RegIIA and its analogues (Fig. 2) were synthesized by solid-phase Fmoc chemistry, as previously described (Yu et al. 2013). To achieve regioselective oxidation, Fmoc-Cys(Trt)-OH was used at Cys I and Cys III positions, and Fmoc-Cys(Acm)-OH was used at Cys II and Cys IV positions. The fully oxidized product was separated and purified by RP-HPLC (reversed-phased-high performance liquid chromatography). The molecular weight and purity were confirmed by electrospray ionization–mass spectrometry (ESI–MS) and analytical HPLC (Supplementary Fig. S1).
We performed molecular dynamics (MD) simulations on the α-Ctx RegIIA and hα3β4 nAChR complex in the presence or absence of oligosaccharide chains. As the carbohydrates are located at the top of the receptor, only the hα3β4 nAChR ECD model was simulated. MD simulations suggested the interaction modes of RegIIA at apo- and carbohydrate-bound hα3β4 nAChRs were similar to the binding mode of RegIIA predicted in our previous studies (Fig. 3A, B, Table 1) (Cuny et al. 2016; Xu et al. 2020). In both models, RegIIA N9 and H5 residues formed hydrogen bonds with β4 K161 and α3 P198, while N11 formed direct interactions with α3 S150 and β4 R83 (Fig. 3C–E, G, H).
Figure 3. Binding modes of RegIIA at the binding site of hα3β4 nAChR. A In the carbohydrate-free model, several hydrogen bonds (dashed lines) are formed between pairs of interacting residues. The α3(+) interface is shown in cyan, β4(-) in purple, and RegIIA in silver. B In the glycoprotein model, the interaction mode is similar to the carbohydrate-free model, except that β-D-mannose forms hydrogen bonds with N12 and H14 (dashed lines). The α3(+) interface is shown in peach, β4(-) in purple, and RegIIA in silver. Residues from the receptor and RegIIA are labeled using normal and italic fonts, respectively. The key interaction sites responsible for the binding of RegIIA are highlighted with dotted circles. C–H, magnification of the key sites in carbohydrate-free model (C–E) and glycoprotein model (F–H) highlighted with circles in (A, B)
Residuea Absence of oligosaccharide chain Presence of oligosaccharide chain +b −c +b −c S4 – D173 K145 W59, D173 H5 Y93, S148, G147, Y190, Y197, P198 – Y93, S148, D199, Y190, Y197, P198 W59 P6 S148, G147, W149 W59, L123 Y93, S148, W149 W59, L123 A7 S148, W149, Y197, S150, Y151 R83 Y93, S148, W149
Y197, S150, Y151
– N9 – K61, W59, L123 – K61, W59, L123 V10 W149, S150 N111, I113, L121, W122, L123 W149, S150 K61, I113, L121, W122, L123 N11 S150, Y197 R83, I113, L121 S150, Y197 R83, L121, R115 N12 C193, E195, Y197 – C193, Y197 K61, BMA H14 – – – BMA, NAG I15 C192, C193 – C192, C193 BMA BMA and NAG oligosaccharides are in bold
aResidues of RegIIA forming direct contact with the hα3β4 nAChR are listed. Contacts between hα3β4 nAChR and RegIIA are defined as van der Waals interactions when the distance between heavy atoms of RegIIA and hα3β4 nAChR is less than 4 Å. Residues of the hα3β4 nAChR and oligosaccharides forming hydrogen bonds with RegIIA are underlined
bResidues from the principal subunit
cResidues from the complementary subunit
Table 1. Pairwise interactions between the RegIIA and the hα3β4 nAChR
One main difference between the two systems was the interaction between the oligosaccharide chain and β4 N117 through N-glycosidic bonds. The presence of linear oligosaccharides mainly affected RegIIA N12 and H14, which involved hydrogen bond interactions between β-D-mannose and the two residues (Fig. 3B, F). The 250 ns-MD simulation also showed change of the distance between β-D-mannose and N12 and H14, (Fig. 4A, B, Supplementary Fig. S2) and although the contact was unstable, interactions with N12 was relatively more stable than H14. Overall, MD simulations indicated that the linear oligosaccharide at β4 N117 positions was flexible and formed relatively weak interactions with residues at the solvent exposed surface of RegIIA.
Figure 4. The distance between RegIIA H14 and β-D-mannose (at β4-N117 position), and β-D-glucosamine (at β4-N117 position) during MD simulations. A Change of the distance between the ND1 of H14 and the O2 of β-D-mannose (cyan). During 150–200 ns, the average distance between two heavy atoms is 2.5 Å, suggesting a direct contact between β-D-mannose and H14 during this period. B Change of distance between O3 of β-D-glucosamine and NE of H14 (orange). Before 150 ns, the interaction between β-D-glucosamine and H14 was via van der Waals force
We also found that the oligosaccharide chain linked to α3 N141 through N-glycosidic bonds can directly interact with α3 H186 (Fig. 5A, Supplementary Fig. S3). The average distance between the CA of α3 C193 and the CA of β4 S40 on the C-loop in apo- and carbohydrate-bound models was about 18 Å and 20 Å, respectively (Fig. 5C). Similarly, the C-loop was also slightly more opened in the carbohydrate-free model than the carbohydrate-bound model (Supplementary Fig. S7). The more closed C-loop suggested that in the presence of carbohydrates, RegIIA had more close contacts with the binding site, especially with the C-loop (Fig. 5B). Therefore, we speculated that the presence of linear oligosaccharide chains did not only affect the binding mode of RegIIA at the hα3β4 nAChR binding pocket, but also had a slight impact on C-loop conformation.
Figure 5. Effects of the carbohydrate at α3 N141 on the conformation of the binding site. A H186 of α3 forms hydrogen bonds with β-D-glucosamine and α-D-mannose on the oligosaccharide chain. The α3 interface is shown in peach, β4 in purple, and oligosaccharide chain in silver white. B Superimposition of carbohydrate-free model (α3 interface is shown in cyan and β4 in purple) and glycoprotein model (α3 interface is shown in peach and β4 in purple).The distance used to characterize C-loop opening in the hα3β4 nAChR is measured between the CA of α3 C193 and the CA of β4 S40. The red and blue dashed lines indicate the distance of C-loop opening in the glycoprotein model and the carbohydrate-free model, respectively. C The 250 ns MD simulation of carbohydrate-free (red) and -bound (black) models showing change of the distance between the CA of α3 C193 and the CA of β4 S40
Through MD simulations, we proposed that H14 of RegIIA played a vital role in interactions with the carbohydrate at β4 N117. Thus, to explore H14 contribution to RegIIA activity at the hα3β4 subtype, H14 RegIIA analogues were synthesized and tested at heterologous hα3β4 nAChRs expressed in Xenopus laevis oocytes.
Due to the close contact between H14 and the carbohydrate at β4 N117, residues with varied physico-chemical properties, such as hydrophilic or aromatic residues were introduced to position 14 of RegIIA. All analogues had substantially decreased potency in comparison to the wild-type RegIIA (Fig. 6A). Therefore, we speculated that the function of H14 may be unique. In addition to direct interactions with the carbohydrate, H14 might also play an essential role at sustaining the conformation of the peptide. Indeed, in MD simulations, the side chain of H14 formed an intraresidue H-bond (Fig. 6B, Supplementary Fig. S4), which contributes to maintaining the three-dimensional structure of RegIIA. Through circular dichroism analysis, we also identified that H14 is essential to the secondary structure of the peptide, and replacement of H14 resulted in certain degree of shift at 200–220 nmol/L (Fig. 6C).
Figure 6. Mutational effects at the activity of RegIIA. A Bar graph of RegIIA and analogues (50 nmol/L) inhibition of ACh-evoked peak current amplitude mediated by hα3β4 nAChRs (*P < 0.0001 compared to RegIIA). Whole-cell currents were activated by 300 μmol/L ACh (mean ± SEM, n = 6–12). B The 250 ns MD simulation showing change of the distance between the N of H14 and the ND1 of H14 (green). The average distance between two heavy atoms is 3.5 Å. C Circular dichroism spectra of RegIIA and its analogues
In consideration of the direct interactions between the carbohydrate and H14, removal of the carbohydrate from β4 N117 might explain decrease the potency of RegIIA. Indeed, in our previous study RegIIA was sixfold less potent at the β4 N117D mutant (Cuny et al. 2016). However, the significant activity decrease was not likely resulted from β4 N117 direct interactions due to the long distance between β4 N117 and RegIIA H14. Thus, it is reasonable to deduce that the decrease is more likely due to the removal of the carbohydrate rather than from the change of the side chain interactions with the toxin.