Role of the C-terminal di-leucine motif of 5-HT1A and 5-HT1B serotonin receptors in plasma membrane targeting.

The 5-HT1A and 5-HT1B serotonin receptors exhibit different subcellular localizations in neurons. Evidence has been reported that the C-terminal domain is involved in the somato-dendritic and axonal targeting of 5-HT1AR and 5-HT1BR, respectively. Here we analyzed the consequences of the mutation of a di-leucine motif and palmitoylated cysteines within this domain. Replacement of I414-I415 by a di-alanine in 5-HT1AR led to endoplasmic reticulum (ER) sequestration of the corresponding mutant expressed in cell lines as well as in hippocampal neurons in culture. Furthermore, di-leucine-mutated receptors were unable to bind 5-HT1A agonists and presented a major deficit in their glycosylation state, suggesting that they are misfolded. By contrast, mutation of the di-leucine motif in the C-terminal domain of 5-HT1BR had no major consequence on its subcellular targeting. However, in the case of the 1ActB chimera (substitution of the C-terminal domain of the 5-HT1BR into 5-HT1AR), this mutation was also found to cause sequestration within the ER. Replacement of palmitoylated cysteines by serines had no consequence on either receptor type. These data indicate that the di-leucine motif of the 5-HT1AR and 5-HT1BR tails is implicated in proper folding of these receptors, which is necessary for their ER export.


Introduction
The 5-HT 1A and 5-HT 1B serotonin receptors are two G-proteincoupled receptors (GPCRs) that exhibit a relatively high degree of homology in their amino acid sequences and share common features. Both are negatively coupled with adenylyl cyclase and act as auto-and heteroreceptors that modulate the activity of numerous neuronal systems (Barnes and Sharp, 1999). However, investigations on their distribution throughout the rat central nervous system showed major differences between these receptors: the 5-HT 1A R is localized on the somas and dendrites of neurons (Kia et al., 1996;Riad et al., 2000), whereas the 5-HT 1B R is found in preterminal unmyelinated axons (Riad et al., 2000;Sari et al., 1999). Interestingly, their neuronal functions depend on these respective localizations. The 5-HT 1A R modulates the firing of neurons (Haj-Dahmane et al., 1991), whereas the 5-HT 1B R participates in a local control of serotonin or other neurotransmitters release from axon terminals in projection areas (for a review, see Sari, 2004).
Recently, we investigated the respective targeting of 5-HT 1A R and 5-HT 1B R by constructing 5-HT 1A R-5-HT 1B R chimeras for the transfection of polarized epithelial Lilly Pork Kidney (LLC-PK1) cells and hippocampal neurons in primary culture (Darmon et al., 1998;Jolimay et al., 2000). These studies showed that the short cytosolic C-terminal tail of both receptors plays a crucial role in their targeting. As also reported for other GPCRs (Bermak et al., 2001;Duvernay et al., 2004;Oksche et al., 1998;Pankevych et al., 2003;Rodriguez et al., 1992;Tai et al., 1999), this region appeared to be necessary for the transport of 5-HT 1A R and 5-HT 1B R to the cell surface, because truncated receptors without the C-terminal domain were sequestrated within the endoplasmic reticulum (ER) in LLC-PK1 cells. Furthermore, these studies also demonstrated that the cytosolic C-terminal tail of 5-HT 1B R contains an axonal-apical targeting signal (Jolimay et al., 2000).
Within the C-terminal domain, the residues I414 and I415 in 5-HT 1A R and L379 and I380 in 5-HT 1B R constitute potential di-leucine motifs (Fig. 1). For numerous integral membrane proteins, such motifs have been shown to play a role in internalization and lysosomal or plasma membrane targeting (Hunziker and Fumey, 1994;Letourneur and Klausner, 1992;Schülein et al., 1998). On the other hand, the C-terminal domains of 5-HT 1A R and 5-HT 1B R contain palmitoylated cysteines (Ng et al., 1993;Papoucheva et al., 2004). In other GPCRs, these residues were found to be involved in several functional aspects, such as membrane targeting and receptor signaling (for a review, see Qanbar and Bouvier, 2003).
In the present study, we investigated the potential role of dileucine motifs and palmitoylated cysteines in the trafficking of 5-HT 1A R and 5-HT 1B R, using site-directed mutagenesis. For this purpose, the di-leucine motifs of 5-HT 1A R (I414-I415) and 5-HT 1B R (L379-I380) were replaced by di-alanine; and palmitoylated cysteines (two in 5-HT 1A R and one in 5-HT 1B R) of the C-terminal domains were mutated to serine. These mutations were also introduced in the chimeras 1ActB and 1BctA, which we used previously to unveil the role of the C- The 5-HT 1A and 5-HT 1B serotonin receptors exhibit different subcellular localizations in neurons. Evidence has been reported that the C-terminal domain is involved in the somato-dendritic and axonal targeting of 5-HT 1A R and 5-HT 1B R, respectively. Here we analyzed the consequences of the mutation of a di-leucine motif and palmitoylated cysteines within this domain. Replacement of I414-I415 by a di-alanine in 5-HT 1A R led to endoplasmic reticulum (ER) sequestration of the corresponding mutant expressed in cell lines as well as in hippocampal neurons in culture. Furthermore, di-leucine-mutated receptors were unable to bind 5-HT 1A agonists and presented a major deficit in their glycosylation state, suggesting that they are misfolded. By contrast, mutation of the di-leucine motif in the C-terminal domain of 5-HT 1B R had no major consequence on its subcellular targeting. However, in the case of the 1ActB chimera (substitution of the C-terminal domain of the 5-HT 1B R into 5-HT 1A R), this mutation was also found to cause sequestration within the ER. Replacement of palmitoylated cysteines by serines had no consequence on either receptor type. These data indicate that the di-leucine motif of the 5-HT 1A R and 5-HT 1B R tails is implicated in proper folding of these receptors, which is necessary for their ER export. 5-HT 1A R and 5-HT 1B R plasma membrane targeting terminal domains (Jolimay et al., 2000). All constructs were used for the transfection of two cell lines (COS-7 and LLC-PK1) and neurons and the localization of expressed proteins was analyzed by immunofluorescence and confocal microscopy. The intracellular localization of the di-leucine mutants was further characterized by double-labeling experiments with ER and Golgi markers. Finally, the characteristics of wild-type and mutated receptors expressed in transfected cells were further investigated by analyzing their glycosylation state, agonist binding properties and coupling to G proteins. Taken together, the data presented here indicate that the di-leucine motif of 5-HT 1A R and 5-HT 1B R tail is implicated in proper folding of the receptor, which is necessary for their ER export.

Results
Di-leucine motif and palmitoylated cysteines contained in the C-terminus of both receptors (Fig. 1) were replaced by alanine and serine residues, respectively, using site-directed mutagenesis. All constructs were tagged by addition of a Flag epitope at their extracellular N-terminus to analyze their subcellular localization.
Di-leucine-mutated 5-HT 1A R is localized in the ER Flag-tagged 5-HT 1A R as well as I414/415A and C417/420S mutants were used to transfect COS-7 ( Fig.  2A,C) or LLC-PK1 (Fig. 2C) cell lines or primary cultures of hippocampal neurons (Fig. 2B,C). Surface labeling was performed by incubating living cells with monoclonal mouse anti-Flag M2 antibody. Transfected cells were then fixed and permeabilized for the subsequent detection of the intracellular receptors using polyclonal rabbit anti-Flag antibody.
As expected, 5-HT 1A R was mostly found at the plasma membrane of each cell type (~55-65% of surface labeling depending on cell type, Fig. 2C). By contrast, the I414/415A mutant was detected only at low levels at the plasma membrane (~10-24% of surface labeling), whereas the amounts of C417/420S mutant at the plasma membrane were very close to those of the wild-type 5-HT 1A R.
Intracellular staining of di-leucine-mutated-5-HT 1A R was distributed in perinuclear ER-like structures. Moreover, in transfected COS-7 cells, we observed a strong co-localization of this I414/415A mutant with the ER luminal marker calregulin but not with the cis and median Golgi marker giantin (Fig. 3A). For these co-localization experiments, cells were treated with the protein synthesis inhibitor cycloheximide before fixation, to lower as much as possible the presence of newly synthesized receptors in ER or Golgi apparatus. Under these conditions, wildtype and C417/420S 5-HT 1A R did not co-localize with calregulin and giantin (Fig. 3A). This result indicates that the I414-I415 motif, but not the palmitoylated cysteines, is necessary for 5-HT 1A R exit from the ER.
We also analyzed the glycosylation state of 5-HT 1A R and related mutants. Western blotting of membrane proteins from transfected LLC-PK1 cells showed that both wild-type and C417/420S 5-HT 1A R migrated mainly as a broad band of ~65 kDa and, to a lesser extent, a thinner band of ~50 kDa (Fig. 3B). By contrast, the I414/415A mutant migrated only as a band of ~50 kDa, which suggests that this construction was not correctly glycosylated. We thus treated membranes with endoglycosidase H (Endo H) to remove high-mannose N-glycosylation or peptide Nglycosidase F (PNGase F) to remove both core and complex  N-glycosylation. The band observed after treatment with PNGase F should correspond to fully deglycosylated receptors. By contrast Endo H is active only on partially glycosylated proteins. After treatment of wild-type and C417/420S 5-HT 1A R with PNGase F, both 50 kDa and 65 kDa bands shifted to a band of ~44 kDa, corresponding to the molecular mass calculated for non-glycosylated Flag-tagged 5-HT 1A R. In the case of digestion with Endo H, the broad band of 65 kDa was still visible, confirming that this band corresponded to a fully glycosylated receptor. Only the thin 50 kDa band (partially glycosylated) was eliminated and converted to the ~44 kDa band. These results showed that the majority of wild-type and C417/420S 5-HT 1A R was completely glycosylated. Concerning the I414/415A mutant, both treatments with Endo H and PNGase F converted the ~50 kDa band into thẽ 44 kDa band, suggesting that this mutant was only partially glycosylated (core glycosylated). These data are consistent with ER retention of the latter mutant, as complex glycosylation occurs only in the Golgi apparatus (Kornfeld and Kornfeld, 1985).
The di-leucine motif is necessary for ligand-binding capacity of 5-HT 1A R We compared the ligand binding capacity of 5-HT 1A R and mutants using the mixed 5-HT 1A -5-HT 1B agonist radioligand, [ 3 H]LSD (Darmon et al., 1998). Membranes of LLC-PK1 cells Journal of Cell Science 119 (20) transfected with wild-type or C417/420S-5-HT 1A R specifically bound equivalent amounts of tritium after incubation with 1.6 nM [ 3 H]LSD (~1 pmol/mg of protein, depending on transfection efficiency). By contrast, membranes of cells transfected with I414/415A mutant specifically bound only a very low amount of the radioligand (Fig. 4A).
Interestingly, deglycosylation of wild-type 5-HT 1A R with PNGase F did not affect its binding capacity under the same assay conditions (not shown). Thus, the reduced [ 3 H]LSD binding capacity of the I414/415A mutant was very probably not caused by its incomplete glycosylation state. Because ligand binding requires correct folding, it can be inferred that the I414-I415 motif is necessary for correct folding of 5-HT 1A R.
Role of the dileucine motif and palmitoylated cysteines in 5-HT 1A R coupling to G proteins We first analyzed the interaction of I414/415A and C417/420S 5-HT 1A R with ␣ subunits of G proteins. As illustrated in Fig.  4B, 5-CT-stimulated [ 35 S]GTP␥S binding onto membranes from LLC-PK1 cells did not statistically differ whether cells were transfected with wild-type or C417/420S 5-HT 1A R. However, this binding was significantly impaired in the case of membranes transfected with I414/415A mutant.
Furthermore, it was shown that 5-HT 1A Rs also interact with G protein ␤␥ subunits to modulate the activity of ERK1/2 (Garnovskaya et al., 1996). We thus tested the ability of 5-HT 1A R and related mutants to activate ERK in transfected LLC-PK1 cells. After treatment with the agonist 8-OH-DPAT for 5 minutes, a ~sixfold increase in ERK2 phosphorylation was observed in cells expressing wild-type or C417/420S 5-HT 1A R (Fig. 4C,D). These results demonstrate that wild-type and C417/420S 5-HT 1A R activate ERK in our experimental conditions. By contrast, 8-OH-DPAT treatment of cells transfected with I414/415A mutant only induced a ~2.7-fold increase in ERK phosphorylation.
Mutations in the C-terminus of 5-HT 1B R have only minor effects on its subcellular localization The percentages of 5-HT 1B R and related mutants at the plasma membrane were also examined. In COS-7 ( Fig. 5A,C) and LLC-PK1 cells (Fig. 5C), wild-type 5-HT 1B R displayed a lower level of surface staining than 5-HT 1A R (COS-7, 22.7±2.7%; LLC-PK1, 30.7±6.0%; P<0.001 compared with data in Fig. 2C for 5-HT 1A R). However, this difference between 5-HT 1A R and 5-HT 1B R did not reach statistical significance in neurons. Replacement of the di-leucine motif in the C-terminus of 5-HT 1B R by alanines (LI379/380A) significantly reduced its amount at the plasma membrane in LLC-PK1 cells but not in COS-7 cells and neurons ( Fig. 5A-C).
As observed in the case of 5-HT 1A R (Fig. 2C), mutation of the palmitoylated cysteine into a serine (C384S) had no significant effect on the subcellular localization of 5-HT 1B R in all cell types analyzed ( Fig. 5A-C).
1ActB chimera reveals the role of the 5-HT 1B R di-leucine motif In a previous study, we substituted the cytosolic tail of 5-HT 1B R into the 5-HT 1A R and vice versa (Jolimay et al., 2000). Analysis of these chimeras expressed in LLC-PK1 cells and in Cells were permeabilized and anti-Flag labeling is shown in green and anti-calregulin or anti-giantin labeling is shown in red. Bar, 10 m. (B) Glycosylation state of receptors was analyzed by western blotting of crude membranes from transfected COS-7 cells with anti-5-HT 1A polyclonal antibody. Membranes were treated with Endo H, PNGaseF or untreated. Similar results were obtained in three independent experiments. 5-HT 1A R and 5-HT 1B R plasma membrane targeting neurons showed that an apical/axonal targeting signal is located in the C-terminus of 5-HT 1B R. The resulting chimeric receptors, 1ActB (5-HT 1A R with C terminus of 5-HT 1B R) and 1BctA (5-HT 1B R with the C-terminus of 5-HT 1A R), were tagged with the Flag epitope and mutants were constructed for both.
On the other hand, 1ActB chimera was mostly localized at the plasma membrane ( Fig. 6D-F), like the 5-HT 1A R (Fig. 2). Di-leucine mutation of the chimera (LI414/415A) resulted in a very low level of plasma membrane localization. By contrast, the amounts of C419S-1ActB mutant at the plasma membrane were very close to those of non-mutated 1ActB (Fig. 6D-F).

Discussion
As with numerous other GPCRs, the cytosolic C-terminal region of 5-HT 1A R and 5-HT 1B R plays an important role in their subcellular localization. Indeed, previous studies showed that this region is necessary for receptor exit from the ER and also that the cytosolic C-terminal tail of 5-HT 1B R contains a dominant axonal-targeting signal (Jolimay et al., 2000). Here, we investigated the potential role of a di-leucine motif and palmitoylated cysteines contained in this receptor domain using site-directed mutagenesis.
The data reported here clearly showed that the di-leucine motif contained in the C-terminal domain of 5-HT 1A R is implicated in its targeting to the plasma membrane. More precisely, this motif plays a crucial role in the correct folding of the receptor, which is necessary for its exit from the ER towards the plasma membrane. The role of the C-terminal dileucine motif of 5-HT 1B R is less clear. Its mutation into a dialanine motif did not modify the localization of the receptor in transfected neurons. However, in the 1ActB chimera, in which the C-terminal tail of 5-HT 1A R has been replaced by the Cterminal tail of 5-HT 1B R, the same di-leucine motif appeared to be implicated in receptor targeting to the plasma membrane.
However, substitution of the palmitoylated cysteine residues with serines did not affect the subcellular localization of receptors as well as chimeras, and in the case of 5-HT 1A R, did not affect receptor binding or coupling to G proteins.
Role of the di-leucine motif of 5-HT 1A R and 5-HT 1B R In GPCRs, di-leucine motifs localized mainly in the Cterminal cytosolic tail were found to be important for targeting. Some of these motifs were shown to act as clathrindependent endocytosis signals (Fan et al., 2001;Gabilondo et al., 1997;Gaudreau et al., 2004;Orsini et al., 1999). In addition, a role in receptor transport from the ER to the plasma membrane was found for a di-leucine motif with an upstream acidic residue in the case of vasopressin V 2 receptor  and, more recently, for a di-leucine motif associated with an upstream phenylalanine residue in the case Basal phosphorylation was determined in the absence of ligand. Equal amounts of cell lysate were separated by SDS-PAGE, blotted and revealed with anti-ERK or anti-PERK antibody (C). PERK2 and ERK2 signals were quantified and ERK phosphorylation is expressed as fold increase over basal levels after normalization with total ERK2 signal (D). ERK phosphorylation in stimulated untransfected (LLC-PK1) cells did not differ from the basal signal. Each bar is the mean ± s.e.m. of three independent experiments. *P<0.05 and ***P<0.001, when compared with the wild type; ns, not significant.
of ␣ 2B -adrenergic and angiotensin II type 1A receptors (Duvernay et al., 2004) or surrounded by three hydrophobic residues in the case of vasopressin V 3 receptor (Robert et al., 2005). Our results concerning the I414-I415 motif in the C-terminal tail of 5-HT 1A R are consistent with these previous findings. Indeed, replacement of this motif by two alanines resulted in a 5-HT 1A R mutant sequestrated within the ER. The loss of [ 3 H]LSD binding capacity of this mutant (29% of wild-type binding) suggests that such sequestration might be caused by an incorrect folding. However, this binding assay was performed with purified membranes of cells expressing the receptors and does not allow distinction of functional characteristics of plasma membrane versus intracellular receptors because ligand could access both pools of receptors. By contrast, for ERK phosphorylation assays, the ligand 8-OH-DPAT was added to living cells, and only plasma-membrane-localized receptors could thus be activated in this protocol. Interestingly, 8-OH-DPATinduced activation of ERK phosphorylation in cells transfected with I414/415A mutant was still ~2.7-fold over basal levels, corresponding to ~45% of the increase observed for the wild-type receptor. We therefore propose that the observed decrease in ERK activation is most probably due to intracellular sequestration of a large amount of this mutant and that the residual plasma-membrane-localized I414/415A 5-HT 1A Rs are functional. This would support the idea of an incorrect folding of sequestrated mutants, as receptors that can exit the ER and reach the plasma membrane seem to be functional, implying their correct folding. However, we cannot entirely exclude that the di-leucine motif of 5-HT 1A R may also participate in receptor ER exit by interaction with COP-II-associated proteins (for a review, see Barlowe, 2003), and that defective interactions caused by the mutation were actually responsible for ER sequestration of the I414/415A mutant.
As noted by Schülein et al.  and Duvernay et al. (Duvernay et al., 2004), this dileucine motif is highly conserved in the C-terminal tail of GPCRs, suggesting a general role in the exit from ER for most of these membrane proteins. However, in our study, replacement of the di-leucine motif by two alanines in the Cterminal tail of 5-HT 1B R only affected its subcellular localization in LLC-PK1 cells, as this mutant did not differ from the wild-type 5-HT 1B R regarding its targeting to the plasma membrane in both COS-7 cells and hippocampal neurons. This apparent discrepancy is not unique among GPCRs. Indeed, mutation of the two conserved leucines in the ␤ 2 -adrenergic receptor, which strongly diminished receptor endocytosis, did not affect its targeting to the plasma membrane and its capacity to bind specific ligands (Gabilondo et al., 1997). Therefore, the implication of the two leucines (or isoleucines) localized in the cytosolic C-terminal tail of GPCRs in the exit of the receptor from the ER may not be universal, but would depend on its environment or be associated with other signals. Concerning the 5-HT 1B R, its predominant intracellular localization found in most cell types tested could also explain why substitution of the di-leucine motif did Journal of Cell Science 119 (20) not generally produce further detectable intracellular sequestration.
To further address this question, we constructed chimeras in which the C-terminal domains of 5-HT 1A R and 5-HT 1B R were switched (Jolimay et al., 2000). In transient transfection experiments, 1BctA chimera (5-HT 1B R core with 5-HT 1A R Ctail) exhibited nearly exclusive perinuclear localization, indicating that this construct is probably not functional. On the other hand, a relatively high proportion (~50%) of 1ActB chimera was localized at the plasma membrane (Fig. 5F), like that observed with the wild-type 5-HT 1A R (Fig. 2C). Accordingly, it can be inferred that the di-leucine motif of the C-terminal domain of 5-HT 1B R allows correct folding of this 1ActB chimera and its exit from ER. This would suggest the need for another signal localized in a different domain of the receptor in addition to the di-leucine motif of the C-terminal tail, which would be present in the 5-HT 1A R but not in the 5-HT 1B R, thereby leading to plasma membrane targeting of the 1ActB but not the 1BctA chimera. Alternatively, it is possible Role of palmitoylated cysteines 5-HT 1A R and 5-HT 1B R were shown to contain palmitoylated cysteines in their C-terminal domain (Ng et al., 1993;Papoucheva et al., 2004). Substitution of these residues with serines did not affect the subcellular localization of either receptors or chimeras. Furthermore, we found that the glycosylation state and the ligand-binding capacity of 5-HT 1A R were not dependent on the presence of the palmitoylated cysteines. These data are consistent with results obtained by Papoucheva et al. (Papoucheva et al., 2004) who recently demonstrated the constitutive palmitoylation of cysteines 417 and 420 of this receptor. These authors also showed that palmitoylated cysteines 417 and 420 are necessary for the receptor coupling to G ␣i3 subunit in transfected insect Sf.9 cells, as well as for its capacity to inhibit adenylyl cyclase activity in NIH3T3 cells and to activate ERK in CHO cells. By contrast, using [ 35 S]GTP␥S binding and ERK activation assays, we showed here that the mutation of both palmitoylated cysteines 417 and 420 did not significantly affect 5-HT 1A R coupling with G ␣ proteins and activation of ERK in LLC-PK1 cells. Such discrepancies might be explained by the use of different cell lines expressing different G ␣i subunits and other signaling molecules, in our studies compared with those of Papoucheva et al. (Papoucheva et al., 2004). First, in Sf.9 cells, the G ␣i3 subunit was cotransfected with receptors. In LLC-PK1 cells, both G ␣i2 and G ␣i3 subunits are endogenously expressed but exhibit different subcellular compartmentations: G ␣i2 is localized at the basolateral membrane, whereas G ␣i3 is restricted to the Golgi apparatus (Ercolani et al., 1990). As we found that 5-HT 1A R is mainly localized at the plasma membrane (Fig. 2), it should interact primarily with G ␣i2 in LLC-PK1 cells. Thus, [ 35 S]GTP␥S binding results agreed with the hypothesis that the palmitoylated cysteines 417 and 420 are necessary for 5-HT 1A R coupling with the G ␣i3 but not the G ␣i2 subunit. In the case of ERK phosphorylation, the differences between cell lines tested is less clear, because CHO cells express both G ␣i2 and G ␣i3 subunits. However, multiple signaling pathways can lead to the activation of ERK after stimulation of a particular GPCR, and some of these pathways may be activated independently of G proteins (for a review, see Werry et al., 2005). In the case of the 5-HT 1A R, the intracellular signaling pathway leading to ERK activation has been shown to implicate G ␤␥ subunits in CHO cells (Garnovskaya et al., 1996;Della Rocca et al., 1999). However, to date, it is not known whether the same pathway is involved in other cell lines, such as LLC-PK1 cells. It would thus be of interest to identify which intracellular signaling molecules contribute to ERK activation by 8-OH-DPAT in LLC-PK1 cells. In any case, these results suggest that palmitoylated cysteines play variable roles in 5-HT 1A R functional characteristics, depending on the cell type and the signaling molecules available. Such differences were already reported for other GPCRs. Thus, substitution of palmitoylated cysteines by alanines in the A subtype of the endothelin receptor (ET A R) has been shown to reduce its capacity to activate G i and G q proteins without affecting its capacity to activate G o protein (Doi et al., 1999).
In conclusion, our data show that the di-leucine motif in the C-terminal domain of 5-HT 1A and 5-HT 1B receptors is necessary for their ER sorting through its implication in the proper folding of receptors. They also provide further support for the statement that 5-HT 1A and 5-HT 1B receptors are routed through distinct intracellular pathways towards their final targeting in neurons.

Antibodies
Anti-rat 5-HT 1A R antibody has been described previously (El Mestikawy et al., 1990;Kia et al., 1996;Riad et al., 2000). This polyclonal rabbit antibody is directed against a peptide sequence located within the third intracellular domain of the receptor. Mouse anti-Flag M2 monoclonal antibody, rabbit anti-Flag polyclonal antibody and mouse monoclonal anti-diphosphorylated-ERK1/2 (PERK) were purchased from Sigma (St Louis, MO). Rabbit polyclonal anti-ERK1/2 (ERK) was purchased from Upstate Biotechnology (Charlottesville, VA). Rabbit polyclonal anti-calregulin antibody was purchased from Santa Cruz Biotechnology (Santa Cruz, CA) and rabbit polyclonal anti-giantin antibody from CRP (Berkeley, CA).
Neuronal cultures were made as described previously (Goslin et al., 1998) with some modifications. Hippocampi of rat embryos were dissected at day 17-18. After trypsinization, tissue dissociation was achieved with a Pasteur pipette. Cells were counted and plated on poly-L-lysine-coated 12-mm-diameter coverslips, at a density of 60,000-75,000 cells per 16-mm dish (300-375 cells per square millimeter), in complete Neurobasal medium supplemented with B27 (Invitrogen), containing 0.5 mM L-glutamine, 10 U/ml penicillin G, and 10 mg/ml streptomycin. Five hours after plating, the coverslips were transferred to a 90-mm dish containing conditioned medium obtained by incubating glial cultures (70-80% confluency) for 24 hours in the complete medium described above. Experiments were performed in agreement with the institutional guidelines for use of animals and their care, in compliance with national and international laws and policies (Council directives no. 87-848,

Cell transfections
For immunofluorescence experiments, LLC-PK1 and COS-7 cells were transferred to 12-mm-diameter coverslips 16 hours before transfection to obtain 30-50% confluency cultures. LLC-PK1 cells were transfected using Lipofectin reagent (Invitrogen). For each coverslip, 1 g plasmid DNA and 1-3 l Lipofectin were both diluted separately in 125 l serum-free DMEM. After a 30-45 minute incubation at room temperature, the two dilutions were combined and the resulting mix was left for another 10-15 minutes at room temperature. Cells were washed with 500 l serum-free DMEM and mix was added for an overnight incubation at 37°C. COS-7 cells were transfected using FuGENE reagent (Roche, Meylan, France). For each coverslip, 2 l FuGENE were diluted in 50 l D-PBS (Dulbecco's phosphate-buffered saline, Invitrogen) and incubated for 5 minutes at room temperature. The dilution was then mixed with 1 g plasmid DNA, and incubation proceeded for another 15 minutes. This mix was added to the growth medium (250 l) overlaying the cells and transfection lasted 24 hours at 37°C.
Hippocampal neurons were transfected on the 7-8th day in vitro as follows: for each coverslip, plasmid DNA (0.8 g) was mixed with 50 l Neurobasal medium without B27 supplement. After 15 minutes at room temperature, 0.8 l Lipofectamine 2000 (Invitrogen) in 50 l Neurobasal medium were added and incubation continued for another 20 minutes. After the addition of 150 l of complete Neurobasal medium containing B27 supplement, the mix was applied onto the neuronal culture, and transfection lasted for 3 hours at 37°C. Typically, 5-10% of neurons expressed the receptors after transfection.
For both cell lines and hippocampal neurons, receptor expression was allowed in growth medium for 24 hours after transfection.
For preparation of membranes and ERK phosphorylation assays (see below), LLC-PK1 cells were transfected by electroporation using Gene Pulser Xcell electroporation system (Bio-Rad, Hercules, CA; 135 V, 1800 F in 200 l DMEM containing 5-10ϫ10 6 cells and 5-10 g plasmid DNA; relaxation time: ~40 milliseconds). Cells were then transferred to a 90-mm dish and grown for 3 days in LLC-PK1 growth medium.

Indirect immunofluorescence
Cells on coverslips were washed with D-PBS+ (D-PBS containing 0.1 mM CaCl 2 and 0.1 mM MgCl 2 ) at 37°C, then fixed with paraformaldehyde (3%) containing 4% sucrose at 37°C in D-PBS+, and permeabilized with 0.1% Triton X-100 in D-PBS+. After two 10-minute washes in D-PBS+, cells were incubated for 30 minutes in antibody buffer (3% bovine serum albumin, 2% normal goat serum, 2% normal donkey serum in D-PBS+). Incubation with primary antibodies was then performed in antibody buffer for 1 hour at room temperature. After two 10-minute washes in D-PBS+, incubation with secondary antibodies proceeded for 1 hour. The secondary antibodies used were Cy3-conjugated donkey anti-rabbit IgG (1:1600 dilution; Jackson ImmunoResearch, West Grove, PA), and Alexa Fluor 488-conjugated goat anti-mouse IgG (1:1600; Molecular Probes, Eugene, OR). The coverslips were finally mounted in Fluoromount-G solution (Clinisciences, Montrouge, France). For ER and Golgi co-localization experiments, cells were treated with the protein synthesis inhibitor cycloheximide (70 M) for 4 hours before fixation, to lower as much as possible the presence of newly synthesized receptors in ER or Golgi apparatus. ER and Golgi labeling was performed using anti-calregulin antibody (1:100 dilution) and anti-giantin antibody (1:2000 dilution), respectively, and receptors were labeled using anti-Flag M2 monoclonal antibody. For surface detection, anti-Flag M2 antibody (2.5 g/ml) was incubated with living cells for 20 minutes at room temperature. Cells were washed in D-PBS+, fixed with paraformaldehyde (3%) containing 4% sucrose, and incubated for 1 hour with Alexa Fluor 488-conjugated goat anti-mouse IgG in antibody buffer. After permeabilization with 0.1% Triton X-100 in D-PBS+, intracellular epitopes were detected using rabbit anti-Flag polyclonal antibody (0.85 g/ml) subsequently revealed by Cy3-conjugated donkey anti-rabbit IgG.
Immunofluorescence images were generated using a Leica laser-scanning confocal microscope. For relative surface label analysis, unsaturated acquisitions were made with the same exposure settings and laser gain for each condition. For each cell type, at least ten cells were analyzed. Quantification of surface and intracellular staining were performed using ImageJ software (NIH) according to Jaskolski et al. (Jaskolski et al., 2004) with modifications, and statistical analysis was carried out using GraphPad Prism 4 (GraphPad Software, San Diego, CA). Background was lowered using Gaussian blur (radius 1 pixel) and an intensity threshold was fixed just above the background level to maximally reduce nonspecific staining. Single cells were selected and carefully traced manually. Surface (S) and intracellular (I) areas with labeling above threshold were measured and the percentage of surface receptor labeling calculated as Sϫ100/(S+I).
Contrast and brightness of images displayed in figures were modified using Adobe Photoshop 7.0 (Adobe Systems, San Jose, CA) for clearer demonstration and do not correspond to the analysis conditions. 5-HT 1A R and 5-HT 1B R plasma membrane targeting

Preparation of membranes
Transfected LLC-PK1 cells were washed with D-PBS, scraped into Tris buffer (50 mM Tris-HCl, pH 7.4), and homogenized with a Polytron. After each of four successive washings in Tris buffer, the membranes were collected by centrifugation at 31,000 g for 20 minutes at 4°C. An incubation for 10 minutes at 37°C was performed after the first washing to eliminate 5-HT (from the serum in the culture medium), and the final pellet was suspended in the same Tris buffer to be stored at -80°C until use. Protein concentration was measured using BCA protein assay kit (Pierce, Rockford, IL).
For membranes subsequently used for binding assays, deglycosylation was done without denaturation.

Radioligand binding assays
Binding assays were performed using 20-25 g membrane proteins in 500 l of 50 mM Tris-HCl buffer, pH 7.4, supplemented with 1.6 nM [ 3 H]lysergic acid diethylamide ([ 3 H]LSD; 79.2 Ci/mmol; Amersham Biosciences). Incubations were performed for 90 minutes at 25°C. Non-specific binding was determined in the presence of 10 M 5-HT. Assays were stopped by rapid filtration through Whatman GF/B filters coated with polyethylenimine (0.5% v/v). Subsequent washing and counting of entrapped radioactivity were as described by Fabre et al. (Fabre et al., 1997). Specific binding is expressed as a percentage of wild-type receptor specific binding. Data were corrected for individual variations in transfection efficiency by relative quantification of receptors as detailed in the deglycosylation procedures. Data analysis was done using GraphPad Prism 4.

[ 35 S]GTP-␥-S binding assays
Binding of [ 35 S]GTP␥S onto transiently transfected LLC-PK1 cell membranes stimulated by 5-carboxamido-tryptamine maleate (5-CT; Sigma) was measured according to a procedure adapted from Alper and Nelson (Alper and Nelson, 1998) and Fabre et al. (Fabre et al., 2000). Briefly, membranes (~40-50 g protein) were incubated for 30 minutes at 37°C in a final volume of 800 l assay buffer (50 mM Tris-HCl, 3 mM MgCl 2 , 120 mM NaCl, 0.2 mM EGTA) containing 0.1 nM [ 35 S]GTP␥S (1000 Ci/mmol, Amersham Biosciences), 300 M GDP and 1 M 5-CT. The reaction was terminated by addition of 3 ml ice-cold 50 mM Tris buffer and rapid vacuum filtration through Whatman GF/B filters. Each filter was then washed twice with 3 ml ice-cold Tris buffer, placed into 4.5 ml scintillation fluid and its entrapped radioactivity measured. Basal [ 35 S]GTP␥S binding was determined from samples without 5-CT, and non specific binding from those supplemented with both 5-CT and WAY 100,635 (1 M), a selective 5-HT 1A R antagonist (Fabre et al., 2000).

ERK phosphorylation assays
Electroporated LLC-PK1 cells were transferred to 35-mm dishes 24 hours after transfection and grown for another 10 hours in LLC-PK1 growth medium. Cells were then starved for at least 8 hours in serum-free medium and subsequently treated with 1 M 8-OH-DPAT (8-hydroxy-N,N-dipropyl-2-amino-tetralin, Sigma) for 5 minutes at 37°C under 5% CO 2 . Cells were then washed in ice-cold D-PBS and lysed in 50 l sodium dodecyl sulphate sample buffer (Laemmli, 1970). For each transfection, an equal portion of the cells was set aside for protein determination with BCA kit. Equal quantities of extracts from cells exposed to 8-OH-DPAT or none (controls) were separated by electrophoresis, transferred to nitrocellulose, and probed with rabbit anti-ERK (1:10,000 dilution) or mouse anti-PERK (1:2500 dilution) antibodies. After incubation with anti-rabbit (1:10,000 dilution, Sigma) or anti-mouse (1:2500 dilution, Sigma) antibodies coupled to horseradish peroxidase, revelation was performed with the ECL+ kit. ERK-2 and PERK2 bands were quantified as described in deglycosylation section. PERK2/ERK2 ratio was calculated for each sample for normalization and data were expressed as the fold increase over basal levels. Data analysis was done using GraphPad Prism 4.

Statistical analyses
For relative surface label analysis, statistical significance was assessed using a oneway ANOVA followed by a Bonferroni test. For [ 3 H]LSD and [ 35 S]GTP␥S binding assays as well as for ERK activation assays, differences were evaluated using unpaired and paired Student's t-tests, as appropriate. The significance level was set at P<0.05.