help button home button Endocrine Society Endocrine Reviews
HOME HELP FEEDBACK SUBSCRIPTIONS ARCHIVE SEARCH TABLE OF CONTENTS

This Article
Right arrow Abstract Freely available
Right arrow Full Text (PDF)
Right arrow Purchase Article
Right arrow View Shopping Cart
Right arrow Alert me when this article is cited
Right arrow Alert me if a correction is posted
Right arrow Citation Map
Services
Right arrow Email this article to a friend
Right arrow Similar articles in this journal
Right arrow Similar articles in PubMed
Right arrow Alert me to new issues of the journal
Right arrow Download to citation manager
Right arrow Request Copyright Permission
Citing Articles
Right arrow Citing Articles via HighWire
Right arrow Citing Articles via Google Scholar
Google Scholar
Right arrow Articles by Sealfon, S. C.
Right arrow Articles by Millar, R. P.
Right arrow Search for Related Content
PubMed
Right arrow PubMed Citation
Right arrow Articles by Sealfon, S. C.
Right arrow Articles by Millar, R. P.
Endocrine Reviews 18 (2): 180-205
Copyright © 1997 by The Endocrine Society

Molecular Mechanisms of Ligand Interaction with the Gonadotropin-Releasing Hormone Receptor

Stuart C. Sealfon, Harel Weinstein and Robert P. Millar

Fishberg Research Center in Neurobiology (S.C.S.), Departments of Neurology (S.C.S.), Physiology and Biophysics (H.W.), and Pharmacology (H. W.), Mount Sinai School of Medicine, New York, New York 10029; and Medical Research Council Regulatory Peptides Unit (R.P.M.), Departments of Chemical Pathology (R.P.M.) and Medicine (R.P.M.), University of Cape Town Medical School, Observatory 7925, South Africa


    Abstract
 Top
 Abstract
 I. Introduction
 II. Amino Acid Sequences...
 III. Structure-Activity...
 IV. Structure and Conformation...
 V. Functional Structure of...
 VI. Conclusions
 References
 

I. Introduction
II. Amino Acid Sequences of GnRH Receptors
A. GnRH receptor cDNAs
B. General structural features
C. Covalent modifications
D. Gene structure
III. Structure-Activity Relations of GnRH Peptides
A. Overview
B. Comparative structures and activities of vertebrate GnRHs
C. Roles of individual amino acids in GnRH activity at the mammalian receptor
D. Conclusions from peptide structure-activity data
IV. Structure and Conformation of GnRH and Its Analogs
A. Early studies of GnRH conformation
B. Integrated computational and experimental studies
C. Exploration of the entire conformation space of GnRH analogs
V. Functional Structure of the Receptor and Ligand-Receptor Complex
A. Extracellular domains
B. Helix domains
C. Intracellular loop domains
D. Computational modeling of three-dimensional receptor structure
VI. Conclusions


    I. Introduction
 Top
 Abstract
 I. Introduction
 II. Amino Acid Sequences...
 III. Structure-Activity...
 IV. Structure and Conformation...
 V. Functional Structure of...
 VI. Conclusions
 References
 

GONADOTROPIN-RELEASING hormone plays a central role in the biology of reproduction (1), and synthetic GnRH analogs have proven valuable in the treatment of a wide variety of endocrinological and nonendocrinological disorders (2, 3, 4, 5, 6, 7, 8, 9, 10). The decapeptide GnRH is generated in neurons of the medial basal hypothalamus through enzymatic processing of a larger precursor. Released in a pulsatile manner into the portal circulation, GnRH interacts with high-affinity receptors on the gonadotropes in the anterior pituitary, leading to the biosynthesis and release of the gonadotropins LH and FSH. The pulse-timing and concentration levels of GnRH are critical for the maintainence of gonadal steroidogenesis and for normal reproductive function.

Chronic, high concentration agonist stimulation of the pituitary GnRH receptors induce regulatory changes that lead to gonadal hypoactivity. This paradoxical suppression of gonadal function in response to pharmacological levels of agonist is the basis for the utility of GnRH analogs in the treatment of gonadal-steroid sensitive tumors, such as prostate cancer.

The GnRH receptor has been an unabatedly intense and productive subject of research for several decades because of its dual significance both for understanding reproductive biology and for developing medical therapies. The landmark elucidation of the primary sequence of GnRH by the laboratories of Schally (11) and Guillemin (12) inaugurated the field. Previous reviews have documented the subsequent evolution of research into GnRH and its receptors over the intervening decades. The complex regulation of the mammalian receptor, which is critical both for normal reproduction and for therapeutic response to analogs, has been studied in many species (reviewed in Refs. 13–15). The various signal transduction pathways used by the receptor have been investigated (reviewed in Refs. 15–20). Several thousand GnRH analogs have been synthesized and characterized (reviewed in Ref.21) and the amino acid and cDNA sequences for GnRHs have been determined from many vertebrates (reviewed in Refs. 22 and 23).

GnRH receptor clones have recently been isolated. These clones provide the tools and impetus for recent progress in studies of the structure-activity of the receptor-ligand complex (24). The availability of the primary amino acid sequences and cDNAs has made possible the study of the molecular mechanism of action of GnRH and its analogs through receptor mutagenesis and computational modeling of the receptor and peptide (e.g. see Refs. 51 and 174). Thus, the structure-activity of GnRH and its analogs can now begin to be placed in the context of the receptor itself. The present review aims to summarize such work on the structure-activity relations and computational modeling of GnRH analogs and of the receptor. Recent developments will be emphasized, with earlier studies presented in illustrative rather than comprehensive fashion.


    II. Amino Acid Sequences of GnRH Receptors
 Top
 Abstract
 I. Introduction
 II. Amino Acid Sequences...
 III. Structure-Activity...
 IV. Structure and Conformation...
 V. Functional Structure of...
 VI. Conclusions
 References
 
A. GnRH receptor cDNAs
The cDNA sequence of the mouse GnRH receptor was the first determined. The receptor cDNA was cloned by three groups using the murine gonadotrope cell line, {alpha}T3–1 cells (25), as a source of RNA. Efficient heterologous expression of the mammalian GnRH receptor in oocytes using {alpha}T3–1 RNA suggested that this cell line would be a suitable source for cloning the receptor (26). The first clone was isolated using a PCR-based homology cloning strategy (27). Mouse GnRH receptor clones were also identified using Xenopus oocyte (28) and mammalian cell line (29) expression cloning. After the elucidation of the mouse receptor sequence, the homologous pituitary cDNAs were identified in five additional mammalian species and one nonmammalian vertebrate: human (30, 31), rat (29, 32, 33), sheep (34, 35), cow (36), pig (37), and catfish (38). An alignment of the cloned GnRH receptor sequences is shown in Fig. 1Go. The predicted amino acid sequence for the GnRH receptors is more than 85% conserved overall in the six mammalian species reported and is nearly identical within the putative transmembrane domains.



View larger version (58K):
[in this window]
[in a new window]
 
Figure 1. Sequence alignment of cloned GnRH receptors. The putative transmembrane domains are indicated. m, Murine; r, rat; h, human; o, ovine; b, bovine; P, pig; cf, catfish. The helix boundaries were determined by sequence comparison, using the helix ends proposed by Donnelly et al. (233) for the human ß2-adrenergic receptor. The consensus sequence reflects the mammalian sequences only. The junctions between exons in the murine and human receptors are indicated by **.

 
The cow, sheep, and human receptors are 328 amino acids long, while the mouse and rat receptors are 327 amino acids, due to the absence of a residue in the second extracellular domain. According to the consensus-numbering scheme used in this review, this residue is Lys4.77(191) in the human receptor and Glu4.77(191) in the three receptors of ungulates cloned to date. (For a description of the consensus-numbering scheme used in this review, see Fig. 3Go or Ref. 39.) The catfish receptor is 370 amino acids in length and is notable for having a 49-amino acid cytoplasmic carboxy terminus domain not present in the mammalian receptors (38). GnRH receptor cDNAs were also isolated from human breast and ovarian tumors (40) and from rat gonads (41). The sequences obtained from these extrapituitary sources were identical to the pituitary GnRH receptor cDNAs of the corresponding species.



View larger version (51K):
[in this window]
[in a new window]
 
Figure 3. Helical net representation of the human GnRH receptor. The known glycosylation site is marked, and certain key functional residues are numbered. To allow for the generalization of findings in the GnRH receptor to other GPCRs, we utilize a consensus numbering scheme that has been described elsewhere (39) and is used in the NIH GPCR mutation database available via World Wide Web at http://mgddk1.niddk.nih.gov:8000/MutationAnalysis.html (237). In this system, the most conserved residue in a particular TMD (e.g. helix 3) is designated by the index 50. Thus the conserved Arg at the cytoplasmic side of helix 3 is numbered 3.50. The other residues are numbered in relation to the conserved residue. For example, the Lys in TMD 3 is denoted as Lys3.32(121). The reference residue in each helix is indicated by bold type.

 
B. General structural features
Analysis of the primary sequence identifies the GnRH receptor as a member of the rhodopsin-like G protein-coupled receptor (GPCR) family. Three distinct families of G protein-coupled receptors have been identified by molecular cloning. The three classes are: the metabotropic glutamate receptors (42, 43), the secretin-calcitonin-PTH class (44, 45, 46, 47, 48), and the large rhodopsin-like GPCR superfamily (49, 50), of which the GnRH receptor proved to be a member. Sequences of the members of the three classes of G-protein coupled receptors, when analyzed for hydrophobicity, all contain seven putative transmembrane domains (see Fig. 1Go). However, the three classes do not share any discernible sequence homology. Included within the rhodopsin-like family of G protein-coupled receptors are the opsins, G protein-coupled neurotransmitter receptors (adrenergic, serotonergic, dopaminergic, muscarinic acetylcholine, etc.), glycoprotein hormone receptors (FSH, LH/CG, and TSH), and a variety of peptide receptors (49, 50).

A given receptor can be identified as belonging to the rhodopsin-like GPCR family by the presence of certain amino acid motifs conserved within the transmembrane helix domains (TMD), a pattern of conservation that has also facilitated the cloning of a large number of these receptors (see Fig. 2Go). Variations in the pattern of conservation in the GnRH receptor in comparison with other rhodopsin-like GPCRs has proven valuable in elucidating the functional and structural roles of some of these side chains (51) (see Figs. 2Go, and 3Go and below).



View larger version (33K):
[in this window]
[in a new window]
 
Figure 2. Alignment of the TMD domains of the human GnRH receptor primary sequence with the human NK1 receptor (234), ß2 adrenergic receptor (235), and rhodopsin (236) sequences. Some residues conserved among most GPCRs are marked.

 
The identification of the GnRH receptor as a member of the GPCR superfamily is consistent with previous studies of guanyl-nucleotide modulation of binding and coupling, which suggested that the GnRH receptor interacts with a G protein (52, 53). Recent studies with {alpha}T3–1 cells have demonstrated that the G protein involved in coupling to phospholipase C in this cell line is pertussis-toxin insensitive, being Gq and/or G11 (54, 55).

A representation of the putative topology of the human GnRH receptor sequence is presented in Fig. 3Go. The receptor is composed of a single polypeptide chain. Hydrophobicity analysis of the receptor’s primary sequence confirms the presence of seven hydrophobic stretches corresponding to putative transmembrane helical domains, with an extracellular amino terminus and an intracellular carboxy terminus. Direct structural information is available for only two heptahelical membrane proteins. The structure of bacteriorhodopsin, which has little if any sequence similarity with the mammalian receptor proteins (see Ref.56), has been elucidated at 3.5 Å resolution by cryo-electron microscopy of two-dimensional crystals (57). Using the same approach, a 9-Å projection map of bovine rhodopsin, shown in Fig. 4Go (58), and a 6-Å projection map of frog rhodopsin (59) have been obtained. The maps of rhodopsin are consistent with the presence of seven transmembrane domains, as had been predicted from primary sequence analysis. The transmembrane domains of all GPCRs are believed to be {alpha}-helical and arranged around a hydrophilic core in a manner similar to the rhodopsin map (39).



View larger version (37K):
[in this window]
[in a new window]
 
Figure 4. Projection density map of bovine rhodopsin at 9 Å resolution. [Reprinted with permission from G. F. Schertler et al.: Nature 362:770–772, 1993 (58). ©1993 Macmillan Magazines, Limited]

 
Several features conserved among GPCRs are altered in the mammalian GnRH receptors. The mammalian GnRH receptor is the only rhodpsin-like GPCR identified to date that lacks the entire intracellular C-terminal domain. Also unique in the GnRH receptor is the presences of an Asn at the position of a conserved Asp residue in other GPCRs. The GnRH receptor also has unusual substitutions at loci that are highly conserved, such as the modification of the more common DRY sequence of the proximal second intracellular domain to DRS and the substitution of Asp for Asn in the TMD 7 consensus NPXXY motif. Notably, most of these unique features are not conserved in the nonmammalian catfish GnRH receptor sequence (see Fig. 1Go). Mutation of DRS to DRY in the mammalian receptor was reported to cause only a small increase in agonist affinity with no discernible change in signal transduction (60). The other unique features of the mammalian receptors seem to have functional significance and will be discussed below in light of recent experimental results.

C. Covalent modifications
1. Glycosylation. Most GPCRs have consensus glycosylation sites, and several receptors have been found to be glycosylated at these sites (61, 62, 63). Biochemical studies of the GnRH receptor have suggested that it is a sialic acid residue-containing glycoprotein (64, 65). The cow, sheep, pig, and human receptor sequences contain two potential sites for N-linked glycosylation (N-X-S/T), one in the amino terminus and one in the first extracellular domain (see Figs. 1Go and 3Go). The rodent species contain an additional potential glycosylation site in the amino terminus. The glycosylation at these sites of the mouse GnRH receptor was investigated by site-directed mutagenesis and photoaffinity labeling (66). Mutation of Asn1.01(4) or Asn1.15(18) in the N-terminal domain to Gln caused a lower apparent molecular weight in gel electrophoresis, whereas mutation of Asn2.65(102) in the putative first extracellular loop did not affect mobility. Whereas the ligand-binding affinities of the amino terminus domain mutants were unchanged, these receptors were expressed in transfected cells at a lower level than the wild type receptor. These results suggest that only the amino-terminal domain sites are glycosylated and that the glycosylation contributes to the level of receptor expression, consistent with earlier studies in which sialidase and tunicamycin were reported to decrease the level of receptor expression but not to alter affinity (65). The mutagenesis studies do not support an earlier suggestion, based on studying the effects of periodate on binding, that glycosylation contributes to high affinity binding (66). The receptor levels of glycosylation-deficient mutants show similar decreases in both membrane preparations and in whole cell assays (67). Thus, in contrast to the ß-adrenergic receptor, for which glycosylation is required for proper transport of the receptor to the cell surface (62), in the case of the GnRH receptor the observed decrease in receptor number does not appear to represent altered receptor transport.

The possibility that differential glycosylation contributes to the differing level of expression observed with transfection of the mouse and human receptor has been examined by introducing a second glycosylation site into the human receptor sequence, thereby recreating the pattern of sites found in the mouse. The second site was found to be glycosylated, and its presence increased the level of receptor expression (68). These studies indicate that the glycosylation of the GnRH receptor does not contribute to receptor affinity but does improve the level of receptor expression, possibly by decreasing the rate of receptor degradation.

2. Phosphorylation sites. Many intracellular serine and threonine residues are within phosphorylation consensus sequences, and phosphorylation could be involved in modulating receptor responsiveness or intracellular trafficking. While desensitization of the GnRH-induced responses in pituitary cells has been observed, it is not clear whether this occurs at the level of the receptor (for review see Refs. 13–15). Of note, however, a lack of rapid receptor-mediated desensitization of the GnRH receptor has been reported in {alpha}T3–1 cells and in transfected cells (69, 70, 71).

3. Disulfide bridges. Most GPCRs contain single conserved cysteines in the first and second extracellular loops that may form a disulfide bond to stabilize the structure of the functional protein. Mutation of these conserved cysteine residues disrupts the function of rhodopsin, muscarinic, ß-adrenergic, and TRH receptors, suggesting that this disulfide bond is required for proper receptor folding (72, 73, 74, 75, 76). Experiments using site-directed mutagenesis and photoaffinity cross-linking support the presence of two extracellular disulfide bridges in the GnRH receptor. The presence of a conserved cystine bridge between C3.25(114)-C5.23(196) was demonstrated in the mouse receptor, and evidence for a second disulfide bond between C1.11(14)-C5.27(200) has been obtained for the human receptor (J. Davidson, personal communication).

D. Gene structure
The chromosomal locations of the mouse, sheep, and human genes have been reported. The human gene was assigned to chromosome 4 by PCR analysis of somatic hybrid cell lines (77, 78) and to the 4q13.1–4q21.1 region using cell hybrid mapping panels (79). Using chromosomal in situ hybridization, three groups have reported the gene localization at band 4q13.2–13.3 (78, 80, 81) and one group at band 4q21.2 (82). Mapping the gene relative to 4q microsatellite markers in GnRH receptor YAC clones supports the 4q13.2–13.3 assignment (80). The mouse gene has also been mapped by linkage analysis to within 1.2 ± 1.2 centimorgans of the chromosome 5 marker Pmv-11 (79), and the sheep gene has been localized near the FecB locus of chromosome 6 (83).

The structures of the mouse and human GnRH receptor genes have been investigated. In contrast to the genes of many GPCRs, which are intronless and are believed to have arisen by retroposition (84), the GnRH receptor contains introns within the coding region. The mouse gene is composed of at least three exons spanning more than 22 kb (85). The open reading frame is distributed among the three exons, which encode amino acids 1–174, 175–247, and 248–327, respectively (see Fig. 1Go for location of exon junctions). Variant transcripts of the mouse receptor that are generated by alternative splicing and do not encode functional receptors have been isolated from {alpha}T3–1 cells (85). The alternative transcripts found all include exon 1 but lack either exon 2 or 3. The alternative transcripts form a minority of the cDNAs isolated from an {alpha}T3–1 cell library, and the biological function of the proteins encoded by these cDNAs is not known. The human gene is also distributed over three exons that span 18.9 kb (77, 86). The amino acid locations of introns 1 and 2 are homologous to their positions in the mouse receptor gene. Intron I is located between amino acids 174–175 in the putative TMD 4 domain, and intron 2 is located between amino acids 248–249. Exon 2 is three nucleotides longer in the human gene in comparison with the mouse, reflecting the presence of an additional amino acid (Lys4.77(191)) in the second extracellular loop of the human receptor (see Fig. 1Go). Southern blot analysis is consistent with the presence of a single gene in the mouse (85), rat (85), and human (77) genomes.

Fan et al. (86) have mapped the 5'- and 3'-flanking regions of the human receptor gene. Multiple initiation sites and multiple polyadenylation signals are present. Five consensus TATA sequences distributed over 669 bases are present in the 5'-flanking region. Primer extension analysis using human brain RNA indicates the utilization of several initiation sites. The longest extension was confirmed by PCR and represents a transcript with 1393 bp of 5'-untranslated sequence. A putative cAMP response element is found at -1490, a putative glucocorticoid/progesterone response element is located at -92, and consensus binding sites for several transcription factors are present. At the 3'-end of the gene, five polyadenylation signals are found, distributed over 800 bp. In the largest possible transcript, the 3'-untranslated sequence is 3.1 kb in length. Thus the exons of the human gene identified appear to account for the largest ~5-kb transcript identified on Northern blot analysis (31).

The 5'-flanking region of the mouse gene has been investigated by Albarracin and co-workers (87). In contrast to the human gene, the mouse gene appears to have a smaller 5'-untranslated segment. The major initiation site is found 62 bases upstream of the translation initiation site. Also in contrast to the human gene, the 5'-sequence of the mouse gene lacks TATA sequences. Preliminary studies on the regulation of a 1.2-kb fragment of the 5'-flanking region of the mouse gene have been reported (87).


    III. Structure-Activity Relations of GnRH Peptides
 Top
 Abstract
 I. Introduction
 II. Amino Acid Sequences...
 III. Structure-Activity...
 IV. Structure and Conformation...
 V. Functional Structure of...
 VI. Conclusions
 References
 
A. Overview
A prerequisite for a meaningful investigation of ligand-receptor interactions is an understanding of the roles of ligand residues involved in receptor binding and in agonist activation of the receptor. This knowledge base for GnRH may be gleaned from the large number of structure-activity studies on synthetic GnRH analogs and on the 11 naturally occurring GnRH structural variants and their synthetic chimeras. The structural features of GnRH agonist and antagonist analogs were comprehensively reviewed a decade ago (21), and a number of more specialized reviews on antagonists and structural constraints have subsequently appeared (88, 89, 90, 91). The structure-activity relations of naturally occurring vertebrate GnRHs have been extensively reviewed (23, 92, 93, 94, 95, 96, 97, 98, 99). However, these reviews do not attempt to specifically and systematically analyze the roles of each of the individual amino acids in GnRH. The comprehensive review of Karten and Rivier (21) addressed the substitutions that produced superactive agonists and antagonists, whereas the reviews on comparative activities of vertebrate GnRHs focused on the contributions of substitutions in positions 5, 7, and 8. In this section we shall attempt to synthesize these diverse sources of information into a more complete review of the roles of the constituent amino acids of GnRH.

Several thousand GnRH analogs have been synthesized to date, and information on their activities potentially provides a very large data base for the purpose of identifying functional residues in GnRH. However, interpretation is complicated by a number of factors:

1. Frequently, multiple substitutions have been incorporated simultaneously in single GnRH analogs, often without a systematic approach due to the large number of combinatorials involved.

2. Even the effects of single-amino acid substitutions may be difficult to interpret. A single substitution may alter affinity and agonist activity via modification of a side chain that interacts with the binding pocket and/or by altering the conformation of the peptide and thus affecting the presentation of other peptide moieties that interact with the receptor. Substitutions that have a conformational effect cannot be differentiated through structure-activity data alone from those that eliminate receptor contact sites. Any substitution may establish new contacts with the receptor and disturb the normal contacts by altering the families of conformations of the peptide. This difficulty in the interpretation of structure-activity data would be obviated by achieving a more complete understanding of the conformational effects of substitutions and by the analysis of ligand-receptor interactions in a structural context. Progress toward predicting the effects of amino acid substitutions on peptide conformation is described in the next section, whereas the development of three-dimensional models of receptor molecules is discussed in Section V.

3. While substitutions of residues that produce antagonists may remove a contact interaction, they most likely establish new compensatory contact sites, presumably with different sites in the receptor, to retain high-affinity binding.

4. For much of the available data, the activities of analogs cannot be rigorously compared because they have been tested in different assay systems (see Ref. 100 for review). The most commonly employed assays have been in vivo bioassays (e.g. inhibition of ovulation) in which activity is a composite of pharmacokinetics of absorption from the injection site, association with lipophilic compartments (e.g. fat and cell membranes), binding to plasma proteins, degradation, metabolic clearance (including renal clearance), receptor binding affinity, and efficacy. Ideally, comparative data on binding affinity and signal transduction (e.g. second messenger generation) are required. However, as these data are available for relatively few analogs, in adjudging the effects of single-amino acid substitutions we have relied extensively on in vivo data, particularly from early studies. It has been necessary, therefore, to consider possible pharmacodynamic contributions to the activity of the analogs when making inferences about receptor binding and receptor activation based on data obtained from in vivo bioassays (see below).

Even the direct measure of analog affinity in receptor-binding assays may yield misleading results. These assays are usually conducted on membrane preparations (which expose all receptors), as opposed to whole cells, and employ conditions (buffers, pH, temperature, etc.) optimized to give maximal binding. These nonphysiological conditions may affect the binding of a substituted analog or mutated receptor differently than physiological conditions. The radiolabeled GnRH analogs used in binding assays mostly rely on incorporating 125I into Tyr5 of GnRH analogs that have a D-amino acid in position 6. This incorporation of the large electron-withdrawing iodine atom considerably alters the properties of the ligand. Since a large, bulky side chain is allowable when a D-amino acid substitutes for Gly6 in the superactive analogs (see below), we have attempted to overcome this problem by substituting the Tyr5 with His (as in the active chicken GnRH II) and incorporating D-Tyr in position 6 (101). This analog has a higher affinity and increased total binding.

A feature of ligand-receptor complexing is that receptor interaction (affinity) and capacity of the bound ligand to activate the receptor (efficacy) are separable phenomena. Thus particular residues of GnRH are more critical for agonist activity (e.g. His2, Trp3), and others are critical for ligand binding (e.g. Pro9). Various models have been proposed to explain the differing contributions of ligand substituents to affinity and efficacy (102). In the "conformational induction" model, agonists bind to an inactive receptor state and induce the receptor to assume an altered active state that leads to coupling with G proteins. In the "conformational selection" model, the receptor spontaneously fluctuates between inactive and active conformers, and agonists have a higher affinity for the active state whereas antagonists (or inverse agonists) have a higher affinity for the inactive state. Consequently, agonist binding causes the concentration of active receptor to increase by mass action, and inverse agonists have the opposite effect. The separate effects of ligand substitutions on affinity and efficacy can be interpreted within either model. In the case of the induction model, some receptor interaction sites are critical for ligand docking, whereas others are critical for inducing a change in the receptor. In the selection model, some receptor contact sites are accessible in both active and inactive states and thus contribute to affinity, whereas other contact sites are accessible or properly positioned for agonist complexing only when the receptor assumes an activated state. In the following sections, the term "activity" will be used to refer to data derived from functional assays, usually LH release. When radioligand binding data are available, the term "affinity" will be employed.

With the preceding caveats in mind, it is nevertheless possible and useful to review the extant data. When evaluated in concert with studies of the receptor-binding pocket and of analog conformation (reviewed in subsequent sections), the data on the structure-activity of analogs provide insight toward elucidating the interactions in the ligand-receptor complex. For this reason, we present here some indications of the roles of the individual constituent amino acids of GnRH in receptor binding and activation.

B. Comparative structures and activities of vertebrate GnRHs
Identification of the GnRHs present in more than 70 species (for review see Refs. 22, 23, 92–99, and 103–105) has demonstrated that two or more forms of GnRH are present in most vertebrate species and in a protochordate, the tunicate (22, 106, 107, 108, 109, 110, 111, 112, 113, 114, 115, 116) (Fig. 5Go). One form is represented by mammalian GnRH and its nonmammalian counterparts, which have a predominant function as hypophysiotropic peptides regulating the pituitary. The second form of GnRH, first identified in chicken brain (His5Trp7Tyr8GnRH), is the most ubiquitous form in vertebrates, and most species have this form along with one or two other GnRHs. As His5Trp7Tyr8GnRH is present in fish, amphibians, reptiles, birds, and mammals, referring to this peptide as "chicken GnRH II" is confusing. We propose that it be designated "GnRH II." The original mammalian GnRH is then "GnRH I." Specific chemical identification may be accomplished by designating the variable amino acids 5–8. Thus chicken GnRH I, salmon GnRH, and catfish GnRH would be YGLQ GnRH, YGWL GnRH, and HGLN GnRH, respectively.



View larger version (66K):
[in this window]
[in a new window]
 
Figure 5. Amino acid sequences of naturally occurring vertebrate and protochordate GnRHs. The conserved NH2- and COOH-terminal domains are boxed. Only Gly6 is conserved among the higher vertebrates in the central domain.

 
The similarity in organization of genes encoding the different GnRHs indicates that the GnRHs arose from a common ancestral gene (see Ref.23) for review). The distribution of GnRH II in the extrahypothalamic central nervous system and peripheral nervous system suggests a neuromodulatory role for the peptide. The most thoroughly documented effect of GnRH II is the inhibition of K+ channels in the bullfrog sympathetic ganglion (117). The presence of GnRH II in bullfrog and Xenopus sympathetic ganglia was recently demonstrated (118). Using PCR, GnRH receptor sequences have been obtained from fish, amphibian, and reptile (B. Blackman, Y.-M. Sun, N. Illing, J. Hapgood, E. Rumbak and R. P. Millar, unpublished). Multiple GnRH receptors are present in these species, suggesting that the duplication of the GnRH gene was accompanied by a coordinated structural evolution of the cognate receptors.

All natural GnRH peptides isolated to date are highly conserved with respect to their length, to the NH2-terminal domain (residues pGlu-His-Trp-Ser), and to the COOH-terminal domain (Pro-Gly.NH2), suggesting that these domains are functionally essential. However, residue conservation does not invariably imply functional signficance. For example, Ser4 is highly conserved and yet can be substituted with the retention of high activity at the mammalian receptor (see below). Among vertebrate GnRHs, position 8 is most variable, and positions 5 and 7 are highly variable (Fig. 5Go). Position 6 is invariably Gly in the higher vertebrates but varies considerably in the lamprey and tunicate GnRHs. No variation in GnRH sequence is found among mammalian GnRHs.

The comparative activities of the GnRH variants in vertebrates provide insight into structure-activity relations. In mammals, mammalian GnRH is highly active at low doses while the other vertebrate GnRHs, with the exception of GnRH II, have poor activity (binding affinity or EC50 and/or maximal gonadotropin release) (107, 110, 119, 120, 121) (Fig. 6Go). Since the single residue that distinguishes mammalian GnRH from all of the other vertebrate GnRH structural variants is Arg8 (Fig. 5Go), this residue was identified as being critical for high-affinity binding to the mammalian receptor. However, the substantial activity of GnRH II (20–30%) (120, 121, 122, 123) suggests that the loss of activity when substituting a neutral amino acid for Arg8 can be overcome by the simultaneous substitution of His in position 5 and Trp in position 7.



View larger version (31K):
[in this window]
[in a new window]
 
Figure 6. Comparative LH releasing activities of chimeras of vertebrate GnRHs in sheep and chicken pituitary cells. [Derived from Ref. 121.]

 
In contrast with the limited activity of most nonmammalian GnRHs at the mammalian receptor, all of the vertebrate GnRHs (with the exception of lamprey GnRHs, which have an acidic residue in position 6) have similar high activities in all of the nonmammalian vertebrates tested (107, 119, 120, 121, 124, 125). Thus, the nonmammalian vertebrate receptors are promiscuous in interacting well with most of the vertebrate GnRHs, whereas the mammalian pituitary GnRH receptor is selective for mammalian GnRH [(see reviews (23, 92, 93, 94, 95, 96, 97, 98)].

Differences in the pharmacology of the GnRH receptors in vertebrate species are illustrated in studies with certain mammalian GnRH antagonist analogs that are pure antagonists in mammals but exhibit agonist activity in chicken (126) and goldfish (127) gonadotropes. Intriguingly, some antagonists stimulate gonadotropin secretion while others release GH in the goldfish (127). It is apparent, therefore, that the structural requirements of GnRH receptors for activation by ligands are variable among vertebrates. Differences, albeit more subtle, between mammalian GnRH receptors in agonist and antagonist binding have also been noted (30, 31, 35, 51, 128).

The studies on GnRH chimeras have also revealed effects of amino acid substitutions on agonist efficacy at the mammalian GnRH receptor by comparing relative potencies of the chimeric analogs in stimulating LH release from sheep pituitary cells and in binding to sheep pituitary GnRH receptors (Fig. 7Go) (121). These data showed that Arg8 substitution by neutral amino acids in mammalian GnRH resulted in a low binding potency but relatively higher LH-releasing potency (ratios of 10–150). In contrast, Tyr5 substitution by His enhanced binding potency but reduced LH-releasing potency (ratios of 0.14–0.2). Thus, once bound, analogs with a neutral amino acid in position 8 are more efficient at activating the receptor. On the other hand, His5 enhances binding, but reduces efficacy. The role of specific amino acids in affinity and activation of the mammalian receptor will be addressed in the subsequent section.



View larger version (18K):
[in this window]
[in a new window]
 
Figure 7. LH stimulation and receptor binding of vertebrate GnRHs in sheep gonadotropes. [Derived from Ref. 121.]

 
C. Roles of individual amino acids in GnRH activity at the mammalian receptor
The contributions of individual amino acids to GnRH receptor binding and activation may be explored by the systematic substitution of single amino acids. Although a considerable body of research was conducted along these lines in the early 1970s (e.g. see Ref.129), this approach has not been revisited with the advent of specific binding assays and more recently with expressed cloned GnRH receptors.

The conservation of the NH2- and COOH-terminal sequences of GnRH through vertebrate evolution and the conclusion that these domains are critically important for receptor binding and activation are substantiated by extensive structure-activity data. Indeed, cognizance of the evolutionary constraints on acceptable structures could have obviated much of the endeavor to produce agonist and antagonist analogs through empirical approaches. It is now clear that both the NH2- and COOH-terminal domains are involved in receptor binding while the NH2-terminal domain plays the major role in receptor activation (see reviews in Refs. 10 and 21). Although the lack of conservation of amino acids 5–8 suggests that these residues are not critical for ligand activity, this is not entirely so, as Arg8 is important for high-affinity binding to the mammalian receptor (see below). This central domain is thus a determinant of receptor selectivity. The role of individual amino acids will be considered within these designated NH2-terminal, central, and COOH-terminal domains.

1. The NH2-terminal domain (pGlu-His-Trp-Ser).
pGlu1. The essential requirement of pGlu was first noted with the loss of activity of native purified GnRH when treated with pyroglutamyl aminopeptidase and confirmed by a series of substitutions in this position. While Leu1, Gly1, Pro1, Gln1 and (O=)Thr1-substituted analogs were essentially inactive in in vivo bioassays (129, 130, 131, 132), several acylated Gly1 analogs (formyl, acetyl, and propionyl) had low but significant activity (~1%), indicating that the -CO-NHCHCO- group is the minimal structure required for activity (131). The cyclic (O=) Ser1 structure (130), which resembles the pGlu structure, and D-pGlu (133) both had about 5% activity, suggesting that a change in conformation in the NH2 terminus is not completely detrimental (134). Although a specific role for pGlu in binding and/or activation of the receptor has not been revealed by these early studies, the subsequent universal substitution of pGlu in GnRH antagonists (see reviews in Refs. 10, 21, and 88–91) identifies this residue as important in receptor activation.

His2.
His2 of GnRH is a good candidate for interaction with the receptor (134). The imidazole ring of histidine has a number of features appropriate for participation in enzyme actions and also in hormone interactions with receptors. These include its aromatic character, hydrogen-bonding capacity, and acid-base properties. The very low activities of Ser-, Ile-, Leu-, Gln-, Gly-, Thr-, Ala-, Lys-, and Arg-substituted analogs (129, 130, 135) and substantial activity of Phe2 GnRH (1–7%) (130, 135, 136), 3-Me His2 (1%) (129), and Trp2 (40%) (137) demonstrate the need for aromaticity and possibly the imidazole moiety. The lack of activity of Gln2 GnRH indicates that the role of His2 does not involve {epsilon}-position hydrogen-bonding because, were that the case, Gln would be a suitable substitute. The high activity of Trp2 GnRH has been confirmed by examining binding to the human GnRH receptor expressed in COS-1 cells (138). On the other hand, the acid-base and hydrogen-donor and hydrogen-acceptor capability of His can be modified with reasonable retention of activity, as demonstrated by I-Nim-His2 (6%) (139) and ß-pyrazolyl-3-alanine (19%) (140), although the latter is a much weaker base. A major breakthrough in understanding the function of His2 evolved from the observations that the Gly2 was a partial agonist and the des-His2 analog was an antagonist (141). Bulky hydrophobics (e.g. D-4-Cl-Phe) have subsequently become the hallmark of substitutions of His2 in GnRH antagonists (for review see Refs. 10, 21, and 88–91). It is therefore likely that His2 plays a role in GnRH interaction with the receptor, which leads to signal propagation and G protein activation. The nature of this interaction appears to demand an element of aromaticity and may be enhanced by basicity/H-bonding capability. Mutagenesis studies identify Lys3.32(121) in the receptor as a possible site of interaction with His2 (142) (see Section V).

Trp3.
Trp3 is clearly a critical residue in GnRH. Trp2 His3 GnRH and Des-Trp3 GnRH were inactive (143). Substitution with nonaromatic amino acids (e.g. Gly, Leu, and Ala) gives rise to very low activity (129, 143, 144, 145), whereas some activity is present with Tyr (0.1%), as well as Phe (0.5%) substitution, and this is increased substantially in pentamethyl-Phe GnRH (30–70%) (129, 143). Notably, the latter residue resembles Trp in its ability to form {pi}-{pi} complexes with aromatic molecules (134). The introduction of an electron-withdrawing fluorine atom of similar atomic radius in position 5 of Trp3 leads to a marked reduction in activity (6%), presumably due to reorienting the dipole and forming hydrogen bonds itself so that an aromatic interaction does not occur (145). The role of an aromatic side chain in this position is further emphasized by the natural substitution of Tyr for Trp3 in lamprey GnRH I (Fig. 5Go). Even 2-napthyl-Ala substitution results in 50% activity (129). Since D-Trp3 GnRH has poor gonadotropin-releasing activity (133) but has been commonly incorporated in antagonists, it is possible that Trp3 plays a role in receptor activation. The altered stereochemistry evidently has a critical effect on agonistic activity. The role of Trp3 in receptor activation is further suggested by an early study demonstrating antagonist activity of Leu3 GnRH (145) and recent work showing that incorporation of NMe in the amino acid in position 3 converts the peptide to an antagonist (146).

Ser4.
The last of the conserved residues in the NH2-terminal domain, Ser4, can be substituted with a number of amino acids (Ala, Thr, Gln, NMeSer) with reasonable retention of activity (10–20%) (129, 130, 132, 143), yet this is the most conserved residue in the empirically generated analogs. Because substitution with larger amino acids such as Ser (But) and Leu is very detrimental (129), it appears that spatial constraints are paramount. Recent work has shown that constraint of the peptide bond with NMe does not decrease activity (146, 147) unlike most other positions in GnRH. The conclusion from early studies that large side chains are not tolerated is supported by the recent observation that biotinylated Ser4 GnRH is inactive (148).

2. The COOH-terminal domain (Pro-Gly·NH2).
Pro9. The conservation of Pro9 in the natural GnRHs and the expected conformational limitations imposed by Pro on the peptide chain suggest that substitution would not be readily tolerated. Sarcosine9 GnRH and Ala9 GnRH had low activity (<1%) while N-Me-Gly9 had 10% activity (129, 132). The exchange of amino acids 8 and 9 (Pro8 Arg9 GnRH) also results in very low activity (129). The discovery that Pro9 may be hydroxylated in fetal brain and decreases activity to 10% (149) underlines the importance of this conserved residue and suggests that this may be a regulatory mechanism.

Gly-amide10.
Removal of the amide to yield the free acid of GnRH results in very low activity (150). This has recently been confirmed for the human GnRH receptor expressed in COS-1 cells (67). Replacement of the Gly-NH2 moiety with Ala resulted in a mild reduction in activity (10%) (130), and a similar reduction was observed with Gly-NMe2 (14%) (132). On the other hand, substitution of Gly-NH2 with alkylamides maintained (methylamide and ethanolamide) or increased activity up to 600% (propylamide and ethylamide) (150, 151) whereas substitution with larger amides (pyrolidineamide and morpholineamide) (150) or D-amino acids (129) decreased activity. The incorporation of electron-withdrawing fluorine atoms into the ethylamide (2,2,2-trifluoroethylamide) further enhanced activity to about 900% (151). These findings suggest that the terminal Gly-NH2 is not essential for activity and that small, uncharged moieties are acceptable at the COOH terminus. Larger groups are inhibitory, possibly by sterically hindering ligand access to the binding site. The findings also suggested that the total chain length might have an important role in the binding of GnRH to its receptor (150, 151). Recent mutagenesis of a receptor site (N2.65(102)->A) has demonstrated a much greater decrease in binding affinity of Gly-NH2 ligands than N-ethylamide ligands (152) (see Section V).

3. The central nonconserved domain (Tyr-Gly-Leu-Arg).
Try5. In accordance with the lack of conservation of Tyr5 in vertebrate GnRHs, substitution in this position is well tolerated. The 44–64% activity of Phe5 (143, 153) demonstrated that the hydroxyl group is not required. Interestingly this substitution has yet to be found in naturally occurring GnRHs, although it would require only one base change. Substitution of the hydroxyl group of tyrosine resulted in activities of 37% (amino), 24% (methoxy), and 5% (nitro) (154). NMe Tyr substitution, which has been proposed to constrain the peptide backbone and to eliminate one of two postulated H bonds with Arg8 in a ß-II turn conformation, led to a reduction in binding affinity to 10–20% (155) Interestingly, mono-iodo-Tyr-GnRH (129) and mono-chloro-Tyr-GnRH (130) had activities of 30–80% and 10%, respectively, while di-iodo-Tyr5 GnRH and di-chloro-Tyr GnRH were devoid of LH-releasing activity (129, 130). His5 GnRH has very high binding affinity for mammalian GnRH receptors (121). These findings demonstrate that the hydroxyl group of Tyr5 is not required, and that simply an aromatic side chain (Trp, Phe, or His) is adequate for high LH-releasing activity. The findings suggest that Tyr5 contributes only to receptor binding and does not play a role in the process of receptor activation. However, substitution of Tyr5 with His, as in GnRH II, results in an analog with high receptor potency (aromaticity maintained) but reduced LH release (partial agonism; (121). Partial agonism is also observed with His5 D-Trp6 GnRH and in His5 D-Tyr6 GnRH and Arg5 D-Tyr6 GnRH (R. P. Millar, unpublished). However, when Arg8 is substituted by Tyr in analogs with His or Arg in position 5, efficacy is restored. Thus the motif His5/Arg5-Xxx-Xxx-Arg8 produces compounds with high binding but diminished receptor activation, indicating that the Tyr5 does play a role, albeit possibly indirect, in receptor activation in the mammalian ligand-receptor complex.

Gly6.
Gly6 is conserved in all vertebrates except the ancient jawless lamprey and is also absent in the tunicate GnRHs (Fig. 5Go). The presence of this small residue in this position allows for flexibility and the assumption of the postulated ß-II-type bend and the preferred conformation for receptor binding (see below and Section IV). This bend would be energetically unfavorable in analogs with larger L-amino acid substitutions for Gly6, and Ile, Val, and Ala analogs were found to have low activity (132, 137). However, the folded conformation is favored by the stereochemistry of D-amino acid substitutions (10, 21, 88, 89, 90, 91, 137). The proposal of a ß-II-type bend for the active conformation of GnRH was first proposed by Monahan et al. (137) after demonstrating that D-Ala6 substitution increased activity to about 400%. This seminal work led to the exploration of numerous substitutions with D-amino acids in this position (see reviews in Refs. 10, 21, and 88–91). In general, substitution with D-amino acids having bulky hydrophobic side chains, particularly aromatics, was most effective (10, 21), and this has been confirmed in numerous binding studies using pituitary membranes and, more recently, with receptors expressed in COS-1 cells. A correlation between hydrophobicity (HPLC retention time) of the D-amino acid and potency has been noted (100). It appears that there is a large "allowable space" facing away from the NH2- and COOH-domains which interacts with the receptor, and this will accommodate the D-amino acids with large side groups (147). In addition to further favoring the ß-II-type conformation, the large side chains of the D-amino acids may interact with nearby residues in the receptor, thereby enhancing the binding affinity. These potential alternative interactions probably account for species differences and are likely to be prevalent in GnRH antagonists with numerous unusual side chains, often aromatic. These features must be taken into account when analyzing the effects of mutagenesis of receptor residues on the binding of these analogs.

Leu7.
The comparative studies of activities of vertebrate GnRHs indicate that substitutions of Leu7 with uncharged L-amino acids with varying size side chains are generally well tolerated. This supposition is confirmed by the demonstration that Val, Ile, Nle, Ser, ethoxycarbonyl-Lys, butoxycarbonal-Lys, and Boc-Lys all had high activities (16–45%) (129, 130) while Ala and Gly had lower (3–6%) activities (129, 130). Potential disruption of conformation by D-amino acid D-Leu (133) or Pro (129) substitution resulted in very low activities as did substitution with basic residues (Lys, Arg) (see Ref.129). Tolerance of the large, bulky substitution of Trp for Leu7 was recently demonstrated by the high LH-releasing activity (110% in sheep pituitary cells) and receptor binding (37% for sheep and 230% for rat) for this analog (121, 122). The original proposal of a type-II ß turn conformation of GnRH also envisaged a hydrogen bond between the C=0 of Ser and NH of Leu. However, substitution of Leu7 with N-Me-Leu, which would eliminate this H bond, did not reduce activity (155a).

Arg8.
Comparative activities of vertebrate GnRHs had indicated that Arg8 is required for high-affinity binding to mammalian receptors (91, 105, 119, 120, 121). A number of early studies had shown that D-Arg, Gln, Leu, Orn, His, diaminobutyryl, and Cit substitution for Arg8 results in a substantial decline in activity (1–6%) while homoArg, Narg, and Lys retained good activity (10–20%) (92, 119, 129, 130, 133, 156, 157, 158, 159, 160, 161). A systematic study on the LH-releasing activities from sheep pituitary cells of Gln-, Ser-, Tyr-, Phe-, Glu-, His-, Leu-, Lys-, Ile-, and Trp-substituted analogs confirms the requirement of a basic amino acid in position 8 for high activity (92, 93). Since receptor-binding is correlated with LH-releasing activity in all position 8-substituted analogs studied, it appears that the role of Arg8 may be in receptor binding. However, as noted above, analogs with neutral amino acid substitutions display improved efficacy. Two hypotheses may be invoked to explain the basis of the higher affinities of Arg8-containing GnRHs. An ionic interaction of Arg8 with one or more negatively charged residues, either an amino acid side-chain (162) or a polysaccharide sialic acid residue (163) in the receptor, were proposed. An alternative or additional possibility was that the side chain of Arg8 affects the structure of the ligand, stabilizing the active conformation of GnRH by hydrogen bonding with the side chains of His2 and Tyr5 (159, 160). Low pK values were measured for His2 and Tyr5 in GnRH, and it was suggested that the more acidic nature of these amino acid side chains was due to their proximity to the cationic side chain of Arg8 (160). GnRH analogs with neutral substitutions, Gln and {omega}-nitro-Arg (159), in position 8, exhibited normal pK values for His2 and Tyr5 and extended titration ranges. These results were interpreted as indicating a decreased interaction of the His2 and Tyr5 side chains with the neutral substituents in position 8 (161). The decreased side-chain interaction was proposed to decrease stabilization of the bioactive conformation and thus cause the lower bioactivity in the neutral GnRH analogs. Based on these findings, a folded conformation of GnRH was proposed (159, 160) similar to models of GnRH that were based on energy minimization and database sequence comparison (137, 164, 165, 166, 167).

The role of Arg8 in determining the preferred conformation of GnRH and in receptor interaction is explored in detail in the following sections. In recent mutagenesis studies, an acidic amino acid residue in extracellular loop III was shown to convey specificity for Arg8 such that mutation to an isosteric amide resulted in the loss of the preferential binding of Arg8 GnRH compared with GnRH with a neutral amino acid in this position (see Section V). The requirement for Arg in position 8 and the acidic residue in the receptor is obviated if the GnRH structure is constrained by incorporation of a D-amino acid in position 6 (see Section V). A recent reexploration of the role of Arg8 in antagonists concluded that this residue may be significant for receptor binding (168) while the demonstration in another study that (Orn(2,4-NAPS0)8 GnRH cross-linked with the receptor (169) was interpreted to support the proposal that Arg8 interacts with receptor moieties.

D. Conclusions from peptide structure-activity data
We have attempted to identify individual residues in GnRH that are involved in receptor binding and receptor activation, as this information is critical in undertaking receptor mutagenesis studies directed at defining ligand contact sites. Although thousands of GnRH analogs have been synthesized and biologically characterized, the complexity of most analogs and the predominance of in vivo testing have complicated the task of clearly identifying the roles of individual amino acids in ligand conformation and in receptor binding and activation. Although a more controlled and systematic examination of the functions of GnRH residues using expressed cloned receptors has begun (see Section V), relatively few analogs have been studied to date. Nevertheless it is possible to generate working hypotheses about ligand requirements of the mammalian pituitary GnRH receptor for binding and activation (Fig. 8Go). The following generalizations may be proffered:



View larger version (31K):
[in this window]
[in a new window]
 
Figure 8. Schematic of GnRH summarizing functional properties of individual amino acid residues.

 
1. It is evident that no single residue is crucial for GnRH activity.

2. The NH2- and COOH-terminal domains are the most important in receptor binding and activation.

3. Although both domains are involved in receptor binding, residues in the NH2-terminal domain are predominantly responsible for receptor activation.

4. The only residues for which good evidence exists for a role in receptor activation are His2 and Trp3, but pGlu1 may also be involved.

5. Substitution of residues outside of the NH2-terminal domain can affect receptor activation, possibly through effects on the conformation that change the presentation of activating residues, or through restrictions in dynamic ligand conformation changes that occur on binding to the receptor.

6. Nonconserved residues of the central domain are less critical, but Arg8 is required for high-affinity binding to the mammalian receptor. However, the requirement for Arg8 may be obviated in conformationally constrained analogs with D-amino acids in position 6, and also when His5 is present as in chicken GnRH II.

7. The achiral Gly or D-amino acids are required in position 6, presumably to allow assumption of the folded active conformation.

8. Nonmammalian GnRH receptors have different requirements for the nonconserved residues in the central domain. Examples include the lack of requirement for a basic residue in position 8 and the nonacceptance of His5 substitution when Arg8 is present. Nonmammalian GnRH receptors also tend to be less dependent on conformational constraint, and D-amino acid substitution may not enhance activity to the same degree as in mammalian receptors.


    IV. Structure and Conformation of GnRH and Its Analogs
 Top
 Abstract
 I. Introduction
 II. Amino Acid Sequences...
 III. Structure-Activity...
 IV. Structure and Conformation...
 V. Functional Structure of...
 VI. Conclusions
 References
 
A. Early studies of GnRH conformation
The structure-activity data reviewed in the previous section provide important hypotheses about the role of specific groups in GnRH binding and receptor activation. However, it is becoming increasingly likely that the elucidation of structure-activity relations for GnRH, and of the molecular basis for agonist and antagonist properties at the GnRH receptor, will not be possible at the level of the ligands alone (170). To understand the action of GnRH and its congeners from a structural perspective, it may become necessary to gain an understanding of the structural and dynamic properties of the ligands, as well as of the ligand/receptor complexes. Such insights are sought from experimental explorations of structure-function relations, as well as with novel computational methods that focus on the structural properties of both the receptor and the ligands.

The inherent flexibility of the hormone decapeptide makes it likely that interactions with various sites in the receptor-binding pocket will affect the conformation of any GnRH analog and reduce the ability to define a single "biologically relevant" conformation for the isolated peptide. Moreover, the pharmacophoric patterns of different peptides depend on the specific residues available for interaction with the various receptor sites, even if their conformations are the same. Consequently, the structural determinants for action on the GnRH receptor will have to be sought from a comprehensive characterization of all the conformations accessible to the peptides under given conditions of temperature and environment, i.e. the conformational space of these peptides, as well as of the three-dimensional pattern of the pharmaco-phoric elements that their amino acids present to the receptor. If the ability to adopt certain conformations determines the receptor activity of peptides with similar pharmacophoric elements, then differences in their conformational spaces can reveal the conformations required for receptor interactions. The conformational properties of the most active peptides can thus serve to define the spatial and dynamic requirements for optimal interaction with the receptor. A useful ranking of structure-activity characteristics can be constructed on this basis, provided that peptides with pharmacologically distinct activities such as agonism and antagonism are differentiated. For these reasons, it is not realistic to expect a full understanding to emerge entirely from structure-activity data obtained from probing the activities of various synthetic analogs without specific analysis of their conformational properties. Rather, the mechanistic insights are more likely to emerge when the powerful approaches offered by current experimental and computational methods for conformational analysis (for reviews see Refs. 171 and 172; also Refs. 166, 173, and 174) are applied to the exploration of peptide structure and design.

Pioneering efforts were undertaken to achieve such a comprehensive exploration of the conformation of GnRH and its active analogs by computational methods (164, 165, 175). The impetus for such studies continues to be the assumption that if the peptide conformation recognized by the receptor (i.e. the bioactive conformation) corresponds to the most abundant form of the peptide in solution, then the peptide will have high affinity for the receptor. Because the most abundant conformers in solution are those corresponding to the lowest free energy, the computational approaches concentrated on the calculation of the conformational energies of the peptides. The early studies (164, 165, 175) identified low-energy conformations of GnRH that were considered to occur also in solution, although it was not possible at the time to account for the effects of aqueous solvation. In spite of the significant limitations of the methods for energy-based evaluation of peptide conformations (for a review of methods see Ref.176) that were available at the time for studies of peptide molecules of the size of GnRH, the early studies identified the central characteristic of the bioactive conformation of GnRH, the ß-bend involving the Tyr-Gly-Leu-Arg in positions 5–8 (see Fig. 9Go).



View larger version (19K):
[in this window]
[in a new window]
 
Figure 9. View of hinged conformer of GnRH obtained using empirical energy calculations. [Reprinted with permission of the publisher from F. A. Momany: J Am Chem Soc 98:2990–2996 (165). ©1976 American Chemical Society.]

 
B. Integrated computational and experimental studies
Subsequently, the most incisive studies of structure-activity relationships used in the design of GnRH analogs combined the computational conformational studies with a variety of experimental (mostly spectroscopic) approaches to characterize the structural features of the peptides (for illustrations and reviews see Refs. 177–179). Such studies often took advantage of cyclization as a means of restricting the conformational freedom of the peptide molecules to reduce the complexity of the problem. The initial inferences from conformational analysis of GnRH that residues Tyr5-to-Arg8 may form a ß-type turn supported the efforts to produce cyclic analogs in the search of high-affinity ligands for the GnRH receptor. Much attention was devoted to the conformational properties of these cyclic analogs in computational studies. It was quickly recognized that, even in these cyclic analogs, the intramolecular motions were complex, and conformational variability persisted. The overall conformational properties of the cyclic analogs were found to be generally in good agreement with the results from energy-based calculations and with the information on the structure and dynamics of the analogs obtained from nuclear magnetic resonance (NMR) (173, 179). The points of agreement include the all-trans nature of the peptide bonds in the major conformational families of the analogs and the ß-type turn structure between residues 4–8 (177) (Fig. 10Go). However, the computational simulations of molecular conformation and dynamics indicated the large variability in the structural options of the uncyclized 1–3 fragment, as well as the fact that several different conformations with equivalent energies and geometrical features compatible with the NMR data, can be adopted even by the constrained dicyclic compound (177).



View larger version (17K):
[in this window]
[in a new window]
 
Figure 10. Schematic representation of dicyclic GnRH analog studied using two-dimensional nuclear magnetic resonance and molecular dynamics simulation. [Reprinted with permission of the publisher from R. J. Bienstock et al.: J Med Chem 36:3265–3273, 1993 (177).]

 
The insights from computational studies were confirmed by results from NMR spectroscopy (e.g. Refs. 177–179 and references therein), but uncertainty persisted with respect to the exact location of the ß-type turn and to the spatial orientation of the side chains assumed to be involved in affinity-determining interactions with receptor sites (180). Thus, cyclo(1, 2, 3, 4, 5, 6, 7, 8, 9, 10) analogs were compared with 4–9 and 4–10 bridged analogs (170) and more recently with the dicyclic analogs (4–10; 5–8) designed to incorporate the conformational constraints of both classes of cyclic analogs (177). It is quite clear from these studies that the multiple conformational forms exist even for the most constrained analogs (170, 178) and that these analogs must be studied with detailed conformational approaches. To be most useful, such conformational studies must not only identify the lowest energy conformations, but must be able to provide reliable information on the probability distribution of the various conformational families, i.e. their relative abundance in solution (174, 181).

Early attempts to decompose the structural determinants for activity of cyclized GnRH analogs into contributions from the length of the bridge and the orientation of certain functional groups (e.g. amide bonds considered to be involved in direct interactions with receptor sites) were not successful (170). A main reason is the residual flexibility of the cyclized GnRH analogs, which lose only a portion of their conformational freedom. This insight (e.g. see Refs. 170 and 178) led to the suggestion that a complete structural characterization, rather than the mere identification of minimal energy conformations, will be necessary for both linear and cyclic analogs (170).

C. Exploration of the entire conformation space of GnRH analogs
The results from experimental and computational studies of the structure and conformation of GnRH analogs emphasize the importance of a complete exploration of the conformational space of the peptides (176). Whether the analogs are conformationally constrained or linear, the complete structural characterization is necessary before a reliable consensus on the structural features most important for receptor interaction can be reached (174). Current methods (174, 181) make possible such extensive explorations of both the conformational and dynamic properties of the peptides, offering the ability to explore the entire conformational space of decapeptides such as GnRH and its analogs (174). Specifically, the application of a recently developed technique of Conformational Memories (182) to the study of GnRH conformational properties illustrates the first complete exploration of the entire conformational space of the peptide using a method of simulation that includes a satisfactory model for the aqueous solvent.

The novel method overcomes some of the shortcomings of modern molecular dynamics approaches to the study of the peptide hormone: although the molecular dynamics techniques are useful for their ability to describe the molecular motions of the peptide in short time scales, they are unable to explore all the conformational states of the peptide in solution, and hence are not able to characterize their relative abundance. In contrast, the Conformational Memories method utilizes a two-stage process of computation to map, and then characterize, the conformational space of a flexible molecule (174). In the first, exploratory stage, repeated runs of the Monte Carlo method (183) combined with simulated annealing (184) are carried out to map the entire conformational space of a flexible molecule by heating it to very high temperatures and cooling it slowly to body temperature. Once the Conformational Memories are established, the method proceeds to a new Monte Carlo search of the conformations of the peptide, performed at 310K and sampling only from the populated regions. Because only about 50% of the torsional space of the 35 bonds of GnRH is populated at 310K, the two-tiered approach reduces by many orders of magnitude the conformational space that must be explored in this second phase (174). The configurations sampled from the Conformational Memory can be any part of the populated space of dihedral angles defining the conformations of the peptide. Consequently, the notion of a barrier restricting access to any part of the conformational space is eliminated in this procedure without approximations.

1. Conformational families of GnRH. In the application of the conformational memories approach to GnRH, the second step of the procedure involved 500,000 steps (174). Structures of the peptide obtained from the run were clustered in conformational families, resulting in the five basic structures depicted in Fig. 11Go. Notably, families of conformations having a ß-turn between residues 5–8 occur in GnRH with a frequency of approximately 70%. A distribution showing a superimposition of 70 of these structures is illustrated in Fig. 12Go; the ß-type turn common to all the structures in this family is clearly evident. In contrast, families that have an extended backbone occur with a frequency of about 5%. The distribution of side-chain orientations of Arg8 in all conformational families was found to be wider than that of any other residues in GnRH.



View larger version (22K):
[in this window]
[in a new window]
 
Figure 11. Backbone trace of representative members of the conformational families of GnRH obtained from Conformational Memories.

 


View larger version (112K):
[in this window]
[in a new window]
 
Figure 12. Superimposition of structures that make up the major conformational family of GnRH obtained from Conformational Memories. GnRH is colored in purple, with Arg8 colored in red. The ß-turn common to all structures is evident.

 
2. Conformational families of Lys8-GnRH. The conformational properties of the Lys8 analog of GnRH that served to explore the role of Arg8 in the receptor interactions of GnRH (21, 121) were compared with the conformational profile of GnRH. In contrast to the wild type hormone, the major conformational family of the Lys8-GnRH congener was found to have an extended backbone, while the ß-turn conformation exists as a very minor family (174). A backbone trace of a representative of each family is shown in Fig. 11Go. The family of conformations with an extended backbone has an abundance of more than 70%, while the ß-type turn conformation of Lys8-GnRH, which is virtually identical to the major conformational family of the GnRH, has a probability of only about 3%. A distribution of the members of the predominant Lys8-GnRH family superimposed upon each other is shown in Fig. 13Go, with the entire molecule shown in red, except for Lys8, which is colored green. Because Lys8-GnRH is a low affinity agonist for the GnRH receptor, adoption of a large population of ß-type turn con