p62 improves AD-like pathology by increasing autophagy.

The multifunctional protein p62 is associated with neuropathological inclusions in several neurodegenerative disorders, including frontotemporal lobar degeneration, amyotrophic lateral sclerosis and Alzheimer's disease (AD). Strong evidence shows that in AD, p62 immunoreactivity is associated with neurofibrillary tangles and is involved in tau degradation. However, it remains to be determined whether p62 also plays a role in regulating amyloid-β (Aβ) aggregation and degradation. Using a gene therapy approach, here we show that increasing brain p62 expression rescues cognitive deficits in APP/PS1 mice, a widely used animal model of AD. The cognitive improvement was associated with a decrease in Aβ levels and plaque load. Using complementary genetic and pharmacologic approaches, we found that the p62-mediated changes in Aβ were due to an increase in autophagy. To this end, we showed that removing the LC3-interacting region of p62, which facilitates p62-mediated selective autophagy, or blocking autophagy with a pharmacological inhibitor, was sufficient to prevent the decrease in Aβ. Overall, we believe these data provide the first direct in vivo evidence showing that p62 regulates Aβ turnover.

Introduction

An imbalance between protein production and degradation contributes to the accumulation of proteinaceous inclusions characteristic of several neurodegenerative disorders, including amyloid-β (Aβ) and tau in Alzheimer’s disease (AD)[1, 2]. Indeed, several laboratories have shown that increasing protein turnover may have beneficial effects on disease outcome[3-6]. Autophagy and the ubiquitin-proteasome system represent two major intracellular degradation systems that contribute to the removal of proteinaceous inclusions. Proteasome dysfunction has been linked to AD and other neurodegenerative disorders[7-11].

Macroautophagy, herein referred to as autophagy, is induced by the formation intracellular autophagosomes that deliver the cargo to be degraded to lysosomes[12]. The formation of the autophagosome is negatively regulated by the mammalian target of rapamycin (mTOR), and involves sequential reactions carried by several autophagy related proteins[13]. LC3-I is a cytosolic autophagy-related protein that during autophagosome formation is post-translationally modified to form LC3-II, which is then incorporated into the growing autophagosome[14]. Notably, the formation of the autophagosome can also be initiated independently of mTOR[15].

p62 is a multifunctional protein involved in protein turnover, via autophagy and proteasome, oxidative stress, and other cellular functions[16, 17]. p62 knockout mice show age-dependent accumulation of NFTs and synaptic deficits[18], underlying the role of p62 in tau aggregation and degradation. In AD, p62 strongly binds to NFTs[19], most likely in an attempt to target them for degradation. Thus, p62 is sequestered in NFTs, which may limit its cellular function by creating reduced levels of available p62 in the cytosol[20]. p62 has several functional domains, including a ubiquitin-binding domain (UBA) an a LC3-interacting region (LIR). p62 binds to polyubiquitinated proteins, including tau, through the UBA domain and targets them for proteasome degradation[21, 22]. Through its LIR domain, p62 binds to LC3 to facilitate selective autophagy[23, 24].

Materials and Methods

Mice

The generation of the APP/PS1 mice was described previously[25]. We have backcrossed the APP/PS1 mice for 12 generations into a pure 129/SvJ background. All animal studies were performed with an approved from the Arizona State University Institutional Animal Care and Use Committee.

Viral constructs generation and injections

AAVs were generated by Vector BioLabs (Malvern, PA). The final titers were 1.0 × 1013 GC/ml for the AAV-GFP and 8.8 × 1012 GC/ml for the AAV-p62-GFP. AAVs were injected bilaterally into the lateral ventricles of newborn, P0 pups, using a 5 µl Hamilton syringe (2 µl of viral suspension per ventricle). Pups were anesthetized by hypothermia. The ventricle were found using the following coordinates: 1.0 mm posterior of bregma, 2.0 mm lateral on each side and 1 mm dorso-ventral from the skull. A plastic stopper was placed in the syringe prior the injection and only 1 mm of the syringe needle was exposed through the stopper. The virus was injected at 1 µl/minute, after which the needle is left in place for 2 additional minutes before it was slowly removed. The experimenter was blinded to the genotype of the mice.

Inclusion/exclusion criteria

Overall, we infected 20 mice per group, 10 females and 10 males. The sample size was chosen based on our previous experience performing similar experiments with these mice. The behavioral experiments were performed in all 20 mice per group. Eight mice per group were randomly selected to perform the biochemical and histological analyses.

Morris water maze

The experimental protocol is detailed in[26]. This test was conducted in a circular pool (diameter = 1 meter) filled with water maintained at 23°C and made opaque by the addition of non-toxic paint. The experimenter was blinded to the group allocation.

Brain tissue processing and immunostaining

Mice were euthanized by CO2 asphyxiation and their brains removed for analyses. One hemisphere of the brain was post-fixed in 4% paraformaldehyde for 48 hours and used for immunohistochemical evaluation. The other hemisphere was flash-frozen on dry ice and used for biochemical experiments and stored at −80°C. From the fixed tissue, using a sliding vibratome, we obtained 50 µm-thick brain sections. Sections were then stored in 0.02% sodium azide in PBS. The endogenous peroxidase activity was quenched with 3% H2O2 in 10% methanol for 30 minutes. For Aβ1–42 staining tissue was incubated for 7 minutes in 88% formic acid to retrieve the epitope. Tissue was incubated overnight at 4°C with an appropriate primary antibody. Sections were washed to remove excess primary antibody and incubated in the appropriate secondary antibody for 1h at room temperature. Excess secondary antibody was washed and sections were developed with diaminobenzidine substrate using the avidin–biotin horseradish peroxidase system (Vector Labs, Burlingame, CA). The experimenter was blinded to the group allocation.

Western blots and ELISA

Frozen tissue was homogenized in T-PER solution (Thermo Fisher Scientific, Waltham, MA) containing complete protease inhibitor (Roche, Basel, Switzerland) and phosphatase inhibitor (Life Technologies, Carlsbad, CA). Brain homogenates were ultracentrifuged at 100,000 g for 1 hour at 4 °C. The supernatant was recovered and stored at −80°C until used for western blots and to measure soluble Aβ levels by ELISA. The pellet was resuspended in 70% formic acid, re-homogenized, and centrifuged as described above. The supernatant of this second centrifugation was recovered and stored at −80°C until used as the insoluble fraction for ELISA experiments. Western blots were performed using precast Novex gels (Life Technologies, Carlsbad, CA). Proteins were transferred to nitrocellulose membranes (iBlot, Life Technologies) and were then incubated for 60 minutes in 5% non-fat powdered milk (Great Value) in tris buffered saline with Tween-20 (TBST, 0.1M Tris, 0.15M NaCl, & 0.1% Tween-20). Primary antibodies specific to the experiment were then applied overnight at 4°C in TBST 5% milk. The next day, blots were washed in TBST three times for 10 minutes/wash and then incubated in the appropriate fluorescent secondary antibody(s) for 1 hour at room temperature (RT). The blots were then washed as described above, and imaged/quantified using a LI-COR Oddesy CLx (LI-COR Biosciences, Lincoln, NE). ELISA measurements were conducted using specific Life Technologies kits, and following the manufacturer’s instructions. The experimenter was blinded to the group allocation.

Cell culture

7PA2 cells were grown in Dulbecco’s Modified Eagles Media (DMEM, Life Technologies) with 10% fetal bovine serum (FBS, Life Technologies) at 37°C with humidified environment (5% CO2) in 6 well plates (5 × 105 cells/well). Cells were transfected with p62, p62-ΔUBA (a gift from Dr. Wei Ding, Capital Medical University, China), and p62-ΔLIR (a gift from Dr. Zhu, University of Kentucky). Aβ levels were measured 48 hours after transfections. The experimenter was blinded to the group allocation.

Hippocampal primary neurons were isolated from APP/PS1 P0 pups and cultured in 6 well plates coated with poly-D-lysine (Sigma-Aldrich, Saint Louis, MO). Neurons were plated in Neurobasal media (Life Technologies) supplemented with 2% B27 (Life Technologies) and 50 U/mL penicillin/streptomycin (Life Technologies). After 24h in culture, neurons were infected with 1µl/ml CamKIIa-p62-GFP AAV (virus titer 8.8×1012 GC/ml). At day 13 post infections, neurons were treated for 14 hours either with 5 mM 3MA (Sigma-Aldrich) or 10 mM MG132 (Thermo Fisher Scientific, Waltham, MA). An untreated plate was used as a control. Aβ levels were measured after this last treatments. The experimenter was blinded to the group allocation.

Proteasome activity

We used the fluorogenic substrates Bz-VGR-AMC and Suc-LLVY-AMC to measure trypsin-like and chymotrypsin-like activities. Specifically, we incubated 10 µl of brain homogenate with each substrate to probe in a total of 200 µl of assay buffer (25mM HEPES, pH 7.5, 0.5mM EDTA, 0.05% NP-40) using black 96-well plates. We obtained kinetic readings at 37°C every 1.5 minutes for 60 min (excitation 360 nm, emission 460 nm) using the Synergy HT multi-mode microplate reader with the Gen5 software (BioTek, Winooski, VT). Readings were normalized to total protein concentrations assayed via a Coomassie Protein Assay Kit (Bradford, Thermo Scientific, Waltham, MA) following the manufacturer’s instructions. The experimenter was blinded to the group allocation.

Antibodies

All of the antibodies used in this study were validated by the manufacturers for use in mouse and/or human tissue (see manufacturers webpages). From Cell Signaling: Total p70S6K (dilution 1:1000, catalog number 9202), p70S6K Thr389 (dilution 1:1000, catalog number 9204), β-actin (dilution 1:10000, catalog number 3700), rpS6 (dilution 1:1000, catalog number 5364), Atg3 (dilution 1:1000, catalog number 3415), Atg5 (dilution 1:1000, catalog number 2630), Atg7 (dilution 1:1000, catalog number 2631), Atg12 (dilution 1:1000, catalog number 2010). From Millipore, anti-Aβ42 (dilution 1:200, catalog number AB5078P), p62 (dilution 1:1000, catalog number MABC32), tau t (dilution 1:1000, catalog number 577801). From BioLegend, 6E10 (dilution 1:3000, catalog number SIG-39320). From Calbiochem, CT20 (dilution 1:3000, catalog number 171610). From Novus Biologicals, LC3 (dilution 1:1000, catalog number NB100-2331). CP13 (dilution 1:000) was a gift from Dr. Peter Davies.

Statistical analyses

Data were analyzed by ANOVAs or student’s t-test as specified in the results and figure legends. When appropriate, post hoc tests were conducted using the Bonferroni correction. These analyses were conducted using GraphPad Prism 5 (GraphPad Software, Inc.). An a priori power analysis was not performed our sample sizes are similar to those reported in previously published papers. Examination of descriptive statistics revealed no violation of any test assumptions that would warrant using statistical test other than the ones used. For all analyses, the variance was approximately the same among groups.

Results

To study the role of p62 in AD pathogenesis, we generated an adeno-associated virus (AAV) expressing mouse p62 under the CamKIIα promoter. To allow co-expression of the green fluorescent protein (GFP) with p62, we cloned a 2A-like peptide sequence[27] between the p62 cDNA and the GFP cDNA (Fig. 1a). We injected new born P0 APP/PS1 and non-transgenic (NonTg) mice with the AAVs expressing p62 or GFP alone (viral titer 8.8 × 1012 and 1.0 × 1013 GC/ml, respectively). The AAVs were injected into the lateral ventricles, 2 µl per each side. Mice that received the p62 virus will be referred to as APP/PS1-p62 and NonTg-p62 mice, while mice that received the GFP virus will be referred to as APP/PS1-GFP and NonTg-GFP mice. Mice were then left to age for seven months, during which all groups gained body weight at a similar rate (Fig. 1b).

Figure 1

p62 gene transfer increases p62 levels in neurons. (a) Diagram depicting the structure of the two adeno associated viruses (AAVs) bilaterally injected into the lateral ventricles of new born APP/PS1 and NonTg mice. (b) Age-dependent change in body weight. (c–d) Western blot and quantitative analysis shows that p62 levels were significantly higher in the two groups infected with the AAV-p62 compared to the two groups infected with the AAV-GFP (n = 8/mice per group). Quantitative analyses were performed by normalizing p62 levels to β-actin, used as a loading control. (e–f) Microphotographs of hippocampal and cortical sections obtained from mice infected with the AAV-p62 virus. Sections were stained with an anti-GFP antibody and show the degree of diffusion of the virus. (g–h) Microphotographs of hippocampal sections co-labeled with GFP (green), and NeuN or GFAP (red). These data indicate that the AAVs almost exclusively infected neurons. Abbreviations: Hp, hippocampous; Cx, cortex. Error bars represent mean ± SEM.

We first measured p62 levels in the hippocampi of 7-month-old mice and found that they were different among the four groups (p = 0.003; Fig. 1c–d). Specifically, p62 levels were significantly higher in the APP/PS1-p62 and NonTg-p62 mice compared to mice infected with the GFP AAVs. In contrast, p62 levels were similar between APP/PS1-p62 and NonTg-p62 mice. To determine the extent of the viral diffusion, we stained sections from APP/PS1-p62 mice with an anti-GFP antibody. The virus was highly expressed throughout the hippocampus and in to a lesser extent in the cortex (Fig 1f). Consistent with the neuronal specificity of CamKIIα, p62 expression was restricted to neurons and was not found in astrocytes (Fig. 1g–h).

Mice received four training trials per day for five consecutive days, in the spatial version of the Morris water maze. For the escape latency there was an effect for days (p < 0.0001) and group (p < 0.0001; Fig. 2a). The day effect indicates that all groups improved across the five days. The group effect indicates that one or more groups learned at a different pace. Specifically, the NonTg-GFP group performed significantly better than the APP/PS1-GFP group at days 4 and 5. In contrast, the APP/PS1-p62 group performed as well as the two NonTg groups throughout the training and significantly better than APP/PS1-GFP on days 3, 4, and 5. We obtained similar results when we analyzed the distance traveled to reach the platform as we found a significant effect for days (p < 0.0001) and group (p = 0.023; Fig. 2b). Specifically, the APP/PS1-GFP group was significantly impaired compared to NonTg-GFP group at days 4 and 5. However, the APP/PS1-p62 group performed significantly better than the APP/PS1-GFP group at day 5 and as well as the two NonTg groups throughout the five days. These results indicate that overexpression of p62 rescued spatial learning deficits in APP/PS1 mice.

Figure 2

p62 overexpression improves cognitive function. (a, b) Learning curves of mice trained in the spatial reference version of the Morris water maze (n = 20/mice per group). The escape latency and the distance traveled to find the hidden platform were plotted against the days of training. Each day represents the average of four consecutive training trials. Both measurements indicate that all four groups significantly improved across the five days. For the escape latency: p < 0.0001 for days and group. Distance traveled: p < 0.0001 and p = 0.023 for days and group. Specifically, the APP/PS1-GFP group was significantly impaired compared to the other three groups at days 3, 4, 5 for the escape latency and day 5 for the distance traveled. (c) The APP/PS1 group spent significantly less time in the target quadrant compared to the other three groups (p = 0.017). (d) The APP/PS1-GFP group spent significantly more time in the opposite quadrant compared to the other three groups (p < 0.0001). (e) The APP/PS1-GFP group needed significantly more time to cross over the platform location compared to the other three groups (p = 0.002). (f) The APP/PS1-GFP group crossed over the platform location significantly fewer times compared to the other three groups (p = 0.04). * indicates that the APP/PS1-p62 group performed significantly better than the APP/PS1-GFP group. # indicates that the APP/PS1-GFP group performed significantly different than the other three groups. Learning data (a–b) were analyzed by two-way ANOVA; probe trials (c–f) were analyzed by one-way ANOVA. Bonferroni’s was used for post hoc tests. Error bars represent mean ± SEM.

Twenty-four hours after the last training trial, we conducted a single 60-second probe trial, during which mice were free to swim in the pool without the hidden platform. We measured the time mice spent in the target quadrant and opposite quadrant, the latency to cross over the platform location, and the total number of platform location crosses. The values for these four measurements were different among the four groups (p = 0.017, p < 0.0001, p = 0.002, p = 0.04, respectively; Fig. 2c–f). Specifically, APP/PS1-GFP group performed significantly different from the other three groups in all measurements (Fig. 2c–f). These data indicate that increasing p62 expression rescued spatial memory deficits in APP/PS1 mice as the APP/PS1-p62 group performed as well as the two NonTg groups.

At the end of the behavioral tests, we stained sections from transgenic mice with an Aβ42-specific antibody. Aβ immunoreactivity was significantly lower in the hippocampus and cortex of APP/PS1-p62 mice compared to APP/PS1-GFP mice (p = 0.01 and p < 0.05, respectively; Fig. 3a–f). We then measured Aβ levels by sandwich ELISA. Soluble and insoluble Aβ40 levels were significantly reduced in APP/PS1-p62 mice compared to APP/PS1-GFP mice (p = 0.01 and p = 0.001, respectively; Fig. 3g–h). Similarly, soluble and insoluble Aβ42 levels were reduced in APP/PS1-p62 mice compared to APP/PS1-GFP mice (p = 0.02 for both measurements; Fig. 3i–j). Given the reported interaction between p62 and tau, we measured the total and phosphorylated tau levels. Overall, we found that neither total tau nor tau phosphorylated at Ser202 were significantly altered among NonTg-GFP, NonTg-p62, APP/PS1-GFP, and APP/PS1-p62 groups (Supplementary Fig. 1).

Figure 3

p62 overexpression reduces Aβ pathology (a, b) Microphotographs of hippocampal sections stained with an Aβ42-specific antibody. (c) Quantitative analysis of the Aβ42 immunoreactivity shows reduced Aβ load in the APP/PS1-p62 mice compared to APP/PS1-GFP mice (p = 0.01). (d, e) Microphotographs of cortical sections stained with an Aβ42-specific antibody. (f) Quantitative analysis of Aβ42 immunoreactivity shows reduced cortical Aβ load in the APP/PS1-p62 mice compared to APP/PS1-GFP mice (p < 0.04). (g–j) Soluble and insoluble Aβ40 and Aβ42 measured by sandwich ELISA. Aβ40 levels were significantly reduced in APP/PS1-p62 mice compared to APP/PS1-GFP mice (p = 0.01 and p = 0.001 for soluble and insoluble, respectively). Aβ42 levels were significantly reduced in APP/PS1-p62 mice compared to APP/PS1-GFP mice (p = 0.02 for both measurements). For all the experiments shown here, n = 8/mice per group. Data were analyzed by student’s t-test and are presented as mean ± SEM.

To begin understanding the mechanisms linking higher p62 levels to reduced Aβ levels, we measured APP processing. APP levels were significantly higher in the APP/PS1-GFP and APP/PS1-p62 mice compared to the two NonTg groups (p < 0.0001; Supplementary Fig. 2a–b). We found similar results for C99 and C83 levels (p < 0.0001 for both measurements; Supplementary Fig. 2c–d). Overall, APP, C99, and C83 levels were unaltered by p62 overexpression as they were similar between APP/PS1-GFP and APP/PS1-p62 mice.

To determine the role of protein turnover in the p62-mediated decrease in Aβ levels, we measure proteasome function, given the role of p62 in this process[16]. Using the fluorogenic substrates Bz-VGR-AMC and Suc-LLVYAMC, we measured trypsin-like and chymotrypsin-like activities, respectively. The chymotrypsin-like activity was significantly different among the four groups, as indicated by the slope of the curve (p = 0.022) and the area under the curve (p = 0.020; Supplementary Fig. 3a–b). These differences were not driven by p62, as the APP/PS1-GFP and the APP/PS1-p62 groups were not statistically significant from each other. The trypsin-like activity was not statistically different among the four groups (Supplementary Fig. 3c–d).

p62 also functions by targeting proteins to autophagy for degradation[28, 29]. To determine the effects of increasing p62 on autophagy, we measured the expression levels of several autophagy related proteins (Atgs), which are routinely used to assess overall autophagy induction[12]. Atg3, Atg5, and Atg7 levels were different among the four groups (p = 0.001, p = 0.004, p = 0.027, respectively; Fig. 4a–d). For all three proteins, the NonTg-p62 and APP/PS1-p62 groups were significantly different than NonTg-GFP and APP/PS1-p62 groups. In contrast, the levels of Atg12 were not statistically significant among the four groups (Fig. 4a, e). We also found that while LC3-I levels were similar among the four groups (Fig. 4a, f), LC3-II levels were significantly higher in NonTg-p62 and APP/PS1-p62 compared to NonTg-GFP and APP/PS1-GFP groups (p = 0.002; Fig. 4a, g). These data show that increasing p62 expression does not change proteasome activity but facilitates autophagy induction, suggesting a possible mechanism for the p62-mediated changes in Aβ. Once autophagosomes are formed, they deliver their cargo to lysosomes for degradation. Thus, if p62-mediated decrease in Aβ levels are linked to an increase in autophagy, one would expect an increase in Aβ in the lysosomes. We labeled sections from APP/PS1-GFP and APP/PS1-p62 with Lamp2A (a lysosomal marker) and 6E10 (an antibody that recognize Aβ and other Aβ-containing fragments). We found that the amount of co-localization was significantly increased in the APP/PS1-p62 mice (Fig. 4h–I; p = 0.01). These data indicate that p62 increases autophagy induction and facilitates the delivery of Aβ to lysosomes.

Figure 4

p62 overexpression increases autophagy induction. (a) Western blots of hippocampal proteins probed with the indicated antibodies. (b–d) Atg3, Atg5, and Atg7 levels were different among the four groups (p = 0.001, p = 0.004, p = 0.027, respectively). Post hoc analyses indicated that the two p62 groups were significantly different than the two groups GFP groups. (e–f) Atg12 and LC3-I levels were not statistically different among the four groups. (g) LC3-II levels were different among the four groups (p = 0.002). Post hoc analyses indicated that the two groups infected with p62 were significantly different than the two groups infected with GFP. (h–i) Representative confocal images and quantitative analyses. Sections were labeled with Lam2a (green) and 6E10 (red), to mark lysosomes and Aβ-containing fragments, respectively. The number of yellow pixels was significantly increased in APP/PS1-p62 mice, which indicates higher Aβ levels in lysosomes. For all the experiments shown here, n=8 mice per group. Quantitative analyses of the blots were performed by normalizing the levels of the protein of interest to β-actin, used as a loading control. Data were analyzed by one-way ANOVA and Bonferroni’s post-hoc test and are presented as mean ± SEM.

The mammalian target of rapamycin (mTOR) is a negative regulator of autophagy and it is known to interact with p62[13, 30]. Further, mTOR hyperactivity is linked to AD pathogenesis[31-36]. To determine whether the p62-mediated autophagy induction is linked to mTOR, we measured mTOR and its downstream targets S6K1, rpS6, and 4EBP1. While total levels of mTOR and S6K1 were not statistically significant among the four groups (Supplementary Fig. 4a–c), the levels of S6K1 phosphorylated at Thr389 were (p = 0.004; Supplementary Fig. 4a, d). Notably, these changes were independent of p62. Further, we found that the total and the levels of phosphorylated rpS6 and 4EBP1 were similar among the four groups (Supplementary Fig. 4e–h). These data indicate that the p62-mediated increase in autophagy induction is independent of mTOR.

p62 has two main functional domains, the UBA and the LIR domains, which regulate protein degradation via the proteasome or autophagy, respectively[16]. To further dissect the mechanisms underlying the p62-mediated decrease in Aβ levels, we used p62 lacking the LIR domain (p62-ΔLIR[37]) or the UBA domain (p62-ΔUBA[38]). We transiently transfected p62-ΔLIR, p62-ΔUBA, or p62 wild type into 7PA2 cells, a widely used cell line known to secrete low molecular weight Aβ oligomers and N-terminal extended peptides[39, 40]. Notably, all three constructs also expressed a fluorescent tag, which allowed us to determine transfection efficiency (Fig. 5a). We ran a western blot of proteins extracted from transfected 7PA2 cells using a p62 antibody whose epitope is located within the LIR domain. We observed a shift in the molecular weight of p62-ΔUBA, while p62-ΔLIR was not detectable (Fig. 5b). We then measured Aβ levels by sandwich ELISA and found that they were different among the four groups (Fig. 5c, p = 0.001). Notably, transfection of wild type p62 or p62-ΔUBA decreased Aβ levels. In contrast, p62-ΔLIR was unable to significantly reduce Aβ levels, suggesting that the effects of p62 on Aβ are mediated by its LIR domain. Given that 7PA2 cells produce different forms of Aβ, including N-terminal extended peptides, it is possible that the sandwich ELISA used here detected different fragments of Aβ. Therefore, we isolated primary neurons from APP/PS1 mice and infected them with the AAVs expressing p62. Thirteen days after viral infections, neurons were treated with 10 mM MG132 or 5 mM 3MA for 14 hours to inhibit proteasome and autophagy, respectively. To determine infection efficiency, we measured the levels of the GFP tag and p62 (Fig. 5d). GFP levels were not detectable in the uninfected group, and were not statistically different among the other three groups (Fig. 5d–e). Similarly, p62 levels were increased in neurons infected with the p62 virus compared to uninfected neurons (p = 0.002; Fig. 5d, f). Most notably, Aβ42 levels, as measured by sandwich ELISA, were different among the four groups (p < 0.0001; Fig. 5g). Importantly, p62-infected neurons treated with the vehicle had significantly lower levels of Aβ42 than non-infected neurons. This decrease was prevented by the administration of the autophagy inhibitor (3MA) but not by the proteasome inhibitor (MG132; Fig. 5g). Overall, our results show that the p62-mediated decrease in Aβ levels are mediated by autophagy.

Figure 5

p62 overexpression decreases Aβ by an autophagy-mediated mechanism. (a) Microphotographs of 7PA2 cells transfected with different p62 constructs, as indicated. (b) Western blot from proteins isolated from transfected 7PA2 cells and probed with a p62 antibody, which recognize p62 within the LIR domain. Therefore ΔLIR is not detectable, while there is a clear shift in molecular weight for ΔUBA. (c) Aβ levels obtained from 7PA2 cells and measured by sandwich ELISA were different among the four groups (p = 0.001). Post hoc analyses revealed that transfection of wild type p62 or p62-ΔUBA decreased Aβ42 levels. In contrast, p62-ΔLIR was unable to significantly reduce Aβ levels. (d) Western blot of proteins extracted from APP/PS1 primary neurons infected with the p62 AAVs and treated with compounds to block autophagy (3MA) or proteasome function (MG132), as indicated. (e) GFP levels were similar among the three groups of neurons infected with the p62 AAVs. (f) p62 blot showed that p62 levels were significantly higher in the three groups infected with the p62 AAVs, compared to uninfected neurons. (g) Aβ42 levels obtained from these APP/PS1 neurons and measured by sandwich ELISA, were significantly different among the four groups (p < 0.0001). Post hoc analyses revealed that the p62 and the p62 + MG132 groups had significantly lower Aβ42 levels compared to the other two groups. For all the experiments shown here, n=9 mice per group. Quantitative analyses of the blots were performed by normalizing the levels of the protein of interest to β-actin, used as a loading control. Data were analyzed by one-way ANOVA and Bonferroni’s post hoc test and are presented as mean ± SEM.

Discussion

Here we provide the first in vivo evidence indicating that p62 plays a role in Aβ degradation. These data underscore a novel mechanism by which Aβ can be targeted for autophagic degradation. p62 is frequently found in protein inclusions of several neurodegenerative disorders[41, 42]. It has been stipulated that by binding to these inclusions (including NFTs), p62 is sequestered from the cytosol creating a p62 loss-of-function environment[16, 20]. Furthermore, expression levels of p62 are reduced in AD and this appears to be linked to oxidative damage to its promoter region[20, 43]. Thus, consistent with the data reported here, increasing p62 levels might be a valid approach to restore neuronal function in proteinopathies.

Several aspects of the autophagy-lysosomal systems are impaired in AD[1, 2, 44]. For example, elegant work by Nixon and colleagues showed accumulation of autophagosomes in postmortem human AD brains, suggesting deficits in autophagy flux[45]. At the same time, autophagy induction might also be impaired in AD, further highlighting the notion that promoting autophagy clearance might be a valid therapeutic approach for AD[46]. Consistently, increasing lysosomal function or autophagy induction improves AD-like pathology in transgenic mice[3-6]. mTOR is a negative regulator of autophagy induction. We and others have shown that decreasing mTOR signaling increases autophagy and improves AD-like pathology in several mouse models of AD[5, 31–33, 47, 48]. Here we report that the p62-mediated increase in autophagy induction is independent of mTOR activity, indicating that the LIR domain of p62 might engage alternative pathways to induce autophagy. Activation of LC3 by Atg4 can lead to autophagosome formation by a series of reactions mediated by Atg7, Atg3, and Atg5, which culminates into the conversion of LC3-I to LC3-II and the formation of the autophagosome[14]. Consistently, we found that the mTOR-independent autophagy induction mediated by p62 was linked to increased levels of several of these Atg proteins, including LC3-II.

The data presented here, together with reports from the literature, suggest that increasing p62 levels might be a valid therapeutic approach for AD as it facilitates the removal of tau and Aβ by activating autophagy. Given the well-established deficits in autophagosome clearance in AD, the timing at which one would increase autophagy via p62 needs to be very closely considered. Increasing autophagy induction before deficits in autophagosome clearance manifest might be beneficial as it will lead to increased turnover of Aβ and tau. However, increasing autophagy induction after deficits in autophagosome clearance manifest may not have the desired effects on Aβ and tau and may further contribute to the pathological accumulation of autophagosomes.