Lysosomal dysfunction in Down syndrome and Alzheimer mouse models is caused by v-ATPase inhibition by Tyr682-phosphorylated APP βCTF

Abstract

Lysosome dysfunction arises early and propels Alzheimer's disease (AD). Herein, we show that amyloid precursor protein (APP), linked to early-onset AD in Down syndrome (DS), acts directly via its β-C-terminal fragment (βCTF) to disrupt lysosomal vacuolar (H⁺)-adenosine triphosphatase (v-ATPase) and acidification. In human DS fibroblasts, the phosphorylated ⁶⁸²YENPTY internalization motif of APP-βCTF binds selectively within a pocket of the v-ATPase V0a1 subunit cytoplasmic domain and competitively inhibits association of the V1 subcomplex of v-ATPase, thereby reducing its activity. Lowering APP-βCTF Tyr⁶⁸² phosphorylation restores v-ATPase and lysosome function in DS fibroblasts and in vivo in brains of DS model mice. Notably, lowering APP-βCTF Tyr⁶⁸² phosphorylation below normal constitutive levels boosts v-ATPase assembly and activity, suggesting that v-ATPase may also be modulated tonically by phospho-APP-βCTF. Elevated APP-βCTF Tyr⁶⁸² phosphorylation in two mouse AD models similarly disrupts v-ATPase function. These findings offer previously unknown insight into the pathogenic mechanism underlying faulty lysosomes in all forms of AD.

Fig. 1.. Assembly of the lysosomal v-ATPase complex is impaired in 2year DS fibroblasts.

(A) Lysosomal pH measured ratiometrically using LysoSensor Yellow/Blue (Y/B) dextran (n = 3, three independent, triplicate). (B to D) Lysosomal v-ATPase activity measured colorimetrically as ATP hydrolysis (B) (n = 4, four independent, duplicate) and fluorometrically (ACMA method) as H⁺ transport (C and D) (n = 3, three independent) using Lyso fractions. (E) Membrane fractions were resolved using native PAGE and immunoblotted with anti-V1B2 antibody. (F) The graph represents relative ratio of full complex divided by total (full plus sub complexes) complexes of v-ATPase (n = 7, seven independent). (G) Immunoblots of v-ATPase subunits distribution in cytosol (Cyto.) and membrane (Memb.) fractions of 2N and DS fibroblasts. Na, K-ATPase 1 (NKA) served as a Memb. marker, and glyceraldehyde-3-phosphate dehydrogenase (GAPDH) served as a Cyto. marker. (H) The graphs represent the ratio of Memb. versus Cyto. band of each v-ATPase subunits (n ≥ 3, at least three independent). (I) Immunoblot of v-ATPase subunit distributions in total lysates (Total) and lysosome-enriched (Lyso.) fractions. LAMP1 served as a marker for lysosomes, and actin served as a loading control for total lysates. (J) The graphs show relative band intensity for each v-ATPase subunit (n ≥ 3, at least three independent). (K and L) Double-immunofluorescence labeling shows colocalization of v-ATPase (V1D) and CTSB (K). Scale bar, 10 μm. Quantification analysis of v-ATPase V1D and lysosomal marker, CTSB, shows colocalization as calculated by Pearson’s correlation coefficient (n ≥ 131 cells, three independent) (L). Quantitative data are presented as mean values with ±SEM, two-tailed unpaired t test (A, B, D, F, and L), ordinary one-way analysis of variance (ANOVA) with Šidák’s multiple comparisons test (H), ordinary two-way ANOVA with Šidák’s multiple comparisons test (J). Each dot represents average value of technical replicates from each independent experiment.

Fig. 2.. Impairment of lysosomal v-ATPase in 2-year DS fibroblasts is dependent on APP expression.

(A) Lysosomal pH of 2N and DS fibroblasts transfected with siRNA for either negative control (siNC) or APP (siAPP) for 72 hours determined by LysoSensor Y/B dextran (n = 7, seven independent, quadruplicate). The immunoblot represents APP levels in 2N and DS fibroblasts after transfection with 40 nM siNC or siAPP. (B) ATP hydrolysis activity of v-ATPase measured using lysosomal fractions from siRNA transfected 2N and DS fibroblasts (n = 3, three independent, duplicate). (C) Membrane fractions of siRNA transfected 2N and DS fibroblasts were resolved using native PAGE and immunoblotted with anti-V1B2 antibody. (D) The graph represents relative ratio of full complex divided by total (full plus sub complexes) complexes of v-ATPase (n = 4, four independent). (E) Immunoblots of v-ATPase subunit distributions in total and Lyso. fractions of siRNA transfected 2N and DS fibroblasts. LAMP1 served as a Lyso marker and GAPDH served as a loading control for total lysates. (F) The graphs show band intensity of each v-ATPase subunit from Lyso fraction (n = 3, three independent). Quantitative data are presented as mean values with ±SEM, ordinary one-way ANOVA with Šidák’s multiple comparisons test (A, B, and D) and ordinary two-way ANOVA with Šidák’s multiple comparisons test (F). Statistical significance between groups is shown by symbols: *2N + siNC versus others, #DS + siNC versus DS + siAPP. Each dot represents average value of technical replicates from each independent experiment.

Fig. 3.. Lysosomal v-ATPase activity is specifically affected by APP-βCTF among metabolites of APP.

(A) Lysosomal pH of 2-year 2N and DS fibroblasts treated with either dimethyl sulfoxide (DMSO) (No Tx), γ-secretase inhibitor, L685,458 (γ-Sec INH; 10 μM), or BACE1 inhibitor (BACE INH; 10 μM) for 24 hours determined by LysoSensor Y/B dextran (n = 5, five independent, triplicate). (B) ATP hydrolysis activity of v-ATPase measured using lysosome fractions from 2-year 2N and DS fibroblasts after treatment indicated inhibitors (n = 3, three independent, duplicate). (C) Membrane fractions from 2-year 2N and DS fibroblasts treated with indicated inhibitors were resolved using the native PAGE and immunoblotted with anti-V1B2 antibody. (D) The graph represents relative ratio of full complex divided by total (full plus sub complexes) complexes of v-ATPase (n = 3, three independent). (E and F) Double-immunofluorescence labeling shows colocalization of V1D and CTSB in 2N and DS fibroblasts treated with indicated inhibitors (E). Scale bar, 10 μm. Quantification analysis shows colocalization as calculated by Pearson’s correlation coefficient (n ≥ 123 cells, three independent) (F). (G) Lysosomal pH of 2-year 2N fibroblasts transfected with either empty vector (Vec) or pcDNA-APP-βCTF (βCTF) for 48 hours determined by LysoSensor Y/B dextran (n = 3, three independent, triplicate). (H) Membrane fractions from 2-year 2N transfected with either empty vector (−) or pcDNA-APP-βCTF (+) for 48 hours were resolved using the native PAGE and immunoblotted with anti-V1B2 antibody. (I) The graph represents relative ratio of full complex divided by total complexes of v-ATPase (n = 5, five independent). Quantitative data are presented as mean values with ±SEM, ordinary one-way ANOVA with Šidák’s multiple comparisons test (A, B, D, and F) and two-tailed unpaired t test (G and I). Statistically significance between groups is shown by symbols: *2N No Tx versus others, #DS No Tx versus DS + inhibitor. Each dot represents average value of technical replicates from each independent experiment.

Fig. 4.. The interaction between APP-βCTF v-ATPase V0 subunits increased in the lysosome-enriched fraction from 2-year DS fibroblasts.

(A to C) Cell lysates (A and B) or Lyso fraction (C) of 2N and DS fibroblast were immunoprecipitated (IP-ed) with anti-APP (C1/6.1) antibody, followed by immunoblotting (IB) with antibodies of v-ATPase V0 subunits (A and C) and V1 subunits (B and C). The values at the bottom of the left IP-ed blot indicate the relative intensities of IP-ed V0a1 and V0d1 normalized by IP-ed APP. The quantification in these experiments is confirmed in two other experiments. (D) Cell lysates of 2N and DS fibroblast were IP-ed with anti-V0a1 antibody, followed by IB with anti-APP (6E10). Nonspecific immunoglobulin G (IgG) used as a control IP against 2N cells. The values at the bottom of the left IP blot indicate the relative intensities of IP-ed APP-βCTF normalized against IP-ed V0a1. Actin and GAPDH served as a loading control and LAMP2 served as a lysosomal marker. (E) Left: Schematic diagram of in situ PLA. PLA performed using V0a1 antibody and C1/6.1 for APP C terminus in 2N and DS fibroblasts. The red box represents signal of PLA (Duolink). Right: Representative fluorescent microscopy image of PLA signals demonstrating association of APP with the V0a1. Scale bar, 10 μm. (F) The graph show number (#) of dots per cell (n ≥ 173 cells, three independent). Quantitative data is presented as mean values with ±SEM, two-tailed unpaired t test (F). Each dot represents number of dot per field of each independent experiment.

Fig. 5.. pYENPTY motif of APP-βCTF interrupt the lysosomal pH via interaction with V0a1 that form a structural pocket within the complex.

(A) Cell lysates were incubated with varying amounts of each peptide for 24 hours, then IP with anti-APP (C1/6.1) antibody, followed by IB with anti-V0a1 antibody. (B) Lysosomal pH measured after treated with either DMSO (−) or 5 μM peptides for 24 hours (n = 3, three independent, triplicate). (C) Membrane fractions from 2-year 2N treated with either DMSO (−) or 5 μM Pep-2 (+) for 24 hours were resolved using the native PAGE. (D) The graph represents relative ratio of full complex divided by total complexes of v-ATPase (n = 3, three independent). (E) Cell lysates were IP-ed with anti-pY⁶⁸²APP antibody, followed by IB with anti-V0a1 antibody. The values at the bottom of the left IP blot indicate the relative intensities of IP-ed V0a1 normalized against IP-ed APP-CTFs (A) or pY⁶⁸²βCTF (E). The quantification in this experiment is confirmed in two other experiments. (F) Schematic diagram of binding with V0a1 and pY⁶⁸²APP-βCTF (left). The surface of the pYENPTY binding pocket of V0a1 (right) is labeled with F196 and R167 of V0a1 subunits. (G) The binding poses of the pYENPTY in the pocket of V0a1. Hydrogen bond and salt bridge interactions by S151, S153, R198, and T250 are labeled with yellow dash lines. The methylene units of R167 and F170 form π-hydrophobic contacts with the aromatic ring of the pY⁶⁸². Proline at the pY+3 position has π-hydrophobic interaction with F196. The pYENPTY is yellow and the unsolved region is light blue. Quantitative data are presented as mean values with ±SEM, ordinary one-way ANOVA with Šidák’s multiple comparisons test (B), two-tailed unpaired t test (D). Each dot represents average value of technical replicates from each independent experiment.

Fig. 6.. Decreased levels of phosphorylated Tyr⁶⁸² APP-βCTF rescues v-ATPase dysfunction in 2-year DS fibroblasts.

(A) Immunoblots of phospho-Y⁶⁸²APP-βCTF and APP-βCTFs distribution in total lysates of 2N and DS fibroblasts. (B and C) Lysosomal pH (C) (n = 5, five independent, triplicate) and ATP hydrolysis activity of v-ATPase using lysosomal fractions (D) (n = 4, four independent, triplicate) measured after treated with either DMSO (−) or 2 μM a Src family of nonreceptor tyrosine kinase inhibitor (AZD0530) for 24 hours. (D and E) Double-immunofluorescence labeling shows colocalization of V1D and CTSB in DMSO- or AZD0530-treated fibroblasts (D). Scale bar, 10 μm. Quantification shows colocalization in DMSO- or AZD0530-treated fibroblasts as calculated by Pearson’s correlation coefficient (n ≥ 118 cells, three independent) (E). (F) After treated with DMSO (−) or AZD0530 (+), cell lysates were IP with anti-APP (6E10) antibody, followed by IB with anti-V0a1 antibody. (G and H) Lysosomal pH (G) (n = 6, six independent, quadruplicate) and ATP hydrolysis activity of v-ATPase using lysosomal fractions (H) (n = 4, four independent, triplicate) measured after transfected with siRNA for either negative control (−) or 100 nM siFyn (+) for 48 hours. Immunoblot represents Fyn kinase protein levels. (I) Membrane fractions from siRNA transfected fibroblasts were resolved using the native PAGE. (J) The graph represents relative ratio of full complex divided by total complexes of v-ATPase. (K and L) Double-immunofluorescence labeling shows colocalization of V1D and CTSB in siRNA transfected fibroblasts (K). Scale bar, 10 μm. Quantification shows colocalization in siRNA transfected fibroblasts, as calculated by Pearson’s correlation coefficient (n ≥ 139 cells, three independent) (L). Quantitative data are presented as mean values with ±SEM, ordinary one-way ANOVA with Šidák’s multiple comparisons test. Statistically significance between groups is shown by symbols: *2N versus others, #DS versus DS + AZD0530/siFyn. Each dot represents average value of technical replicates from each independent experiment.

Fig. 7.. Defective lysosomal v-ATPase function in Ts2 mouse model and human DS brain.

(A and B) Representative fluorescence images of tfLC3 color changes of TRGL or Ts2xTRGL adult brain (A) and respective quantification of puncta representing ALs (fully or less acidified) (B). For data analysis, color change of tfLC3-positive vesicles was calculated with CTSD colabeling. Bar colors denote the colors of puncta; purple bars indicate tfLC3 with CTSD (fully acidified ALs, quenched eGFP), and white bars indicate tfLC3 with CTSD (less acidified ALs, unquenched eGFP) (n = 504 neurons, 6 TRGL mice; n = 537 neurons, 6 Ts2xTRGL mice). (C) ATP hydrolysis activity of v-ATPase was measured using Lyso fraction (15 to 18) of the iodixanol step gradient from adult mice (n = 5, triplicate). (D) Adult mouse brain from control (2N) and Ts2 homogenates were fractionated through an iodixanol step gradient. Each fraction [lysosomal enriched fraction (Lyso): 15 to 18; exclude lysosomal enriched fraction (Lyso−): 1 to 14 and 19 to 22] and post-nuclear supernatant (PNS) were resolved using the SDS-PAGE and immunoblotted with anti-V1A, anti-V1C1, anti-V1D, anti-V0a1, and anti-V0d1 antibodies. (E) The graphs represent levels of v-ATPase subunits from each fraction (n = 4). (F and H) Membrane fractions from mouse brain (F) or frozen human brain tissue (H) were resolved using the native. (G and I) The graph represents relative ratio of full complex divided by total (full plus partial complexes) complexes of v-ATPase (n = 5). Quantitative data (B, C, E, G, and I) are presented as mean values with ±SEM, two-tailed unpaired t test (C, G, and I) and ordinary one-way ANOVA with Šidák’s multiple comparisons test (B and E). Each dot corresponds to average value of technical replicates from one mouse.

Fig. 8.. Reduction in levels of phosphorylated Tyr⁶⁸² APP-βCTF in vivo reverses v-ATPase dysfunction in brains of DS model mice.

(A and B) Immunoblot (A) and the graphs (B) represent pY⁶⁸²APP-βCTF and CTFs levels in mouse brain (n = 7). (C and D) Immunoblot (C) and the graphs (D) represents pY⁶⁸²APP-βCTF and CTFs levels after treated with either vehicle (−) or AZD0530 (+) for 4 weeks (n = 5). GAPDH served as a loading control. (E and F) Representative fluorescence images of tfLC3 color changes of adult brain after treatment with vehicle (Veh) or AZD0530 for 3 weeks (E) and respective quantification of puncta representing ALs (fully or less acidified) (n = 452 neurons, 4 TRGL; n = 491 neurons, 4 AZD0530-treated TRGL; n = 476 neurons, 4 Ts2xTRGL; n = 484 neurons 4 AZD0530-treated Ts2xTRGL) (F). (G) ATP hydrolysis activity of v-ATPase measured using Lyso fraction (15 to 18) of iodixanol step gradient from adult mouse brain after treatment with vehicle (−) or AZD0530 (+) for 4 weeks (n = 4). (H) Membrane fractions from AZD0530-treated adult mouse brain were resolved using native PAGE. (I) The graph represents relative ratio of full complex divided by total (full plus partial complexes) complexes of v-ATPase (n = 4). (J) Homogenates from AZD0530-treated adult mice brain were fractionated through an iodixanol step gradient. The Lyso fraction and the PNS were immunoblotted using indicated antibodies. (K) The graphs represent levels of lysosomal v-ATPase subunits (n = 5 TRGL; n = 4 Ts2xTRGL; n = 4 AZD0530-treated Ts2xTRGL). Quantitative data are presented as mean values with ±SEM, two-tailed unpaired t test (B) and ordinary one-way ANOVA with Šidák’s multiple comparisons test (D, F, G, I, and K). Statistically significance between groups is shown by symbols: *2N + Veh versus others, #DS + Veh versus DS + AZD0530. Each dot corresponds to average value of technical replicates from one mouse.

Fig. 9.. Defective v-ATPase dysfunction in APP-based AD model mouse.

(A) Lysosomal v-ATPase activity measured colorimetrically as ATP hydrolysis with and without ConA. Activity assays were performed on lysosomal enriched fraction (15 to 18) of iodixanol step gradient from 6-month Tg2576 female adult brain pretreated with o-vanadate to minimize nonspecific ATPase activity (n = 6, triplicate). (B) Immunoblot represents pY⁶⁸²APP-CTFs and APP-CTFs protein levels in 6-month female Tg2576. Actin served as a loading control. (C) The graph shows relative band intensity of pY⁶⁸²APP-βCTF proteins (n = 4). (D) Membrane fractions from 6-month Tg2576 female adult brain were resolved using the native PAGE and immunoblotted with anti-V1B2 antibody. (E) The graph represents relative ratio of full complex divided by total (full plus partial complexes) complexes of v-ATPase (n = 3). (F) Lysosomal v-ATPase activity measured colorimetrically as ATP hydrolysis with and without ConA. Activity assays were performed on lysosomal enriched fraction (15 to 18) of iodixanol step gradient from 6-month 5xFAD female adult brain pretreated with o-vanadate to minimize nonspecific ATPase activity (n = 4, triplicate). (G) Immunoblot represents pY⁶⁸²APP-CTFs and APP-CTFs protein levels in 6-month female 5xFAD. (H) The graphs show band intensity of pY⁶⁸²APP-βCTF protein (n = 4). (I) Membrane fractions gradient from 6-month 5xFAD female adult brain were resolved using the native PAGE and immunoblotted with anti-V1B2 antibody. (J) The graph represents relative ratio of full complex divided by total (full plus partial complexes) complexes of v-ATPase (n = 3). Quantitative data (A, C, E, F, H, and J) are presented as mean values with ±SEM, two-tailed unpaired t test. Each dot corresponds to average value of technical replicates from one mouse.