Cytosolic free-Ca²⁺ concentrations were recorded using fura-2. ^*P<0.0001 versus the absence of the inhibitor.

Thus, under conditions of clustered integrins, anandamide yielded detachment of the CB₁R from β1 integrin, which, in turn, may preserve anandamide-induced signaling pathways via e-aR leading to cytosolic Ca²⁺ elevation.

Depending on integrin clustering, anandamide triggered either nuclear accumulation of NFκB or NFAT while MAPK activation occurred under all conditions

To investigate transcriptional consequences of the two distinct signaling pathways initiated by anandamide via either CB₁R or e-aR, the nuclear recruitment of nuclear factor κB (NFκB) and nuclear factor of activated T-cells (NFAT) upon anandamide stimulation was investigated under conditions of clustered and unclustered integrins. In the presence of extracellular Ca²⁺ (i.e. unclustered integrins), anandamide evoked nuclear accumulation of the p65 subunit of NFκB (Fig. 8A) but not NFAT (Fig. 9A), whereas if integrins were clustered by Mn²⁺ in the presence of external Ca²⁺, anandamide failed to stimulate p65 translocation (Fig. 8B) but evoked strong nuclear accumulation of NFAT (Fig. 9B).

In contrast to the selective activation of the transcription factors NFκB and NFAT, Erk1 and Erk2 were activated by anandamide under all circumstances. Notably, Erk1 and Erk2 activation in response to anandamide was slightly faster (+42±26% after 2 minutes) but less pronounced (maxima +430±88% after 45 minutes; n=3) under conditions of unclustered integrins (Fig. 9C) than if integrin clustering was evoked by Mn²⁺ (–5±17% after 2 minutes and maxima +730±185% after 45 minutes; n=3) (Fig. 9D) or in the absence of extracellular Ca²⁺ (data not shown).

• Download figure
• Open in new tab
• Download powerpoint

Fig. 8.

NFκB is activated independently of anandamide-triggered Ca²⁺ signals. The effect of 10 μM anandamide on the (trans)location of GFP-tagged p65 subunit of NFκB was visualized in endothelial cells in buffer containing 2 mM Ca²⁺ in the absence (A) or presence (B) of 70 μM Mn²⁺. Images of endothelial cells that were transiently transfected with p65-GFP were taken 48 hours after transfection using a conventional fluorescence microscope. Representative images of at least 10 cells out of three experiments on alternate days are shown. Scale bars: 10 μm.

Receptor(s) involved in anandamide-induced endothelial cell activation

Endocannabinoids interact with two cannabinoid receptors identified by molecular cloning: CB₁R, which is expressed predominantly in the brain (Matsuda et al., 1990) but is also present in peripheral tissues (Shire et al., 1995), and CB₂R, expressed mainly by cells of the immune system (Munro et al., 1993) but also calf pulmonary aortic endothelial cells (Zoratti et al., 2003). In human and murine vasculature, the actions of endogenous cannabinoids have been widely interpreted as being mediated by CB₁R, although there is a growing amount of evidence, particularly in isolated heart and blood vessel preparations, that another cannabinoid receptor, referred as e-aR may exist (Offertaler et al., 2003).

• Download figure
• Open in new tab
• Download powerpoint

Fig. 10.

Schemes of the two signaling cascades activated by anandamide in endothelial cells. (A) Anandamide-induced Ca²⁺ signaling under conditions of unclustered integrins originates from CB₁R that is linked to β1 integrin and stimulates Syk via Gα_i/o, subsequently resulting in the activation of NFκB, Erk1 and Erk2. Additionally, Syk inhibits PI3K and, thus, the downstream signaling of the GPR55 is prevented. (B) Once integrins are clustered, CB₁R uncouples from β1 integrin and, thus, no negative feedback on PI3K via Syk occurs. Consequently, GPR55-initiated signaling becomes promoted, resulting in the activation of a PI3K-Bmx-PLCγ cascade that triggers in the production of IP₃ and subsequent intracellular Ca²⁺ mobilization, yielding activation of NFAT, Erk1 and Erk2. Based on the data presented, involvement of G_q/s in GPR55-triggered signaling cannot be not confirmed or excluded at this stage. Abbreviations: α1β3, α1β3 integrin; αvα5, αvα5 integrin; Bmx/Etk, bone marrow kinase, X-linked/epithelial and endothelial tyrosine kinase; CB₁R, cannabinoid 1 receptor; ER, endoplasmic reticulum; Erk1/2, extracellular signal regulated kinases 1 and 2; GPR55, G-protein-coupled receptor 55 (e-aR, `atypical' endothelial anandamide receptor); IP₃, Ins(1,4,5)P₃; G_i, G_i protein; G_q, G_q (G_α13) protein; NFAT, nuclear factor of activated T-cells; NFκB, nuclear factor κB; p50, nuclear factor κB p50; p65, nuclear factor κB, subunit 3; PI3K, phosphoinositide 3-kinase; PLC, phospholipase C; Syk, spleen tyrosine kinase; Tk, tyrosine kinase.

Our findings that, despite the presence of CB₁R, its very selective synthetic agonist HU210 [(6aR)-trans-3-(1,1-dimethylheptyl)-6a,7,10,10a-tetrahydro-1hydroxy-6,6-dimethyl-6H-dibenzo[b,d]pyran-9-methanol] (Felder et al., 1995) failed to stimulate intracellular Ca²⁺ signaling while the e-aR agonist O1602 (Bukoski et al., 2002; Jarai et al., 1999) evoked a strong Ca²⁺ signal in Ca²⁺ containing solution, indicate that anandamide-triggered Ca²⁺ signaling is mediated by e-aR in the human umbilical vein-derived endothelial cell line used in this study. However, anandamide failed to induce a Ca²⁺ signal in the presence of extracellular Ca²⁺. Accordingly, this points to a repressive effect of CB₁R-originated signal transduction on signaling pathways beyond e-aR. This hypothesis was further confirmed by our findings that activation of CB₁R with HU210 strongly diminished the O1602-initiated Ca²⁺ signal. Notably, caution is advised on the specificity of CBR antagonist/agonists, and, thus, additional efforts to characterize involved receptors were undertaken.

However, in the absence of external Ca²⁺, anandamide initiated strong cytosolic Ca²⁺ signaling. Notably, this signal was insensitive to the CB₂R antagonist SR144528 but was prevented by rimonabant, an inhibitor of CB₁R as well as of e-aR (Offertaler et al., 2003). Importantly, as O1918, the antagonist of the e-aR (Jarai et al., 1999; Mo et al., 2004; Offertaler et al., 2003), prevented anandamide-induced Ca²⁺ signaling in the absence of external Ca²⁺, these data indicate that in the nominal absence of extracellular Ca²⁺ the CB₁R-originated suppressive effect is uncoupled from the e-aR pathway, and, thus, allows the generation of a Ca²⁺ signal. Hence, using RT-PCR and siRNA, the putative anandamide receptor GPR55 (Brown, 2007; Hiley and Kaup, 2007; Ryberg et al., 2007) was identified as the e-aR. Hence, overexpression of GPR55 strongly increased Ca²⁺ signaling upon anandamide and O1602 treatment in Ca²⁺-free and Ca²⁺-containing buffer, respectively. However, the potency of anandamide was less in this study than previously reported (Ryberg et al., 2007), which may be due to the different cell models (i.e. transfected HEK 293 cells versus endothelial cells). Hence, lysophosphatidylinositol, a recently described agonist of the GPR55 (Oka et al., 2007), was found to trigger Ca²⁺ signaling via constitutive GPR55 in endothelial cells. Accordingly, there are several questions arising: first, how does the e-aR/GPR55 trigger Ca²⁺ signaling; second, what is the molecular mechanism of the CB₁R-mediated suppressive effect on anandamide-induced Ca²⁺ signaling; third, why and how is the Ca²⁺ signal via e-aR/GPR55 shielded from CB₁R-mediated suppression in the absence of extracellular Ca²⁺; and, fourth, what are the physiological consequences of both signaling pathways for transcriptional control in endothelial cells?

GPR55 interacts with integrins and triggers Ca²⁺ signaling via PI3K and Bmx

In the absence of extracellular Ca²⁺, integrins might be involved in the transition of GPR55-originated signaling that yields intracellular Ca²⁺ mobilization from the ER (Fig. 10B). In particular, the inhibitory properties of the crosslinking antibody against αvβ3 integrin [LM609 (Charo et al., 1990; Cheresh, 1987)] and the functional-blocking antibodies against the β1 [JB1A (Mohri et al., 1996)] and β3 subunits [B3A (Vignali et al., 1990)] on anandamide/O1602-induced Ca²⁺ signaling fuel the hypothesis that, upon anandamide/O1602 stimulation, GPR55 interacts with αvβ3 and α5β1 integrins, which is prerequisite to transmitting outside-in signaling towards intracellular downstream targets (Fig. 10A). This conclusion is further supported by the inhibitory effects of fibronectin (Rupp and Little, 2001) or its integrin-binding motif RGD (Bar-Shavit et al., 1991; Ruoslahti, 2003; Ruoslahti and Pierschbacher, 1987) on anandamide-/O1602-induced Ca²⁺ signaling. The importance of integrin clustering was additionally confirmed by our results that the inhibition of ROCK, which is required for clustering of the integrins (Rodriguez-Fernandez et al., 2001), by Y27632 (Uehata et al., 1997) strongly diminished anandamide-induced Ca²⁺ signaling.

Anandamide-induced Ca²⁺ signaling in Ca²⁺-free medium was sensitive to the inhibition of PI3K, PLC and the Ins(1,4,5)P₃ receptor. Thus, the involvement of PI3K upstream of PLC activation and Ins(1,4,5)P₃-mediated intracellular Ca²⁺ mobilization can be assumed (Fig. 10B). The 70-80 kDa protein Bmx/Etk, a unique tyrosine kinase of the Tec family in endothelial and epithelial cells (Takesono et al., 2002), was identified as mediator between PI3K and PLC. PI3K-induced activation of Bmx/Etk has been described in prostate cancer cells and in breast cancer cells (Bagheri-Yarmand et al., 2001; Qiu et al., 1998). In support of this, the Tec family inhibitor LFM-A13 (Chau et al., 2002; Mahajan et al., 1999), as well as Bmx/Etk-targeted siRNA inhibited anandamide- and O1602-induced Ca²⁺ signaling. Our data that the wide-range tyrosine kinase inhibitors PP1 and PP2 (Liu et al., 1999) abolished anandamide- and O1602-induced Ca²⁺ signaling, and the enhanced tyrosine phosphorylation lead us to suggest that Bmx/Etk undergoes tyrosine phosphorylation by Src family kinases.

Overall, our data point to anandamide-triggered Ca²⁺ signaling as the apparent result of a signaling cascade starting with anandamide binding to GPR55 that clusters with αvβ3 and α5β1 integrins to transmit outside-in signaling, leading to PI3K-mediated activation/translocation of Bmx/Etk. After Bmx/Etk phosphorylation by a member of the Src family (Stephens et al., 2001), it activates PLCγ (Qin and Chock, 2001) that, in turn, instigates Ins(1,4,5)P₃-triggered intracellular Ca²⁺ mobilization from the ER (Fig. 10B).

CB₁-receptor mediated suppression on anandamide-triggered Ca²⁺ signaling is mediated via Gα_i/o and Syk

Notably, inhibition of Gα_i/o by pertussis toxin unmasked intracellular Ca²⁺ signaling upon anandamide in the presence of extracellular Ca²⁺, pointing to the involvement of Gα_i/o downstream of CB₁R (Howlett et al., 2002) in a negative-feedback loop that prevents intracellular Ca²⁺ signaling upon anandamide under these conditions (Fig. 10A).

A potential executor of the inhibitory effects on GPR55-PI3K-Bmx signaling downstream of CB₁R and Gα_i/o is Syk. This tyrosine kinase with splicing variants of ∼72 kDa and ∼68 kDa (Wang et al., 2003) inhibits activation of PI3K in breast cancer cells (Mahabeleshwar and Kundu, 2003) and is expressed in endothelial cells (Inatome et al., 2001). Accordingly, the long form of Syk could be found in the human endothelial cell line used for this study. Moreover, selective inhibition of Syk by piceatannol (Ashikawa et al., 2002) and the respective siRNA unmasked anandamide-triggered Ca²⁺ signaling in the presence of extracellular Ca²⁺. Additionally, wortmannin-resistant tyrosine phosphorylation of a 70-80 kDa protein was found in response to anandamide in the presence but not absence of extracellular Ca²⁺. These data indicate that endothelial Syk indeed executes repression of anandamide-triggered Ca²⁺ signaling downstream to CB₁R and Gα_i/o (Fig. 10A).

Integrins shield GPR55-triggered signaling from CB₁R-controlled suppression

Neither NMR spectroscopy nor anandamide uptake experiments provided evidence for a Ca²⁺-anandamide chelate to prevent binding/uptake of free anandamide and, thus, Ca²⁺ signaling in the presence of extracellular Ca²⁺. Therefore, we speculated that the interaction of extracellular Ca²⁺ with cell membrane constituents is responsible for the observed inhibitory effect of extracellular Ca²⁺ on anandamide-induced Ca²⁺ signaling in endothelial cells. Hence, the existence of a rather specific Ca²⁺-binding site was further supported by the findings that the Ca²⁺ surrogates Sr²⁺ and Ba²⁺ also prevented anandamide-induced Ca²⁺ signaling in a concentration-dependent manner. Notably, the IC₅₀ value of Ca²⁺ (∼8 μM) to prevent anandamide-triggered Ca²⁺ signaling is strikingly similar to that found for Ca²⁺-induced inhibition of integrin clustering (Leitinger et al., 2000; Leitinger et al., 1999). Consequently, the involvement of these dynamically regulated transmembrane heteromultimers in unmasking GPR55-triggered Ca²⁺ signaling in the absence of extracellular Ca²⁺ was postulated. Clustered integrins (i.e. in the absence of extracellular Ca²⁺) may protect GPR55-initiated signaling against the repressive effects of CB₁R-originated pathways. In line with this assumption, Mn²⁺, a potent activator of integrin clustering (Mould et al., 1998; Smith and Cheresh, 1990; Thamilselvan et al., 2003), allowed anandamide-induced Ca²⁺ signaling even in the presence of extracellular Ca²⁺, while Mn²⁺ did not induce any Ca²⁺ signal by itself.

In addition, binding of CB₁R to the β1-integrin subunit was released upon anandamide only if integrins were clustered, which further supports the concept that clustering of integrins shields anandamide-induced PI3K-Bmx-PLCγ-Ca²⁺ signaling from the inhibitory effect of CB₁R activation (Fig. 10). Besides CB₁R, phospho-Syk was also found to interact directly with the cytoplasmic domain of β1-, β2- and β3-integrin subunits (Woodside et al., 2002).

Accordingly, the interaction of CB₁R and Syk with inactive (i.e. unclustered) β1 integrin might either disturb the cytoplasmic integrin rearrangement necessary to transmit the GPR55-initiated signal to its downstream target or allow Syk-mediated inhibition of PI3K owing to the proximity of both proteins (Fig. 10A). However, once integrins cluster either by activation of ROCK, extracellular Mn²⁺ or removal of extracellular Ca²⁺, CB₁R and Syk are released from β1 integrin and no further suppression of PI3K occurs, thus allowing GPR55-originated Ca²⁺ signaling to be transmitted (Fig. 10B). Though the involvement of a G_α13 in GPR55-triggered Ca²⁺ signaling (Ryberg et al., 2007) seem unlikely in the cells used, owing to the involvement of integrins, it cannot be excluded.

In contrast to anandamide that binds to CB₁R and GPR55, activation of GPR55 only by either O1602 or LPI yielded Ca²⁺ signaling independently of extracellular Ca²⁺, thus indicating that the negative feedback by CB₁R needs to be activated in order to achieve inhibition of the GPR55-originated pathway.

Transcription factor activation upon anandamide depends on the signaling pathway activated

Syk induces tyrosine phosphorylation of IκBα, causing dissociation, phosphorylation and nuclear translocation of the p65 subunit of NFκB (Takada et al., 2003). This is in line with the data indicating that, in the presence of extracellular Ca²⁺, anandamide triggered nuclear translocation of p65 (Fig. 10A). However, under conditions of clustered integrins, anandamide failed to evoke the translocation of p65. The lack of NFκB activation under these conditions is consistent with reports pointing to the importance of Ca²⁺ oscillation for the activation of this transcription factor (Hu et al., 1999). By contrast, nuclear translocation of Ca²⁺/calcineurin activated NFAT (Clipstone and Crabtree, 1992; Loh et al., 1996) upon stimulation with anandamide was exclusively obtained under conditions of clustered integrins (Fig. 10B). These findings are in agreement with our previous report that intracellular Ca²⁺ mobilization activates NFAT in human endothelial cells (Bochkov et al., 2002). Notably, although the contribution of NFκB and NFAT to gene expression and regulation of cell function often overlaps, there are considerable differences as NFκB is frequently referred to as being involved in endothelial cell apoptosis (von Albertini et al., 1998) and the initiation of inflammatory processes (Badrichani et al., 1999), while NFAT has been implicated in angiogenesis (Qin et al., 2006) and cell growth (Li et al., 2005).

Interestingly, anandamide-activated mitogenic signaling results in phosphorylation of Erk1 and Erk2 under all conditions, which is consistent with frequent reports that integrins and cannabinoid receptors induce mitogen-activated protein kinase (MAPK) signaling pathways (Liu et al., 2000). The phenomenon of Erk1 and Erk2 phosphorylation under the various conditions (i.e. with and without Ca²⁺ signaling) might be related to the activation of different pathways, resulting in phosphorylation of Erk1 and Erk2. Consistently, Syk stimulates Erk1 and Erk2 via the Ras-MAPK pathway (Kawakami et al., 2003) but also Ca²⁺ that is known to influence cell proliferation and differentiation (Berridge et al., 2000a; Berridge et al., 2000b) stimulates the Ras-MAPK pathway (Cullen and Lockyer, 2002).

In addition to such alternative activation of distinct transcription factors, the observed GPR55-triggered Ca²⁺ signaling may also mediate Ca²⁺-sensitive cell functions such as activation of nitric oxide synthase, formation of prostaglandins or endothelium-derived hyperpolarizing factor (Graier et al., 1994), proliferation (Munaron, 2002), or permeability (Tiruppathi et al., 2006). Accordingly, depending on the status of integrins, anandamide initiates alternate processes in endothelial cells that may even lead to opposite cell functions (e.g. apoptosis versus cell proliferation).

Although both signaling cascades and their mutual interrelation could be described in herein, important questions still remain to be answered. What are the events that favor one of the two alternative signaling pathways? What are the physiological and/or pathological consequences of the respective signaling cascades? Particularly in view of the known initiators of integrin clustering such as inflammation (Staunton et al., 2006), adhesion (Shattil and Newman, 2004) and migration (Lindbom and Werr, 2002), the physiological implication of the dual signaling induced by anandamide in endothelial cells urges its investigation.

Chemicals, supplies and materials

Fura-2/AM was from Molecular Probes Europe (Leiden, Netherlands). ³H-anandamide was obtained from APB (Vienna, Austria). Anandamide (arachidonylethanolamide), R-(+)-methandamide, U73122, 2APB, LMF-A13, HU210 and piceatannol were purchased from Tocris Cookson (Northpoint, Avonmouth, Bristol, UK). O1602, O1918 and AM251 were from Cayman Europe (Tallinn, Estonia). SR144528 and rimonabant were kind gifts from Sanofi-Recherche (Monpellier, France). Bradykinin, arachidonic acid, ATP and lysophosphatidylinositol were from Sigma Chemicals (St Louis, MO, USA). RNeasy Mini kit and the EndoFree Plasmid Maxi kit were purchased from Qiagen (VWR, Vienna, Austria). RQ1 RNase-free DNase I and TransFas were from Promega (Mannheim, Germany). RNAguard and random hexamer primers were from APB. First-strand buffer, moloney murine leukemia virus reverse transcriptase, cell culture media and media substitutes were obtained from Life Technologies (Gibco, Invitrogen, Vienna, Austria). Fetal calf serum was from PAA (Linz, Austria). DyNAzyme II DNA polymerase was from Finnzymes Oy (Vienna, Austria). Primers were synthesized at MWG Biotech (Ebersberg, Germany). Restriction enzymes and T4 DNA ligase were purchased at New England BioLabs (Frankfurt, Germany) or Promega. The expression vectors pcDNA 3 and pBudCE4.1 were obtained from Invitrogen (Vienna, Austria). The dNTPs and all other chemicals were from Roth (Karlsruhe, Germany). The CB1R antibody (H-150) was from Santa Cruz Biotechnology (Heidelberg, Germany). Antibodies for β1 integrin (JB1A) and αV integrin (SAM-1) were obtained from Biomedica/Chemicon (Millipore, Vienna, Austria) and Protein G Sepharose 4 Fast Flow for immunoprecipitation was from GE Healthcare (Vienna, Austria).

Cell culture and transfection

The human umbilical vein derived endothelial cell line, EA.hy926 (Edgell et al., 1983) at over 45 passages was grown in DMEM containing 10% FCS and 1% HAT. In addition, we have performed experiments in short-term cultured human umbilical vein endothelial cells and human uterine artery endothelial cells (Trenker et al., 2007). If necessary, dishes were coated with 20 μg/ml fibronectin (Short et al., 1998). Transient transfection was performed using Transfast according to the manufacturer's protocol.

Reverse transcriptase-PCR

Total RNA was isolated using the RNeasy Mini Kit according to the manufacturer's protocol and subjected to RT-PCR as described previously (Zoratti et al., 2003). The identity of all PCR products was verified on TAE agarose gels and was further confirmed by automated sequencing of the isolated PCR.

Reduction of gene expression by siRNA

The siRNA against human Bmx/Etk was selected at position 1482 bp of the human mRNA and ligated into pSuppressor using the SalI/XbaI sites. For identification of siRNA-expressing cells, the U6 promotor and Bmx/Etk or Syk siRNA were cloned into pBudCE4.1 encoding mtDsRed (Malli et al., 2003) via KpnI/XhoI at the second multicloning site. The siRNA against human Syk in pSuppressor was purchased by Imgenex. siRNA against GPR55 was from Quiagen (Hs_GPR55_5_HP). The efficiency of siRNAs and of an appropriate negative control was approved by real-time PCR (supplementary material Fig. S2E; see Fig. 4A).

Western blot analysis

EA.hy926 cells were scraped in lysis buffer and disrupted by three cycles of freezing/thawing. Equal amounts of total protein were subjected to western blot and detected by enhanced chemiluminescence as previously described (Schaeffer et al., 2003). If required, membranes were stripped with 0.1 M glycine (pH 2.5), and re-probed.

Immunoprecipitation

Cell lysates were incubated with 2-5 μg (200 μg/ml) of specific antibody overnight. The obtained immune complexes were precipitated with Protein G Sepharose 4 Fast Flow, according to the manufacturer's protocol and subjected to western blot analyses.

Immunohistochemistry

Fixed and permeabilized cells grown on glass cover were incubated overnight at 4°C with anti-NFAT antibody. After incubation with Alexa Fluor 488 secondary antibody (2 μg/ml), the glass cover slips were sealed and fluorescence was visualized using deconvolution microscopy.

Measurement of tyrosine kinase activity

Tyrosine kinase activity was monitored in single endothelial cells using a probe for tyrosine phosphorylation of the CrkII adaptor protein [Picchu-936X (Kurokawa et al., 2001)], as previously described (Schaeffer et al., 2003).

Cytosolic free Ca²⁺ measurements

Cytosolic free-Ca²⁺ was measured using Fura2/AM as previously described (Frieden et al., 2002; Malli et al., 2005; Trenker et al., 2007). [Ca²⁺]_cyto was expressed as (F₃₄₀/F₃₈₀)/F₀.

FRET measurement of endoplasmic free-Ca²⁺ concentration and kinase activity

Tyrosine kinase activity was monitored in single endothelial cells using a probe for tyrosine phosphorylation of the CrkII adaptor protein [Picchu-936X (Kurokawa et al., 2001)] as previously described (Schaeffer et al., 2003). For the assessment of the intraluminal free-Ca²⁺ concentration of the ER, cells were transiently transfected with Cam4_er and measurements were performed as described previously (Frieden et al., 2002; Malli et al., 2003; Malli et al., 2005). FRET-based sensors were excited at 440 nm (440AF21, Omega Optical, Brattleboro, VT, USA) and emission was collected simultaneously at 535 and 480 nm using an optical beam splitter (Dual-View MicroImager, Optical Insights, Visitron Systems, Puchheim, Germany). Kinase activity or alterations in ER Ca²⁺ were expressed as (F₅₃₅/F₄₈₀)/F₀.