Fig 2: The GluN2B-GluN2A switch in PSD enrichments from the frontal cortex and hippocampus. a Proportion of GluN2B relative to GluN2A (GluN2B:GluN2A) in PSD enrichments from the frontal cortex and hippocampus. b Levels of GluN1, GluN2A, and GluN2B in PSD fractions from the frontal cortex. c Levels of GluN1, GluN2A, and GluN2B in PSD fractions from the hippocampus. **p ≤ 0.005; ***p < 0.0005; ****p ≤ 0.0001; *****p < 0.00001. PSD postsynaptic density, FCx frontal cortex, HIP hippocampus. Error bars represent standard error of the mean (SEM)

Fig 3: Rd inhibits NMDAR NR2b phosphorylation at Ser-1303 after injuries. (a–c) Western blotting analysis showed that in NMDA- (a) or OGD-injured (b) cultured neurons and MCAO-injured rat brains (c), the levels of NR2b phosphorylation at Ser-1303 but not at Tyr1472 and Ser1480 sites were increased, which was attenuated by Rd treatment markedly. GAPDH was used as an internal control. For comparison easily, shown blots were cropped, and the full-length blots were presented in Supplementary Fig. 7. #p < 0.05 vs. the control or the sham group; *p < 0.05 vs. NMDA, OGD, or MCAO injury group. Error bars = S.E.M.

Fig 4: DTNBP1 genotype differences in pY1472-GluN2B phosphorylation, Fyn, and PLCγ in PSD enrichments of the frontal cortex and hippocampus. a, e, i, m Main effects of DTNBP1 genotype on the pY1472-GluN2B:total GluN2B ratio, abundance of Fyn, pY527-Src phosphorylation, and abundance of PLCγ in PSD enrichments of the frontal cortex. b, f, j, n Developmental trajectories of pY1472-GluN2B:total GluN2B ratio, abundance of Fyn, pY527-Src phosphorylation, and abundance of PLCγ in mice of different DTNBP1 genotypes in the PSD of the frontal cortex. c, g, k, o Main effects of DTNBP1 genotype on the pY1472-GluN2B:total GluN2B ratio, abundance of Fyn, pY527-Src phosphorylation, and abundance of PLCγ in PSD enrichments of the hippocampus. d, h, l, p Developmental trajectories of pY1472-GluN2B:total GluN2B ratio, abundance of Fyn, pY527-Src phosphorylation, and abundance of PLCγ in mice of different DTNBP1 genotypes in the PSD of the hippocampus. *p < 0.05; ***p ≤ 0.0005. N.S. not significant. Error bars represent SEM

Fig 5: Rd does not bind to NMDAR directly. (a) Western blotting results showed successful co-transfection of recombinant NR1/NR2a or NR1/NR2b in HEK293 cells. GAPDH was used as an internal control. For comparison easily, shown blots were cropped, and the full-length blots were presented in Supplementary Fig. 6. (b,c) NMNA-evoked currents in HEK293 cells co-transfected with NR1/NR2a and NR1/NR2b were eliminated by applying NVP-AAM077 (b) and ifenprodil (c), respectively. (d,e) Rd did not affect NMNA-evoked currents in HEK293 cells co-transfected with NR1/NR2a (d) or NR1/NR2b (e). (f) The inhibition percentages of [3H]CGP39653 specific binding to NMDAR in the presence of PPD ginsenosides (Rb1, Rb2, Rd, Rg3, and Rh2) and PPT ginsenosides (Rg1 and Rg2) were revealed by radioligand binding assay. Error bars = S.E.M.