Differential Internalization of Mammalian and Non-mammalian Gonadotropin-releasing Hormone Receptors UNCOUPLING OF DYNAMIN-DEPENDENT INTERNALIZATION FROM MITOGEN-ACTIVATED PROTEIN KINASE SIGNALING*

Sustained stimulation of most G-protein-coupled receptors (GPCRs) 1 causes their desensitization and internalization from the cell surface. For many GPCRs this is mediated by G-protein receptor kinases, which phosphorylate the receptor and pro-mote the binding of (cid:1) -arrestin and a consequent reduction in coupling of the receptor to its heterotrimeric G-proteins (1). (cid:1) -arrestin also serves as an adapter, targeting the desensitized receptor to clathrin-coated vesicles (CCVs) for internalization (2). Once formed, the CCV is “pinched” off from the plasma membrane by an oligomeric dynamin “collar”. This effect of dynamin is dependent upon its intrinsic GTPase and can be blocked by expression of GTPase-inactive (dominant negative) mutants such as K44A dynamin (3). The internalized receptors are then either recycled back to the surface membrane or targeted to lysosomes for degradation so that agonist-induced GPCR internalization can reinforce desensitization or underlie resensitization

mitogen-activated protein (MAP) kinase activation by GPCRs (13,14), an effect that could reflect a requirement of GPCR internalization for MAP kinase activation (15) or a requirement for dynamin-dependent internalization of downstream proteins such as MAP/ERK kinase (MEK) (16). Indeed, since transactivation of the epidermal growth factor (EGF) receptors (17)(18) or release of EGF (19) can underlie GPCR-mediated MAP kinase activation, the dynamin dependence of this effect could also reflect the known dynamin dependence of tyrosine kinase receptor-mediated MAP kinase activation (3).
The gonadotropin-releasing hormone receptor (GnRH-R) is a phospholipase C-coupled GPCR that mediates neuroendocrine control of gonadotropin secretion from pituitary gonadotropes and thereby plays a central role in control of reproduction. GnRH-Rs have a number of atypical structural characteristics (20), notably the unique absence of C-terminal tails in type I mammalian GnRH-Rs. In contrast, all cloned non-mammalian GnRH-Rs possess such tails and, since the serine and threonine residues that are phosphorylated by GRKs are often found within the C-terminal tail, comparative studies have addressed the functional relevance of these structures. These have revealed that mammalian GnRH-Rs do not show rapid homologous desensitization (21)(22), whereas non-mammalian GnRH-Rs do rapidly desensitize (23) and that, although mammalian GnRH-Rs undergo agonist-induced internalization via CCVs, they are internalized much more slowly than nonmammalian GnRH-Rs. Moreover, where investigated, nonmammalian GnRH-Rs show agonist-induced phosphorylation and ␤-arrestin translocation and undergo ␤-arrestin-dependent internalization (24), whereas mammalian GnRH-Rs do not (25). Using recombinant adenovirus expressing human and type I Xenopus GnRH-Rs (Ad hGnRH-R and Ad XGnRH-R) we have recently shown that these functional distinctions (desensitization and internalization) are retained when GnRH-Rs are expressed at physiological density and in gonadotrope lineage cells (26). This supports the notion that GnRH-Rs have undergone a relatively recent period of accelerated molecular evolution that is functionally relevant and pertinent to the development of mammalian reproductive strategies (27).
Here, we have explored the dynamin dependence of GnRH-R internalization and MAP kinase signaling using recombinant adenovirus to express human and Xenopus GnRH-Rs in HeLa cells conditionally expressing wild-type or dominant negative (K44A) dynamin under tetracycline control (tet-OFF). Both receptors were positively coupled to PLC and MAP kinase (ERK 2 phosphorylation) and, as anticipated, the non-mammalian GnRH-Rs rapidly desensitized and internalized, whereas these processes were slower or absent for the mammalian GnRH-Rs. Both receptors were internalized via CCVs (as indicated by sucrose dependence) but internalization of the XGnRH-R was primarily dynamin-dependent, whereas that of the hGnRH-R was not. MAP kinase activation was dynamindependent for both receptors, was reduced by inhibition of internalization or of PKC activity, and was not inhibited by an EGF receptor, tyrophostin. Thus, desensitizing and non-desensitizing GnRH-Rs undergo functionally distinct forms of internalization and show dynamin-dependent signaling that does not reflect internalization of the GPCR or EGF receptor transactivation.

EXPERIMENTAL PROCEDURES
Materials and Cell Culture-GnRH and chicken GnRH II (cGnRH-II) were purchased from Peninsula Laboratories Europe Ltd. (Mersyside, United Kingdom) or from Sigma. Buserelin and 125 I-buserelin (2000 Ci/mmol) were provided by Prof. Sandow (Aventis Pharma GmbH, Frankfurt, Germany). 125 I-cGnRH-II (ϳ3400 Ci/mmol as determined by self-displacement) was prepared using chloramine T and purified by G25 Sephadex column chromatography. 125 I-EGF was purchased from PerkinElmer Life Sciences. AG1478 was purchased from Calbiochem. Culture media, sera, and plasticware were from Life Technologies, Inc. or Falcon (Becton Dickinson, Oxford, UK). Lipofectin, LipofectAMINE, and plus reagent were from Life Technologies, Inc., and FuGENE 6 was from Roche Molecular Biochemicals. 2-[ 3 H]inositol (14 -16 Ci/mmol) was from Amersham International (Little Chalfont, UK). Recombinant adenovirus expressing the human GnRH-R (Ad hGnRH-R), the Xenopus type I GnRH-R (Ad XGnRH-R) and EGFP (Ad EGFP) were generated as previously described (26), and CMV-EGFP was from CLON-TECH. All other reagents were from standard commercial suppliers. HeLa cells stably expressing either the wild-type (WT dynamin cells) or the dominant negative GTPase-deficient mutant dynamin (K44A dynamin cells) (kindly provided by Dr. S. Schmidt, Scripps Institute, La Jolla, CA) were cultured in serum-supplemented Dulbecco's modified Eagle's medium with G418 (300 g/ml), puromycin (100 ng/ml), and tetracycline (1 g/ml) as described (3). For experiments cells were harvested by trypsinization, plated in Dulbecco's modified Eagle's medium supplemented with 2% serum, and incubated for 2 days in flasks or culture plates as described in figure legends. Cells were infected on the second day, and where appropriate the tetracycline was removed. After 8 h, the medium was replaced with fresh medium, with or without tetracycline.

Accumulation of [ 3 H]Inositol phosphates ([ 3 H]IP)-[ 3 H
]IP accumulation was used as a measure of PLC activity as described (28) using cells labeled by pre-incubation with [ 3 H]inositol and stimulated in the presence of LiCl. Cells were cultured in 24-well plates in 1 ml of media, and 2 Ci 2-[ 3 H]inositol (14 -16 Ci/mmol) was added to each well for the final 16 h of incubation. After two washes with PSS (127 mM NaCl, 1.8 mM CaCl 2 , 5 mM KCl, 2 mM MgCl 2 , 0.5 mM NaH 2 PO 4 , 5 mM NaHCO 3 , 10 mM glucose, 0.1% (w/v), bovine serum albumin and 10 mM HEPES, pH 7.4), each well was stimulated for the period indicated in the figures with 200 l of PSS containing 10 mM LiCl and the indicated concentration of buserelin, GnRH, or cGnRH-II. The stimulation was terminated by adding 1 ml of water at 95°C. The cells were lysed by freezing and thawing and [ 3 H]IP was separated from [ 3 H]inositol using anion exchange chromatography in formate form Dowex-1 columns, and the amount of 3 H eluted in each fraction was determined by liquid scintillation spectroscopy (28).
Radioligand Binding and Internalization Assays-GnRH-R expression was assessed and competition curves were constructed using whole cell binding assays in which ϳ50,000 cells were incubated in suspension for 30 min at 21°C in 100 l of PSS containing 1 mg/ml bacitracin with ϳ10 Ϫ10 M radiolabel and 0 or 10 Ϫ10 to 10 Ϫ5 M of the unlabeled competitor peptide (26). Free and bound peptide were then separated by centrifugation through oil. For the human GnRH-R, the radiolabel was 125 I-buserelin and for Xenopus GnRH-R the radiolabel was 125 I-cGnRH-II. Similar assays were performed to quantify cell surface EGF receptors. Approximately 50,000 cells were incubated in suspension for 3 h at 4°C in 100 l PSS containing 1 mg/ml bacitracin with ϳ0.2 nM 125 I-EGF and 0 or 10 Ϫ6 M of unlabeled EGF. Free and bound peptides were then separated by centrifugation through oil.
Internalization of EGF receptors was quantified in a modified whole cell binding assay in which ϳ250,000 cells were grown in 6-well plates. Cells were washed in PSS and then incubated at 37°C in 1 ml of PSS containing ϳ10 Ϫ7 M EGF. After the required incubation period (2-45 min) the cells were rapidly rinsed in ice-cold PSS (two times) and then incubated for 5 min in ice-cold 150 mM NaCl,50 mM acetic acid (pH 3) before scraping, pelleting, and resuspending in PSS. Internalization of GnRH-R was quantified in a modified whole cell binding assay in which ϳ50,000 cells (grown in 24-well plates) were washed in PSS and then incubated at 37°C in 200 l of PSS containing ϳ10 Ϫ10 M radiolabel and 0 (total binding) or 10 Ϫ6 M (nonspecific binding) of buserelin or cGnRH-II. After the required incubation period (5-30 min) the cells were rapidly rinsed in ice-cold PSS (two times) and then incubated for 2 min either in ice-cold PSS or in ice-cold 150 mM NaCl,50 mM acetic acid (pH 3). The cells were then washed three more times in ice-cold PSS and solubilized in 0.5 ml of 0.2 M NaOH with 1% SDS. Radiolabel in the solubilized cells was determined by ␥ counting, and specific cell-associated radioactivity was determined by subtraction of nonspecific from the total. Total specific binding is defined as the specific binding in cells receiving no acid wash, whereas acid-resistant (internalized)-specific binding is defined as that seen in the acid washed cells. An internalization index was calculated by expressing acid-resistant-specific binding as a percentage of total cell-associated-specific binding.
Fluid phase endocytosis was determined by measuring uptake of horseradish peroxidase as described (29). Briefly, cells were cultured at 250,000 cells/well in 6-well plates and infected with Ad hGnRH-R or Ad XGnRH-R (m.o.i. 100) as above. They were then incubated for 60 min at 37°C in 0.5 ml of Dulbecco's modified Eagle's medium containing 5 mg/ml horseradish peroxidase (Sigma) and 0 or 10 Ϫ7 M GnRH or cGnRH-II. Uptake was terminated by repeated washing in ice-cold phosphate-buffered saline prior to cell lysis with 0.1% Triton X-100 in phosphate-buffered saline. Horseradish peroxidase activity was then determined using 3,3-diaminobenzidine as a substrate followed by colorimetric measurement at 450 nm.
Dynamic Video Imaging of Cystolic Ca 2ϩ -Cytosolic Ca 2ϩ concentration was measured by dynamic video imaging using MagiCal hardware and Tardis software with a Nikon Diaphot microscope with ϫ40 quartz oil immersion objective (30). Cells were cultured on glass coverslips in flat-bottomed 12-well plates and were washed and then incubated for 30 min at 37°C with PSS containing 2 M fura-2 acetoxmethyl ester (fura-2/AM). Immediately before imaging the coverslip was dipped into PSS, blotted, and mounted on a greased incubation chamber. The cells were kept immersed in media constantly. Media changes were carried out by pipetting 4 ml of solution into the chamber, while an aspiration tube removed excess liquid and maintained a constant volume of around 400 l. The samples in the chamber were illuminated alternately at 340 and 380 nm, and the ratio of the light emitted at 510 nm was measured, averaging 16 video frames to minimize the pixel to pixel standard error. This ratio was proportional to the intracellular ionized Ca 2ϩ . For calibration, a dissociation constant of 225 nM for fura-2 and Ca 2ϩ at 37°C was used, and minimal and maximal fluorescence at 340 and 380 nm were determined in cells loaded with dye and made permeable to extracellular Ca 2ϩ with 10 M ionomycin in solutions containing either 10 mM Ca 2ϩ or 10 mM EGTA.
Measurement of ERK 2 Activation by Western Blotting-Activation of ERK 2 was measured by Western blotting according to standard techniques. Briefly HeLa cells were plated in 6-well plates at 250,000 cells/well, infected with either Ad XGnRH-R or Ad hGnRH-R at an m.o.i. of 50 -100, and left to express for 24 h. Following treatment, the cells were washed twice in ice-cold phosphate-buffered saline before being lysed on ice for 10 min in 400 l of extraction buffer (10 mM Tris, pH 7.6, 5 mM EDTA, 1 mM EGTA, 50 mM NaCl, 30 mM sodium pyrophosphate, 50 mM NaF, 1 mM dithiothreitol, 100 M sodium orthovanadate, 1% Triton X-100, 1 mM phenylmethylsulfonyl fluoride, 10 g/ml antipain, 2 g/ml leupeptin, 2 g/ml pepstatin). Samples were then centrifuged (13,000 ϫ g, 4°C, 10 min) to clear the supernatant before 40 l of aliquots were boiled with 40 l of sample buffer. Proteins were separated by SDS-polyacrylamide gel electrophoresis (8% gel), transferred to polyvinylidene difluoride membrane and blocked with 5% skimmed milk/Tris-buffered saline Tween. ERK 2 was detected using monoclonal anti-ERK 2 (Santa Cruz) and visualized using enhanced chemiluminescence (ECL, Amersham Pharmacia Biotech). The unphosphorylated and phosphorylated forms of ERK 2 were distinguished by retardation of the latter in SDS-polyacrylamide gel electrophoresis. Both bands were scanned and quantified by densitometry (Quantity 1 Software, Bio-Rad), and the phosphorylated ERK 2 was expressed as a percentage of total ERK 2 (both bands).
Statistical Analysis and Data Presentation-The figures show the mean Ϯ S.E. of data pooled from "n " independent experiments (raw data or data normalized as described in the figure legends). Data are typically reported in text as mean Ϯ S.E., and statistical analysis was by analysis of variance and Student's t test, accepting p Ͻ 0.05 as statistically significant. EC 50 values were estimated by non-linear regression using "Graphpad Prism" (Graphpad Prism, San Diego, CA).

RESULTS
To explore the dynamin dependence of GnRH-R desensitization, internalization, and signaling, we sought to express human and Xenopus GnRH-Rs in HeLa cells expressing wild-type or K44A dynamin under control of the tetracycline-response element and a tetracycline-controlled transactivator. However, transfection efficiencies were found to be extremely low in these cells when conventional strategies (CaPO 4 , Lipofectin, LipofectAMINE Plus, and FuGENE) were used. Using flow cytometry to assess protein expression after transfection with a CMV-EGFP plasmid we estimated that Ͻ20% of cells were transfected. In contrast transfection efficiencies approaching 100% were obtained using recombinant EGFP-expressing adenovirus (not shown). We therefore made use of previously generated recombinant adenovirus expressing human and Xenopus laevis type I GnRH-Rs (Ad hGnRH-R and Ad XGnRH-R, respectively).
Since these receptors have not (to our knowledge) been previously expressed in HeLa cells, it was necessary to first perform a basic pharmacological characterization. This was undertaken using WT dynamin cells maintained in medium with tetracycline (conditions not permissive for transgene expression). To determine the relationship between viral titer and receptor expression we constructed competition binding curves using 125 I-buserelin and varied amounts of unlabeled buserelin. No specific binding was seen in untransfected cells, but infection with Ad hGnRH-R with increasing titer (from m.o.i. of 3-100) increased 125 I-buserelin binding, and this was inhibited in a concentration-dependent manner by unlabeled buserelin. Fitting this data to a single site competition model revealed no dependence of K d value on viral titer, so B max values were estimated by refitting the data with the K d fixed at the mean value of 2.4 nM (2.4 nM Ϯ 0.3 nM, n ϭ 3). This revealed receptor densities of 3000, 14,600, 25,500, and 46,500 sites/cell at m.o.i. values of 3, 10, 30, and 100, respectively (Fig. 1). Similar experiments with 125 I-cGnRH-II and unlabeled cGnRH-II in cells infected with Ad XGnRH-R again revealed high affinity binding sites with no relationship between Ad titer and K d . The data was therefore fitted through the mean K d value of 3.1 nM (3.1 nM Ϯ 0.9 nM, n ϭ 3). This revealed receptor densities of 6600, 17,000, 49,000, and 215,000 sites/cell at m.o.i. values of 3, 10, 30, and 100, respectively (data not shown).

H]IP accumulation was measured in cells infected with Ad hGnRH-R at varied titer. No stimulation was seen in untransfected cells or in cells infected
Since activation of PLC mediates elevation of [Ca 2ϩ ] i by GnRH in pituitary cells, we also assessed possible effects of GnRH on Ca 2ϩ in Ad hGnRH-R-infected HeLa cells. As shown ( Fig. 1), GnRH did not increase [Ca 2ϩ ] i in control fura-2-loaded HeLa cells but caused a sustained and titer-dependent increase after infection at m.o.i. values of 3-100. Since these data were acquired by video imaging it was also possible to determine the proportion of cells responding to GnRH. This revealed that the vast majority of cells (Ͼ80%) responded to GnRH at relatively low titer (m.o.i. values of 10 -30), and that increasing viral titer above an m.o.i. of 10 increased the amplitude of the responses without increasing the proportion of cells responding. These data (not shown) are very similar to those previously obtained when Ad GnRH-R were used to infect pituitary cells (26).
To determine the ligand specificity of GnRH-Rs expressed in these cells competition binding assays were performed using unlabeled GnRH, buserelin, and cGnRH-II and 125 I-buserelin or 125 I-cGnRH-II in cells infected with Ad hGnRH-R or Ad XGnRH-R, respectively (both at an m.o.i. of 100). The rank order of potency for competition at the hGnRH-R was buserelinϾGnRHϾcGnRH-II, whereas that at the XGnRH-R was cGnRH-IIϾ ϾbuserelinϾ ϾGnRH (not shown). Identical rank orders of potency were seen when HeLa cells infected with Ad hGnRH-R or Ad XGnRH-R were used to construct doseresponse curves for [ 3 H]IP accumulation (not shown).
We next investigated desensitization of these receptors. To do so WT dynamin cells were infected with either Ad hGnRH-R or Ad XGnRH-R (at m.o.i. values of 100) and then used in time-course experiments in which [ 3 H]IP accumulation was measured. As shown (Fig. 2, upper panel) both receptors mediated comparable initial rates of [ 3 H]IP accumulation, but this initial rate was not maintained beyond 2 min for XGnRH-Rs stimulated by cGnRH-II. As an index of desensitization, the rate of accumulation between 2-5 min was only 9% Ϯ 3% of the initial rate (0 -2 min) for the XGnRH-R but was as high as 73% Ϯ 15% for the hGnRH-R.  . The binding was terminated by transfer to ice-cold PSS and washing before solubilization of the cells in 0.2 M NaOH and 1% SDS. In a parallel experiment, cells were treated exactly as described above except that, after binding, the cells were incubated in acid medium to remove radioligand binding from cell surface receptors. Data shown are mean Ϯ S.E. (n ϭ 3) of specific binding (nonspecific binding subtracted) pooled from repeated experiments, each having triplicate observations that were expressed as a percentage of total specific binding (acid-labile plus acid-resistant) at 30 min.

FIG. 2. Time course of [ 3 H]IP accumulation and internalization in WT dynamin cells infected with Ad hGnRH-R or Ad
To monitor receptor internalization, WT dynamin cells were infected with either Ad hGnRH-R or Ad XGnRH-R (m.o.i. 100) and used in time-course experiments in which both total specific binding and acid resistant-specific binding were measured. Both ligands associated rapidly with their receptors with halfmaximal binding at 5-10 min (not shown). However, the acidresistant fraction of specific binding was greater in cells expressing XGnRH-Rs than hGnRH-Rs at all time points (Fig. 2,  lower panel). To test whether these receptors stimulate fluid phase endocytosis, uptake of horseradish peroxidase was quantified in cells stimulated for 60 min with 0 or 10 Ϫ7 M GnRH. Enzyme uptake was temperature-dependent (increased 2-3fold by increasing from 4°C to 37°C) but was not measurably influenced by GnRH in Ad hGnRH-R-infected cells (93 Ϯ 8% of control, n ϭ 4) or by cGnRH-II in Ad cGnRH-II-infected cells (109 Ϯ 9% of control, n ϭ 4).
We next determined the effectiveness of the K44A dynamin blockade of internalization using the endogenous EGF receptors as controls. Pretreatment with EGF caused a time-dependent loss of cell surface EGF receptors (as judged by subsequent whole cell binding assays at 4°C). In K44A dynamin cells cultured with tetracycline, surface binding was reduced to 26.8% Ϯ 10.4%, 16.2% Ϯ 1.3%, 9.1% Ϯ 2.9%, and 8.1% Ϯ 1.1% of control after 5, 15, 30, and 45 min, respectively (not shown). In contrast, no significant loss was seen in cells cultured in the absence of tetracycline (conditions permissive for expression of the dominant negative transgene). The specificity of this effect was established by control experiments in which EGF pretreatment caused a clear loss of cell surface receptors in WT dynamin HeLa cells, irrespective of the absence or presence of tetracycline in the culture medium.
The dynamin dependence of GnRH-R internalization was next investigated by infecting K44A dynamin cells with Ad hGnRH-R or Ad XGnRH-R, culturing these cells in the presence or absence of tetracycline, and then quantifying receptor internalization at 37°C as above. As shown (Fig. 3, upper  panel), 125 I-buserelin internalization (hGnRH-R-mediated) proceeded relatively slowly in control cells (cultured with tetracycline) to a maximum at 120 min with a half-time of 30 -60 min. Omission of tetracycline from the culture medium (permissive for K44A dynamin expression) did not measurably influence this time-course. 125 I-cGnRH-II internalization (via XGnRH-R) was relatively rapid in control cells cultured with tetracycline (no measurable increase after 30 min and a halftime of ϳ10 min) and was clearly reduced in cells cultured without tetracycline (Fig. 3, lower panel). In control experiments (not shown), tetracycline had no direct effect on radioligand binding to either of these GnRH-R, and omission of tetracycline from the culture medium did not measurably influence internalization of either receptor in cells expressing wild-type dynamin (under tet-off control).
To determine whether differences in receptor number underlie the observed differences in dynamin dependence, internalization was assessed in cells infected with Ad hGnRH-R or Ad XGnRH-R at varied m.o.i. (from 3-100) and cultured with and without tetracycline. Fig. 4 shows the relationship between total specific binding (an index of receptor number) and acidresistant binding (an index of receptor internalization) using data accumulated over four separate experiments. As expected, XGnRH-R internalization was greater than that for the hGnRH-R (72 Ϯ 4% (n ϭ 28) and 23 Ϯ 2% (n ϭ 25), respectively, data pooled irrespective of m.o.i.), and omission of tetracycline reduced the rate of XGnRH-R internalization (to 17 Ϯ 2%, n ϭ 30, p Ͻ 0.05) without altering that of the hGnRH-R (p Ͼ 0.1). Linear regression revealed that the rate of internalization of the XGnRH-R reduced as receptor number increased (gradient different from 0 at p Ͻ 0.05), but most importantly, the dynamin dependence of XGnRH-R internalization was maintained at a range of receptor densities encompassing those at which the hGnRH-R showed no dynamin dependence (Fig. 4).
We next addressed the dependence of GnRH-R internalization on CCVs using hypertonic sucrose (Fig. 5), which prevents the formation of clathrin lattices on coated pits and thereby prevents CCV-mediated receptor internalization (31). Using 15-min incubations, 72% Ϯ 8% of 125 I-EGF binding was acidresistant under control conditions, but this was reduced to 30% Ϯ 5% (p Ͻ 0.05) by tetracycline omission (K44A expression). Similarly 51% Ϯ 7% of 125 I-cGnRH-II binding to Ad XGnRH-R-infected cells was acid-resistant under control conditions, but this was reduced to 19% Ϯ 4% (p Ͻ 0.05) by allowing K44A expression. In contrast, only 22% Ϯ 4% of 125 I-buserelin binding was acid-resistant in control Ad hGnRH-R-infected cells, and this was not measurably altered by allowing K44A expression. Hypertonic sucrose (0.4 M) reduced the acid-resistant binding of all three ligands, to less than 5% without altering total specific binding (not shown), irrespective of the presence or absence of tetracycline (not shown). These data suggest that CCVs mediate dynamin-dependent and -independent internalization of GnRH-R.
Since GnRH is known to activate MAP kinase signaling in pituitary cells and GPCR-mediated MAP kinase activation can be dynamin-dependent, we next explored the possible effects on ERK 2 phosphorylation in Ad GnRH-R-infected K44A dynamin cells cultured with and without tetracycline. As shown (Fig. 6), GnRH and cGnRH-II both caused pronounced activation of ERK 2 phosphorylation in control cells (expressing hGnRH-Rs and XGnRH-Rs, respectively) with comparable time-courses (maxima achieved at ϳ10 min. with subsequent reduction to near basal levels at 60 min). Stimulation of ERK 2 phosphorylation was also seen in cells cultured without tetracycline, but for both receptors the response was significantly reduced (area under the curve 75.9% Ϯ 5.8 and 73.6% Ϯ 7.4% (p Ͻ 0.05) of control for XGnRH-Rs and hGnRH-Rs, respectively). Thus, although the internalization of the hGnRH-R is apparently not dependent upon dynamin in HeLa cells, its effect on ERK 2 phosphorylation clearly is.
To explore the relationship between receptor internalization and signaling, we blocked internalization with sucrose and tested for possible inhibition of GnRH-R-mediated ERK 2 activation. However, these experiments were complicated by the fact that sucrose itself activated ERK 2 (not shown) and we therefore sought an alternative means of blocking internalization. As shown (Fig. 7), the lectin, concanavalin A, had no measurable effect on XGnRH-R-mediated ERK 2 phosphorylation in control cells (not expressing K44A dynamin), despite the fact that it reduced 125 I-cGnRH-II internalization by ϳ70% (from 68 Ϯ 4% (n ϭ 4) to 17 Ϯ 1% (n ϭ 5), p Ͻ 0.05). Similar results (inhibition of internalization but not of signaling) were obtained with Ad hGnRH-R infected cells (not shown).
To further explore the mechanism of ERK 2 activation in these cells, effects of PKC activation and inhibition were determined. As shown (Fig. 8), the PKC activator PMA (like GnRH and cGnRH-II) increased ERK 2 phosphorylation in control cells (not expressing K44A dynamin). PMA had no significant effect in the presence of the PKC inhibitor bisindolylmaleimide 1 (demonstrating the efficiency of the blocker), and the responses to activation of hGnRH-Rs and XGnRH-Rs were both inhibited by the inhibitor (to 50% Ϯ 8 and 40% Ϯ 8% (p Ͻ 0.05) of maximum, respectively), demonstrating a mediatory role for PKC. Expression of K44A dynamin inhibited hGnRH-R-and XGnRH-R-mediated ERK 2 phosphorylation as expected (to 26% Ϯ 10 and 45% Ϯ 7% (p Ͻ 0.05) of maximum stimulation, respectively) and also inhibited PMA-stimulated ERK 2 phosphorylation (to 59% Ϯ 3%, p Ͻ 0.05), demonstrating that such inhibition is not GnRH-R-specific. When K44A dynamin was expressed, the dynamin-independent component of PMA-stimulated ERK 2 activation was blocked by bisindolylmaleimide 1 (again, demonstrating dependence on PKC activation), but we were unable to detect any such inhibition for responses to GnRH or cGnRH-II, implying that the dynamin-independent signaling of these receptors to ERK 2 is not PKC-mediated.
In a final series of experiments the effects of GnRH, cGnRH-II, and EGF were assessed in the presence and absence of the EGF receptor tyrosine kinase inhibitor AG1478. As shown (Fig.  9) EGF, GnRH, and cGnRH-II caused comparable phosphorylation of ERK 2 and were all sensitive to inhibition by expression of K44A dynamin (not shown). In contrast, 1000 nM AG1478 prevented EGF-stimulated ERK 2 phosphorylation without measurably altering the ERK 2 phosphorylation mediated by activation of either GnRH-R. DISCUSSION The GnRH receptor family has undergone a period of rapidly accelerated molecular evolution in which the C-terminal tails, present on all known non-mammalian GnRH-R, have been lost with the advent of mammalian type I GnRH-R. This provides a unique opportunity to study the functional relevance of these features with normal (non-mutated) receptors. Such studies with (open symbols) or without (closed symbols) tetracycline for 8 h to allow infection before the medium was changed to fresh containing no Ad with or without tetracycline to allow expression of the mutant dynamin and then cultured for a further 16 h. They were then used in whole cells receptor internalization assays as above (Fig. 3 legend). The figures show total specific binding (an index of receptor number, x axis) plotted against the proportion of binding that was acid-resistant (% internalized, y axis). The data are pooled from multiple experiments, with each dot representing a single determination of each parameter from a single internally controlled experiment with triplicate replicates. Linear regression revealed a significant negative correlation of log transformed binding data and internalization data for the XGnRH-R with and without tetracycline (slope different from 0 at p Ͻ 0.05 in both cases) but not for the hGnRH-R (p Ͼ 0.1 in both cases). have revealed that mammalian GnRH-R do not rapidly desensitize whereas the two non-mammalian GnRH-R investigated to date (catfish (23) and Xenopus GnRH-R (26)) do show rapid homologous desensitization. Similarly, mammalian GnRH-Rs undergo agonist-induced internalization, but at much lower rates than non-mammalian GnRH-R (24, 26, 27). The resist-  FIG. 7. Influence of concanavalin A on XGnRH-R-mediated ERK 2 activation. ERK 2 phosphorylation was determined and quantified as described above in K44A HeLa cells infected with Ad XGnRH-R at an m.o.i. of 100. The cells were pretreated with 0 or 0.25 mg/ml concanavalin A (Con A) for 30 min prior to stimulation for 15 min with 0 or 10 Ϫ7 M cGnRH-II as indicated. The figure shows mean Ϯ S.E. (n ϭ 3) calculated after expressing the phosphorylated ERK 2 as a percentage of total ERK 2 (both determined by densitometry). ERK 2 phosphorylation was not measurably influenced (p Ͼ 0.1) by concanavalin A in the presence or absence of cGnRH-II. Similar data were obtained in cells infected with Ad hGnRH-R and stimulated with 0 or 10 Ϫ7 M GnRH (not shown).

FIG. 8. PKC-dependence of GnRH-R-mediated activation of ERK 2 in K44A dynamin cells infected with Ad hGnRH-R or Ad
XGnRH-R. K44A dynamin cells were infected with Ad hGnRH-R or Ad XGnRH-R at m.o.i. values of 50 -100 before being cultured with or without tetracycline. Cells were then pretreated with PSS with or without 1 M bisindolylmaleimide 1 for 30 min before being stimulated with PSS containing 30 nM PMA (top panel), 100 nM GnRH (hGnRH-R, middle panel) or cGnRH-II (XGnRH-R, bottom panel) for 10 min before extraction. Data is expressed as a percentage of maximal stimulation determined by the proportion of ERK 2 that is in the phosphorylated form from pooled observations (n ϭ 3), as determined by densitometry measurement. The PMA control is pooled from both hGnRH-R-and XGnRH-R-infected cells (n ϭ 6) (* p Ͻ 0.05 compared with maximal stimulation (no inhibitor), ** p Ͻ 0.05 compared with both maximal stimulation and dynamin blocked stimulation).
ance of mammalian GnRH-R to desensitization and internalization has been attributed to the lack of required C-terminal tail phosphorylation sites since catfish GnRH-R do undergo agonist-induced phosphorylation and bind ␤-arrestin (causing it to translocate to the plasma membrane), whereas mammalian GnRH-R do not (32). Similarly, we have found that XGnRH-R mediate translocation of ␤-arrestin-GFP to plasma membranes, whereas hGnRH-R do not (not shown). Moreover, internalization of catfish GnRH-R is ␤-arrestin-dependent (24), whereas internalization of the human GnRH-R is not (25). Thus non-mammalian GnRH-Rs appear to follow the scheme outlined above for ␤-arrestin-mediated desensitization and internalization (and possible ␤-arrestin-mediated signaling), but this is not the case for the mammalian GnRH-Rs investigated to date (20,22).
Here we have explored the dynamin dependence of desensitization, internalization, and signaling of human and Xenopus GnRH-Rs using recombinant adenovirus to express these receptors in HeLa cells expressing either wild-type or K44A (dominant negative) dynamin. GnRH-Rs expressed in this way had high affinity for receptor-specific ligands (low nM K d for 125 I-buserelin binding to hGnRH-Rs and for 125 I-cGnRH-II binding to XGnRH-Rs) and mediate PLC activation, Ca 2ϩ mobilization, and MAP kinase (ERK 2) activation. They also show different ligand specificity (buserelinϾGnRHϾcGnRH-II at hGnRH-Rs and cGnRH-IIϾ ϾbuserelinϾGnRH at XGnRH-Rs) and different rates of internalization and desensitization (both being more rapid for the XGnRH-R). Receptor density, and receptor-mediated effects on  (33). Thus, recombinant Ad provide an efficient means of expressing GnRH-Rs in HeLa cells at physiological density, and these receptors retain the pharmacological characteristics anticipated from earlier studies with endogenous GnRH-Rs and GnRH-Rs expressed heterologously in pituitary gonadotrope progenitor cells (26,34,35).
The dynamin dependence of EGF receptor internalization has previously been demonstrated in K44A dynamin cells (3) and was confirmed here as a positive control for blockade of dynamin-dependent endocytosis. Using this system, we have found, as expected, that internalization of the XGnRH-R is more rapid than that of the hGnRH-R in K44A dynamin cells cultured with tetracycline. Blockade of dynamin-dependent internalization by tetracycline omission did not reduce hGnRH-R internalization. This was unexpected because transient expression of K44A dynamin has been reported to inhibit internalization of the rat GnRH-R by 20% in COS 7 cells (32). Although a tendency for inhibition was observed at the 60-and 120-min time points (Fig. 3), and in some subsequent experiments (e.g. Fig. 4), this did not attain statistical significance (even when data were pooled from multiple series of experiments), and we were therefore unable to demonstrate any measurable dynamin dependence of hGnRH-R internalization in HeLa cells. In stark contrast, the internalization of XGnRH-Rs was dramatically reduced when dynamin-dependent internalization was prevented by tetracycline omission.
Internalization of several GPCRs is inhibited by expression of dominant negative mutants of dynamin (6,15) and by prevention of CCV formation using hypertonic sucrose (31). Although these treatments are not entirely specific for CCVs (4,11,37,38), electron microscopy has revealed that mammalian GnRH-R are internalized via coated vesicles (39) and co-internalization with transferrin (a marker for CCV-mediated internalization) implies that these are CCVs (25). We have recently found that the XGnRH-R also co-internalizes with transferrin (not shown). Accordingly, the sucrose-dependent internalization of both GnRH-R in HeLa cells is also most likely CCV-mediated.
We were concerned that the observed differences in GnRH-R internalization might reflect differences in receptor number and tested this by expressing receptors at varied density (varying viral titer between m.o.i. values of 3-100). This revealed a negative correlation between XGnRH-R binding and internalization rate (Fig. 3), supporting the notion that the stoichiometry of receptors and other proteins can indeed influence the observed internalization. However, the key distinctions between hGnRH-R and XGnRH-R internalization (faster internalization and dynamin sensitivity of the later) are maintained over a wide range of receptor density (Fig. 4). The retention of these distinctions with a range of XGnRH-Rs estimated to encompass that of the hGnRH-Rs clearly demonstrates dependence on receptor structure rather than receptor density.
The best-explored pathway for ERK 2 activation by tyrosine kinase receptors involves Shc/Grb2-mediated activation of a Ras guanine nucleotide exchange factor that activates the monomeric G-protein, Ras. This in turn activates Raf, which phosphorylates MEK causing it to phosphorylate ERK 2. Numerous GPCRs, including GnRH-Rs, have also been shown to activate ERK 2 by feeding into this pathway at multiple sites (41,42). Up-stream activation can be achieved by GPCR-mediated transactivation of EGF receptor involving stimulated liberation of HB-EGF in the extracellular environment (19). For Gq-coupled GPCRs activation of PKC can lead to Raf activation, and Ca 2ϩ elevation can activate a Ras G-protein regulatory factor. More recently, ␤-arrestin, recruited to ␤ 2 -adrenergic receptors, was shown to bind Src, which can regulate Shc/ GRb2 (12,36,43), and the observation that dominant negative mutants of ␤-arrestin or dynamin-blocked ERK 2 activation implied that GPCR endocytosis was necessary for signaling to the MAP kinase. These observations are intriguing, not only FIG. 9. Tyrosine kinase dependence of GnRH-R-mediated activation of ERK 2 in K44A dynamin cells infected with Ad hGnRH-R or Ad XGnRH-R. K44A dynamin cells were infected with Ad hGnRH-R or Ad XGnRH-R at m.o.i. values of 50 -100 before being cultured with tetracycline. Cells were then pretreated with PSS with or without 1000 nM AG1478 for 30 min before being stimulated with PSS containing 10 Ϫ7 M GnRH (hGnRH-R) or cGnRH-II (XGnRH-R) or 10 Ϫ7 M EGF (EGF-R) for 10 min before extraction. Data is expressed as the percentage of ERK 2 that is in the phosphorylated form from pooled observations (n ϭ 3) as determined by densitometry measurement. * p Ͻ 0.05 compared with maximal stimulation (no inhibitor). because they demonstrate signaling of the "desensitized" receptor, but also because Src phosphorylates and activates dynamin (44), providing a mechanism for activation of dynamin-dependent internalization by the GPCR. However, EGF receptor signaling and MEK signaling are also dynamin-dependent (3,16), so the dynamin dependence of GPCR signaling can actually reflect the fact that GPCR activation impinges on the EGF receptor/ERK 2 signaling pathway (reviewed in Ref. 42).
From these observations, it is clear that understanding the relationship between GPCR cycling and ERK 2 activation will depend upon the loci at which the GPCR signaling pathway feeds into the MAP kinase cascade. Here, we have found that hGnRH-R and XGnRH-R both mediate ERK 2 phosphorylation (indicative of activation of the ERK 1/2 signaling cassette) in K44A cells cultured with tetracycline. To our knowledge, this is the first demonstration of MAP kinase activation by a nonmammalian GnRH-R, and it is therefore of interest that timecourses and amplitudes of the responses were indistinguishable. This clearly implies that rapid GPCR desensitization (seen for the XGnRH-R but not for the hGnRH-R) is not an important determinant of the kinetics of this response. It remains to be determined whether ␤-arrestin-mediated signaling (which may occur for the XGnRH-R but presumably not for the hGnRH-R) supports the response to XGnRH-R activation.
In addressing the mechanisms of ERK 2 activation we found that the tyrphostin AG 1478 abolished EGF-stimulated ERK 2 activation at a concentration that does not inhibit the response to activation of either of the GnRH-Rs used. ERK activation by the endogenous mouse GnRH-R of ␣T3-1 pituitary cells has also been shown to be resistant to inhibition by AG1478 and by dominant negative EGF receptors (40) despite the fact that earlier work had implied a role for EGF receptor transactivation in GnRH action (18). In the HeLa cells used here, the differential sensitivity to AG1478 clearly implies that EGF-R activation (17)(18)(19)) does not underlie GnRH-R-mediated ERK 2 activation. However, we have also found that activation of ERK 2 by both GnRH-Rs was partially inhibited by dynamin despite the fact that internalization of the hGnRH-R was not. This uncoupling clearly argues that dynamin-dependent internalization of the GPCR is not required for dynamin-dependent ERK 2 activation.
The interpretation above is reinforced by the fact that inhibition of XGnRH-R internalization with concanavalin A did not reduce XGnRH-R-mediated ERK 2 activation. Moreover, we found that PMA-stimulated ERK 2 activation is also dynamindependent and that the PKC inhibitor (bisindolylmaleimide 1) inhibited ERK 2 activation by both GnRH-Rs. Although multiple mechanisms are likely to be involved, the simplest interpretation of these data is that GnRH-R-mediated ERK 2 activation is largely PKC-mediated in these cells and that it is dynamin-dependent because signaling from PKC to ERK 2 is, at least in part, dynamin-dependent. Recent studies in pituitary cells indicate that GnRH activates ERK primarily by causing a PKC-mediated activation of Raf with only a minor contribution from Src and Ras to Raf activation (40). GnRH-stimulated ERK activation was also found to be partially dynamin-sensitive in this system, but this effect was confined to inhibition of the Src/Ras signaling step (e.g. up-stream of, or unrelated to, PKC activation). However, this does not hold true in HeLa cells where events downstream of PKC are also dynamin-dependent (e.g. PMA-stimulated ERK 2 activation is inhibited by K44A dynamin). Signaling by MEK, which lies down-stream of PKC in the GnRH-R signaling cascade, has been shown to be dynamin-dependent in COS 7 cells (16). In this system, the dynamin-dependent endocytosis of activated MEK was critical for MAPK activation, and this relation-ship may well explain the dynamin dependence of GnRH-R signaling shown here.
In the preceding sections, inhibition by K44A dynamin has been equated with "dynamin dependence", but this may be an over-simplification because internalization of different GPCRs has been found to be differentially sensitive to distinct mutants of dynamin (11). This raises the possibility that the K44A dynamin-independent hGnRH-R internalization described here might prove sensitive to other dynamin mutants, but even if this is the case, the difference between the hGnRH-R and the XGnRH-R (and EGF receptor) in terms of sensitivity to K44A dynamin 1 implies that they access functionally distinct internalization routes.
In summary, a functional comparison of human and Xenopus GnRH-Rs has revealed that the tail-less mammalian GnRH-R internalizes and desensitizes slowly as compared with a tailed non-mammalian GnRH-R, and that this distinction is a function of receptor type rather than receptor density. Expression of K44A dynamin did not measurably alter desensitization or hGnRH-R internalization but dramatically reduced internalization of the XGnRH-R and EGF receptor, whereas blockade of CCV formation with sucrose abolished internalization of all three receptors. The tyrphostin, AG 1478, blocked EGF-stimulated, but not GnRH-R-mediated, ERK 2 activation, demonstrating that EGF receptor transactivation is not mediating GnRH action in these cells. K44A dynamin expression also inhibited ERK 2 activation mediated by either GnRH-R, whereas inhibition of GnRH-R internalization with concanavalin A did not influence GnRH-R-mediated ERK 2 activation. PMA also caused a dynamin-dependent activation ERK 2, and the effects of PMA, GnRH, and cGnRH-II on ERK 2 phosphorylation were all blocked by a PKC inhibitor. Thus we have established that a) desensitizing and non-desensitizing GnRH-Rs are targeted for CCV-mediated internalization by functionally distinct mechanisms, b) GnRH-R signaling to ERK