Fig 1: PGM3 knockdown reduces N-glycans and improves social memory in PS19 mice.a, Experimental design: 8-month-old PS19 mice receive hippocampal injections of shScr or shPGM3 and undergo social memory testing 2 weeks later. b,c, MALDI imaging and violin plots of relative intensities for representative complex N-glycans showing reduced brain N-glycan levels in PS19-shPGM3 mice compared with PS19-shScr controls (P<0.0001). d, Schematic of the social memory paradigm across four interaction trials. e, Social memory performance across trials, expressed as time interacting relative to trial 1, showing improved performance in PS19-shPGM3 mice (n = 6 animals) compared with PS19 controls (n = 5 animals). Data are mean ± s.e.m.; P values are indicated (two-tailed t-test for glycan comparisons; two-way repeated-measures ANOVA for behavior). Panel a created in BioRender; Sun, R. https://BioRender.com/y4c87wr (2026); d, Sun R. https://biorender.com/n8icm0r (2026).Source Data.
Fig 2: Targeting hexosamine pathway-dependent N-glycosylation reduces brain N-glycans and improves social memory in 5xFAD mice.a, A schematic of the hexosamine pathway and ER N-glycosylation, highlighting inhibition of PGM3 and OST (by NGI-1). b, The experimental design for 8-month-old 5xFAD mice injected with shScr or shPGM3 and tested for social memory 2 weeks later. c, The experimental design for 5xFAD mice treated with vehicle or NGI-1 and tested for social memory 2 weeks later. d,e, MALDI imaging and violin plots of representative two unique complex N-glycans (N-glycan structure shown on the left) showing reduced brain N-glycan abundance in 5xFAD-shPGM3 mice compared with 5xFAD-shScr controls. f,g, MALDI imaging and violin plots of representative two unique complex N-glycans (N-glycan structure shown on the left) showing decreased levels in NGI-1–treated 5xFAD brains compared with vehicle (veh)-treated 5xFAD mice. h, A schematic of the social memory test across four interaction trials. i, Social memory performance across trials showing improved memory retention in 5xFAD-shPGM3 mice (n = 5 animals) compared with 5xFAD-shScr mice (n = 4 animals). j, Social memory performance in NGI-1-treated (n = 5 animals) and vehicle-treated 5xFAD (n = 5 animals) mice showing partial behavioural rescue with NGI-1. Data are mean ± s.e.m. P values are indicated (two-tailed t-test for glycan comparisons; two-way repeated-measures ANOVA for behaviour). Panels created in BioRender: a, Sun, R. https://biorender.com/wa4lg3j (2026); b, Sun, R. https://biorender.com/n41h483 (2026); c, Sun, R. https://biorender.com/dzklqk3 (2026); h, Sun, R. https://biorender.com/n8icm0r (2026).Source data.
Fig 3: Glucosamine treatment and PGM3 knockdown do not induce astrocytosis or Aβ deposition in WT mice.a, Immunofluorescence images of GFAP and Aβ in brains from WT mice treated with water or glucosamine, showing similar astrocyte staining and absence of Aβ signal between groups. b, GFAP and Aβ staining in WT mice injected with shScr or shPGM3 lentivirus, demonstrating comparable astrocyte morphology and no detectable Aβ accumulation in either condition. Experiments are repeated three times with the same results.Source Data.
Supplier Page from OriGene Technologies for Pgm3 Mouse shRNA Lentiviral Particle (Locus ID 109785)