Evidence that the rab5 effector APPL1 mediates APP-βCTF-induced dysfunction of endosomes in Down syndrome and Alzheimer's disease

Abstract

β-Amyloid precursor protein (APP) and its cleaved products are strongly implicated in Alzheimer's disease (AD). Endosomes are highly active APP processing sites, and endosome anomalies associated with upregulated expression of early endosomal regulator, rab5, are the earliest known disease-specific neuronal response in AD. Here, we show that the rab5 effector APPL1 (adaptor protein containing pleckstrin homology domain, phosphotyrosine binding domain and leucine zipper motif) mediates rab5 overactivation in Down syndrome (DS) and AD, which is caused by elevated levels of the β-cleaved carboxy-terminal fragment of APP (βCTF). βCTF recruits APPL1 to rab5 endosomes, where it stabilizes active GTP-rab5, leading to pathologically accelerated endocytosis, endosome swelling and selectively impaired axonal transport of rab5 endosomes. In DS fibroblasts, APPL1 knockdown corrects these endosomal anomalies. βCTF levels are also elevated in AD brain, which is accompanied by abnormally high recruitment of APPL1 to rab5 endosomes as seen in DS fibroblasts. These studies indicate that persistent rab5 overactivation through βCTF-APPL1 interactions constitutes a novel APP-dependent pathogenic pathway in AD.

Figure 1

β-Cleaved carboxy-terminal fragment of APP (βCTF) activates rab5 on endosomes and increases endosome size. (a) Rab5 activation, reflected by the rate of fluorescence recovery after photobleaching (FRAP), is measured as the rate at which photobleached GTP-bound rab5 (activated rab5) on individual endosomes is replaced by fluorescent cytosolic GDP-rab5 (inactive rab5). The rate of FRAP for endosomal GFP-rab5 is significantly decreased when βCTF levels are increased by overexpressing wild-type APP (APPwt) in N2a cells (N2a APPwt) or blocking βCTF cleavage with a 10 μm γ-secretase inhibitor L685 458 (N2a+L685), as compared with control cells expressing βCTF at endogenous levels (N2a) or transfected with an APP mutant construct unable to be cleaved to βCTF (N2a APPmv). The FRAP rate after transfecting a dominant-active mutant GFP-rab5 Q79L as a positive control is extremely reduced, indicating persistent rab5 activation. The summary graph reflects by the percent of average fluorescence recovery at 270 s after photobleaching in each condition (n=20 endosomes, one endosome per cell, total 20 cells, mean±s.e.m., one-way analysis of variance (ANOVA), Tukey's test, ***P<0.001). (b) βCTF activates rab5 in human embryonic kidney 293 (HEK293) cells. APPwt overexpression increases the levels of active rab5 (GTP-bound rab5) detected with a GTP-rab5-specific antibody (first lane), whereas APPmv mutant has no effect (second lane), as compared with untransfected control (ctrl) cells (third lane). A bar graph presents mean immunoreactive GTP-rab5 signal±s.e.m. for three separate immunoblot experiments (representative blot shown) (**P<0.01, one-way ANOVA, Tukey's test). Ten percent of cell lysates is used as total rab5. (c) βCTF induces endosomal enlargement in N2a cells. Cross-sectional area of rab5-positive endosomes is increased by APPwt overexpression or 10μM L685 458 (lanes 2 and 3), which raise αCTF and βCTF levels compared with those in ctrl cells (lane 1). APPmv mutant expression elevates αCTF levels but not βCTF levels (lane 4) and does not enlarge endosomes. Each graph bar indicates the number of measured endosomes from 20 cells (*P<0.05 and **P<0.01, respectively, mean±s.e.m., one-way ANOVA, Tukey's test).

Figure 2

β-Cleaved carboxy-terminal fragment of APP (βCTF) generated on endosomes binds via its YENPTY domain to the PTB domain of APPL1 (adaptor protein containing pleckstrin homology domain, phosphotyrosine binding domain and leucine zipper motif). (a) βCTF interacts with APPL1: immunoblot analysis shows co-immunoprecipitation (co-IP) of βCTF and APPL1 from cell lysates of human embryonic kidney 293 (HEK293) cells overexpressing wild-type human APP (APPwt) and APPL1 using antibodies against APPL1 or against APP (APP C1/6.1 antibody is raised against amino-acid residues 676–695 of human APP695 and recognizes full-length APP (flAPP), αCTF and βCTF). (b) APPL1 interaction with APP requires β-secretase cleavage: APPL1 does not co-IP with a β-cleavage-incompetent mutant APP (APPmv) from lysates of HEK293 cells overexpressing both proteins. (c) βCTF–APPL1 interaction requires the PTB domain of APPL1: βCTF does not co-IP with a mutant APPL1 (APPL1ΔPTB) lacking the PTB domain when this mutant and APPwt are overexpressed in HEK293 cells. (d) APPL1 interacts via the YENPTY domain of βCTF: APPL1 does not co-IP with an APP mutant containing a YENPTY domain variant (APP_AENATA) from lysates of HEK293 cells overexpressing both proteins. (e) FE65 and APP-BP1 are immunoprecipitated with APP using 6E10 antibody, raised against the amino acids 1–17 of human Aβ, detects human flAPP, βCTF and Aβ but not with APPL1 from lysates of HEK293 cells overexpressing APPwt and APPL1. (f) APPL1 selectively interacts with endocytosed βCTF: in N2a cells overexpressing APPwt and APPL1, APPL1 does not co-IP with βCTF when βCTF generation on endosomes is prevented by blocking endocytosis with MiTMAB (M), a dynamin inhibitor. Treatment with promyristic acid (P), an inactive form of MiTMAB, a negative control for endocytic blockade, does not prevent APPL1wt co-IP with βCTF. In all co-IP experiments, 10% of total cell lysates is used as an input standard. (+) and (−) indicates the presence and absence of the antibody used in IP, respectively.

Figure 3

APPL1 (adaptor protein containing pleckstrin homology domain, phosphotyrosine binding domain and leucine zipper motif) mediates β-cleaved carboxy-terminal fragment of APP (βCTF)-induced endosomal enlargement and rab5 activation. (a) APPL1 small interfering RNA (siRNA) knockdown blocks wild-type human APP (APPwt)-mediated Rab5 activation: APPwt overexpression in human embryonic kidney 293 (HEK293) cells significantly elevates the ratio of activated (GTP-bound) rab5 to total rab5 relative to the ratio in cells lacking APPwt overexpression (lane 2 vs lane 1) (**P<0.01, one-way analysis of variance (ANOVA), Tukey's test). siRNA knockdown of APPL1 prevents rab5 activation in the presence of APPwt overexpression (fourth lane) but has no effect on rab5 activation when APP is expressed at endogenous levels (third lane). The bar graph depicts mean±s.e.m. for four separate experiments (one representative immunoblot shown). (b) APPL1 mediates APPwt-induced rab5 enlargement. Increases in average size (cross-sectional area) of GFP-rab5 endosomes induced by APPwt overexpression in N2a cells are reversed with APPL1 siRNA treatment but not with scrambled control siRNA (ctrl) (***P<0.001, one-way ANOVA, Tukey's test). APPL1 siRNA does not affect endosomal size in cells expressing APP at endogenous levels. Each graph bar indicates the number of measured endosomes from 20 cells and shows values as mean±s.e.m. Ten percent of total cell lysates is used as a standard. (c) APPL1 mediates APPwt-induced rab5 activation on endosomes: APPL1 siRNA, but not control siRNA, blocks the slowed fluorescence recovery after photobleaching (FRAP) of GFP-rab5 in N2a cells overexpressing APPwt but does not alter endosomal rab5 activation (FRAP) in cells expressing APP at normal endogenous levels endosomes. The summary graph reflects by the percent of average fluorescence recovery at 270 s after photobleaching in each condition (n=20 endosomes, one endosome per cell, total 20 cells, mean±s.e.m., one-way ANOVA, Tukey's test, *P<0.05 and *P<0.01). (d) The PTB domain of APPL1 is required for βCTF-induced rab5 enlargement. The increased GFP-rab5 endosome size induced by APPwt and APPL1wt overexpression in N2a cells is prevented when either the APPmv mutant or the APPL1ΔPTB (PTB domain deleted APPL1) mutant is substituted for APPL1wt as the corresponding overexpressed construct. Each graph bar indicates number of measured endosomes from 20 cells (*P<0.05, mean±s.e.m., one-way ANOVA, Tukey's test). (e) The PTB domain of APPL1 is required for βCTF-induced GFP-rab5 activation on endosomes: FRAP in N2a cells. Overexpression of APPwt and APPL1wt in N2a cells reduces the FRAP on endosomes, indicating rab5 activation, whereas APPmv mutant or APPL1ΔPTB overexpression do not alter the FRAP relative to that in control cells. The summary graph reflects the percent of average fluorescence recovery at 270 s after photobleaching in each condition (n=20 endosomes, one endosome per cell, total 20 cells, mean±s.e.m., one-way ANOVA, Tukey's test, ***P<0.001).

Figure 4

APPL1 (adaptor protein containing pleckstrin homology domain, phosphotyrosine binding domain and leucine zipper motif) mediates β-cleaved carboxy-terminal fragment of APP (βCTF)-induced enlargement and impaired transport of endosomes in neurons. (a) Overexpression of wild-type human APP (APPwt) or rab5 increases average cross-sectional areas of individual rhoB-positive endosomes in primary cultures of mouse cortical neurons (one-way analysis of variance (ANOVA), Tukey's test, *P<0.05 and **P<0.01, respectively). (b) Rab5 or APPwt overexpression reduces average transport velocities of rhoB-positive endosomes in cultured neurons (one-way ANOVA, Tukey's test, ***P<0.001). (c) Rab5 or APPwt overexpression has no effect on transport interruption of rhoB-positive endosomes in cultured neurons. (d–f) Raising βCTF levels by APPwt overexpression or exposure to the γ-secretase inhibitor L685 458 (L685) in cultured neurons increases average cross-sectional area of rab5-positive endosomes (d), decreases average speeds of rab5-positive endosomes (e) and increases numbers of stationary rab5 endosomes (f) (one-way ANOVA, Tukey's test, **P<0.01 and ***P<0.001). Overexpressing the β-secretase cleavage-incompetent APPmv mutant does not alter these parameters (d–f). (g–i) Treatment with APPL1 small interfering RNA (siRNA) but not a control scrambled siRNA prevents from the increased cross-sectional area of rab5 endosomes (g), the reduced rab5 endosome transport velocity (h) and the transport interruption of rab5 endosomes (i.e. increased stationary endosome number) (i) that are induced in primary cultures of mouse cortical neurons by APPwt overexpression (unpaired two-tailed t-test, **P<0.01 and ***P<0.001). (j–l) The PTB domain of APPL1 is required for βCTF-induced endosome alterations in primary cortical neurons. Increased endosomal cross-sectional area (j), lowered transport rate of endosomes (k) and transport interruption of rab5 endosomes (elevated stationary endosomes) (l) are induced by overexpressing APPwt and APPL1wt (one-way ANOVA, Tukey's test, **P<0.01 and ***P<0.001) but are not induced by overexpression of either APPL1wt with APPmv or APPwt with APPL1ΔPTB (PTB domain deleted APPL1 mutant). Results are presented as mean±s.e.m.

Figure 5

APPL1 (adaptor protein containing pleckstrin homology domain, phosphotyrosine binding domain and leucine zipper motif)- and βCTF (β-cleaved carboxy-terminal fragment of APP)-dependent endosomal abnormalities in Down syndrome (DS) fibroblasts and Alzheimer's disease (AD) brain. (a) Greater APPL1 colocalization with rab5 in DS fibroblasts compared with control (ctrl) fibroblasts is seen by immunocytochemistry as shown in representative images of cells double immunolabeled with antibodies to APPL1 (green) and rab5 (red). The graph shows a higher APPL1 and rab5 colocalization coefficient (R) in DS fibroblasts, as calculated by Pearson's correlation coefficient in 30 DS and 30 control cells (mean±s.e.m., unpaired two-tailed t-test *P<0.05). (b) APPL1 mediates endosomal enlargement in DS cells. Treatment with APPL1 small interfering RNA (siRNA), but not scrambled control siRNA, blocks the increase in cross-sectional area of endosomes in DS fibroblasts but has no effect on normal endosomes in age-matched 2N ctrl fibroblasts (n=30 cells, one-way analysis of variance (ANOVA), Tukey's test, mean±s.e.m., *P<0.05 and ***P<0.001). A representative blot with a glyceraldehyde 3-phosphate dehydrogenase (GAPDH) loading control and graphic quantitation from three separate experiments are shown. (c) APPL1 mediates the abnormally elevated endocytosis in DS fibroblasts. Fluorescent horseradish peroxidase (HRP)-positive puncta are higher in DS fibroblasts at 30 min after the addition of HRP to the medium compared with control cells and are reduced in DS cells treated with APPL1 siRNA compared with cells treated with scrambled (ctrl) siRNA, whereas APPL1 siRNA has no effect on HRP uptake in control fibroblasts. Quantitative analysis of fluorescence at 0, 5, 15 and 30 min after the addition of HRP to the medium shows significantly reduced HRP uptake after APPL1 siRNA compared with cells treated with scrambled (ctrl) siRNA (ctrl–ctrl-siRNA; 15, 15, 20 and 19 cells; ctrl-siRNA: 16, 28, 17 and 25 cells; DS-ctrl-siRNA: n=26, 10, 18 and 36 cells; DS-APPL1 siRNA: n=12, 16, 39 and 42 cells in 0, 5, 15 and 30 min from two experiments, mean±s.e.m., one-way ANOVA, Tukey's test, mean±s.e.m., ***P<0.001). (d) APPL1 mediates abnormally increased nuclear factor-κB (NF-κB) pathway activation in DS fibroblasts. Functional evidence for APPL1-mediated rab5 activation on endosomes is reflected in greater APPL1/rab5-endosomal-mediated nuclear localization of the NF-κB p65 subunit (p65/RelA). Levels of nuclear p65/RelA normalized to nuclear histone H3 as a loading control (third lane) are elevated in DS fibroblasts compared with those in control (ctrl) cells (first lane). This elevation is reversed by APPL1 siRNA (fourth lane) but not scrambled (ctrl) siRNA as shown on an immunoblot representative of three experiments quantified in the graph (mean±s.e.m., two-way ANOVA with interaction, *P<0.05). (e) βCTF levels, but not full-length APP (flAPP) or αCTF levels, are significantly elevated in AD brain compared with age-matched control (ctrl) brain, as shown on an immunoblot representative of three experiments quantified in the graph. C1/6.1 antibody was used to measure βCTF and αCTF, which are normalized to the level of APP (n=13 control and 13 AD brains, mean±s.e.m., unpaired two-tailed t-test, *P<0.05). Identities of βCTF (C1–C4) and αCTF (C5) were confirmed using 6E10 (data not shown). Actin is used as a loading control. (f) Membrane-associated APPL1 levels are increased in AD brains compared with control (ctrl) brains as shown on an immunoblot representative of three experiments quantified in the graph as a ratio of membrane-associated APPL1 in a 50 μg pellet (P) of brain homogenate to the APPL1 level in the 50 μg supernatant (S) (n=13 control and 13 AD brains, mean±s.e.m., unpaired two-tailed t-test, *P<0.05). (g) APPL1 association with rab5-positive endosomes is increased in AD brain. Double immunofluorescence labeling confirms rab5-positive endosome enlargement in neocortical pyramidal neurons of AD brain and demonstrates greater APPL1 colocalization (arrowheads) in these enlarged endosomes as compared with neuropathologically normal control brains. The graph shows a higher APPL1 and rab5 colocalization coefficient (R) in AD brains, as calculated by Pearson's correlation coefficient (n=90 cells, mean±s.e.m., unpaired two-tailed t-test, *P<0.05). (h) Recruitment of APPL1 to rab5 endosomes is higher in neuronal endosomes of AD brain compared with that in control (ctrl) brain, as reflected by an increased percent of rab5 endosome surface area occupied by APPL1-immunoreactive signal quantified from images similar to those in panel g. Comparison of rab5 endosomes of different size ranges demonstrates increased APPL1 colocalization on endosomes of greater size in AD brains (n=90 cells; 3824, 408 and 224 endosomes for control, 3194, 269 and 156 endosomes for AD in each size bin, mean±s.e.m., unpaired two-tailed t-test, ***P<0.001).