Allosteric Modulation of the Calcium-sensing Receptor by γ-Glutamyl Peptides

γ-Glutamyl peptides were identified previously as novel positive allosteric modulators of Ca2+o-dependent intracellular Ca2+ mobilization in HEK-293 cells that bind in the calcium-sensing receptor VFT domain. In the current study, we investigated whether γ-glutamyl-tripeptides including γ-Glu-Cys-Gly (glutathione) and its analogs S-methylglutathione and S-propylglutathione, or dipeptides including γ-Glu-Ala and γ-Glu-Cys are positive allosteric modulators of Ca2+o-dependent Ca2+i mobilization and PTH secretion from normal human parathyroid cells as well as Ca2+o-dependent suppression of intracellular cAMP levels in calcium-sensing receptor (CaR)-expressing HEK-293 cells. In addition, we compared the effects of the potent γ-glutamyl peptide S-methylglutathione, and the amino acid l-Phe on HEK-293 cells that stably expressed either the wild-type CaR or the double mutant T145A/S170T, which exhibits selectively impaired responses to l-amino acids. We find that γ-glutamyl peptides are potent positive allosteric modulators of the CaR that promote Ca2+o-dependent Ca2+i mobilization, suppress intracellular cAMP levels and inhibit PTH secretion from normal human parathyroid cells. Furthermore, we find that the double mutant T145A/S170T exhibits markedly impaired Ca2+i mobilization and cAMP suppression responses to S-methylglutathione as well as l-Phe indicating that γ-glutamyl peptides and l-amino acids activate the CaR via a common mechanism.

The extracellular calcium-sensing receptor (CaR) 3 is a multi-modal class C G-protein coupled receptor that is widely distributed in mammalian tissues (1). It controls the plasma Ca 2ϩ ion concentration by mediating negative feedback regulation of parathyroid hormone secretion as well as Ca 2ϩ o -induced inhibition of Ca 2ϩ reabsorption in the distal nephron (review: (2)). Analysis of the receptor's physiological significance has been based on its tissue distribution, analyses of its actions in cultured cells and tissues, and demonstrations of the impact of global knock-outs (3) and, more recently, conditional knock-outs in endocrine and mesenchymal tissues (4). These studies demonstrate that the CaR is a pluripotent sensor, regulator and modulator. In addition to regulating whole body calcium metabolism, it senses nutrient signals in the gastrointestinal tract (reviews: Refs. 5, 6), modulates synaptic transmission (7) and cellular differentiation in mesenchymal tissues (4), as well as engraftment of stem cells in specific niches (8), and organogenesis e.g. in the lung (9) and central nervous system (10,11). Accompanying this remarkable plasticity, the CaR, like several other class C GPCRs, binds multiple physiologically relevant ligands to control intracellular signaling pathways (Reviews: Refs. 12,13). In addition to sensing multivalent cations including Ca 2ϩ and Mg 2ϩ , the CaR is modulated by various L-amino acids (review: Ref. 1). Based on chimeric receptor and mutational analyses, L-amino acids bind in the receptor N-terminal Venus Fly Trap (VFT) domain (14) and the effects of L-amino acids are selectively impaired by a double mutant (T145A/S170T), which exhibits normal Ca 2ϩ o -sensing (15). Comparative molecular modeling of the mGlu-1 and CaR VFT domain ligand binding surfaces indicates that, whereas the amino acid side-chain binding region is tightly defined by a cluster of positively charged residues in mGlu-1 and other mGlus (16,17) it is relatively unrestricted in the CaR and closely related class C GPCRs including GPRC6A and T1R1 (18). The larger side-chain binding surfaces in this subgroup of receptors appear to explain, at least in part, their notable promiscuity for various sub-classes of amino acids (reviews: Refs. 19,20) and has led to the successful prediction that small peptides that retain ␣-amino and ␣-carboxylate functional groups at their N termini, including the tri-peptide glutathione would bind and activate the CaR expressed in HEK-293 cells (21). Building on this observation, ␥-glutamyl peptides were recently found to be potent taste enhancers (22) raising the possibility that they may be physiologically im-portant receptor modulators in various tissues in which the CaR is expressed.
To further evaluate the physiological significance of these observations we have now tested the effects of the ␥-glutamyl-tripeptides ␥-Glu-Cys-Gly (glutathione), S-methylglutathione and S-propylglutathione, and dipeptides ␥-Glu-Ala and ␥-Glu-Cys, on extracellular Ca 2ϩ o -induced intracellular Ca 2ϩ mobilization and suppression of cAMP levels in CaRexpressing HEK-293 cells. We then extended the investigation to consider the effects of these peptides on intracellular Ca 2ϩ mobilization and PTH secretion in normal human parathyroid cells, and whether a double mutant T145A/S170T that selectively impairs amino acid-dependent activation of the CaR (15) might also impair the actions of ␥-glutamyl peptides.
The results indicate that various ␥-glutamyl peptides including glutathione and its metabolite S-methylglutathione are potent positive allosteric modulators of the CaR that promote Ca 2ϩ i mobilization and lower intracellular cAMP levels in CaR-expressing HEK-293 cells and promote Ca 2ϩ i mobilization and suppress PTH secretion in normal human parathyroid cells. Results of a comparative analysis of HEK-293 cells that express either the wild-type or double mutant T145A/ S170T CaR indicate that, distinct from Ca 2ϩ o and phenylalkylamines, ␥-glutamyl peptides and L-amino acids modulate the receptor via a common mechanism linking binding of L-amino acids or ␥-glutamyl peptides at closely related sites in the VFT domain to a characteristic change in receptor conformation that couples to the modulation of cytoplasmic Ca 2ϩ and cAMP levels.

EXPERIMENTAL PROCEDURES
Cell Culture of Control and Stably Transfected HEK-293 cells-Human embryonic kidney (HEK)-293 cells stably expressing the CaR (HEK-CaR cells), untransfected HEK-293 cells, and HEK-293 cells that stably expressed a double mutant form of the CaR (HEK-CaR-T145A/S170T cells) were thawed from frozen stocks and cultured in DMEM (Invitrogen)/10% fetal bovine serum as described previously (15). In general, cultured cells were studied between passage numbers 4 and 20. For microfluorimetry experiments, HEK-293 cells were transferred onto sterilized 15-mm diameter coverslips in 24-well plates and cultured for a further 24 -48 h.
Preparation of Human Parathyroid Cells-Samples from normal human parathyroid auto-transplants were obtained during thyroid surgery at the Royal North Shore Hospital and North Shore Private Hospital, St Leonards, New South Wales and the Mater Misericordiae Hospital, North Sydney, New South Wales, Australia. All procedures were performed under guidelines established by the relevant human research ethics committees, and all patients provided written informed consent for the use of the tissue for experimental purposes.
Parathyroid tissue was transported to the laboratory in MEM that contained 1.25 mM CaCl 2 . Upon arrival in the laboratory, it was either used immediately or, more typically, stored overnight in MEM at 4°C as described previously (23). Prior to collagenase digestion, parathyroid tissue retains viability under these conditions (24). For digestion, it was transferred to MEM that contained 1 mg/ml collagenase and 0.1 mg/ml DNase I. After brief oxygenation it was incubated at 37°C for 20 min. The enzyme suspension was then decanted and the parathyroid tissue was transferred into 5 ml of MEM that contained 1 mg/ml of bovine serum albumin (BSA). The tissue was then subjected to trituration (10 -15 times) through the tip of a disposable 5-ml syringe (no needle attached). The cloudy suspension containing clumped parathyroid cells was then passed through a 200-m pore size nylon filter, sedimented (70 ϫ g, 2.5 min), and the cell pellet gently resuspended then washed twice with 5 ml of BSA-containing MEM. It was finally resuspended in 2 ml of bovine serum albumin-containing MEM. Remaining pieces of undigested parathyroid tissue were then incubated for a further 20 min at 37°C in fresh medium containing collagenase and DNase I. The enzyme suspension was once more decanted, and the parathyroid tissue subjected to trituration and centrifugal isolation as above.
Amino acid and Peptide-containing Solutions-Stock amino acid-containing solutions were routinely made up in physiological saline at 50 or 100 mM and stored at Ϫ20°C then diluted in amino acid-free physiological saline as required. Dipeptide and tri-peptide containing solutions, on the other hand, were made up fresh in physiological saline and used on the day of experiment. The control physiological saline solution (PSS) used in all microfluorimetry experiments had the following composition: 125 mM NaCl, 4.0 mM KCl, 0.5 mM CaCl 2 , 1.0 mM MgCl 2 , 20 mM HEPES (NaOH), 0.1% D-glucose (pH 7.4).
Microfluorimetry of HEK-293 Cells and Human Parathyroid Cells for Determination of Intracellular Ca 2ϩ Concentration-HEK-293 cells that had been cultured on coverslips in 24-well plates were loaded with 5 M fura2-AM in PSS that contained 1 mg/ml bovine serum albumin at 37°C for 2 h in the dark. This solution was then removed and replaced by fresh albumin-containing PSS and the cells were incubated for a further 30 min to promote the conversion of fura-2 to its fully ionized, Ca 2ϩ -binding form.
Suspensions of normal human parathyroid cells, which had been prepared by collagenase digestion, were loaded with 1 M fura-2 AM for 20 min at 37°C in MEM containing bovine serum albumin (1 mg/ml). The cells were sedimented (70 ϫ g, 2.5 min) and then resuspended in albumin-free physiological saline solution.
Fura-2-loaded HEK-293 cells that had been cultured on coverslips or fura-2 loaded parathyroid cells, which readily attach to glass surfaces, were transferred into a superfusion chamber, placed in the light path of a Zeiss Axiovert fluorescence microscope (63ϫ objective) and perifused with inorganic phosphate-free physiological saline solutions that were modified to contain various concentrations of Ca 2ϩ o , L-Phe, and ␥-glutamyl di-and tri-peptides. A Lambda DG-4 150 watt Xenon light source (Sutter) was programmed to alternate between two excitation wavelengths (340 and 380 nm) at a frequency of 1 s Ϫ1 . Areas of interest were selected and digital images were then captured and downloaded using a high resolution AxioCam camera controlled by Stallion SB.4.1.0 PC software (Zeiss, Australia). Fura-2-loaded cells were imaged at 510 nm. For concentration-response analysis, fura-2 ratio data were integrated and then either plotted directly or expressed as a ratio of a control response (e.g. L-Phe or R-467).
Detection of Changes in Intracellular cAMP Levels in CaRexpressing HEK-293 Cells-HEK-CaR, HEK-CaR-T145A/ S170T, or control HEK-293 cells were seeded onto 15-mm coverslips in 24-well plates for 24 h and then transfected for 48 h with pcDNA3.1 containing the cAMP bio-sensor CFPnd-EPac1-cpVenus (EPac1; the generous gift of Dr Kees Jalink, Netherlands Cancer Institute) using Lipofectamine-2000 according to the manufacturer's instructions (Invitrogen). The media were replaced with PSS containing 0.5 mM Ca 2ϩ for 30 min at 37°C, and coverslips were then transferred to a chamber placed in the light path of a Zeiss Axiovert epi-fluorescence microscope (63ϫ objective) and perifused with PSS that contained various concentrations of Ca 2ϩ o , L-Phe, ␥-glutamyl di-or tri-peptides, or the type-II calcimimetic NPS R-467 as required. Epac1-transfected cells were excited continuously with light centered on a wavelength of 436 nm using the Lambda DG-4 light source and emitted light corresponding to CFP and Venus YFP emissions were sampled at 1s intervals using filters centered on 488 nm (CFP) and 528 nm (Venus-YFP) to permit the detection of cAMP-dependent changes in FRET. Cells of interest were selected and digital images were captured and downloaded as described above. The ratios of the fluorescence readings at 488 nm and 528 nm were plotted as a function of time after correcting for background in the absence of cells.
Determination of PTH Secretion from Perifused Human Parathyroid Cells-Perifusion of normal human parathyroid cells was undertaken in low molecular mass (4 -5 kDa) cut-off gel filtration micro beads so that intact PTH (ϳ9 kDa) would appear in the void volume as described previously (23,25). Gel filtration media were pre-equilibrated with physiological saline: 125 mM NaCl, 4.0 mM KCl, 1.25 mM CaCl 2 , 1.0 mM MgCl 2 , 0.8 mM Na 2 HPO 4 , 20 mM HEPES (NaOH), 0.1% D-glucose (pH 7.4) that contained 1ϫ basal amino acid mixture (total concentration 2.8 mM; (26)), and 1 mg/ml bovine serum albumin. Around 40,000 -50,000 cells were loaded onto the surface of a 0.4-ml bed volume of Bio-gel P-4 (nominal exclusion limit 4 kDa) and then gently covered with a 0.4-ml bead volume of Sephadex G-25 (nominal exclusion limit 5 kDa) in a small perifusion column. Tubing connections were then established downstream to a peristaltic pump and up-stream to a reservoir and the column was suspended in a water bath (37°C) and perifused at 1.5 ml/min with 37°C equilibrated control physiological saline that contained the 1-fold L-amino acid mixture and 1 mg/ml bovine serum albumin. Routinely, 2-min (i.e. 3 ml) samples were collected into tubes immersed in an ice bath and then transferred to dry ice. As required, solutions were changed to permit variations in the concentrations of Ca 2ϩ o , amino acids or ␥-glutamyl peptides. All samples were stored at Ϫ80°C until analysis of intact human PTH using an Immulite 2000 autoanalyzer. For estimates of steadystate levels of PTH secretion, the first 2-min samples of all treatment periods, during which the cells were re-equilibrating, were routinely ignored and the means of the second and third samples were used for plotting the concentration-response relationships and curve-fit analysis.
Statistical Analysis and Curve Fitting-Microfluorimetry experiments yielded single cell data-sets consisting of fluorescence excitation ratios (F 340 /F 380 ) and the corresponding times (s) since the start of the experiment. The fluorescence ratio data were integrated and the baseline responses corresponding to an initial control period were subtracted to yield activator-and concentration-dependent receptor-response values. The resulting data were plotted unmodified or expressed as ratios of control responses. For ␥-glutamyl peptide responses, either 10 mM L-Phe or 5 M NPS R-467 were used as controls. Experimental data were plotted using DeltaGraph 5.0 or GraphPad Prism 5.0b and curve-fitting was performed using the Hill equation: r ϭ b ϩ (a-b) C n /(e n ϩC n ) where r ϭ response, a ϭ maximum response, b ϭ basal response, C ϭ extracellular Ca 2ϩ concentration (in mM), e ϭ EC 50 (the Ca 2ϩ o concentration that induced a half-maximal response) and n ϭ Hill coefficient. PTH secretion data were fitted to the following equation: S ϭ a Ϫ (aϪb) C n /(i n ϩC n ) where S ϭ secretory response, a ϭ maximum secretory response, b ϭ basal secretory response, C ϭ extracellular Ca 2ϩ concentration (in mM), i ϭ IC 50 (the Ca 2ϩ o concentration that yielded half-maximal inhibition) and n ϭ Hill coefficient. The data were expressed routinely as means Ϯ S.E. (number of experiments). Statistical comparisons between treatments within PTH secretion data sets were performed using Student's paired t test (two-tailed) in SPSS Statistics 17.0. Statistical comparisons between the effects of positive allosteric modulators on pEC 50 values, maximum responses, or specific data points in Ca 2ϩ o concentration-response curves were performed using Student's unpaired t test (two-tailed) in SPSS. Statistically significant comparisons were accepted at p Ͻ 0.05.  Table 1).

Effects of ␥-Glutamyl
All four ␥-glutamyl peptides were effective positive allosteric modulators of Ca 2ϩ i mobilization in the presence of 2.5 mM Ca 2ϩ . The order of potency for Ca 2ϩ i mobilization was as follows (at Ca 2ϩ o ϭ 2.5 mM; Fig. 1, B and C and Table 1): S-methylglutathione Ͼ glutathione Ͼ ␥-Glu-Ala Ͼ L-Phe. In separate experiments, similar potencies were observed for ␥-Glu-Cys and ␥-Glu-Ala (Table 1). Co-administration of the recognized CaR-active amino acid, L-Phe and high potency ␥-glutamyl peptide S-methylglutathione, at maximal concentrations i.e. 10 mM and 10 M respectively, induced no additional effect when compared with either L-Phe or S-methylglutathione alone (not shown). On the other hand, at a Ca 2ϩ o concentration of 2.5 mM, maximally effective concentrations of S-methylglutathione and NPS R-467 exhibited significant positive interactions when compared with the effects of either compound alone (Fig. 2).

Effects of ␥-Glutamyl Peptides on Ca 2ϩ o -dependent Intracellular Ca 2ϩ Mobilization in Normal Human Parathyroid
Cells-Of the amino acids and peptides that had been identified as positive allosteric modulators of Ca 2ϩ i mobilization in HEK-CaR cells, L-Phe, ␥-Glu-Ala, glutathione, and S-methylglutathione, together with another glutathione analog, Spropylglutathione were selected for further studies of intracellular Ca 2ϩ mobilization in fura-2 loaded normal human parathyroid cells. At Ca 2ϩ o concentrations at or below 0.5 mM, none of the compounds was effective. At Ca 2ϩ o concentrations greater than 1.0 mM, however, all of the compounds acutely mobilized intracellular Ca 2ϩ (Fig. 3A). At a Ca 2ϩ concentration of 1.5 mM, the concentration-response relations indicated that the ␥-glutamyl peptides were effective in the micromolar concentration range with S-methylglutathione exhibiting a near maximal effect at around 2-5 M (Fig. 3, B and C). The order of potency for intracellular Ca 2ϩ mobilization was as follows: S-methylglutathione Ͼ glutathione Ͼ ␥-Glu-Ala Ͼ S-propylglutathione ( Fig. 3C and Table 2).
Effects of ␥-Glutamyl Peptides on PTH Secretion from Normal Human Parathyroid Cells-We next tested the effects of ␥-glutamyl peptides including ␥-Glu-Ala, glutathione and S-methylglutathione (Fig. 4) Fig. 4, B and C). For ␥-Glu-Ala, the IC 50 value was ϳ30 M (supplemental Fig. S1). The results indicate that ␥-glutamyl peptides and S-methylgluta-   (14) and glutathione (21) binding sites to the VFT domain and we previously identified T145A/S170T as a double mutant form of the VFT domain with near-normal Ca 2ϩ o -sensitivity but markedly impaired L-amino acid sensing (15). In the current study, we first con-firmed that this mutant exhibited normal or near-normal Ca 2ϩ o sensitivity in the absence of L-Phe or S-methylglutathione (Fig. 5, A and B). In the presence of 10 mM L-Phe, HEK-293 cells expressing the wild-type CaR exhibited markedly enhanced Ca 2ϩ o sensitivity (Fig. 5C) whereas HEK-293 cells expressing the T145A/S170T double mutant CaR exhibited no change in Ca 2ϩ o sensitivity (Fig. 5D) as reported previously (15).
We next tested the impacts of the potent ␥-glutamyl peptide, S-methylglutathione on the Ca 2ϩ o sensitivity of Ca 2ϩ i mobilization in wild-type CaR and T145A/S170T CaR-expressing cells. In HEK-CaR cells, 10 M S-methylglutathione induced a marked increase in Ca 2ϩ o potency (Fig. 5E) but had little or no effect on Ca 2ϩ o potency in HEK-293 cells express- ing the CaR double mutant T145A/S170T (Fig. 5F). Consistent with these findings, concentration-response relations for the wild-type CaR and T145A/S170T double mutant demonstrated very similar Ca 2ϩ o sensitivities in the absence of allosteric modulators (Fig. 6, A and B) and the EC 50 values for     Fig. 6, A and B). Thus, the EC 50 values for Ca 2ϩ o (in mM) were, respectively, 3.6 Ϯ 0.3 (n ϭ 5), 3.4 Ϯ 0.2 (n ϭ 7), and 3.5 Ϯ 0.2 (n ϭ 7) in the absence of modulators or in the presence of 10 mM L-Phe or 10 M S-methylglutathione (p ϭ 0.78 for L-Phe and p ϭ 0.82 for S-methylglutathione with respect to control).
To further investigate the impacts of L-Phe and S-methylglutathione on the CaR double mutant T145A/S170T, we examined the effects of stepwise increases in the concentrations of L-Phe and S-methylglutathione in the presence of a submaximal Ca 2ϩ o concentration, 2.5 mM using 5 M NPS R-467 as a positive control (Fig. 7). In the case of the wild-type CaR, both L-Phe and S-methylglutathione induced robust concentration-dependent increases in Ca 2ϩ i mobilization in the presence of 2.5 mM Ca 2ϩ o (Fig. 7, A and C). In the case of the CaR double mutant T145A/S170T, however, the effects of both L-Phe and S-methylglutathione were attenuated (Fig. 7, B and D). Concentration-response relations for L-Phe and Smethylglutathione derived from experiments including those shown in Fig. 7 demonstrated marked impairment of Ca 2ϩ i mobilization responses in HEK-293 cells expressing the T145A/S170T double mutant when compared with wild-type (Fig. 8) and similar results were obtained whether the data were expressed as integrated fluorescence ratio responses (Fig. 8) or, alternatively, as fractional responses with respect to a control, 5 M R-467 (not shown).
Interestingly, the double mutant exhibited markedly suppressed maximum responses to L-Phe and S-methylglutathione but appeared to have little or no effect on the potencies of the modulators (Fig. 7, B and D; Fig. 8). These results suggest that the double mutant has little or no effect on modulator binding but instead impairs the coupling between Ca 2ϩ o and positive allosteric modulators such as L-Phe and glutathione analogs that bind in the VFT domain.
Impacts of L-Phe and S-Methylglutathione on Ca 2ϩ o -dependent cAMP Suppression in HEK-293 Cells that Stably Express the Wild-type CaR or T145A/S170T Double Mutant-Finally, we investigated the impact of L-Phe, S-methylglutathione and elevated Ca 2ϩ o on intracellular cAMP levels in HEK-CaR cells that were transfected with the cAMP reporter CFPnd-EPac1-cpVenus since the CaR is known to suppress cAMP levels in a G i and Ca 2ϩ i -dependent manner (27). cAMP levels were first elevated by exposure to 5 M forskolin and the cells were then exposed to one or more CaR activators in the continuing presence of forskolin. Elevated Ca 2ϩ o induced a concentration-dependent inhibition of cAMP levels that appeared to reach a maximum at around 6 mM (Fig. 9A). Restoring Ca 2ϩ o to the baseline level (0.5 mM) led to a prompt recovery in intracellular cAMP levels, which took the form of an overshoot following exposure to high Ca 2ϩ o concentration (Ն 5 mM) (Fig. 9A), consistent with the phenomenon of sensitization or superactivation of adenylyl cyclase (review: Ref. 28). No effects of elevated Ca 2ϩ o , L-Phe, or S-methylglutathione were observed in control HEK-293 cells.
Consistent with their behaviors as positive allosteric modulators, L-Phe and S-methylglutathione were without effect on intracellular cAMP levels in HEK-CaR cells at a Ca 2ϩ o concentration of 0.5 mM (not shown). In the presence of submaximal Ca 2ϩ o concentrations (1.5 or 2.5 mM), however, 10 mM L-Phe and 10 M S-methylglutathione acutely and reversibly suppressed cAMP levels (Fig. 9B), and the phenylalkylamine type-II calcimimetic NPS R-467 (0.2-5 M) had a similar effect (not shown). We next tested the effects of L-Phe, S-methylglutathione, and elevated Ca 2ϩ o on intracellular cAMP levels in HEK-293 cells that stably expressed the T145A/S170T CaR double mutant. The effects of 10 mM L-Phe and 10 M S-methylglutathione were markedly attenuated, however, the effects of 5 mM Ca 2ϩ o (Fig. 9C) and 5 M R-467 (not shown) were unaffected.

DISCUSSION
The CaR has several binding sites for Ca 2ϩ ions and other multivalent cations in its VFT (29) and heptahelical (14,30) domains. These so-called type-I (cationic) calcimimetics are generally regarded as full agonists of the receptor since they activate multiple downstream signaling pathways linked to the activation of G q/11 , G i/o or G 12/13 (reviews: Refs. 12, 13, 31). Thus, their binding site(s) define the location(s) of the receptor orthosteric site(s). In addition, the CaR binds socalled type-II calcimimetics, which are typically uncharged organic compounds that act as positive allosteric modulators. Two main sites for the binding of positive allosteric modulators have been identified: one in the heptahelical domain for type-II calcimimetics such as the phenylalkylamines NPS R-467 and cinacalcet (30,32); and one or more in the VFT domain for L-amino acids such as L-Phe or L-Trp (14) and glutathione and analogs (21). Previous work suggested that the L-amino acid binding site is selectively disabled by the double mutant T145A/S170T (15).
In the current study, we provide further evidence of the efficacy of ␥-glutamyl peptides as positive allosteric modulators of Ca 2ϩ o -induced Ca 2ϩ i mobilization in CaR-expressing HEK-293 cells ( Fig. 1 and Table 1) and demonstrate their efficacy in normal human parathyroid cells (Fig. 3 and Table 2). In addition, we found that two glutathione analogs, S-methylglutathione and S-propylglutathione and two other ␥-glu-  Tables 1 and 2) and ␥-Glu-Cys (Table 1)  i -dependent mechanisms have been identified (27). In the current study, we transiently expressed a FRET-based cAMP sensor, CFPnd-EPac1-cpVenus (33) in CaR-expressing HEK-293 cells to permit detection of changes in intracellular cAMP levels in real-time. As expected, resting levels of cAMP were promptly elevated upon exposure of CaR-expressing HEK-293 cells to forskolin and, in its continuing presence, were suppressed in a concentration-dependent manner upon exposure to elevated Ca 2ϩ o (Fig. 9A). In addition, the positive allosteric modulators L-Phe and S-methylglutathione (Fig. 9B), as well as NPS R-467 (not shown) all reversibly suppressed cAMP levels in the presence of 1.5 mM Ca 2ϩ o . Thus, the positive allosteric modulators of the CaR, L-Phe, and S-methylglutathione both effectively suppressed cAMP levels. It is not yet known whether these effects arose primarily from activation of G i/o or Ca 2ϩ i mobilization. Previous studies have suggested that L-Phe primarily promotes a CaR-dependent pathway linked to Ca 2ϩ i mobilization (34,35) although it also has a small positive modulatory effect on ERK1/2 phosphorylation (36).
Chimeric receptor analysis previously demonstrated that the binding sites for L-amino acids (14) and glutathione (21) both localize to the receptor VFT domain. To investigate whether these two classes of positive allosteric modulator might bind at a common or overlapping site and/or act via a common mechanism, as suggested by molecular modeling analysis (18), we took advantage of a previously identified double mutant of the CaR VFT domain (T145A/ S170T) that selectively disables L-amino acid sensing with little or no effect on Ca 2ϩ o -sensing (15). This mutant disabled the effects of L-Phe as well as S-methylglutathione, on Ca 2ϩ i mobilization (Figs. 5-8) as well as suppression of forskolin-elevated cAMP levels in CaR-expressing HEK-293 cells (Fig. 9). However, in the current study, the double mutant appeared to markedly impair efficacy with little or no effect on potency in the cases of both L-Phe and S-methylglutathione (Fig. 8) suggesting that its primary impact was not on ligand binding but on the coupling of the amino acid and ␥-glutamyl peptide binding sites to Ca 2ϩ o -dependent receptor activation. This suggests that the mutant receptor is unable to sustain a normal conformational response to either amino acid or peptide binding with impacts on both Ca 2ϩ i mobilization and cAMP signaling pathways. The impact on other CaR-linked signaling pathways remains to be determined. Although these results do not allow a firm conclusion to be drawn regarding the relationship between the binding sites for L-amino acids and glutathione analogs, molecular modeling suggests that the S-methylglutathione binding site in the CaR (supplemental Fig. S2) is closely related to that previously de-scribed for the amino acid L-glutamate in mGlu-1 (16), mGlu-3, and mGlu-7 (17). Thus, L-amino acids and glutathione analogs, distinct from Ca 2ϩ o and R-467, act via a common mechanism as revealed by the selective impacts of the T145A/S170T double mutant and are likely to occupy overlapping binding sites in the receptor VFT domain. or the double mutant T145A/S170T CaR (filled symbols) were loaded with fura-2AM and transferred to the light path of a microfluorescence apparatus as described under "Experimental Procedures." After a short pre-equilibration at 0.5 mM, the Ca 2ϩ o concentration was elevated to 2.5 mM. Panel A shows the L-Phe concentration-response relations for the wild-type CaR (open triangles; n ϭ 3) and CaR double mutant T145A/S170T (filled triangles; n ϭ 5). The difference in the maximal responses was statistically significant (p ϭ 0.03). Panel B shows the SMG concentration-response relations for the wild-type CaR (open squares; n ϭ 3) and CaR double mutant T145A/S170T (filled squares; n ϭ 5). The difference in the maximal responses was statistically significant (p ϭ 0.02). FIGURE 9. Impacts of extracellular Ca 2؉ , L-Phe and S-methylglutathione on cAMP levels in HEK-CaR cells and HEK-CaR-T145A/S170T cells. HEK-CaR cells (panels A and B) or HEK-CaR-T145A/S170T cells (panel C) were transiently transfected with the cAMP FRET reporter CFPnd-EPac1-cpVenus as described under "Experimental Procedures." Intracellular cAMP levels (means Ϯ S.E.) were elevated using the cell-permeable adenylyl cyclase activator forskolin (5 M) and the impacts of various CaR activators including Ca 2ϩ o (0.5-8.0 mM), 10 mM L-Phe and 10 M SMG were then investigated. HEK-293 cells that expressed the wild-type CaR exhibited Ca 2ϩ o -concentration-dependent reversible inhibition of cAMP levels; maximum inhibition was observed at around 6 mM (panel A; n ϭ 4). In these experiments, an overshoot was commonly observed upon return to the control physiological saline solution containing 0.5 mM Ca 2ϩ o and 5 M forskolin. In addition, in the presence of 1.5 mM Ca 2ϩ o , 10 mM L-Phe, or 10 M SMG reversibly suppressed intracellular cAMP levels in HEK-CaR cells (panel B; n ϭ 4). In HEK-CaR-T145A/S170T cells, however, the effects of L-Phe and SMG were markedly attenuated whereas the effect of elevated Ca 2ϩ o was unchanged (panel C; n ϭ 3).
Global and conditional knock-out studies for assessing CaR function have, thus far, identified roles for the CaR in the control of parathyroid hormone secretion (3,4), renal Ca 2ϩ reabsorption (3), and the formation of cartilage and bone (4). However, the realization that an earlier global knock-out strategy, based on the in-frame deletion of exon-5, was incomplete due to residual receptor function (4), together with studies demonstrating diverse pluripotent roles in development and tissue function (review: Ref. 37) suggest that the discrete list above understates the physiological importance of the receptor due to its wide distribution and expression in compartments that differ in their ionic, pH and nutrient compositions (reviews: Refs. 1, 5, 38). The current work, demonstrating that L-amino acids as well as glutathione and its metabolites modulate the CaR by a common mechanism, raises the possibility that the CaR may be modulated primarily by L-amino acids in some compartments and by ␥-glutamyl peptides in others, depending on their relative levels in the compartment concerned. For example, total L-amino acid levels normally exceed 2.5 mM in systemic plasma and are even higher in the gut lumen following the ingestion of proteinrich foods (review: Ref. 19) and recent work has demonstrated the physiological significance of the CaR in luminal amino acid-induced gastrin secretion in vivo (39). However, L-amino acid levels are substantially lower in the cerebrospinal fluid (around 10% of those in plasma) as a result of blood-brain barrier function (1,40). The extracellular levels of glutathione, on the other hand, appear to be higher in the brain, where they can approximate 10 -20 M due to astrocyte-dependent synthesis and release (41,42). These considerations raise the possibility that glutathione and/or a metabolite such as Smethylglutathione may play a specific role in CaR-dependent neuromodulation.
The CaR belongs to GPCR class C, members of which are typified by N-terminal Venus FlyTrap domains linked to heptahelical domains. In mammalian biology, VFT domains are also found in ionotropic, as well as metabotropic, receptors for glutamate. Interestingly, glutathione and analogs including S-methylglutathione and glutathione sulfonate have been previously shown to modulate NMDA and AMPA receptors (43,44). In addition, like the CaR, the NMDA receptor binds multivalent cations including neomycin, spermine, and Mg 2ϩ (45). The current work on the physiological impact of ␥-glutamyl peptides on CaR function strengthens the notion that the CaR and ionotropic, as well as metabotropic, glutamate receptors share conserved VFT domain nutrient binding surfaces and may thus contribute to coordinate control of CNS function in response to a common signal including, perhaps, extracellular glutathione and/or glutathione analogs. Attention has been drawn previously to potential roles for glutathione and its metabolites in the modulation of CNS function (reviews: Refs. 46,47).
In conclusion, we have demonstrated that ␥-glutamyl peptides including glutathione and its S-alkyl derivatives, S-methylglutathione, and S-propylglutathione along, with the di-peptides ␥-Glu-Ala and ␥-Glu-Cys are potent positive allosteric modulators of the calcium-sensing receptor that stimulate Ca 2ϩ i mobilization in CaR-expressing HEK-293 cells and nor-mal human parathyroid cells and powerfully suppress PTH secretion. In addition, they suppress cAMP levels in forskolinstimulated CaR-expressing HEK-293 cells transfected with the cAMP sensor CFPnd-EPac1-cpVenus. Finally, unlike responses to elevated Ca 2ϩ o or the type-II calcimimetic NPS R-467, both ␥-glutamyl peptide-induced Ca 2ϩ i mobilization and suppression of cellular cAMP levels were markedly attenuated in the case of the CaR double mutant T145A/S170T indicating that L-amino acids and ␥-Glu peptides activate the CaR via a common mechanism.