Mutation of Alanine 623 in the Third Cytoplasmic Loop of the Rat Thyrotropin (TSH) Receptor Results in a Loss in the Phosphoinositide but Not cAMP Signal Induced by TSH and Receptor Autoantibodies*

Thyrotropin (TSH) and IgG preparations from pa- tients with Graves’ disease increase inositol phosphate as well as cAMP formation in cos-7 cells transfected with rat TSH receptor cDNA. Mutation of alanine 623 in the carboxyl end of the third cytoplasmic loop of the TSH receptor, to lysine or glutamic acid, results in the loss of TSH- and Graves’ IgG-stimulated inositol phosphate formation but not in stimulated cAMP formation. There is no effect of the mutations on basal or Pz-purinergic receptor-mediated inositol phosphate for- mation. The mutations do not affect transfection efficiency or the synthesis, processing, or membrane in- tegration of the receptor, as evidenced by the un- changed amount and composition of the TSH receptor forms on Western blots of membranes from transfected cells. The mutations increase the affinity of the TSH receptor for [‘2sI]TSH and decrease B,,,; however, cells with an equivalently decreased B,,, as a result of transfection with lower levels of wild type receptor do not lose either TSH-induced inositol phosphate formation or cAMP signaling activity. Thus, in addition to discriminating between ligand-induced phosphatidylinositol

Thyrotropin (TSH) and IgG preparations from patients with Graves' disease increase inositol phosphate as well as cAMP formation in cos-7 cells transfected with rat TSH receptor cDNA. Mutation of alanine 623 in the carboxyl end of the third cytoplasmic loop of the TSH receptor, to lysine or glutamic acid, results in the loss of TSH-and Graves' IgG-stimulated inositol phosphate formation but not in stimulated cAMP formation. There is no effect of the mutations on basal or Pzpurinergic receptor-mediated inositol phosphate formation. The mutations do not affect transfection efficiency or the synthesis, processing, or membrane integration of the receptor, as evidenced by the unchanged amount and composition of the TSH receptor forms on Western blots of membranes from transfected cells. The mutations increase the affinity of the TSH receptor for ['2sI]TSH and decrease B,,,; however, cells with an equivalently decreased B,,, as a result of transfection with lower levels of wild type receptor do not lose either TSH-induced inositol phosphate formation or cAMP signaling activity. Thus, in addition to discriminating between ligand-induced phosphatidylinositol bisphosphate and cAMP signals, the mutation appears to cause an altered receptor conformation which affects ligand binding to its large extracellular domain.
The thyrotropin (TSH)' receptor is coupled to the PIP, as well as the cAMP signal system in FRTL-5 rat thyroid cells; the a,-adrenergic receptor (al-AR) couples only to the former (1)(2)(3)(4)(5). The rat TSHR, like the a'-AR, has seven transmembrane domains (6,7); it is, however, an example of a single receptor coupled to more than one G protein (8). Thus, there is a single TSHR gene (6,9,10) and transfected TSHR cDNA * The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
The abbreviations used are: TSH, thyrotropin; TSHR, thyrotropin receptor; AR, adrenergic receptor; PIP2, phosphatidylinositol bisphosphate. confers both cAMP and PIP, responses in Chinese hamster ovary cells (11). Understanding how the TSHR couples to two rather than one G protein is important to understand differences between the effects of TSHR and al-AR on signal transduction, growth, and function (1)(2)(3)(4)(5). It is also important to understanding the action of the autoantibodies in patients with Graves' disease. Graves' IgG can induce the release of arachidonic acid from rat FRTL-5 thyroid cells (12); however, it has not been established that this results from perturbation of the PIP2 cascade via the TSHR. Thus, autoantibody initiation of the PIP, cascade could not be demonstrated when evaluated in a human thyroid cell system (13).
The present report shows that alanine 623 in the third cytoplasmic loop of the TSHR seven-transmembrane domain ( Fig. 1) is critical for the ability of TSH and Graves' IgG to initiate the PIP, but not the cAMP signal of cos-7 cells transfected with rat TSHR cDNA. The selectivity and novelty of the mutation is underscored by studies of the alB-AR, which showed that the same mutation of the comparably located alanine constitutively activated the PIP, signal while minimally impacting the maximal hormone response (7), and by studies of the TSHR, which suggested all three intracytoplasmic loops were important for TSH-induced cAMP signal transduction (14).
Mutagenesis-Oligonucleotide-mediated, site-directed mutagenesis (15) was used to convert alanine 623 to glutamic acid (6233) or lysine (62310 and the derived clones inserted in the EcoRI site of pSG5. Amplified DNA was purified by CsCl gradient centrifugation; each mutant had the sequence predicted and produced RNA appropriate in size and amount, relative to both wild-type TSH receptor and p-actin RNA. Transfection and Assays-Transfection of cos-7 cells with mutant DNA was performed by electroporation (6,15); 25 pg of purified plasmid DNA was used unless otherwise noted. Transfection efficiency was measured by CotransfectingpSVGH and measuring growth hormone concentration in the media (10). The same batch of transfected cells was plated in Dulbecco's modified Eagle's medium with 10% fetal calf serum using six-well plates (5 X lo5 cells/well) for TSH binding assays or 24-well plates (1 X 10% cells/well) for cAMP and inositol phosphate assays; medium was inositol-free in the latter assays and supplemented with 2.5 pCi/ml my0-[2-~H]inositol (Du Pont-New England Nuclear; specific activity approximately 20 Ci/ mmol). All three assays were simultaneously initiated 48 h after transfection and after washing with assay buffer: NaC1-free Hanks' balanced salt solution containing 0.5% bovine serum albumin, 222 mM sucrose, and 20 mM HEPES at pH 7.4 (15). [''"IITSH binding was measured after incubation for 2 h a t 22 "C in 1 ml of assay buffer containing 1.5 X lo5 cpm ['''I]TSH (60 pCi/pg) and 0 to lo-' M unlabeled TSH (6,15). Specific TSH binding was obtained by subtracting values obtained in the presence of 10" M unlabeled TSH. Total cAMP and inositol phosphate levels were measured in the same wells after incubation for 1 h a t 37 "C with 0.2 ml of assay buffer containing 10 mM LEI, 0.5 mM 3-isobutyl-l-methylxanthine, and, as noted, 10"' to 10" M TSH, 0.5 or 5.0 mg/ml Graves' or normal IgG, 10 p~ ATP, 10 ng/ml cholera toxin, or 2 p~ forskolin. After the addition of 1.0 ml of 5% perchloric acid, samples were centrifuged to remove protein debris, then neutralized with KOH and centrifuged t o remove insoluble salts. Total cAMP in aliquots of the supernatant was measured by radioimmunoassay (6,15); inositol phosphate formation was determined using Dowex AGl-X8 columns (mesh size 100-200) eluted with 1 M ammonium formate, 0.1 M formic acid after removal of free inositol with HzO and glycerophosphoinositol with 5 mM sodium tetraborate, 60 mM sodium formate (5).
All assays were performed in duplicate, on a t least three separate occasions with different batches of cells, and with simultaneously run positive and negative controls: cells transfected with wild type receptor or pSG5 vector alone. Values in each well were corrected for cell protein (6,15). The program LIGAND (16) and a single high affinity binding site model, which best fit all data, were used to calculate K d values for TSH binding and ECm values for TSH-increased inositol phosphate and cAMP levels (15).
Western Blots-Western blots of membrane proteins derived from cos-7 cells transfected with wild type or mutant receptor cDNA were performed as described (17) and used an antibody against residues 352-366, which are specific to the TSH receptor (17).

RESULTS
Mutation of Alanine 623 Does Not Appear to Alter Synthesis, Processing, or Integration of the TSHR into the Plasma Membrane but Does Alter r2'I]TSH Binding-Mutation of alanine 623 to glutamic acid (6233) or lysine (623K) had no effect on transfection efficiency as measured by cotransfecting 0.1 pg of pSVGH; thus, growth hormone levels were 413 f 32 ng (mean f S.E.) in all transfections. In addition, the mutations did not alter the composition or amount of TSHR forms detected on Western blots of membranes from cells. Thus, similar amounts of the same three major TSHR forms, 230, 180, and 95 kDa, were identified (Fig. 2). The 230-kDa form is the unprocessed precursor of the receptor, the 95-kDa protein is the fully glycosylated receptor form associated with cell surface activity, and the 180-kDa form appears to be a biosynthetic intermediate, not a dimer of the 95-kDa form (17).
The two mutants exhibit specific and saturable ['2sII]TSH binding. Displacement data (Fig. 3A) or saturation binding assays using concentrations of ['251]TSH between 400 and 700,000 cpm/well (data not shown) indicated that K d values for the 623K and 6233 mutants were 16-or 8-fold lower than wild type receptor, 14 k 2, 28 f 7, and 231 f 53 pM, respectively (Fig. 3A, insert); B,,, was, however, decreased by 8and 5-fold to 12 k 1.5 and 21 f 1%, respectively, of wild type receptor (Fig. 3A, insert). Decreases in B,,, comparable to the mutants, i.e. to 43.1 k 1.8 or 13.2 +. 2.1%, could be obtained by decreasing the amount of transfected wild type plasmid DNA from 25 to 2.5 or 1 pg, respectively; in this circumstance  (17). After blotting, ing 5% 8-mercaptoethanol, and 10 pg of membrane protein was nitrocellulose membranes were sequentially incubated with a rabbit antibody against the synthetic TSHR peptide (residues 352-366) and 'ZsI-labeled donkey anti-rabbit Ig F(ab'h The 230-, 180-, and 95-kDa proteins are not identified with preimmune serum nor with immune serum preincubated with peptide 352-366 (17). there was, however, no comparable change in K d .
Mutation of Alanine 623 Results in a Loss in TSH-induced Inositol Phosphate Formation but Not TSH-induced cAMP Signal Generation-TSH-increased inositol phosphate formation in cos-7 cells transfected with the wild type rat TSHR (Fig. 3B) is concentration-dependent and is evident at physiologic TSH concentrations (Fig. 3B); it is not evident in control cells transfected with pSG5 alone (Fig. 3B) or with antisense receptor constructs (data not shown). Maximal activity, 3.2-fold above basal, can be decreased to 2.5-and 2.2-fold above basal in cells transfected with 2.5 or 1 pg of plasmid DNA, respectively (Table I). The decrease in maximal activity in this case is not associated with a change in EC50 and is less than the decrease in the Bmax of TSH binding. For example, with 1-pg transfections, the decrease in maximal activity is only 43% whereas the decrease in maximal binding is 87%.
In contrast to cells transfected with 25 pg of wild type receptor plasmid, TSH did not increase inositol phosphate formation in cells transfected with the same amounts of the 6233 mutation and increased inositol phosphate levels only slightly at very high concentrations of TSH ( M) in cells transfected with the 623K mutation (Fig. 3B). The loss in TSH-induced inositol phosphate formation reflected a loss both in measurable EC50 and maximal responsiveness in both mutants (Table I, Fig 1). The loss could not be accounted for by the decrease in B,,, exhibited by these mutants. Thus, even the 62313 response was negligible compared with cells transfected with 1 pg of wild type receptor (Fig. 3B) despite nearly identical B,,, values. The loss of response by both mutants could, therefore, be attributed to the effect of the mutation on signal transduction not receptor number.
The loss of TSH responsiveness in the 6233 and 623K mutations was receptor-specific. Thus, ATP increased inositol phosphate formation equally in cells transfected with wild type receptor, 6233, 623K3, or the pSG5 vector alone (Fig.  30).
IgG from a Graves' patient with autoantibodies to the TSHR, which increased cAMP levels (see below), also increased inositol phosphate formation in cells transfected with wild type rat TSHR plasmid (Fig. 3C). As was the case for TSH, IgG-increased inositol phosphate formation was lost in cells transfected with the same amount of 6233 and 623K mutant plasmid (Fig. 3C). Similar results were obtained using two separate Graves' IgG preparations; IgG from four normal individuals had no effect on inositol phosphate formation in transfectants with wild type receptor.
In contrast to the loss of TSH and autoantibody responsiveness in the inositol phosphate assays, cells transfected with the 623K or 6233 mutants still retained significant TSHinduced (Fig. 4A) and Graves' IgG-induced (Fig. 4B) cAMP signal generation. In the case of the 623K mutant, the TSH effect was similar to that of the wild type receptor as a function of TSH concentration, having a similar EC50 value and no major change in maximal response relative to basal values (Fig. 4 A ; Table I). In the case of the 6233 mutant, the TSH-induced cAMP response is also retained reasonably well, the maximal response above basal actually being better (8.4-fold uersus 6.0-fold; p < 0.05) than in cells transfected with C

FIG. 4. The ability of TSH ( A ) , a Graves' IgG ( B ) , and forskolin or cholera toxin (C) to increase cAMP levels in Cos-7 cells transfected with wild type and mutant TSH receptors.
Data are expressed as the mean * S.E. of all experiments as in Fig.   3; the number of experiments is given in Table I. the same amount (25 pg) of wild type receptor (Fig. 4 A ; Table   I).
Cells transfected with comparable amounts (25 pg) of the 623K and wild type, but not the 6233 plasmid, exhibited increased basal cAMP levels relative to pSG5 (Fig. 4 A ; Table   I) and antisense controls (data not shown). The increase was a function of the amount of transfected receptor (Table I; wild type basal) and was specific for cAMP accumulation, since there was no comparable increase in basal (Table I) or ATP-induced inositol phosphate formation (Fig. 3 0 ) . The increase in basal activity was not paralleled by a change in cholera toxin-stimulated activity (Fig. 4C) but did correlate with changes in forskolin-induced activity (Fig 4C). Thus, in addition to a low basal activity, it was not surprising that cells transfected with 25 pg of the 6233 mutant had a forskolin-induced response similar to cells transfected with 1-2.5 pg of wild type plasmid DNA (Fig. 4C) but a cholera toxin response the same as cells transfected with 25 pg of wild type DNA. In sum, the 6233 mutation may have an altered interaction with the G,-adenylylcyclase complex by comparison to wild type or 623K receptor, but this only minimally effects the TSH-induced cAMP response.

DISCUSSION
The TSHR is one of a number of receptors for physiologically important hormones, i.e. parathormone, calcitonin, and gonadotropins, which couple to different G proteins associated with stimulation of the cAMP and PIPz cascade (8). This is the first report of a point mutation which discriminates between the two ligand-induced signals of this group of seven transmembrane receptors. Thus, we show that mutation of alanine 623 of the TSHR to glutamic acid (6233) or lysine (623K) results in a loss of TSH-increased inositol phosphate but not cAMP formation. These results may be compared to mutations that selectively perturb coupling to only one of the effector pathways used by aZA-and ,8-AR receptors. Thus, mutation of aspartic acid 79 in WAR (18) results in a loss in agonist increased potassium currents but no change in agonist-inhibited adenylylcyclase or calcium currents; deletions of residues 222-229 and 258-270 in 8-AR (19) lose G.-dependent stimulation of adenylylcyclase but not non-G-coupled Na-H exchange.
Although the cAMP signal is important for growth and function of the thyroid, it is not the only signal implicated (1-6, 11, 12). The hyperfunction of Graves' patients does not correlate with goiter; neither correlates quantitatively with measurements of cAMP elevations (12). This led to the possibility that receptor autoantibodies perturbing signals other than cAMP were important to understand the pleiotropic expression of the disease (1-4, 6, 12). The present report is the first to unequivocally demonstrate that autoantibodies to the TSHR stimulate the PIPz cascade. Furthermore, the discrepancy between the effect of the two concentrations of Graves' IgG on inositol phosphate uersus cAMP elevations in cells transfected with 25 pg of wild type receptor (Figs. 3C uersus 48, respectively) raise the possibility that receptor autoantibodies exist which perturb the two signals differently. This conclusion has been confirmed by observations of multiple Graves' IgG preparations' and is consistent with studies measuring autoantibody-modulated arachidonic acid release (12) and receptor mRNA levels (6).
Residues 252-259 in the middle of the third cytoplasmic loop of the aI-AR are critical for coupling to a pertussis toxininsensitive G protein involved in activation of the PIP, cascade (20). The third cytoplasmic loop of the TSHR has twothirds fewer residues than the aI-AR and no homologous residues to this region (6,7,20). Alanine 623 of the TSHR receptor is, however, comparably located to alanine 293 in the third cytoplasmic domain of the aIB-AR (6, 7). Mutation of alanine 293 of the aIB-AR to glutamic acid or lysine causes, in contrast to data herein, a significant enhancement of basal inositol phosphate formation, from approximately 2 to 200% above control, but no change in maximal ligand-increased activity despite a lower ECbO (7). Nevertheless, it is clear that the comparable alanines in TSHR and a1B-AR are specifically linked to initiation of the PIP, cascade, probably by influencing receptor affinity for a pertussis toxin-insensitive G protein involved in phospholipase C activation. These data are particularly relevant to gonadotropin receptors, whose sequences are identical in this area (6), but may be generally important to other seven-transmembrane domain receptors in the group.
Transfections with mutant receptors may result in decreased synthesis, processing, and insertion into the membrane. This is not the case here since there is no change in receptor forms present in the membranes on Western blots. Nevertheless, both alanine 623 mutations lower the K d and decrease the B,,, of ligand binding, as was the case when alanine 293 of the cul~-AR was mutated (7, 20). There is no comparable change in and even an increase (6233) in the maximal TSH-increased cAMP response relative to basal. Loss of PIPz coupling cannot be accounted for by alterations in absolute receptor number, since comparable changes are not evidenced when the Bmax of TSH binding is decreased nearly 90% by using less wild type receptor plasmid in the transfection. In the case of the a1B-AR (7), altered ligand binding is believed to result from an altered conformation associated with abnormal coupling to the G protein. In the case of the TSHR, altered ligand binding resultant from the conformation change may reflect the selective loss of coupling between mutant receptors and the G protein involved in phospholipase C activation but not G.. The TSHR binds ligand via its large extracellular domain (15, 17). The present data appear, therefore, to establish that there is a conformational relationship between the signaling domain on the cytoplasmic surface of the cell and the large extracellular binding domain of the receptor. This was a possibility suggested for the lutropin receptor when aspartic acid 383 in the 2nd transmembrane domain was mutated (21).
The integrity of all three cytoplasmic loops of the TSHR is important for the TSH-induced increase in cAMP levels (14). As in the case of the P-AR, receptor coupling to G. can thus be presumed to involve a broad array of determinants (19,22). Moreover, it has been shown (14) that simultaneous substitution of all residues between 617-625, other than isoleucine 622 and alanine 623, results in a loss of the TSHincreased cAMP signal, whereas substitution of each residue individually has no effect (23). Caution must be used in interpreting the former TSHR result (14), since there is an associated 10-fold decrease in receptor affinity and since no measurements of Bmax or inositol phosphate formation were made nor was there evidence of normal receptor synthesis or integration into the bilayer (14). Nevertheless, the data (14, 23) are consistent with the possibility that G. interacts with residues surrounding alanine 623 in this region. These data (14) may account for the effect of the 6233 mutation on basal and forskolin-induced cAMP levels; more importantly, they emphasize the selectivity and novelty of the present mutation.