Induction of Intracellular Calcium Concentration by Environmental Benzo(a)pyrene Involves a β2-Adrenergic Receptor/Adenylyl Cyclase/Epac-1/Inositol 1,4,5-Trisphosphate Pathway in Endothelial Cells*

Background: Calcium signal induced by the environmental polycyclic aromatic hydrocarbon (PAH) benzo(a)pyrene (B(a)P) contributes to its aryl hydrocarbon receptor (AhR)-related toxic effects. Results: β2-Adrenoreceptor (β2ADR) mediates B(a)P-induced calcium signal through a G protein/adenylyl cyclase/cAMP/Epac-1/inositol 1,4,5-trisphosphate pathway. Conclusion: An important crosstalk between β2ADR and AhR pathways can be activated by PAHs. Significance: β2ADR is potentially involved in deleterious effects of PAHs. Polycyclic aromatic hydrocarbons (PAHs) such as benzo(a)pyrene (B(a)P) are widely distributed environmental contaminants, known as potent ligands of the aryl hydrocarbon receptor (AhR). These chemicals trigger an early and transient increase of intracellular calcium concentration ([Ca2+]i), required for AhR-related effects of PAHs. The mechanisms involved in this calcium mobilization were investigated in the present study. We demonstrated that B(a)P-mediated [Ca2+]i induction was prevented in endothelial HMEC-1 cells by counteracting β2-adrenoreceptor (β2ADR) activity using pharmacological antagonists, anti-β2ADR antibodies, or siRNA-mediated knockdown of β2ADR expression; by contrast, it was strongly potentiated by β2ADR overexpression in human kidney HEK293 cells. B(a)P was shown, moreover, to directly bind to β2ADR, as assessed by in vitro binding assays and molecular modeling. Pharmacological inhibition and/or siRNA-mediated silencing of various signaling actors acting downstream of β2ADR in a sequential manner, such as G protein, adenylyl cyclase, Epac-1 protein, and inositol 1,4,5-trisphosphate (IP3)/IP3 receptor, were next demonstrated to prevent B(a)P-induced calcium signal. Inhibition or knockdown of these signaling elements, as well as the use of chemical β-blockers, were finally shown to counteract B(a)P-mediated induction of cytochrome P-450 1B1, a prototypical AhR target gene. Taken together, our results show that B(a)P binds directly to β2ADR and consequently utilizes β2ADR machinery to mobilize [Ca2+]i, through activation of a G protein/adenylyl cyclase/cAMP/Epac-1/IP3 pathway. This β2ADR-dependent signaling pathway activated by PAHs may likely be crucial for PAH-mediated up-regulation of AhR target genes, thus suggesting a contribution of β2ADR to the health-threatening effects of these environmental pollutants.

In addition, PAHs, including the prototypical PAH benzo-(a)pyrene (B(a)P), have been shown to elicit an early and transient increase in intracellular calcium concentration ([Ca 2ϩ ] i ), that is required for up-regulation of AhR-mediated induction of PAHs target genes. For example, inhibition of the Ca 2ϩ signal evoked by PAHs fully prevents their AhR-related inducing effects toward CYP1A1 or CCL1 expression (5). These data suggest that calcium is a major player of the AhR-signaling pathway activated by PAHs (5,6). Early PAH-related calcium mobilization, however, occurs independently of AhR. Indeed, PAHs known to poorly activate AhR, such as benzo(e)pyrene or pyrene, are able to increase [Ca 2ϩ ] i in a similar manner to that of PAH agonists of AhR such as B(a)P (7); in the same way, 2,3,7,8-tetrachlorodibenzo-p-dioxin (TCDD), an halogenated aromatic hydrocarbon and potent AhR agonist, increases [Ca 2ϩ ] i even in cellular models that do not express AhR (8). Moreover, knockdown AhR expression failed to counteract [Ca 2ϩ ] i signal triggered by B(a)P (7).
Mechanisms that contribute to [Ca 2ϩ ] i modulation by PAHs have been linked to inhibition of sarcoendoplasmic reticulum calcium ATPase (9) or to activation of protein-tyrosine kinases (10), ryanodine receptor (11), store-operated calcium channel, or inositol 1,4,5-trisphosphate (IP 3 ) receptor (5,12). However, the initial events that trigger calcium signaling in response to PAH exposure still remain unknown.
␤-Adrenergic receptors (␤ADRs) belong to the family of G protein-coupled receptors and include the three isoforms ␤1, ␤2, and ␤3 ADR (13). These receptors participate to the control of many physiological processes, like the regulation of smooth muscle contraction, blood pressure, bronchodilation, and glycogenolysis. ␤ADR stimulation by ligands, such as epinephrine, commonly leads to the activation of adenylyl cyclase (AC) via a G s protein and, subsequently, increases the production of cAMP (14). This second messenger is a central player in intracellular signaling and is known to directly activate protein kinase A (PKA), special types of membrane ionic channels (cAMP-gated channel), and a family of guanine nucleotide exchange factors known as exchange protein factor directly activated by cAMP (Epac) and composed of two members, Epac-1 and Epac-2 (15,16).
Signaling pathways, dependent on ␤ADRs, especially ␤2ADR, are known to be modulated by PAHs and related AhR ligands, such as TCDD. For example, TCDD can decrease ␤-adrenergic responsiveness in cardiac muscle cells, with a concomitant increase in [Ca 2ϩ ] i . This effect seems to be caused by the interaction of TCDD with ␤ADRs upstream of AC (17)(18)(19). Accordingly, TCDD was shown to increase intracellular cAMP level via the activation of a G protein (20). In the same way, exposure to PAHs or to cigarette smoke (known to contain several PAHs) decreases the number of ␤ADR, especially of ␤2ADR, on the surface of blood mononuclear cells, airway smooth muscle cells, or tracheal epithelial cells (21)(22)(23). Authors suggest that this decrease of ␤ADR expression corresponds to a desensitization mechanism, a phenomenon usually triggered by ␤ADR activation (24,25).
The fact that PAHs interact with ␤2ADR function and/or expression and that signaling pathway related to this receptor can increase [Ca 2ϩ ] i via cAMP-mediated Epac activation (26 -28), indicates that ␤2ADR might play a role in B(a)P-induced [Ca 2ϩ ] i increase. The present study was thus designed to gain insights into this hypothesis. Our data show that B(a)P binds to ␤2ADR and consequently utilizes ␤2ADR machinery to mobilize [Ca 2ϩ ] i , through activation of a G protein/AC/cAMP/ Epac-1/IP 3 pathway. Furthermore, we observe that this ␤2ADR-dependent signaling pathway activated by PAHs is crucial for PAH-mediated up-regulation of AhR target genes such as CYP1B1, thus suggesting a contribution of ␤2ADRs to the deleterious effects of these environmental pollutants toward health.
Intracellular Calcium Concentration Measurements-Variations in [Ca 2ϩ ] i were analyzed in PAH-exposed HMEC-1 or HEK293 cells by microspectrofluorometry using the Ca 2ϩ -sensitive probe Fura-2AM, as previously reported (12). Briefly, HMEC-1 or HEK293 cells were incubated at 37°C in cell suspension buffer (134.8 mM NaCl, 4.7 mM KCl, 1.2 mM K 2 HPO 4 , 1 mM MgCl 2 , 1 mM CaCl 2 , 10 mM glucose, 10 mM HEPES, pH 7.4) supplemented with 1.5 M Fura-2AM and 0.006% pluronic acid. Following probe loading, cells were placed in a continuously perfused recording chamber mounted on the stage of an epifluorescence microscope (Nikon), and trapped dye fluorescence was measured at 510 nm. The ratio of fluorescence intensities recorded after excitation at 340 nm (F 340 ) and 380 nm (F 380 ) was used to estimate [Ca 2ϩ ] i . Results are presented as normalized calcium level, knowing that basal [Ca 2ϩ ] i was arbitrary normalized to 1. The monochromator and the photome-ters, which allow emission and detection of fluorescence from three to five cells in the field of view, were part of a DeltaRAM system from Photon Technology International (PTI, Birmingham, UK), which also provided software systems to acquire and process data.
siRNA Transfection-Chemically synthesized, doublestranded, ON-TARGETplus SMARTpool siRNAs targeting ␤2ADR or Epac-1 were purchased from Dharmacon (Chicago, IL). ON-TARGETplus non-targeting siRNAs were used as a control. Semi-confluent cells were transfected with 100 nM siRNAs using Dharmafect-1 transfection reagent diluted in antibiotic-free culture medium. Forty-eight hours after transfection, cells were exposed to treatments. Transfection efficacy was verified by Western blotting analysis of ␤2ADR and Epac-1 expression.
Crude Membrane Preparation-Crude membranes were prepared by differential ultracentrifugation as previously reported (29). Briefly, after washing, cells were lysed in buffer containing 1 mM EDTA, 0.2 mM phenylmethylsulfonyl fluoride, and protease inhibitors in 10 mM Tris, pH 7.4, and centrifuged at 500 ϫ g for 5 min to remove nuclei and unbroken cells. Supernatant was next ultracentrifuged at 40,000 ϫ g for 30 min. Pellet, containing membranes, was resuspended in lysis buffer and centrifuged at 40,000 ϫ g for 30 min. The resulting pellet was suspended in binding buffer (0.5 mM MgCl 2 , in 50 mM Tris, pH 7.4), aliquoted, and stored at Ϫ80°C until use. Ci/mmol). The final concentrations of the tritiated PAH were 0 -100 nM. Total volume was adjusted to 200 l with binding buffer, supplemented with 2% bovine serum albumin, and tubes were incubated for 30 min at 37°C. Incubation was terminated by adding 5 ml of ice-cold binding buffer to the tube and rapidly filtering contents through a Whatman GF/C glass microfiber filter under low vacuum. The filter was rinsed twice with 10 ml of ice-cold washing buffer containing 10 mM Tris and 0.5 mM MgCl 2 at pH 7.4. Radioactivity trapped in the filters was then measured by scintillation counting. "Nonspecific" binding was determined in parallel assay tubes that contained crude membranes of HEK wt cells instead of HEK␤2 cells. "Specific" binding was defined as the difference between the total and the nonspecific binding.
Molecular Modeling-The complete docking procedure consists of an initial docking/scoring step, followed by a pharmacophore filtering. Docking and scoring were performed by LigandFit, with CFF force field, a softened van der Walls parameter, and a dielectric constant of 1. The most probable binding site was the orthosteric one, with a total volume of 715 Å 3 , partitioned onto seven subsites for extensive screening. The conformers of flexible ligands were calculated before docking by using the catalyst diverse conformation generation method. The filtering was based on the thiadiazole part of crystallographic timolol (PDB: 3D4S) (30). The aromatic pharmacophore was defined as a centroid of 1-Å radius and two orthogonal and opposite projections from the aromatic plane of 1.5-Å radius. All the algorithms used are implemented in Discovery Studio 2.1 (Accelrys Inc., San Diego, CA). For calibration of the scoring function, a dataset of ten ␤2ADR ligands (K d ranging from 0.1 to 4500 nM) (31) was used to calibrate and validate the docking/scoring protocol. Starting from the Ludi free energy expression form (32), we considered its five components as adjustable parameters. Hence, they were recalibrated by leastsquares fitting to the experimental K d .
Intracellular cAMP and IP 3 Measurements-Cellular concentration of cAMP or IP 3 was quantified by the chemiluminescent immunoassay cAMP-Screen TM System (Applied Biosystems, Foster City, CA) or IP-One elisa (Cisbio, Bedford, MA), respectively, according to the manufacturer's instructions.
RNA Isolation and Analysis-Total RNA was isolated from cells using the TRIzol method (Invitrogen); it was then subjected to reverse transcription real-time quantitative PCR (RT-qPCR) analysis, as previously described (33). Primers were as follows: CYP1B1-forward: 5Ј-TGATGGACGCCTTTATC-CTC-3Ј; CYP1B1-reverse: 5Ј-CCACGACCTGATCCAAT-TCT-3Ј; 18S-forward: 5Ј-CGCCGCTAGAGGTGAAATTC-3Ј; 18S-reverse: 5Ј-TTGGCAAATGCTTTCGCT-3Ј. The specificity of each gene amplification was verified at the end of qPCR reactions through analysis of dissociation curves of the qPCR products. Amplification curves were analyzed with ABI Prism 7000 SDS software using the comparative cycle threshold method. Relative quantification of the steady-state target mRNA levels was calculated after normalization of the total amount of cDNA tested to an 18 S RNA endogenous reference.
Immunoblotting Analysis-Immunoblotting was performed on crude membrane (for analysis of ␤2ADR protein level) or on total cellular extracts (for analysis of Epac-1 protein level) as previously described (34). Briefly, protein samples (40 g) were subjected to electrophoresis in a 10% acrylamide gel and electrophoretically transferred to a nitrocellulose membrane (Bio-Rad). After blocking with Tris-buffered saline containing 4% bovine serum albumin and 0.1% Tween 20 at room temperature, membranes were incubated with specific primary antibody overnight at 4°C and, subsequently, with appropriate horseradish peroxidase-conjugated secondary antibody for 1 h. Immunolabeled proteins were finally visualized by autoradiography using chemiluminescence.
Statistical Analysis-Results are usually presented as means Ϯ S.D. They were statistically analyzed using the Student's t test. The criterion of significance was p Ͻ 0.05; data from binding assays were analyzed using SigmaPlot 11 software.

RESULTS
␤2ADR Is Involved in B(a)P-related [Ca 2ϩ ] i Induction-Owing to the fact that endothelium constitutes one of the well known targets of PAHs (35)(36)(37)(38), human microvascular endothelial cells HMEC-1 were mainly used in the present study. To investigate the potential role of ␤2ADR in the calcium signal triggered by PAHs, we first verified that ␤2ADR stimulation can modulate [Ca 2ϩ ] i in this model. According to previous reports (26,28), ␤2ADR activation by salbutamol (a ␤2ADR agonist) results in an increase in [Ca 2ϩ ] i in HMEC-1 cells. This increase started after ϳ5 min of exposure and was maximum at 30 min (Fig. 1A). As expected, this effect was strongly abolished by antagonizing ␤2ADR using a selective ␤2-blocker, ICI-118,551, or non-selective ␤-blockers, propranolol and carazolol (Fig. 1B). Interestingly, the prototypical PAH B(a)P was able to generate a [Ca 2ϩ ] i increase with kinetics similar to that observed with salbutamol (Fig. 1A). Moreover, the B(a)P-induced [Ca 2ϩ ] i increase was also diminished by ␤2ADR inhibition using ICI-118,551, propranolol, or carazolol (Fig. 1B), thus supporting an implication of ␤2ADR in B(a)P-mediated [Ca 2ϩ ] i increase. This hypothesis was next confirmed by the fact that inhibiting ␤2ADR activity in HMEC-1 cells using specific antibodies targeting ␤2ADR, markedly prevented the increase of [Ca 2ϩ ] i due to B(a)P ( Fig. 2A); in the same way, transfection of siRNAs targeting ␤2ADR, which markedly reduced ␤2ADR expression in HMEC-1 cells as shown by Western blotting (Fig. 2B), counteracted the [Ca 2ϩ ] i increase triggered by B(a)P (Fig. 2B). Finally, we used HEK293 cells, known to constitutively poorly express ␤2ADR (39), and transfected them either with control plasmid (HEK wt ) or with a ␤2ADR plasmid (HEK␤2). As shown by Western blotting (Fig.  2C), HEK␤2 cells exhibited a very large overexpression of ␤2ADR when compared with HEK wt ; they also displayed a marked increase of [Ca 2ϩ ] i in response to B(a)P treatment, in contrast to HEK wt (Fig. 2C).

G s /AC/cAMP Pathway Is Involved in B(a)P-related [Ca 2ϩ ] i
Induction-It has been demonstrated that ␤2ADR-mediated [Ca 2ϩ ] i increase requires G s protein-mediated activation of AC as a first step and the subsequent production of cAMP (26,28). In addition, it is well known that TCDD can rapidly increase cAMP level in C3H10T1/2 cells in a G protein-dependent manner (20). Therefore, we first explored if B(a)P exposure can also induce cAMP levels in HMEC-1 cells. As shown in Fig. 3A, B(a)P transiently enhanced cAMP level in HMEC-1 cells, with a rapid onset after 2 min of exposure and a maximum effect at 10 min. Moreover, this induction was significantly abolished in the presence of the selective ␤2-blocker ICI-118,551, or a competitive inhibitor of AC, dd-Ado (Fig. 3B), pointing to the involvement of ␤2ADR-related AC activation in the cAMP level increase by B(a)P. We noted that the induction peak of cAMP levels in B(a)P-exposed HMEC-1 cells occurred earlier than that of [Ca 2ϩ ] i (10 min versus 30 min), which suggests that PAHs-related induction of [Ca 2ϩ ] i occurs downstream of the activation of cAMP. We then investigated the impact of cAMP increase on B(a)P-related calcium signal. As shown in Fig. 3 (C  and D), B(a)P failed to significantly increase [Ca 2ϩ ] i in HMEC-1 cells when cAMP production was inhibited using a potent inhibitor of G protein (suramin, Fig. 3C) or competitive inhibitors of AC such as dd-Ado (Fig. 3D) or MDL-12,330A (data not shown).

Epac-1 Is Involved in B(a)P-related [Ca 2ϩ
] i Induction-An important target of cAMP is the family of Epac proteins, Epac-1 and -2. These guanine nucleotide exchanger factors were initially identified by De Rooji and co-workers to explain the PKAindependent activation of Rap by cAMP (40). There is now considerable evidence that these proteins play major roles in cellular physiology (41); they are notably involved in regulation of calcium channels and endothelium physiology and, according to Schmidt and co-workers (28), the increase of [Ca 2ϩ ] i by ␤2ADR requires the cAMP-dependent activation of Epac-1. Two approaches were used to test the implication of Epac-1 in B(a)P-related [Ca 2ϩ ] i variation in HMEC-1 cells. First, we used the Epac inhibitor BFA (42), which was found to markedly block B(a)P-related [Ca 2ϩ ] i increase (Fig. 4A). Second, we down-regulated Epac-1 expression using siRNA transfection. As shown in Fig. 4B, the decrease of Epac-1 expression assessed by Western blotting was associated with an inhibition of B(a)Prelated [Ca 2ϩ ] i increase. 3 is a universal intracellular messenger that mediates Ca 2ϩ release from intracellular stores via the activation of IP 3 receptor (IP 3 R) located at the endoplasmic reticulum membrane (43). This messenger usually originates from phospholipase C-dependent hydrolysis of phosphatidylinositol diphosphate after hormonal activation of the cell (28,44). Because phospholipase C can be activated by Epac-1 following ␤ADR stimulation (27), it is tempting to speculate that the ␤ADR/AC/Epac-1-signaling cascade activated by B(a)P may result in IP 3 up-regulation. Consistent with this hypothesis, B(a)P was found to significantly increase the IP 3 level in HMEC-1 cells (Fig. 5A). This IP 3 up-regulation was moreover inhibited in the presence of the ␤2ADR inhibitor ICI-118,551 or the AC inhibitor dd-Ado, thus likely indicating it was dependent on activation of the ␤2ADR/ AC-signaling pathway in response to B(a)P. In agreement with this conclusion, IP 3 accumulation elicited by exposure to B(a)P  was observed in ␤2ADR-overexpressing HEK␤2 cells, but not in HEK wt cells (Fig. 5B).

IP 3 Is Involved in B(a)P-related [Ca 2ϩ ] i Increase-IP
To determine whether B(a)P-mediated IP 3 induction occurs upstream of B(a)P-related [Ca 2ϩ ] i induction, cells were preincubated with IP 3 recycling-blocking agents, namely L-690.330 and LiCl, known to disrupt IP 3 metabolism by inhibiting inositol-monophosphatase, thus leading to the reduction of myoinositol necessary for the building of neo-synthesized IP 3 , thereby finally down-regulating the IP 3 level (45,46). Calcium increase by B(a)P was found to be strongly repressed in the presence of L-690.330 (Fig. 5C) or LiCl (data not shown), suggesting an important contribution of IP 3 to BaP-mediated [Ca 2ϩ ] i variation. This conclusion was next fully confirmed by the fact that B(a)P-related [Ca 2ϩ ] i increase was abolished by preincubation with 2-APB, a specific antagonist of IP 3 R (47) (Fig. 5D), or with TMB-8, frequently used as an inhibitor of calcium release from intracellular stores (48) (Fig. 5E).
Contribution of ␤2ADR Pathway to AhR-dependent Regulation of B(a)P Target Genes-To determine whether the different elements, acting in the ␤2ADR-related signaling cascade putatively activated by B(a)P and described above, may be implicated in the regulation of the AhR gene target CYP1B1, we next analyzed the effect of their chemical inhibition or downregulation on B(a)P-mediated induction of CYP1B1 mRNA expression. The use of the ␤ADR blockers propranolol and ICI-118,551 was found to markedly inhibit CYP1B1 up-regulation in response to B(a)P (Fig. 6A); similar results were observed with the G protein inhibitor suramin or the AC inhibitor dd-Ado (Fig. 6B). In the same way, BFA-mediated pharmacological inhibition of Epac-1 or the siRNA-related knockdown of Epac-1 expression prevented B(a)P-related CYP1B1 induction (Fig. 6, C and D). Finally, inhibiting calcium release from intracellular stores using TMB-8 or the IP 3 R antagonist 2-APB also counteracted CYP1B1 up-regulation (Fig. 6E).
B(a)P Directly Binds ␤2ADR-The implication of ␤2ADR in B(a)P-related [Ca 2ϩ ] i induction may correspond to a direct activation of these receptors by the PAH, or it may alternatively reflect a positive and required cooperation between the ␤2ADR system and another yet unknown system activated by B(a)P. To gain insights into this issue, we first performed B(a)P saturation binding assays using ␤2ADR-positive crude membranes. Total binding was determined by incubating crude membranes from HEK␤2 cells with different concentrations of [ 3 H]-B(a)P (from 0 to 100 nM), whereas nonspecific ␤2ADR-independent binding was concomitantly evaluated using crude membranes from HEK wt cells instead of HEK␤2 counterparts. Subtraction of nonspecific binding from total binding revealed a specific and saturable binding of B(a)P to HEK␤2 crude membranes, with a B max of ϳ2.5 pmol/mg and a K d of ϳ10 nM (Fig. 7A). To further investigate the possibility of direct interaction between B(a)P and ␤2ADR, we then docked B(a)P into the x-ray structure of human ␤2ADR (PDB: 2RH1) (49). The most favorable binding domain for B(a)P was located between TM3, TM5, and TM6 segments of ␤2ADR (Fig. 7B). This binding mode most likely allows extensive interactions of B(a)P with Phe-193, Tyr-199, Phe-289, and Phe-290 on ␤2ADR and partially overlaps the carazolol binding site (Fig. 7C).

DISCUSSION
Exposure to PAHs is well known to trigger an early increase of [Ca 2ϩ ] i , which is thought to participate to the up-regulation of various genes targeted by the PAHs-activated transcription factor AhR (5,50). The aim of the present study was to characterize the initial events implicated in this calcium mobilization by PAHs. Taking B(a)P as an example of PAH, the present work strongly suggests that [Ca 2ϩ ] i induction in endothelial HMEC-1 cells is initially caused by interaction of the PAH with ␤2ADR and subsequent activation of a G protein/AC/cAMP/ Epac-1/IP 3 pathway (Fig. 8).
The direct involvement of ␤2ADR in calcium mobilization by PAHs is thus supported by (i) the specific binding of B(a)P to ␤2ADR as revealed by saturation binding assay and molecular docking, (ii) the inhibition of B(a)P-mediated [Ca 2ϩ ] i induction by chemical ␤ADR blockers, including the selective ␤2ADR blocker ICI-118,551, by a monoclonal antibody anti-␤2ADR, or by siRNA-mediated knockdown of ␤2ADR expression, and (iii) the fact that the ␤2ADR-transfected HEK␤2 cells exhibited a marked increase of [Ca 2ϩ ] i in response to B(a)P, in contrast to control HEK wt cells. Downstream events following ␤2ADR interaction with B(a)P likely successively involve G s protein, AC, cAMP, Epac-1 and finally IP 3 (Fig. 8). Indeed, the calcium signal generated by B(a)P is reduced by suramin, a G protein inhibitor. Involvement of the downstream factors AC/cAMP was further demonstrated by (i) the ␤2ADRand AC-dependent induction of cAMP levels by B(a)P in HMEC-1 cells and (ii) the inhibition of B(a)P-triggered calcium mobilization by blocking AC activity and consequently cAMP production by dd-Ado or MDL-12,330A. Epac-1, well known to be activated by cAMP (40), was most likely the following actor in the signaling cascade contributing to B(a)P-mediated [Ca 2ϩ ] i increase, because the Epac inhibitor BFA, as well as siRNAs-mediated down-regulation of Epac-1 expression, counteracted calcium mobilization due to the PAH. ␤2ADR-mediated Epac-1 activation is known to induce IP 3 formation in a phospholipase C-dependent manner (27,28), and such a cellular process is likely to also occur in B(a)P-exposed HMEC-1 cells. These cells were found to exhibit induction of IP 3 levels, counteracted by chemical inhibition of ␤2ADR or AC. The fact that B(a)P treatment of ␤2ADR-overexpressing HEK␤2 cells, but not of control HEK wt cells, induced IP 3 levels also supports a link between B(a)P-␤2ADR interaction and IP 3 . Finally, IP 3 is likely a major final actor involved in calcium mobilization triggered by PAHs, because the use of IP 3 recycling-blocking agents like L-690.330 prevented [Ca 2ϩ ] i increase in B(a)P-exposed HMEC-1 cells. Terminally, IP 3 likely acts on IP 3 R found in membranes of endoplasmic reticulum to release calcium, because the IP 3 R antagonist 2-APB was found to abolish calcium mobilization in HMEC-1 cells treated by B(a)P. Similarly, 2-APB has been found to abolish calcium increase occurring in macrophages exposed to B(a)P (5).
It should be kept in mind that, besides the ␤ADR machinery leading to IP 3 formation, various mechanisms contributing to calcium modulation by PAHs have been previously reported, such as protein-tyrosine kinases (10), sarcoendoplasmic reticulum calcium ATPases (9), or ryanodine receptor (11). The exact relevance of these mechanisms with respect to calcium mobilization and their putative interplays with the ␤ADR system remain to be determined. Nevertheless, they may be particularly operant in cells weakly expressing ␤2ADR.
Our study with B(a)P showed that ␤2ADR-related adrenergic-like effects are required for the up-regulation of AhR target gene by PAHs. Indeed, the pharmacological inhibition of various actors of the ␤2ADR/G protein/AC/cAMP/Epac-1/IP 3 -signaling pathway, as well as the down-regulation of Epac-1 expression, were shown to counteract B(a)P-mediated induction of CYP1B1 mRNA levels (Fig. 6). The crucial element implicated in this interaction between ␤2ADRand AhR-signaling cascades is likely to be calcium. Indeed, inhibition of the terminal step of calcium release from intracellular stores through the use of 2-APB was sufficient for inhibiting CYP1B1 induction in HMEC-1 cells exposed to B(a)P. Similarly, 2-APB and the calcium chelating agent 1,2-bis(2-aminophenoxy)ethane-N,N,NЈ,NЈ-tetraacetic acid tetrakis have been shown to prevent the up-regulation of various AhR targets such as CYP1A1 and the chemokine CCL1 in response to AhR agonists (5,6,12). The exact way that calcium interacts with the AhR-signaling cascade remains to be determined, even if a role for calcium-activated signaling pathways such as calmodulindependent protein kinases may be suspected (6,51). In addition to calcium, cAMP may also play a role in the AhR signaling cascade, because cAMP and its downstream kinase PKA have been reported to constitute players of the AhR pathway by themselves (20,52).
The affinity of B(a)P (K d ϭ 10 nM) is among the highest reported for ␤2ADR ligands. Amazingly the neutral lipophilic nature of PAHs structure does not match the known ␤2ADR pharmacophore (53). Thus, the exact nature of B(a)P binding to ␤2ADR remains to be characterized. Nevertheless, the great accessible surface area of B(a)P is likely to favor large van de Walls contacts. Besides, docking analysis suggest extensive interactions of B(a)P with the amino acids Phe-193, Tyr-199, Phe-289, and Phe-290 on ␤2ADR. The aromatic nature of these amino acids combined with that of B(a)P, as well as the geometry of predicted complexes, argue strongly in favor of -stacking. The reduced potency of 6-nitro-B(a)P for elevating [Ca 2ϩ ] i when compared with that of B(a)P strongly supports this hypothesis (data not shown). Indeed, electron-withdrawing substituents, such as the nitro group found in 6-nitro-B(a)P, reduce the interaction energy of T-shaped aromatic dimers (54). Interestingly, -stacking has already been emphasized for ␤ADRs ligands (55,56). Other target proteins of PAHs, such as ␣1-acid glycoprotein or the estrogen receptor (57,58), also rely on aromatic interaction.
The low reported K d (10 nM) for B(a)P binding to ␤ADR is in the range of B(a)P concentrations to which humans may be commonly exposed through diet, air pollution, or cigarette smoke (59,60). This suggests that environmental exposure to PAHs may result in deleterious interactions with the ␤-adrenergic system in humans. Such putative adrenergic-like effects of PAHs remain to be determined. However, administration of B(a)P at low doses has been demonstrated to impair ␤2ADRmediated stimulation of adipose tissue lipolysis and cause weight gain in mice (61), supporting in vivo interactions of PAHs with ␤ADRs. It should be kept in mind that the overall interplay between PAHs and the adrenergic system may be more complex than simply reflecting an activation of ␤2ADR by PAHs. Indeed, exposure of mice to the halogenated hydrocarbon TCDD decreases ␤ADR responsiveness, especially in the heart (17,18), whereas PAH-containing cigarette smoke has been shown to decrease expression of ␤ADRs in blood lymphocytes (23). Such data may be consistent with a desensitization of the ␤-adrenergic system, which occurs in response to initial  stimulation of the adrenergic system by agonists (25) and involves the down-regulation of ␤ADR expression as well as the internalization of membrane ␤ADRs. In agreement with this hypothesis, exposure to a mixture of PAHs for 24 h dramatically diminishes expression of ␤2ADR mRNAs and protein in airway smooth muscle cells (21).
In summary, the prototypical PAH B(a)P was found to bind to ␤2ADR in endothelial HMEC-1 cells and to consequently activate a G protein/AC/cAMP/Epac-1/IP 3 pathway, which in turn resulted in a [Ca 2ϩ ] i increase, required for B(a)P-mediated induction of the AhR gene target CYP1B1. Such data are therefore consistent with a previously unsuspected implication of the ␤2ADR system in deleterious effects of environmental PAHs.