Identification of a Follicle-stimulating Hormone Receptor-binding Region in hFSH-P-(81-95) Using Synthetic Peptides*

Pituitary and placental glycoprotein hormones are heterodimers with alpha-subunits of identical primary structure, but dissimilar beta-subunits. Regions of structural similarity between the beta-subunits might be involved in interaction with the homologous alpha-subunits, and regions of structural dissimilarity could, therefore, be candidates for receptor interactions. A restrained matrix dot-plot analysis identified hFSH-beta-(8-32) and hFSH-beta-(55-65) as candidates for interaction with alpha-subunit. Therefore, by subtraction, hFSH-beta-(33-54) and hFSH-beta-(66-111) seemed candidates for regions of interaction with receptor. In a previous report we demonstrated that hFSH-beta-(33-53) represented a receptor-binding region of hFSH-beta. Analysis of structural parameters (flexibility, surface probability, secondary structure prediction, etc.) indicates similarities between hFSH-beta-(33-53) and hFSH-beta-(85-95), suggesting the latter might be the component of hFSH-beta-(61-111) interacting with the receptor. Testing of 11 synthetic peptides, corresponding to the primary structure of hFSH-beta, demonstrated that hFSH-beta-(31-45)-peptide amide, were unique in ability to inhibit 125I-follicle-stimulating hormone binding to receptor. hFSH-beta-(81-95)-peptide amide also stimulated estradiol biosynthesis in Sertoli cell cultures. The correlation between binding inhibition and surface probability, flexibility, and predicted secondary structure (alpha, extended, and turn) was highly significant (R2 = 0.87, p less than 0.0001). Regression significance for these parameters, taken individually, were very poor. Receptor-binding regions, therefore, appear to be characterized by a particular and complex arrangement of secondary structure motifs, surface probability, and flexibility.

subunit should have similar structural characteristics. Therefore conserved P-domains might contain regions of subunitsubunit interaction. By the same reasoning, nonhomologous P-subunit regions might be expected to represent receptorbinding regions of the hormone (Pierce et al., 1971;Steward and Steward, 1977;Ward, 1978;Pierce and Parsons, 1981). By analysis of various structural characteristics of glycoprotein hormones P-subunits, we deduced that the FSH-P-subunit may have two receptor-binding regions around amino acids 33-54 and 85-95. We have previously shown that hFSH-P-(34-37)-peptide amide (KTCT) and hFSH-/3-(49-52)-peptide amide (TRDL) inhibit binding of lZ51-hFSH to FSH receptor in calf testis membranes , Schneyer et al., 1988 and that hFSH+(33-53)-peptide amide, which encompasses these tetrapeptides, inhibits binding with increased potency (Andersen et al., 1987). hFSH-P-(33-53)peptide amide binds to FSH receptor, behaving as a partial agonist of FSH with regard to stimulation of basal estradiol production and as a partial antagonist of FSH-stimulated estradiol biosynthesis (Santa Coloma et al., 1990). In order to examine the hypothesis of a second receptor-binding region around hFSH-P-(85-95), to better delineate the already identified receptor-binding region (hFSH+(33-53)) and to search for possible additional binding regions, we examined 11 overlapping peptides corresponding to the primary structure of the FSH-P-subunit. Our results support the notion that an additional receptor-binding region is located in the hFSH-@-(81-95) sequence.  (Bidlingmeyer et al., 1984).

Preparation of Radiolabeled
FSH-Radiolabeled hFSH was prepared and purified by polyacrylamide gel electrophoresis as described previously (Schneyer et al., 1986). This method involves lactoperoxidase-catalyzed iodination of highly purified human FSH. The resulting "'1-hFSH had a specific activity of 20-25 &i/rg and a minimum of 35% bindability to excess receptors.

Region
Radioligand Receptor Assay (RRA)-Calf testis membranes containing FSH receptors were prepared as described elsewhere (Dattatreyamurty et al., 1987). The RRA was performed as previously described (Schneyer et al., 1986) with the following modifications: 3 mg wet weight of membrane preparation, 2.5 ng of radioligand, and samples or buffer (50 mM HEPES, 250 mM sucrose, 5 mM MgC12, adjusted to pH 6.5) to a total volume of 0.25 ml, were incubated for 18 hat 20 "C. After incubation, hormone bound to membrane receptor was separated from unbound hormone by centrifugation (30,000 x g, 15 min). Nonspecific binding was defined as radioligand bound in the presence of peptide plus a 300-fold excess of unlabeled hFSH.
In Vitro Bioassay for FSH Actiuity-Assessment of FSH agonist or antagonist activity was based on ability of peptides to affect either basal or hormone-stimulated conversion of androstenedione to estradiol by cultures of Sertoli cells taken from immature rat testis (Grass0 and Reichert, 1989

Assessment of Receptor-binding Domains by Computer
Analysis-As outlined under the Introduction and considered further under "Discussion," regions of homology among @subunits of glycoprotein hormones may be involved in subunit interaction and regions of low sequence homology might contain receptor-binding regions. Analysis of aligned hFSH and hGC p-subunit sequences indicated several regions of sequence homology (Fig. 1). Using a constrained dot-plot approach a simpler profile was obtained, suggesting two possible a-subunit-binding regions ( Fig. 2) located between amino acids 8-32 and 55-65. There also appeared to be two likely receptor-binding regions on the hFSH$-subunit. The first theoretical binding region, suggested by the dot-plot analysis (Fig, 2), was located between amino acids 33 and 54. In a separate study, based on a different approach (see "Discussion"), we have already reported that the sequence 33-53 contains a receptor-binding region for FSH (Andersen et al., 1987;Santa Coloma et al., 1990). The location of the second theoretical receptor-binding region on hFSH-/I was not well defined by the alignment of sequences or the dot-plot analysis. A broad region with receptor-binding potential was suggested by dot-plot analysis to be between amino acids 66 and 111.
To obtain evidence for more precise location of the second binding region, several structural and physicochemical char- acteristics of hFSH+subunit were examined. Prediction of secondary structure (Chou andFasman, 1978, Garnier et al., 1978), as well as hydrophilicity (Kyte and Doolittle, 1982), surface probability (Emini et al., 1985), flexibility (Karplus and Schulz, 1985), and antigenic index (Jameson and Wolf, 1988), were determined (Fig. 3). We noted that hFSH-P-(33-53), shown previously to bind to receptor (Santa Coloma et al., 1990), possessed high values for hydrophilicity, surface probability, flexibility, and antigenicity. Assuming that general characteristics which allow P-subunit interaction with the FSH receptor were similar among binding regions and therefore similar to that of hFSH-P-(33-53), it appeared that the second receptor-binding region may be located in a relative smaller segment of the 61-111 region, between amino acids 85 and 95. In order to test the validity of our theoretical approach to more precisely localize the receptor-binding region in hFSH-P-(33-53) and to search for possible additional binding regions, we examined several overlapping synthetic peptides corresponding to the primary structure of the FSH-@ subunit (Fig. 4) on the basis of their ability to inhibit '251-hFSH binding to receptor.

Inhibition
of lz51-FSH Binding to Calf Testis Membranes-Peptides corresponding to sequences 31-45 and 81-95 were found to possess maximum Y-hFSH inhibitory activity (Fig.  5). FSH-P-(31-45) is a segment of the sequence 33-53 already implicated in FSH receptor binding (Santa Coloma et al., 1990). The location of the second binding region was in sequence FSH-P-(81-95), as predicted on the basis of our structural analysis (sequence 85-95). It is noteworthy that  Coloma et al., 1990) and hFSH-P-(81-95) were the more potent inhibitors.
The results suggest the presence of a second binding region in hFSH-P-(81-95).
An interesting observation, derived from the matrix of correlation between variables (Table I), indicates that hydrophilicity (H), surface probability (S), or flexibility (F) are highly correlated. A high correlation between AI and hydrophilicity, Ps, or flexibility (F) was also noted, although not surprising since the function to determine AI includes these parameters (Jameson and Wolf, 1988). For this reason, AI is not likely to contribute additional information to the regression equation. Furthermore, the contribution of AI to the correlation coefficient is very low (R* = 0.28). We therefore eliminated the antigenic index (AI) from calculation of regression coefficients. After elimination of AI, a low significance for hydrophilicity (p > 0.1) became evident, implying that hydrophilicity (hydrophobicity), by itself, is not an appropriate parameter to characterize binding regions between hFSH$-subunit and the FSH receptor. The statistics obtained for the "best subset" are shown in Table II. The multiple correlation coefficient was excellent (R2 = 0.87) and highly significant (p < O.OOOl), suggesting that the Garnier's parameters, and to a lesser extent flexibility and surface probability, are important parameters to describe (or predict) binding regions. The individual contribution to R2 reflects the relative importance of each parameter (Table II). By using the regression coefficients it was possible to define a linear function, referred to as "binding index," BI = 526.8 + 2.178 (S) -381.6 (F) + 40.8 (H) -48.6 (E) -40.9 (T). The predicted values of this function, together with experimental values, are represented in Fig. 6. These parameters seem to have questionable significance when considered individually (Table I), but taken together, appear to contain sufficient information to describe or predict receptor-binding regions with some reliability. Therefore, we calculated the predicted values for peptides corresponding to the N-terminal region,  The contents of a-helix (low) and p-extended (high) or turn (high) structure seem to be more important parameters to describe or predict binding sites in hFSH-P than surface probability or flexibility.
which were not included in the calculation of the regression coefficients.
exist around the sites of carbohydrate attachment in the fully glycosylated molecule, which could not be expressed in the inhibition assay due to the low solubility of the synthetic peptides corresponding to this region or to the absence of carbohydrates.
The latter possibility seems less probable because binding activity is only slightly affected by carbohydrate removal (Matzuk et al., 1989).
Effect of hFSH-P-(81-95)-peptide Amide on Estradiol Synthesis in Cultured Rat Sertoli Cells-We also examined the effect of hFSH-P-(81-95)-peptide amide on the conversion of androstenedione to estradiol in cultures of Sertoli cells taken from testis of immature rats. We studied the effect of this peptide on basal secretion of estradiol and on FSH-stimulated estradiol biosynthesis.
A synergistic effect on FSH- at low peptide concentrations (5 and 10 pM). Over 20 pM, a reduction in the effect occurs, probably reflecting the partial agonist activity of hFSH-P-(81-95)-peptide amide at high doses. This synergistic effect was not seen with hFSH-P-(31-45) (not shown) or with hFSH-P-(33-53) (Santa Coloma et al., 1990). Together with the stimulation of estradiol biosynthesis over basal levels, these data establish further evidence that hFSH+-(81-95) is involved in the interaction with the FSH receptor.

DISCUSSION
Ward and Moore (1979) postulated that the hCG p-subunit region 93-100 may be the determinant of hormone specificity among the glycoprotein hormones and designated this region the determinant loop. This determinant loop concept has proved useful in attempts to understand details of the interaction of the glycoprotein hormones with their receptors. Recently, Keutmann et al. (1989) reported that a synthetic peptide corresponding to hCG-P-(93-100) inhibited hCG binding to receptor, but did not stimulate testosterone production in cultured rat Leydig cells. Keutmann et al. (1987) also found that synthetic peptides corresponding to hCG-p-(38-57), comparable to hFSH-P-(32-51), inhibited hCG binding to receptor, and stimulated testosterone production in Leydig cell cultures. In an earlier study , we reported inhibition of lZ51-hFSH binding to testicular FSH receptors by synthetic tetrapeptides corresponding to amino acids 34-37 (TRDL) and 49-52 (KTCT) of the P-subunit of human follitropin (FSH). A subsequent report (Schneyer et al., 1988) concluded that the interaction of FSH with its receptor may involve multiple, discrete binding sites, which included hFSH-P-(34-37) (TRDL). In a preliminary report (Andersen et al., 1987), we noted that an extended synthetic peptide encompassing TRDL and KTCT, hFSH-P-(33-53)peptide amide, also inhibited 'Y-hFSH binding to receptor, at a potency significantly greater than seen with either individual tetrapeptide.
In a further study we obtained evidence that hFSH+?-(33-53)-peptide amide binds to the FSH receptor in a functional manner, affecting both basal and FSHstimulated steroidogenesis (Santa Coloma et al., 1990). Using sequence alignment and a "dot-plot" approach to localize nonhomologous regions between hFSH-P and hCG-p, and considering some parameters related to the hFSH-P-(33-53)-binding region (surface probability, hydrophilicity, etc.), we developed the hypothesis that a second binding region should be located around FSH-P-(85-85).
This was the approximate region that Ward and Moore (1979) designated as the determinant loop (hFSH-P-(87-94)). Using 11 overlapping synthetic peptides corresponding to hFSH+subunit, we obtained experimental evidence suggesting that hFSH-P-(81-95)-peptide amide, which includes the predicted 85-95 region and the determinant loop, represent a second receptor-binding region. Due to solubility problems encountered with peptides corresponding to the N-terminal segment of hFSH+-subunit in the RRA, we do not exclude the possibility that an additional binding site could exist in the N-terminal region, as postulated by Stewart and Stewart (1977) for leuteinizing hormone. No evidence for interaction between the C-terminal region (amino acids 101-111) of hFSH-@-subunit and the FSH receptor were founded.
Multiple linear regression analysis of the FSH-binding inhibition data identified parameters appropriate to describe these receptor-binding regions. Hydrophilicity was a poor estimator even when it was considered together with a number of other structural parameters.
The suggestion that hydrophilicity is not adequate to describe (or predict) receptorbinding sites in the FSH molecule is interesting because it is widely held that receptor-binding sites are exposed at the surface, and therefore should be hydrophilic.
A linear combination of these parameters, however, provides an improved prediction of receptor-binding regions. Regression analysis also indicates that hydrophilicity and surface probability, flexibility, or antigenic index are highly correlated and therefore, all together provide redundant and probably irrelevant information to the regression model. Among the secondary structure parameters of hFSHp, the percentage of predicted P-turns or extended structures and a-helixes (with a contribution to R2 of 0.35, 0.23, and 0.15, respectively) seem to be more meaningful than flexibility or surface probability.
Perhaps the more interesting observation derived from the regression analysis is that, at least for hFSH-P-subunit, binding regions are characterized by a unique combination of different structural and physicochemical properties.
The identification of a second binding region was further confirmed by the ability of FSH-P-(81-95) to modify the response to FSH in Sertoli cell cultures. Since FSH is a complex molecule with two subunits, each probably contributing to the interaction with receptor, it would be surprising if a small peptide totally mimicked the effect of the intact hormone.
However, the discovery of partial agonists was crucial to the development of several syntropic antagonists (Black, 1989). FSH+  and FSH-P-(81-95) peptides, are each partial agonists of FSH-stimulated estradiol biosynthesis. Selective and appropriate chemical modifications could lead eventually to the development of a full antagonist or agonist of FSH.