Hormone interactions to Leu-rich repeats in the gonadotropin receptors. II. Analysis of Leu-rich repeat 4 of human luteinizing hormone/chorionic gonadotropin receptor.

The luteinizing hormone receptor (LHR) consists of an approximately 350-amino acid-long N-terminal extracellular exodomain and a membrane-associated endodomain of similar size. Human chorionic gonadotropin (hCG) binds to the exodomain, and then hCG/exodomain complex is thought to make a secondary contact with the endodomain and generate hormone signals. The sequence alignment of the exodomain shows imperfectly matching eight to nine Leu-rich repeats (LRRs). In the preceding article (Song, Y., Ji, I., Beauchamp, J., Isaacs, N., and Ji, T. (2001) J. Biol. Chem. 276, 3426-3435), we have shown that LRR2 and LRR4 are crucial for hormone binding. In this work, we have examined the residues of LRR4, in particular Leu(103) and Ile(105) in the putative beta strand. Our data show that Leu(103) and Ile(105) are involved in the specific, hydrophobic interaction of the LRR4 loop, likely to form the hydrophobic core. This loop is crucial for the structural integrity of all of the LRRs. In contrast, the downstream sequence consisting of Asn(107), Thr(108), Gly(109), and Ile(110) of LRR4 is crucial for cAMP induction but not for hormone binding, folding, and surface expression. This implicates, for the first time, its involvement in the interaction with the endodomain and signal generation. The evidence for the interaction is presented in the following article.

The luteinizing hormone/chorionic gonadotropin receptor (LHR) 1 plays crucial roles in development of the gonads in both sexes and the ovulation cycle of the females (1,2). LHR is a member of the glycoprotein hormone receptor subfamily, which includes the follicle-stimulating hormone receptor and the thyroid-stimulating hormone receptor in the G protein-coupled receptor superfamily. It consists of an extracellular N-terminal half (exodomain) and a membrane-associated C-terminal half (endodomain) (3,4). The ϳ350-amino acid-long exodomain is responsible for high affinity hormone binding (5-7) and hormone specificity (8 -10). The resulting hCG-exodomain complex is thought to make a secondary contact with the endodomain, which generates hormone signal (11)(12)(13). Despite the importance of this secondary contact, it has been difficult to find any clues for the secondary contact points. These contact points and residues are likely to be the site of signal generation and play a key role in the signal generation.
The exodomain contains imperfect Leu/Ile-rich repeats (LRRs) of 22-29 amino acids (3, 14 -18), which are a common structural motif found in a large family of proteins, including glycoprotein hormone receptors (19). In the crystals of ribonuclease inhibitors, individual LRRs consist of a ␤ strand connected to a parallel ␣ helices despite divergent sequences. In each ␤ strand, there are two conserved Leu and/or Ile residues in the alternate positions as Leu/Ile-X-Leu/Ile. Therefore, the primary homology among LRRs is the Leu/Ile-X-Leu/Ile motif in ␤ strands. Furthermore, the ␤ strands in ribonuclease inhibitors are involved in the interaction with ribonuclease. However, it has been unknown whether the putative LRRs of the LHR are active at all and what their nature and function are. In the preceding article (20) we have shown that some, but not all, LRRs of the LHR and the FSH-R are crucial for hormone binding. In LHR, LRR2 and LRR4 are crucial.
In this study, the residues around the Leu-Ser-Ile motif in LRR4 of LHR were examined. Our results suggest that the Leu and Ile are involved in the specific and tightly packed hydrophobic interaction in the core of LRR4 loop. In addition, our data implicate, for the first time, the involvement of LRR4 in the interaction of the hCG-exodomain complex with the endodomain. The evidence is presented in the following companion article (21).

EXPERIMENTAL PROCEDURES
Mutant human LHR cDNAs were prepared, expressed in HEK 293 cells, and assayed for hormone binding and intracellular cAMP production as described previously (20,22). All assays were carried out in duplicate and repeated four to six times. Means and standard variations were calculated. FLAG-LHR was prepared by inserting the FLAG epitope (23), Asp-Tyr-Lys-Asp-Asp-Asp-Asp-Lys, between the C terminus (Ser 26 ) of the signal sequence and the N terminus (Arg 27 ) of mature receptors. (20).

RESULTS
To investigate the Leu-Ser-Ile motif of LRR4 in LHR, we decided to examine the extended sequence encompassing the motif and the flanking residues, Arg 99 -Leu-Lys-Tyr-Leu-Ser-Ile-Cys-Asn-Thr-Gly-Ile-Arg-Lys 112 . These residues were individually substituted with Ala, and each of the Ala substitution mutants was stably expressed in HEK 293 cells. They were assayed for 125 I-hCG binding on intact cells or after solubilization in Triton X-100 and for hCG-dependent cAMP induction. As shown by the binding displacement data and Scatchard FIG. 1. Ala scan and characterization of residues around the conserved Leu 103 /Ile 105 motif in LRR4 of LHR. Amino acids from Arg 99 to Lys 112 encompassing the conserved Leu 103 /Ile 105 motif in the putative ␤ strand of LRR4 were individually substituted with Ala. The resulting mutant receptors were stably expressed on human 293 cells. Cells were assayed for 125 I-hCG binding in the presence of increasing concentrations of nonradioactive hCG (A), and Scatchard analysis (B) was plotted against specific binding. Cells were also solubilized in Triton X-100 and assayed for 125 I-hCG binding in the presence of unlabeled hCG (C), and the results were converted to Scatchard plots (D). In addition, intact cells were treated with increasing concentrations of unlabeled hCG and intracellular cAMP was measured (E) as described under "Experimental Procedures." Experiments were repeated four to six times in duplicate. NS stands for not significant. plots in Fig. 1, the cells transfected with the vector carrying the Arg 99 3 Ala mutant bound 125 I-hCG with a K d value of 1,410 pM as intact cells (Fig. 1, A and B) or 1,640 pM after solubilization ( Fig. 1, C and D). These values were higher than the corresponding wild type values of 600 pM and 1,100 pM. In contrast, the cells that were transfected with the vector containing no receptor cDNA or were not transfected at all did not bind the hormone (data not included). These results indicate that the cells transfected with the Arg 99 3 Ala plasmid expressed the mutant receptor on the cell surface as well as in the cells and that the mutant was capable of binding hCG. In addition, the cells produced cAMP in response to hCG in a dose-dependent manner with an EC 50 value of 230 pM and a maximum cAMP level of 72 fmol/1,000 cells. The EC 50 value of cAMP induction, 230 pM, is 39% of the wild type value of 89 pM. Therefore, the higher K d value resulted in the higher EC 50 value. However, the inverse was not observed. For example, a higher binding affinity does not necessarily correlates with lower EC 50 for cAMP induction. Furthermore, maximum cAMP production does not consistently correlates with an EC 50 value. Taken together, the affinities of hormone binding and cAMP induction of the Arg 99 3 Ala mutant are consistently less than one-half of the corresponding wild type affinities. Consequently, the lower affinity and efficacy to induce cAMP is likely due to the lower hormone binding affinity rather than to a defective intrinsic mechanism in cAMP induction. However, this correlation is limited, because the opposite is not true.
In contrast to the hormone-responsive Arg 99 3 Ala mutant, the Leu 100 3 Ala did not show hCG binding to intact cells ( Fig.  1). To test whether the Leu 100 3 Ala mutant was capable of binding hCG but trapped in cells, the cells transfected with the Leu 100 3 Ala mutant vector were solubilized in Triton X-100 and assayed for 125 I-hCG binding ( Fig. 1, C and D). Again, 125 I-hCG binding was not observed. In addition, the cells did not respond to hCG to produce cAMP (Fig. 1E). These results are generally considered as evidence for a receptor defective in hormone binding (18,20,24,25). To demonstrate their expression on the cell surface, the mutant receptor containing the FLAG tag and monoclonal anti-FLAG antibody was used. The antibody recognized the intact cells expressing the FLAG-Leu 100 3 Ala mutant (35 Ϯ 9% of the wild type expression), indicating the surface expression of the FLAG-Leu 100 3 Ala mutant (20).
Next, the Lys 101 3 Ala mutant was examined. 125 I-hCG bound to intact cells expressing the Lys 101 3 Ala mutant with a K d value of 1,640 pM and cells solubilized in Triton X-100 with a K d value of 1,860 pM ( Fig. 1). cAMP was produced in response to hCG with an EC 50 of 214 pM and a maximum cAMP level of 68 fmol/1,000 cells. These results are similar to those of the Arg 99 3 Ala mutant, indicating that the Arg 99 3 Ala and Lys 101 3 Ala mutants behaved the same and the Ala substitution for Arg 99 and Lys 101 had the similar effect. The effect of the Ala substitution for the next residue, Tyr 102 , was more severe, although it did not abolish the activity of the mutant (Fig. 1). For example, the K d values for binding to intact cells and solubilized cells were 4,610 and 7,800 pM, respectively, while the EC 50 for cAMP production was 720 pM. These affinities of the mutant are only 12-14% of the wild type values, although the maximum cAMP level was 120 fmol/1,000 cells, which is nearly 80% of the wild type level (Table I).
The effects of the Ala substitution for the next three residues were similar to or more severe than the effect of the Tyr 102 3 Ala substitution. For example, the Ser 104 3 Ala mutant had K d values of 3,600 and 5,370 pM for hormone binding to intact cells and solubilized cells, respectively ( Fig. 1, F-I). The EC 50 value for cAMP induction was 599 pM (Fig. 1J). The affinities for hormone binding to intact cells and cAMP induction were, therefore, only 15-17% of the wild type affinities (Table I). Leu 103 3 Ala and Ile 105 3 Ala resulted in the complete loss of hormone binding both to intact cells and solubilized cells as well as hCG-dependent cAMP induction (Fig. 1, F-J). These two residues correspond to the conserved Leu/Ile-X-Leu/Ile motif in the ␤ strands of LRRs. Ala substitutions for Cys 106 , Asn 107 , Thr 108 , Gly 109 , Ile 110 , Arg 111 , and Lys 112 did not abolish hormone binding but attenuated the binding affinity (Fig. 1, F-O, and Table I). All of the mutant receptors were also capable of inducing cAMP synthesis. These results indicate that the conserved Leu/Ile-X-Leu/Ile motif, Leu 103 and Ile 105 , is indeed more sensitive to Ala substitution than the flanking residues except Leu 100 . Interestingly, this was independent of the chemical and physical properties of the side chains of the flanking residues.
Since these results are consistent with the LRR hypothesis, we investigated the nature of the Leu 103 3 Ala and Ile 105 3 Ala substitutions by replacing Leu 103 and Ile 105 with a panel of amino acids containing hydrophobic, hydrophilic, neutral, anionic, and cationic side chains. First, Leu 103 was substituted with various hydrophobic amino acids. The substitutions of Val, Phe, and Ile increased the K d values up to 10-fold (Fig. 2,  A-D), thus reducing the affinity to 11-44% of the wild type (Table II), but did not abrogate hormone binding and cAMP induction (Fig. 2E). It is notable that the Leu 103 3 Ile mutant  (Fig. 2, F-H). Since these substitutions introduced an amino acid with a different side chain, it is difficult to tell whether the substitution effects were due to the missing side chain of Leu or the newly introduced side chain. To test these possibilities, Leu 103 was deleted, and the resulting Leu 103 3 deletion mutant did not bind the hormone (Fig. 2,  F-H). Next, Ile 105 was examined after substitutions. The substitutions of Phe, Val, and Leu were tolerable, as the resulting mutants were capable of hormone binding and inducing cAMP (Fig. 3, A-E), although the affinities were reduced to 16 -56% of the wild type affinity (Table II). This is similar to the Phe, Val, and Ile substitutions for Leu 103 . Interestingly, the Val substitution impacted the least for both Leu 103 , whereas the Phe substitution did the least for Ile 105 , suggesting the distinct environment of the two residues. On the other hand, the Trp substitution resulted in the complete loss of the activity for both of Ile 105 and Ile 103 . The substitution for Ile 105 with nonhydrophobic Thr, Asn, Asp, or Lys completely impaired the receptor activity as was the case with Leu 103 (Fig. 3, F-H). To investigate the relationship between Leu 103 and Ile 105 , a reciprocal mutant with Leu 103 3 Ile and Ile 105 3 Leu was generated. The resulting double mutant. Leu 103 3 Ile/Ile 105 3 Leu, poorly bound the hormone with a K d value of 3,340 pM on intact cells and with a K d value of 6,970 pM in solution (Fig. 4). These Leu 103 was substituted with a panel of amino acids with hydrophobic, hydrophilic, neutral, anionic, and cationic side chains. In addition, Leu 103 was deleted in a mutant. The resulting mutant receptors were stably expressed on human 293 cells. Cells were assayed for 125 I-hCG binding and hCG-dependent cAMP induction as described in the legend to Fig. 1. values indicate that the binding affinities are Ͻ20% of the corresponding wild type affinities, consistent with the view of the distinct environment of Leu 103 and Ile 105 .
To determine whether the double mutant receptor was expressed at all, the FLAG-Leu 103 3 Ile/Ile 105 3 Leu receptor was tested. 125 I-Monoclonal anti-FLAG antibody bound to the intact cells as well as solubilized cells that were transfected with the FLAG-Leu 103 3 Ile/Ile 105 3 Leu plasmid (data not included). In contrast, cells that were not transfected or transfected with the LHR with Leu 103 3 Ile/Ile 105 3 Leu plasmid did not bind the antibody. These results clearly indicate expression of FLAG-LHR with Leu 103 3 Ile/Ile 105 3 Leu on the cell surface and in cells, as well as expression of the Leu 103 3 Ile/Ile 105 3 Leu mutant lacking the FLAG epitope. When the Leu 103 3 Ile/Ile 105 3 Leu mutant was reverted to the wild type receptor, the revertant was capable of binding the hormone and inducing cAMP production. Therefore, the inability of the Leu 103 3 Ile/Ile 105 3 Leu mutant to bind the hormone and induce cAMP was due to the double substitutions, not due to unexpected random mutations during the mutagenesis and cloning.

DISCUSSION
The results observed in this work show that the Ala substitutions for Leu 103 and Ile 105 abolished the hormone binding activity of the receptor, whereas the Ala substitution for Tyr 102 and Ser 104 severely impaired the receptor. In contrast to these four crucial and tandem residues, other residues among the sequence of the 14 amino acids, Arg 99 -Leu-Lys-Tyr-Leu-Ser-Ile-Cys-Asn-Thr-Gly-Ile-Arg-Lys 112 , in the putative LRR4 are marginally or less severely impacted by Ala substitution except Leu 100 . Therefore, the sequence of Tyr 102 -Leu 103 -Ser 104 -Ile 105 appears to be crucial for hormone binding and may constitute the core of the putative ␤ strand of LRR4. This configuration would orient Leu 103 and Ile 105 at one side and Tyr 102 and Ser 104 at the other side of the ␤ strand, suggesting a hydrophobic core comprising Leu 103 and Ile 105 and a hydrophilic phase with Tyr 102 and Ser 104 . The hydrophobic core may include Leu 100, as it is equally sensitive to Ala substitution. In fact, the result of multiple substitutions for Leu 103 -Ile 105 indicate such a hydrophobic core, since only hydrophobic residues larger than Ala, but less bulky than Trp, are tolerable at the positions of Leu 103 and Ile 105 . Substitutions with hydrophilic, neutral, cationic, and anionic residues totally impaired the receptor activity. These results suggest that the hydrophobic core is compact and specific.
It is interesting to note that the hormone binding affinity of the reciprocal double mutant, Leu 103 3 Ile/Ile 105 3 Leu was similar to the binding affinity of the single substitution mutant with Leu 103 3 Ile or Ile 105 3 Leu. These results indicate that Leu 103 and Ile 105 in the putative LRR4 ␤ strand could not be switched with each other, consistent with the compact and specific nature of the hydrophobic core. Furthermore, they suggest that the substitutions do not have a synergistic effect. A simple explanation in this particular case is that Leu 103 or Ile 105 may interact with each other as well as other residues. As a result, the interruption of either or both of them would have the similar effect. This hypothesis is consistent with the remarkably similar trends of substitutions for Leu 103 and Ile 105 . For example, the substitutions of Val and Phe were tolerable to both Leu 103 and Ile 105 , whereas the substitutions of Ala, Trp, Thr, Asn, Asp, and Lys abolished the receptor activity. The Val substitution was most tolerable for Leu 103 with a K d value of 1,410 pM, whereas the Phe substitution for Leu 103 is less tolerable with a K d value of 3,730 pM. The trend was opposite for Ile 105 . The Phe substitution was most tolerable with a K d value of 1,110 pM, whereas the Val substitution is less tolerable with a K d value of 3,820 pM. The common and specific nature of their hydrophobic environment is consistent with our model (Fig. 5), showing the direct interaction of Leu 103 with Ile 105 and Leu 100 in a hydrophobic pocket of LRR4.
Our studies have been focused on the two primary targets, the interaction between the exodomain and hormone (initial high affinity binding) and the interaction between the hCGexodomain complex and the endodomain (signal generation). It has been very difficult to find any clues for the contact site(s) of the exodomain for the endodomain. Our data presented in Figs. 1 and 2 as well as Table I are the first to implicate the involvement of LRR4 in contacting the endodomain, in particular, Gly 109 . The reason is as follows: most of Ala substitutions for LRR4 residues reduced the maximum cAMP induction up to 1 and 2 suggested a hydrophobic core of LRR4. Based on these results, LRR4 was modeled to form a hydrophobic core (20), similar to the hydrophobic cores of the Leu-rich repeats found in the crystal structure of ribonuclease inhibitors (19). Once the hydrophobic core is factored in with the side chains of Leu 100 , Leu 103 , Ile 105 , and Ile 110 , the orientation of the side chains of other residues are determined by energy minimization. Yellow indicates an natural residue, red indicates Ala substitutions with a maximum effect on hormone binding (K d of hormone binding to cells) and/or expression, and blue indicates a minimum effect on hormone binding.

TABLE III
Comparison of the LRR4 sequence The sequences encompassing the Leu-X-Ile in the ␤ strand of LRR4 of glycoprotein hormone receptors were aligned. Absolutely conserved residues are marked with "*". PD*LKE**L *KF*G*F*** LGV******* 41% of the wild type value. The only exception is Gly 109 3 Ala substitution, which nearly abolished cAMP induction to 14% of the wild type value. This trend is more obvious when the (maximum cAMP mut/wt) /(K d wt/mut , where mut indicates mutant and wt indicates wild type) ratios are compared among the Ala mutants (Table I). The ratios are more than 1.00 for all except Gly 109 3 Ala and Asn 107 3 Ala. The ratios for the two mutants are 0.52 and 0.71, and such exceptionally low ratios are not found with any LRR2 Ala mutants (20). The sequence alignment reveals the striking conservation of the amino acids from Leu 97 -Pro-Gly-Leu-Lys-Tyr-Leu-Ser-Ile-Cys-Asn-Thr-Gly 109 ( Table III). Out of the 13 amino acids, 8 are absolutely identical among LHR, follicle-stimulating hormone receptor, and thyroid-stimulating hormone receptor of all species. Interestingly, Asn 107 -Thr-Gly 109 are uniquely in tandem and their maximum cAMP mut/wt )/(K d wt/mut ) ratios are 0.71, 1.08, and 0.52, respectively. These residues are not essential for folding and surface expression, since the Ala mutants were successfully expressed on the cell surface and capable of binding the hormone. However, Ala substitution of them notably impaired cAMP induction as obvious by their significantly high EC 50 for cAMP and/or low maximum cAMP level (Table I), suggesting an intriguing possibility of their interaction with the endodomain.
In conclusion, the Leu 103 -Ile 105 sequence is crucial for the specific interaction to form the hydrophobic core of LRR4. In addition, the downstream sequence consisting of Asn 107 , Thr 108 , and Gly 109 is crucial for cAMP induction but not for hormone binding, folding, and surface expression. Therefore, they are likely involved in the interaction with the endodomain. This is the first evidence suggesting an endodomain contact point in the exodomain. Then, the inevitable question is whether LRR4 interacts with hCG at all. In the following article (21) we show evidence that LRR4 does interact with hCG and, furthermore, is involved in the interaction of hCGexodomain complex with the endodomain.