Structure-guided macrocyclisation of the Gila monster venom peptide helospectin I: a computational design study of VPAC-targeted analogues for vasculogenic erectile dysfunction and female sexual arousal disorder
Amr Kamel Khalil Ahmed
Abstract
Background. Phosphodiesterase-5 (PDE5) inhibitors depend on preserved endothelial nitric-oxide (NO) output and lose efficacy in long-standing diabetes, after pelvic surgery and in radiation-injured vasculature. A vasodilator that acts downstream of, and in parallel to, NO–cGMP would address that gap. Helospectins I and II, 37–38-residue peptides of the glucagon/vasoactive-intestinal-peptide (VIP) superfamily isolated from Heloderma venom, are VPAC1/VPAC2 agonists that relax vascular smooth muscle through Gs–adenylate cyclase–cAMP–PKA signalling. Their therapeutic use is limited by proteolysis and by the systemic hypotension that unselective VPAC agonism produces.
Objective. To select, model and re-engineer a Heloderma venom peptide into a conformationally constrained VPAC agonist suitable for local, on-demand genital vasodilation, and to define precisely where covalent constraints can be placed without touching the receptor-binding surface.
Methods. The Heloderma venom proteome was triaged by molecular target and toxicity to a single candidate, helospectin I (exendin-1; UniProt P0DJ94/P04203; 38 aa). A receptor-bound model was built on a dual template. The core (residues 1–27) was taken from PACAP-27 as bound to human VPAC2 (PDB 7VQX, 2.74 Å) and the C-terminal extension (28–38) from the micelle-bound PACAP-38 NMR ensemble (PDB 2D2P). The two were threaded 1:1 (24 substitutions) with PDBFixer and relaxed in OpenMM (amber14 with GBn2 implicit solvent) under positional restraints on the receptor-bound core. Solvent-accessible surface area was computed for the free and receptor-bound peptide to define the binding footprint. Every i,i+3, i,i+4 and i,i+7 pair in the helical region was scored for Cβ–Cβ distance, solvent exposure in the complex, interface involvement and conservation across the glucagon/VIP superfamily. The top-ranked bridge was built explicitly, and native and cyclised peptides were compared in single 350 K implicit-solvent molecular dynamics trajectories Results. Helospectin I is 56% identical to VIP and 89% identical to helodermin over residues 1–27. It carries Lys12 natively, where VIP has arginine, and a long C-terminal extension, which VIP lacks; both were engineered into VIP to obtain the VPAC2-selective agonist Ro 25-1553. Lys21, the anchor of the Ro 25-1553 lactam, is native to VIP as well as to helospectin and is not an engineered feature. In the receptor-bound model, residues 1–20 together with Tyr22, Leu23, Ile26 and Leu27 are buried at the VPAC2 interface, whereas Lys21, Glu24, Ser25, Gly28, Ser29 and the C-terminal tail lose no accessible surface in that pose (ΔSASA = 0 Ų). The bridge scan places the two highest-scoring macrocycles on exactly that free face: a Lys21–Glu24 i,i+3 lactam requiring no sequence change (Cβ–Cβ 5.7 Å) and a Lys21–Asp25 i,i+4 lactam requiring the single substitution S25D (Cβ–Cβ 6.4 Å), the latter reproducing the ring topology of Ro 25-1553 on a scaffold that natively supplies Lys12 and the C-terminal extension. Closing the Lys21–Asp25 ring perturbs the receptor-bound conformation by 0.42 Å Cα RMSD and leaves the 2–31 helix intact. Under 350 K stress dynamics the cyclised peptide retains more helical hydrogen bonding than the native peptide (17.7 ± 1.7 versus 14.8 ± 1.9 i→i+4 backbone H-bonds over the final 50 ps, single trajectory per construct), although backbone RMSD, dominated by the unconstrained C-terminal tail, does not separate the two. Transplanting the cyclised peptide into VPAC1 (PDB 8E3Z) gives an interface of almost identical size (1643 versus 1652 Ų buried), with the difference confined to two local regions: the C-terminal tail engages VPAC1 more, while the Ser11–Tyr22 helix face engages VPAC2 more.
Conclusions. In this model a single substitution yields a macrocyclic analogue of a natural VPAC agonist whose constraint sits entirely off the receptor interface. We propose HSP-3 ([Asp25]helospectin I-NH₂, cyclo(Lys21–Asp25)) as a lead for local genital vasodilation, and specify the assay cascade, including the ICP/MAP ratio that separates local efficacy from systemic hypotension, required to test it. All conclusions are computational and conditional on a homology model that has not been independently validated; no binding, functional or in-vivo data are reported.