Natural aminosterols inhibit NMDA receptors with low nanomolar potency

Abstract

Abnormal functions of N-methyl-D-aspartate receptors (NMDARs) are associated with many brain disorders, making them primary targets for drug discovery. We show that natural aminosterols inhibit the NMDAR-mediated increase of intracellular calcium ions in cultured primary neurons and neuroblastoma cells. Structural comparison with known NMDAR-negative allosteric modulators, such as pregnanolone-sulfate-2 (PAS), raises the hypothesis that aminosterols have the same mechanism of action. Fluorescence resonance energy transfer (FRET) measurements using labeled NMDAR and the labeled aminosterol trodusquemine (TRO) indicate close spatial proximity, likely arising from binding. Other indirect yet plausible mechanisms for NMDAR inhibition by TRO were excluded. Electrophysiological patch clamp measurements on primary neurons indicate that pre-incubated TRO inhibits NMDA-induced ion currents with a IC₅₀ of 5 nm. Inhibition is observed only after cell membrane pre-adsorption, indicating accessibility to NMDAR from the cell membrane and binding to the transmembrane domains (TMDs) and TMD-ligand-binding domain (LBD) linkers, similarly to PAS. The TRO IC₅₀ is 5000-fold higher than that of PAS and 20-16 000 times higher than those of other inhibitors binding to TMD/TMD-LBD regions, identifying aminosterols as promising and potent NMDAR modulators.

Keywords: ENT‐03; NMDA antagonist; ionotropic glutamate receptors; negative allosteric modulators; squalamine.

Fig. 1

Chemical structures of aminosterols, pregnanolone‐sulfate‐2 (PAS), and derivatives. Chemical structures of squalamine (SQ, A), trodusquemine (TRO, B), ENT‐03 (C), PAS (D), PA with L‐argininyl replacing the sulfate group at C3 (E), PA with 4(trimethylammonio)‐butanoyl replacing the sulfate group at C3 (F), PAS with isobutyl group replacing the acetyl group at C17 (G).

Fig. 2

Intracellular free Ca²⁺ levels in primary rat cortical neurons treated with N‐methyl‐d‐aspartate (NMDA) after treatment with different trodusquemine (TRO) concentrations. (A) Representative confocal scanning microscopy images of untreated cells and cells treated for 10 min with 1.5 mm NMDA without or with pretreatment with 15, 10, 7.5, and 5 μm TRO for 10 min. Scale bar: 30 μm. (B) Semi‐quantitative analysis of intracellular free Ca²⁺‐derived fluorescence as shown in panel A. Variable numbers of cells (12–22) in three different experiments (n = 3) were analyzed for each condition, error bars are SEM. Comparisons between the different groups were performed by Student's t‐test, double (**) and triple (***) asterisks refer to P values <0.01 and <0.001, respectively, relative to untreated cells. Single (§) and triple (§§§) symbols refer to P values <0.05 and <0.001, respectively, relative to NMDA treatment.

Fig. 3

Presence of functionally active N‐methyl‐d‐aspartate receptors (NMDARs) in SH‐SY5Y cells. (A) GluN2B‐derived fluorescence in representative confocal scanning microscopy images of SH‐SY5Y cells treated with a primary antibody against GluN2B, labeled with Alexa Fluor‐488, following pretreatment with vehicle, 25 nm negative control siRNA, and 25 nm siRNA against GluN2B. Scale bar: 30 μm. (B) Semi‐quantitative analysis of GluN2B‐derived fluorescence as shown in panel A. (C) Representative confocal scanning microscopy images of intracellular Ca²⁺ levels in untreated cells and cells treated with 1.5 mm NMDA for 10 min without and with pretreatment with 10 μm memantine (Mem). Scale bar: 30 μm. (D) Semi‐quantitative analysis of intracellular free Ca²⁺‐derived fluorescence as shown in panel C. Variable numbers of cells (12–22) in three different experiments (n = 3) were analyzed for each condition, error bars are SEM. Comparisons between the different groups were performed by Student's t‐test, triple (***) asterisks refer to P values <0.001, relative to vehicle or untreated cells.

Fig. 4

Intracellular free Ca²⁺ levels and ROS levels in SH‐SY5Y cells treated with N‐methyl‐d‐aspartate (NMDA) after treatment with different trodusquemine (TRO) concentrations. (A, C) Representative confocal scanning microscopy images of (A) intracellular Ca²⁺ levels and (C) intracellular ROS levels in untreated cells and cells treated with 1.5 mm NMDA for 10 min (A) and 30 min (B), without or with pretreatment with 10, 5, 3, 2, 1, 0.5, 0.3, and 0.1 μm TRO for 10 min. Scale bar: 30 μm. (B, D) Semi‐quantitative analysis of (B) intracellular free Ca²⁺‐derived fluorescence and (D) intracellular ROS levels, as shown in panels A and C, respectively. Variable numbers of cells (12–22) in three different experiments (n = 3) were analyzed for each condition, error bars are SEM. Comparisons between the different groups were performed by Student's t‐test, single (*), double (**), and triple (***) asterisks refer to P values <0.05, <0.01, and <0.001, respectively, relative to untreated cells. Single (§), double (§§), and triple (§§§) symbols refer to P values <0.05, <0.01, and <0.001, respectively, relative to NMDA treatment.

Fig. 5

GluN2B expression, TMA‐DPH fluorescence anisotropy, and N‐methyl‐d‐aspartate‐ (NMDA‐) induced intracellular free Ca²⁺ levels in SH‐SY5Y cells treated with trodusquemine (TRO) and other compounds. (A) Representative confocal scanning microscopy images of SH‐SY5Y cells treated with a primary antibody against GluN2B, labeled with Alexa Fluor‐488, in the presence or absence of pretreatment with 5 μm TRO for 10 min. Scale bar: 30 μm. (B) Semi‐quantitative analysis of GluN2B‐derived fluorescence as shown in panel A. (C) Fluorescence anisotropy (r) of TMA‐DPH incorporated in the phospholipid bilayer of SH‐SY5Y cell membranes left untreated or treated with 2 μm lysophosphatidylcholine (LPC) for 2 h, 10 μm arachidonic acid (AA) for 2 h, 0.5 mM  cholesterol (CHOL) for 3 h, or 5 μm TRO for 10 min. (D) Representative confocal scanning microscopy images of intracellular Ca²⁺ levels following no treatment (left), or treatment with 1.5 mm NMDA for 10 min (right), with or without pretreatment with 5 μm PTP1B inhibitor for 2 h or 5 μm TRO for 10 min. Scale bar: 30 μm. (E) Semi‐quantitative analysis of intracellular free Ca²⁺‐derived fluorescence as shown in panel D. Variable numbers of cells (12–22) in three different experiments (n = 3) were analyzed for each condition, error bars are SEM. Comparisons between the different groups were performed by Student's t‐test, double (**) and triple (***) asterisks refer to P values <0.01 and <0.001, respectively, relative to untreated cells.

Fig. 6

Fluorescence resonance energy transfer (FRET) analysis of the interaction between N‐methyl‐d‐aspartate receptor (NMDAR) and trodusquemine (TRO). (A) Representative confocal scanning microscopy images of SH‐SY5Y cells treated with unlabeled TRO in the presence of anti‐GluN2B antibody labeled with Alexa Fluor‐488 as a donor (D) on the donor channel (left), acceptor channel (middle), and FRET channel (right). Scale bar: 10 μm. (B) Cells treated with TRO labeled with Alexa Fluor‐594 as an acceptor (A) in the presence of unlabeled anti‐GluN2B antibody, on the donor channel (left) acceptor channel (middle) and FRET channel (right). Scale bar: 10 μm. (C) Cells treated with both D and A. From left to right: donor channel, acceptor channel, FRET channel, and colocalization image obtained by overlapping the donor and acceptor channels. Scale bar: 10 μm. (D) Semi‐quantitative analysis of the donor channel in the presence of only D (primary antibody against GluN2B labeled with D), only A (TRO labeled with A), both D and A (primary antibody against GluN2B labeled with D and TRO labeled with A), and the mathematical sum of only D and only A values. Variable numbers of cells (12–22) in three different experiments (n = 3) were analyzed for each condition. Comparisons between the different groups were performed by Student's t‐test, double (**) asterisks refer to P values <0.01. Error bars are SEM.

Fig. 7

Representative confocal scanning microscopy images of SH‐SY5Y cells for the negative control of fluorescence resonance energy transfer (FRET). (A) Cells treated with unlabeled LPS in the presence of primary antibody against GluN2B labeled with Alexa Fluor‐488 (D) on the donor channel (left, excitation 488 nm, emission 499–535 nm), the acceptor channel (middle, excitation 561 nm, emission 640–700 nm) and the FRET channel (right, excitation 488, emission 640–700 nm). Scale bar: 10 μm. (B) Cells treated with LPS labeled with Alexa Fluor‐594 (A) in the presence of unlabeled primary antibody against GluN2B, on the donor channel (left) the acceptor channel (middle) and the FRET channel (right). Scale bar: 10 μm. (C) Cells treated with both D and A. From left to right, the donor channel, the acceptor channel, the FRET channel, and the colocalization image obtained by overlapping the donor and acceptor channels. Scale bar: 10 μm. (D) Semi‐quantitative analysis of the GluN2B derived fluorescence in the donor channel in the presence of only D (primary antibody against GluN2B labeled with D), only A (LPS labeled with A), both D and A (primary antibody against GluN2B labeled with D and LPS labeled with A), and the sum of the only D fluorescence and only A fluorescence. Variable numbers of cells (12–22) in two different experiments (n = 2) were analyzed for each condition. Comparisons between the different groups were performed by Student's t‐test, error bars are SEM. n.s., nonsignificant.

Fig. 8

Patch clamp recordings of N‐methyl‐d‐aspartate receptor‐ (NMDAR‐) activated currents in primary rat hippocampal neurons in the absence or presence of different trodusquemine (TRO) concentrations. (A–D) Representative current traces recorded in cultured primary rat hippocampal neurons exposed twice to NMDA (100 μm, 2 min each, 7‐min interval). Tetrodotoxin (TTX; 200 nm) was co‐applied with NMDA to avoid excessive neuronal firing. Glycine (10 μm) was applied throughout the experiment. The second NMDA application was performed either in the absence (A; n = 16) or presence of TRO, applied also 5 min before the agonist (1, 3, or 10 nm from B to D). Downward deflections are spontaneous synaptic currents. Scale bars: 100 pA, 100 s. (E) Concentration–response curve of TRO‐mediated inhibition of NMDA‐induced current, obtained by normalizing the current recorded during the second NMDA application vs the first (taken as 100%). An IC₅₀ of 5.1 nm (95% confidence limits: 3.0–8.6 nm) was determined. The number in brackets indicates the number of experiments (n) performed in TRO. Error bars: SEM.

Fig. 9

Intracellular free Ca²⁺ levels in SH‐SY5Y cells treated with N‐methyl‐d‐aspartate (NMDA) after treatment with different concentrations of ENT‐03. (A) Representative confocal scanning microscopy images of intracellular Ca²⁺ levels in untreated cells and cells treated with 1.5 mm NMDA for 10 min without or with pretreatment with 10, 5, 3, 2, 1, 0.5, 0.3, and 0.1 μm ENT‐03 for 10 min. Scale bar: 30 μm. (B) Semi‐quantitative analysis of intracellular free Ca²⁺‐derived fluorescence as shown in panel A. Variable numbers of cells (12–22) in three different experiments (n = 3) were analyzed for each condition, error bars are SEM. Comparisons between the different groups were performed by Student's t‐test, single (*), double (**), and triple (***) asterisks refer to P values <0.05, <0.01, and <0.001, respectively, relative to untreated cells. Double (§§) and triple (§§§) symbols refer to P values <0.01 and < 0.001, respectively, relative to NMDA treatment.

Fig. 10

Intracellular free Ca²⁺ levels in SH‐SY5Y cells treated with AMPA and kainate after treatment with different concentrations of trodusquemine (TRO). (A, C) Representative confocal scanning microscopy images of intracellular Ca²⁺ levels in untreated cells and cells treated with (A) 50 μm AMPA or (B) 5 μm kainate for 10 min without or with pretreatment with 10, 5, 3, 2, 1, 0.5, 0.3, and 0.1 μm TRO for 10 min. Scale bar: 30 μm. (B, D) Semi‐quantitative analysis of intracellular free Ca²⁺‐derived fluorescence, as shown in panels A and C, respectively. Variable numbers of cells (12–22) in three different experiments (n = 3) were analyzed for each condition, error bars are SEM. Comparisons between the different groups were performed by Student's t‐test, single (*), double (**), and triple (***) asterisks refer to P values <0.05, <0.01, and <0.001, respectively, relative to untreated cells. Single (§), double (§§), and triple (§§§) symbols refer to P values <0.05, <0.01, and <0.001, respectively, relative to AMPA and kainate treatment.

Fig. 11

Schematic representation of the interaction of trodusquemine (TRO) with cell membranes and N‐methyl‐d‐aspartate receptor (NMDAR). (A) Representation of the insertion and localization of TRO within cell membranes resulting from experimental data and molecular dynamics simulations (adapted from [33]). Distances and angles indicated in the figure were measured experimentally [32]. (B) Schematic representation of the mode of approach of TRO to NMDAR (crystal structure of GluN1b‐GluN2B NMDA receptor structure in nonactive‐2 conformation 5FXI, adapted from [84]). The structure of NMDAR is represented embedded in the membrane, and its domains are indicated as extracellular amino‐terminal domain (ATD), extracellular ligand‐binding domain (LBD) and transmembrane domain (TMD), containing three transmembrane segments (M1, M3, and M4) and a re‐entrant pore loop (M2). The figure implies that TRO binds to the membrane and then approaches NMDAR for binding. It does not imply a precise molecular interaction.