Phosphate acts directly on the calcium-sensing receptor to stimulate parathyroid hormone secretion

Elevated Pi concentrations inhibit the CaSR

We first evaluated the effect of acute increases in Pi concentration in CaSR-transfected HEK-293 cells (CaSR-HEK) by measuring CaSR-induced ${\mathrm{Ca}}_i^{2+}$-mobilization upon stimulation with different concentrations of ${\mathrm{Ca}}_o^{2+}$ (Fig. 1a, b). Then, CaSR-specific positive allosteric modulators (PAMs) were included so that the lowest possible ${\mathrm{Ca}}_o^{2+}$ concentration could be used to achieve CaSR-induced ${\mathrm{Ca}}_i^{2+}$ mobilization, thus reducing the risk of Ca × P association in the presence of increased Pi concentrations. Thus, Fura2-loaded CaSR-HEK cells were incubated in experimental buffer containing 0.5 mM Ca²⁺ plus either spermine (Fig. 1c), NPS-R568, hereafter R568, or cinacalcet, a CaSR PAM used clinically (Fig. 1d–f). Cells were switched to buffers containing 0.8 mM Pi (a normal plasma concentration in humans) and then 2 mM Pi (a pathophysiologic concentration observed in CKD patients) before being returned to experimental buffer (without Pi). CaSR responsiveness at 0.8 mM Pi (i.e., 350/380 Fura2 ratio area under the curve) was not significantly decreased when compared with 0 mM Pi (Fig. 1f), whereas the CaSR response in 2 mM Pi was significantly decreased (by 40%) in the experiments performed, regardless of the CaSR agonist or PAM used (Fig. 1c–f). This inhibition was immediately reversible upon removal of Pi in all cases. Furthermore, the decline in CaSR responsiveness elicited by Pi cotreatment was much greater than that for time-matched controls (Fig. 1f). It should be noted that increasing Pi concentration from 0.8 to 2 mM decreased the free ionized Ca²⁺ concentration by 7%, and the Pi inhibitory effect was still observed upon correction of Ca²⁺ levels (Supplementary Fig. 1).

Carbachol induces non-CaSR-mediated ${\mathrm{Ca}}_i^{2+}$-mobilization in CaSR-HEK cells by activating muscarinic acetylcholine receptors (mAChRs) expressed endogenously in the cells. However, Pi cotreatment failed to alter the response to carbachol suggesting that it elicits a specific inhibitory effect on CaSR-mediated signaling (Supplementary Fig. 2a).

If the apparent inhibitory effect of Pi on CaSR was due solely to buffering ionized Ca²⁺, then lowering extracellular pH should reverse this effect by releasing free Ca²⁺. Therefore, ${\mathrm{Ca}}_i^{2+}$-mobilization experiments were repeated but adding a buffer with 2 mM Pi at pH 7.2 (equivalent to moderate-to-severe metabolic acidosis observed in end-stage CKD patients). The results confirmed that even under acidic conditions, increased Pi concentration resulted in a substantial (60%) reduction of CaSR activity that was immediately reversed upon removal of Pi and return to pH 7.4 (Supplementary Fig. 2b). To confirm that the Pi effect was not readout-dependent, we also quantified the effect of Pi on CaSR-induced extracellular signal-regulated kinase phosphorylation (pERK). Indeed, hyperphosphatemia (2 mM Pi) significantly reduced the maximal CaSR-induced pERK in CaSR-HEK cells by 30% and this inhibitory effect was again maintained in acidic conditions (pH 7.2) (Fig. 2a).

Pi acts as a noncompetitive antagonist of CaSR

According to the recently reported human CaSR-ECD crystal structure²⁷, multivalent anions bind to CaSR at defined sites, distinct from the binding sites described for Ca²⁺ or l-amino acids. To elucidate the molecular mechanism underlying Pi-mediated CaSR inhibition, we first examined the ${\mathrm{Ca}}_o^{2+}$ concentration-effect relationship in the absence and presence of 0.8 and 2 mM Pi by measuring ${\mathrm{Ca}}_i^{2+}$-mobilization in CaSR-HEK cells (Fig. 2b). As before, R568 was included to induce CaSR responses at lower ${\mathrm{Ca}}_o^{2+}$ concentrations in order to reduce the risk of Ca × P association. The CaSR maximal response (E_max) to ${\mathrm{Ca}}_o^{2+}$ was inhibited in the presence of 2 mM Pi (by 32 ± 3%; n = 8) and this inhibitory effect was not overcome by further increases in Ca²⁺ concentration. Interestingly, Pi did not alter ${\mathrm{Ca}}_o^{2+}$ potency (EC₅₀) (Fig. 2b). Thus, Pi inhibited the CaSR via an apparent noncompetitive antagonism, reducing receptor efficacy but not potency. Moreover, the inhibitory effect of Pi was concentration-dependent following stimulation with either 1 or 1.5 mM Ca²⁺ and 1 μM cinacalcet (IC₅₀ 1.21 mM) (Fig. 2c) or with 1.5 mM Ca²⁺ and 1 μM R568 (IC₅₀ 1.27 mM) (Supplementary Fig. 3a). These IC₅₀ values for Pi indicate CaSR sensitivity to Pi both over its physiological range (0.8–1.4 mM), and also sensitivity to the hyperphosphatemia of CKD, indicating the potential pathophysiological relevance of this effect.

Inorganic SO₄ is the other multivalent anion proposed to bind the CaSR ECD according to crystal structure data²⁷. SO₄ may act as a ligand at the same sites as Pi, and therefore we tested whether this anion could also inhibit the CaSR. Indeed, SO₄ inhibited ${\mathrm{Ca}}_i^{2+}$-mobilization in a concentration-dependent manner with a similar potency to Pi (IC₅₀ 1.31 mM) (Fig. 2c). However, as physiological SO₄ concentrations are lower than those observed for Pi (~0.3 mM in serum), SO₄ would appear to be of lesser physiological significance as a CaSR inhibitor. We also tested different Pi concentrations in the presence of 0.3 mM SO₄ (physiological), and the IC₅₀ obtained was not significantly different from that obtained with Pi alone (Supplementary Fig. 3b). Interestingly, the inhibitory concentration-effect curves for both Pi and SO₄ had Hill slopes close to −1, suggesting the existence of a unique inhibitory anion-binding site, or at least the lack of cooperativity between binding sites (Fig. 2 and Supplementary Fig. 3).

Pi increases PTH secretion in human parathyroid cells

Elevated PTH secretion levels together with hyperphosphatemia are major complications of CKD. We measured the direct effect of Pi on PTH secretion from freshly isolated parathyroid cells obtained from samples of healthy human parathyroid tissue. In order to minimize CaSR downregulation, human parathyroid tissue was stored as undigested sliced tissue cubes. Tissue digestion and cell preparation was performed immediately before PTH assay. In addition, at the beginning of each experiment, we evaluated the responsiveness and quality of the specimen by decreasing ${\mathrm{Ca}}_o^{2+}$ from 1.2 to 1 mM, leading to the expected CaSR-mediated rise in PTH secretion (Supplementary Fig. 4). We measured PTH secretion every 2 min in a perifusion model (Ca²⁺, 1.2 mM) in the presence of increasing Pi concentrations (0.8, 1.4, 2, and 3 mM) encompassing the physiological and pathophysiological range. A significant increase in PTH secretion was observed at 2 mM Pi (hyperphosphatemia), +56 ± 10%, with a further increase at 3 mM Pi (severe hyperphosphatemia), +77 ± 13% (Fig. 3, individual traces shown in Supplementary Fig. 4). This stimulatory effect of Pi on PTH secretion was quickly overcome, within minutes, when Pi concentration was returned to 0.8 mM (physiologic). These rapid and reversible effects are consistent with a receptor-mediated response (Fig. 3). In addition, we confirmed that the addition of Pi to the buffer had only a marginal effect on ionized Ca²⁺ concentration (−6% 0.8 vs. 2 mM Pi; Supplementary Fig. 1a). We further confirmed the inhibitory effect of Pi on CaSR-induced ${\mathrm{Ca}}_i^{2+}$ mobilization in human parathyroid cells obtained by collagenase digestion of parathyroid tissue (Supplementary Fig. 5).

Pi increases PTH secretion via the CaSR

To assess whether the rise in PTH secretion observed in hyperphosphatemia was mediated by the CaSR, we compared PTH secretion levels from control and CaSR-knockout (KO Casr) murine parathyroid glands (Fig. 4a). KO Casr mice have a homozygous parathyroid gland-specific deletion of exon 7 (transmembrane domain) of the CaSR. To be considered as full KO Casr, only mice exhibiting (a) high serum PTH levels (greater than tenfold higher than controls), and (b) lack of a response of their parathyroid glands to Ca²⁺ concentration changes (<10% response when changing from 1.2 to 0.8 mM or from 0.8 to 1.6 mM) were included in the results (Supplementary Fig. 6)²⁸. The sera of control and KO Casr mice were collected and assayed for their intact PTH (1–84) content, revealing levels ~17-fold higher than those in control mice, due to the lack of functional CaSR (Fig. 4b). Two parathyroid glands from each mouse were dissected free of the surrounding thyroid and connective tissues and then incubated for 1 h in different buffers containing increasing concentrations of Pi in high and low ${\mathrm{Ca}}_o^{2+}$ (Fig. 4c). Parathyroid glands from the control mice exhibited significantly increased PTH secretion in response to pathophysiological Pi concentrations. In contrast, in glands from the KO Casr mice in which CaSR had been functionally knocked out, Pi-mediated stimulation of PTH secretion was not observed (Fig. 4c).

CaSR anion-binding sites are conserved across species

Four different multivalent anion-binding sites (1–4), hereafter termed Pi-binding sites, were identified in the inactive and active CaSR-ECD structures (Fig. 5)²⁷. We examined the level of amino acid conservation at the predicted Pi-binding sites, by using the phylogenetic family C GPCR alignment of Herberger et al.²⁹. Herberger’s data set comprised 138 nucleotide sequences including 42 CaSR sequences, from different phylogenetic groups. WebLogo revealed higher bit scores at the predicted residues involved in Pi binding in the vertebrate CaSR subgroup, corresponding to strong amino acid conservation, compared with all collected family C GPCR sequences (Fig. 6). The four CaSR-predicted Pi-binding sites are characterized by an abundance of positively charged residues (here represented with blue), with arginine being particularly conserved. Interestingly, predicted binding sites 1–3 are more conserved within the CaSR subgroup as illustrated in the WebLogo analysis (Fig. 6). The CaSR subgroup analysis revealed high conservation of Pi-binding sites across the 42 different species with diverse evolutive origin. This analysis further confirmed that the residues involved in the Pi-binding sites which are present in the inactive conformation, including R62 and E277 (site 1), and R66 and S301 (site 3) are highly conserved (Fig. 6).

To identify the site of Pi-mediated CaSR inhibition, we focused on the two inactive conformation-specific Pi-binding sites 1 and 3 (Fig. 5). In the absence of Pi, R62 and E277 (binding site 1) form a salt bridge across the upper and lower domains of the Venus flytrap that stabilizes the CaSR's active conformation by promoting ECD closure (Supplementary Movie 1). When Pi is present, this salt bridge is broken and the positive charge of R62 is stabilized by direct interaction with the negative charges of Pi. By breaking the salt bridge, Pi displaces E277 on the lower domain, thus reducing the free energy needed to shift the equilibrium toward the inactive (open) conformation of the receptor (Fig. 5). In binding site 3, when Pi is absent, R66 interacts with S301 forming a hydrogen bond across the upper and lower domains that also promotes receptor closure and this stabilizes the CaSR’s active conformation. In contrast, Pi breaks the hydrogen bond by interacting directly with R66 and partially with R69, which also contributes to the inactive conformation by displacing S301 in the lower domain (Fig. 5). In addition, Ca²⁺ binding site 3 stabilizes the loop conformation that supports formation of the interdomain bond between R66 and S301. These features highlight the importance of Ca²⁺ binding site 3 and Pi-binding sites 1 and 3 in regulating CaSR’s activity.

Pi-mediated CaSR inhibition requires residue R62

Based on the in silico analysis, we selected residues R62 and R66 as candidates for mutagenesis to determine their possible contribution to Pi-mediated CaSR inhibition. These residues were mutated separately to alanines (CaSR^R62A and CaSR^R66A) and the CaSR constructs arising were expressed transiently in wild-type HEK-293 cells. Immunoblotting revealed successful expression of the mutated receptors at levels similar to the WT receptor (Supplementary Fig. 7a). We expected decreased sensitivity of CaSR^R62A and CaSR^R66A to ${\mathrm{Ca}}_o^{2+}$ due to likely disturbances of Venus flytrap closure. This was confirmed experimentally whereby the potencies of ${\mathrm{Ca}}_o^{2+}$-induced ${\mathrm{Ca}}_i^{2+}$ mobilization for these mutants were reduced (EC₅₀ for ${\mathrm{Ca}}_o^{2+}$, 3.4 mM WT; 5.2 mM CaSR^R62A; 5.3 mM CaSR^R66A). The ${\mathrm{Ca}}_{\mathrm{o}}^{2+}$ E_max values for these mutants, however, were not altered (Supplementary Fig. 7b).

We next evaluated the effect of pathophysiologic Pi concentrations on CaSR-induced ${\mathrm{Ca}}_i^{2+}$ mobilization in CaSR^R62A and CaSR^R66A. It was found that 2 mM Pi had no effect on CaSR^R62A-induced ${\mathrm{Ca}}_i^{2+}$ mobilization upon stimulation with different concentrations of Ca²⁺ (Fig. 7a) or cotreatment with R568 (Fig. 7b). Due to the mutants’ decreased sensitivity to Ca²⁺, higher Ca²⁺ concentrations were used (1 mM Ca²⁺ for CaSR^R62A and 1.5 mM Ca²⁺ for CaSR^R66A) to achieve equivalent responses to CaSR^WT (0.5 mM Ca²⁺) (Fig. 7b and Supplementary Fig. 9). Similarly, CaSR^R62A was also insensitive to 2 mM SO₄ (Fig. 7c). CaSR^R62A responses were also identical to time-matched controls when exposed to either 1 or 1.5 mM ${\mathrm{Ca}}_o^{2+}$, in the presence of 1 μM R568, with or without Pi (Supplementary Fig. 8a,b). Cell surface receptor expression levels were also evaluated and confirmed to be equivalent for CaSR^WT and CaSR^R62A, and significantly greater than those for the well-known trafficking mutants, E799A and I642Del (Fig. 7d). Next, we measured ${\mathrm{Ca}}_i^{2+}$ mobilization in response to increasing ${\mathrm{Ca}}_o^{2+}$ concentration (0.5–5 mM), in the presence of R568, in CaSR^R62A-stably transfected HEK-293 cells (CaSR^R62A-HEK) and found that E_max values were not altered in the presence of Pi (P > 0.05) (Fig. 7e). We did note that the EC₅₀ values for ${\mathrm{Ca}}_i^{2+}$mobilization in response to ${\mathrm{Ca}}_o^{2+}$ were significantly reduced in the presence of Pi, from 2.6to 1.6 mM (in 0.8 mM Pi) and 1.7 mM (2 mM Pi), showing partial recovery of the loss of function. There were no significant differences between 0.8 and 2 mM Pi responses in either potency (EC₅₀) or efficacy (E_max) (Fig. 7e).

The effect of Pi on CaSR^R62A was also studied over a more extended range of Pi concentrations (0–5 mM) (Supplementary Fig. 8c). Again in CaSR^R62A, physiologically relevant Pi concentrations (0.8–1.4 mM) increased ${\mathrm{Ca}}_o^{2+}$-induced ${\mathrm{Ca}}_i^{2+}$ mobilization, when compared with 0 mM Pi (control). Nevertheless, the inhibitory effect of Pi was absent in CaSR^R62A at any of the Pi concentrations tested (up to 5 mM Pi) (Supplementary Fig. 8c). Next, by using a separate readout of receptor activity increased Pi concentration failed to inhibit pERK in CaSR^R62A-HEK cells, whereas in wild-type CaSR-HEK cells the pERK response was significantly reduced (by up to 30%) as before (Fig. 7f).

In contrast, we found that CaSR^R66A exhibited Pi-induced inhibition of ${\mathrm{Ca}}_o^{2+}$-induced ${\mathrm{Ca}}_i^{2+}$-mobilization that closely resembled responses by CaSR^WT (Supplementary Fig. 9a). In the presence of 2 mM Pi, the maximal response of CaSR^R66A to ${\mathrm{Ca}}_o^{2+}$ (E_max) was reduced significantly (by up to 60%), whereas the EC₅₀ for ${\mathrm{Ca}}_o^{2+}$ was unchanged (Supplementary Fig. 9b). Together, these data implicate R62 (Pi-binding site 1) as the key residue responsible for Pi-induced CaSR inhibition.

Cell culture

HEK-293 cells (from ATCC) were used as a model for the functional expression of CaSR protein. Wild-type HEK-293 and stably transfected CaSR-HEK-293 (CaSR-HEK) cells were grown in Dulbeccos’ modified Eagle’s medium containing 1.8 mM CaCl₂ and 0.9 mM Na₂HPO₄ from Sigma (D5796), and fortified with 10% fetal bovine serum (FBS). Cells were dissociated with trypsin upon reaching 80–90% confluence and grown at 37 °C with 5% CO₂ until passage 30.

ERK activation assay

CaSR-HEK cells were grown on 30-mm dishes and then incubated at 37 °C in experimental buffer containing (mM) 20 HEPES (pH 7.4), 125 NaCl, 4 KCl, 0.5 CaCl₂, 0.5 MgCl₂, and 5.5 glucose with Pi added as Na₂HPO₄ and KH₂PO₄ in a 4:1 ratio (pH 7.4). Following equilibration (20 min), cells were treated for 10 min with experimental buffer including different Pi concentrations and then lysed on ice for 15 min in a RIPA-like buffer supplemented with protease and phosphatase inhibitors ((mM) 12 HEPES (pH 7.4), 300 mannitol, 1 EGTA, 1 EDTA, 1% (v/v) Triton-X 100, 0.1% (w/v) sodium dodecyl sulfate (SDS), 1 NaF, 250 μM sodium pyrophosphate, 100 μM sodium vanadate, 1.25 μM pepstatin, 4 μM leupeptin, 4.8 μM PMSF, and 1 N-ethylmaleimide). Phosphorylation of ERK (pERK) was quantified by semiquantitiative immunoblotting.

Immunoblotting protein analysis

Protein samples were first diluted (4:1) with 5× Laemmli buffer followed by resolution of the proteins by SDS-PAGE and electrophoretic transfer into a nitrocellulose membrane (GE Healthcare Life Sciences). Membranes were incubated in blocking buffer solution containing Tween-TBS solution and 5% (w/v) bovine serum albumin (BSA) for 1 h and probed with primary antibodies for pERK [Phospho-p44/42 MAPK (Erk1/2) (Thr202/Tyr204) (E10) Mouse (1:3000 dilution) from Cell Signaling (Cat. number 9106)], CaSR [Calcium Sensing Receptor monoclonal mouse antibody (5C10, ADD) (1:3000) from Thermo Fisher Scientific (Cat. number MA1-934)], or β-actin [β Actin-Peroxidase monoclonal (AC-15) mouse antibody (1:25,000) from MERCK (Cat. number A3854)] for 1 h. Secondary antibody was horseradish peroxidase-conjugated horse anti-mouse IgG (1:10,000) from Cell Signaling (Cat. number 7076). Chemiluminescence was generated by using BIO-RAD clarity western ECL substrate, detected by using a Chemidoc gel analyzer (BIO-RAD), and quantified by densitometry by using Image Lab 6. Uncropped and unprocessed immunoblot scans shown in Figs. 2a and 7f and in Supplementary Fig. 7a are supplied in the Source Data file.

Site-directed mutagenesis

Alanine mutations were introduced into the human wild-type CaSR gene by using QuickChange Lightning Site-directed mutagenesis kit (Stratagene). Mutations were confirmed by the Genomic Technologies Core Facility at The University of Manchester. HEK-293 cells were then transiently transfected with wild-type or mutant receptors by using FuGENE-6 (Roche Diagnostics). Cells were used experimentally within 72 h of transfection. For stable transfection, vectors were first linearized (Ssp1) and then CaSR-expressing cells were selected by using hygromycin (400 μg/ml).

Ionized calcium measurement

Free ionized Ca²⁺ was measured by using a GEM premier 5000 gas analyzer with an ion-selective electrode by the Clinical Biochemistry department of the Manchester Royal Infirmary.

Intracellular calcium mobilization imaging

Dual-excitation-wavelength microfluorometry was carried out on cells cultured on 13-mm glass coverslips for 48 h in DMEM with 10% FBS. Cells were loaded with 1 μM FURA2-AM in 1.2 mM Ca²⁺ experimental buffer supplemented with 0.1% (w/v) BSA for 1–2 h at room temperature in the dark. Experimental buffer contained (mM) 0.5 Ca²⁺, 20 HEPES (pH 7.4), 118–125 NaCl, 0.5 MgCl₂, 4 KCl, and 5.5 glucose supplemented with Pi as Na₂HPO₄ and K₂HPO₄ added in a 4:1 ratio. Cells were mounted in a perfusion chamber and observed through ×40 oil-immersion lens on a Nikon Diaphot inverted microscope.

Receptor cell surface assay

HEK-293 cells were transiently transfected with pcDNA3.1(+) vector (Life Technologies, USA) containing CaSRs that included a Flag tag inserted after the signal peptide. Cells were then grown in a 96-well plate until 80–90% confluent, then washed with Tween-TBS buffer (15 mM Tris (pH 8.0), 150 mM NaCl, and 0.1% (v/v) Tween 20), and then fixed with either 100% methanol (total expression of CaSR) or 4% paraformaldehyde (receptor cell surface expression levels) for 15 min at 4 °C. Following blocking with 1% skimmed milk (1 h), cells were incubated for 1 h with mouse Anti-Flag antibody H2-peroxidase (A8592, Sigma-Aldrich, USA). Excess antibody was removed by washing with Tween-TBS and 3,3′,5,5′-tetramethylbenzidine (TMB) substrate added with the peroxidase reaction stopped after 10 min by using 1 M HCl. Absorbance was measured at 450 nm.

Origin and preparation of human parathyroid cells

Normal human parathyroid tissue was obtained with consent from patients undergoing total thyroidectomy for major thyroid disease (e.g., large multinodular goiters or Graves' disease) when the surgeon identified and removed one parathyroid gland, subjected it to mincing, and then returned it via a syringe to the ipsilateral sternomastoid muscle, to reduce postoperative hypoparathyroidism⁴⁸. Any parathyroid gland tissue fragments left in the syringe were provided under informed consent and de-identified from the Royal North Shore Hospital (St. Leonards, NSW, Australia) and the Mater Misericordiae Hospital (North Sydney, NSW). Approval for access to human parathyroid tissue was provided by the Human Research Ethics Committee of St. Vincent’s Health Network, Darlinghurst, NSW 2010 (SVH File Number 11/067). We confirm that we have complied with all relevant ethical regulations for research.

The parathyroid specimen was transported in ice-cold Hanks’ balanced salt solution (Invitrogen) containing 1.25 mM CaCl₂ and stored as sliced tissue cubes at 4 °C in MEM (11380-0.37, Invitrogen) for up to 48 h. Digestion of the tissue was performed as follows. First, the specimen was incubated at 37 °C for 20 min in a MEM-oxygenated solution containing 1 mg/ml collagenase (Type I, Worthington, Scimar, Victoria) and 0.2 mg/ml DNase I (Type IV, Sigma). After 20 min, the enzyme solution was decanted and the parathyroid tissue transferred into fresh MEM containing 1 mg/ml bovine serum albumin (BSA, protease-free, Sigma A-3059). The tissue was then triturated by repeated passage through the tip of a disposable 5-ml syringe (without a needle). The cloudy suspension containing clumps of parathyroid cells was filtered with a 200-μm pore-size nylon filter and centrifuged for 2 min at 100 × g (Sorvall H-1000B rotor). The pellet was resuspended and washed twice in 5 ml of MEM supplemented with 1 mg/ml BSA. The remaining undigested tissue was returned to MEM medium containing collagenase and DNase I and incubated for a further 20 min at 37 °C followed by additional trituration and centrifugation as before. The first and second cell harvests were combined for analysis^49,50.

Human PTH assay

Parathyroid cells obtained from human parathyroid tissue were used immediately after tissue disaggregation and digestion. Parathyroid cells (~20,000–50,000) were loaded onto the cell surface of a 1-ml bed volume of Bio-Gel P-4 (nominal exclusion limit 4 kDa) and then gently covered with a 1-ml bed volume of sephadex G-25 (nominal exclusion limit 5 kDa) in a small perifusion column. The column was equilibrated for 20 min by using (mM) 125 NaCl, 4 KCl, 1.25 CaCl₂, 1 MgCl₂, and 20 HEPES (pH 7.4) buffer supplemented with 0.1% d-glucose 1× basal amino acid mixture, 1 mg/ml BSA, as well as with 0.8–3 mM Na₂HPO₄ and K₂HPO₄ added in a 4:1 ratio. Cells were perifused (at 37 °C and 1.5 ml/min) by gel filtration with intact PTH (~9 kDa) detected in the void volume. Tubing connections were established downstream to a peristaltic pump and upstream to a reservoir. Samples were collected every 2 min (3 ml) into chilled tubes and then frozen (−80 °C) until analysis for intact human PTH by using a third-generation, two-site chemiluminescence assay on an Immulite 2000 autoanalyzer (Diagnostic Products Corporation, CA). As required, perifusion solutions were changed to allow variations in Ca²⁺ and Pi concentrations in experimental buffers.

Murine parathyroid gland isolation and ex vivo culture

Mice with homozygous parathyroid gland-specific deletion of exon 7 (encoding the seven-transmembrane domain, three intracellular loops, and intracellular tail) of the CaSR (hereafter termed KO Casr) were generated cross-breeding of floxed Casr mice with mice expressing parathyroid-specific Cre^28,51. Littermates that did not express both transgenes simultaneously were used as controls. Two parathyroid glands from each mouse were dissected free of the surrounding thyroid and connective tissues and submerged in a microdroplet (10 μl) of secretion medium [MEM Eagles with Earle’s balanced salts supplemented with 0.5 mM Mg²⁺, 0.2% bovine serum albumin, and 20 mM HEPES (pH 7.4)] and placed in the middle of a 13-mm track-etched (0.1 μM pore) polycarbonate membrane (Whatman), floating on a large drop (500 μl) of ice-cold secretion media supplemented with 3 mM Ca^2+ 51. The time used to find, dissect, and clean the parathyroid glands from each mouse was <5 min per animal. The dissected glands were equilibrated in fresh secretion media (at 37 °C) for 1 h. Then, each pair of glands were incubated for 30 min in 500 μl of secretion media (37 °C) containing different concentrations of Ca²⁺ and Pi. Intact PTH (1–84) secreted into the bathing culture media was then collected and quantified by ELISA assay (Immunotopics Inc.) and expressed as pg PTH/30 min/gland. Mouse serum was also collected and assayed for intact PTH. Only KO Casr mice displaying high serum PTH levels (more than tenfold higher than control) and whose excised parathyroid glands were insensitive to ${\mathrm{Ca}}_o^{2+}$ are reported. Breeding of parathyroid cell-specific CaSR KO mice and procedures for isolation and ex vivo culture of their parathyroid glands have been approved (Protocol 18-017) by the Institutional Animal Care and Use Committee (IACUC) at San Francisco Veterans Affairs Medical Center. We confirm that we have complied with all relevant ethical regulations for animal testing and research.

Amino acid conservation analysis

A data set of family C GPCR alignments was generated by Herberger et al.²⁹ and included 138 sequences, with 42 of them comprising the CaSR subgroup. Exploratory data analysis, data visualization, and database management was performed with R-studio by using the “seqinr” package^52,53. Weblogo (version 3.0⁵⁴) was used to visualize amino acid conservation in the four Pi-binding sites predicted by Geng et al.²⁷. Sequence conservation by position was assessed by bit-score calculation for each amino acid residue position in the four Pi-binding sites, with the relative frequency of each amino acid displayed pictorially by height of letter.

In silico study of CaSR-ECD anion-binding sites

CaSR-ECD original structures were obtained from Protein Data Bank (PDB) by using the accession numbers 5K5S (active conformation) and 5K5T (inactive conformation). Anion-binding interactions were studied by using PyMol Molecular Graphics System, Schrodinger⁵⁵.

Data analysis

Changes in Fura2 ratio (340/380) were measured with MetaFluor imaging system and assessed by determining the area under the curve by using the R-studio “PKNCA Perform Pharmacokinetic Non-Compartmental Analysis” package^52,56. Responses were measured for 190 s, but excluding the first 40–50 s to avoid change artifacts. Densitometry of immunoreactive bands was performed by using Image Lab 6.0 2D software (BIO-RAD). GraphPad Prism 7 (GraphPad Software, San Diego, CA, USA) was used for nonlinear regression analysis, mean, and SEM. Orthosteric concentration−response curves were normalized to 100% as the maximum orthosteric agonist/antagonist response. Normalized data were fitted to a four-parameter concentration response for ${\mathrm{Ca}}_i^{2+}$-mobilization with Hill Eq. (1), where Top and Bottom are the maximal and the minimal asymptotes of the concentration–response curve.

Statistical analysis

Statistical differences between the EC₅₀ and E_max in the presence and absence of Pi were determined by an extra sum-of- squares F-test. A one-way ANOVA with Dunnett’s multiple comparisons post test was used to determine statistical differences between EC₅₀ and E_max. A one-way repeated measure ANOVA, with Dunnett’s for normally distributed data or Friedman Dunn’s for non-normally distributed data, was used to determine statistical differences between 0, 0.8, and 2 mM Pi in ${\mathrm{Ca}}_i^{2+}$-mobilization, PTH secretion, and pERK assays. Statistical differences between 0.8 and 2 mM Pi in pERK in WT and R62A were determined by Student’s t test. Data are presented as mean ± SEM or median with range, except for EC₅₀ and IC₅₀ values, which are presented as mean 95% confidence interval, and P values <0.05 were considered statistically significant. Replicates are reported either as n for the number of coverslips used per experiment, or as N for the number of independent experiments or animals.

Reporting summary

Further information on research design is available in the Nature Research Reporting Summary linked to this article.