Two defective heterozygous luteinizing hormone receptors can rescue hormone action.

Luteinizing hormone receptor is a G protein-coupled receptor and consists of two halves: the N-terminal extracellular half (exodomain) and C-terminal membrane-associated half (endodomain). Hormone binds to the exodomain, and the resulting hormone-exodomain complex modulates the endodomain to generate signals. There are mutations that impair either hormone binding or signal generation. We report that the coexpression of a binding defective mutant and a signal-defective mutant rescues signal generation to produce cAMP. This rescue requires both types of mutant receptors and is dependent on the human chorionic gonadotropin dose, the surface concentration of mutant receptors, and the amino acid position of mutations. Furthermore, random collisions among mutant receptors are not involved in the rescue. Our observations provide new insights into the mechanisms of the functional and structural relationship of the exo- and endodomain, signal transduction, and receptor genetics, in particular for defective heterozygotes.

The luteinizing hormone receptor (LHR) 1 plays a crucial role in the development of the gonads in both sexes and ovulation in females. Defective mutations of the receptor often cause infertility (1). Gain of function mutations are generally dominant, whereas loss of function mutations are recessive. The genetic prediction of mutations is not straightforward, because the effects of some mutations are partial and some patients are defective heterozygotes. For example, there are patients with two defective heterozygous LHR mutations (2,3) and the precise relationship of two mutant receptors in a patient is unclear. This is particularly relevant for LHR, which has two distinct domains, one for hormone binding and the other for signal generation (4 -6). We raise the question of whether the exodomain of an LHR can modulate the endodomain of another LHR. As a first step toward understanding this novel and intriguing question, we have investigated the relationship of two different LHR mutants, one with defective hormone binding and the other with normal hormone binding but defective signal generation.
LHR belongs to the structurally unique glycoprotein hormone receptor subfamily of the G protein-coupled receptor family (5). Unlike other receptor subfamilies, they comprise two equal halves, an extracellular N-terminal half (exodomain) and a membrane-associated C-terminal half (endodomain) (7)(8)(9)(10)(11). The exodomain is ϳ350 amino acids long, and it alone is capable of high affinity hormone binding (12)(13)(14)(15) with hormone selectivity (16 -18) but without hormone action (14,19,20). Hormone signal is generated in the ϳ350-amino acid-long endodomain (4), which is structurally equivalent to the entire molecule of many other G protein-coupled receptors such as rhodopsin and adrenergic receptors (5). Growing evidence suggests that glycoprotein hormones initially bind to the exodomain (5) and that the resulting hormone-exodomain complex undergoes a conformational change (21) and modulates the endodomain. This secondary interaction is thought to generate a signal in the endodomain (4 -6, 22). These findings are consistent with the observations that signal generation is generally impacted by endodomain mutants (23), whereas mutations in the exodomain tend to affect hormone binding (24 -26).
Considering the existence of heterozygous mutant LHRs in a patient (2,3), we wondered about the relationship between the two alleles as to whether they would be dependent on or independent of each other. Particularly, there is the intriguing possibility that two heterozygous mutants, one defective in hormone binding and the other with normal hormone binding but defective signal generation, might interact with each other to rescue hormone action. Obviously, this would require the novel intermolecular interaction of the exodomain of one LHR with the endodomain of another LHR. Although it has never been described, it would have significant impact on the interpretation of receptor genetics and provide new insights into clinical treatments. To test the hypothesis, we co-expressed various pairs of heterozygous defective LHRs and tested for their functional rescue.

EXPERIMENTAL PROCEDURES
Mutagenesis and Functional Expression of Receptors-Mutant rat LHR and FSHR cDNAs were prepared in a pSELECT vector using the non-polymerase chain reaction-based Altered Sites Mutagenesis System (Promega), sequenced, and subcloned into pcDNA3 (Invitrogen) as described previously (27). After subcloning pcDNA3, the mutant cDNAs were sequenced again. Varying concentrations of plasmids were transfected into human embryonic kidney (HEK) 293 cells by the calcium phosphate method (28). Transiently transfected cells were assayed 60 -72 h after transfection. Stable cell lines were established in minimum essential medium containing 8% horse serum and 500 g/ml G-418. All assays were carried out in duplicate and repeated three to four times. Means and standard deviations were calculated. 125 I-hCG Binding and Intracellular cAMP Assay-hCG and human follicle-stimulating hormone, provided by the National Hormone and Pituitary Program, were radioiodinated as described previously (29). Cells were assayed for 125 I-hormone (150,000 cpm) binding in the presence of increasing concentrations of nonradioactive hormone. K d values were determined by Scatchard plots. For intracellular cAMP assay, cells were washed twice with minimum essential medium and incubated in the medium containing isobutylmethylxanthine (0.1 mg/ml) for 15 min. Increasing concentrations of hCG were then added, and the incubation was continued for 45 min at 37°C. After the medium was removed, the cells were rinsed once with fresh medium without isobutylmethylxanthine, lysed in 70% ethanol, freeze-thawed in liquid nitrogen, and scraped. After pelleting cell debris at 16,000 ϫ g for 10 min at 4°C, the supernatant was collected, dried under vacuum, and resuspended in 10 l of the cAMP assay buffer provided by the manufacturer. cAMP concentrations were determined with a 125 I-cAMP assay kit (Amersham Biosciences) following the manufacturer's instruction and validated for use in our laboratory.
Radioimmunoassay for Flag-LHR-Flag-LHR was prepared by inserting the Flag epitope, Asp-Tyr-Lys-Asp-Asp-Asp-Asp-Lys (5-GAC TAC AAG GAC GAT GAC GAT AAG-3), between the C terminus of the signal sequence and the N terminus of mature receptors. Mouse anti-Flag monoclonal M2 antibody (Sigma) was iodinated with 125 I according to the published procedure for radioiodination of hCG (29), and 125 Ianti-Flag antibodies were purified on a Sephadex G-150 column. Binding of 125 I-anti-Flag (150,000 cpm) to HEK 293 cells expressing Flag-LH receptors was carried out in the presence of increasing concentrations of nonradioactive anti-Flag antibody in minimum essential medium containing 3 mg/ml of bovine serum albumin for 8 -10 h at 4°C.

RESULTS
To investigate the interaction of heterozygous mutant LHRs, we chose the K583R mutant (LHR K583R ) in which Lys 583 was substituted with Arg. This mutant receptor is normally processed and targeted to the cell surface and is capable of binding hCG but incapable of inducing cAMP production (30). The mutant receptor is referred to as LHR ϩhCG/ϪcAMP . In addition to the LHR ϩhCG/ϪcAMP , other mutant LHRs were selected that were expressed on the cell surface but were incapable of binding hCG (LHR ϪhCG ). They are L20A, C22A, P479A, and P479G mutants (31,32). HEK 293 cells transiently transfected with the LHR K583R plasmid showed hCG binding with the wild type affinity but did not produce cAMP in response to increasing doses of hCG ( Fig. 1). Cells transiently transfected with the plasmid for LHR L20A , LHR C22A LHR P479A , or LHR P479G did not show hCG binding or cAMP induction, consistent with previous reports (31,32). The cells transfected with the blank plasmid, pcDNA3, failed to bind hCG and produce cAMP, indicating that the vector itself was not involved in hCG binding.
Co-expression of Two Mutant LHRs and Successful Rescue of cAMP Production-Next, cells were cotransfected with a pair of LHR ϩhCG/ϪcAMP and LHR ϪhCG , for example, K583R and L20A mutants or K583R and C22A mutants. The cells that were cotransfected with either LHR K583R and LHR L20A or LHR K583R and LHR C22A were capable of binding hCG, and the K d values were similar to the wild type value (Fig. 2, A and B). In addition, these cells induced cAMP production in an hCG dosedependent manner (Fig. 2C). The maximal levels of cAMP were approximately one-third of the wild type value, and their EC 50 values were 15-25-fold higher than the wild type value (Fig. 2,  table), suggesting the rescue of cAMP induction with lower potency.
The data shown in Fig. 2 suggest that the nonbinding receptors were expressed on the cell surface, as rigorously demonstrated by different methods in previous reports (31,32). However, to validate the surface expression, cells were transfected with the mutant receptors that carry the Flag epitope at the N terminus (Flag-LHR C22A ) and assayed for binding of anti-Flag monoclonal antibody as described previously (32,33). The cells were incubated with 125 I-anti-Flag monoclonal antibody with increasing concentrations of nonradioactive antibody. The cells showed specific binding of the 125 I-antibody, which was gradually displaced by nonradioactive antibody, as did the cells transfected with Flag-LHR wt (Table I). However, there was no specific antibody binding to the cells transfected with the LHR WT , LHR C22A , or LHR L20A plasmid as previously reported (32). These results show that the Flag-LHRs were indeed expressed on the cell surface in these experiments. To determine whether Flag-LHR C22A can rescue cAMP induction, it was coexpressed with LHR K583R . As expected, the cells bound hCG and produced cAMP (Fig. 3).
Specificity for the Rescue of cAMP Induction-It is unclear whether cAMP was induced by accidental collisions between the endodomains of two different mutant receptors. To test this possibility, several pairs of two different LHR ϪhCG mutants were co-expressed. As shown in Fig. 4, none of the co-expressed pairs (LHR C22A and LHR L20A , LHR C22A and LHR P479A , LHR C22A and LHR P479G , LHR L20A and LHR P479A , and LHR L20A and LHR P479G ) was capable of inducing cAMP or binding hCG. These results show that one of the mutant pairs has to be capable of binding hCG to rescue cAMP induction.
To test the dependence of the rescue on hCG binding, cells were co-transfected with varying concentrations (6, 12, and 18 g) of the LHR K583R plasmid and a constant amount (6 g) of the LHR L20A or LHR C22A plasmid. The cells were assayed first for hCG binding to determine the relationship of the surface concentration of LHR K583R with the plasmid concentration (Fig. 5). The results show that the surface concentration of LHR K583R increased in parallel to the plasmid concentration used for transfection. The range of the LHR K583R concentration was 5,000 -21,000 receptors/cell, which compares favorably with the in vivo LHR concentration on porcine granulosa cells, several thousand per cell. 2 In addition, the variation in the receptor concentration does not appear to impact the hormone binding affinity. However, the maximum cAMP levels show an interesting trend. When 6 g of the LHR L20A plasmid was cotransfected with 6, 12, or 18 g of the LHR K583R plasmid, the maximum cAMP levels were 52.3, 71.1, and 36.5 fmol/1000 cells, respectively. The differences among the three values are statistically significant with p values of Ͻ0.05 to Ͻ0.001. Therefore, the maximum cAMP level increased by 36% at 12 g and then decreased by 30% at 18 g as compared with the cAMP level at 6 g of the plasmid. The result was similar when 6 g of the LHR C22A plasmid was co-transfected with 6, 12, or 18 g of the LHR K583R plasmid. These observations suggest that the cAMP rescue requires LHR K583R and is dependent on the concentration of this mutant receptor.
One may question whether the surface expression levels of the LHR ϪhCG mutants shown in Fig. 5 were constant, although 6 g of the plasmids was used for transfection of the cells throughout the experiment. To address this problem, we took another approach to keep the expression level of LHR ϪhCG mutants constant. Cell lines were established after stably transfecting them with the LHR L20A plasmid or LHR C22A plasmid. These cell lines were transfected again with varying concentrations (6, 12, and 18 g) of the LHR K583R plasmid. The doubly transfected cells showed increasing concentrations of LHR K583R (Fig. 6). Transfection with 12 g of the LHR K583R plasmid increased the maximum cAMP level by 29 -58% over that of the cells transfected with 6 g of the plasmid. Transfection with 18 g of the plasmid resulted in a 3-fold increase in the EC 50 value for the cAMP rescue, although the maximum cAMP levels remained high. The observations described in Figs. 5 and 6 indicate that the cAMP rescue is dependent on the LHR K583R concentration. However, there is a notable difference in the results of Figs. 5 and 6. In Fig. 5 the maximum levels of cAMP peaked as the LHR K583R concentration increased, whereas it plateaued in Fig. 6. The difference in the two experiments was LHR ϪhCG , which was transiently expressed in the Fig. 5 experiment and stably expressed in the Fig. 6 experiment. A molecule is expressed in stable cell lines generally more than in transiently expressing cells because of the associated antibiotic selection. Therefore, another experiment was performed using the stable cell line expressing Flag-LHR C22A , which appears to express less than 12,800 receptors/cell. It was transiently transfected with increasing concentrations of the LHR K583R plasmid from 3 to 18 g. The cells produced cAMP in response to hCG, and the maximum cAMP levels peaked (Fig. 3). These results taken together with the data shown in Figs. 5 and 6 show that there are optimal concentrations of LHR K583R to pair with LHR ϪhCG and rescue cAMP induction. They indicate the importance of the number of the hCG binding receptor and/or the ratio of the hCG binding receptor to the nonbinding receptor.
The nonbinding receptors tested so far have mutations in the exodomain that impair hormone binding. In addition to these nonbinding receptors with a defective exodomain, there are nonbinding receptors that have a normal exodomain but mutation in the endodomain, such as P479A and P479G of the transmembrane helix 4 (31). These mutations in the endodomain block hCG binding to the exodomain by constraining the exodomain although the exodomain itself is intact (34,35). To test whether these mutants could pair with LHR K583R and induce cAMP production, LHR K583R was co-expressed with LHR P479A or LHR P479A . The cells co-expressing LHR K583R and LHR P479A or LHR K583R and LHR P479A failed to induce cAMP production although they were capable of binding hCG (Fig. 7). These results indicate that not all of the mutant pairs of LHR ϪhCG and LHR ϩhCG/ϪcAMP are capable of rescuing the hCG dependent cAMP induction, suggesting a specificity for pairing. Furthermore, these results suggest that LHR ϪhCG with a mutation in the exodomain, but not in the endodomain, could be rescued.
In addition, we tested the affect of another receptor species on the activity of wild type LHR. When LHR wt was co-expressed with FSHR wt , the functional FSHR did not impact the hCG binding affinity or the EC 50 value and maximum level of cAMP induction by LHR wt (data not shown). These results show that the cAMP induction by LHR ϪhCG and LHR ϩhCG/ϪcAMP was not rescued by accidental collisions between them or with different hormone receptor species.  Our observations described in this work show that cells co-expressing a pair of two differently defective mutants, one defective in hCG binding at the exodomain (LHR ϪhCG ) and the other defective in signal generation at the endodomain (LHR ϩhCG/ϪcAMP ), can induce cAMP production. This successful rescue of cAMP induction requires both types of mutant receptors. However, not all LHR ϪhCG were capable of pairing with LHR K583R , an LHR ϩhCG/ϪcAMP , and rescuing cAMP induction. Rescue is observed when hormone binding of an LHR ϪhCG is impaired by a mutation in the exodomain but not by mutations in the endodomain. These results suggest specificity for the rescue of cAMP induction. For example, the rescue is dependent on hCG dose, the surface concentration of the mutant receptors, and the amino acid positions of the mutations. Furthermore, random collisions among mutant receptors are not involved in the rescue.
It is known that LHR binds hCG first at the exodomain (12)(13)(14), and the resulting hCG-exodomain complex undergoes conformational changes (21,36,37) and modulates the endodo- main (38,39). This secondary interaction is responsible for signal generation and receptor activation (4 -6). Based on these observations and the results described in this work, the cooperation between the two types of mutant LHRs includes the exodomain of LHR K583R and the endodomain of LHR L20A or LHR C22A . Furthermore, the two domains most likely interact with each other. Therefore, our results suggest an intermolecular interaction between the exodomain of one receptor and the endodomain of another receptor and implicate at least partial substitution of the hCG-functional exodomain complex of a receptor for the defective exodomain of another receptor. This is supported by several pieces of evidence. The rescue is observed when hormone binding of an LHR ϪhCG is impaired by a mutation in the exodomain but not by mutations in the endodo-main. The exodomain and endodomain are dependent on each other before (34,35) and after (38,39) hormone binding. The exodomain modulates signal generation using a suppressor in the hinge region (38,39) and an activator in Leu-rich repeat 4 (40). On the other hand, the endodomain constrains hormone binding at the exodomain through exoloops and transmembrane helices (31,34,35). The interaction between the exodomain and endodomain involves exoloop 2 of the endodomain and the hinge region of the exodomain (22,39). In addition, other exoloops are likely to be involved (35).
The intermolecular exodomain-endodomain interaction is also consistent with the dependence of the rescue on receptor concentrations and the existence of optimal concentrations. The observation that too few or too many LHR ϩhCG/ϪcAMP can FIG. 6.
Coexpression of stable LHR ؊hCG and transient LHR ؉hCG/؊cAMP . Cells stably expressing LHR ϪhCG were transiently transfected with increasing concentrations of the LHR K583R plasmid. The cells were assayed for hormone binding and cAMP production as described in the legend for Fig. 1.   FIG. 7. cAMP rescue is dependent on the location of mutation in LHR ؊hCG . Cells were transiently coexpressed with LHR K583R and LHR P479A or LHR K583R and LHR P479G . The cells were assayed for hormone binding and cAMP production as described in the legend for Fig. 1. interfere with the collaboration between LHR ϩhCG/ϪcAMP and LHR ϪhCG is of interest and reminiscent of the antibody and antigen interaction. One wonders whether too many LHR ϩhCG/ϪcAMP might nonproductively compete for a limited number of LHR ϪhCG , which could lead to less effective induction of cAMP. It also suggests the intriguing possibility of pleiotropic activation of other LHRs by a liganded LHR, in addition to intramolecular activation of its own endodomain. The intermolecular exodomain-endodomain interaction would allow a heterozygote consisting of LHR ϩhCG/ϪcAMP and LHR ϪhCG to rescue the LHR activity.
The rescue observed in this study differs from experiments performed by the Hsueh group (20). They have elegantly demonstrated the rescue of two defective mutant LHRs: one with the exodomain connected to the transmembrane domain 1 and lacking the rest of the transmembrane helices; and the other possessing the first five transmembrane helices without an exodomain (20). The first transmembrane helix played a crucial role in the cAMP rescue, apparently by its interaction with the five transmembrane helices. Perhaps because of the different mechanisms of rescue in the previous report (20) and this study, the maximum cAMP levels rescued differ considerably; it was 12% of the wild type value in the previous study and 50% in this study. This is also substantially (ϳ9-fold) higher than the normal basal cAMP level. This is a significant difference when compared with the maximum cAMP level of activating mutant LHRs, which is 5-8-fold greater than the normal cAMP basal level (41).