Proc. Nati. Acad. Sci. USA Vol. 88, pp. 760-764, February 1991 Medical Sciences

Conversion of human choriogonadotropin into a follitropin by protein engineering (chimeric hormones/receptor binding/structure-function relationships)

ROBERT K. CAMPBELL, DIANA M. DEAN-EMIG, AND WILLIAM R. MOYLE Department of Obstetrics & Gynecology, University of Medicine and Dentistry of New Jersey, Robert Wood Johnson Medical School, Piscataway, NJ 08854

Communicated by Seymour Lieberman, October 9, 1990 (received for review August 22, 1990)

ABSTRACT Human reproduction is dependent upon the actions of follicle-stimulating hormone (hFSH), lutenizing hormone (hLH), and chorionic gonadotropin (hCG). While the a subunits of these heterodimeric proteins can be interchanged without effect on receptor-binding specificity, their (3 subunits differ and direct hormone binding to either LH/CG or FSH receptors. Previous studies employing chemical modifications of the hormones, monoclonal antibodies, or synthetic peptides have implicated hCG (-subunit residues between Cys-38 and Cys-57 and corresponding regions of hLHP and hFSH(3 in receptor recognition and activation. Since the P subunits of hCG or hLH and hFSH exhibit very little sequence similarity in this region, we postulated that these residues might contribute to hormone specificity. To test this hypothesis we constructed chimeric hCG/hFSH ( subunits, coexpressed them with the human a subunit, and examined their ability to interact with LH and FSH receptors and hormone-specific monoclonal antibodies. Surprisingly, substitution of hFSH(3 residues 33-52 for hCG(3 residues 39-58 had no effect on receptor binding or stimulation. However, substitution of hFSH( residues 88-108 in place of the carboxyl terminus of hCG(3 (residues 94-145) resulted in a hormone analog identical to hFSH in Its ability to bind and stimulate FSH receptors. The altered binding specificity displayed by this analog is not attributable solely to the replacement of hCGI3 residues 108145 or substitution of residues in the "determinant loop" located between hCG(3 residues 93 and 100.

numbering) has also been proposed to play an important role in receptor binding and stimulation. Synthetic peptides containing hCG, hLH (5), or hFSH (8, 9) ,3-subunit residues from this region inhibit hormone binding and stimulate their respective receptors. In addition, amino acid residues in these regions have been implicated in the recognition of hCG and hFSH by antibodies which inhibit hormone binding to receptors (10-12). Proteolytic cleavage of this region of the 83 subunit in ovine LH greatly reduces hormone activity (13). We devised a strategy to identify (3-subunit sequences that determine hormone specificity. We postulated that the homologous placement of their 12 cysteine residues (Fig. 1 Upper) would permit the creation of chimeric hCG/hFSH (3 subunits which would combine with the a subunit. Since hCG and hFSH exhibit low cross-reactivity in receptor binding (17), we anticipated that we would be able to identify determinants of receptor-binding specificity by comparing the abilities of these heterodimers to bind to LH and FSH receptors. By focusing on a positive endpoint-i.e., switching receptor binding specificity-rather than a negative one such as loss of binding, we expected to be able to distinguish between specific effects of the mutations on binding determinants and nonspecific disruption of hormone structure.

METHODS Plasmids. Plasmids pSVL-hCG(3 and pSVL-hCGa were constructed by transferring the hCG(3 cDNA from pBR322chCGj3 (18) or hCGa cDNA from a-hCG/pBR322 (14) into the Sma I site of the transient expression vector pSVL (Pharmacia) using standard techniques (19). Plasmid pKBM is a derivative of pUC18 (20) in which the polylinker has been replaced by an oligonucleotide cassette encoding recognition sites for HindIII, Xho I, Xba I, Sst I, BamHI, BssHII, Sal I, Sst II, and EcoRI. Plasmid pKBM-hCG( was constructed by transferring the hCG(3 cDNA from pSVL-hCG(3 into pKBM, using Xho I and BamHI restriction sites common to the two plasmids. Plasmids were maintained in Escherichia coli strain

Follicle-stimulating hormone (FSH; follitropin), luteinizing hormone (LH; lutropin), chorionic gonadotropin (CG; choriogonadotropin), and thyroid-stimulating hormone (TSH; thyrotropin) constitute a family of heterodimeric glycoproteins that regulate the development and function of the ovary, testis, and thyroid (1). Within a species the a subunit is encoded by a single gene and can be interchanged between hormones without effect on receptor binding, whereas the (3 subunits differ and direct binding specificity (2, 3). Both subunits are required for full biological activity. On the basis of sequence differences between the P subunits and the results of chemical modification studies it has been proposed that amino acid residues between Cys-93 and Cys-100 in human CG ( subunit (hCG() and the corresponding regions of the other hormones form a "determinant loop" which directs hormone specificity (4). Recent studies have demonstrated that synthetic peptides containing the determinant loop region from hCG(3 (5) or hFSH,8 (6) inhibit the binding of 125I-labeled hormone to receptors, albeit at high concentrations (10-100 EM). The FSH peptide also stimulates granulosa cell steroidogenesis. In addition, phosphorylation of hCG(3 residue Thr-97 disrupts binding to LH receptors (7). The region of the ( subunit between Cys-38 and Cys-57 (hCG

DH5a or HB101. Cassette Mutagenesis. To facilitate synthesis of the hCG/ hFSH (3-subunit chimeras we altered the hCG,8 cDNA to provide unique recognition sites for the restriction endonucleases Nae I and PpuMI at nucleotides 189 and 388, respectively. An oligonucleotide cassette prepared by using the phosphoramidite method (21) and which contained silent mutations in the codons for hCG(3 residues Val-48 (GTC -GTQ) and Leu-49 (CTG -* CTC) was inserted between the Pst I and Bsu36I sites of pKBM-hCG,8. This destroyed unwanted recognition sites for Nae I and PpuMI. Briefly, 100 ng of cassette DNA was combined with Pst I/Bsu36I-cut pKBM-hCGf and ligated in a reaction with 1 unit of T4 DNA Abbreviations: FSH, follicle-stimulating hormone (follitropin); LH, luteinizing hormone (lutropin); CG, chorionic gonadotropin (choriogonadotropin); h prefix, human; r prefix, recombinant; a or f3 suffix, a or (3 subunit; mAb, monoclonal antibody.

The publication costs of this article were defrayed in part by page charge payment. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. §1734 solely to indicate this fact. 760

Medical Sciences: 10

Campbell et al. 20

Proc. Natl. Acad. Sci. USA 88 (1991)

30

140

50

60

761

70

hCGP

SKEPLRPRCRPINATLAVEKEGCPVCITVNTTICAGYCPTMTRVLQGVLPALPQWVCNYRDVRFESIRLPGCP : : 1 I . I 1: 1 . : I ...................................... 1 hFSHP NSCELTNITIAIEKEECRFCISINTTWCAGYCYTRDLVYKNPARPKIQKTCTrFKELVYETVRVPGCA .

1

.

10

80

hCGP

.

90

I

1

20

30

100

110

40

50

120

1,30

1

60

FIG. 1. (Upper) Amino acid seof naturally occurring hCG (14) and hFSH (15) (8 subunits. The standard single-letter amino acid code is used; cysteines ( I ), other identical residues (:), and positions with conservative substitutions (16) (.) are marked. Regions of low Receptor similarity chosen for boxed. Spec if ic i ty Both hCG(3 (top) and study hFSH/3are(bottom) LH numbering schemes are provided. (Lower) Schematic depiction of chimeric P FSH subunits and summary of receptor bind-

140 145 RGVNPVVSYAVALSCQCALCRRSTTDCGGPKDHPLTCDDPRFQDSSSSKAPPPSLIPSPSRLPGPSDTPILPQ

....1.. ..::I

.

I

hFSHP HHADSLYTYPVATQCGKCDSDSTDCrVRGLGPSYCSFGEMKE 70

80

90 1

hCGP hFSHP CF39-58 CF94-117 CF39-58+94- 117 CF39-58+94-97 CF94-114 CF94-97 CF108-114

100 39

(33)

58

(52)

111 94

(88)

117

(111)

I

LH

[

C [ [ [

145

I

FSH

MME....

FSH

LH FSH

I

[

LH

C

LH

ligase (overnight incubation, 15°C), after which the modified plasmid, called pKBM-hCGf3', was isolated from ampicillinresistant colonies of E. coli DH5a. Cassettes encoding the hFSH(3 amino acid sequences illustrated in Fig. 1 Lower were ligated into pKBM-hCGf' between Nae I and Taq I restriction sites (for replacement of hCG,3 amino acid residues 39-58), ApaLl and PpuMI (to replace residues 94-97), or PpuMI and BamHI (to replace residues 108-145). The chimeric constructs were subsequently transferred from pKBM to pSVL, using the unique Xho I and BamHI restriction sites. Plasmid pSVL-3CF94-97 was used to construct CF94-117 and CF94-114 by cassette replacement using the BamHI site and a Sca I site which had been introduced at the codons for residues 97 and 98 (hCG numbering). Constructs containing multiple regions of hFSH,6 in hCG,-i.e., CF39-58+94-97, CF39-58+94-114, and CF39-58+94-117-were assembled by swapping amino acid residues 1-88 from pSVL-,BCF94-97, pSVL-,3CF94-114, or pSVL-13CF94-117 with the corresponding region of pSVL-,8CF39-58, using Pvu II. Cesiumgradient-pure DNA used in transfections was sequenced by using T7 DNA polymerase (22). Transfection of COS-7 Cells. A previously described DEAE-dextran method (23) was used to cotransfect COS-7 cells (24) with pSVL-derived plasmids containing native or chimeric, subunits and pSVL-hCGa. Eighteen hours after transfection the medium was replaced with serum-free Dulbecco's modifled Eagle's medium (DMEM). After another 72 hr the medium was harvested and centrifuged to remove cellular debris. Analysis of Biological and Immunological Activity. Monoclonal antibody (mAb) "sandwich assays" used to map or quantify recombinant proteins were performed as described previously (10). Media harvested from transfected COS-7 cells were concentrated by using Amicon Centriprep-10 concentrators. The recombinant hormones and chimeric

an-

alogs were quantified by dimer-specific sandwich assays employing mAbs to human a subunit (A110, A113) and hCG P subunit (B105, B108). Similar results were obtained with either a- or ,B-subunit capture and with different pairs of antibodies. In addition, analogs that bound anti-hFSHB antibody B601 gave comparable results when quantified in an

quences

ing specificity; open,

hCG(3

sequences;

shaded, hFSHP sequences. The nomenclature for the constructs identifies the first and last nonconserved amino acid residues in each region of substitution according to the hCGf3 numbering scheme in Upper. Residue numbers for

hFSH.B

are

in parentheses. Receptor

specificities (LH

or

FSH)

determined

from receptor binding assays (see Fig. 2) are also indicated.

hFSH dimer-specific sandwich assay with mAb B601 and anti-a-subunit mAb E501 (25). Membrane-bound LH receptors for the hCG radioligandreceptor assays were obtained from rat corpora lutea induced to superovulate with pregnant mare serum gonadotropin and hCG as described (26). The FSH radioligand-receptor assays were done with a membrane preparation from bovine testes (8). Highly purified urinary hCG (CR121) and pituitary hFSH (NIH-AFP-4822-B) were used as reference standards. Labeled hormones (hCG CR121 and NIH-FSH-I-3) were prepared by using an Iodo-Gen-based method (26). LH-receptor-mediated steroidogenesis assays were performed as previously described (27), using a testosterone antiserum obtained from Cambridge Medical Diagnostics. The FSH-receptor-mediated steroidogenesis assays were performed essentially as previously described (28), using granulosa cells obtained from rats implanted with pellets containing diethylstilbestrol (DES). Approximately 40,00050,000 cells were added to each well of a standard 96-well culture plate to which sterile hormone or analog had already been added to give a final volume of 100 ,ul per well. After 72-hr incubation estradiol was measured directly from aliquots (75 ,ul) of the culture fluid by radioimmunoassay using a highly specific antiserum. At high concentrations (-7.5 x 10-10 M) of hFSH or analogs with FSH activity estradiol production declined. Inhibition did not occur when large incubation volumes (500 ,ul per well of a 24-well plate) were used, suggesting that it was due to the effects of high cell concentration, as noted previously (28). RESULTS Recognition of Chimeric .8 Subunits by mAbs. The structures of the chimeras and the effects of the mutations on 13-subunit conformation were characterized by sandwich assays employing mAbs to hCG and hFSH (Table 1). Among the antibodies used were A113 (10), which binds the a subunit in both hCG and hFSH, B101 (29) and B105 (30), which bind the , subunit of hCG but not hFSH, and B601, which binds the ,B subunit of hFSH but not hCG. B101 and B601 inhibit hormone-receptor binding and do not bind to hormone

762

Medical Sciences: Campbell et al.

Table 1. Sandwich immunoassays of chimeric hCG/hFSH analogs adsorbed to mAb B105 Net mAb bound, cpm Analog A113 B101 B601 rhCG 5719 12,820 -31 a + CF39-58 224 13,620 10,416 a + CF94-97 14,250 4626 -57 a + CF39-58+94-97 13,453 -92 11,663 a + CF94-117 12,699 4519 -55 a + CF39-58+94-117 8,944 -50 6,878 a + CF94-114 13,362 5154 -58 a + CF108-114 9,711 4441 -53 In these sandwich assays anti-hCGfi mAb B105 was adsorbed to the surface of a microtiter plate and used to capture recombinant hCG (rhCG) or chimeric analogs (10). Bound analytes were detected with 125I-labeled mAbs to human a subunit (A113), hCG (3 subunit (B101), or hFSH ,B subunit (B601). Results for each detection antibody are presented as cpm bound minus the blank, averaged from triplicate samples.

receptor complexes (ref. 29 and D.M.D.-E. and W.R.M., unpublished observations), whereas B105 binds well to hCGreceptor complexes (30). All of the chimeric 3 subunits combined with a subunit and retained the B105 epitope as indicated by their ability to form sandwiches between B105 and A113 (Table 1). Chimeras in which hCGB residues 39-58 were replaced with the corresponding region from hFSH(3 (analogs CF39-58, CF3958+94-97, and CF39-58+94-117) did not bind anti-hCG mAb B101. This finding is consistent with our previous studies, which implicated this region of hCGI3 in B101 binding (10, 11). These same chimeras were recognized by anti-hFSHB mAb B601 (Table 1), indicating that hFSH/3 residues between 33 and 52 participate in B601 binding and have similar conformations in hFSH and the hCG/hFSH chimeras. Replacement of hCGP3 residues 94-145 in the absence of changes in residues 39-58 (e.g., analog CF94-117) had no apparent effect on recognition by either Bl01 or B601 (Table 1). Receptor Binding. We studied the abilities of the recombinant hormones and analogs to displace 125I-labeled hCG and hFSH from LH and FSH receptors, respectively. Analog CF94-114 exhibited greatly diminished binding to LH receptors and recognized FSH receptors indistinguishably from recombinant hFSH (Fig. 2). This effect is not attributable to the removal of hCGP3 residues 108-145, since CF108-114, which also lacks these residues, retained LH-receptor specificity. It is also not attributable solely to the replacement of hCGB residues 94-97 in the "determinant loop." Analog CF94-97 did not exhibit any tendency to bind FSH receptors, although it did display a 30-fold decrease in LH-receptor binding relative to recombinant hCG. Conversely, receptor binding specificity was unaffected by the substitution of hFSH,3 residues for hCG,3 39-58. Analog CF39-58 bound LH receptors as well as recombinant hCG, and it did not bind any better than recombinant hCG to FSH receptors (Fig. 2). In addition, the presence of hCGf residues 39-58 in analog CF94-117 did not adversely affect its ability to bind FSH receptors, since it bound as well as both CF3958+94-117 and recombinant hFSH. Thus, the region of the 8 subunits corresponding to hCGB Cys-38-Cys-57 (Cys-32Cys-51 in hFSHB) does not appear to contribute to binding specificity and may be interchangeable among glycoprotein hormones despite the low conservation of its sequence. Receptor Stimulation. Since receptor binding and stimulatory activity in hCG and hFSH can be separated (27, 31) we sought to determine whether the observed changes in binding specificity were associated with similar changes in the ability of the analogs to stimulate LH and FSH receptors. To do this we used steroidogenesis assays specific for hCG (testoster-

Proc. Natl. Acad Sci. USA 88 (1991)

m CL

280 0

0

80

40 m 60 -

0 CD 120

8C iz

100ii

L-

0

C 80 z o

a'

60

I (A _

40 Z

-6

E x 0

&*

20 1L 110-10

109

106

Chimera (M) FIG. 2. (Upper) Inhibition of 1251-hCG binding to rat corpora lutea LH receptors by recombinant (r) hCG, hFSH, or hCG/hFSH chimeras. (Lower) Inhibition of 1251-hFSH binding to bovine testes FSH receptors. Data for analogs with hCG-like behavior are illustrated with * (-, rhCG; - . -, CF39-58; and , CF108-114); data for analogs with hFSH-like behavior are shown as ( rhFSH; --, CF94-117; ---, CF94-114; and - , CF39-

rhCG, rhFSH,

or

58+94-117); and data for analogs that behave differently from hCG and hFSH are shown as A (--, CF94-97 and----, CF39-58+9497). Results from triplicate tubes are presented as percent displacement of total 1-1I-hCG or 1251-hFSH binding measured in the absence of inhibitor (23). Maximal displacement (100%6) is defined as that obtained in the presence of 1.25 x 10-8 M highly purified urinary hCG (CR121) or pituitary hFSH (NIH AFP-4822-B). The data illustrated are a composite of several typical assays.

one production by rat Leydig cells) or hFSH (estradiol production by rat granulosa cells). These assays indicated that the stimulatory activities of the analogs were closely related to their binding activity (Fig. 3). This suggests that the determinants of receptor binding and activation may be closely linked, or that a region of homologous structure in the different hormones participates in receptor stimulation. In-

terestingly, while CF94-114, CF94-117, and CF39-58+94117 were similar to hFSH in their ability to bind and stimulate FSH receptors, they were 10-20 times more potent than hFSH in stimulating Leydig cell steroidogenesis, an "LHspecific" response.

DISCUSSION We have identified a sequence from hFSHB that acts as a determinant for FSH receptor binding when transferred into hCGI3. This finding suggests that despite the complex struc-

Medical Sc'iences:

Campbell et al.

Proc. Natl. Acad. Sci. USA 88 (1991)

ion.

08

-C 1001101

_100;

4E 40 820

if

Ii

20

C

081071

09

A

60

012

10

10

a

10

C

cn94-117 73

80

114

UJ 60

x

CFSH-

-

o

Cimra(M

40-

C

20

0

10-11 rhCG,

10-10

rhFSH,

or

Chimera

10-9 CM)

FIG. 3. (Upper) Dose-response curve of testosterone production rat Leydig cells incubated with rhCG, rhFSH, or hCG/hFSH ,8-subunit chimeras coexpressed with a subunit. (Lower) Doseresponse curve of estradiol production by rat granulosa cells incubated with recombinant hormones and analogs. As in Fig. 2, data for analogs having predominantly hCG activity are illustrated with *,

by

, CF108-114); data for (-, rhCG; -, CF39-58; and analogs having predominantly hFSH activity are shown with ( rhFSH; --, CF94-117; ---, CF94-114; and * -, CF3958+94-117); and data for other analogs are shown as A (--, CF94-97 and --, CF39-58+94-97). The data are depicted as percent maximal testosterone or estradiol synthesis induced by rhCG or rhFSH, respectively, and are a composite from several typical assays (24). Similar to the results shown for hCG, incubation of granulosa cells with up to 2.5 x 10-9 M CF39-58, CF94-97, CF108-114, or CF39-58+94-97 did not stimulate estradiol production. .

9

-

tures of glycoprotein hormones their specificity may be determined by a single contiguous region of the subunit. While we cannot exclude a contribution from the residues corresponding to hCGP 94-97 or 108-114, our data suggest that differences between hCG,8 98-107 and the corresponding region in hFSHP (i.e., 92-101) are particularly important to binding specificity. Within this region hCG and hFSH differ at six residues (hCGP 101-106). All of the chimeric analogs containing hCGP residues 101-106 preferentially bind LH receptors, whereas analogs with the corresponding hFSH residues (i.e., hFSHB 95-100) preferentially bind FSH receptors. The amino acid sequence in this region is completely conserved among the known FSH subunits but is only partly conserved within the LH/CG family, suggesting that it may play a greater role in FSH than in LH/CG function. Interestingly, the ,3 subunit of chicken LH (32) is homologous to hFSH in this region (i.e., Thr-Val-Gln-Gly-Leu-Gly vs.

763

Thr-Val-Arg-Gly-Leu-Gly). This may explain the preferential binding of chicken LH to mammalian FSH receptors (33). However, the sequences in this region do not offer as ready an explanation for the ability of equine LH (eLH) or equine CG (eCG) to interact with bovine or rat FSH receptors, since this region of the eLH/eCG p subunit (34) is much more similar to hCGP (i.e., Gly-Val-Phe-Lys-Asp-His vs. Gly-GlyPro-Lys-Asp-His) than to hFSHpS. The observation that the equinep subunit is inactive when combined with a nonequine a subunit (35) suggests that unique features of both equine subunits may contribute to the mixed function seen with eLH and eCG. Thef/-subunit residues that determine binding specificity may exert their effect through a direct interaction with the target receptor or they might act indirectly by influencing another region of the a or p subunit. Certain antibodies to the a subunit can distinguish between hCG and hFSH, suggesting that the conformation of the a subunit differs between the hormones (36, 37). Several lines of evidence imply that portions of the region encompassing hCGP residues 94-114 (hFSHp 88-108) may be near the a subunit. Asp-111 in the bovine LH ,8 subunit can be cross-linked to the a subunit (38), and residues 111-113 of hCG,8 appear to interact with an antibody that binds to determinants on both the a and /8 subunits (39). In addition, hCGP Thr-97 can be phosphorylated in the free , subunit, but not in the ap heterodimer (40). An indirect effect of this region on receptor binding through an induced change in hormone structure could explain why peptides in this region of hCG13 and hFSHP8 do not appear to interact with LH or FSH receptors (5, 6). The data also reinforce the notion that the various glycoprotein hormone p subunits have highly similar conformations. Even though we chose the least conserved sequences in the proteins for modification and introduced numerous changes in side-chain size, charge, and hydrophobicity, all of the constructs retained binding sites for the a subunit and two or more p-subunit mAbs, and directed binding to LH and/or FSH receptors. Importantly, transfer of hFSHP sequences into hCGP was associated with the development of hFSHspecific activity (antibody or receptor binding), indicating that the transplanted sequences assume conformations similar to those in hFSH. Thus, not only can the intact p subunits be exchanged between hormones to create novel analogs, but portions of their sequences can be exchanged as well. Since transfer of hFSHO residues 88-108 into hCG,8 results in a hormone analog identical to hFSH in its binding and stimulatory activity, we conclude that these residues and/or conserved features ofhCG and hFSH (in the a subunit and/or ,p subunit) are responsible for activating the receptor. This is consistent with data from previous studies indicating that sequences corresponding to hCG8 residues 93-100 (5) and multiple regions of the a subunit (25) participate in hormonereceptor interactions. However, it is difficult to reconcile our data with models of the hormones which identify the "large loop" regions in the p subunits of hCG (residues 38-57) and hFSH (32-51) as principal binding domains. While high concentrations of synthetic peptides consisting of hCG ,p-subunit residues 38-57 inhibit hCG binding to LH receptors and induce testosterone synthesis (5), the activity of these peptides has been shown to be sensitive to nonconservative amino acid substitutions, particularly substitution for Arg-43 (41). The activity of the intact hormone appears to be much less sensitive to amino acid substitutions, since the entire 38-57 region of the hCGP subunit can be replaced by residues from hFSH,8 without altering hormone-receptor interactions. This lack of effect is surprising, since 12 of the 16 residues in the loops differ, including substitutions at 7 of the 9 residues which are invariant among the known sequences of mammalian LH and CG p subunits (34)-i.e., Pro-39 -* Tyr, Met-41 -> Arg, Arg-43 -o Leu, Leu-49 -- Ala,

Pro-50

Proc. Natl. Acad Sci. USA 88 (1991)

Medical Sciences: Campbell et al.

764 --

Arg, Pro-53

-*

Ile, and Val-56

--

Thr.

Similarly,

while peptides containing hFSH/3 residues 34-37 or 33-53 reportedly bind, and in the latter case, stimulate, FSH receptors (8, 9), the "large loop" of hFSH appears to be neither sufficient nor required for full FSH activity of the chimeric subunits. Analogs containing hFSH.3 residues 88-108 are as effective as hFSH,3 in directing binding and stimulation of FSH receptors, regardless of whether they contain other hFSH,3 sequences. Although the hormone receptors appear to be blind to differences in the "large loops," antibodies directed against hCG and hFSH are not, since substitution of the loops is associated with dramatic changes in antibody recognition. This indicates that the two loops do have distinguishing features that can direct specific interactions with other proteins. Perhaps the observed activities of these peptides is unrelated to their conformation in the intact hormone, as has been suggested for the immunological characteristics of certain peptides (42). Alternatively, conserved structural motifs or residues (Val-44, Gln-54) in the hCG and hFSH P subunits may participate in binding or stimulation of both LH and FSH receptors, despite the overall lack of homology between the two hormones in this region.

In summary, we have shown that hormone-specific characteristics such as receptor or antibody binding specificity can be transferred between glycoprotein hormones as independent "domains" by substitution of contiguous regions in their subunits. In particular we have identified a region of the hFSH p subunit (residues 88-108) that can be used to convert hCG into a follitropin-like protein. When combined with recent progress toward determining the tertiary structures of the glycoprotein hormones (43, 44) and the cloning of their receptors (45, 46), the ability to build informative mutants and to transfer functional properties between hormones will facilitate our understanding and practical manipulation of their structure. We thank Dr. Kelly M. McMasters for the many discussions that contributed to this work. We also thank the following individuals for graciously providing materials used in these studies: Dr. John C. Fiddes for the hCGa and hCG,8 cDNAs; Dr. Robert E. Canfield for the urinary hCG standard (CR121) and mAbs B101, B105, and B108; Dr. Robert J. Ryan for mAb E501; Dr. Douglas Bolt for bovine testes membranes and suggestions for the hFSH radioligand-receptor assays; Dr. Scott C. Chappel for recombinant hFSH; Drs. E. Glenn Armstrong and Robert Wolfert for mAbs A113 and B601; and Drs. Richard Krogsrud and S. Berube for mAb A110. Pituitary hFSI for radioiodination (NIH-FSH-I-3) and unlabeled standard (NIH-AFB4822-B) were provided by the National Hormone and Pituitary Distribution Program, National Institute of Diabetes and Digestive and Kidney Diseases. This work was supported by the National Institutes of Health (Grants HD14709 and HD24650) and the New Jersey Center for Advanced Biotechnology and Medicine. 1. Pierce, J. G. & Parsons, T. F. (1981) Annu. Rev. Biochem. 50, 465-495. 2. Fiddes, J. C. & Talmadge, K. (1984) Recent Prog. Horm. Res. 40, 43-74. 3. Liao, T.-H. & Pierce, J. G. (1970) J. Biol. Chem. 245, 3275-3281. 4. Ward, D. N. & Moore, W. T., Jr. (1979) in Animal Models for Research on Contraception and Fertility, ed. Alexander, N. J. (Harper & Row, New York), pp. 151-164. 5. Keutmann, H. T., Charlesworth, M. C., Mason, K. A., Ostrea, T., Johnson, L. & Ryan, R. J. (1987) Proc. Natl. Acad. Sci. USA 84, 2038-2042. 6. Santa Coloma, T. A. & Reichert, L. E., Jr. (1990) J. Biol. Chem. 265, 5037-5042. 7. Ratanabanangkoon, K., Keutmann, H. T., Kitzmann, K. & Ryan, R. J. (1983) J. Biol. Chem. 258, 14527-14531. 8. Sluss, P. M., Krystek, S. R., Jr., Andersen, T. T., Melson, B. E., Huston, J. S., Ridge, R. & Reichert, L. E., Jr. (1986) Biochemistry 25, 2644-2649.

9. Santa Coloma, T. A., Dattatreyamurty, B. & Reichert, L. E., Jr. (1990) Biochemistry 29, 1194-1200. 10. Moyle, W. R., Matzuk, M. M., Campbell, R. K., Cogliani, E., Dean-Emig, D. M., Krichevsky, A., Barnett, R. W. & Boime, I. (1990) J. Biol. Chem. 265, 8511-8518. 11. Moyle, W. R., Dean-Emig, D. M., Lustbader, J. V. & Keutmann, H. T. (1988) ICSU Short Rep. 8, 116. 12. Schneyer, A. L., Sluss, P. M., Huston, J. S., Ridge, R. J. & Reichert, L. E., Jr. (1988) Biochemistry 27, 666-671. 13. Bousfield, G. R. & Ward, D. N. (1988) J. Biol. Chem. 263, 1260212607. 14. Fiddes, J. C. & Goodman, H. M. (1979) Nature (London) 281, 351-356. 15. Watkins, P. C., Eddy, R., Beck, A. K., Vellucci, V., Leverone, B., Tanzi, R. E., Gusella, J. F. & Shows, T. B. (1987) DNA 6,205-212. 16. Dayhoff, M. O., Schwartz, R. M. & Orcutt, B. C. (1978) in Atlas of Protein Sequence and Structure, ed. Dayhoff, M. 0. (Natd. Biomed. Res. Found., Silver Spring, MD), pp. 345-352. 17. Lee, C. Y. & Ryan, R. J. (1972) Proc. Nati. Acad. Sci. USA 69, 3520-3523. 18. Fiddes, J. C. & Goodman, H. M. (1980) Nature (London) 286, 684-687. 19. Maniatis, T., Fritsch, E. F. & Sambrook, J. (1982) Molecular Cloning: A Laboratory Manual (Cold Spring Harbor Lab., Cold

Spring Harbor, NY). 20. Yanisch-Perron, C., Vieira, J. & Messing, J. (1985) Gene 33, 103-119. 21. Beaucage, S. L. & Caruthers, M. H. (1981) Tetrahedron Lett. 22, 1859-1862. 22. Tabor, S. & Richardson, C. C. (1987) Proc. NatI. Acad. Sci. USA 84, 4767-4771. 23. Cullen, B. R. (1987) Methods Enzymol. 152, 684-704. 24. Gluzman, Y. (1981) Cell 23, 175-182. 25. Ryan, R. J., Keutmann, H. T., Charlesworth, M. C., McCormick, D. J., Milius, R. P., Calvo, F. 0. & Vutyavanich, T. (1987) Recent Prog. Horm. Res. 43, 383-422. 26. Moyle, W. R., Anderson, D. M., MacDonald, G. J. & Armstrong, E. G. (1988) J. Recept. Res. 8, 419-436. 27. Moyle, W. R., Bahl, 0. P. & Marz, L. (1975) J. Biol. Chem. 250, 9163-9169. 28. Skaf, R., MacDonald, G. J., Sheldon, R. M. & Moyle, W. R. (1985) Endocrinology 117, 106-113. 29. Moyle, W. R., Ehrlich, P. H. & Canfield, R. E. (1982) Proc. NatI. Acad. Sci. USA 79, 2245-2249. 30. Moyle, W. R., Pressey, A., Dean-Emig, D. M., Anderson, D. M., Demeter, M., Lustbader, J. & Ehrlich, P. (1987) J. Biol. Chem. 262, 16920-16926. 31. Calvo, F. O., Keutmann, H. T., Bergert, E. R. & Ryan, R. J. (1986) Biochemistry 25, 3938-3943. 32. Noce, T., Ando, H., Ueda, T., Kubokawa, K., Higashinakagawa, T. & Ischii, S. (1989) J. Mol. Endocrinol. 3, 129-137. 33. Bousfield, G. R. (1981) Ph.D. thesis (Indiana Univ., Bloomington). 34. Bousfield, G. R., Liu, W.-K., Sugino, H. & Ward, D. N. (1987) J. Biol. Chem. 262, 8610-8620. 35. Bousfield, G. R., Liu, W.-K. & Ward, D. N. (1985) Mol. Cell. Endocrinol. 40, 69-77. 36. Strickland, T. W. & Puett, D. (1981) Endocrinology 111, 95-100. 37. Hojo, H. & Ryan, R. J. (1985) Endocrinology 117, 2428-2434. 38. Weare, J. A. & Reichert, L. E., Jr. (1979) J. Biol. Chem. 254, 6972-6979. 39. Bidart, J.-M., Troalen, F., Bohoun, C. J., Hennen, G. & Bellet, D. H. (1987) J. Biol. Chem. 262, 15483-15489. 40. Keutmann, H. T., Ratanabanangkoon, K., Pierce, M. W., Kitzmann, K. & Ryan, R. J. (1983) J. Biol. Chem. 258, 14521-14526. 41. Keutmann, H. T., Charlesworth, M. C., Kitzmann, K., Mason, K. A., Johnson, L. & Ryan, R. J. (1988) Biochemistry 27, 89398944. 42. Laver, W. G., Air, G. M., Webster, R. G. & Smith-Gill, S. J. (1990) Cell 61, 553-556. 43. Harms, D. C., Machin, K. J., Evin, G. M., Morgan, F. J. & Isaacs, N. W. (1989) J. Biol. Chem. 264, 6705-6706. 44. Lustbader, J. W., Birken, S., Pileggi, N. F., Gawinowicz Kolks, M. A., Pollak, S., Cuff, M. E., Yang, W., Hendrickson, W. A. & Canfield, R. E. (1989) Biochemistry 28, 9239-9243. 45. McFarland, K. C., Sprengel, R., Phillips, H. S., Kdhler, M., Rosemblit, N., Nikolics, K., Segaloff, D. L. & Seeburg, P. (1989) Science 245, 494-499. 46. Sprengel, R., Braun, T., Nikolics, K., Segaloff, D. L. & Seeburg, P. H. (1990) Mol. Endocrinol. 4, 525-529.