Constrained Corticotropin-Releasing Factor (CRF) Agonists and Antagonists with i−(i+3) Glu-Xaa-dXbb-Lys Bridges

We hypothesized that covalent constraints such as side-chain to side-chain lactam rings would stabilize an α-helical conformation shown to be important for the recognition and binding of the human corticotropin-releasing factor (hCRF) C-terminal 33 residues to CRF receptors. These studies led to the discovery of cyclo(20−23)[dPhe12,Glu20,Lys23,Nle21,38]hCRF(12-41) and of astressin {cyclo(30−33)[dPhe12,Nle21,38,Glu30,Lys33]hCRF(12-41)}, two potent CRF antagonists, and of cyclo(30−33)[Ac-Leu8,dPhe12,Nle21,Glu30,Lys33,Nle38]hCRF(8-41), the shortest sequence equipotent to CRF reported to date (Rivier et al. J. Med. Chem. 1998, 41, 2614−2620 and references therein). To test the hypothesis that the Glu20−Lys23 and Glu30−Lys33 lactam rings were favoring an α-helical conformation rather than a turn, we introduced a d-amino acid at positions 22, 31, and 32 in the respective rings. Whereas the introduction of a d-residue at position 31 was only marginally deleterious to potency (ca. 2-fold decrease in potency), introduction of a d-residue at position 22 and/or 32 was favorable (up to 2-fold increase in potency) in most of the cyclic hCRF, α-helical CRF, urotensin, and urocortin agonists and antagonists that were tested and was also favorable in linear agonists but not in linear antagonists; this suggested a unique and stabilizing role for the lactam ring. Introduction of a [dHis32] (6) or acetylation of the N-terminus (7) of astressin had a minor deleterious or a favorable influence, respectively, on duration of action. In the absence of structural data on these analogues, we conducted molecular modeling on an Ac-Ala13-NH2 scaffold in order to quantify the structural influence of specific l- and dAla6 and l- and dAla7 substitutions in [Glu5,Lys8]Ac-Ala13-NH2 in a standard α-helical configuration. Models of the general form [Glu5,lAla6 or dAla6,lAla7 or dAla7,Lys8]Ac-Ala13-NH2 were subjected to high-temperature molecular dynamics followed by annealing dynamics and minimization in a conformational search. A gentle restraint was applied to the 0−4, 1−5, and 8−12 O−H hydrogen bond donor−acceptor pairs to maintain α-helical features at the N- and C-termini. From these studies we derived a model in which the helical N- and C-termini of hCRF form a helix−turn−helix motif around a turn centered at residue 31. Such a turn brings Gln26 in close enough proximity to Lys36 to suggest introduction of a bridge between them. We synthesized dicyclo(26−36,30−33)[dPhe12,Nle21,Cys26,Glu30,Lys33,Cys36,Nle38]Ac-hCRF(9-41) which showed significant α-helical content using circular dichroism (CD) and had low, but measurable potency {0.3% that of 6 or ca. 25% that of [dPhe12,Nle21,38]hCRF(12-41)}. Since the 26−36 disulfide bridge is incompatible with a continuous α-helix, the postulate of a turn starting at residue 31 will need to be further documented.