Structure-activity relationships of simplified resiniferatoxin analogues with potent VR1 agonism elucidates an active conformation of RTX for VR1 binding

We previously described a series of N-(3-acyloxy-2-benzylpropyl) homovanillate and N'-(4-hydroxy-3-methoxybenzyl) thiourea derivatives that were potent VR1 agonists with high-affinities and excellent analgesic profiles. The design of these simplified RTX analogues was based on our RTX-derived pharmacophore model which incorporates the 4-hydroxy-3-methoxyphenyl (A-region), C(20)-ester (B-region), orthophenyl (C1-region) and C(3)-keto (C2-region) groups of RTX. For the purpose of optimizing the spatial arrangement of the four principal pharmacophores on the lead agonists (1-4), we have modified the distances in the parent C-region, 3-acyloxy-2-benzylpropyl groups, by lengthening or shortening one carbon to vary the distances between the pharmacophores. We find that two of the amides, 4 and 19, possess EC(50) values <1 nM for induction of calcium influx in the VR1-CHO cells. As observed previously, the structure-activity relations for inhibition of RTX binding to VR1 and for induction of calcium uptake were distinct, presumably reflecting both intrinsic and methodological factors. In order to find the active conformation of VR1 ligands, the energy-minimized conformations of seven selected agonists were determined and the positions of their four pharmacophores were matched with those of five low energy RTX conformations. The rms values for the overlaps in the pharmacophores were calculated and correlated with the measured binding affinities (K(i)) and calcium influx (EC(50)) values. The binding affinities of the agonists correlated best with the RMS values derived from RTX conformation E (r(2)=0.92), predicting a model of the active conformation of RTX and related vanilloids for binding to VR1. Poorer correlation was obtained between any of the conformations and the EC(50) values for calcium influx.