PDZ-dependent Activation of Nitric-oxide Synthases by the Serotonin 2B Receptor*

Taking advantage of three cellular systems, we established that 5-HT 2B receptors are coupled with NO signal- ing pathways. In the 1C11 serotonergic cell line and Mastomys natalensis carcinoid cells, which naturally express the 5-HT 2B receptor, as well as in transfected LMTK 2 fibroblasts, stimulation of the 5-HT 2B receptor triggers intracellular cGMP production through dual activation of constitutive nitric-oxide synthase (cNOS) and inducible NOS (iNOS). The group I PDZ motif at the C terminus of the 5-HT 2B receptor is required for re- cruitment of the cNOS and iNOS transduction pathways. Indeed, the 5-HT 2B receptor-mediated NO cou- pling is abolished not only upon introduction of a competitor C-terminal 5-HT 2B peptide in the three cell types but also in LMTK 2 fibroblasts expressing a receptor C-terminally truncated or harboring a point mutation within the PDZ domain. The occurrence of a direct functional coupling between the receptor and cNOS activity

diversity of 5-HT receptors and the lack of specific pharmacological tools have impaired the investigation of the precise roles of these proteins in the signaling networks that mediate 5-HT physiological functions. This is particularly true in the case of the 5-HT 2B receptor, which, like all the 15 known mammalian 5-HT receptor subtypes but one (5-HT 3 ), belongs to the G protein-coupled receptors (GPCRs). 5-HT 2B receptors cDNA were cloned from the rat stomach fundus, mouse brain, human liver, and neuroblastoma cell lines (see Ref. 3 for review). The 5-HT 2B receptor could be expressed either by transfection or endogeneously during the serotonergic neuronal differentiation of 1C11 cells, referred to as 1C11* /5HT cells (4,5). In both cases, the receptor couples with PIP 2 hydrolysis, similarly to the two other members (5-HT 2A and 5-HT 2C ) of the 5-HT 2 receptor family. In contrast, the 5-HT 2B receptor stimulation fails to stimulate PIP 2 hydrolysis in rat fundus (6) and in rat vasculature (7). Signaling processes involving 5-HT 2B receptors and distinct from PIP 2 hydrolysis might thus occur. Accordingly, 5-HT 2B -dependent activations of both the ras-mitogen-activated protein kinase cascade (8) and phospholipase A 2 (5) have already been reported. The possibility that NO, a major messenger molecule in the cardiovascular, immune, and nervous systems (see Ref. 9 for review), may behave as a signaling effector associated to the 5-HT 2B receptor, is suggested by data obtained in rat vasculature (7), rat fundus (6), and human endothelial cells (10).
The present work provides biochemical evidence for a functional link between NO signaling and the 5-HT 2B receptor. NO synthase (NOS), the enzyme responsible for NO synthesis, occurs as three distinct isoforms (11,12). Two isoforms are constitutive (cNOS), and their activities are dependent upon the calcium-calmodulin complex. One is expressed in some neurones (13) and in skeletal muscles (14). It corresponds to the neuronal isoform (NOS-1). The other is the endothelial isoform (NOS-3), abundant in endothelial and epithelial cells (15). The third NOS isoform is calcium-independent and inducible (iNOS or NOS-2) (11). It responds to endotoxins of Gramnegative bacteria or inflammatory cytokines in a wide variety of cells (most notably in macrophages, glia, or vascular cells). Here, we show that both cNOS and iNOS contribute to NO production upon 5-HT 2B receptor activation in the murine serotonergic lC11* /5HT cells (4) or in Mastomys natalensis carcinoid tumor primary cultured cells (16). These two in vitro systems, which naturally express 5-HT 2B receptors within a complete serotonergic phenotype, allow NO couplings to be studied under conditions close to in vivo physiological situations.
The PDZ domains present in a diverse family of structural proteins and enzymes such as cNOS appear to be important elements in protein-protein interactions at the plasma mem-brane. Interactions between proteins containing either PDZ domains or PDZ recognition motifs are assumed to mediate formation of macromolecular signaling complexes (see Ref. 17 for review). For instance, in rat cerebellum and forebrain (18), the resulting multiprotein clusters contribute to the functional coupling of the NMDA receptor to NOS-1 activity. PDZ target motifs typically consist of 4 -11 amino acid residues at the extreme C terminus of the target protein (17). Interestingly, a group I PDZ motif, VSYI(D/E), is present at the C terminus of the 5-HT 2B receptor. This led us to examine the role of this motif in NO signaling pathways.
In mouse LMTK Ϫ fibroblasts stably transfected with the full-length cDNA encoding the mouse 5-HT 2B receptor (LM6 cells), a 5-HT 2B -NO coupling mediated by cNOS and iNOS is observed. Instead, if the transfected cDNA expresses a C-truncated form of the receptor (LM5 cells), NO production is fully abolished. An involvement of the 5-HT 2B PDZ target domain in the coupling to cNOS and iNOS could be further indicated upon introduction, in the three cell types studied, of a 20-amino acid competitor peptide whose sequence corresponds to the C terminus of the 5-HT 2B receptor. In all cases, the peptide abolished the 5-HT 2B -dependent NO production. The observation that the 5-HT 2B -NOSs coupling is canceled in LMTK Ϫ fibroblasts expressing a 5-HT 2B receptor with a point mutation in the VSYI(D/E) motif (LMTK Ϫ -VSSI cells) confirms the implication of the PDZ domain of the receptor in its coupling to cNOS and iNOS.
The ability of G proteins to directly regulate NO production remains largely to be elucidated. One recent report indicates a regulation of iNOS expression by G␣ 13 in an epithelial renal cell line (19) and suggests that proinflammatory agents such as thrombin or thromboxane might induce cellular responses similar to those initiated by bacterial endotoxins or cytokines. In the present study, we observed that introduction of anti-G␣ 13 antibodies in either LM6, 1C11* /5HT or Mastomys permeabilized cells systematically blocked the 5-HT 2B receptor-mediated iNOS activation. Altogether, these findings highlight novel GPCR-dependent regulations of NO production.
Cell Cultures-To avoid any interfering iNOS induction because of cytokines or undefined components of the media, all cell types were grown, as previously reported (5) in a serum-free medium (Dulbecco's modified Eagle's medium/Ham's F-12 medium (1:1) with 5 g/ml insulin, 5 g/ml transferrin, 30 nM selenium, 20 nM progesterone, and 100 M putrescine). M. natalensis carcinoid tumor primary cultured cells were obtained as previously reported (16). 1C11 progenitor cells were grown and induced to differentiate toward the serotonergic pathway in the presence of 1 mM dibutyryl cAMP and 0.05% cyclohexane carboxylic acid (20). Experiments were performed on 1C11 serotonergic cells 2 days after addition of the inducers (1C11* /5HT cells, (4)). Mouse LMTK Ϫ fibroblasts, lacking 5-HT receptors and thus nonresponsive to DOI stimulation, were stably transfected with cDNA encoding either the full-length (LM6 cells) or a truncated (LM5 cells) mouse 5-HT 2B receptor and cloned. The 5-HT 2B receptor expressed in LM5 cells was deleted for 77 amino acid residues of the C-terminal domain from amino acid 403. LMTK Ϫ cells were also transfected with a cDNA encoding the full-length 5-HT 2B receptor harboring a point mutation in the VSYI motif (LMTK Ϫ -VSSI).
Cell Permeabilization-The cells were washed twice with phosphatebuffered saline containing 0.1% bovine serum albumin and exposed to 1 hemolytic unit of alveolysin/10 6 cells at 37°C under agitation (21). Alveolysin was purified from Bacillus alvei as described (22). One hemolytic unit is equivalent to ϳ1 ng of protein (193 pmol).
Radioligand Binding Experiments-Radioligand binding experiments were performed as previously detailed (4,23 Determination of Intracellular cGMP Levels-Cells were washed twice in fresh serum-free medium and incubated for various times at 37°C with l00 M isobutylmethylxanthine and test agents. The reaction was stopped by aspiration of the medium followed by addition of 500 l ice-cold 95% ethanol/5% formic acid (1:1, v/v). After 1 h at 4°C, the ethanolic phase was collected and lyophilized. cGMP was quantified using an iodinated radio immunoassay kit (cGMP RIA kit RPA 525, Amersham Pharmacia Biotech).
Determination of Intracellular Inositol 1,4,5-Trisphosphate Levels-Cells were washed twice with fresh serum-free medium and incubated for various times at 37°C with 100 nM DOI. After incubation, cells were washed twice in cold phosphate-buffered saline and scraped with a rubber policeman. Intracellular IP 3 levels were then measured radioimmunologically as previously detailed (4).

L-[
3 H]Citrulline in the supernatant was quantified by liquid scintillation counting. Specific NOS activity was calculated as the nitro-Larginine-sensitive formation of L-[ 3 H]citrulline/min/mg of protein. L-Citrulline was quantitatively recovered by this batch assay as determined by using L-[ 14 C]citrulline. In parallel experiments, cell cytosols were incubated in Ca 2ϩ -free buffer to determine Ca 2ϩ -independent NOS (iNOS) activities.
Isolation of Total RNA and Reverse Transcription PCR of NOS Isoforms-First-strand cDNA was synthesized from 1 g of total RNA with oligo(dT) 17  . PCR reactions were performed using the Gene-Amp kit (Perkin-Elmer) in a 50-l volume containing 2 l of the reverse transcription products (40 cycles, 60°C). PCR products were analyzed on 2% agarose gels. NOS-2 mRNA steady state levels were monitored during 1 h upon 5-HT 2B receptor activation by real-time RNA quantification (25).
Site-directed Mutagenesis by Recombinant Circle PCR-Point mutations were introduced into the wild-type mouse 5-HT 2B R cDNA by oligonucleotide-directed mutagenesis as described previously (26). A single-stranded uracil-containing DNA template was generated in Escherichia coli CJ236 (dut Ϫ , ung Ϫ ) (Invitrogen) using R408 helper phage. Using oligonucleotides encoding the mutations, the following mutations were introduced: m5-HT 2B Y478S and R403del. Identity of the mutant cDNAs was confirmed by dideoxynucleotide sequencing using a Sequenase II kit (U. S. Biochemical Corp.). To verify the absence of additional mutations in the parental cDNAs, rescue constructs of all mutations were constructed and tested for recovery of NOS activations. In all cases, the rescue constructs demonstrated wild-type NOS couplings (data not shown), confirming that the disruption of NOS couplings observed for both mutants resulted from the mutations.
Data Analysis and Statistics-The statistical analyses on small groups were performed using nonparametric tests. The significance criterion of p Ͻ 0.05 was adopted. All values are given as the arithmetic means Ϯ S.E.

Activation of the 5-HT 2B Receptor Triggers Intracellular
cGMP Accumulation-Most of the transductional actions of NO are mediated through an activation of cytosolic guanylate cyclases with concomitant increase of cellular cGMP (27). Thus, we first examined whether exposure of 5-HT 2B receptor-expressing cells (i.e. Mastomys, 1C11* /5HT , and LM6 cells) to DOI, a specific agonist of 5-HT 2 receptors, triggered cGMP formation. All cells were grown in serum-free cultures to preclude signaling interferences mediated by cytokines or undefined serum components. As shown in Fig. 1A, saturation of the receptor with 100 nM DOI resulted in an accumulation of cGMP observable after 10 min in all three cell types. The increase in cGMP was lower in 1C11* /5HT and LM6 cells than in Mastomys cells. The elevated response of Mastomys cells can be related to the overexpression of 5-HT 2B receptors in these carcinoid tumoral cells, as previously evidenced by anti-5-HT 2B antibodies (8). It may also reflect cell type-specific features of the 5-HT 2B receptor-mediated signaling responses. Indeed, as observed in rat fundus (6), Mastomys 5-HT 2B receptors do not couple to PLC␤, whereas DOI treatment (100 nM) activates PIP 2 hydrolysis in both LM6 and 1C11* /5HT cells (Fig. 1B).
The DOI-induced cGMP formation observed for the three cell types supports the idea of a 5-HT 2B -NO coupling. We therefore decided to search for the occurrence of a DOI-mediated NOS activation.

5-HT 2B Receptors Are Coupled to Both cNOS and iNOS
Activities-To assess the contributions of Ca 2ϩ -dependent NOSs (cNOS) and/or Ca 2ϩ -independent iNOS to NO synthesis in the different cell systems, NOS activities were monitored in cell lysates in the presence or absence of calcium ions. In each cell investigated, activation of 5-HT 2B receptors by 100 nM DOI resulted in a time-dependent increase in both cNOS ( Fig. 2A) and iNOS (Fig. 2B) activities. cNOS activity becomes maximal 5-15 min after addition of the agonist. Noticeably, the enhancement of cNOS activity lasts longer in Mastomys cells than in the two other strains. This discrepancy is consistent with the high level of cGMP accumulating in Mastomys cells (Fig. 1A).
In the absence of agonist stimulation, low levels of cNOS activity were measured in all three cell lysates but not in LMTK Ϫ untransfected cells. These levels might thus reflect a constitutive basal activity of the 5-HT 2B receptor. Such a basal activity has already been observed for its PLC␤ (4,23), ras-mitogenactivated protein kinase (8), and phospholipase A 2 (5) couplings.
It is noteworthy that DOI-mediated 5-HT 2B stimulation also induces a rapid and transient increase of iNOS activity (Fig.   2B). Activation of iNOS is only very shortly delayed (lag is about 5 min or less) with respect to cNOS activation. This extremely rapid activation of iNOS is reproducibly observed in all three cell types. It contrasts with the iNOS activation by proinflammatory agents, for which the time of response ranges between 3 h (28) and up to several days. In all three cell types, iNOS mRNA steady state levels were assessed by real-time quantitative PCR and could be observed not to vary (data not shown). This suggests that 5-HT 2B -mediated iNOS activation depends on post-transcriptional controls.
The distinct NOS isoforms expressed in each cell type were further characterized through reverse transcription PCR experiments. As shown in Fig. 3, all three cell types contain transcripts encoding iNOS (NOS-2) and either of the two cNOS (NOS-1 or NOS-3) isoforms. As expected, NOS-1 mRNA is present in 1C11* /5HT neural-like cells and in Mastomys neuroendocrine cells, whereas NOS-3 mRNA can only be detected in LM6 fibroblasts, which, like endothelial cells, have a mesodermal origin.
Correlations between NOS Activities and 5-HT 2B Receptor Binding Constants-The functional relation between the 5-HT 2B receptor and NOS activities could be further evidenced by examining the effects of a series of known agonists or antagonists on both DOI binding and NOS activities. Their binding to the receptor was assessed through competition with DOI, and the corresponding IC 50 values were determined. K i values were deduced using the Cheng-Prusoff equation (29). The doseresponse effect of agonists on NOS activities could be directly followed. In the case of antagonists, effects were estimated through the capacity of each drug to inhibit the DOI-dependent stimulation of NOS activities.
For the three cells studied, highly significant correlations (r s ϭ 0.976, n ϭ 8, p Ͻ 0.01) were found between the 5-HT 2B receptor binding constants of the drugs, and the apparent equilibrium constants deduced from their effect on cNOS activities (NOS-1 was followed in Mastomys (Fig. 4A) and 1C11* /5HT (Fig. 4B) cells; NOS-3 was followed in LM6 fibroblasts; Fig. 4C). Such pharmacological correlations strongly support the occurrence of a direct functional coupling between the 5-HT 2B receptor and cNOS activity.
With the same set of drugs, lower but still significant correlations (r s ϭ 0.643, n ϭ 8, p Ͻ 0.05) were found in the case of iNOS activity in either Mastomys (Fig. 4D) or 1C11* /5HT (Fig.  4E) serotonergic cells. Using LM6 fibroblasts, the correlation did not reach significance (Fig. 4F, r s ϭ 0.571, n ϭ 8, p Ͼ 0.05). We conclude from these observations that coupling of the 5-HT 2B receptor to the iNOS activity most likely follows more complex mechanisms than those involved in the cNOS coupling and that these mechanisms may vary with the cell context.
Activation of cNOS and iNOS Requires the 5-HT 2B Receptor C-terminal PDZ Motif-In the central nervous system, the NMDA receptor-mediated stimulation of NOS-1 occurs via PDZ-PDZ interactions (18). Like the NR 2 subunit of the NMDA receptor, the sequence of the 5-HT 2B receptor polypeptide ends in a type I consensus target sequence for PDZ domain-containing proteins. This led us to search for a possible role of the C-terminal domain of the 5-HT 2B receptor in its coupling to NOS activities.
LMTK Ϫ fibroblasts were thus stably transfected with a 3Јterminal deleted version of the cDNA encoding the 5-HT 2B receptor (LM5 cells). As shown in Fig. 5, upon expression of this truncated 5-HT 2B receptor, DOI-dependent cNOS and iNOS activations were no longer observed. Although deletion of the C-terminal domain of the 5-HT 2B receptor did not impair its functional coupling to PIP 2 hydrolysis (Fig. 1B), it completely abolished DOI-dependent cGMP formation (Fig. 1A). We therefore conclude that, at least in LMTK Ϫ fibroblasts, the C-terminal domain of the 5-HT 2B receptor participates to the recruitment of both the cNOS and iNOS signaling pathways.
We finally examined whether a peptide, the sequence of which overlaps that of the VSYI(D/E) motif at the C terminus of the 5-HT 2B receptor, could interfere with the NOS activities. As shown in Fig. 5, addition to permeabilized cells of a 20amino acid peptide identical to the C terminus of the receptor prevents stimulation by DOI of cNOS and iNOS activities in all three cell types. Together, these results clearly establish that the C-terminal tail of the 5-HT 2B receptor is critical for functional coupling to cNOS and iNOS.
The role of the VSYI(D/E) motif in 5-HT 2B receptor-mediated NOSs activation was further tested by introducing in permeabilized cells various 20-amino acid C-terminal peptides harboring mutated versions of the VSYI(D/E) motif, i.e. VSYV(D/ E), VSFI(D/E), VAYI(D/E), and GSYI(D/E). In all three cell types, DOI-induced stimulation of cNOS and iNOS activities still occurred in the presence of these peptides (Fig. 5). In parallel, LMTK Ϫ cells were transfected with a cDNA encoding a 5-HT 2B receptor in which the tyrosine residue of the VSYI motif was substituted by a serine. DOI stimulation of the resulting LMTK Ϫ -VSSI cells does not trigger NOSs activation (Fig. 5) while producing the same IP 3 response as in LM6 cells (data not shown). These data strongly suggest that the VSYI(D/E) motif of the 5-HT 2B receptor represents the binding site for association with PDZ-containing complexes involved in 5-HT 2B /NOSs couplings. For the first time, an involvement of PDZ interactions in a GPCR signaling pathway is evidenced.
G␣ 13 Is Involved in the 5-HT 2B Receptor-mediated Stimulation of iNOS-The initial evidence for the involvement of heterotrimeric G protein in the regulation of NO production arose from transfection experiments of renal epithelial (MCT) cells by cDNAs encoding various G protein ␣ chains (␣ 13 , ␣ s , ␣ i , and ␣ q/11 ). The only G␣ chain having a marked stimulatory effect on iNOS expression was G␣ 13 (19). To assess a possible role of

5-HT 2B Receptor and Nitric Oxide
G␣ 13 on 5-HT 2B -mediated NO production, each of the three cellular systems were permeabilized and then incubated for 30 min with either anti-G␣ 13 , anti-G␣ q/11 , anti-G␣ s , or anti-G␣ i antibodies, prior to the addition of 100 nM DOI. cNOS and iNOS activities were measured 10 min after the agonist treatment. As shown in Fig. 6 (A and B), the anti-G␣ 13 antibodies fully abolished the DOI-mediated iNOS activation in all cell types while having no effect on cNOS activities. As a control, anti-G␣ 13 antibodies preincubated overnight with the related blocking peptide no longer inhibit iNOS activity (Fig. 6B). Furthermore, antibodies against G␣ q/11 , G␣ s , or G␣ i were without any effect on cNOS (data not shown). These antibodies slightly reduced the iNOS response in LM6 and Mastomys cells but not in 1C11* /5HT cells (Fig. 6C). As positive controls, the inhibitory effect of these antibodies could be verified on 1C11* /5HT cells either on the 5-HT 2B receptor-mediated stimulation of PLC␤ (G␣ q/11 ) or the 5-HT-mediated negative adenylate cyclase coupling of the 5-HT 1B/D receptor (G␣ i ). This set of results demonstrates that neutralization of endogenous G␣ 13 by specific antibodies cancels the iNOS response. It thus confirms that a functional link between G␣ 13 and iNOS activation exists. Eventually, it establishes that G␣ 13 is one of the protagonists participating to the coupling of the 5-HT 2B receptor to iNOS.

DISCUSSION
The present study provides strong biochemical evidence for a functional coupling of 5-HT 2B receptors to the NOS signaling pathways. In the three cell types assayed, i.e. Mastomys carcinoid tumor primary cultures, clonal neural-like 1C11* /5HT serotonergic cells and 5-HT 2B receptor transfected LM6 fibroblasts, the 5-HT 2 receptor agonist DOI (or 5-HT, data not shown) always triggers intracellular cGMP accumulation resulting from both cNOS and iNOS activation. Such a link between 5-HT and NO could already be suspected in view of the simultaneous involvement of these two agents in smooth muscle contractility (6,30), vascular contraction (7,31), and related migraine pathogenesis (32). Because 5-HT 2B receptors are present in stomach, intestine, pulmonary smooth muscles, kidney, as well as in myocardium, vascular endothelium, and meningeal tissues (3,33,34), they are obvious candidates to trigger the NO response associated to 5-HT. However, a direct 5-HT 2B receptor-NO coupling is difficult to evidence in vivo (35), because of the paucity of 5-HT 2B -bearing cells in intact tissues, on the one hand, and the diversity of signals likely to mobilize NOSs responses, on the other hand.
Here, the occurrence of a direct coupling between 5-HT 2B FIG. 4. Comparison of the effects of various agonists and antagonists of the 5-HT 2B receptor measured through inhibition of DOI binding to the receptor (abscissa) and through inhibition of NOS activities (ordinate). Calcium-dependent cNOS activity (A-C) and calcium-independent iNOS activity (D-F) were measured in Mastomys (A and D), 1C11* /5HT (B and E), and LM6 (C and F) extracts. Shown are the apparent pK i or pK d values of drugs in each process. In the case of the agonists and the antagonists, their binding to the receptor was assessed through competition with DOI binding, and the corresponding IC 50 value was determined. K i values were deduced using the Cheng-Prusoff equation (29). The dose-response effect of agonists on NOS activity could be directly followed. In the case of antagonists, effects were estimated through the capacity of each drug to inhibit the DOI-dependent stimulation of NOS activities. Ten different concentrations of each competing drug were used. receptors and cNOS is strongly supported by the highly significant correlations found between the binding constants of agonists and antagonists on the receptor and the dose-response effects of the drugs on cNOS activity. cNOS activation occurs whether 5-HT 2B receptors couple (1C11* /5HT or LM6) or do not couple (Mastomys) to phosphoinositide hydrolysis. It is therefore likely that the recruitment of cNOS by 5-HT 2B receptors does not depend on PLC␤-dependent intracellular Ca 2ϩ mobilization. An indirect effect of calcium ions on the length and/or the amplitude of the cNOS response, depending on 5-HT 2Bmediated PLC␤ activation and involving a Ca 2ϩ /calmodulin complex, cannot, however, be excluded. Indeed, interactions between the IP 3 and NO signaling pathways have already been reported (36,37).
This work also establishes that the cytoplasmic VSYI(D/E) containing C-terminal domain of the 5-HT 2B receptor is involved in the recruitment of NOS activity. 5-HT 2B /NO coupling is fully abrogated either in LM5 fibroblasts stably transfected with a cDNA encoding a C-terminally truncated form of the 5-HT 2B receptor or by disrupting PDZ interactions in all three cell types. The latter effect is obtained through introduction of a competitor peptide corresponding to the 20 C-terminal amino acids of the 5-HT 2B receptor. Versions of this peptide harboring mutations in the VSYI(D/E) motif no longer abolish the 5-HT 2B -mediated cNOS and iNOS activation. Finally, the direct involvement of the VSYI motif could be evidenced in LMTK Ϫ -VSSI fibroblasts through site-directed mutagenesis on the 5-HT 2B receptor VSYI motif itself. These results allow us to extend the role of VSYI(D/E)-PDZ interactions (38) in the organization of signaling complexes to the case of the 5-HT 2B receptor. By making a parallel with the NMDA receptor (18), we may therefore postulate that the VSYI(D/E) motif of the C-terminal cytoplasmic domain of the 5-HT 2B receptor contributes to the assembly of submembrane protein complexes in-volved in the 5-HT 2B /NO coupling.
Although interactions of GPCRs with PDZ-containing proteins have already been reported, notably for the ␤2-adrenoreceptor (see Ref. 39 for review), our study indicates for the first time that VSYI(D/E)/PDZ domain interactions play a role in the coupling of a GPCR (namely the 5-HT 2B receptor) to downstream transductional NOS activities. Such findings may represent a first step in the demonstration that NO participates, in a general manner, to the cellular responses obtained upon activation of 5-HT 2 receptors. In agreement with this idea, stimulations of 5-HT 2A and 5-HT 2C receptors have already been shown to trigger increase in cGMP concentration in C6 glioma cells (40) and in choroid plexus (41), respectively. Although an interaction of these two receptors with NOSs has not yet been described, it is worthwhile to note that type I-PDZ recognition motifs also occur at the C termini of these two proteins (42). Moreover, the possibility that the recruitment of NOSs by these 5-HT 2 receptors modify their coupling with other macromolecular complexes partners has to be considered. Indeed, such networks might explain the 5-HT-mediated inhibition of the NMDA receptor/NO/cGMP pathway in rat cerebellum and the inhibition of NO synthesis in interleukin-1␤-stimulated rat vascular smooth muscle, attributed to the 5-HT 2C (43) and the 5-HT 2A (37) receptors, respectively. Further work is necessary to reach general conclusions. However, the present case of the 5-HT 2B receptor provides new insight for analyzing the physiological role of 5-HT 2 receptors whose localizations (choroid plexus, meningeal tissues, platelets, kidney, lung, etc.) mainly occur at sites of NO production.
PDZ interactions are also involved in the coupling of the 5-HT 2B receptor to the iNOS signaling pathway. Indeed, iNOS activation is impaired in LM5 and LMTK Ϫ -VSSI cells as well as in the three cell types studied here after introduction of the C-terminal 5-HT 2B receptor peptide. Interestingly, the 5-HT 2B receptor-mediated iNOS activation is extremely fast if compared with the length of the iNOS response normally observed during infection or inflammation. A 5-HT 2B -dependent transcriptional regulation of iNOS activity can be excluded because the steady state level of iNOS transcripts is found constant during the cell response. Therefore the possibility that, upon DOI-mediated induction, iNOS is turned on from an inactive state by post-translational controls only, has to be considered (44,45).
This study provides evidence that the ␣ subunit of the G␣ 13 protein plays a critical role in the mechanisms of 5-HT 2B /iNOS coupling. In the three cell types studied, introduction of anti-G␣ 13 antibodies fully inhibits iNOS activation, without interfering with the 5-HT 2B receptor/cNOS signaling. Inhibition of other endogenous G proteins with the help of anti-G␣q 11 , G␣ s , or G␣ i antibodies has no consequence on NOS activities. Thus, although the C-cytoplasmic domain of the 5-HT 2B receptor plays a pivotal role in the recruitment of all three NOS isoforms, the mechanism that couples the receptor to iNOS differs from that which regulates cNOS. In addition, the relatively low significance of pharmacological correlations between the 5-HT 2B receptor binding constants of drugs and the dose-re-FIG. 6. G␣ 13 is involved in the 5-HT 2B receptor-mediated iNOS but not cNOS-mediated coupling in LM6, 1C11* /5HT , and Mastomys cells. Anti-G␣ 13 antibodies do not interfere with the 5-HT 2B receptor-mediated cNOS activation (A) but selectively cancel iNOS activation (B) in the three cell types. The 5-HT 2B receptor-mediated iNOS activation is unaffected upon incubation with anti-G␣ 13 antibodies neutralized with the corresponding blocking peptide (B). Antibodies against G␣ s , G␣ i , or G␣ q have no effect on the iNOS response in 1C11* /5HT cells and slightly reduce iNOS activation in LM6 and Mastomys cells (C). In all cases (A-C), NOS activities were measured 10 min after the addition of 100 nM DOI. White bars, control; shaded bars, anti-G␣ 13 ; hatched bars, anti-G␣ 13 incubated with peptide.

5-HT 2B Receptor and Nitric Oxide
sponse effect of the same drugs on iNOS activity (Fig. 4) suggests that the mechanism coupling the 5-HT 2B receptor to iNOS is more complex than that acting on cNOS. For instance, instead of resulting from a direct 5-HT 2B receptor/G␣ 13 coupling, iNOS activation might integrate cross-talk(s) with other 5-HT 2B receptor transductional pathways.
Knowledge of the signaling pathways mediating 5-HT 2B receptors physiological functions in vivo is of the utmost interest in view of the probable involvement of these receptors in pathological situations such as migraine (32) and hypertension (46). Nevertheless, these pathways may greatly depend on the cell type considered (47), and in vitro cell cultures appear to be an indispensable tool to progressively solve the complexity of 5-HT 2B -coupling mechanisms. For instance, the present study established that in addition to its role in phosphoinositide hydrolysis (4), p21 ras mitogen-activated protein kinase signaling (8), and phospholipaseA 2 /AA release (5), the 5-HT 2B receptor also controls NO/cGMP production through dual activation of cNOS and iNOS. To eventually describe the contribution of such 5-HT-mediated signals to the control of serotonergic functions, corresponding intracellular targets will have to be identified. Hopefully, the availability of the 1C11 cell line, which exhibits all the couplings of the 5-HT 2B receptor described to date in the same integrated serotonergic phenotype, may help dissecting these pathways and their cross-talks.