The binding orientations of structurally-related ligands can differ; A cautionary note

Abstract

Crystal structures can identify ligand-receptor interactions and assist the development of novel therapeutics, but experimental challenges sometimes necessitate the use of homologous proteins. Tropisetron is an orthosteric ligand at both 5-HT₃ and α7 nACh receptors and its binding orientation has been determined in the structural homologue AChBP (pdbid: 2WNC). Co-crystallisation with a structurally-related ligand, granisetron, reveals an almost identical orientation (pdbid; 2YME). However, there is a >1000-fold difference in the affinity of tropisetron at 5-HT₃ versus α7 nACh receptors, and α7 nACh receptors do not bind granisetron. These striking pharmacological differences prompt questions about which receptor the crystal structures most closely represent and whether the ligand orientations are correct. Here we probe the binding orientation of tropisetron and granisetron at 5-HT₃ receptors by in silico modelling and docking, radioligand binding on cysteine-substituted 5-HT₃ receptor mutants transiently expressed in HEK 293 cells, and synthetic modification of the ligands. For 15 of the 23 cysteine substitutions, the effects on tropisetron and granisetron were different. Structure-activity relationships on synthesised derivatives of both ligands were also consistent with different orientations, revealing that contrary to the crystallographic evidence from AChBP, the two ligands adopt different orientations in the 5-HT₃ receptor binding site. Our results show that even quite structurally similar molecules can adopt different orientations in the same binding site, and that caution may be needed when using homologous proteins to predict ligand binding.

Keywords: 5-HT(3); Agonist; Antagonist; Binding; Crystal; Cys-loop; Granisetron; Ion channel; Ligand; Radioligand; Receptor; Structure; Synthesis; Tropisetron.

Fig. 1

Positions of the residues mutated in this study. (A) A cartoon showing the orthosteric binding site of the 5-HT₃ receptor formed by binding loops A–F. The extracellular binding site is found at the interface of two adjacent subunits. For clarity only two subunits are shown. (B) An amino acid sequence alignment showing the positions of mutated residues (white text, black boxes). The six recognised binding loops are shown as grey lines. The proteins are sequences from human 5-HT3A (P46098), mouse 5-HT3A (Q6J1J7), chick α7 nACh (F1P4Y5) and sequences taken from the AChBP crystal structures 2YME, 2WNC and 3SQ6. c_α7nACh and h_5HT3 are the sequences of receptors used in the binding studies presented here. EMBOSS Needle shows that the sequence of 2WNC_AChBP has a closer identity and similarity to c_α7nACh (Id = 27.4%; Sim = 41.8%) than to h_5HT3 (Id = 19.4%; Sim = 31.8%) (Rice et al., 2000). To facilitate comparisons with previous work, the numbering used throughout this manuscript refers to residues at equivalent positions of the mouse 5-HT3A subunit (Q6J1J7).

Fig. 2

Binding of tropisetron. Predicted binding clusters for tropisetron docked into (A) a homology model of the 5-HT₃ receptor binding site based upon the template 2WNC (Model-1), and (B) the crystal structure 2YME (Model-2). (C) Co-crystal structures 2WNC and 2YME show AChBP bound with tropisetron and granisetron respectively. In panels 2A & 2B all 10 predicted ligand poses are overlaid. 5-HT₃ receptor residues within 5 Å of tropisetron in each of these poses are shown in Table 1.

Fig. 3

Binding site residues and the corresponding change in affinity when they are mutated. All positions in this figure are shown as cysteine and are colour coded according to the extent of the change in affinity this mutation caused. Residues that are thought to have a structural role have been omitted. The structure is Model-1, and more details of the effects of these cysteine substitutions can be found in Table 2, Table 3.

Fig. 4

Expression of mutant 5-HT₃ receptors. Three of the mutant receptors did not bind either [³H]tropisetron or [³H]granisetron (W90C, E129C, W183C). To confirm that this could be attributed to altered binding rather than changes in expression, these mutants were probed by Western blot using a 5-HT₃ receptor-specific antibody. Expression of the mutants and wild type receptors were comparable, and there was no detection in untransfected cells. In each lane 10⁵ cell equivalents were loaded and an antibody against the cleavage and polyadenylation specificity factor (CPSF) included as an internal control. Note that whole-cell homogenates were used for both radioligand binding and Western blot to enable monitoring of the same populations of both intracellular and cell-surface receptors.

Fig. 5

The three docked-pose clusters predicted by docking tropisetron into a homology model of the 5-HT₃ receptor that was based on the template 2WNC (Model-1). Representative orientations of tropisetron are shown for each of the clusters A, B and C (for an overlay of all docked poses see Fig. 2A). In panel 5D, tropisetron (Cluster-A, pink) , and granisetron (2YME, green) are overlaid. The residues shown are those that abolished binding in Fig. 3 (Cluster-A, red; 2YME, cyan). Other residues are omitted for clarity. For a full list of the identified residues in Cluster-A and 2YME see Table 1. The docked-pose predicted in Cluster-A is the most consistent with the binding results presented here. (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)