Identification of an l-Phenylalanine Binding Site Enhancing the Cooperative Responses of the Calcium-sensing Receptor to Calcium*

Background: The calcium-sensing receptor (CaSR) is a key mediator of Ca2+ homeostasis in vivo. Results: An l-Phe binding site at the CaSR hinge region globally enhances its cooperative activation by Ca2+. Conclusion: Communication between the binding sites for Ca2+ and l-Phe is crucial for functional cooperativity of CaSR-mediated signaling. Significance: The results provide important insights into the molecular basis of Ca2+ sensing by the CaSR. Functional positive cooperative activation of the extracellular calcium ([Ca2+]o)-sensing receptor (CaSR), a member of the family C G protein-coupled receptors, by [Ca2+]o or amino acids elicits intracellular Ca2+ ([Ca2+]i) oscillations. Here, we report the central role of predicted Ca2+-binding site 1 within the hinge region of the extracellular domain (ECD) of CaSR and its interaction with other Ca2+-binding sites within the ECD in tuning functional positive homotropic cooperativity caused by changes in [Ca2+]o. Next, we identify an adjacent l-Phe-binding pocket that is responsible for positive heterotropic cooperativity between [Ca2+]o and l-Phe in eliciting CaSR-mediated [Ca2+]i oscillations. The heterocommunication between Ca2+ and an amino acid globally enhances functional positive homotropic cooperative activation of CaSR in response to [Ca2+]o signaling by positively impacting multiple [Ca2+]o-binding sites within the ECD. Elucidation of the underlying mechanism provides important insights into the longstanding question of how the receptor transduces signals initiated by [Ca2+]o and amino acids into intracellular signaling events.


Introduction
It has long been recognized that Ca 2+ acts as a second messenger that is released from intracellular stores and/or taken up from the extracellular environment in response to external stimuli to regulate diverse cellular processes. The receptor is present in the key tissues involved in [Ca 2+ ] o homeostasis (e.g., parathyroid, kidney, bone) and diverse other non-homeostatic tissues (e.g., brain, skin, etc.) (8-11). CaSR consists of a large N-terminal extracellular domain (ECD) (~600 residues) folded into a Venus Fly Trap (VFT) motif, followed by a 7-pass transmembrane region (7TM) and a cytosolic C-terminus. The ECD has been shown to play an important role in the cooperative response of the CaSR to [ (15)(16)(17). Several of these naturallyoccurring mutations of CaSR exhibit altered functional cooperativity (15).
Functional cooperativity of CaSR (i.e., based on biological activity determined using functional assays rather than a direct binding assay), particularly the functional positive homotropic cooperative response to [Ca 2+ ] o , is essential for the receptor's ability to respond over a narrow physiological range of [Ca 2+ ] o (1.1-1.3 mM) (3). CaSR has an estimated Hill coefficient of 3-4 for its regulation of processes such as activating intracellular Ca 2+ signaling and inhibiting PTH release. Under physiological conditions, L-amino acids, especially aromatic amino acids (e.g., L-Phe), as well as short aliphatic and small polar amino acids (18) (19,20). In aggregate, the levels of amino acids in human serum in the fed state are close to those activating the CaSR in vitro (19,21) and can further enhance functional cooperativity via positive heterotropic cooperativity. Recently, several groups have reported that the CaSR in cells within the lumen of the gastrointestinal (GI) tract is activated by L-Phe and other amino acids, which have long been recognized as activators of key digestive processes. Hence, the CaSR enables the GI tract to monitor events relevant to both mineral ion and protein/amino acid metabolism in addition to the CaSR's sensing capability in blood and other extracellular fluids (19,22,23). Glutathione and its γ-glutamylpeptides also allosterically modulate the CaSR at a site similar to the L-amino acid-binding pocket but with over 1,000-fold higher potencies (20,24). Thus, CaSR is essential for monitoring and integrating information from both mineral ions/nutrients/polyamines in blood and related extracellular fluids. Nevertheless, we still lack a thorough understanding of the molecular mechanisms by which CaSR is activated by [Ca 2+ ] o and amino acids, which, in turn, regulate CaSR functional positive cooperativity. In addition, in a clinical setting, the molecular basis for the alterations in this cooperativity caused by diseaseassociated mutations is largely unknown due to the lack of knowledge of this receptor's structure and its weak binding affinities for [Ca 2+ ] o and amino acids (13, 15,25,26).
In the present study, we use two complementary approaches--monitoring [Ca 2+ ] i oscillations in living cells and performing molecular dynamic (MD) simulations--to provide important insights into how the CaSR functions and the behavior of the receptor at the atomic level. We first demonstrate that the molecular connectivity between [Ca 2+ ] o -binding sites that is encoded within the key Ca 2+ -binding Site 1 in the hinge region of the CaSR's ECD is responsible for the functional positive homotropic cooperativity in the CaSR's response to [Ca 2+ ] o . We further identify an L-Phe-binding pocket adjacent to Ca 2+binding Site 1. We show that occupancy of this binding pocket by L-Phe is essential for functional positive heterotropic cooperativity by virtue of its having a marked impact on all five of the predicted Ca 2+ -binding sites in the ECD with regard to [Ca 2+ ] o -evoked [Ca 2+ ] i signaling. Furthermore, with MD simulations we show that the simulated motions of Ca 2+ -binding Site 1 are correlated with those of the other predicted Ca 2+binding sites. Finally, the dynamic communication of L-Phe at its predicted binding site in the hinge region with the CaSR's Ca 2+ -binding sites not only influences the adjacent [Ca 2+ ] o binding site 1 but globally (i.e., by exerting effects widely over the ECD) enhances cooperative activation of the receptor in response to alterations in [Ca 2+ ] o .

Computational Prediction of L-Phe-binding
Site and Ca 2+ -Binding Sites from a Model Structure. The structure of the extracellular domain of CaSR (residues 25-530) was modeled based on the crystal structure of mGluR1 (1EWT, 1EWK and 1ISR), and the potential Ca 2+ -binding sites in the CaSR ECD were predicted using MetalFinder (25,27). Prediction of the L-Phebinding site was performed by AutoDock-Vina (28). In brief, the docking center and grid box of the model structure and the rotatable bonds of L-Phe were defined by AutoDock tools-1.5.4. The resultant L-Phe coordinates were combined back to the pdb file of the model structure for input into the LPC/CSU server to analyze interatomic contacts between the ligand and receptor (29). The residues within 5 Å around L-Phe were considered as L-Phe-binding residues.

Measurement of [Ca 2+ ] i responses in single cells
transfected with WT or mutant CaSRs with or without L-Phenylalanine. Measurement of intracellular free Ca 2+ was assessed as described by Huang,et al. (30). Briefly, wild type CaSR or its mutants were transiently transfected into HEK293 cells grown on coverslips and cultured for 48 h. The cells were subsequently loaded for 15 min using 4 μM Fura-2 AM in 2 mL physiological saline buffer (10 mM HEPES, 140 mM NaCl, 5 mM KCl, 1.0 mM MgCl 2 , 1 mM CaCl 2 and pH 7.4). The coverslips were mounted in a bath chamber on the stage of a Leica DM6000 fluorescence microscope, and the cells were incubated in calcium-free physiological saline buffer for 5 mins. The cells were then alternately illuminated with 340 or 380 nm light, and the fluorescence at an emission wavelength 510 nm was recorded in real time as the concentration of extracellular Ca 2+ was increased in a stepwise manner in the presence or absence of 5 mM L-Phe. The ratio of the emitted fluorescence intensities resulting from excitation at both wavelengths was utilized as a surrogate for changes in [ and function. The initial coordinates for all the simulations were modeled from the 2.20 Å resolution x-ray crystal structure of mGluR1 with PDBID 1EWK (31). The AMBER 10 suite of programs (32) was used to carry out all of the simulations in an explicit TIP3P water model (33), using the modified version of the all-atom Cornell et. al. (34) force field and the re-optimized dihedral parameters for the peptide ω-bond (35). An initial 2 ns simulation was performed using NOE restraint during the equilibration in order to reorient the side chains residues in the Ca 2+binding site, but no restraints were used during the actual simulation. A total of three MD simulations were carried out for 50 ns each on the apo-form and ligand loaded forms. During the simulations, an integration time step of 0.002 ps was used to solve Newton's equation of motion. The longrange electrostatic interactions were calculated using the Particle Mesh Ewald method, (36) and a cutoff of 9.0 Å was applied for non-bonded interactions. All bonds involving hydrogen atoms were restrained using the SHAKE algorithm (37). The simulations were carried out at a temperature of 300 K and a pressure of 1 bar. A Langevin thermostat was used to regulate the temperature with a collision frequency of 1.0 ps-1. The trajectories were saved every 500 steps (1ps). The trajectories were analyzed using the ptraj module in Amber 10.

Accelerated Molecular Dynamics Simulation.
Accelerated MD (aMD) was carried out on the free CaSR ECD using the RaMD method (38) implemented in a pmemd module of AMBER on the rotatable torsion. A boost energy, E, of 2000kcal/mol was added to the average dihedral energy and a tuning parameter, α, of 200 kcal/mol was used. The dual boost was also applied to accelerate the diffusive and solvent dynamics as previously described (39). The simulation conditions were similar to that of the normal MD simulations above. Principal Component Analysis was carried out on the trajectories using the ptraj module in AMBER. The directions of the eigenvectors for the slowest modes were visualized using the Interactive Essential Dynamics (IED) plugin (40).
Docking Studies of Phe, Asp, and GLUT. The binding energies for the ligands were calculated using an ensemble-docking method and Autodock vina (28). The ensemble of conformations of CaSR was generated using molecular dynamics simulations as described above. Gasteiger charges were assigned to the ligands and CaSR using the Autodock ADT program. The ligands were flexible during docking to each conformation of CaSR using the following parameters: the grid spacing was 1.0 Å; the box size was 25 Å in each dimension, and the center of the box was chosen as the center of the active site of CaSR, with a large enough space to sample all possible ligand conformations within the box. The maximum number of binding modes saved was set to 10. The conformation with the lowest binding energy was used and assumed to be the best binder. Distributions of the binding energies for each ligand were calculated based on the lowest binding energy of each ligand to each conformation in the ensemble of CaSR conformations.
Principal Component Analysis (PCA). Using the ptraj module of AMBER 10, the Principal Component Analysis (41,42) was performed on all the atoms of the residues that are 5 Å away from Site 1 of CaSR ECD. The covariance matrix of the x, y, and z coordinates of all the atoms obtained from each snapshot of the combined trajectories of the ligand-free CaSR ECD, the Ca 2+ -loaded form, the form loaded with only L-Phe, and the form loaded with both Ca 2+ and L-Phe were calculated. The covariance matrix was further diagonalized to produce orthonormal eigenvectors and their corresponding eigenvalues, ranked on the basis of their corresponding variances. The first three eigenvectors, the Principal Components, which contributed the majority of all the atomic fluctuations, were used to project the conformational space onto them, i.e., along two dimensions.
Statistics. The data are presented as means ± SE for the indicated number of experiments. Statistical analyses were carried out using the unpaired Student's t-test when two groups were compared. A P value of < 0.05 was considered to indicate a statistically significant difference.

Molecular connectivity among predicted calcium-binding sites is required for functional positive cooperativity of CaSR
It has been documented that in several regions of the CaSR and mGluRs the amino acid residues are highly conserved (43). Those conserved elements provide a structural framework for the modeling of the CaSR ECD. Among all the available crystal structures of the mGluRs, studies on mGluR1 give concrete structural information about ligand-free as well as various ligand-bound forms of the receptor. Moreover, CaSR and mGluR1 share similar signaling pathways and can form heterodimers with one another either in vivo or in vitro (44). Thus, the crystal structures of mGluR1 were employed for modeling the CaSR ECD. By using our own computational algorithms, we previously identified five putative Ca 2+ -binding sites in the modeled CaSR ECD (Fig. 1) (25,26,45). Among these, Site 1 is located in the hinge region between the two lobes in the VFT motif. Among 34 newly identified naturally-occurring missense mutations within the ECD, 18 are located within 10 Å of one or more of the predicted Ca 2+ -binding sites (15). Interestingly, a few disease-associated human mutations severely impair the functional cooperativity of CaSR (46) . Functional positive homotropic cooperativity here refers to [Ca 2+ ] o -induced changes in CaSR activity that can be ascribed to interactions between the five predicted Ca 2+binding sites, which are located in different regions of the ECD (47)(48)(49). To understand the observed cooperativity and the origin of changes in cooperativity caused by disease-associated mutations at the atomic level, we have carried out MD simulations on the modeled CaSR ECD to predict correlated motions. MD simulation provides an approach complementary to the experiments in live cells for understanding biomolecular structure, dynamics, and function (50). We calculated the cross-correlation coefficients of each residue with those of all of the other residues in the CaSR ECD from the simulations (Material and Methods). Fig. 1 (lowerright panel) shows the normalized correlation matrix map of both negative (blue) and positive (red) correlated motions between each pair of residues. Negative and positive correlated motions indicate movements in the opposite direction or in the same direction, respectively. Positive correlations occur between groups of residues if they are within the same domain or directly interact with each other. Fig. 1 shows strong correlations amongst residues from S169 to A324. Notably, the negative correlated motions between residues K47-L125 and residues S240-A300 suggest that the two lobes undergo a dynamic change similar to that of mGluR1 upon interactions with its ligands. Closer analysis indicates that residues involved in Ca 2+ -binding Site 1 exhibit negative correlations with residues in Sites 2, 3, 4 and 5 (Table 1).
We then predicted a putative amino acidbinding site in the modeled CaSR based on its sequence homology to mGluR1 and its ligandloaded form using AutoDock-Vina (28). As shown in Fig. 1 (upper-left panel), this potential amino acid-binding pocket, formed by residues K47, L51, W70, T145, G146, S169, S170, I187, Y218, S272, H413 and R415, partially overlaps Ca 2+ -binding Site 1 in the modeled CaSR ECD. Its predicted location is consistent with previous functional studies, suggesting important roles for S170 and T145 in amino acid-potentiated intracellular calcium responses (51)(52)(53). This L-Phe-binding site is also located within the hinge region of the ECD of CaSR with a relatively localized configuration during MD simulations. Other Ca 2+binding Sites (Sites 2-5) are more than 10 Å away from the L-Phe-binding site (26). This partial colocalization of predicted Ca 2+ -binding site 1 and the amino acid-binding site at the hinge domain in the ECD of the CaSR is also observed in other members of the family C GPCRs, including mGluRs and taste receptors, which share some degree of sequence similarity with the CaSR (  (19).
We define functional positive heterotropic cooperativity as that which occurs when the functional positive cooperative effect of interaction with one ligand (e.g., Ca 2+ ) affects the functional response resulting from interaction of a different ligand with the protein (i.e., an aromatic amino acid) (57). This term can be applied in the case of CaSR when it simultaneously senses Ca 2+ and L-Phe. We have also observed greater correlated motions among the multiple Ca 2+binding sites after docking both Ca 2+ and L-Phe compared with docking of L-Phe alone to the ECD domain of the CaSR (Fig. 2c). Taking these results together, we propose that there is molecular connectivity centered at predicted calcium-binding Site 1 that plays an essential role in regulating the correlated motions among the multiple Ca 2+binding sites. Further communication of this site with the amino acid-binding site is likely to mediate functional heterotropic cooperativity of CaSR-mediated signaling, as shown later.

Functional positive homotropic cooperativity among Ca 2+ -binding sites.
Given that it is not readily possible to perform radioligand binding assays on CaSR due to its low affinities for its ligands, especially Ca 2+ (e.g., mM K d ) as well as difficulty in purification of CaSR, we monitored intracellular calcium ( . Such population studies were also reported in our previous studies (25,26). Results from western blot and immunofluorescence staining using an anti-CaSR antibody indicated essentially equivalent expression of WT CaSR as well as its variants on the cell surface (Fig. 5). As shown in Fig. 3a, the level of [Ca 2+ ] o required to initiate oscillations in mutant E297I at predicted Ca 2+ -binding Site 1 or D215I at Site 2 increased markedly from 3.0 ± 0.1 mM to 17.0 ± 0.4 mM and 13.9 ± 0.2 mM, respectively (n>30, p<0.05). Correlating well with these results, the two mutants had significantly impaired responses to [Ca 2+ ] o in the population assay with increased EC 50 values (Table 3). The Hill coefficients in Table 3 and Fig. 3b indicate that the cooperativity among the various Ca 2+binding sites was impaired by mutating each of them separately. Strikingly, removal of Ca 2+binding ligand residues, such as E297I and Y218Q at Site 1, converted the single process for functional activation of the WT CaSR by [Ca 2+ ] o to biphasic functional processes, suggesting that the underlying cooperative binding mechanism had been substantially perturbed (Figs. 3b & 6d).  Table  2). Concurrently, L-Phe also increased the oscillation frequency from 1.5 ± 0.1 to 2.2 ± 0.2 peaks/min (n>30, p<0.05) in the presence of 3.0 mM [Ca 2+ ] o in the single cell assay (Fig. 6c, Table  2). Meanwhile, L-Phe produced functional positive heterotropic cooperativity of the receptor, as it facilitated the response of the WT CaSR to [Ca 2+ ] o by significantly decreasing the EC 50 from 2.9 ± 0.2 mM to 1.9 ± 0.2 mM (n=3, p<0.05) and increasing the Hill coefficient from 3.0 to 4.0 in the cell population assay (Fig. 6d, Table 3).
We then performed detailed analyses to understand the role of residues in the modeled L-Phe-binding site in the functional positive heterotropic cooperativity contributed by L-Phe (Fig. 6, and Table 4). Five out of 12 residues located within 5 Å of the modeled L-Phe-binding site exhibited impaired L-Phe-sensing ability. Mutants L51A and S170T exhibited impaired L-Phe-sensing capability as indicated by the absence of any change in the starting point (Fig. 6b) as well as constant oscillatory frequencies (~1.7 and 1.4 peaks/min) in the presence of L-Phe (Fig. 6c), while they maintained relatively unaltered calcium-sensing functions. Consistent with the single cell assay results, cell population studies revealed that the EC 50 values of L51A, S170T and Y218Q remained the same with or without L-Phe (Fig. 6d) [the effect of L-Phe on S170T has previously been reported by Zhang et. al. in a cell population assay (51)]. Addition of 5 mM L-Phe lowered the [Ca 2+ ] o required to initiate oscillations in cells transfected with mutations S272A or T145A but failed to increase the oscillatory frequency at 2.5 mM [Ca 2+ ] o (the level at which the majority of the cells began to oscillate), nor did it reduce the EC 50 (Table 4, Fig. 7). Y218 is predicted to be involved in binding of both L-Phe in its binding pocket and of Ca 2+ in Site 1. Indeed, the mutation Y218Q largely disrupted the functional positive homotropic cooperativity with transformation of the single cooperative response to [Ca 2+ ] o of the WT CaSR to a biphasic process in the cell population assay. Y218Q also exhibited less sensitivity to [Ca 2+ ] o , as [Ca 2+ ] i oscillations did not start until [Ca 2+ ] o was increased to more than 10 mM, reflecting its role in this Ca 2+ -binding site. Of note, however, addition of 5 mM L-Phe failed to restore the calcium sensitivity of this mutant as manifested by an unchanged oscillation pattern. An oscillation frequency of ~1.5 peaks/min was observed at 20 mM [Ca 2+ ] o both with and without L-Phe for this mutant. In contrast, mutations such as K47A, Y63I, W70L, G146A, I162A, S169A, I187A, H413L, and R415A did not abrogate the positive allosteric effect of 5 mM L-Phe (Table 4). Taken together, these results suggest that residues located at the predicted L-Phebinding site, including L51, T145, S170, S272, and Y218, play key roles in sensing L-Phe. The sensing of L-Phe at the hinge region adjacent to Site 1 has marked global (i.e., extending widely over the ECD) effects on the five predicted Ca 2+ -binding sites predicted earlier, which are spread over several different locations in the CaSR's ECD. Interestingly, addition of 5 mM L-Phe significantly rescued the [Ca 2+ ] i responses of the two mutants, E297I and D215I, in Sites 1 and 2, respectively, that exhibited disrupted cooperativity. Notably, L-Phe converted the biphasic Ca 2+ -response curve for E297I back to a uniphasic curve (Fig. 8). Fig. 8 shows that their starting points for [Ca 2+ ] o -initiated oscillations were reduced from 17.0 ± 0.4 mM to 7.3 ± 0.2 mM (E297I) and from 13.9 ± 0.2 mM to 6.7 ± 0.3 mM (D215I), respectively, in the presence of L-Phe (Table 2). Both of the mutants exhibited more than 2-fold-shifts in their starting points in the presence of L-Phe. The frequencies of their oscillations increased to more than 2 peaks/min in both cases compared to the frequencies without L-Phe ( Fig. 8c and Table 2). Similarly, addition of L-Phe decreased their EC 50 s in the cell population assay (Table 3) 50 was reduced and the maximum response was rescued by 5 mM L-Phe.
E353 is part of Ca 2+ -binding Site 4 in lobe 2, while D398 and E399 are involved in Site 5 (26). Table 3 and Fig. 4 show that removal of these negatively charged residues at Sites 4 and 5 increased the level of [Ca 2+ ] o required to initiate [Ca 2+ ] i oscillations and also the mutant receptors' EC 50 s. L-Phe at 5 mM was able to enhance these mutants' sensitivity to [Ca 2+ ] o and decrease their EC 50 s. Therefore, as shown in Table 2, not only Ca 2+ -binding residues adjacent to the predicted L-Phe-binding site (e.g., Sites 1 and 2) but also sites farther away from the hinge region exhibited L-Phe-induced decreases in EC 50 and starting point as well as increases in oscillation frequency and Hill coefficient.. This result suggests the global nature of the functional positive heterotropic cooperative interaction between L-Phe and extracellular Ca 2+ .

The ensemble of conformations of calcium-and L-Phe-loaded CaSR ECD is distinguishable from the non-loaded forms.
To provide a more detailed description of the CaSR's mechanism of action at the atomic level, we again used MD simulations to analyze the trajectories of these simulations using principal component analysis (PCA) (Material & Methods), which separates out the protein motions into principal modes ranked according to their relative contributions (41). Projection of the trajectories of the different states of CaSR onto the first three modes, which accounted for the majority of the total fluctuations, is shown in Fig. 9a Based on all the experimental results and the directions of the eigenvectors from the longtime simulation (Fig. 9b) we propose a model to illustrate the possible mechanism by which Ca 2+ and L-Phe regulate the function of the CaSR mainly through the molecular connectivity encoded at the hinge region of the ECD of the protein. Our model (Fig. 9c) suggests that a local conformational change upon interaction of the CaSR with L-Phe might affect the overall conformation of the receptor, thereby influencing the cooperativity between multiple Ca 2+ -binding sites and enhancing the receptor's overall response to Ca 2+ .

Discussion
Several major barriers have hampered our understanding of how CaSR integrates its activation by two different classes of nutrients, divalent cations and amino acids, to regulate the functional cooperativity of the receptor and of the alterations of this cooperativity caused by disease mutations. These include "invisible" binding pockets for these two key physiological agonists of the CaSR, namely Ca 2+ and amino acids, challenges in obtaining structural information associated with membrane proteins, and the lack of direct binding methods in determining the mechanism underlying cooperative activation of the CaSR by Ca 2+ and amino acids (13, 19,59). To overcome these limitations, we have developed several computer algorithms and a grafting approach for identifying and predicting Ca 2+binding sites in proteins, and we have successfully verified the intrinsic Ca 2+ -binding capabilities of predicted Ca 2+ -binding sites in the CaSR and mGluR1α (25)(26)(27)60).
Our studies, shown in Figs. 1, 3 and 6, suggest that mutations in Ca 2+ -binding Site 1, such as E2971 and Y218Q, not only disrupt the CaSR's Ca 2+ -sensing capacity but also have an impact on the positive homotropic cooperative interactions of Ca 2+ with the other Ca 2+ -binding sites. The biphasic behavior of these mutants with a large disruption of cooperativity is very similar to our previously reported metal-binding concentration response curves of subdomain 1 and its variants with increases in [Ca 2+ ] o (26). Subdomain 1 of CaSR contains a protein sequence encompassing Ca 2+ -binding Sites 1, 2 and 3, but not Ca 2+ -binding Sites 4 and 5. It also exhibits both a strong and a weak metal-binding component. This strong metal-binding process can be removed by further mutating Site 1 (E297I). In contrast, mutations at Sites 2 and 3 in subdomain 1 have less impact on the first binding process (26). These experimental results are consistent with molecular dynamics simulation studies carried out here showing that residues located at Site 1 have strong correlated motions with other residues involved in Sites 2, 3, 4 and 5 (Table 1). The results suggest that the dynamics of Site 1 are intricately coupled to those of the other binding sites; therefore, any changes in the dynamics of Site 1 could affect those of the other sites. The observation of this molecular connectivity and its relationship to positive cooperativity from the molecular dynamics simulations provides a description at the atomic level of the crosstalk between the different sites of the CaSR suggested by the experimental results in live cells.
Here, we have also identified and characterized an L-Phe-binding pocket formed by residues L51, S170, T145, Y218 and S272 that is adjacent to and partially overlaps the key Ca 2+ -binding Site 1 at the hinge region of the Venus Fly Trap (VFT) of the CaSR. This Ca 2+ -binding Site is also conserved in other family C GPCRs, including the mGluR1 VFT (Fig. 2) (27,31,56). Y218 is involved in sensing both Ca 2+ and L-Phe. The aromatic ring from residue Y218 could form delocalized pi bonds with the side chain of L-Phe and the hydrophobic interaction between L51 and L-Phe would further stabilize this interaction. Mutating S170 might interfere with H-bonding of the ligated amino acid to the α-amino group of S170 based on the structure of mGluR1α (53). S170T has been reported by different groups to interfere with the CaSR's L-Phe-sensing ability (51,53). Consideration of the crystal structure of the glutamate-bound form of mGluR1 (31), together with our docking analysis, implies that residues T145 and S272 may not directly participate in the interaction with L-Phe but could possibly interact with L-Phe by ligation of water molecules, which is a relatively weaker type of interaction.
We have observed essentially equivalent expression of WT CaSR as well as its variants on the cell surface (Fig. 5). These data suggest that the difference in the Ca 2+ -sensing capacities among the WT and mutant receptors are due to perturbation of the cell surface receptors' functions, rather than, for example, impaired trafficking of the receptor proteins to the cell surface. L-Phe rescued the calcium responses of the tested mutants located in all five predicted Ca 2+ -binding sites, and it had more dramatic rescuing effects on mutants E297I and D215I compared with the other mutants (Fig. 8d). Thus, the importance of the hinge region, where L-Phe likely interacts with the CaSR ECD, is once again highlighted.
The PCA results suggest that the need for Ca 2+ in initially activating CaSR, as suggested by these experiments, is related to shifting the ensemble of conformations of CaSR from an inactive state to an active state. The activity of the Ca 2+ -loaded form of CaSR is then further enhanced by the binding of L-Phe, which produces an additional change in the ensembles of conformations of CaSR. Therefore, the global modulation of receptor activity by Ca 2+ and L-Phe at GEORGIA STATE UNIV on September 23, 2016 http://www.jbc.org/ Downloaded from might be explained by a combination of an induced fit and population shift models (61), that is the receptor's overall structure could vary in the equilibrium distributions of conformations that can interchange dynamically in the absence of Ca 2+ and L-Phe . Our experimental results suggest that binding of Ca 2+ at its various sites is associated with motions of these sites that are highly correlated with one another. Consequently, the shift in the ensemble of conformations of CaSR induced by the initial binding of Ca 2+ at Site 1 will alter the equilibrium population of the unbound conformations of other Ca 2+ -binding sites due to their crosstalk with Site 1. The binding of Ca 2+ to Site 1 and the subsequent interaction of CaSR with L-Phe can further shift the conformations of the ECD from one part of the free energy landscape to another. In this way Ca 2+ -binding to other sites is more readily favorable. Our findings here also enhance our understanding of the role of Ca 2+ in modulating key Ca 2+ -binding proteins, such as calmodulin, to mediate signal transduction via correlated motions among their multiple Ca 2+binding sites, thereby generating cooperative responses with critical biological consequences (62,63).
The CaSR's co-activation by these two classes of ligands may be particularly important in the gastrointestinal tract, where high concentrations of amino acids resulting from protein digestion would promote activation of the CaSR and its stimulation of digestive processes even when there are relatively low levels of [Ca 2+ ] o . Moreover, as reported in clinical studies, there are 33 disease-related variants near Ca 2+binding Site 1 associated with receptor activation or inactivation and, in some cases, with reduced cooperativity (15). Our work suggests that it is likely that such mutations disrupt the molecular connectivity encoded in the receptor and provide a better understanding of the molecular basis of some of the CaSR-related clinical disorders. Although the circulating levels of Ca 2+ and L-Phe in vivo are lower than in the in vitro experiments, the discrepancy noted here (2-3-fold for the WT receptor for Ca 2+ ) is substantially less than variety of more classical hormone-receptor systems. For the cloned parathyroid hormone (PTH) receptor, for example, the K d 's for activation of adenylate cyclase and stimulation of PLC, are 1 nM and 20-50 nM, respectively, while the normal circulating levels of PTH are ~1-7 pM, i.e., resulting in a >100-fold to even a >1000-fold discrepancy between in vivo and in vitro results (64). Our finding of the capacity of L-Phe to rescue diseaselinked mutations suggests the possibility of enhancing the activities of such mutant receptors using calcimimetics of various types as pharmacotherapy. Thus our results provide insights into key factors regulating the receptor's overall activity, which can lay the foundation for a new generation of therapeutics and drugs.
In addition to the Ca 2+ -sensing receptor, [Ca 2+ ] o regulates 14 of the other members of the family C G protein-coupled receptors (GPCRs), including the metabotropic glutamate receptors (mGluR), γ-aminobutyric acid GABA B receptors and receptors for pheromones, amino acids and sweet substances (1,5,10, 16,[54][55][56]65). The observed molecular connectivity centered at predicted calcium-binding Site 1 of the CaSR, which is adjacent to an amino acid-binding pocket at the hinge region of the receptor, may be shared by other members of the family C GPCRs (66,67) . Besides the strong conservation of the predicted calcium-binding site 1 and the adjacent amino acid-binding pocket, several lines of evidence support this suggestion (Fig. 2) (25,26,56,68). We have predicted a Ca 2+ -binding site partially sharing a Glu-binding site in the ECD of mGluR1 α, and both of them co-activate the receptor (27). A Ca 2+ -binding pocket was proposed to be present in the ligand-binding site of the GABA B receptor (55). Many animals and humans can detect the taste of calcium via a calcium taste receptor that is modulated by an allosteric mechanism (69).
In summary, our present study provides a mechanistic view of the interplay among extracellular Ca 2+ , amino acids and the CaSR via molecular connectivity that modulates the positive homotropic and heterotropic cooperativity of CaSR-mediated intracellular Ca 2+ signaling. The positive cooperative co-activation of the CaSR by Ca 2+ and L-Phe and the importance of the positive homotropic and heterotropic cooperativity, respectively, exhibited by the two agonists may be

Competing financial interests
The author(s) declare no competing financial interests.         The cross-correlation coefficients of each residue to all of the other residues of the modeled CaSR ECD structure after docking with calcium were calculated from the simulation. The strongest positive correlation of a residue with itself is given the value 1; the strongest negative correlation between two residues is given the value -1. A cut off at > 0.7 and < -0.4 is considered as a strong correlation between residues. Strongly correlated residues in predicted calcium-binding sites are listed in the table.   HEK293 cells were transiently transfected with the WT CaSR or Ca 2+ -binding-related CaSR mutants, and after 48 h the cells were loaded with fura-2 as described under "Materials and Methods". The cells on glass coverslips were then transferred into the cuvette for fluorimetry and exposed to various increases in [Ca 2+ ] o (from 0.5~30 mM) in the absence or presence of 5 mM L-Phe as described above..