Activating Mutations of the G-protein Subunit α 11 Interdomain Interface Cause Autosomal Dominant Hypocalcemia Type 2

Abstract Context Autosomal dominant hypocalcemia types 1 and 2 (ADH1 and ADH2) are caused by germline gain-of-function mutations of the calcium-sensing receptor (CaSR) and its signaling partner, the G-protein subunit α 11 (Gα 11), respectively. More than 70 different gain-of-function CaSR mutations, but only 6 different gain-of-function Gα 11 mutations are reported to date. Methods We ascertained 2 additional ADH families and investigated them for CaSR and Gα 11 mutations. The effects of identified variants on CaSR signaling were evaluated by transiently transfecting wild-type (WT) and variant expression constructs into HEK293 cells stably expressing CaSR (HEK-CaSR), and measuring intracellular calcium (Ca2+i) and MAPK responses following stimulation with extracellular calcium (Ca2+e). Results CaSR variants were not found, but 2 novel heterozygous germline Gα 11 variants, p.Gly66Ser and p.Arg149His, were identified. Homology modeling of these revealed that the Gly66 and Arg149 residues are located at the interface between the Gα 11 helical and GTPase domains, which is involved in guanine nucleotide binding, and this is the site of 3 other reported ADH2 mutations. The Ca2+i and MAPK responses of cells expressing the variant Ser66 or His149 Gα 11 proteins were similar to WT cells at low Ca2+e, but significantly increased in a dose-dependent manner following Ca2+e stimulation, thereby indicating that the p.Gly66Ser and p.Arg149His variants represent pathogenic gain-of-function Gα 11 mutations. Treatment of Ser66- and His149-Gα 11 expressing cells with the CaSR negative allosteric modulator NPS 2143 normalized Ca2+i and MAPK responses. Conclusion Two novel ADH2-causing mutations that highlight the Gα 11 interdomain interface as a hotspot for gain-of-function Gα 11 mutations have been identified.

To date, 4 FHH2 and 6 ADH2 different mutations have been identified in the GNA11 gene on chromosome 19p13.3 (Fig. 1B), which encodes Gα 11 , and studies of the location of such mutations has provided insight into Gα 11 structure function (2, 9-11, 15, 16). Thus, FHH2 and ADH2 mutations cluster within 3 regions ( Fig. 1C): the Gα 11 -GPCR interaction region; the interdomain interface between the helical and GTPase domains; and the sites at which Gα 11 interacts with Gβγ and PLC (2, 9-11, 15, 16). This indicates that these 3 structural regions play a critical role in Gα 11 -mediated CaSR signaling. Additionally, previous studies of these mutations have indicated that CaSR negative allosteric modulators, which are known as calcilytic compounds, can normalize the gain-of-function caused by Gα 11 mutations both in vitro and in mouse models of ADH2 (17)(18)(19), and thus represent a potential targeted therapy for this disorder.
Here, we report the clinical and genetic findings in 2 unrelated families with ADH, in whom novel heterozygous germline gain-of-function Gα 11 mutations, were identified.

Patients and families
Family 1. This family comprised 3 affected members (a mother, her son, and daughter) ( Fig. 2A). The son (individual II.1, Fig. 2A) at the age of 10 years was referred with a chronic motor tic disorder, which was subsequently diagnosed as Tourette syndrome. He was also experiencing paresthesia, and biochemical investigations showed him to have a mildly low serum calcium of 2.12 mmol/L (normal 2.20-2.70 mmol/L) in association with an inappropriately normal plasma PTH of 2.8 pmol/L (normal 1.0-7.0 pmol/L) and insufficient serum 25-hydroxyvitamin D of 42 nmol/L (adequate >50 nmol/L). He had a normal serum phosphate concentration of 1.57 mmol/L (normal 0.90-1.80), magnesium of 0.90 mmol/L (normal 0.70-1.0), creatinine of 59 μmol/L (normal 28-63), alkaline phosphatase activity of 146 IU/L (normal 60-425), and low urinary calcium-to-creatinine ratio of 0.08 mmol/mmol (normal 0.30-0.70). He was commenced on oral calcium and cholecalciferol, which increased his serum 25-hydroxyvitamin D to 87 nmol/L; however, his serum calcium remained low at 2.11 mmol/L. His mother and younger sister ( Fig. 2A) were also found to be mildly hypocalcemic with serum calcium concentrations of 2.08 mmol/L and 2.15 mmol/L, respectively. This family was investigated for ADH as a possible cause of the mild hypocalcemia and leukocyte DNA was obtained from affected family members following informed consent for analysis of the CASR and GNA11 genes. Family 2. This family comprised 5 affected members (a mother and her 2 sisters, and her daughter and son) (Fig. 2B). The daughter (individual III.1, Fig. 2B), at the age of 38 years, was referred with a 6-month history of fatigue, myalgia, dizziness, and bilateral hip pain. She had no history of paresthesia, muscle cramps, seizures, or renal calculi. She had not previously undergone neck surgery and had no history of deafness, renal or cardiac abnormalities, candidiasis, or Addison's disease. Her only comorbidity was recently diagnosed autoimmune hypothyroidism, which was treated with levothyroxine 75 μg daily. Her height was 170 cm and weight was 84.3 kg. Biochemical investigations showed her to have a low serum calcium of 1.97 mmol/L (normal 2.20-2.60) in association with an inappropriately normal plasma PTH of 4.1 pmol/L (normal 1.0-7.0), and a borderline low urine calcium to creatinine ratio of 0.30 mmol/mmol (normal 0.30-0.70), and a fractional excretion of calcium 0.01 (normal >0.01). She had a normal serum magnesium concentration of 0.78 mmol/L (normal 0.70-1.00), phosphate of 1.12 mmol/L (normal 0.70-1.45), creatinine of 69 μmol/L (normal 45-90), alkaline phosphatase activity of 45 U/L (normal 30-130), 25-hydroxyvitamin D of 125 nmol/L (normal>50 nmol/L), 1,25-dihydroxyvitamin D of 111 pmol/L (normal 43-144), and TSH of 1.76 mU/L (normal 0. 30-4.20). Thyroid peroxidase antibodies were elevated at >1518 (normal < 60 IU/mL), and anti-parathyroid antibodies, as assessed by indirect immunofluorescence (20), were not detected. Her hypocalcemia was initially treated with 1.0 to 2.5 g of oral elemental calcium daily. However, she remained hypocalcemic and also became hypomagnesemic (lowest serum magnesium = 0.62 mmol/L), and was commenced on alfacalcidol 1.0 μg daily, as well as oral magnesium aspartate 10 mmol twice daily. Her mother, brother, and 2 maternal aunts were also hypocalcemic ( Fig. 2B), and these findings were suggestive of either ADH or familial isolated hypoparathyroidism. Leukocyte DNA was obtained from affected family members following informed consent for analysis of the CASR, GNA11, GCM2, GATA3, AIRE, and PTH genes.

Intracellular calcium measurements
Ca 2+ e -induced Ca 2+ i responses were measured by Fluo-4 calcium assays as previously described (16). HEK-CaSR cells were plated in 12-well plates and transiently transfected with 1000 ng/mL pBI-CMV2-GNA11. Following 24 hours' incubation, cells were replated at 30 000 cells/well in black-walled 96-well plates (Corning). Cells were treated with serum-free media (SFM) overnight. Fluo-4 dye was prepared according to manufacturer's instructions (Invitrogen), and cells loaded for 1 hour at 37°C. Baseline measurements were made and increasing concentrations of CaCl 2 injected automatically into each well. Changes in Ca 2+ i were recorded on a PHERAstar instrument (BMG Labtech) at 37°C with an excitation filter of 485 nm and an emission filter of 520 nm. The peak mean fluorescence ratio of the transient response after each individual stimulus was measured using Cytomation Summit software (Beckman Coulter), and expressed as a normalized response. Nonlinear regression of concentration-response curves was performed with GraphPad Prism using the normalized response at each [Ca 2+ ] e for each separate experiment for the determination of the mean half-maximal concentration (EC 50 ) (i.e., [Ca 2+ ] e required for 50% of the maximal response). Assays were performed in 4 to 8 independent transfections. Statistical analysis was performed using the F-test.

Luciferase reporter assays
HEK-CaSR cells were plated in 24-well plates and transiently transfected with 100 ng/mL pBI-CMV2-GNA11 WT or mutant construct, 100 ng/mL luciferase construct (either pGL4-nuclear factor of activated T-cell response element [NFAT-RE] or pGL4-serum response element [SRE] and 10 ng/mL pRL null control luciferase reporter. Following 48 hours incubation, cells were treated with SFM overnight. Cells were then treated with SFM containing 0.1 to 10 mM CaCl 2 and incubated for 4 hours. Cells were lysed and assays performed using Dual-Glo Luciferase (Promega) on a Veritas Luminometer (Promega) as previously described (16,18). Luciferase:renilla ratios were expressed as fold changes relative to responses at low CaCl 2 concentrations (0.1 mM). For studies with NPS 2143 (Abcam), drug was added to cells 4 hours before reporter assays were performed. All assay conditions were performed in 4 to 12 independent transfections. Statistical analysis was performed by 2-way ANOVA with Tukey's multiple-comparisons test using GraphPad Prism 6.

Results
Identification of novel missense mutations in Gα 11 in 2 ADH2 probands DNA sequence analyses in the 2 ADH families identified abnormalities only in the GNA11 gene. Thus, in family 1 (Fig. 2A), a heterozygous G-to-A transition at nucleotide c.196 within exon 2 of GNA11 ( Fig. 1B  and 2C) was identified and in family 2 (Fig. 2B), a heterozygous G-to-A transition at nucleotide c.446 within exon 3 of GNA11 was identified ( Fig. 1B and  2D). The G-to-A transition in family 1 is predicted to lead to a missense substitution of Gly to Ser at codon 66 of the Gα 11 protein (Fig. 2E), and in family 2 to a missense substitution of Arg to His at codon 149 of the Gα 11 protein (Fig. 2F). Bioinformatic analyses using Polyphen-2 and MutationTaster software (22,23) predicted the p.Gly66Ser and p.Arg149His variants to be damaging and likely disease causing (Polyphen-2 score 1, MutationTaster score 0.99). The p.Arg149His Gα 11 variant was not detected in the gnomAD database, whereas the p.Gly66Ser variant was detected in 2 of 281 488 alleles, yielding a rare allele frequency of <0.001%. The p.Gly66Ser and p.Arg149His variants were detected in all hypocalcemic members of families 1 and 2, respectively ( Fig. 2A and 2B), and these findings with the demonstration of evolutionary conservation of the Gly66 and Arg149 residues in Gα 11 orthologs and Gα paralogs ( Fig. 3A and 3B), indicated that the p.Gly66Ser and p.Arg149His abnormalities likely represented pathogenic mutations rather than benign polymorphic variants. Thus, 2 heterozygous novel missense germline mutations (Fig. 2C-F) were likely identified in the 2 ADH families, and structural and functional characterization of these potential Gα 11 mutations were therefore undertaken.

Structural characterization of the p.Gly66Ser and p.Arg149His Gα 11 mutant proteins
The Gly66 residue is located within the linker 1 peptide that acts as a flexible hinge between the helical and GTPase domains of Gα 11 and connects the α1 helix of the GTPase domain with the αA helix of the helical domain (Figs. 1C, 3A, and 3C-D). The linker 1 peptide comprises 5 residues that form a hydrogen bond network with residues within the α1-and αA-helices to stabilize the G-protein structure (27) (Figs. 1C and  3C). The Gly66 residue represents the central amino acid of the linker 1 peptide and forms a hydrogen bond with the Arg60 residue (27), mutations of which have been reported to cause ADH2 (Figs. 1 and 3C) (9, 10). However, the Ser66 mutation is not predicted to disrupt the interaction with the Arg60 residue (Fig. 3D), but the mutant Ser66 residue instead leads to the introduction of a bulky polar side chain (Fig. 3D), which may destabilize the linker 1 region.
The Arg149 residue is located within the αD helix of the helical domain, which lies close to switch 3, a flexible region within the GTPase domain that undergoes conformational changes during Gα 11 activation (28) (Figs. 1C, 3A and 3E). Arg149 projects into the interdomain interface and is predicted to form 2 contacts (dotted black line) with the switch 3 Asp236 residue (Fig. 3E). Mutation of the Arg149 residue to His149 is predicted to lose both contacts with the Asp236 residue (Fig. 3F).

Functional characterization of the p.Gly66Ser and p.Arg149His Gα 11 mutant proteins
The effects of the p.Gly66Ser and p.Arg149His mutations on Gα 11 function could not be predicted from the homology modeling studies described previously, and we therefore characterized these mutations in vitro to determine their effects on CaSR-mediated signaling. HEK-CaSR cells were transiently transfected with pBI-CMV2-GNA11 constructs expressing either the WT (Gly66 or Arg149) or mutant (Ser66 or His149) Gα 11 proteins. This bidirectional pBI-CMV2 vector allows for coexpression of Gα 11 and GFP at equivalent levels (2); and expression of the CaSR, Gα 11 and GFP was confirmed by fluorescence microscopy and/or Western blot analyses (Fig. 4A, B). Gα 11 expression was shown to be similar in cells transiently transfected with WT or mutant proteins, and greater in transfected cells than endogenous Gα 11 protein expression in untransfected cells, by Western blot analyses in which calnexin was used as a loading control (Fig. 4B).

Effect of the p.Gly66Ser and p.Arg149His Gα 11 mutant proteins on CaSR-mediated Ca 2+
i responses. The effects of the Gα 11 mutants, Ser66 and His149, on Ca 2+ e -induced Ca 2+ i responses using the Fluo-4 calcium assay were assessed, as reported (16). The Ca 2+ i responses in WT and mutant Gα 11 -expressing cells were shown to increase in a dose-dependent manner following stimulation with increasing concentrations of Ca 2+ e . The responses of the mutant Ser66 and His149 expressing cells were similar to WT cells at low (0.1 mM) Ca 2+ e , but were significantly elevated compared with WT cells following Ca 2+ e stimulation (Fig. 4C, D). Thus, the Ser66 and His149 mutant expressing cells showed a leftward shift in the concentration-response curve (Fig. 4C, D) Fig. 4C, D). Therefore, a 10-nM dose of NPS 2143 is effective at normalizing Ser66 and His149 Ca 2+ i responses, whereas 30 mM of NPS 2143 leads to a dose-dependent "overcorrection" that is equivalent to a loss-of-function of the CaSR.
To provide further evidence that the Ser66 and His149 Gα 11 mutant proteins affect Ca 2+ i signaling, the gene transcription induced by a NFAT-RE containing luciferase reporter construct was measured, as NFAT is a downstream mediator of Ca 2+ i signaling (29) (Fig. 1A). HEK-CaSR cells were transiently transfected with WT or mutant Ser66 or His149 mutant Gα 11 proteins, and NFAT-RE reporter fold-change responses measured in response to increasing concentrations of Ca 2+ e . NFAT-RE reporter responses were significantly elevated in cells expressing the Ser66 and His149 mutant Gα 11 proteins (Fig. 5A, B). The effects of 10 nM NPS 2143 on these NFAT-RE responses were assessed at 7.5 mM Ca 2+ e concentration. This confirmed the significantly increased NFAT-RE reporter fold-change responses in Ser66 and His149 cells, compared with WT expressing cells (Ser66 = 4.12 ± 0.27 and His149 = 4.85 ± 0.45, compared with 2.26 ± 0.06 for WT expressing cells, P < 0.001 and P < 0.0001, respectively) and demonstrated that addition of 10 nM NPS 2143 to the cells rectified NFAT-RE reporter fold-change responses to WT values (Ser66 + 10 nM NPS 2143 = 2.63 ± 0.09 and His149 + 10 nM NPS 2143 = 2.92 ± 0.06) (Fig. 5C, D).
Effect of the p.Gly66Ser and p.Arg149His Gα 11 mutant proteins on CaSR-mediated MAPK responses. Previous studies of Gα 11 mutations have demonstrated an increase in MAPK signaling in cells expressing ADH2-causing mutant proteins (9,17). To investigate the effect of the Ser66 and His149 mutant proteins on MAPK signaling, gene transcription induced by a SRE containing luciferase reporter construct, which is a downstream mediator of MAPK signaling (21) (Fig. 1A), was measured in HEK-CaSR cells transiently expressing WT or mutant Ser66 or His149 Gα 11 proteins (Fig. 6A, B). Cells expressing the Ser66 and His149 mutant proteins showed no alterations in SRE reporter fold-change responses at low (0.1 mM) Ca 2+ e (Fig. 6A, B). However, stimulation with increasing Ca 2+ e concentrations led to significantly elevated SRE reporter fold-change responses at 2.5 to 10 mM Ca 2+ e in cells expressing the Ser66 and His149 Gα 11 mutants compared with WT expressing cells (Fig. 6A, B). The effects of 10 nM NPS 2143 on these SRE responses were assessed at 7.5 mM Ca 2+ e concentration. This revealed that the SRE reporter fold-change responses in Ser66 and His149 expressing cells were significantly elevated compared with WT expressing cells (Ser66 = 7.86 ± 1.28, compared with 3.21 ± 0.32 for WT expressing cells, P < 0.0001, and His149 = 13.26 ± 0.85, compared with 9.56 ± 1.18 for WT, P < 0.01) (Fig. 6C, D), whereas addition of 10 nM NPS-2143 to the Gα 11 mutant expressing cells rectified SRE reporter responses to that of WT Gα 11 -expressing cells (Ser66 + 10 nM NPS-2143 = 3.96 ± 0.16; and His149 + 10 nM NPS-2143 = 10.15 ± 0.57) (Fig. 6C, D). Thus, a 10-nM dose of NPS-2143 is effective at normalizing mutant Gα 11 Ser66 and His149 MAPK responses.

Discussion
Our studies have identified 2 novel heterozygous germline Gα 11 mutations associated with ADH2. The affected individuals harboring the gain-of-function p.Gly66Ser and p.Arg149His Gα 11 mutations had a generally mild clinical phenotype with serum calcium concentrations of >1.90 mmol/L and were either asymptomatic or experienced paresthesiae. In addition, there were no alterations in serum concentrations of phosphate or magnesium, and plasma PTH concentrations were detectable and inappropriately within the normal range. These findings are similar to that reported for other patients with ADH2, which is characterized by mild-to-moderate hypocalcemia, normal or elevated serum phosphate, normomagnesemia, and low/normal PTH values (2,(9)(10)(11)(12). Urinary calcium excretion was normal or low in the affected individuals in this report, which is also consistent with the phenotype of ADH2. However, some ADH2 patients are susceptible to treatment-related hypercalciuria, nephrocalcinosis, and nephrolithiasis (9,11,12).
The p.Gly66Ser and p.Arg149His Gα 11 mutations reported in this study are located at the interface between the GTPase and helical domains (Figs. 1 and 3). The interdomain interface represents a highly conserved region of the Gα subunit (30), and is the site of multiple interactions between the GTPase and helical domains, including between the linker 1 peptide and the α1 and αA helices (27), and also between the αD-helix and the switch III region (31). This region has a critical structural role within the G-protein and is important for binding guanine nucleotides. In support of this, engineered mutations of the Gα subunit interdomain interface residues have been shown to destabilize the GDP-bound state, and it is likely that such mutations enhance the separation of the GTPase and helical domains, which in turn leads to the release of GDP (30). The interdomain interface region has previously been associated with 4 germline mutations of Gα 11 , three associated with ADH2 (p.Arg60Cys, p.Arg60Leu, and p.Arg181Gln), and 1 associated with FHH2 (p.Thr54Met) (Fig. 1) (2,9,10,15). Additionally, the germline Gα 11 hypermorphic variant, p.Ile62Val, identified in an N-ethyl-N-nitrosourea generated mouse, which is a model for ADH2, and the somatic constitutively activating mutations in Gα 11 identified in patients with uveal melanoma, also affect the interdomain interface (18,32,33). Thus, this region likely represents a hotspot for disease-causing Gα 11 mutations.
Our finding that the germline p.Gly66Ser and p.Arg149His Gα 11 mutations led to a gain-of-function contrasts with engineered mutagenesis studies involving homologous residues in other Gα subunits (27,31). Thus, an engineered p.Gly66Asp mutation in the closelyrelated Gα q protein did not cause a gain-of-function, but instead increased coupling of non-G q -GPCRs to G q effectors (27,34). The Gly66 residue is located within the linker 1 peptide, which is not fully conserved between the Gα 11 and Gα q proteins (Fig. 3A); this lack of sequence conservation may explain the differences observed in these studies. Mutations of the αD-helix Asn167 Gαs residue, which is homologous to the Arg149 Gα 11 residue, also did not lead to a gain-of-function (31). Indeed, an engineered p.Asn167Ala mutation had no effect on Gαs function, whereas an engineered p.Asn167Arg mutation impaired GPCR-mediated activation of Gαs (31). However, mutagenesis studies of Arg144 in Gα i , which is homologous to Arg149 in Gα 11 , did show an increase in GDP dissociation rates, which may increase signaling activity (35). Moreover, the Ser140-Asp227 interdomain contact in Gα t , equivalent to Gα 11 Arg149-Asp236, is important for conformational transitions between active and inactive states (36). Thus, it is difficult to predict the structure-function consequences of the His149 Gα 11 mutation, and the introduction of the mutant residue, rather than loss of the WT residue in the αD-helix is likely to be responsible for influencing Gα subunit function.
Our in vitro studies have shown that that the germline p.Gly66Ser and p.Arg149His Gα 11 mutations do not enhance CaSR-mediated signaling at low (0.1 mM) Ca 2+ e concentrations, and thus these mutations are not constitutively activating. This observation is in keeping with other reported germline ADH2causing Gα 11 mutations, but contrasts with somatic uveal melanoma-causing Gα 11 mutations, which cause a marked increase in MAPK activation in unstimulated cells (17). The p.Gly66Ser and p.Arg149His Gα 11 mutations were associated with an overall mild increase in CaSR-mediated Ca 2+ i and MAPK responses, and these findings may explain the mild hypocalcemia observed in the patients harboring these mutations. These cellular studies involving the mutant Ser66 and His149 Gα 11 proteins have also provided further evidence of the utility of calcilytic compounds in rectifying signaling abnormalities in the Gα 11 protein, which we have previously shown in vitro and in a mouse model of ADH2 (17,18). Importantly, our studies showed that a low dose (10 nM) of the NPS 2143 calcilytic compound can successfully correct the gainof-function associated with both the Ser66 and His149 ADH2-causing Gα 11 mutations, and this is similar to the p.Arg181Gln mutation, which is also located in the interdomain interface (17), but contrasts to the p.Phe341Leu mutation, which affects the α5-helix of the GTPase domain that directly binds to the GPCR transmembrane domains and intracellular loops, and requires a higher dose (30 nM) of NPS 2143 to normalize CaSR signaling (17). Thus, mutations affecting residues in the interdomain interface require a lower dose of allosteric modulator to rectify CaSR signaling than Gα 11 mutations located in the G-protein-GPCR interface, and further investigation of these may provide insights into the mechanism by which allosteric modulators rectify CaSR-mediated signaling abnormalities associated with G-protein mutations.
In summary, our studies have identified disease-causing mutations located in the linker 1 peptide and αD-helix of the Gα 11 protein. These findings demonstrate that the Gα 11 interdomain interface represents a hotspot for germline gain-of-function mutations causing ADH2.