Altered Signaling and Desensitization Responses in PTH1R Mutants Associated with Eiken Syndrome

Plasma membrane expression of WT and mutant PTH1Rs

The positions in the PTH1R of the three studied mutations of Eiken syndrome, E35K, Y134S and R485X, as well as the H223R mutation of JMC are shown in Fig. 1a and b. Cell surface expression of the receptors in transiently transfected HEK293-derived Gs22a cells was assessed via antibody binding to an HA epitope tag incorporated into the non-essential E2 region of the receptor’s ECD²⁷. The PTH1R-E35K mutant was expressed at levels comparable to those of PTH1R-WT, while the other mutants were expressed at 30-70% of the level of PTH1R-WT (Fig. 1c and Supplemental Fig. 1). These expression assays and subsequent functional assays were performed using an equivalent amount of plasmid DNA for each PTH1R variant (100 ng/well in 96-well plate assays), which provided the maximum level of expression possible for each variant, as indicated by the results of DNA-titration experiments (Supplemental Fig. 2a, b).

Basal cAMP generation and effects of β-arrestin2 overexpression

Basal levels of cAMP generation monitored via the glosensor reporter stably expressed in Gs22a cells²⁸ were markedly elevated in cells expressing PTH1R-R485X relative to those in cells expressing PTH1R-WT and were comparable to those observed for PTH1R-H223R (Fig. 2a and Supplemental Fig. 2b). Modest yet consistent elevations in basal cAMP were observed in cells expressing PTH1R-E35K, while basal cAMP levels for PTH1R-Y134S were comparable to those for PTH1R-WT. The basal cAMP signaling activity of PTH1R-R485X was similar to that observed for PTH1R-PD, in which serine residues at positions 489, 491, 492, 493, 495, 501 and 504 in the C-tail are replaced by alanine^11,28, whereas the basal signaling activity of PTH1R-546X, in which the C-tail is truncated at a site downstream of the phosphorylation cluster was comparable to that of PTH1R-WT (Fig. 2b). These results confirm prior findings of elevated basal cAMP signaling for PTH1R mutants lacking serine-threonine phosphorylation residues in the mid-region of the receptor’s C-tail^14,28 and also establish that the increase in basal signaling of PTH1R-R485X is due to the absence of these residues and not to the loss of other potential docking determinants in the more distal C-tail region, as used, for example, by NHERF scaffolding proteins^29,30.

Quite strikingly, co-transfection with β-arrestin2^YFP strongly suppressed the basal cAMP signaling activity of PTH1R-R485X, such that the resulting cAMP levels were comparable to those in cells expressing PTH1R-WT (Fig. 2c). These findings imply that that PTH1R-R485X can interact with β-arrestin despite the absence of most of the receptor’s C-tail. This interpretation is consistent with recent cryogenic electron microscopy (cryo-EM) structures of other GPCRs in complex with β-arrestin showing that interactions occur to the TMD core region in addition to the receptor C-tail^31,32,33, as well as a recent cross-linking study that identifies specific proximity points between β-arrestin2 and the both the TMD and C-tail regions of the PTH1R²⁶, as depicted in Fig. 1d. Co-transfection with β-arrestin2^YFP resulted in more modest, albeit measurable decreases in the basal cAMP signaling activity of PTH1R-WT and PTH1R-H223R. A previous study also found that the constitutive cAMP signaling of PTH1R-H223R could be at least partly suppressed by β-arrestin2 co-transfection¹². The marked difference we observe in the capacity of β-arrestin2 to suppress basal cAMP signaling by PTH1R-R485X as compared to that of PTH1R-H223R highlights the distinct mechanisms by which these two mutations lead to increased receptor activity, as the C-tail truncation most likely alters the basal-state interaction with β-arrestins while the H223R mutation in TM2 likely perturbs a conserved polar network that controls receptor activation and deactivation processes²³.

Agonist-induced signaling responses of PTH1R variants

We then assessed the capacities of the PTH1R mutants to activate cAMP signaling in response to stimulation by either PTH(1-34) or PTHrP(1-36) (Supplemental Table 1). PTH1R-E35K and PTH1R-Y134S responded to each ligand with potencies and efficacies that were comparable to those observed on PTH1R-WT while PTH1R-R485X exhibited a potency for each ligand that was enhanced 2- to 4-fold relative to that on PTH1R-WT, albeit the difference was significant only for PTHrP(1-36) (Fig. 3a, Supplemental Table 3, and Table 1). The response maximum (Emax) attained by each ligand on PTH1R-R485X was reduced to about 80% of that attained on PTH1R-WT, although the difference was again significant only for PTHrP(1-36). PTH1R-H223R exhibited moderately reduced potencies and efficacies for each ligand, which is consistent with prior findings for this PTH1R mutant³⁴.

Comparison of the time-course data for the increases in cAMP generation that occurred in the presence of PTH(1-34) and PTHrP(1-36) at a near-saturating concentration (3.0 nM) revealed that the peak signals attained by PTHrP(1-36) on PTH1R-E35K were moderately but consistently higher than those on PTH1R-WT, whereas the peak signals attained by PTH(1-34) on this mutant receptor were comparable to those on PTH1R-WT, as were those for each ligand on the other mutant receptors (Fig. 3b). After washing away the unbound ligand, the cAMP signals induced by PTH(1-34) decayed at similar rates in cells expressing PTH1R-WT, PTH1R-E35K or PTH1R-R484X, while those in cells expressing PTH1R-Y134S decayed at a moderately faster rate, as compared to with PTH1R-WT. For the cAMP signals induced by PTHrP(1-36), the rates of decay were noticeably faster for both PTH1R-E35K and PTH1R-Y134S, as compared to those for PTH1R-WT or PTH1R-R485X. Thus we observed marked reductions in the duration of the signaling responses induced by PTHrP(1-36) on both the PTH1R-E35K and PTH1R-Y134S variants, which suggests that despite a rapid initial onset of the signaling response, the active ligand-receptor formed by this ligand and these two receptor variants are relatively unstable.

Divergent residues at positions 5 and 19 in PTH and PTHrP as determinants of altered ligand interactions on PTH1R Eiken mutants

The cAMP time course data of Fig. 3b and c suggest that both the E35K and Y134S PTH1R mutations exert distinct effects on the interaction with PTH(1-34) versus PTHrP(1-36). Thus, while PTHrP initially activates each mutant receptor at least as efficiently as it does PTH1R-WT, the mutant complexes formed with PTHrP are less stable as compared to those formed with PTH, and also to those formed by PTHrP on PTH1R-WT. One notable site of amino acid divergence in the two ligands is position-5, which is Ile in PTH and His in PTHrP. These residues are known to be key determinants of altered selectivity effects that have been observed for these ligands on PTH1R-WT, as the His⁵–>Ile substitution in PTHrP peptides markedly enhances receptor binding affinity, likely due to a more optimal interaction of the Ile side chain with hydrophobic residues in the orthosteric pocket of the receptor’s TMD region^3,4,5,35. Consistent with this, we observed that Ile⁵-PTHrP(1-36), as compared to His⁵-PTHrP(1-36) induced a more sustained cAMP signaling response on each PTH1R variant, including PTH1R-E35K and PTH1R-Y134S (Fig. 3d).

We also investigated ligand residue 19, which diverges as positively charged arginine in PTHrP and negatively charged glutamate in PTH, and is known to be a second important determinant of ligand selectivity at the PTH1R³⁶. The cryo-EM structure of the LA-PTH/PTH1R complex shows that Arg¹⁹ in LA-PTH is relatively close (~8 A˚ at closest side chain atoms) to Glu³⁵ in the receptor (Fig. 1b); a proximity also observed in the more recent cryo-EM structures of the PTH1R in complex with native PTH(1-34) and PTHrP(1-34) peptides⁴. As Glu³⁵ is replaced by positively charged lysine in PTH1R-E35K, we assessed whether the differences in cAMP signaling actions we observed for PTH and PTHrP peptides on this variant might involve charge-based differences in interactions between the sidechains of the residue at position 19 in the ligands and at position 35 in the receptor. We used for these studies two pairs of PTH probe peptides, PTH(1-31) and Mc-PTH(1-34) for which in each pair position 19 was either Glu or Arg (Supplemental Table 1). Stimulation of cells expressing PTH1R-E35K revealed for each peptide pair a small, yet consistent enhancement in the cAMP signaling response induced by the peptide containing arginine at position 19, relative to that containing Glu-19 (Supplemental Fig. 3). Although these differences did not attain statistical significance, they nevertheless support a model by which the E35K mutation confers different effects on the interaction with PTH vs PTHrP due to altered interaction with the divergent residue-19 in the ligand. The mild enhancing effect on initial signaling observed for PTHrP on PTH1R-E35K, as well as the increase in the rate of decay of the response, would thus arise from a charge-based repulsive interaction between the positively charged sidechains of Arg-19 in the ligand and that of Lys-35 in the mutant receptor that facilitates initial activation but leads to a relatively unstable complex. For PTH, the negatively charged sidechain of Glu-19 provides a more complementary charge-based interaction with the Lys-35 sidechain that does lead to such changes in the onset of activation and the rate of decay of the signaling response. However, the precise mechanisms underlying such dynamic processes of ligand induced signaling and deactivation at the PTH1R remain to be elucidated.

Preserved iCa signaling in PTH1R-R485X

It was previously suggested that the delayed ossification seen in Eiken syndrome patients with the PTH1R-R485X variant might arise from an impaired capacity of the mutant receptor to activate the Gαq/phospholipase C (PLC)/inositol triphosphate (IP₃)/intracellular calcium (iCa²⁺) signaling⁷. This hypothesis was based on the moderately delayed bone mineralization observed in mice expressing a PTH1R mutant altered in intracellular loop 2 (E³¹⁷KKY³²⁰– > DSEL) that is specifically impaired for Gαq/PLC/iCa²⁺/IP₃ signaling³⁷. We therefore examined iCa²⁺ signaling for the PTH1R Eiken variants and found that PTH1R-R485X as well as PTH1R-E35K, produced responses to either PTH(1-34) or PTHrP(1-36) that were at least as robust as those produced by PTH1R-WT (Supplemental Fig. 4). The results therefore do not support an impairment in iCa⁺⁺ signaling as a major determinant of the Eiken syndrome phenotype in patients with either of these two mutations. We observed only a weak if any increase in iCa²⁺ signaling for PTH1R-Y134S and PTH1R-H223R. A defect in PLC/IP_3/iCa²⁺ signaling for PTH1R-H223R is consistent with prior findings on this mutant PTH1R that causes JMC³⁸. The defect in iCa²⁺ signaling observed for PTH1R-Y134S could be due in part to a reduced level of surface expression (Fig. 1c and Supplemental Fig. 1c)³⁹. Another potential factor, however, is that on the PTH1R-Y134S mutant there is a faster rate of ligand dissociation, as compared to on PTH1R-WT, which is suggested by the faster rate of decay of the cAMP response observed for the Y134S mutant after ligand washout (Fig. 3c and Supplemental Table 3). It therefore remains possible that alterations in both cAMP and IP₃/iCa²⁺ signaling responses to PTH and/or PTHrP ligands contribute to the clinical phenotype in patients with the PTH1R-Y134S variant.

Radioligand binding properties of PTH1R mutants

We assessed the capacities of the PTH1R variants in intact GS22a cells to bind ¹²⁵I-PTHrP(1-36) as well as the higher affinity control radioligand ¹²⁵I-LA-PTH*⁴⁰. The total binding of ¹²⁵I-PTHrP(1-36) on PTH1R-R485X and PTH1R-H223R was comparable to that on PTH1R-WT but was reduced by ~50% of PTH1R-WT on PTH1R-E35K and PTH1R-Y134S (Fig. 4a). The reductions in binding of ¹²⁵I-PTHrP(1-36) to PTH1R-E35K and PTH1R-Y134S were specific for that radioligand, as total binding of ¹²⁵I-LA-PTH* was similar on each PTH1R variant (Fig. 4b). Competition binding assays performed using ¹²⁵I-LA-PTH* as tracer radioligand and unlabeled PTH(1-34) competitor ligand revealed similar apparent affinities for PTH(1-34) on each PTH1R Eiken variant (Fig. 4c). The apparent affinity of PTH(1-34) on PTH1R-H223R was enhanced ~13-fold versus that on PTH1R-WT, which is consistent with prior studies showing enhanced affinities for PTH agonist ligands on the constitutively active PTH1R-H223R mutant⁴¹. Unlabeled PTHrP(1-36) exhibited little if any capacity to inhibit binding of ¹²⁵I-LA-PTH* to even PTH1R-WT, and so apparent affinities for this peptide were not determined. Overall, these binding data are consistent with the notion that the E35K and Y134S PTH1R mutations selectively reduce overall affinity for PTHrP, which in turn leads to relatively more rapid rates of decay in the cAMP responses after initial complex formation (Fig. 3c, d).

Impaired agonist-induced β-arrestin2 recruitment by PTH1R Eiken mutants

We then evaluated the effects of the mutations on the capacity of the PTH1R to recruit β-arrestin2 and internalize with it to endosomes in response to agonist-stimulation. For this purpose, we transiently transfected the PTH1R variants into HEK293-derived GBR24 cells stably expressing β-arrestin2^YFP 40,42 and then treated the cells with PTH(1-34)^TMR for 30 min before fixing and visualizing the cells by fluorescent microscopy. We observed that for each PTH1R variant PTH(1-34)^TMR staining (red) was localized into clusters, which we interpret as ligand-receptor complexes internalized into endosomes, and the clusters formed with PTH1R-WT, PTH1R-E35K, PTH1R-Y134S and PTH1R-H223R were co-localized with β-arrestin2^YFP (Fig. 5a, b and Supplemental Fig. 5a). In contrast, β-arrestin^YFP did not co-localize in clusters with PTH(1-34)^TMR in cells expressing PTH1R-R485X, but instead remained dispersed in the cytoplasm.

We further evaluated PTH(1-34)-induced recruitment of β-arrestin to the receptors using bioluminescence resonance energy transfer (BRET) biosensors. We thus analyzed the movement of a β-arrestin2 construct tagged with renilla luciferase-II (β-arrestin2^rLucII) to either the plasma-membrane or to endosomes, as detected by a renilla green fluorescent protein (rGFP) containing a CAAX or FYVE anchoring motif, respectively⁴³. As shown in Fig. 5c, PTH(1-34) stimulation of PTH1R-WT resulted in a robust recruitment of β-arrestin2^rLucII to both the plasma-membrane and to endosomes, whereas stimulation of PTH1R-R485X resulted in markedly blunted responses, especially to endosomes (fold-BRET increases from baseline, WT vs R485X, 4.79 ± 0.85 vs. 2.69 ± 0.65, P = 0.10 at the plasma membrane; 6.08 ± 1.21 vs. 1.33 ± 0.13, P = 0.02 at endosomes). These findings are consistent with the C-tail of the PTH1R playing a key role in stabilizing interaction of the receptor with β-arrestins and enabling translocation to endosomes^12,13,14,16. PTH1R-Y134SA exhibited a modest reduction in capacity to recruit β-arrestin2^rLucII to the plasma membrane while recruitment to endosomes was comparable to that of PTH1R-WT (Fig. 5c). The responses of PTH1R-E35K for both the plasma membrane and endosomal reporters were similar to those of PTH1R-WT. These BRET data thus concur with our fluorescent microscopy analyses, as they reveal a marked deficiency in β-arrestin2 recruitment to endosomes for PTH1R-R485X, but not for either PTH1R-Y134S or PTH1R-E35K in response to PTH(1-34).

We also noted upon direct inspection of the time course data obtained in these BRET assays that the plasma membrane recruitment response for PTH1R-R485X tended to exhibit a shallower initial phase (t = 0 to 5′) and hence a slower rise to peak signal, as compared to that of PTH1R-WT, while, as expected, no endosomal recruitment signal was detected for PTH1R-R485X at any time point (Supplemental Fig. 6). These data thus seem consistent with the notion that PTH1R-R485X forms only relatively low affinity complexes with β-arrestin2 that assemble relatively slowly at the plasma membrane and then are not stable enough to transit to endosomes, which is also suggested by a recent study on the PTH1R that utilizes cells lacking G protein receptor kinases (GRKs) that mediate receptor phosphorylation²⁵.

We then assessed by fluorescent microscopy in GBR24 cells the capacity of the PTH1R variants to recruit β-arrestin2^YFP translocation in response to PTHrP. Treatment of PTH1R-WT with PTHrP(1-36)^TMR resulted in robust co-clustering of the ligand in endosomes with β-arrestin2^YFP, while treatment of each PTH1R Eiken mutant resulted in generally weaker staining of the ligand in endosomal clusters and there was no co-clustering with β-arrestin2^YFP (Fig. 6a, b and Supplemental Fig. 5b). A lack of clustering of β-arrestin2^YFP into endosomes with each PTH1R mutant, and a robust clustering with PTH1R-WT was also observed in cells treated with unlabeled PTHrP(1-36) (Supplemental Fig. 7). In contrast, Ile⁵-PTHrP(1-36)^TMR induced a robust co-clustering of β-arrestin2^YFP with the ligand in endosomes with each PTH1R variant except PTH1R-R485X, for which β-arrestin2^YFP again remained dispersed in the cytoplasm (Fig. 6c, d and Supplemental Fig. 7). BRET analyses performed with PTH1R-E35K revealed that Ile⁵-PTHrP(1-36) was nearly 10-fold more potent than PTHrP(1-36) and as potent as PTH(1-34) for inducing recruitment of β-arrestin2r^LucII to both the plasma membrane and to endosomes (Fig. 6e, f), confirming the profound rescue effect of the His⁵–>Ile substitution on the capacity of PTHrP(1-36) to promote stable interaction with β-arrestin2 and the recruitment of the complex to endosomes via the PTH1R-E35K mutant receptor.

These studies overall thus show that the R485X Eiken mutation strongly impairs the capacity of the PTH1R to translocate β-arrestin2 to endosomes in response to either PTH or PTHrP ligands, and that the E35K and Y134S mutations impair the β-arrestin translocation response more selectively to PTHrP than to PTH. Although the mechanisms by which these two mutations in the ECD impact the capacity of the PTH1R to interact with cytoplasmic β-arrestins is not known, studies on other GPCRs have shown that stable interaction with β-arrestin is dependent on a strong interaction with an agonist ligand^16,44, which our data suggest does not occur for the PTH1R-Y134S and -E35K mutants and PTHrP.

Impaired PTHrP-induced receptor desensitization for PTH1R variants of Eiken syndrome

β-arrestins are generally thought to play key roles in mediating short-term receptor desensitization and thus limit responsiveness to persistent as well as repeated ligand exposure⁴⁵. We therefore investigated the capacities of the Eiken PTH1R variants to desensitize cAMP signaling responses induced by PTH(1-34), PTHrP(1-36) as well as LA-PTH during both an initial ligand exposure phase and, after initial ligand washout, during a subsequent ligand rechallenge phase. Transfected GS22a cells were thus pre-stimulated with ligand (1 nM) or with vehicle and cAMP generation was monitored for 14 min (Supplemental Fig. 8a). The cells were then rinsed to remove unbound ligand and cAMP generation was monitored for an additional 90-min washout period (Fig. 7a and Supplemental Fig. 8b). The cells were then treated again with the same test ligand used for initial stimulation or with vehicle, and cAMP generation was monitored for a final 60-min re-challenge period (Fig. 7b and Supplemental Fig. 8b). The level of residual cAMP signaling at the end of the washout period was significantly reduced for PTH(1-34) in cells expressing PTH1R-Y134S, as compared to in cells expressing PTH1R-WT, whereas it was significantly increased for PTHrP(1-36) in cells expressing PTH1R-R485X, and elevated for all PTH1R variants with LA-PTH treatment (Fig. 7a and Supplemental Fig. 8b). The responses observed upon ligand rechallenge, assessed relative to the corresponding responses in vehicle-pre-treated cells, were significantly blunted for each test ligand in cells expressing PTH1R-WT or PTH1R-R485X, while they were blunted only for PTH(1-34) and LA-PTH, but not PTHrP(1-36) in cells expressing PTH1R-E35K or PTH1R-Y134S (Fig. 7b and Supplemental Fig. 8c). These results reveal for each Eiken PTH1R mutant a defect in either signal termination (R485X, Fig. 7a) or signal desensitization after ligand rechallenge (E35K, Y134S, Fig. 7b) that was apparent with PTHrP(1-36) but not PTH(1-34). Of note, these effects mirrored the reductions in the capacities to recruit β-arrestin2 to endosomes, which for PTH1R-Y134S and PTH1R-E35K were specific for PTHrP (Fig. 6a–d). They are also consistent with the notion that PTHrP normally signals only transiently from the plasma membrane and not from endosomes, whereas PTH can signal for more extended times even with internalization to endosomes⁴⁶. We also note, however, that while a blunting of the ligand rechallenge response is consistent with receptor desensitization, some of the ligand-receptor pairs tested in this experiment maintained at the time of ligand rechallenge a considerable level of residual cAMP signaling that was derived from the initial ligand pre-treatment phase. This was especially noticeable with LA-PTH and likely reflects a residual level of ligand occupancy on the receptor, which hinders any direct mechanistic interpretation of the blunting effects on the ligand rechallenge responses.

Inverse agonist reduces basal cAMP signaling by PTH1R-R485X

Finally, we evaluated whether the elevated basal cAMP signaling activity observed for PTH1R-R485X could be suppressed by treatment with [Leu¹¹,dTrp¹²,Trp²³,Tyr³⁶]-PTHrP(7–36)NH2 {dTrp¹²-PTHrP(7–36)}, which functions as an inverse agonist on PTH1R-H223R and other constitutively active mutant PTH1Rs of JMC^34,41,47. Addition of this ligand to Gs22A cells expressing either PTH1R-H223R or PTH1R-R485X, resulted in rapid decreases in the intracellular cAMP levels for each receptor, as well as dose-dependent inhibition of cAMP signaling induced by a subsequent addition of PTH(1-34) agonist peptide (Fig. 8a, b and Supplemental Fig. 9a, b).

In studies aimed at parsing out whether PTH1R-mediated signaling responses were derived from the cell surface or from endosomes, we assessed the capacity of dTrp¹²-PTHrP(7-36), as a membrane impermeable antagonist peptide, to inhibit the signaling response induced by previously applied PTH(1-34)^31,48. We found that addition of dTrp¹²-PTHrP(7-36) (1 μM) to cells 15 min after a prior addition of PTH(1-34) (0.5 nM) resulted in little if any change in the cAMP response induced by the agonist in cells expressing either PTH1R-WT or a mutant derivative, even if the cells were pre-treated with the internalization inhibitor Dyngo4A (Supplemental Fig. 9c). In contrast, addition of the presumably membrane-permeable small molecule PTH1R antagonist, SW106^34,49, significantly reduced the cAMP response mediated by previously added PTH(1-34) in cells transfected with any of the tested PTH1R variants, except PTH1R-H223R, for which the bulk of the cAMP signal is likely derived from constitutive receptor activation (see above) and SW106 is not an inverse agonist on PTH1R-H223R³⁴ (Supplemental Fig. 9d). That PTH(1-34)-induced cAMP signaling mediated by PTH1R-WT and the Eiken mutant receptors can be inhibited by a subsequent challenge with SW106 but not dTrp¹²-PTHrP(7-36) is consistent with the notion that this signaling is derived mainly from endosomes³¹. It also seems possible, however, that the higher affinity, and hence slower off rate with which PTH(1-34) binds to the PTH1R, relative to dTrp¹²-PTHrP(7-36), prevents effective displacement of the agonist from the receptor by the antagonist. Although the two peptides bind to at least partially overlapping sites in the receptor’s orthosteric pocket, such inefficient inhibition could in part reflect the relatively short time frame of the assays, even for complexes that are fixed at the cell surface, as we sought to achieve with Dyngo4a pre-treatment. SW106, on the other hand, likely binds to an allosteric site in the receptor⁴⁹, and so could promote a rapid release of bound PTH(1-34) from the complex, independent of its subcellular location.