Fig 2: Brain atrophy, neuronal density, and curvature are reduced by LRGT treatment. (A,B) Brain weight, cortex and hippocampal size, (C,D) NeuN/DAPI ratio, and (E,F) axonal curvature ratio were analyzed in all four genotypes (Control, APP/PS1, db/db, and APP/PS1xdb/db) under study and compared with animals by week 26, after 20 weeks on LRGT treatment (500 µg/kg/day) (complete statistical analysis included as a Supplementary Material). (A) Long-term LRGT limited brain weight loss (p < 0.01 vs. Control, Control-LRGT, APP/PS1, APP/PS1-LRGT, db/db, and APP/PS1xdb/db). Cortical size was significantly improved by the LRGT treatment (p < 0.01 vs. Control, Control-LRGT, APP/PS1, APP/PS1-LRGT, db/db-LRGT, and APP/PS1xdb/db-LRGT; ††p < 0.01 vs. Control, Control-LRGT, APP/PS1, and APP/PS1-LRGT). No differences were observed in the hippocampus. Data are representative of 4–5 animals. (B) Illustrative example of cresyl violet staining showing reduced cortical size in db/db and APP/PS1xdb/db mice. Scale bar = 200 um. (C) Neuronal density was reduced in the proximity of amyloid plaques in APP/PS1xdb/db mice, and LRGT ameliorated this situation (††p = 0.008 vs. APP/PS1 and APP/PS1-LRGT). A similar profile is observed in cortical and hippocampal areas with no amyloid plaques (p < 0.001 vs. Control, Control-LRGT, APP/PS1, APP/PS1-LRGT, db/db-LRGT, and APP/PS1xdb/db-LRGT; ##p < 0.01 vs. Control and Control-LRGT; + + p < 0.01 vs. Control, Control-LRGT, APP/PS1, APP/PS1-LRGT, db/db, and db/db-LRGT). Data are representative of five animals. (D) Illustrative example of NeuN (red) and DAPI (blue) staining in areas located in the proximity of amyloid plaques (TS staining, green) and in areas without amyloid plaques. Zoom-in images of representative regions are marked by white squares and presented next to the original image, including areas with and without amyloid plaques. Scale bar = 50 µm, insets scale bar = 25 µm. (E) LRGT reduced curvature ratio in the proximity of amyloid plaques (**p = 0.004 vs. rest of the groups) and in areas free of amyloid plaques (**p < 0.01 vs. rest of the groups, ##p < 0.01 vs. Control and Control-LRGT, p < 0.01. vs. Control) (complete statistical analysis included in Supplementary Material). Data are representative of five animals per group (308–920 neurons/group). (F) Illustrative examples of SMI-312 (red) and TS (green) staining in the proximity of and far from amyloid plaques (yellow lines mark representative neurites). Scale bar = 10 µm.

Fig 3: Long-term liraglutide (LRTG) treatment ameliorates metabolic alterations in T2D and AD-T2D mice. (A) Non-fasting plasma glucose, (B) insulin, and (C) body weight were measured every 4 weeks in controls, APP/PS1, db/db and APP/PS1xdb/db mice treated with vehicle or LGRT 500 µg/kg for 20 weeks (from weeks 6 to 26) (complete statistical analysis included as a Supplementary Material). (A) LRGT significantly reduced postprandial glucose levels in diabetic mice: (##p < 0.01 Control, Control-LRGT, APP/PS1, APP/PS1-LRGT; p < 0.01 vs. Control; ††p < 0.01 Control, Control-LRGT, APP/PS1, APP/PS1-LRGT, db/db-LRGT, and APP/PS1xdb/db-LRGT; p < 0.001 vs. Control, Control-LRGT, APP/PS1, APP/PS1-LRGT, and db/db-LRGT; p < 0.01 vs. Control-LRGT). (B) Insulin levels were maintained by long-term LRGT treatment. At 6 weeks of age, immediately before the commencement of LRGT treatment, insulin levels were significantly increased in APP/PS1xdb/db mice (##p < 0.01 Control, Control-LRGT, APP/PS1, APP/PS1-LRGT; p < 0.01 vs. Control, Control-LRGT, APP/PS1, APP/PS1-LRGT, db/db, and APP/PS1xdb/db; p < 0.01 vs. Control, Control-LRGT, APP/PS1, APP/PS1-LRGT, and APP/PS1xdb/db). (C) LRGT maintained body weight in APP/PS1-LRGT (##p < 0.01 vs. Control, Control-LRGT, APP/PS1, and APP/PS1-LRGT; ##p < 0.01 vs. Control, Control-LRGT, APP/PS1, and APP/PS1-LRGT; p < 0.01 vs. Control, Control-LRGT, APP/PS1, APP/PS1-LRGT, and APP/PS1xdb/db). Data are representative of 5–12 animals, and differences were detected by one-way ANOVA followed by Tukey’s b or Tamhane test.

Fig 4: PSH homology model.(A) Alignment of the modeled holo form (model 2) of PSH (blue) with APP C83 substrate (orange) and the crystal structure of PSH (PDB 4HYG) in the apo form (green) in side view (left panel) and top view (right panel). An overlay of all three models and their RMSD values are provided in Figure 4—figure supplement 1A and B. (B) Schematic representation of the hydrogen bonds formed between ß3 of the substrate (orange), ß2, and Q272, respectively, of PSH (blue). (C) Interaction of TMD3 and TMD6a in the C83-bound PSH model through residues R70 and E181. (D) Interaction of TMD3 and TMD6a through residues Y159 and R278 in the C83-bound ?-secretase cryo-EM structure (PDB 6IYC). (E) Analysis of WT and mutant PSH activity in DDM and POPC environment after incubation with C100-His6 substrate at 37 °C overnight by immunoblotting for AICD (Y188) and Aß (2D8). Immunoblotting of PSH (6F4) was performed to control for PSH levels. The asterisks mark two substrate degradation bands, which are independent of PSH cleavage. Figure 4—source data 1.Immunoblot images (raw and annotated) of cleavage assays (Source data for Figure 4E).

Fig 5: Cleavage of APP C83 by PSH.(A) Analysis of PSH activity in DDM and POPC environment after incubation with C83-His6 and C100-His6 substrates at 37 °C and pH 7.0 by immunoblotting for AICD (penta-His) and p3 (Aß (22-35)). Immunoblotting of PSH (6F4) was performed to control for PSH levels. (B, C) p3-40/p3-42 ratio (B) and Aß40/Aß42 ratio (C) from PSH activity assays in DDM (red) and POPC (blue) environment analyzed by ECL-IA. Quantitative data are represented as mean ± standard deviation (SD) (n=3 biological replicates). Source data are shown in Figure 3—source data 1. Figure 3—source data 1.Immunoblot images (raw and annotated) of cleavage assays (Source data for Figure 3A). Figure 3—source data 2.Raw values of p3 and Aß concentrations measured in the ECL-IA and calculated p3-40/p3-42 and Aß40/Aß42 ratios (Source data for Figure 3B, C).