NMR Structure of the First Extracellular Domain of Corticotropin-releasing Factor Receptor 1 (ECD1-CRF-R1) Complexed with a High Affinity Agonist*

The corticotropin-releasing factor (CRF) peptide hormone family members coordinate endocrine, behavioral, autonomic, and metabolic responses to stress and play important roles within the cardiovascular, gastrointestinal, and central nervous systems, among others. The actions of the peptides are mediated by activation of two G-protein-coupled receptors of the B1 family, CRF receptors 1 and 2 (CRF-R1 and CRF-R2α,β). The recently reported three-dimensional structures of the first extracellular domain (ECD1) of both CRF-R1 and CRF-R2β (Pioszak, A. A., Parker, N. R., Suino-Powell, K., and Xu, H. E. (2008) J. Biol. Chem. 283, 32900–32912; Grace, C. R., Perrin, M. H., Gulyas, J., Digruccio, M. R., Cantle, J. P., Rivier, J. E., Vale, W. W., and Riek, R. (2007) Proc. Natl. Acad. Sci. U.S.A. 104, 4858–4863) complexed with peptide antagonists provided a starting point in understanding the binding between CRF ligands and receptors at a molecular level. We now report the three-dimensional NMR structure of the ECD1 of human CRF-R1 complexed with a high affinity agonist, α-helical cyclic CRF. In the structure of the complex, the C-terminal residues (23–41) of α-helical cyclic CRF bind to the ECD1 of CRF-R1 in a helical conformation mainly along the hydrophobic face of the peptide in a manner similar to that of the antagonists in their corresponding ECD1 complex structures. Unique to this study is the observation that complex formation between an agonist and the ECD1-CRF-R1 promotes the helical conformation of the N terminus of the former, important for receptor activation (Gulyas, J., Rivier, C., Perrin, M., Koerber, S. C., Sutton, S., Corrigan, A., Lahrichi, S. L., Craig, A. G., Vale, W., and Rivier, J. (1995) Proc. Natl. Acad. Sci. U.S.A. 92, 10575–10579).

Here, we report the 3D NMR structure of the human ECD1-CRF-R1 (14) complexed with the peptide agonist α-helical cyclic CRF (αhcCRF). The motivation for these studies is the identification of the binding surface involved in the recognition of a high affinity agonist, the subsequent comparison with that previously reported for the antagonists and the determination of reciprocal conformational changes in both the receptor and the agonist.

Protein expression and purification
The purification and labeling of the ECD1-CRF-R1 expressed in E.coli was carried out as described in (15,29).

Synthesis of αhcCRF, with C-13 and N-15
isotopically labeled amino acids The synthesis was performed on a methyl-benzhydrylamine resin using the Fmoc strategy. Purification and characterization used established procedures (see (30,31)). Full details are given in the Supplemental Information section.
Radioreceptor assays Radioreceptor assays for full-length receptors expressed in mammalian cells were carried out as described in (14) and for soluble receptors as described in (32).

NMR Experiments and Analysis
All NMR spectra were recorded at 35 ºC using a Bruker 700 MHz spectrometer equipped with four radio-frequency channels and a triple resonance cryo-probe with shielded z-gradient coil. The NMR samples contained either 0.3 mM 13 C, 15 Nlabeled ECD1-CRF-R1 and an equimolar concentration of unlabeled αhcCRF or 0.3 mM 13 C, 15 N-labeled αhcCRF (labeled uniformly by 15 N and 13 C except for residues T 11 , E 30 and K 33 ) and an equimolar concentration of unlabeled ECD1-CRF-R1 in 10 mM Bis-Tris(HCl)/95% H 2 O/5% D 2 O at pH 6.5. The sample for the 2D [ 1 H, 1 H]-NOESY (Nuclear Overhauser enhancement spectroscopy) was prepared with an equimolar concentration of 0.4 mM of unlabeled ECD1-CRF-R1 and αhcCRF, respectively, in 10 mM Bis-Tris(HCl) pH 6.5 in 100% D 2 O.
Sequential assignment was performed with the standard protocol for 13 C, 15 N-labeled samples (33). 1 H, 13 (39) spectra recorded with mixing times of either 100 or 120 ms. All of the spectra quadrature detection in the indirect dimensions was achieved using States-TPPI (40). The water signal was suppressed using spin-lock pulses (41) or the WATERGATE sequence (42). All of the spectra were processed with the program PROSA (43) and were analyzed with the program CARA (44).
Determination of the structure of ECD1-CRF-R1 complexed with αhcCRF Meaningful distance restraints (ca. 1718) and angle restraints (371) were collected for the calculation of the structure of the complex (Table 1). These structural restraints were used as an input for the structure calculation with the program CYANA (43) followed by restrained energy minimization using CNS (45). A total of 100 conformers were initially generated by CYANA, and the bundle of 20 conformers with the lowest target function was used to represent the 3D NMR structure. The structures are validated with the program PROCHECK (46) and the structure has been deposited in the protein data bank database with ID code XXXX.

RESULTS
Selection of a soluble high-affinity agonist A major challenge in the 3D structure determination of ECD1-CRF-R1 complexed with an agonist was to obtain a high affinity agonist that was soluble above pH 5 at concentrations required for the NMR experiments. Comparison of several CRF analogs showed that the novel CRF agonist, αhcCRF, satisfied the requirements i.e., it bound with low nanomolar affinity to both CRF-R1 (~1 nM) and the ECD1-CRF-R1 (~30 nM, Table 2) and was soluble at physiological pH. This peptide is only, in part, related to α-helical-CRF(1-41) and to astressin (3,47

Slow conformational exchange dynamics of free ECD1-CRF-R1
The NMR structure of the free ECD1-CRF-R1 was not determined because resonances for many residues were not observed in the [ 15 N, 1 H]-TROSY spectrum ( Fig. 2A). These include the N-terminal residues 1-24 and the C-terminal residues 110-127 of the construct. In addition, several residues that are part of loop 2 involved in ligand binding (see below) (i.e., R 66 , C 68 , F 71 , F 72 , G 74 , V 75 , Y 77 , N 78 and A 95 ) were absent in the [ 15 N, 1 H]-TROSY spectrum ( Fig. 2A). Changes in pH or temperature failed to improve the quality of the spectrum. The absence of these peaks in the spectrum may be due to slow conformational exchange dynamics involving the backbone, which results in line broadening so severe that peaks cannot be detected. A similar observation was documented for the free ECD1-CRF-R2β, but was limited to only a few residues located in loop 2 [i.e., Y87 (corresponding to F71 in CRF-R1), F88 (F72); N89 (Y73), G90 (G74), I91 (V75), K92 (R76), R97 (N81)] (2). For the ECD1-CRF-R2β, this slow conformational exchange was estimated to be on the time scale of 10 -2 seconds and a similar rate is also estimated for the ECD1-CRF-R1. However, since the number of cross peaks absent in the [ 15 N, 1 H]-TROSY spectrum is much larger for ECD1-CRF-R1 than for ECD1-CRF-R2β, the conformational exchange dynamics of ECD1-CRF-R1 must be of greater amplitude and/or must involve a larger segment of the ECD1-CRF-R1 compared to ECD1-CRF-R2β. For ECD1-CRF-R2β, this slow conformational exchange phenomenon was suppressed, at least in part, when complexed with the antagonist astressin, since all of the missing cross peaks of the amide moieties appeared in the [ 15 N, 1 H]-TROSY spectrum of the complex. Assuming that all of the peaks might appear in the spectrum of ECD1-CRF-R1 complexed with the agonist, we carried out the structural studies of the ECD1-CRF-R1 complexed with αhcCRF.
Mapping the agonist binding site on ECD1-CRF-R1 by agonist-induced chemical shift changes Insight into the agonist binding site on ECD1-CRF-R1 can be obtained from an analysis of the resonance shifts in the NMR spectra upon addition of αhcCRF (34). Fig. 2A shows the [ 15 N, 1 H]-TROSY spectrum of the ECD1-CRF-R1 in the absence and presence of equimolar αhcCRF. Following agonist binding, large chemical shift changes of resonances or appearance of cross peaks was observed at four different regions of the ECD1, namely, (i) L 50 and I 51 , V 65 , (ii) R 66 , C 68 , F 71 -N 78 , (iii) N 82 , G 83 , R 85 , A 95 -S 100 , and (iv) Q 103 -E 108 (Fig. 2B). Most of these residues are of hydrophobic nature and conserved between the two CRF receptors (I 51 , R 66 , C 68 , F 72 , G 74 , V 75 , Y 77 , N 78 , N 98 , Y 99 ) (Fig.  2C). This structural study, together with similar chemical shift perturbation data observed for the ECD1-CRF-R2β-astressin complex (2), suggest that both the agonist and antagonist interact with the same residues in the ECD1s.
3D structure of the ECD1-CRF-R1 complexed with αhcCRF Almost complete sequential assignment of the various resonances of the ECD1-CRF-R1-αhcCRF complex was obtained using standard procedures and the 3D structure was determined (see Materials and Methods). The good quality of the 3D structure is represented by the small root mean square deviation (r.m.s.d.) of 0.88 Å for residues 42-105 of the ECD1 and for residues 27-38 of αhcCRF ( Fig. 3A and 3B), as well as by the small value of residual constraint violations in the 20 refined conformers, and by the small deviations from ideal geometry (Table 1). In addition, the input data represent a selfconsistent set, the restraints are well satisfied in the calculated conformers, and similar energy values were obtained for all of the 20 conformers.
As documented already in the structures of the ECD1 of CRF-R2β (2,29) and of the ECD1s of the other members of the B1 receptor family (25)(26)(27)(28), the overall fold of the ECD1 of CRF-R1 is the short consensus repeat (SCR) motif with the three disulphide bonds between cysteine residues, C 30 -C 54 , C 44 -C 87 and C 68 -C 102 . The SCR comprises a short N-terminal α-helix (D 27 -E 31 ) and two anti-parallel β-sheet regions around residues S 47 -V 48 (β1 strand), C 54 -W 55 (β2strand), L 63 -R 66 (β3-strand) and G 83 -E 86 (β4strand). The chemical shift values of 13 C α , 13 C β and 1 H α support the presence of these secondary structural elements. At the core of the SCR motif, the aliphatic side chain of the conserved R 85 is sandwiched between the highly conserved W 55 and W 93 residues and is in close proximity to the side chain of the conserved D 49 and thus may possibly be involved in a salt bridge interaction (Fig. 3B). Furthermore, the indole nitrogen of W 55 forms a hydrogen bond with the carboxyl group of D 49 stabilizing the β1-β2 hairpin (Fig. 3B). These structural features are very similar to those observed for the ECD1 of CRF-R2β and are supported by the up-field shifted side chain resonances of R 85 (βCH 2 : 0.42, -0.55; γCH 2 : 0.65,1.36; δCH 2 : 0.97,0.18 ppm) that require close proximity to an aromatic side chain since aromatic ring currents cause such high-field shifts. Additional up-field shifted resonances are observed for the methyl protons of I 51 (0.12, -0.28 ppm), T 53 (0.08 ppm), the αprotons of G 83 (4.42, 2.92 ppm) and methyl protons of V 97 (0.41, -0.29 ppm) and they are all attributed to the close proximity of the residues to the aromatic ring of Y 99 .
The conformation of the agonist αhcCRF both free and complexed with the ECD1-CRF-R1 For the free 15 N, 13 C-labeled αhcCRF (amino acid numberings per Fig. 4G) at pH 6.5 and 298 K, all of the expected 30 peaks were observed in the [ 15 N, 1 H]-HMQC spectrum (T 11 , E 30 and K 33 were not labeled, see Materials and Methods) (Fig. 4A, red contours). In addition to the major conformation, a minor conformation (~1/3 of the major conformation) is also present for most of the N-terminal residues up to residue L 10 . This minor conformation is attributed to cis/trans isomerizations of the N-terminal proline residues. The chemical shift values for the C α , C β and H α protons ( Fig. 4C and 4D, red bars) of the free peptide indicate that the peptide segment from residues F 12 to L 37 is partially in a helical conformation. Fig. 4A also shows the [ 15 N, 1 H]-HMQC spectrum of 13 C, 15 N-labeled αhcCRF complexed with unlabeled ECD1-CRF-R1 (black contours). When compared to the [ 15 N, 1 H]-HMQC spectrum of free αhcCRF, two prominent features in the complex spectrum are observed: (i) the larger dispersion of the cross peaks attributed to a higher ordered structure of αhcCRF in the complex, and (ii) broad cross peaks with large line-widths due to the larger size of the complex and/or due to slow conformational exchange. Most prominent chemical shift changes as well as line broadening are observed for the C-terminal residues of αhcCRF with the maximum shift for A 41 (Fig. 4E). The chemical shifts for the C α , C β and H α protons ( Fig. 4C and 4D, black bars) suggest that αhcCRF prefers a helical conformation when bound to the ECD1-CRF-R1 and that the helicity is more pronounced upon complex formation not only at the C-terminus but also towards the N-terminus of the peptide (Figs. 4A and 4B). This is further supported by the α-helical NOEs [i.e., αβ(i, i+3), αN(i, i+3) and αN(i, i+4)] observed for residues F 12 to A 41 (Fig. 4G). In addition, amide proton chemical shifts between A 31 and A 41 (Fig. 4F) show a wave-like pattern attributed to the amphipathic nature of the helix in the complex, observed also for astressin bound to the ECD1 of CRF-R2β (2). In addition, a kink of the long helix is observed between residues E 27 to E 29 of αhcCRF (Fig. 3). The positioning of this kink is similar to the kinks observed for CRF family ligands in the solvent DMSO (48) but is absent in all the other ligands of family B1 when they are bound to their respective ECD1s in the crystal structures. Although the role of this kink may be understood only in the context of the full-length receptor, it enlarges the conformational space of the N-terminal peptide segment which is possibly important for receptor-specific signaling (Fig. 3A) (49,50).

Molecular interactions between αhcCRF and the ECD1
The helical segment of αhcCRF bound to the ECD1 is along the protein's hydrophobic face, covering an area of 2647 Å 2 of the ECD1 (Fig. 5). The C-terminal αhcCRF residues L 37 , L 38 and A 41 are involved in hydrophobic contacts with F 72 , Y 73 , Y 77 , I 51 , C 68 , P 69 , V 97 , C 102 and Y 99 of the ECD1. In particular, the side chain of L 38 is located in a deep hydrophobic pocket surrounded by I 51 , C 68 , P 69 , F 72 , Y 77 , Y 99 and C 102 of the ECD1 (note that up-field shifted resonances for the β-and methyl protons of L 38 are indicative of its interactions with aromatic side chains of the ECD1). The amide group at the C-terminus of αhcCRF is involved in an inter-molecular hydrogen bond with the backbone carbonyl of V 97 . In return, the backbone amide proton of V 97 is involved in a hydrogen bond with the backbone carbonyl of A 41 in αhcCRF (Fig. 5). These hydrogen bonds explain the necessity of C-terminal amidation for high affinity recognition of CRF ligands (48,51). Although there are several NOEs observed between the side chain of N 34 to Y 73 and V 75 , in the 3D structure, these two side chains are not close enough to form an intermolecular hydrogen bond. The side chain of A 31 of αhcCRF also interacts with V 75 , Y 77 through hydrophobic interactions. The side chain of R 35 is in close proximity to the side chain of E 104 , thereby facilitating a salt-bridge interaction. Since the side chain resonances of R 35 could not be observed in the 13 C-resolved [ 1 H, 1 H]-NOESY spectrum except for the delta protons, its interaction must be interpreted with care.

Comparison of the structures of ECD1-CRF-R1
complexed with the agonist αhcCRF and ECD1-CRF-R2β complexed with the antagonist astressin Recently, our group reported the structure of the ECD1 of CRF-R2β complexed with the peptide antagonist astressin (2). Comparison of the structures of the complexes of ECD1-CRF-R1 and ECD1-CRF-R2β enables the identification of the common interaction sites. Both of the structures have the SCR motif characteristic of the ECD1s of family B1 members (Fig. 6A). There is a short N-terminal helix observed for ECD1-CRF1-R1, which was absent in the structure of ECD1-CRF-R2β because in the latter, the N-terminal segment was truncated in the protein construct. While conformation of loop 1 is not defined in either ECD1, loops 2 and 3 are structured and interact with the corresponding ligand in a slightly different manner (Fig. 6B). The backbone of loop 2 of ECD1-CRF-R2β folds closer to the ligand than the corresponding loop in ECD1-CRF-R1, while the opposite holds for loop 3.
There are also structural differences of the ligand. Although both the agonist and the antagonist bind to almost the same region of the ECD1, the orientations of the ligands are slightly different with respect to loop 2 (Fig. 6B, C, D). Furthermore, the C-terminal residues of astressin prefer a 3 10 -helix, whereas in αhcCRF, they are in an α-helical conformation. In both ligands, the backbone carbonyl of the last residue (A/I 41 ) is involved in a hydrogen bond with the amide proton of Val 97/113 . The side chains of the ligand residues, E 39 , L/K 36 , A/H 32 , and Q 29 are completely solvent exposed in both of the structures. Also, the side chain of E 40 of αhcCRF is completely solvent exposed, whereas I 40 of astressin interacts with Q 66 of CRF-R2β. The side chain of L/Nle 38 is completely buried in a hydrophobic core in both structures (Fig. 6). Although the side chains of Phe 72/88 of ECD1 are conserved in their positions and these residues interact with residues L 37 and N 34 of the ligand in both of the structures, the positions of Y 73 /N 89 and Y 77/93 are distinct. The orientation of Y 77 of CRF-R1 is towards the ligand, whereas in CRF-R2β, Y 93 points away from the ligand. Consistent with this is the observation that the mutation [Y77A] in CRF-R1 abrogated high affinity binding of both astressin and sauvagine and significantly increased the EC 50 for sauvagine and urocortin1 stimulated intracellular cAMP accumulation (data not shown). The side chain of N 34 of astressin is in close proximity to the ECD1 to form a hydrogen bond with the backbone carbonyl of F 88 in ECD1-CRF-R2β. Such a hydrogen bond is missing in the structure of the ECD1-CRF-R1 in complex with αhcCRF. The residues V 75 /I 91 are in almost the same position and they interact with A 31 of the ligand. The side chain of R 35 does not directly interact with E 104 of the ECD1-CRF-R1, although it is close enough to form a solvent-exposed intermolecular salt bridge. In contrast, in ECD1-CRF-R2β, R 35 is involved in a buried salt bridge with E 86 , which, in ECD1-CRF-R1, is replaced by A 70 and its side chain is solvent exposed.
Structural plasticity of both the ECD1 and the ligand The absence of cross peaks for amide moieties of residues in loop 2 in the [ 15 N, 1 H]-TROSY spectra of both of the ECD1s of CRF-R1 and CRF-R2β suggests the presence of slow conformational exchange dynamics in the ECD1s of CRF receptors. Although the cross peaks appear in the [ 15 N, 1 H]-TROSY spectra in the complex with the peptide for both receptor studies, the significant line broadening observed for these peaks suggests that the conformational exchange is only partially suppressed. The amide moieties of the ligand αhcCRF also show broad, weak cross peaks in the [ 15 N, 1 H]-HMQC spectra upon complex formation (Fig. 4). The presence of slow conformational exchange dynamics in the millisecond time range for segments of both the ECD1 and the ligand is in agreement with the presence of conformational heterogeneity observed in both the NMR and Xray structures. Furthermore, the nanomolar binding affinity of the antagonist astressin for a CRF-R2β mutant whose corresponding ECD1 shows molten globule-like conformational states (52) supports the dynamic character of the ECD1. Although we can only speculate about the biological role of these conformational exchange dynamics, their presence could account for the multiple recognition, binding and signaling observed for the various hormone ligands.

Refinement of the two step binding mechanism of CRF peptides to its receptors
The two-step model for ligand-binding and signaling of type B1 GPCRs (29,53,54) proposes that the Cterminal segment of the ligand binds to the ECD1, which then may position the aminoterminal portion of the peptide hormone in close proximity to the serpentine regions of the receptor to initiate signaling. The ECD1 is therefore the major peptide-binding domain and conversely, the C-terminal segment of the ligand is important for high binding affinity and selectivity to the receptors. All of the 3D structures of the ECD1-receptor-ligand complexes are consistent with this model, since the C-terminal segment of the peptide ligand interacts with the ECD1 (1,2,(26)(27)(28). Our NMR studies of the complex between ECD1-CRF-R1 and an agonist further indicate that the recognition of the ligand by the ECD1 not only binds the hormone and positions its N-terminal residues for signal activation, but also induces helix formation towards the N-terminus of the ligand to generate a conformationally active state. In a recent review (55), this additional function in receptor activation of the ECD1 in family B1 GPCRs was proposed to be based on (i) a structural comparison between various hormone ligands, free and complexed with ECD1s and, (ii) the importance of helix-capping residues in the N-terminal region of the numerous corresponding ligands (56)(57)(58). The induction of the helix in the ligand upon complex formation with the ECD1 is accompanied by a kink between residues E 24 to E 26 of αhcCRF ( Fig. 3A; Fig. 8); this kink may also play a role in receptor activation by either enlarging the conformational space of the Nterminal peptide segment or by positioning the ECD1 relative to the serpentine region of the receptor important for signaling or/and coreceptor interactions (29). Hence, our 3D structure of the ECD1 of CRF-R1 complexed with an agonist suggests a refined two-step model for receptor activation.
Agonist versus antagonist Previous studies on chemically modified and truncated CRF ligands showed that the first seven residues at the Ntermini of CRF are not necessary for GPCR signaling (3,4) and that residue 8 was critical, whereas the C-terminal (~15) residues are important for binding (47,53,54). Hence, CRF analogs truncated by 8 residues or more at the N-terminus are antagonists. This finding can easily be understood with the help of the twostep model for ligand binding and signaling of type B1 GPCR discussed above (16,19,29). If the N-terminal segment of the ligand is missing, the C-terminal fragment still binds to the receptor but is not able to produce activation. Such a ligand is then evidently an antagonist since it occupies the major binding site thereby blocking peptide agonist binding. The 3D structure of the ECD1-CRF-R1 complexed with the agonist αhcCRF presented here is the first direct experimental proof that supports this hypothesis (Figs. 7 and 8) and shows clearly the similarity of the C-terminal binding of the peptide agonists and antagonists.

CONCLUSION
The information gained from our structural studies complements earlier knowledge of the molecular interactions between ligands and GPCRs of the B1 family. Specifically, the 3D NMR structure presented here of ECD1-CRF-R1 complexed with a high affinity CRF agonist, has identified the residues in the ECD1-CRF-R1 involved in ligand recognition and has highlighted the similarity between agonist and antagonist binding receptor domains. Further, the structure of the complex revealed the extended helicity of the N-terminal domain of the agonist. These data provide further support for the model of ligand-induced receptor signaling in the B1 receptor family.   complexed with its ECD1 receptor domain (25) is not included in this figure because they are completely different from that of the other ligands. Whether this difference is an artifact of the procedure used in structure determination or a true difference needs to be determined,   Hydrophobic residues are colored yellow, positively charged residues are colored blue and negatively charged residues are colored red. (C) (C-G) Plots of the chemical shift difference between αhcCRF and the corresponding 'random coil' values for 13 C α -13 C β and, (D) for 1 H α atoms shown as black and red bars for the complex and free αhcCRF, respectively. Chemical shift changes suggest that the ligand binding induces more helicity, not only in the C-terminal segment, but also in the N-terminal segment. (E) Plot of the normalized chemical shift changes [Δ(δ( 1 H) 2 + Δ(δ( 15 N)) 2 /5] 1/2 observed for αhcCRF in the complex versus the amino acid sequence (34). There is an overall 0.1 ppm chemical shift change due to the difference in the temperatures at which the free (298 ºK) and complex (308 ºK) spectra were recorded. (F) Plot of the observed chemical shifts of the 1 H N moieties of αhcCRF free (in red) and bound (in black) to ECD1-CRF-R1. (G) Intra-molecular NOEs observed in αhcCRF bound to ECD1-CRF-R1 along the amino acid sequence. Thin, medium and thick lines represent weak (4-5 Å), medium (3-4 Å) and strong (<3 Å) NOEs observed between the residues connected by the line. The bar connecting E 30 and K 33 represents the lactam bridge connecting the side chains of these residues. Residues T 11 , E 30 and K 33 are not isotopically labeled and hence their values are not shown.   . The backbone ribbon of the ECD1 along with the highly conserved residues including the disulphide bonds is shown in grey. Loop 2 residues that are observed are shown in light green; it must be noted that the side chains of F 71 , Y 73 , R 76 are missing. Ligand residues 27-41 that are observed are highlighted in dark green and the ligand backbone is shown as a dark green ribbon. (C), (D) Crystal structures of the MBP(F94E)-CRF-R1-ECD-H 6 protein complexed with the truncated CRF(27-41) (pdb code: 3EHT, crystal form III). The backbone ribbon of the ECD1 along with the highly conserved residues including the disulphide bonds is shown in orange. Loop 2 residues that are observed are shown in light green; it must be noted that the conformation of loop 2 is different from that of (A), resulting in different orientations for the residues involved in ligand binding. Ligand residues 31-41 having electron density are highlighted in dark green and the backbone is shown as a dark green ribbon. (E), (F) NMR structures of the ECD1-CRF-R1 complexed with αhcCRF. The backbone ribbon of the ECD1 along with the highly conserved residues including the disulphide bonds is shown in cyan. Side chains of loop 2 residues are shown in light green. The conformation of loop 2 is very close to that of (A) and is different from that of (C). The side chains of F 71 , Y 73 and R 76 were assigned and hence their interactions with the ligand could be observed. with pdb code 3EHU) (B) astressin (in green) complexed with ECD1-CRF-R2β (from the NMR structure with pdb code 2JND), (C) astressin (in royal blue) in DMSO. The structures are shown after superimposing the C-terminal residues 30-41 of astressin.