Introduction

Top Abstract Introduction Experimental Procedures Results Discussion References

Corticotropin-releasing factor (CRF) is a primary regulator of the body's stress response (for review, see Vale et al., 1997) and plays an important role in the development of several psychiatric and eating disorders, including anxiety, major depression, and anorexia nervosa (Behan et al., 1996; Arborelius et al., 1999; Holsboer, 1999). CRF and its structurally related analogs mammalian urocortin (UCN), amphibian sauvagine, and fish urotensin I mediate their effects through two receptors, both of which belong to the class B subfamily of G protein-coupled receptors (GPCRs) (Vale et al., 1997). The principal subtypes of the CRF receptor, type 1 (CRF₁) and type 2 (CRF₂), are 70% identical in amino acid sequence, and have been cloned in several species (Speiss et al., 1998; Lovejoy and Balment, 1999). Several alternatively spliced forms of CRF₁ and CRF₂ have been described, including CRF₂, CRF₂, and CRF₂ (for review, see Kilpatrick et al., 1999).

Despite their high degree of sequence homology, the CRF₁ and CRF₂ receptors differ markedly in their tissue distribution and pharmacological properties. Mammalian CRF₁ has been shown to be a promiscuous receptor that binds human CRF (hCRF), UCN, urotensin I, and sauvagine with comparable affinity. These peptides are also equipotent in their ability to stimulate cAMP production in cells expressing recombinant CRF₁ (Vaughan et al., 1995; Donaldson et al., 1996; Dautzenberg et al., 1997, 1999; Palchaudhuri et al., 1998). Until now, only CRF₁ from Xenopus laevis (xCRF₁) has been shown to be a ligand-selective type 1 receptor (Dautzenberg et al., 1997). xCRF₁ binds hCRF, Xenopus CRF (xCRF), urotensin I, and rat UCN (rUCN) with higher affinity than the structurally related peptide analogs ovine CRF (oCRF) and sauvagine (Dautzenberg et al., 1997, 1998). CRF₂ receptors exhibit a different ligand specificity than CRF₁ receptors. hCRF, oCRF, and xCRF bind to CRF₂ receptors with significantly lower affinity than do UCN, urotensin I, and sauvagine (Donaldson et al., 1996; Dautzenberg et al., 1997, 2000; Ardati et al., 1999; Palchaudhuri et al., 1999). However, the N-terminally truncated sauvagine analog [D-Phe¹¹,His¹²]sauvagine_(11-40) also called antisauvagine-30 (aSVG), has recently been reported to be a highly selective mouse CRF₂-specific antagonist (Rühmann et al., 1998), whereas other commonly used peptide antagonists, -helical CRF_(9-41) (-hel CRF) and astressin (AST) show little or no preference for CRF₁ or CRF₂ receptors (Rühmann et al., 1998).

The differences in ligand-binding specificity between CRF₁ and CRF₂ have been largely attributed to the relative importance of different domains of these receptors in ligand engagement. The N-terminal extracellular domain (EC1) of mammalian and amphibian CRF₁ confers the major determinants for high-affinity ligand binding and specificity (Dautzenberg et al., 1998, 1999; Perrin et al., 1998; Wille et al., 1999). In contrast, the ligand-recognition domains of hCRF₂ have been mapped to the second extracellular domain (EC2) and to the interface between EC3 and transmembrane helix 5 (Liaw et al., 1997). Recently, we reported that ligand binding to the Xenopus CRF₂ receptor (xCRF₂) is also dependent on analogous domains, suggesting that the ligand-binding pocket of hCRF₂ and xCRF₂ are probably very similar (Dautzenberg et al., 1999). However, radioligand binding data suggest some differences in ligand specificity between the human and Xenopus CRF₂ receptors, indicating that additional motifs may be involved in defining full ligand specificity.

To more carefully evaluate the ligand specificity of mammalian and amphibian CRF receptors, we have measured binding of four different ¹²⁵I-labeled CRF analogs, the agonists CRF and sauvagine, and the antagonists AST and aSVG to human and Xenopus CRF₁ and CRF₂. We also measured the ability of many unlabeled CRF receptor agonists and antagonists to compete with these four radioligands for binding. Our results demonstrate distinct binding preferences at hCRF₁, hCRF₂, xCRF₁, and xCRF₂ that suggest distinct functions for mammalian and amphibian CRF receptors.

	Experimental Procedures

Top Abstract Introduction Experimental Procedures Results Discussion References

Materials, Peptides, and Reagents. All cell culture media and reagents were purchased from Life Technologies (Basel, Switzerland). Aprotinin was obtained from Roche Molecular Biochemicals (Mannheim, Germany). The CRF peptides (purity >95%) were obtained from Bachem Corporation (Bubendorf, Switzerland), and antisauvagine-30 (purity >95%) was kindly provided by Dr. Arvind Patel (Amersham Pharmacia Biotech, Little Chalfont, UK).

Radiochemicals. ¹²⁵I-Tyr⁰-hCRF (¹²⁵I-CRF) and ¹²⁵I-AST (¹²⁵I-AST) (2200 Ci/mmol each) were purchased from NEN (Boston, MA). ¹²⁵I-Tyr⁰-sauvagine (¹²⁵I-sauvagine) and ¹²⁵I-aSVG (2000 Ci/mmol each) were obtained from Amersham Pharmacia Biotech.

Membrane Preparations and Radioreceptor Binding Assays. Membranes were prepared from HEK293 cells, stably expressing either hCRF₁, hCRF₂, xCRF₁, and xCRF₂ by previously described methods (Rühmann et al., 1996; Hauger et al., 1997; Dautzenberg et al., 2000).

2

2

cAMP Assays. HEK293 cells, stably expressing hCRF₁, xCRF₁, hCRF₂, or xCRF₂ were plated at 10⁴ cells/well in 96-well dishes. Stimulation was carried out for 30 min (37°C, 5% CO₂) with the indicated peptides. Measurement of cellular cAMP was performed as previously described (Dautzenberg et al., 1997, 1999). The data were analyzed by one-way ANOVA and significance between groups was determined by post hoc analysis using Dunnett's test.

	Results

Top Abstract Introduction Experimental Procedures Results Discussion References

Saturation Binding of ¹²⁵I-Labeled CRF Agonists and Antagonists to hCRF₁, xCRF₁, hCRF₂, and xCRF₂. Preliminary binding experiments revealed differences in the ability of hCRF₁, xCRF₁, hCRF₂, and xCRF₂ to bind the various ¹²⁵I-labeled CRF analogs (Table 1). For hCRF₁, strong binding was observed with ¹²⁵I-CRF, ¹²⁵I-sauvagine, and ¹²⁵I-AST but not with ¹²⁵I-aSVG. The xCRF₁ receptor, like hCRF₁, also bound ¹²⁵I-CRF and ¹²⁵I-AST and not ¹²⁵I-aSVG; however, it differed in that it also did not bind ¹²⁵I-sauvagine. In contrast to hCRF₁ and xCRF₁, the hCRF₂ and xCRF₂ receptors bound ¹²⁵I-CRF poorly, whereas the other three ¹²⁵I-labeled CRF analogs bound with high affinity. These data are consistent with previous reports on the ligand specificity of type 1 and type 2 receptors (Donaldson et al., 1996; Dautzenberg et al., 1997; Rühmann et al., 1998).

Binding of CRF Peptides to the hCRF₁ Receptor. We compared the ability of unlabeled agonists and unlabeled antagonists to compete for binding of either radiolabeled agonists or radiolabeled antagonists to the hCRF₁ receptor. Competitive binding studies with ¹²⁵I-CRF, ¹²⁵I-sauvagine, and ¹²⁵I-AST as radioligands were conducted with the CRF agonists hCRF, oCRF, rUCN, human UCN (hUCN), sauvagine, and urotensin I and with the peptide antagonists -hel CRF, AST, and aSVG. All CRF agonists efficiently competed with radiolabeled agonists (¹²⁵I-CRF and ¹²⁵I-sauvagine) for binding to the hCRF₁ receptor, yielding K_i values in the subnanomolar (rUCN, hUCN, urotensin I, and sauvagine) to low nanomolar (hCRF, oCRF) range. In contrast, the K_i values of the antagonists -hel CRF and AST to compete for binding of the ¹²⁵I-labeled agonists were markedly higher (Table 2). In a corollary experiment, we found that the ability of the agonists hCRF, oCRF, urotensin I, and especially sauvagine to compete for binding of the radiolabeled antagonist ¹²⁵I-AST was strongly impaired compared with the results with displacement of radiolabeled agonists (Table 2). Remarkably, this was not the case for hUCN and rUCN. Although hUCN and rUCN were very potent competitors with the radiolabeled agonists ¹²⁵I-CRF and ¹²⁵I-sauvagine, they were also, in contrast to the other unlabeled agonists, potent competitors when the antagonist ¹²⁵I-AST was used as radioligand (Table 2). In a final observation, the ability of unlabeled antagonists to compete with the antagonist ¹²⁵I-AST revealed few differences compared with competition with the labeled agonists. The binding affinity of -hel CRF to the hCRF₁ receptor was similar with all radioligands (agonists and antagonists), whereas unlabeled AST was slightly more potent in competing with the antagonist ¹²⁵I-AST than with the agonists ¹²⁵I-CRF and ¹²⁵I-sauvagine (Table 2).

Binding of CRF Peptides to the hCRF₂ Receptor. The nine peptides used in this study displayed a similar rank order binding profile at hCRF₂ using the agonists ¹²⁵I-CRF and ¹²⁵I-sauvagine. Rat UCN, hUCN, urotensin I, sauvagine, and the antagonist AST competed for ¹²⁵I-CRF and ¹²⁵I-sauvagine binding to hCRF₂ with high affinity, whereas -hel CRF was bound with ~10-fold lower affinity (Table 3). Finally, hCRF and oCRF displayed very low-affinity binding to hCRF₂ (Table 3). More pronounced differences in the binding affinity of the various CRF agonists to hCRF₂ were observed when the two antagonists ¹²⁵I-AST and ¹²⁵I-aSVG were used instead of the radiolabeled agonists CRF and sauvagine. Sauvagine and urotensin I exhibited a 25- to 50-fold decrease in potency in competition with the radiolabeled antagonists compared with the labeled agonists (Table 3). hCRF and oCRF, already poor ligands for hCRF₂, as determined by agonist competition assays, showed a further decrease in potency to compete with antagonist ligands. However, similar to the observations with the hCRF₁ receptor, we found that hUCN and rUCN retained high-affinity binding to hCRF₂ regardless of whether agonist or antagonist radioligands were used (Table 3). Additionally, the CRF antagonists -hel CRF, AST (Fig. 1), and aSVG displayed similar affinities to hCRF₂ with all four radioligands (Table 3).

View this table: [in this window] [in a new window]   TABLE 3 Competitive binding of various CRF analogs to hCRF2 The data are means ± S.E. from at least three different binding experiments performed in triplicate.

View larger version (15K): [in this window] [in a new window]   Fig. 1.   Competition binding of 125I-CRF (A) and 125I-AST (B) to hCRF1, hCRF2, xCRF1, and xCRF2 by AST. Competitive binding was performed using 125I-CRF (A) or 125I-AST (B) and increasing concentrations (0.01 nM-10 µM) of unlabeled AST. Data represent triplicates from one representative experiment performed three to four times.

Binding of CRF Peptides to the xCRF₁ Receptor. Because the xCRF₁ receptor is not able to bind sauvagine and aSVG, we performed competition binding assays with only ¹²⁵I-CRF and ¹²⁵I-AST. With the agonist ¹²⁵I-CRF as the displacable ligand, we observed the following rank order binding potency: hCRF ~ hUCN ~ rUCN ~ urotensin I > oCRF > -hel CRF > sauvagine > AST (Table 4). With the antagonist ¹²⁵I-AST as radioligand, the affinity of most of the agonists significantly decreased, which was consistent with results we found for binding to hCRF₁ and hCRF₂ receptors. Also consistent with our observations with hCRF₁ and hCRF₂ receptors, we found that binding of hUCN and rUCN decreased minimally when using labeled antagonist compared with labeled agonist (Table 4). In contrast to the two UCN peptides, the potency of hCRF decreased 27-fold (Table 4) when competing with the radiolabeled antagonist. Although the potency of urotensin I binding also decreased by a factor of 30, it is important to note that this peptide occupied only a minor fraction (~40%) of xCRF₁ binding sites when using ¹²⁵I-AST as radioligand (Table 4). Finally, sauvagine displayed only micromolar affinity to xCRF₁ with ¹²⁵I-AST as radioligand (Table 4).

Binding of CRF Peptides to the xCRF₂ Receptor. As with hCRF₂, all four of the radioligands ¹²⁵I-CRF, ¹²⁵I-sauvagine, ¹²⁵I-AST, and ¹²⁵I-aSVG, bound to the receptor and were used in competition binding assays with the nine unlabeled peptides. Each of the four radioligands showed different affinities and rank order binding profiles for the series of competitors. The rUCN and hUCN peptides displayed high-affinity binding (1 nM) to xCRF₂, regardless of which of the four ¹²⁵I-labeled CRF analogs was used (Table 5). In contrast, sauvagine was a high-affinity competitor with labeled agonists (¹²⁵I-CRF and ¹²⁵I-sauvagine) to xCRF₂, but showed about 4- and 8-fold decreased binding affinity with the labeled antagonists ¹²⁵I-aSVG and ¹²⁵I-AST as ligands, respectively (Table 5). We observed even more pronounced differences in the affinity of urotensin I for xCRF₂ receptors. With agonist ¹²⁵I-CRF as radioligand, urotensin I displayed subnanomolar affinity but was 14-fold less potent in competing for binding with the second agonist ¹²⁵I-sauvagine. The measured affinity for urotensin I binding decreased an additional 3.5- to 5.5-fold when the two antagonists ¹²⁵I-aSVG and ¹²⁵I-AST were used as radioligands (Table 5). hCRF also showed a similar profile in binding affinity. Although hCRF had moderate affinity (~20 nM) to xCRF₂ with ¹²⁵I-CRF as ligand, its affinity decreased by about 6-fold with the second agonist ¹²⁵I-sauvagine. However, in contrast to urotensin I, the affinity for hCRF did not decrease further when the antagonists ¹²⁵I-AST and ¹²⁵I-aSVG were used (Table 5). The binding of oCRF, although poor with all radioligands, showed a similar rank order binding profile as hCRF (Table 5).

Stimulation of cAMP Production in HEK293 Cells Expressing Human and Xenopus CRF₁ and CRF₂ Receptors. We generated stable HEK293 cell lines expressing hCRF₁, hCRF₂, xCRF₁, or xCRF₂ receptors. All transfected lines produced maximal increases in cAMP that differed by less than 2-fold in response to stimulation with fully activating doses of nonselective agonists such as urotensin I. The values varied from a low of 15.0 ± 0.5 pmol of cAMP/10⁴ cells for xCRF₂-producing HEK cells to a high of 26.2 ± 0.7 pmol/10⁴ cells for xCRF₁-producing cells. Similar profiles were obtained with maximally stimulating concentrations of all agonists used in this study (data not shown).

2

2

View this table: [in this window] [in a new window]   TABLE 6 Stimulation of cAMP production in HEK293 cells stably transfected with cDNAs coding for hCRF1, hCRF2, xCRF1, or xCRF2 receptors The data are the mean ± S.E. of four stimulations.

View larger version (25K): [in this window] [in a new window]   Fig. 2.   Stimulation of cAMP production in HEK293 cells stably expressing hCRF1, hCRF2, xCRF1, or xCRF2 receptors by hCRF, rUCN, and sauvagine. The stimulations were performed as described under Experimental Procedures. Data are representative of four independent stimulations performed in duplicate. For rUCN the Emax values were significantly lower (p < 0.0001) at each receptor than for the other agonists.

	Discussion

Top Abstract Introduction Experimental Procedures Results Discussion References

Mammalian and amphibian CRF₁ and CRF₂ receptors show distinct substrate specificities due to functional and species differences (Hauger and Dautzenberg, 1999). To gain more insight into ligand-receptor interactions, we compared binding profiles of ¹²⁵I-labeled CRF receptor agonists (hCRF and sauvagine) and antagonists (AST and aSVG) to human and Xenopus CRF₁ and CRF₂ receptors. Binding of radioligands to hCRF₁, hCRF₂, xCRF₁, and xCRF₂ receptors revealed marked differences in ligand preferences with only ¹²⁵I-AST being a high-affinity ligand for all receptors. The hCRF₁ and xCRF₁ receptors showed similar high-affinity binding to ¹²⁵I-CRF and ¹²⁵I-AST, however, only hCRF₁ bound ¹²⁵I-sauvagine with high-affinity, whereas xCRF₁ failed to bind. In contrast, human and Xenopus CRF₂ receptors showed similar binding profiles. hCRF₂ and xCRF₂ showed high-affinity binding to ¹²⁵I-sauvagine, ¹²⁵I-AST, and ¹²⁵I-aSVG, and low-affinity binding to ¹²⁵I-CRF. These data suggest that CRF₂ receptor specificity is more highly conserved between mammals and amphibians than is CRF₁ specificity. Of particular interest were the binding characteristics of aSVG. aSVG has been reported as a superior tool for pharmacological separation of CRF₁ and CRF₂ receptors (Rühmann et al., 1998). In agreement with this observation, we found that radiolabeled ¹²⁵I-aSVG bound both hCRF₂ and xCRF₂ but not hCRF₁ or xCRF₁. More detailed pharmacological characterization of ¹²⁵I-aSVG has been presented previously (Higelin et al., 2000). We also detected several general characteristics of ligand-CRF receptor interactions. For hCRF₁, hCRF₂, and xCRF₁, the K_d values for all high-affinity radioligands were similar at each receptor. In contrast, xCRF₂ showed a rank order binding preference for ¹²⁵I-AST, followed by ¹²⁵I-sauvagine and ¹²⁵I-aSVG. We also found that antagonists generally labeled significantly more binding sites than did agonists. This phenomenon has been described previously for hCRF₁, rat CRF₂, and mouse CRF₂ (Perrin et al., 1999) and appears due to the existence of a large percentage of G protein-uncoupled receptors in recombinant expression systems. Antagonists fail to discriminate between coupled and uncoupled receptors, whereas agonists preferentially bind G protein-coupled receptors. Thus, antagonists detect more binding sites than do agonists.

The competitive binding studies yielded several important observations. First, binding of agonists to human CRF receptors revealed higher apparent affinities when the competed radioligand was also an agonist, but produced rightward shifts in dose response with antagonist radioligands. Important exceptions were the UCN peptides, rUCN, and hUCN, which showed high affinity for all receptors in displacement of both ¹²⁵I-labeled agonists and antagonists. Second, the affinities of unlabeled antagonists for human CRF receptors were independent of the radioligand. Third, the rank order binding preferences of agonists and antagonists for xCRF₁ and xCRF₂ receptors varied considerably in a manner dependent on the competed radioligand.

We commonly observed a rightward shift of the unlabeled agonist competition binding curves to hCRF₁ and hCRF₂ in the presence of the antagonists ¹²⁵I-AST or ¹²⁵I-aSVG. This phenomenon has been observed previously and is a general characteristic of agonist binding to GPCRs using radiolabeled antagonists as competed ligands (Sleight et al., 1996). Although binding of antagonists is independent of the G protein-coupled state of receptors, agonists preferentially bind only coupled receptors (Kenakin, 1997). In recombinant systems, a large proportion of receptors is not coupled to G proteins (Kenakin, 1997) and thus, agonists compete less efficiently for binding in the presence of a radiolabeled antagonist than a radiolabeled agonist. However, this phenomenon may reflect not simply an artifact of recombinant systems but a physiologically important function. Cells in vivo may express a reserve population of G protein-uncoupled receptors. For example, in rat cerebellum, a greater number of CRF binding sites were labeled by a CRF receptor antagonist than by an agonist (Perrin et al., 1999). This suggested the presence of a population of uncoupled CRF receptors in this tissue.

We observed an interesting exception to the trend in decreased potency of agonists in competing with radiolabeled antagonists versus other agonists. The UCN analogs rUCN and hUCN, in contrast to other agonists, competed almost equally efficiently with both radiolabeled agonists and antagonists. This observation is consistent with previous data (Perrin et al., 1999; Rühmann et al., 1999) and likely accounts for some intrinsic antagonistic properties of UCN. It has been shown that UCN binding to mammalian CRF₁ and CRF₂ receptors is relatively insensitive to GTP analogs (Perrin et al., 1999), which normally uncouple GPCRs from their G proteins (Kenakin, 1997; Higelin et al., 2000). Additional support for the possibility that hUCN and rUCN have intrinsic antagonist properties comes from our functional studies. We observed that UCN peptides were significantly less potent in stimulation of cAMP production with human and amphibian CRF₁ and CRF₂ receptors than would have been expected from the K_i values measured in binding experiments. In contrast, all other agonists had lower EC₅₀ than K_i values at all receptors. Additionally, hUCN and rUCN stimulated cAMP production in hCRF₁-, hCRF₂-, xCRF₁-, and xCRF₂-transfected HEK cells to a significantly lesser extent than other agonists. Taken together, this suggests that UCN is unable to fully activate CRF₁ and CRF₂ receptors. This conclusion is also supported by in vivo experiments where UCN was shown to be less efficacious than CRF in stimulating behavioral responses after injection into mouse brain regions where CRF₂ receptors are predominantly expressed (Radulovic et al., 1999). UCN is thought to be the endogenous ligand to CRF₂ receptors (Vale et al., 1997). However, in light of its inability to mediate full agonism in vivo, it is tempting to speculate that another CRF peptide may exist (Bittencourt et al., 1999).

The xCRF₁ and xCRF₂ receptors, unlike their human orthologs, showed different rank order binding profiles in competition binding assays. With ¹²⁵I-CRF as radioligand, xCRF₁ receptors bound hCRF, hUCN, rUCN, and urotensin I with high affinity and oCRF and sauvagine with low affinity (Dautzenberg et al., 1997). Competition with the antagonist ¹²⁵I-AST resulted in a rightward shift in the agonist dose-response curves compared with competition with ¹²⁵I-CRF. However, the affinities of hUCN and rUCN shifted minimally in competition with radiolabeled antagonists, consistent with a potential antagonist character of UCN. For the unlabeled antagonists, -hel CRF binding to xCRF₁ was similar with both agonist and antagonist radioligands and aSVG was inactive at xCRF₁. The most striking difference was observed with the antagonist AST. Although ¹²⁵I-AST bound with high affinity to xCRF₁ and unlabeled AST efficiently competed for ¹²⁵I-AST binding, unlabeled AST was a poor competitor with the agonist ¹²⁵I-CRF as radioligand. This differential binding of AST at xCRF₁ suggests additional functional differences between mammalian CRF₁ and xCRF₁.

Surprisingly, competition binding studies with xCRF₂ revealed substantial differences to its human counterpart. Only sauvagine displayed the classical profile showing a higher affinity with radiolabeled agonist ligand compared with radiolabeled antagonist. hCRF, oCRF, and urotensin I, in contrast to their binding profile at hCRF₂, showed lower affinity in the presence of ¹²⁵I-sauvagine, ¹²⁵I-AST, and ¹²⁵I-aSVG. Binding of hUCN and rUCN was similar with the other CRF receptors in its insensitivity to the nature of the radioligands. Thus, agonist binding profiles to xCRF₂, in contrast to hCRF₁, hCRF₂, and xCRF₁, were not separated by the agonistic versus antagonistic nature of the respective radioligand, but were distinguished instead, by differences between ¹²⁵I-CRF and other radioligands. ¹²⁵I-CRF seems to bind to xCRF₂ with a somewhat different mechanism than do the other radiolabeled peptides. Because previous pharmacological comparisons of hCRF₂ and xCRF₂ were performed only with the ligand ¹²⁵I-CRF (Dautzenberg et al., 1997, 1999), this study provides the first evidence for differences in ligand binding characteristics between mammalian and amphibian CRF₂ receptors.

Binding of unlabeled antagonists to xCRF₂ revealed additional pharmacological heterogeneity of this receptor. Although binding of -hel CRF and aSVG to the xCRF₂ receptor was similar with ligands ¹²⁵I-CRF, ¹²⁵I-sauvagine, and ¹²⁵I-aSVG, an approximately 8-fold lower affinity was measured with ¹²⁵I-AST. Conversely, AST bound xCRF₂ with subnanomolar affinity when ¹²⁵I-AST was used as radioligand, but the affinity of AST decreased up to 22-fold with the three other radioligands. This contrasted with the binding profiles observed for hCRF₂. Although binding specificity of hCRF₂ and xCRF₂ has been shown to be dependent on extracellular loops other than domain EC1 (Liaw et al., 1997; Dautzenberg et al., 1999, 2000), the regions that likely mediate hCRF₂'s ligand selectivity are not well conserved between the two receptors. In light of these results, which now provide tools that allow pharmacological discrimination of human and Xenopus CRF₂ receptors, it may be possible to identify residues essential for ligand binding by comparative site-directed mutagenesis studies.

In conclusion, we have revealed markedly different binding specificities between human and Xenopus CRF₁ and CRF₂ receptors by radioligand binding and competition assays. We found that radiolabeled antagonists generally labeled more binding sites than did agonists. We discovered additional evidence to support the possibility that hUCN and rUCN have intrinsic antagonist characteristics. Finally, we identified different binding modes of CRF agonist and antagonist ligands at human and amphibian CRF₁ and CRF₂ receptors. These differences likely reflect variations in the structure of the binding pocket, suggesting that mammalian and amphibian receptors may have evolved similar but distinct functions.