Paradoxical Block of Parathormone Secretion Is Mediated by Increased Activity of Gα Subunits

The paradox of blunted parathormone (PTH) secretion in patients with severe hypomagnesemia has been known for more than 20 years, but the underlying mechanism is not deciphered. We determined the effect of low magnesium on in vitro PTH release and on the signals triggered by activation of the calcium-sensing receptor (CaSR). Analogous to the in vivosituation, PTH release from dispersed parathyroid cells was suppressed under low magnesium. In parallel, the two major signaling pathways responsible for CaSR-triggered block of PTH secretion, the generation of inositol phosphates, and the inhibition of cAMP were enhanced. Desensitization or pertussis toxin-mediated inhibition of CaSR-stimulated signaling suppressed the effect of low magnesium, further confirming that magnesium acts within the axis CaSR-G-protein. However, the magnesium binding site responsible for inhibition of PTH secretion is not identical with the extracellular ion binding site of the CaSR, because the magnesium deficiency-dependent signal enhancement was not altered on CaSR receptor mutants with increased or decreased affinity for calcium and magnesium. By contrast, when the magnesium affinity of the Gα subunit was decreased, CaSR activation was no longer affected by magnesium. Thus, the paradoxical block of PTH release under magnesium deficiency seems to be mediated through a novel mechanism involving an increase in the activity of Gα subunits of heterotrimeric G-proteins.

Parathormone (PTH) 1 secretion from the parathyroid gland is suppressed by high extracellular calcium and magnesium (1). The calcium-sensing receptor (CaSR) is responsible for the calcium-dependent inhibition of PTH secretion (2). Direct binding of calcium or magnesium activates the CaSR (3). Activation of the CaSR triggers G␣ q /G␣ i -mediated signaling pathways (4). Several mutations have been identified with increased activation of this receptor (5,6). CaSR mutants with increased affinity/potency for the agonist calcium and in part enhanced constitutive activity led to permanent inhibition of PTH secretion (7). Therefore, patients with activated CaSR mutants suffer from hypoparathyroidism. A similar phenotype of blunted PTH secretion is seen in patients with severe magnesium deficiency (8 -10). This finding is unexpected since the effects of high magnesium on parathyroid hormone secretion are similar to those of calcium, and therefore, low magnesium should be expected to result in increased PTH secretion. And indeed, in contrast to patients, rats respond to severe hypomagnesemia with increased secretion of PTH (11,12). It is known that hypomagnesemia reflects intracellular magnesium deficiency (9). Thus, the site of magnesium action has been assumed to lie intracellularly (9). However, causality between blunted PTH secretion and magnesium deficiency is not established, although the magnesium paradox has been known for more than 20 years (8). In search for the mechanism we investigated the relationship between magnesium deficiency, PTH secretion, and CaSR-mediated signaling. We present evidence that increased activity of G␣ subunits leads to enhanced signaling mediated by constitutive activation of the human CaSR. Enhanced CaSR-mediated signaling may thus constitute the link between severe hypomagnesemia and blunted PTH secretion.

EXPERIMENTAL PROCEDURES
Cell Culture and Transfection-Human embryonic kidney cells (HEK-293) were cultivated and transfected with plasmids encoding the wild-type and the different CaSR mutants (7,(13)(14)(15) under the control of the cytomegalovirus promotor as described (16).
Construction of G␣ i and CaSR Mutants-For mutagenesis of G␣ i we used the Escherichia coli expression vector pQE60 containing the G␣ i1 -cDNA with an internal nucleotide sequence after amino acid 121 encoding a hexahistidine tag. The codon coding for arginine 209 of G␣ i1 was exchanged to cysteine by site-directed mutagenesis (16). For mutagenesis of the human CaSR, an additional restriction site for XhoI was introduced (silent mutation of leucine 276). The mutated cDNAs were sequenced entirely to confirm the identity of the mutants.
Expression and Purification of G␣ s and G␣ i1 -The G␣ s and G␣ i1 proteins (wild-type and R209C) were expressed in E. coli BL21(DE3) and purified by Ni 2ϩ -agarose according to the manufacturer's protocol (Qiagen). Purified proteins were desalted and concentrated by centrifugation through a Centricon concentrator with an exclusion limit of 30,000 Da (Amicon). After the addition of 20% (v/v) glycerol, the purified proteins were frozen in liquid nitrogen and stored in aliquots at Ϫ80°C at a protein concentration of 2-10 g/l.
Purification of G␣ o from Bovine Brain-G␣ o from bovine brain was prepared as described (17).
Expression of G␣ q in Sf9 Cells -G␣ q was expressed in Sf9 cells together with G␤ 1 ␥ 2 using recombinant baculoviruses. Forty-eight hours after infection, cells were harvested, and membranes of baculovirus-infected cells were prepared as described (16). Membranes were directly assayed for [ 35 S]GTP␥S binding in comparison to control membranes expressing G␤ 1 ␥ 2 alone, since G␣ q was not stable during further purification.
Preparation of HEK-293 Membranes-Membranes of HEK-293 cells transfected with the cDNA encoding the human CaSR were prepared at 4°C by sucrose density gradient centrifugation as described (18). The membrane pellet was treated with 5 M urea, washed, and stored at a protein concentration of 0.1-0.5 g/l at Ϫ80°C .
Determination of [ 35 S]GTP␥S Binding to G␣ Subunits-Basal and receptor-stimulated binding of [ 35 S]GTP␥S to G␣ i , G␣ s , G␣ o , and G␣ q was determined as described (18,19). Briefly, membranes of HEK-293 cells (control or CaSR-transfected, 5-10 g of protein/assay in 50 l) were reconstituted with the indicated G␣ subunits (40 nM) and bovine brain G␤␥ (100 nM) for 30 min on ice in reaction buffer (25 mM Hepes, pH 7.4, 100 mM NaCl, 1 mM dithiothreitol, 160 nM GDP, and MgCl 2 as indicated). Similar results were obtained with MgSO 4 . For determination of basal guanine nucleotide binding to G␣, membranes and G␤␥ were omitted. The experiment was started by the addition of [ 35 S]GTP␥S (60 nM; 2 ϫ 10 6 cpm/50 l). The reaction tubes were placed at 25°C, and at different time points, samples (50 l) were withdrawn and passed over nitrocellulose filters followed by three washes with ice-cold washing buffer (20 mM Tris, 100 mM NaCl, 25 mM MgCl 2 , pH 7.4), and filter-bound radioactivity was determined. Determination of Cellular Inositol Phosphate Levels-Inositol phosphate levels of dispersed parathyroid or of HEK-293 cells were determined as described (19) with minor modifications. Before the experiment, cells were washed with buffer (138 mM NaCl, 0.5 mM CaCl 2 , 5 mM KCl, 20 mM Na ϩ -HEPES, pH 7.4) containing the indicated concentration of MgCl 2 and stored in the same buffer for 15 min to equilibrate the cells with the Mg 2ϩ concentration tested. Then 10 mM LiCl was added. After 20 min at 37°C, cellular inositol phosphates were extracted and determined.
Determination of Cellular Levels of Cyclic AMP-Cellular cAMP levels were determined in dispersed parathyroid cells or in HEK-293 cells 48 h after transfection. Before the experiment the cells were washed with buffer (138 mM NaCl, 0.5 mM CaCl 2 , 5 mM KCl, 20 mM Na ϩ -HEPES, pH 7.4) containing the indicated concentration of MgCl 2 . Cells were allowed to equilibrate with the indicated Mg 2ϩ concentration for 15 min. Then buffer with the specified Mg 2ϩ concentration and supplemented with 0.2 mM isobutylmethylxanthine was added, and the experiment was started by the addition of forskolin (10 M), isoproterenol (100 nM), or buffer as a control. After 20 min at 37°C, cellular cAMP was extracted and determined by radioimmunoassay (Immunotech).
PTH Release from Dispersed Human Parathyroid Cells-Human adenomatous or primary hyperplastic parathyroid glands removed during surgery from patients with hyperparathyroidism were immediately placed in ice-cold RPMI-medium. Dispersed parathyroid cells from parathyroid tissue were prepared by digestion with collagenase and DNase similarly as described (20,21). For determination of PTH release, dispersed parathyroid cells (1 ϫ 10 5 ) were equilibrated for 1 h in incubation buffer (138 mM NaCl, 5 mM KCl, 20 mM Na ϩ -Hepes, pH 7.4) supplemented with CaCl 2 and MgCl 2 as indicated. After washing, PTH release was performed for 10 min at 37°C in the specified buffer. The concentration of immunoreactive PTH in the incubation medium was measured by immunoradioactive assay determining intact human parathyroid hormone (Intact PTH, Nichols Institute Diagnostics). Inositol phosphate and cAMP levels were determined in dispersed parathyroid cells similarly as described above.
Measurement of the Intracellular Free Mg 2ϩ Concentration, [Mg 2ϩ ] i -For determination of changes in [Mg 2ϩ ] i , HEK-293 cells in incubation buffer (138 mM NaCl, 5 mM KCl, 1 mM CaCl 2 , 1 mM MgCl 2 , 20 mM Na ϩ -Hepes, pH 7.4) were loaded with 5 M mag-fura-2/AM dissolved in dimethyl sulfoxide with 0.02% pluronic acid for 30 min at 37°C in a humidified incubator. Cells were washed three times with warmed incubation buffer and incubated for 30 min at 37°C to ensure complete deesterification. Cells were finally washed once with fresh incubation buffer supplemented with CaCl 2 and MgCl 2 as indicated. Fluorescence was recorded with a PerkinElmer fluorescence photometer (LS50B) at an emission wavelength of 520 nm and an excitation wavelength alternating between 340 and 380 nm. [Mg 2ϩ ] i was determined from the ratio between 340 and 380 nm as described previously (22).

Effect of Magnesium Deficiency on PTH Release from Human
Parathyroid Cells-In patients with severe hypomagnesemia, PTH secretion is blocked (8 -10). To analyze whether this in vivo paradox is related to the release of PTH from parathyroid cells, we determined the effect of low magnesium on PTH secretion in vitro on dispersed parathyroid cells. As a control for calcium-dependent stimulus-secretion coupling of the parathyroid tissue from patients with hyperparathyroidism, stimulation of the CaSR by calcium was measured. Calcium as a CaSR agonist blocked the release of PTH with an EC 50 of 1.5 Ϯ 0.2 mM (n ϭ 4) when the Mg 2ϩ concentration was 1 mM (Fig.  1A). This EC 50 value is in close correlation with the set-point values reported previously for human adenomas or primary hyperplasias (23,24). Severe magnesium deficiency was simulated by decreasing the extracellular Mg 2ϩ to 0.1 mM. When Mg 2ϩ was decreased, PTH secretion was blocked independently of the extracellular Ca 2ϩ concentration (Fig. 1A). These findings resemble the in vivo situation in patients; magnesium deficiency blocks PTH release, leading to concomitant hypocalcemia (8 -10), and calcium replenishment cannot overcome this inhibition of PTH secretion (9). The IC 50 value of magnesium for the inhibition of PTH release was 0.25 Ϯ 0.04 mM, and inhibition of PTH release is not seen before the Mg 2ϩ concentration falls below 0.5 mM (Fig. 1B). This finding correlates again with in vivo data demonstrating that the magnesium levels in patients that induce PTH secretion block are generally very low, with concentrations varying between 0.4 and 0.1 mM (8 -10).
Enhanced Second Messenger Generation under Magnesium Deficiency-PTH secretion is inhibited by higher concentrations (Ͼ1 mM) of Ca 2ϩ and Mg 2ϩ , which are agonists of the CaSR. Two different signaling pathways are activated by the CaSR, generation of inositol phosphates and inhibition of cAMP. Under magnesium deficiency, both signaling pathways were enhanced in dispersed parathyroid cells (Fig. 1, C and D). Generation of cAMP was inhibited to about 80 -85% that of the maximum inhibition observed after stimulation with the CaSR agonist calcium ( Fig. 1C and not shown), and basal inositol phosphates were increased to 30 -35% that of the maximum stimulation by magnesium or calcium (Fig. 1D). The IC 50 values of the magnesium deficiency-dependent signal enhancement were similar as for the PTH secretion, i.e. 0.19 Ϯ 0.03 mM for the cAMP pathway and 0.2 Ϯ 0.02 mM for the inositol phosphate production (Fig. 1, C and D). These findings demonstrate that magnesium deficiency affects similar signaling pathways as those activated by the CaSR.
Desensitization of CaSR-mediated Signaling Affects PTH Release under Magnesium Deficiency-Is the CaSR necessary for mediating the effect of magnesium deficiency on PTH release? To address this question, we desensitized the CaSR on parathyroid cells by prolonged calcium stimulation. Desensitization of the CaSR was visible by a substantially decreased response to the agonist magnesium (6 mM) and calcium (2 mM), i.e. after desensitization, magnesium and calcium inhibited PTH release by only Ϸ10% (Fig. 2, lower panel) compared with a Ϸ60% inhibition of PTH release in nondesensitized control cells (Fig. 2,upper panel). Interestingly, the effect of magnesium deficiency on PTH release was suppressed similar to the response to the CaSR agonists ( Fig. 2, lower panel). This finding strongly suggests that active cell surface-localized CaSRs are a prerequisite for the effect of magnesium deficiency on PTH release.

Effects of Low Magnesium on Inositol Phosphate Generation Mediated by Basal Activation of Recombinantly Expressed
Human CaSR-Since the previous experiments suggested that the site of magnesium action lies within the axis CaSR-Gprotein-effector, we further analyzed the mechanism of the magnesium paradox in a transfected cell system. Direct effects of magnesium on the signaling of the CaSR were analyzed compared with control cells without this receptor. Basal inositol phosphate levels of HEK-293 cells expressing the CaSR increased up to 2.8-fold when the Mg 2ϩ concentration in the buffer was decreased from 1 to 0.1 mM, whereas mock-transfected cells without CaSR expression did not show any significant increase in basal inositol phosphate levels under these conditions (Fig. 3A). The IC 50 value of magnesium was 0.18 Ϯ 0.02 mM. The CaSR was responsible for the increase in basal inositol phosphate levels under magnesium deficiency, since basal inositol phosphate levels increased with increasing CaSR expression levels (Fig. 3B). Thus, the effect of magnesium deficiency on basal activation of recombinantly expressed human CaSR paralleled the effects of magnesium deficiency on parathyroid cell signaling.
CaSR-mediated Inhibition of Adenylyl Cyclase-CaSR-mediated inhibition of adenylyl cyclase activity was determined. In HEK-293 cells, forskolin-stimulated adenylyl cyclase activity was inhibited by the CaSR. Maximal inhibition was obtained with 5 mM Ca 2ϩ in buffer with 0.5 mM or 0.1 mM Mg 2ϩ (not shown). Low magnesium alone resulted in partial inhibition (Fig. 3C). The CaSR-mediated inhibition of cAMP levels at 0.1 mM Mg 2ϩ was 64 Ϯ 8% that of the maximum inhibition by 5 mM Ca 2ϩ (not shown). The IC 50 value of magnesium was 0.25 Ϯ 0.05 mM (Fig. 3C). In contrast, the forskolin-stimulated cAMP levels of mock-transfected control cells were barely affected by a decrease in magnesium (Fig. 3C). Together these findings demonstrate that the G␣ i -and G␣ q -mediated pathways triggered by constitutive activity of the CaSR were affected by magnesium deficiency in HEK-293 cells similarly as in dispersed parathyroid cells (cf. Fig. 1).
Effect of Magnesium on the Activation of a CaSR Mutant with Altered Ion Binding Properties-Magnesium is an agonist of the CaSR. Therefore we asked whether the extracellular magnesium binding site(s) of the CaSR is (are) involved in mediating the signal enhancement under magnesium deficiency. Different CaSR receptor mutants were expressed ( 4A). Similar expression levels of the different CaSR mutants were verified in immunoblot (Fig. 4A). Although CaSR R185Q has a nearly 10-fold decreased potency for calcium (data not shown and Ref. 15) and magnesium (Fig. 4B, panel II, and not shown), the IC 50 value of magnesium in inhibiting basal receptor activation was similar between cells expressing mutated CaSR R185Q and the wild-type CaSR (Fig. 4B, panel I). This finding strongly suggests that the extracellular ion binding site of the CaSR for calcium and magnesium is not involved in mediating the signal enhancement under low magnesium.
Magnesium Deficiency Increases Signaling Mediated by the Basal Activity of Different CaSR Mutants-CaSR mutants with altered basal activity were analyzed. Basal activity of CaSR F128L was increased 1.6-fold compared with the wild-type receptor (Fig. 4B, panel IV). The maximum signal enhancement observed under magnesium deficiency (0.1 mM) in CaSR F128L -expressing cells was also increased 1.5-1.8-fold compared with the wild-type receptor (Fig. 4B, panel III), whereas the IC 50 value of magnesium in inhibiting basal CaSR F128L activation was similar between the wild-type CaSR and the mutated CaSR F128L (Fig. 4B, panel III). As a control, the EC 50 value for the activation of CaSR F128L by magnesium was 3.5 Ϯ 0.2 mM (Fig. 4, panel B, panel IV). Thus, signaling mediated by the basal activity of the wild-type CaSR and of CaSR F128L is enhanced under magnesium deficiency with similar IC 50 values, but the absolute extent of this enhancement is different and depends on the basal activity of the respective CaSR.
To further analyze whether basal CaSR activity determines the absolute extent of the signal enhancement by low magnesium, we coexpressed wild-type CaSR together with mutant CaSR R795W , resulting in receptor heterodimers with altered functional properties (18,25), e.g. decreased basal activity. CaSR R795W displays defective G-protein-coupling. Coexpression of CaSR R795W with wild-type CaSR decreased the affinity/ potency for calcium (not shown) or magnesium about 2-fold (Fig. 4B, panel IV). In parallel, the signal generated by the basal activity of CaSR-CaSR R795W heterodimers was suppressed to 20 -25% that of the signal of the wild-type receptor at 0.5 mM and at 0.1 mM Mg 2ϩ (Fig. 4B, panel III). By contrast, the IC 50 value of the magnesium decrease-dependent signal enhancement was not altered. Thus, a decrease in the basal activity of the CaSR is accompanied by a decrease in the absolute extent of the signal enhancement under magnesium deficiency but no change in the IC 50 value of magnesium. Together these findings confirm that magnesium deficiency enhances signaling mediated by the basal activity of the CaSR.
Comparison of the Signals Generated by the Basal Activity of the Rat and the Human CaSR-Is the basal activity of the CaSR related to the magnesium paradox of blunted PTH secretion in vivo? Since the paradox of blunted PTH secretion under severe magnesium deficiency has been observed in patients but not in rats (11,12), we compared the basal activity of the rat and the human CaSR. With 0.5 mM Mg 2ϩ , the increase in basal inositol phosphate levels of rat CaSR-expressing cells was only 15-20% that of the increase of cells expressing the human CaSR (Fig. 4B, panels V and VI). Interestingly, a decrease in the extracellular magnesium also led to an enhancement of the signaling mediated by basal rat CaSR activity, with an IC 50 value for magnesium similar to that of the human CaSR (Fig. 4B, panel V). However, the absolute extent of the signal enhancement of the rat CaSR at 0.1 mM Mg 2ϩ was again only 15-20% that of the human CaSR (Fig. 4B, panel V). Equally effective levels of the human and the rat CaSR were expressed in these experiments, as determined by maximum stimulation with 10 mM Mg 2ϩ (Fig. 4B, panel VI) and by similar concentration response relationships with EC 50 values of 2.8 Ϯ 0.3 and 5.6 Ϯ 1 mM for calcium (not shown) and magnesium (Fig. 4B), respectively. Together these experiments demonstrate that differences in PTH secretion under magnesium deficiency reported for patients and rats correlate with differences in the basal activity of the respective CaSRs and with differences in the absolute extent of the signal enhancement under magnesium deficiency. The findings that the IC 50 values of magnesium for the signal enhancement were similar between the human and the rat CaSR and between different CaSR mutants with different basal activity or with different affinity for extracellular magnesium binding suggest a common magnesium binding site different from the extracellular ion binding site of the CaSR.
CaSR-mediated Activation of G-proteins-Since the extracellular magnesium binding site of the CaSR did not seem to be involved in the magnesium paradox, we asked whether magnesium acted on the intracellular side of the CaSR, at the CaSR-G-protein interface. To address this question, CaSRmediated activation of G-proteins was determined by the receptor-stimulated enhancement of [ 35 (Table I and Fig.  5A). The IC 50 value for the magnesium-dependent inhibition of CaSR-stimulated GTP␥S binding was 0.18 Ϯ 0.04 mM (Table I). Thus, magnesium suppresses the CaSR-mediated G-protein activation and the CaSR-stimulated activation of G␣ q -and G␣ i -coupled pathways with similar concentration-response relationships. These findings also demonstrate that magnesium acts within the axis receptor-G-protein.
CaSR-mediated G-protein Activation of G␣ i R209C -Is the magnesium binding site responsible for this inhibition of basal CaSR activation localized on the CaSR or on G␣? Since the IC 50 value of magnesium for the inhibition of basal CaSR activation was not affected by CaSR mutants with increased or decreased affinity for extracellular magnesium or calcium (cf. Fig. 4), we focused on the magnesium binding site of the G␣ subunit (26,27). Although magnesium effects on the activation of G␣ subunits have not been detected on G␣ i/o isolated from bovine brain (28), the fact that magnesium suppresses the guanine nucleotide exchange of small GTP-binding proteins (29) suggests the possibility of a similar mechanism for G␣ subunits. And indeed, on wild-type G␣ i , magnesium inhibited basal guanine nucleotide exchange, with an IC 50 of 0.18 mM (Table II). By contrast, basal guanine nucleotide exchange of a G␣ i mutant, G␣ i R209C with impaired magnesium binding (30) was not affected by 1 mM Mg 2ϩ (Table II) (Fig. 5B). Also, the t1 ⁄2 were similar at 0.1 and 1 mM Mg 2ϩ (4.8 Ϯ 0.3 and 4.5 Ϯ 0.4 min, Table I). Together these data are compatible with the concept that the magnesium binding site re-sponsible for the enhancement of CaSR-mediated G-protein activation under low magnesium is localized on the G-protein.
Determination of the Intracellular Free Mg 2ϩ Concentration at Different Extracellular Mg 2ϩ Concentrations-Since the previous experiments suggested that the magnesium binding site responsible for the enhancement of CaSR-mediated G-protein activation was localized intracellularly on the G␣ subunit, we asked whether a decrease in the extracellular free Mg 2ϩ concentration, [Mg 2ϩ ] e , was followed by a decrease in the intracellular free Mg 2ϩ concentration, [Mg 2ϩ ] i . HEK-293 cells were loaded with magfura-2, and [Mg 2ϩ ] i was determined. The [Mg 2ϩ ] i of resting HEK-293 cells incubated in buffer with 1.0 mM Mg 2ϩ was 0.7 Ϯ 0.06 mM (Fig. 6A). This value is in good agreement with the [Mg 2ϩ ] i of resting vascular smooth muscle cells (22). By contrast, when the HEK-293 cells were incubated for 20 min in buffer with 0.1 mM Mg 2ϩ , the [Mg 2ϩ ] i was reduced to 0.15 Ϯ 0.03 mM (Fig. 6B). This finding demonstrates that a decrease in [Mg 2ϩ ] e leads to a concomitant decrease in [Mg 2ϩ ] i . Upon the addition of 1.0 mM extracellular Mg 2ϩ , the [Mg 2ϩ ] i of magnesium-depleted HEK-293 cells increased to the initial value within 15 min (Fig. 6C). These experiments are in agreement with previous data demonstrating that alterations in [Mg 2ϩ ] e are accompanied by concomitant alterations in [Mg 2ϩ ] i (31). Considering that effects of magnesium on guanine nucleotide exchange were observed with a Mg 2ϩ concentration below 1 mM (cf. Fig. 5, Tables I and II), the observed decrease in [Mg 2ϩ ] i upon a decrease of [Mg 2ϩ ] e is sufficient to mediate effects on guanine nucleotide exchange of Gproteins in intact cells.
Inhibition of G␣ i and the Release of PTH from Parathyroid Cells-To further analyze whether G-proteins are indeed involved in mediating the magnesium paradox of blunted PTH secretion, we determined the effect of G␣ i/o -protein inhibition on parathyroid cells (Fig. 7). Pertussis toxin treatment abolished the effect of high magnesium (6 mM) in suppressing PTH release (Fig. 7, B versus A). Pertussis toxin also abolished the effect of low magnesium (0.1 mM) on PTH release (Fig. 7B). These findings demonstrate that G␣ i/o proteins are essentially involved in mediating the magnesium paradox of PTH release.
Pertussis Toxin Treatment Reveals Magnesium Effects of G␣ s -mediated Signaling-The magnesium binding site on the G␣ subunit that seems responsible for the enhancement of CaSR-mediated signaling under magnesium deficiency is conserved between various G␣ subunits (26). In accordance with this fact, GTP␥S binding to various G␣ subunits was inhibited by magnesium with similar concentration response relationships (Table II). However, in parathyroid cells, magnesium deficiency selectively enhanced CaSR-mediated inhibitory signaling, whereas the stimulation of PTH secretion by isoproterenol was not enhanced under magnesium deficiency (Fig. 7C). Since G␣ i/o proteins are the most abundant G␣ subunits in the cell, we asked whether inhibition of G␣ i/o proteins could reveal effects of low magnesium on G␣ s -mediated signaling. And indeed, when G␣ i/o proteins were blocked by pertussis toxin,   isoproterenol-stimulated PTH release was slightly but significantly enhanced under 0.1 mM Mg 2ϩ compared with 1 mM or 6 mM Mg 2ϩ (Fig. 7D). Thus, in parathyroid cells the enhancement of G␣ s -mediated signaling under magnesium deficiency is overcome by the increased activity of G␣ i/o proteins. Effect of Magnesium Deficiency on Signals Generated by Activation of Different G␣ q/i -coupled Receptors-If the magnesium binding site responsible for the enhancement of CaSRmediated signaling under magnesium deficiency is localized on the G␣ subunit, signals mediated by the basal activity of receptors other than the CaSR should be similarly enhanced under magnesium deficiency. To analyze whether magnesium deficiency enhances the basal activity of G␣ q/i -coupled receptors in intact cells, we determined the effect of magnesium deficiency on signals generated by the basal activity of two different G␣ q/i -coupled receptors, the angiotensin II AT 1 and the bradykinin B 2 receptor. Both receptors were expressed at high levels in HEK-293 cells to increase signaling mediated by basal receptor activity. Inositol phosphate levels stimulated by basal receptor activity of the AT 1 and the B 2 receptor were increased 2-3-fold under low magnesium (0.1 mM) compared with 1 mM Mg 2ϩ (Table III). Since extracellular magnesium is not an agonist of the AT 1 and the B 2 receptor as of the CaSR, we finally asked whether signals generated by agonist-mediated receptor activation were enhanced similarly under magnesium deficiency as were the signals generated by constitutive receptor activity. And indeed, signaling of angiotensin II and bradykinin was enhanced under magnesium deficiency (0.1 mM Mg 2ϩ ) compared with 1 mM Mg 2ϩ as detected by a 2-3-fold decrease in the EC 50 values of angiotensin II and bradykinin (Table III). Together these experiments demonstrate that signaling mediated by activation of different G␣ q/i -coupled recep-tors is enhanced under magnesium deficiency. Furthermore, these data are in agreement with the newly identified mechanism of magnesium suppressing the guanine nucleotide exchange of G␣ subunits in intact cells.

DISCUSSION
Cloning of a CaSR-cDNA (3) and the identification of several CaSR mutants from patients with hypo-and hyperparathyroidism (1, 5-7) clearly established the mechanistic relationship between serum calcium and magnesium and PTH secretion. Despite this progress, the paradox of blunted PTH secretion under magnesium deficiency is still an open question. Previous work suggested that the target of magnesium action was intracellular (9). Since PTH synthesis was not affected by magnesium deficiency (32) and PTH levels rose within minutes after parenteral magnesium replacement (33), the defect was pinpointed to the level of PTH secretion.
And indeed, we found that in vitro PTH secretion from parathyroid cells was blocked under low magnesium similarly to that reported in patients. Since the effect of low magnesium was dependent on the axis CaSR-G-protein, i.e. desensitization of the CaSR or inhibition of G␣ i -proteins by pertussis toxin also suppressed the effect of magnesium deficiency on PTH secretion, we reconstituted this system of CaSR-G-protein in vitro in a cell line other than the parathyroid cell. The characteristics of the magnesium deficiency-mediated signal enhancement of basal CaSR activity were similar in parathyroid and in CaSRexpressing HEK-293 cells. The magnesium binding site was localized on the G␣ subunit; the presence of a G␣ i mutant with decreased affinity for magnesium abolished the effect of magnesium on CaSR-mediated G-protein activation, whereas on CaSR mutants with increased or decreased affinity for magnesium, the IC 50 value of the magnesium deficiency-mediated signal enhancement did not change. Together these findings let us conclude that the enhancement of G-protein activation under magnesium deficiency enhances signals mediated by constitutive CaSR activity. The absolute extent of this enhancement is greater for the human CaSR than for the rat CaSR because the human receptor has a higher constitutive activity. This difference between the rat and the human CaSR leads to blunted PTH secretion under severe hypomagnesemia in patients but not in rats. Although our data suggest that the magnesium binding site responsible for this effect is located on the G␣, we cannot rule out the possibility that the human CaSR differs from the rat CaSR by an additional yet unknown magnesium binding site acting in concert with the magnesium binding site on the G␣ subunit.
Magnesium-dependent inhibition of guanine nucleotide exchange by stabilizing guanine nucleotide binding is a common feature of small GTP-binding proteins (29). Our finding that magnesium suppresses the release of GDP on G␣ subunits parallels the role of magnesium on c-Ras mechanistically (34).  In contrast to GDP-bound Ras, which has a more than 10-fold higher affinity for magnesium than G␣ (34), the affinity of magnesium for GDP-bound G␣ subunits is in the submillimolar range. Therefore pathophysiological alterations in the magnesium homeostasis will affect selectively the activity of heterotrimeric G-proteins. Guanine nucleotide exchange of G␣ i , G␣ o , G␣ s , and G␣ q was similarly affected by magnesium. Therefore the detected mechanism of magnesium stabilizing the GDPbound G␣ may define a novel function of magnesium on heterotrimeric G-proteins. Why does magnesium deficiency in vivo selectively enhance signaling mediated by the CaSR although low magnesium enhances guanine nucleotide exchange of many different G␣proteins. Several characteristics of receptor-mediated signaling may be responsible for this phenotype. In vivo, G␣ i/o proteins are the most abundant G-proteins. Therefore the enhancement of G␣ i/o -mediated signaling will be predominant. This suggestion is true for parathyroid cells, where magnesium effects on G␣ s -mediated signaling are not detectable unless G␣ i/o proteins are inhibited. The absence of magnesium effects on G␣ s -mediated signaling has also been found for G␣ s -dependent secretion processes other than PTH (35). Concerning G␣ q/11 -mediated signaling, minor or absent effects of magnesium deficiency on G␣ q/11 -stimulated responses in vivo may also be due to enhanced activation of G␣ i/o because many G␣ q/11 -coupled receptor systems are functionally antagonized by G␣ i -mediated responses and vice versa (36 -38). Therefore magnesium deficiencydependent signal enhancement of G␣ q/11 -coupled receptor systems may only become apparent in vitro when compensatory mechanisms are not effective (39) and/or under conditions of overexpression (Table III). By contrast, CaSR-mediated inhibition of PTH release is one of the few physiological systems that needs simultaneous stimulation of G␣ i and G␣ q proteins. Since there exists no functional antagonism of CaSR-mediated signaling, as demonstrated by the phenotype of activating CaSR mutations from patients with hypoparathyroidism (7), magnesium deficiency in vivo may selectively affect CaSRgoverned PTH release due to an increase in G␣ i/o/q/11 activation.