Different Biochemical Properties Explain Why Two Equivalent Gα Subunit Mutants Cause Unrelated Diseases*

Background: Gα subunit mutants cause different human diseases. Results: We characterized the Albright's Hereditary Osteodystrophy mutant Gαs-R265H and the equivalent Gαo-R243H cancer mutant to find that they have different biochemical properties. Conclusion: The same mutation in two homologous Gα-subunits causes either gain-of-function or loss-of-function in two unrelated diseases. Significance: Not all disease-associated mutations in conserved residues have similar consequences among homologous G proteins. There is an increasing number of disease-associated Gα mutations identified from genome-wide sequencing campaigns or targeted efforts. Albright's Hereditary Osteodystrophy (AHO) was the first inherited disease associated with loss-of-function mutations in a G protein (Gαs) and other studies revealed gain-of-function Gα mutations in cancer. Here we attempted to solve the apparent quandary posed by the fact that the same mutation in two different G proteins appeared associated with both AHO and cancer. We first confirmed the presence of an inherited Gαs-R265H mutation from a previously described clinical case report of AHO. This mutation is structurally analogous to Gαo-R243H, an oncogenic mutant with increased activity in vitro and in cells due to rapid nucleotide exchange. We found that, contrary to Gαo-R243H, Gαs-R265H activity is compromised due to greatly impaired nucleotide binding in vitro and in cells. We obtained equivalent results when comparing another AHO mutation in Gαs (D173N) with a counterpart cancer mutation in Gαo (D151N). Gαo-R243H binds nucleotides efficiently under steady-state conditions but releases GDP much faster than the WT protein, suggesting diminished affinity for the nucleotide. These results indicate that the same disease-linked mutation in two different G proteins affects a common biochemical feature (nucleotide affinity) but to a different grade depending on the G protein (mild decrease for Gαo and severe for Gαs). We conclude that Gαs-R265H has dramatically impaired nucleotide affinity leading to the loss-of-function in AHO whereas Gαo-R243H has a mild decrease in nucleotide affinity that causes rapid nucleotide turnover and subsequent hyperactivity in cancer.

Heterotrimeric G proteins are molecular switches that alternate between "on" (GTP-bound) and "off" (GDP-bound) states to control signal transduction in eukaryotes (1)(2)(3). Signal relay by heterotrimeric G proteins is determined by the lifetime of the GTP-bound form, which is in turn controlled by a delicate equilibrium between the rates of nucleotide exchange and hydrolysis. In resting G proteins, the ␣-subunit is loaded with GDP and tightly bound to the G␤␥ obligatory dimer. G proteincoupled receptors (GPCRs) 3 as well as some other non-receptor proteins (4 -6) work as guanine nucleotide exchange factors (GEFs) that accelerate the rate of GDP release. In cells, GTP is loaded spontaneously upon GDP release due to the physiological ϳ10:1 GTP:GDP concentration ratio, which ensures the directionality of the reaction toward G protein activation. Both GTP-bound G␣ and uncomplexed G␤␥ subunits regulate the activity of a wide array of effectors. G protein signaling is terminated by the intrinsic GTPase activity of the G␣ subunits, which can be further enhanced by GTPase-activating proteins (GAPs). Upon conversion of GTP to GDP ϩ Pi, G␣ reassociates with G␤␥ to complete a cycle. Given the pivotal role of heterotrimeric G proteins in physiology it is not surprising that dysregulation of the biochemical reactions involved in this process gives rise to human disease. In fact, multiple germline and somatically acquired mutations in G␣ subunits that alter their biochemical properties have been described in different human diseases (7,8).
Albright's Hereditary Osteodystrophy (AHO) was the first hereditary disease associated with mutations in G protein ␣-subunits (9). Heterozygous loss-of-function mutations in G␣s cause AHO, a syndrome characterized by skeletal and developmental abnormalities including brachydactyly, short stature, obesity, and mental deficits, among others (10). More than 100 different mutations have been described in AHO to date, out of which ϳ40 are missense changes (11,12). An interesting feature of AHO is that maternal inheritance of the G␣s loss of function mutations also produces multihormone resis-tance whereas paternal inheritance does not. This is caused by imprinting of the gene encoding for G␣s (10 -12). The multihormone resistance associated with maternal inheritance of G␣s AHO mutations is termed pseudohypoparathyrodism Ia (PHP-Ia), whereas pseudopseudohypoparathyroidism (PPHP) refers to patients with the features of AHO lacking hormonal resistance. PHP-Ia is characterized by resistance to hormones that exert their action via G␣s-coupled GPCRs, and it is most prominent for thyroid-stimulating hormone (TSH) and parathyroid hormone (PTH). The molecular basis for the loss of function associated with several G␣s mutations found in AHO have been characterized and include defective GTP binding, enhanced GTP hydrolysis, impaired GPCR coupling, and decreased protein stability (10,(13)(14)(15)(16)(17).
As opposed to loss of function mutations in AHO, gain of function mutations have been described in cancer for multiple G␣ subunits. In addition to mutations in G␣s codons 201 (R3 C) and 227 (Q3 L), which are found in ϳ25% of pituitary adenomas (18), mutations in the same codons have been described for other G␣ subunits (19). A notorious example of this is the high frequency (up to 80%) of G␣q or G␣11 mutations in uveal melanomas (20 -22). The amino acid residues present in these mutated positions are universally conserved across G␣ subunits as is the mechanism by which they trigger oncogenic transformation: i.e. R3 C and Q3 L mutations promote a gain of function by impairing GTPase activity and rendering the G proteins constitutively active in the GTP-bound conformation. In fact, this high degree of sequence and functional conservation has been traditionally exploited as a tool to investigate the function of G proteins. Artificial mutations equivalent to the oncogenic R3 C or Q3 L are routinely used to investigate the biological functions of virtually every G␣ subunit. For example, introduction of equivalent mutations in G␣o, G␣12, G␣13, or G␣z, which are not found in carcinomas, induces a gain of function and promotes oncogenic transformation when artificially introduced into cultured cells (23)(24)(25)(26)(27)(28). R3 C and Q3 L mutants of G␣i1 were also used in the pioneering crystallographic studies that revealed the structural basis of GTP hydrolysis that are common to all G␣ subunits (29).
Cancer genome sequencing campaigns are revealing a far more complex landscape of G protein mutations (8) that contrasts with the reduced number of "classical" G protein oncomutants described above. More than a hundred different mutations have been identified and although some of them occur at conserved positions in the G␣ protein sequences, it is still unclear how conserved their impact on G protein function is.
Here we compared the functional consequences of a G␣o cancer somatic mutation (R243H) with an identical G␣s-inherited mutation (R265H) present in a family with AHO. We provide evidence indicating that the same mutation in two homologous G␣ subunits causes either a gain-of-function or loss-of-function phenotype associated with two unrelated human diseases.

EXPERIMENTAL PROCEDURES
Reagents-Unless otherwise indicated all reagents were of analytical grade and obtained from Sigma-Aldrich or Fisher Scientific. Escherichia coli strain BL21(DE3) codonϩ was purchased from Invitrogen. Pfu ultra DNA polymerase was pur-chased from Agilent. [␥-32 P]GTP and [␣-32 P]GTP were from Perkin-Elmer.
Patients and Sequence Analysis-Two members of a single family (mother and son, patients II.1 and III.1 in Fig. 2A, respectively) presented skeletal abnormalities of AHO, i.e. brachydactyly assessed by x-ray radiography (30). The son (patient III.1) presented characteristics of abnormal endocrine function (elevated TSH with normal thyroxine (T 4 )) and was diagnosed with PHP-Ia whereas the mother (patient II.1) had normal endocrine markers and was diagnosed with PPHP. Another member of the family (grandmother, patient I.2, deceased) presented external signs of AHO but was not formally diagnosed. 4 Samples from patients II.1 and III.1 were provided by Dr. Bastida Eizaguirre (Hospital Santiago Apostol, Vitoria, Spain) with written consent from the patients and approval by the institutional bioethics committee. We amplified exon 10 of the human GNAS gene by PCR using primers annealing the flanking introns using as template genomic DNA extracted from peripheral blood. The amplicon was run in an agarose gel, purified using a GenJet Gel extraction kit (Thermo) and sequenced.
Plasmid Constructs-Rat G␣o (isoform 1; G␣o1, hereafter referred to as G␣o) with a short N-terminal hexahistidine (His) tag was cloned into pET28b as described previously (33). Histagged bovine G␣s with a short N-terminal His tag cloned into pHIS6 was a kind gift from Dr. Artemyev (University of Iowa) (34). To generate constructs for mammalian expression, G␣s was PCR amplified from pHIS6-G␣s and cloned into the EcoRI and XhoI sites of the plasmid pMSCV-IRES-GFP (Addgene 20672). In this construct GFP expression is driven by the same promoter as the G protein by virtue of an internal ribosome entry site (IRES), making the levels of both transcripts proportional. Point mutations were generated using specific primers following the manufacturer's instructions (QuikChange II, Agilent). All constructs were verified by DNA sequencing.
Cell Culture and Immunoblotting-HEK293 cells were grown at 37°C in DMEM supplemented with 10% FBS, 100 units/ml penicillin, 100 g/ml streptomycin, 1% L-glutamine, and 5% CO 2 . Equal amounts DNA plasmids were transfected using the calcium phosphate method and cells harvested after 48 h. For the experiments to assess the stability of G␣s WT and G␣s R265H proteins, HEK293 cells transfected with the indicated plasmids were incubated with 20 g/ml cycloheximide for 0, 0.5, 1, 2, 4, or 6 h prior harvesting. Lysates were prepared by resuspending the cells in lysis buffer (20 mM HEPES, pH 7.2, 5 mM Mg(CH 3 COO) 2 , 125 mM K(CH 3 COO), 0.4% Triton X-100, 1 mM DTT) supplemented with a protease inhibitor mixture (Sigma) and cleared at 14,000 ϫ g for 10 min. Samples were boiled after addition of Laemmli buffer, run by SDS-PAGE 4 Dr. Bastida Eizagurre, personal communication.
cAMP Measurements-Changes in intracellular levels of cAMP were measured by the cAMP-Glo Assay kit (Promega) following the manufacturer's instructions. Briefly, HEK293 cells were transfected with G␣s WT, G␣s R265H, or an empty vector and harvested 48 h later. 20-l aliquots in duplicate of a 700 ϫ 10 3 cell/ml suspension were incubated in the presence of IBMX (1 mM) for 15 min at 37°C. Reactions were stopped using the lysis buffer provided with the kit, and subsequent steps carried out following the manufacturer's instructions to determine the increase in cAMP compared with control cells (empty vector-transfected). Aliquots of the cells were processed in parallel to quantify total amount of protein and verify the expression of the transfected constructs. cAMP levels were normalized by the total amount of protein.
Limited Proteolysis Assay with HEK293 Cell Lysates-Approximately half of the HEK293 cells of a 10 cm dish transfected with G␣s WT or G␣s R265H were lysed in 45 l of buffer (20 mM HEPES, pH 7.2, 5 mM Mg(CH 3 COO) 2 , 125 mM K(CH 3 COO), 0.4% Triton X-100, 1 mM DTT) supplemented with 125 M GDP, 125 M GTP␥S, or 125 M GDP plus 125 M AlCl 3 and 10 mM NaF. Samples were vortexed, incubated in ice for 10 min and centrifuged at 14,000 ϫ g for 10 min. 40 l of the supernatants were incubated at 30°C for 30 min to allow the loading of nucleotide. Trypsin (12.5 g/ml) was added to the tubes and incubated at 30°C for 5 min. Reactions were stopped by addition of Laemmli sample buffer and boiling. Samples were run by SDS-PAGE and immunoblotted with a rabbit anti-G␣s antibody (1:500, G␣s (C-18) Santa Cruz Biotechnology).
Protein Purification-His-G␣o was purified exactly as described previously (33). His-G␣s was purified from BL21(DE3) codonϩ bacteria by nickel affinity chromatography followed by ion exchange chromatography as described in Ref. 35 with minor modifications. Briefly, His-G␣s expression was induced by 0.1 mM IPTG at 23°C overnight when the OD600 reached ϳ0.5. Subsequent steps were exactly as in (35) except that a HiTrapQ HP column (GE Healthcare) was used for the ion exchange chromatography and that proteins were buffer exchanged into 50 mM Tris-HCl, pH 8.0, 150 mM NaCl, 5 mM MgCl 2 , 10 M GDP, 5% (v:v) glycerol, and 5 mM ␤-mercaptoethanol before storage at Ϫ80°C.
Limited Proteolysis Assay with Purified Proteins-This assay was carried out exactly as described previously (36) for His-G␣o and with minor modifications for His-G␣s. Briefly, His-G␣s (0.2 mg/ml) was incubated for 60 min at 30°C in the presence of GDP (30 M), GTP␥S (30 M) or GDP⅐AlF 4 Ϫ (30 -100,000 M GDP, 30 M AlCl 3 , 10 mM NaF). After this incubation, trypsin was added to the tubes (final concentration 80 g/ml), and samples were incubated for 20 min at 30°C. Reactions were stopped by adding Laemmli sample buffer and boiling. Proteins were separated by SDS-PAGE and stained with Coomassie Blue.

Measurement of GTP␥S Binding by Intrinsic Tryptophan
Fluorescence-Purified His-G␣o (1 M) was equilibrated at room temperature in a cuvette with 1 ml of buffer (20 mM Na-HEPES, pH 8, 100 mM NaCl, 1 mM EDTA, 2 mM MgCl 2 , 1 mM DTT, 0.05% (w:v) C12E10). GTP␥S (2 M) was added to the cuvette after ϳ3 min, and the G protein activation rate monitored by measuring the change in the intrinsic fluorescence (excitation at 284 nm, emission at 340 nm) due to structural rearrangement of the Switch II tryptophan residue Trp-212 (37). Data were collected with a Hitachi F-4500 Fluorescence Spectrophotometer, background corrected (buffer fluorescence) and expressed as % of maximal binding (fluorescence plateau at Ͼ40 min). Each protein (WT or mutant) was normalized to its own maximal binding to facilitate the interpretation of the results. Maximal binding values were similar for WT and mutants and are indicated in the figure legends of the corresponding experiment. Fluorescence measurements for His-G␣s (0.25 M) were performed identically except that the buffer composition was 50 mM Tris-HCl, pH 8.0, 150 mM NaCl, 5 mM MgCl 2 , 0.5 M GDP, 5% (v:v) glycerol, and 5 mM ␤-mercaptoethanol and that the results were expressed as a ratio of the basal fluorescence (F/F 0 ) because mutants did not reach a maximum.

RESULTS AND DISCUSSION
Identification of Two Analogous Mutations in Cancer and AHO-We recently characterized a novel oncogenic mutation in G␣o (36). We found that the molecular mechanism by which this gain-of-function mutation, namely R243H, triggers oncogenic transformation differs from any previously characterized oncogenic mutations in other G␣ subunits: i.e. it renders the G protein hyperactive by accelerating the rate of nucleotide exchange instead of by impairing GTP hydrolysis (36). The residue corresponding to G␣o Arg-243 is conserved in all 16 human G␣ subunits. Bioinformatics searches in the cancer genomics databases cBioPortal (31) and COSMIC (32) revealed somatic mutations affecting this residue in 6 different G␣ subunits (Fig. 1A). Remarkably, we did not find in these databases any other residue that was mutated in as many different G proteins (not shown). The 6 mutated ␣-subunits collectively belong to 3 out of the 4 G␣ subfamilies (Gi/o, Gq/11, and G12/ 13) and the missense changes are either identical (R3 H) or different (R3 C, W, Q, or L) from that originally found in G␣o (R243H). The high conservation of this residue and the fact that it is altered in different G protein families suggests an impact of its mutation on G protein function. It is tempting to speculate that mutation of this conserved arginine in different G␣-subunits promotes a gain-of-function, hyperactive phenotype anal-ogous to that observed for the previously characterized G␣o R243H oncomutant (36). This would be analogous to what occurs with the "classical" R3 C/Q3 L oncomutants. However, we came across a clinical case report of an AHO patient (30) bearing a G␣s mutation, namely R265H, analogous to G␣o R243H (Fig. 1A). Because AHO is caused by the acquisition of inherited G␣s mutations leading to a loss-of-function, hypoactive phenotype, this suggests that identical, naturally occurring mutations in homologous ␣-subunits cause different phenotypes associated with two unrelated human diseases. We set out to investigate this quandary by comparing the properties of the disease-associated mutants G␣o R243H and G␣s R265H.
As a first step to compare these two mutants we investigated the structural features of this conserved residue. Arg-243 is located in the switch III region of G␣o and the most prominent feature associated with it in the structure of active G␣o is its close proximity to Glu-43, suggestive of an electrostatic interaction between the two residues (Fig. 1B). Glu-43 is located in the P-loop, a highly conserved sequence in G␣ subunits which is required for nucleotide binding (39). Because of this, we previously postulated (36) that the increased rate of nucleotide exchange observed in G␣o R243H was due to an alteration of the P-loop associated with the disruption of the electrostatic interaction with Glu-43. However, when we overlaid the structure of active G␣s (40) over that of G␣o (41) no significant differences were observed: i.e. the proximity and orientation of the Arg-265 and Glu-50 in G␣s is almost identical to Arg-243 and Glu-43 in G␣o (Fig. 1B). The overall conformation of the P-loop and SwIII are also virtually identical. These observations suggest that the different functional consequences of the R3 H mutation on either G␣o or G␣s do not arise from a dramatic structural difference. In fact, Glu-43 is highly conserved across all G␣ subunits with the exception of G␣z (which has unique biochemical properties among ␣-subunits (42) (Fig. 1B) and the electrostatic contact with Arg-243 is observed in all the G␣ subunits crystallized to date (not shown).
Validation of G␣s R265H as a Bona Fide Loss of Function Mutation in AHO-Next we decided to verify the previously described G␣s R265H mutation to rule out a possible error in the assignment of the mutation, which would explain the difference in behavior from G␣o R243H. This mutation, which was reported 12 years ago in a non-English written journal (30) and cited only once (43), has not been seen in other AHO patients or confirmed by others. More importantly, the sequencing data result was not shown in the original report and the description of the mutation was only stated (30). We obtained samples of the two living members of the family with diagnosed AHO, i.e. mother and son ( Fig. 2A) directly from the physician in charge of the case. It was not possible to obtain samples of a third member of the family (grandmother) with possible signs of AHO because she was deceased. We amplified by PCR and sequenced a region of the genomic DNA spanning GNAS exon 10. We found two overlapping peaks corresponding to A and C at position 794 of the G␣s coding sequence in the sequencing chromatogram for both mother and son. These findings are consistent with a heterozygous 794AϾC mutation and confirm the presence of R265H as an inherited mutation in this family.
AHO-linked mutations are frequently associated with destabilization of G␣s and decreased expression in the cellular environment (10). To test if the stability of G␣s was affected by the R265H mutation we compared the levels of G␣s WT and R265H exogenously expressed in HEK293 cells. To control for the amount of plasmid transfected we monitored GFP expression from an internal ribosome entry site (IRES) under the control of the promoter driving G␣s expression in the same plasmid. We found that G␣s R265H expression was comparable to WT upon transfection of equal plasmid amounts (Fig. 2B), sug- gesting that the stability of the mutant in cells is not compromised. We further validated this by assessing the half-life of G␣s WT or G␣s R265H proteins expressed in HEK293 cells after blocking protein translation with cycloheximide. G␣s WT and G␣s R265H protein levels decreased at similar rates (t 1/2 ϳ1.5 h) in the course of the next hours after cycloheximide treatment (Fig. 2, C and D). Taken together, our results indicate that protein destabilization is not the underlying cause of the loss of function in AHO associated with the R265H mutation.
To further investigate the underlying cause of the impaired G␣s function associated with the R265H mutation we analyzed its ability to bind nucleotides in cells. For this we used a well-established assay that monitors G protein activation upon nucleotide binding by measuring its susceptibility to limited trypsinolysis (33,44). Briefly, inactive, GDP-bound G␣ is readily digested by trypsin whereas activation upon GTP␥S or GDP-AlF 4 Ϫ binding renders it into a conformation in which only a short N-terminal sequence can be cleaved and the rest remains trypsin-resistant. We expressed G␣s WT and G␣s R265H in HEK293 cells and assessed their susceptibility to proteolysis upon loading with different nucleotides. We found that G␣s R265H protection from trypsinolysis upon GTP␥S or GDP-AlF 4 Ϫ loading was dramatically decreased compared with G␣s WT (Fig. 2E). As an alternative assessment of G␣s function in FIGURE 2. G␣s R265H is a bona fide loss-of-function mutation in AHO. A, left, pedigree of the family affected by the G␣s R265H mutation. Squares, male; circles, female; black symbols, AHO. The question mark indicates a deceased patient who had external signs of AHO but was not formally diagnosed before death. The table below summarizes clinical features of the two patients with confirmed AHO. Right, sequence analysis of a fragment of G␣s exon 10 displaying a heterozygous substitution of A for G in codon 265. B, G␣s R265H protein expression levels in mammalian cells are similar to G␣s WT. HEK293 cells were transfected with equal amounts of plasmids encoding for G␣s WT or G␣s R265H, lysed, electrophoresed, and immunoblotted with the indicated antibodies. Equal G␣s WT and G␣s R265H plasmid transfection is verified by the equal levels of GFP expression, which is driven by the same promoter as G␣s via an IRES. One experiment representative of three is shown. C and D, G␣s R265H protein half-life in mammalian cells is similar to G␣s WT. HEK293 transfected with G␣s WT or G␣s R265H were treated with cycloheximide (20 g/nl) for the indicated times, lysed, and immunoblotted with G␣s antibodies. A representative blot is shown in C, and the quantification of two independent experiments (mean Ϯ S.D.) fitted to a single exponential decay curve is shown in D. E, G␣s R265H expressed in mammalian cells has impaired nucleotide binding as assessed by trypsin protection assays. HEK293 cells expressing G␣s WT or G␣s R265H were lysed and incubated in the presence of GDP, GTP␥S, or GDP plus AlCl 3 /NaF (AlF 4 Ϫ ) before treatment with trypsin as described in "Experimental Procedures." Samples were subjected to SDS-PAGE and immunoblotted with the indicated antibodies. One experiment representative of three is shown. cells we measured changes in intracellular cAMP levels. Expression of G␣s WT in HEK293 cells caused an increase of 60 Ϯ 13 fmol of cAMP/g protein over the levels in control cells (empty vector-transfected), whereas with G␣s R265H it was only 15 Ϯ 7 fmol of cAMP/g protein (average Ϯ S.E., n ϭ 3, p Ͻ 0.05). These results indicate that G␣s R265H activity in cells is impaired and together with the previous results validate G␣s R265H as a bona fide loss-of-function mutation in AHO.
G␣s R265H Has Impaired Activity in Vitro due to Defective GTP Binding-Next we investigated the biochemical properties of purified His-G␣s R265H and compared it side by side with His-G␣o R243H, which had been previously characterized (36). We first measured the steady-state GTPase activity of His-G␣o and His-G␣s mutants and their corresponding wild-types (WT). It is important to note that G␣o and G␣s WT have very similar rates of nucleotide exchange (k ϭ 0.27 and k ϭ 0.34 min Ϫ1 , respectively) and nucleotide hydrolysis (k ϭ 2.4 and k ϭ 3.5 min Ϫ1 , respectively) (45,46). The steady-state GTPase activity of G␣ subunits reflects a reaction that proceeds in two major steps: GDP for GTP nucleotide exchange followed by GTP hydrolysis. It is known (45,46) that for both G␣o and G␣s WT the GTP hydrolysis step is faster than nucleotide exchange and that GDP release is rate-limiting in this process. Thus, the steady-state GTPase activity of G␣o and G␣s WT reflects the rate of GDP release and subsequent nucleotide exchange. Consistent with our previous findings (36), steady-state GTP hydrolysis by His-G␣o R243H was ϳ4 -5 fold faster than by His-G␣o WT (Fig. 3A), which is in good agreement with its increased rate of nucleotide exchange (36). On the contrary, steady-state GTP hydrolysis by His-G␣s R265H was virtually suppressed compared with His-G␣s WT (Fig. 3A). These results suggest that R265H mutation dramatically impairs GTP binding by G␣s. However, it is possible that the diminished steady-state GTPase activity observed for G␣s R265H arises from a defect in nucleotide hydrolysis rather than a defect in nucleotide binding/exchange. To rule out this possibility we measured directly the rate of nucleotide binding by monitoring the increase of G␣ intrinsic tryptophan fluorescence induced upon GTP␥S binding (37). The His-G␣o R243H tryptophan fluorescence time course revealed a rate of GTP␥S binding 5-6-fold faster than WT (Fig. 3B), which is in excellent agreement with previous observations using a radioligand binding assay (36). On the other hand, the increase of His-G␣s tryptophan fluorescence upon GTP␥S binding was dramatically impaired by the R265H mutation (Fig. 3B). These results indicate that the decreased steady-state GTPase activity observed for G␣s R265H (Fig. 3A) is most likely caused by impaired GTP binding rather than impaired GTP hydrolysis. Moreover, this interpretation is fully consistent with the loss-of-function phenotype commonly associated with AHO mutations, whereas GTPase deficiency would be expected to promote a gain-of-function.
To further confirm the defect in GTP binding we used limited proteolysis assays as in Fig. 2C but with purified His-G␣s. We found that both His-G␣s WT and His-G␣s R265H were readily digested when loaded with GDP, as expected, but that His-G␣s R265H incubated with GTP␥S was only marginally protected from trypsin digestion compared with His-G␣s WT (Fig. 3C). This result is in excellent agreement with that obtained with the same assay on G␣s expressed in cells (Fig.  2C). On the other hand, and consistent with previous observations (36), G␣o WT and R243H were equally protected from trypsin upon incubation with GTP␥S. These results not only confirm that G␣s R265H has impaired GTP binding but also FIGURE 3. Comparison of the nucleotide binding properties of G␣o R243H and G␣s R265H. A, mutation of Arg-243 to His accelerates the steadystate GTPase activity of His-G␣o whereas mutation of Arg-265 to His in His-G␣s dramatically impairs it. The amount of [␥-32 P]GTP (25-50 cpm/fmol) hydrolyzed by G␣o (left) or G␣s (right) WT (blue) or R3 H mutant (red) was determined as described in "Experimental Procedures." Results are shown as mean Ϯ S.D. of one experiment representative of three performed in duplicate. B, mutation of Arg-243 to His accelerates GTP␥S binding to His-G␣o whereas mutation of Arg-265 to His in His-G␣s dramatically impairs it. GTP␥S binding by G␣o (left) or G␣s (right) WT (black traces) or R3 H mutant (red traces) was determined by intrinsic fluorescence measurements as described in "Experimental Procedures." For G␣o proteins, data are expressed as % of their own maximal binding which was F/Fo ϭ 1.20 and F/Fo ϭ 1.26 for WT and R243H, respectively. One experiment representative of three is shown. C, mutation of R265 to H dramatically impairs His-G␣s protection from trypsin hydrolysis upon GTP␥S incubation whereas mutation of Arg-243 to His has no effect on His-G␣o. His-G␣s WT or R265H (top) and His-G␣o WT or R243H (bottom) were incubated in the presence of 30 M GDP or 30 M GTP␥S before treatment with trypsin as described in "Experimental Procedures." One representative experiment is shown on the left, and a graph with the mean Ϯ S.E. of three independent experiments is shown on the right. that purified His-G␣s R265H recapitulates the biochemical properties of the protein expressed in cells.
Both G␣s R265H and G␣o R243H Bind GDP with Diminished Affinity-To characterize if the impaired GTP binding observed for G␣s R265H is associated with an overall defect in nucleotide binding we performed additional limited proteolysis experiments. G␣ subunits are activated by AlF 4 Ϫ in addition than by GTP␥S. Because AlF 4 Ϫ -mediated activation requires the presence of GDP in the nucleotide binding pocket (to form the GTP mimetic GDP-AlF 4 Ϫ ), we used AlF 4 Ϫ induced protection from trypsinolysis as a surrogate measurement of GDP binding. We found that the R265H mutation dramatically impairs AlF 4 Ϫinduced protection from trypsinolysis at physiological concentrations (25 M) of GDP (Fig. 4A). Once again, this is consistent with the result of analogous experiments with G␣s expressed in cells (Fig. 2C). This defect is most likely due to decreased affinity for GDP rather than an intrinsic inability of His-G␣s R265H to adopt an active conformation because it was partially rescued at higher, supraphysiological concentrations of the nucleotide (Fig.  4A). Taken together these results indicate that G␣s R265H has dramatically impaired affinity for both GDP and GTP nucleotides.
In a previously published work (36), we showed that, contrary to His-G␣s R265H, His-G␣o R243H binds GDP and GTP␥S at physiological concentrations as efficiently as His-G␣o WT under steady-state conditions. However, the faster rate of GTP␥S binding described for this mutant (36) suggests that it probably binds GDP loosely and releases it at a faster rate than in the WT protein to allow rapid GTP loading. We demonstrated that this is actually the case by directly measuring the rate of GDP-release (Fig. 4B): GDP release in His-G␣o R243H is ϳ5-fold faster than in His-G␣o WT. This acceleration of the rate of GDP release is identical when either GTP␥S or GDP are used as the displacing competitor (Fig. 4B), further confirming that the faster rate of G␣o R243H activation is due to the unfavorable retention of GDP in its nucleotide binding pocket.
Based on these results we propose a model (Fig. 4C) in which the overall phenotype and disease associated with either G␣s R265H or G␣o R243H arise from a decrease in nucleotide affinity but that it occurs to a different grade depending on the G protein. In G␣o, the R3 H mutation promotes a modest decrease in nucleotide affinity that does not preclude nucleotide loading at physiological concentrations. Instead, this mild decrease in affinity favors the rapid exchange of nucleotides and subsequent hyperactivation of the G protein that triggers oncogenesis (36,43). On the other hand, in G␣s, the R3 H mutation translates into a profound decrease of nucleotide affinity that impairs the overall binding of nucleotides even at concentrations several orders of magnitude larger than physiological. As a consequence, G␣s R265H is hypoactive and causes the impaired hormonal responsiveness associated with AHO.
We attempted to dissect the structural basis for the different functional consequences of the R3 H mutation in G␣s versus G␣o. We reasoned that this difference could arise from a difference in the amino acid sequence of the two proteins in the region adjacent to the mutation and the nucleotide binding pocket. Among the divergent residues between G␣s and G␣o in this region, only G␣s Arg-258 (which corresponds to His-236 in G␣o) seemed to be able to directly contribute to the nucleotide binding properties: it forms an electrostatic contact with Glu-50 (supplemental Fig. S1A), and its mutation in AHO impairs nucleotide binding (14). We hypothesized that the presence of an Arg in position 258 of G␣s could contribute to the impaired nucleotide binding observed upon R2653 H mutation, and that G␣o R243H does not manifest such defect because the corresponding residue is replaced by a histidine (His-236). If this was true, the double G␣o mutant H236R/ R243H would mimic the properties of G␣s R265H by displaying impaired nucleotide binding. However, we found that this is not the case because G␣o H236R/R243H bound GTP␥S efficiently and with properties analogous to G␣o R243H (supplemental Fig. S1B). This result suggests that other amino acid differences among G␣s and G␣o may account for the observed differences, but this was not investigated further because we could not rationally design additional substitutions predicted to influence nucleotide binding.
Divergent Functional Consequences in Other Disease-related G␣ Mutations-To our knowledge the results presented in this work provide the first example of two naturally occurring, identical mutations associated with human disease that render two homologous G proteins with opposite behavior (gain-of-function versus loss-of-function). This contrasts with the "classical" G protein cancer mutants R3 C and Q3 L, which have highly conserved functional consequences when introduced in virtu- ally any G protein. In addition, the only G␣s mutation associated with AHO (and testotoxicosis) (17) that has been introduced in another G␣ subunit (G␣i1, Ref. 47) also seems to have conserved functional consequences. Nevertheless, the behavioral disparity of disease-associated mutants described here may not be an exception. Analysis of the literature (48 -50) and the cBioportal/COSMIC databases revealed additional examples of conserved G␣ residues that are mutated both in G␣s of AHO patients and in other G␣ proteins in cancer (Fig. 5). It is likely that these mutations shared by AHO and cancer represent additional examples of differing mechanisms analogous to that described for the G␣s R265H/G␣o R243H pair.
To test this idea we investigated the biochemical properties of the G␣s D173N mutant described in AHO, for which analogous mutations have been found in other G␣ subunits in cancer (Fig. 5). It is important to note that G␣s D173N has been previously characterized in cells from AHO patients (49), in which it is expressed at normal levels but its function is compromised. Based on this and the fact that Asp-173 is located in the ␣D-␣E loop important for the stabilization of the nucleotide binding pocket we investigated the nucleotide handling properties of the purified protein. We found that G␣s D173N, like G␣s R265H, has greatly impaired GTP␥S binding as determined by tryptophan fluorescence measurements (Fig. 6A) and trypsin protection assays (Fig. 6B). It also displayed reduced protection from trypsin upon GDP-AlF 4 Ϫ loading (Fig. 6B), suggesting a defect in GDP binding as well. We compared side by side the properties of the corresponding mutant in G␣o (i.e. D151N) as a surrogate for the cancer mutation found in G␣t2, G␣16, and G␣z. We decided to investigate G␣o instead of the G␣ subunits actually mutated in cancer (G␣t2, G␣16, G␣z) because they are extremely challenging to purify in the required quantities. Moreover, their rates of nucleotide binding and hydrolysis are very different from those of G␣s, which would complicate the interpretation of the results. We found that GTP␥S binding to G␣o D151N was much faster than to WT (Fig. 6C) and that trypsin protection upon GTP␥S or GDP-AlF 4 Ϫ was indistinguishable from that observed in G␣o WT (Fig. 6D). This suggests that the D151N mutation in G␣o leads to a gain-of-function due to accelerated nucleotide exchange. Thus, the results with G␣s D173N/G␣o D151N pair highly resemble the same loss/gain of function dichotomy observed for the G␣s R265H/ G␣o R243H pair, thereby supporting the concept that analogous mutation in different G␣ subunits can be associated with unrelated diseases.
Final Remarks-Overall, the observations reported in this work raise an important issue related to large genome sequencing campaigns like those carried out by The Cancer Genome Atlas. Although the recent advances in DNA sequencing technologies are bound to provide a wealth of information on the molecular basis of disease, the data generated through these efforts must be analyzed with caution to avoid misinterpretation of datasets. Assumptions on the functional outcome of given mutations in systems level analyses must be subjected to rigorous scrutiny even if conserved among homologous proteins, There are previous examples of identical mutations having significantly different functional consequences depending on the G protein, although all of them are synthetic, man-made mutations. Particularly relevant to the work presented here was the investigation of G␣s R265E in comparison to the same mutation in a G␣t/i chimera (51). The R3 E mutation was originally reported to render the G␣t/i chimera with dominantnegative properties, i.e. it is incompetent for GTP␥S-mediated signaling activation and irreversibly binds to receptors (52). On the contrary, the same mutation in G␣s renders it a G protein capable of activating effectors upon GTP␥S treatment and binds to receptors with unaltered affinity (51). However, a common feature of the R3 E mutants among G␣s and G␣t/i as well as with the R3 H mutants in the same position described in this work for G␣o and G␣s, is the apparent reduced affinity for nucleotides. Thus, the position corresponding to Arg-265 in G␣s appears to play a critical role in controlling the nucleotide binding properties of multiple, if not all, G proteins, although the precise consequences of mutating it vary from G protein to G protein. It is possible that such differences arise from minor sequence differences found in the SwIII region of different G proteins (51). In summary, this work provides novel information about the importance of conserved G protein mutations in the molecular basis of disease and raises a cautionary note for future research on this topic.