Improved proteostasis in the secretory pathway rescues Alzheimer’s disease in the mouse

See Duran-Aniotz et al . (doi: 10.1093/brain/awv401 ) for a scientific commentary on this article.

Many neurodegenerative diseases are characterized by accumulation of toxic protein aggregates in specific subcellular locations. Using mouse models, Peng et al. show that inhibition of the endoplasmic reticulum acetylation machinery enhances autophagy-mediated disposal of aggregates that form within the secretory pathway and rescues Alzheimer’s disease.

See Duran-Aniotz et al . (doi: 10.1093/brain/awv401 ) for a scientific commentary on this article.

Many neurodegenerative diseases are characterized by accumulation of toxic protein aggregates in specific subcellular locations. Using mouse models, Peng et al. show that inhibition of the endoplasmic reticulum acetylation machinery enhances autophagy-mediated disposal of aggregates that form within the secretory pathway and rescues Alzheimer’s disease.

Abstract

See Duran-Aniotz et al . (doi: 10.1093/brain/awv401 ) for a scientific commentary on this article.

The aberrant accumulation of toxic protein aggregates is a key feature of many neurodegenerative diseases, including Huntington’s disease, amyotrophic lateral sclerosis and Alzheimer’s disease. As such, improving normal proteostatic mechanisms is an active target for biomedical research. Although they share common pathological features, protein aggregates form in different subcellular locations. Nε-lysine acetylation in the lumen of the endoplasmic reticulum has recently emerged as a new mechanism to regulate the induction of autophagy. The endoplasmic reticulum acetylation machinery includes AT-1/SLC33A1, a membrane transporter that translocates acetyl-CoA from the cytosol into the endoplasmic reticulum lumen, and ATase1 and ATase2, two acetyltransferases that acetylate endoplasmic reticulum cargo proteins. Here, we used a mutant form of α-synuclein to show that inhibition of the endoplasmic reticulum acetylation machinery specifically improves autophagy-mediated disposal of toxic protein aggregates that form within the secretory pathway, but not those that form in the cytosol. Consequently, haploinsufficiency of AT-1/SLC33A1 in the mouse rescued Alzheimer’s disease, but not Huntington’s disease or amyotrophic lateral sclerosis. In fact, intracellular toxic protein aggregates in Alzheimer’s disease form within the secretory pathway while in Huntington’s disease and amyotrophic lateral sclerosis they form in different cellular compartments. Furthermore, biochemical inhibition of ATase1 and ATase2 was also able to rescue the Alzheimer’s disease phenotype in a mouse model of the disease. Specifically, we observed reduced levels of soluble amyloid-β aggregates, reduced amyloid-β pathology, reduced phosphorylation of tau, improved synaptic plasticity, and increased lifespan of the animals. In conclusion, our results indicate that Nε-lysine acetylation in the endoplasmic reticulum lumen regulates normal proteostasis of the secretory pathway; they also support therapies targeting endoplasmic reticulum acetyltransferases, ATase1 and ATase2, for a subset of chronic degenerative diseases.

Introduction

Integral membrane proteins and secretory proteins are typically synthesized at the surface of the endoplasmic reticulum (ER) where they also enter the secretory pathway ( Wickner and Schekman, 2005 ). Proteins that are not required to enter the secretory pathway are instead synthesized in the cytosol. Independently from where they are synthesized, all new polypeptides are selected based on their ability to fold. The quality control machinery that selects correctly folded and unfolded/misfolded polypeptides is tightly linked to the degrading machinery so that unfolded/misfolded polypeptides can be disposed of, thus ensuring fidelity of the protein code ( Trombetta and Parodi, 2003 ).

Autophagy is an essential component of the degrading machinery. It helps dispose of large toxic protein aggregates that form within the secretory pathway as well as in the cytosol. Malfunction of autophagy contributes to the progression of many chronic diseases, including neurodegeneration, cancer, nephropathies, immune and cardiovascular diseases (reviewed in Nixon, 2013 ; Frake et al. , 2015 ; Levine et al. , 2015 ). In addition, many chronic degenerative diseases are characterized by the aberrant accumulation of toxic protein aggregates. As such, improving normal proteostatic mechanisms is an active target for biomedical research ( Mizushima et al. , 2008 ; Levine et al. , 2015 ).

Compelling data indicate that both hypoactive and hyperactive autophagy can be detrimental for the organism (reviewed in Frake et al. , 2015 ; Levine et al. , 2015 ). The same data also indicate that increased levels of autophagy, which are pathogenic in wild-type mice in the absence of toxic protein aggregates, can be beneficial in mouse models of diseases characterized by increased accumulation of toxic protein aggregates ( van Dellen et al. , 2000 ; Pickford et al. , 2008 ; Hetz et al. , 2009 ; Madeo et al. , 2009 ; Bhuiyan et al. , 2013 ). As toxic protein aggregates can form in different locations (i.e. within the secretory pathway or in the cytosol), it is likely that different signals are used to trigger autophagy.

Many integral membrane proteins and secretory proteins that enter the secretory pathway undergo transient Nε-lysine acetylation within the lumen of the ER ( Choudhary et al. , 2009 ; Pehar et al. , 2012 a ). The acetylation of ER cargo proteins requires active transport of acetyl-CoA from the cytosol into the ER lumen by AT-1/SLC33A1, and the acetyltransferase activity of two ER membrane proteins, ATase1 and ATase2 (reviewed in Pehar and Puglielli, 2013 ). ATase1 (encoded by NAT8B ) and ATase2 (encoded by NAT8 ) were recently found to associate with the oligosaccharyl transferase complex and acetylate correctly folded polypeptides ( Ding et al. , 2014 ). Studies conducted with two substrates of the ATases, BACE1 and CD133 (encoded by PROM1 ), suggest that Nε-lysine acetylation might be part of normal quality control to select correctly folded polypeptides ( Costantini et al. , 2007 ; Ko and Puglielli, 2009 ; Ding et al. , 2014 ; Mak et al. , 2014 ). Importantly, the promoter of AT-1 has an X-box binding protein 1 (XBP1) binding element and, as a result, the expression of AT-1 is activated by XBP1 during the unfolded protein response to partially repress the induction of autophagy ( Pehar et al. , 2012 b ). Downregulation or inactivation of AT-1 in isolated cells or in the animal leads to increased autophagy ( Jonas et al. , 2010 ; Pehar et al. , 2012 b ; Peng et al. , 2014 ). Therefore, the above studies suggest that the ER acetylation machinery might participate in the regulation of quality control as well as ER-associated degradation type II/autophagy.

Data from yeast, Drosophila melanogaster , Caenorhabditis elegans , and mammals indicate that Nε-lysine acetylation can serve as a master regulator of the autophagic response to a large variety of insults (reviewed in Madeo et al. , 2009 , 2010 ). In addition to possible global epigenetic control of autophagy proteins induced by changes in acetyl-CoA levels ( Eisenberg et al. , 2014 ; Marino et al. , 2014 ), acetylation and deacetylation of selective members of the autophagy machinery, such as ATG9A, ATG5, ATG7, Atg8/GABARAP) and ATG12, can also regulate the induction and/or progression of autophagy. More specifically, Nε-lysine acetylation inhibits while Nε-lysine deacetylation stimulates autophagy ( Lee et al. , 2008 ; Lee and Finkel, 2009 ; Pehar et al. , 2012 b ). Improved autophagic functions that result from reduced acetylation and/or increased deacetylation have been associated with more efficient protein and organelle homeostasis, cytoprotection, lifespan extension, and rescue of proteotoxic phenotypes (reviewed in Madeo et al. , 2015 ).

Here, we report that inhibition of the ER acetylation machinery stimulates the disposal of toxic protein aggregates that form within the secretory pathway but not those that form in other compartments. Consistently, genetic or biochemical inhibition of the acetylation machinery in the mouse rescued the Alzheimer’s disease phenotype, but not the Huntington’s disease or the amyotrophic lateral sclerosis phenotypes.

Materials and methods

The following experimental approaches have been described in detail previously: lactate dehydrogenase (LDH) activity in the conditioned media ( Costantini et al. , 2005 ); electrophysiology of hippocampal brain slices ( Pehar et al. , 2010 ); and enzyme-linked immunosorbent assay (ELISA) of soluble amyloid-β ( Costantini et al. , 2006 ; Pehar et al. , 2010 ).

Cells and animals

Mouse embryonic fibroblasts (MEFs) from wild-type and AT-1 ^S113R/+ mice were described previosuly ( Peng et al. , 2014 ). MEFs, Chinese Hamster Ovary (CHO), and human neuroglioma (H4) cells were maintained in Dulbecco’s modified Eagle medium supplemented with 10% foetal bovine serum (FBS) and 1% penicillin/streptomycin/glutamine solution (Mediatech). CHO cell transfection was performed using Lipofectamine™ 2000 (Invitrogen/Life Technologies). MEFs were transfected with Amaxa™ Basic Nucleofector™ Kit for Primary Mammalian Fibroblasts (Lonza). Cells were harvested 48 h later for western blot or immunostaining.

AT-1 ^S113R/+ and APP _695/swe mice were described previously ( Pehar et al. , 2010 ; Peng et al. , 2014 ). mHtt ^Q160 (also known as R6/2) and hSOD1 ^G93A mice were from The Jackson Laboratory. The rodent diet with Compound 9 was manufactured by Bio-Serv. The same diet without compound served as the control diet.

All animal experiments were carried out in accordance with the NIH Guide for the Care and Use of Laboratory Animals and were approved by the Institutional Animal Care and Use Committee of the University of Wisconsin-Madison and the Madison Veterans Administration Hospital.

Protein extraction and western blotting

Protein extracts ( Peng et al. , 2014 ) and extracellular enriched proteins ( Lesne et al. , 2006 ; Pehar et al. , 2010 ) were recovered as before. Detergent-soluble and -insoluble fractions were prepared as described ( Gan et al. , 2012 ). Briefly, cells were lysed in lysis buffer (50 mM Tris-HCl, pH 7.4, 150 mM NaCl, 2 mM EDTA, 1 mM dithiothreitol) completed with protease inhibitors (Roche) and 1% Triton™ X-100 (Buffer A), following centrifugation at 100 000 g for 30 min at 4°C. Supernatants were recovered as Triton-soluble fractions. Pellets were washed with Buffer A three times, and then resuspended in lysis buffer containing Buffer A, 1% sodium dodecyl sulphate (SDS) and 0.5% sodium deoxycholate. After sonication and a brief spin down, the lysates were recovered as Triton-insoluble (SDS-soluble) fractions.

Differential detergent extraction of human SOD1 (hSOD1) from the spinal cord of early symptomatic mice was performed as previously described ( Wang et al. , 2003 ). Briefly, tissue was lysed in TEN buffer (10 mM Tris-HCl pH 8.0, 1 mM EDTA, 100 mM NaCl, 0.5% NP-40 and protease inhibitors). After sonication the lysate was centrifuged at 100 000 g for 5 min. The supernatant S1 was recovered as the non-ionic detergent soluble fraction. The pellet P1 was washed twice in TEN buffer by sonication and centrifuged at 100 000 g for 5 min to obtain pellet P2. Pellet P2 was resuspended by sonication in TEN buffer supplemented with 0.5% sodium deoxycholate and 0.25% SDS. After centrifugation the supernatant was recovered as the non-ionic detergent insoluble fraction.

Protein concentration was measured by the bicinchoninic acid method (Pierce). Protein electrophoresis was performed on a NuPAGE® system using 4–12% Bis-Tris gels (Invitrogen).

The following primary antibodies were used: anti-Beta Amyloid (clone 6E10, 1:1000, Signet); anti-Amyloid Precursor Protein, C-Terminal (1:1000, Millipore); anti-phospho-PHF-tau (pSer202+Thr205; clone AT8, 1:750, Thermo Scientific); anti-Tau (clone T46, 1:1000, Invitrogen); anti-Tau (3-repeat isoform RD3; clone 8E6/C11, 1:500, Millipore); anti-alpha Synuclein (clone LB509, 1:1000, Abcam); anti-Huntingtin (clone mEM48, 1:1000, Millipore); anti-hSOD1 (clone EPR1726, 1:10 000, Epitomics); anti-BACE1 (1:1000, Abcam); anti-p62 (1:1000, Cell Signaling); anti-actin (1:1000, Cell Signaling); anti-LC3B (1:1000, Cell Signaling); anti-ATG9A (1:1000, Epitomics); anti-acetylated lysine (1:100, Cell Signaling); anti-IDE (1:1000, Abcam); anti-NEP (1:1000, Millipore); anti-ATF6 (1:250, Imgenex); anti-Bip (1:1000, Cell Signaling); anti-phospho-eIF2a (1:1000, Cell Signaling); anti-eIF2a (1:1000, Cell Signaling); anti-phospho-PERK (1:200, Santa Cruz); anti-PERK (1:1000, Cell Signaling); anti-Calreticulin (1:1000, Abcam). Blots were visualized with goat anti-rabbit Alexa Fluor® 680-conjugated or anti-mouse Alexa Fluor® 800-conjugated secondary antibodies on infrared imaging (LICOR Odyssey Infrared Imaging System; LI-COR Biosciences), or with HRP-conjugated anti-mouse or anti-rabbit secondary antibodies on chemiluminescent detection (ImageQuant LAS4000; GE Healthcare).

cDNA and plasmids

The plasmid containing human α-synuclein (A53T SYN) cDNA was a generous gift from Dr Jeffrey A. Johnson. This plasmid was used as a template to generate the cDNA of α-synuclein with an initiator methionine (M-A53T syn) or the signal peptide from human APP (SP-A53T syn) at the N-terminus. Primers for M-A53T syn were: 5′-AACCCAAGCTTGCCATGGATGTATTCATGAAAGGAC-3′ (forward) and 5′-AAGGCCTCGAGTCATTAGGCTTCAGGTTCGTAGTCT-3′ (reverse). Primers for SP-A53T syn were: 5′-AACCCAAGCTTGTCGCGATGCTGCCCGGTTTGGCACTGCTCCTGCTGGCCGCCTGGACGGCTCGGGCGATGGATGTATTCATGAAAGGAC-3′ (forward) and 5′-AAGGCCTCGAGTCATTAGGCTTCAGGTTCGTAGTCT-3’ (reverse). The PCR fragments were subsequently cloned (HindIII/XhoI) in vector pcDNA™3.1/ myc -His (+) B (Invitrogen) resulting in plasmids M-A53T syn and SP-A53T syn.

The p5xATF6-GL3 plasmid was a gift from Ron Prywes (Addgene plasmid 11 976) ( Wang et al. , 2000 ). For ATF6-luciferase reporter activity, MEFs were transfected with 5 µg promoter-reporter construct as well as the empty vector along with 0.1 µg of Renilla luciferase (Promega) by using Amaxa™ Basic Nucleofector™ Kit for Primary Mammalian Fibroblasts (Lonza). Firefly and Renilla luciferase activities were measured 24 h after transfection with a dual luciferase kit (Promega) and expressed as relative luciferase activity. Co-transfected Renilla luciferase was used to normalize for transfection efficiency ( Ko and Puglielli, 2007 ).

XBP1 quantitative PCR was carried out as described ( Sha et al. , 2009 ). Primers for XBP1 were: 5′>GAGTCCGCAGCAGGTG>3′ (forward) and 5′>TCCAGAATGCCCAAAAGG>3′ (reverse). Primers for total XBP1 were: 5′>ACATCTTCCCATGGACTCTG>3′ (forward) and 5′>TAGGTCCTTCTGGGTAGACC>3′ (reverse). Primers for GAPDH were: 5′> AGGTCGGTGTGAACGGATTTG>3′ (forward) and 5′> TGTAGACCATGTAGTTGAGGTCA>3′ (reverse).

Histology and immunostaining

Histology and immunostaining techniques were described before ( Pehar et al. , 2010 ; Peng et al. , 2014 ). The following primary antibodies were used: anti-phospho-PHF-Tau (clone AT8, 1:100, Thermo Scientific); anti-synaptophysin (clone YE269, 1:250, Abcam); anti-alpha Synuclein (clone LB509, 1:100, Abcam); anti-LC3B (1:100, Cell Signaling); anti-Beta Amyloid (clone 6E10, 1:100, Signet); anti-Beta Amyloid (clone 4E12, 1:100, MBL); anti-NeuN (clone A60, 1:100, EMD-Millipore). Secondary antibodies were Alexa 488- and Alexa 594-conjugated goat anti-rabbit and anti-mouse (5 µg/ml; Molecular Probes-Invitrogen). For phospho-PHF Tau-AT8 immunofluorescence, the secondary antibodies were biotin-labelled goat anti-mouse (5 µg/ml; Molecular Probes-Invitrogen) followed by Alexa 488- or Alexa 594-conjugated streptavidin (5 µg/ml; Molecular Probes-Invitrogen). Beta-amyloid staining was performed after pretreatment of tissue sections with 70% formic acid for 30 min. Processed slides were imaged on a Zeiss Axiovert 200 inverted fluorescent microscope.

Statistical analysis

Data analysis was performed using GraphPad InStat 3.06 statistical software (GraphPad Software Inc.). Data are expressed as mean ± standard deviation (SD). Comparison of the means was performed using Student’s t -test or one-way ANOVA followed by Tukey-Kramer multiple comparisons test. For lifespan assessment, data were analysed with the Kaplan-Meier lifespan test and log-rank test using GraphPad Prism version 4.0 (GraphPad Software). Differences were declared statistically significant if P < 0.05.

Results

AT-1 activity regulates the disposal of protein aggregates within the secretory pathway

To determine whether the increased activation of autophagy that results from reduced influx of acetyl-CoA into the ER preferentially degrades certain toxic protein aggregates, we used MEFs from AT-1 ^S113R/+ mice. AT-1 _S113R is a mutant version of AT-1 that is devoid of acetyl-CoA transport activity. As a result, AT-1 ^S113R/+ knock-in mice have increased activation of autophagy ( Peng et al. , 2014 ). Both wild-type and AT-1 ^S113R/+ MEFs were transfected with A53T α-synuclein, a mutant version of α-synuclein that is associated with autosomal dominant Parkinson’s disease ( Polymeropoulos et al. , 1997 ). α-Synuclein has high propensity to aggregate and is found in Lewy bodies of sporadic and familial forms of Parkinson’s disease, cortical dementia with Lewy bodies, as well as other forms of synucleinopathies ( Galvin et al. , 2001 ). To discriminate between aggregates that form in the cytosol and in the secretory pathway we transfected the above MEFs with two different versions of α-synuclein: one that had an initiator methionine (M-A53T syn) to direct translation in the cytosol and one with a signal peptide (SP-A53T syn) to direct translation on the ER and insertion into the secretory pathway ( Fig. 1 A). To differentiate between soluble and aggregated species of α-synuclein, MEFs were sequentially lysed with Triton ^TM X-100 (for soluble/non-aggregated α-synuclein) and SDS (for aggregated α-synuclein) ( Gan et al. , 2012 ).

Figure 1.

Increased autophagy in AT-1 ^S113R/+ mice targets protein aggregates in the secretory pathway. ( A ) Western blot showing the migration profile of M-A53T syn and SP-A53T syn. ( B–F ) MEFs from wild-type (WT) and AT-1 ^S113R/+ mice were transfected with mutant α-synuclein (A53T syn). Levels of soluble (Triton™ X-100) and insoluble/aggregated (SDS) A53T syn were detected by western blotting. M-A53T syn, α-synuclein with an initiator methionine; SP-A53T syn, α-synuclein with a signal peptide at the N-terminus. Selected images are shown in B–E , while quantification of changes is shown in F . Values are mean ± SD * P < 0.05. Loading controls are shown in Supplementary Fig. 1 . ( G ) Media from SP-A53T syn transfected MEFs were used to immunoprecipitate α-synuclein prior to western blotting. Total cell lysate served as positive control. ( H ) Lactated dehydrogenase (LDH) activity was assayed in the media of SP-A53T syn transfected MEFs. Values are mean ( n = 4) ± SD. ** P < 0.005. ( I ) Immunolabelling showing co-localization of SP-A53T syn with LC3β puncta in AT-1 ^S113R/+ MEFs. As expected, LC3β displayed a diffuse cytosolic distribution in wild-type MEFs; puncta were only visible in AT-1 ^S113R/+ MEFs ( Peng et al. , 2014 ). No co-localization of α-synuclein with LC3β puncta was observed in AT-1 ^S113R/+ MEFs transfected with M-A53T syn. ( J ) Western blot showing increased levels of SP-A53T syn aggregates following BPE treatment to arrest the autophagy flux. Increased levels of p62, an autophagy-cargo protein that is normally degraded as part of the autophagy process, served as a marker of successful blockage of autophagy. Representative images are in the left panel; quantitative changes of α-syn/SDS are in the right panel.

The results show striking differences across the experimental set-up ( Fig. 1 B–F). Specifically, the levels of Triton-soluble α-synuclein were overall similar when comparing wild-type and AT-1 ^S113R/+ MEFs as well as M-A53T syn and SP-A53T syn ( Fig. 1 B and D) suggesting no overall differences in the aggregation of α-synuclein. There was also no significant difference when we compared levels of SDS-soluble M-A53T syn in wild-type and AT-1 ^S113R/+ MEFs ( Fig. 1 C and F) suggesting that the increased levels of autophagy in AT-1 ^S113R/+ MEFs do not influence the disposal of syn aggregates that form in the cytosol. In contrast, the levels of SDS-soluble SP-A53T syn were significantly decreased ( Fig. 1 E and F). To determine whether the reduced levels of SDS-soluble SP-A53T syn in AT-1 ^S113R/+ MEFs was simply due to a more efficient secretion of the protein aggregates, we immunoprecipitated α-synuclein from the media. However, as expected, the immunoprecipitation did not yield significant levels of the protein ( Fig. 1 G) confirming that AT-1 ^S113R/+ MEFs dispose of SP-A53T syn aggregates more efficiently. The increased efficiency in disposing of the toxic protein aggregates was accompanied by reduced cell toxicity, as assessed by determining LDH release in the media ( Fig. 1 H). Direct assessment of transfected cells revealed a marked co-localization of SP-A53T syn with LC3β, a commonly used marker of autophagy ( Pehar et al. , 2012 b ; Peng et al. , 2014 ). We previously published that the autophagy flux is maintained in AT-1 ^S113R/+ ( Peng et al. , 2014 ); therefore, the co-localization of SP-A53T syn with LC3β ( Fig. 1 I) and consequent reduced levels of SDS-soluble SP-A53T syn can be interpreted as a result of more efficient autophagy-mediated degradation of the aggregated protein. Finally, we blocked the progression of autophagy with bafilomycin (500 nM)/pepstatin A (10 µg/ml)/E64 (10 µg/ml) (BPE) and observed a significant increase in the levels of SDS-soluble SP-A53T syn ( Fig. 1 J), supporting our conclusion that autophagy is responsible for the clearance of SP-A53T syn aggregates.

When taken together the above results suggest that the increased autophagy activation described in AT-1 ^S113R/+ MEFs ( Peng et al. , 2014 ) preferentially targets toxic protein aggregates that form within the secretory pathway.

Reduced AT-1 activity in the mouse rescues Alzheimer’s disease but not Huntington disease or amyotrophic lateral sclerosis

To confirm the above results in mouse models, we crossed AT-1 ^S113R/+ mice, which display reduced AT-1 transport activity and increased activation of autophagy in neurons ( Peng et al. , 2014 ), with mouse models of Huntington’s disease, amyotrophic lateral sclerosis, and Alzheimer’s disease. Specifically, for Huntington’s disease we used mHtt ^Q160 (also known as R6/2) mice ( Mangiarini et al. , 1996 ); for amyotrophic lateral sclerosis we used hSOD1 ^G93A mice ( Gurney et al. , 1994 ); and for Alzheimer’s disease we used APP _695/swe mice ( Borchelt et al. , 1996 ; Pehar et al. , 2010 ). Both huntingtin (HTT) and superoxide dismutase 1 (SOD1) have an initiator methionine and are translated on cytosolic ribosomes. Protein aggregates in mHtt ^Q160 are mainly observed in the nucleus ( Mangiarini et al. , 1996 ; Davies et al. , 1997 ), whereas in hSOD1 ^G93A they are observed in the cytosol, ER-Golgi compartment and mitochondria ( Ferri et al. , 2006 ; Kikuchi et al. , 2006 ). In contrast to HTT and SOD1, the amyloid precursor protein (APP) is a type 1 membrane protein with a signal peptide at the N-terminus; it is translated on ER-bound ribosomes and inserts into the secretory pathway. Because of the topology of APP, the amyloid-β peptide that results from proteolytic cleavage of APP within the secretory pathway can only be released in the lumen of the organelle (or, eventually, secreted to the extracellular milieu) ( Haass et al. , 1995 ; Cook et al. , 1997 ; Takami et al. , 2009 ). APP _695/swe mice develop amyloid-β aggregates both inside and outside the neuronal cell body ( Duyckaerts et al. , 2008 ).

Crossing mHtt ^Q160 or hSOD1 ^G93A mice with AT-1 ^S113R/+ mice did not rescue the Huntington’s disease-like ( Fig. 2 ) or the amyotrophic lateral sclerosis-like ( Fig. 3 ) phenotypes. However, crossing APP _695/swe with AT-1 ^S113R/+ mice resulted in a dramatic rescue of the Alzheimer’s disease-like phenotype ( Fig. 4 ). Specifically, we observed a drastic increase of the lifespan of the animals ( Fig. 4 A), reduced intraneuronal amyloid-β labelling ( Fig. 4 B), reduced levels of soluble amyloid-β aggregates ( Fig. 4 C and D), and improved synaptic plasticity, as assessed by long-term potentiation ( Fig. 4 E). To assess whether the phenotypic rescue was due to reduced generation of amyloid-β rather than to increased disposal of intracellular amyloid-β aggregates, we determined levels of BACE1 and APP in AT-1 ^S113R/+ mice. The results showed no significant changes on either protein ( Fig. 4 F, left). Consistently, no overall effect on APP processing was observed in APP _695/swe ;AT-1 ^S113R/+ mice ( Fig. 4 F, right). Finally, to assess whether changes in amyloid-β were due to increased levels of amyloid-β-degrading proteases rather than to autophagy activation, we also determined levels of neprilysin and insulin-degrading enzyme, the two most prominent amyloid-β-degrading proteases ( Wang et al. , 2006 ). However, no changes were observed ( Fig. 4 F, left).

Figure 2.

Increased autophagy in AT-1 ^S113R/+ mice did not rescue the phenotype of mHtt ^Q160 mice. ( A–C ) Lifespan ( A ), body weight ( B ) and clasping ( C ) of indicated animals. Values in B are mean ± SD. Total numbers: females mHtt ^Q160 , n = 17; females mHtt ^Q160 ;AT-1 ^S113R/+ , n = 14; males mHtt ^Q160 , n = 18; males mHtt ^Q160 ;AT-1 ^S113R/+ , n = 18. ( D ) Western blot assessment of Htt aggregates in the striatum (Sm), cortex (Cx), and cerebellum (Cb). Lane 1: mHtt ^Q160 ;AT-1 ^S113R/+ mice; Lane 2: mHtt ^Q160 mice. Animals (males) were 2 months old when analysed.

Figure 3.

Increased autophagy in AT-1 ^S113R/+ mice did not rescue the phenotype of hSOD1 ^G43A mice. ( A ) Median survival in hSOD1 ^G93A (163 days; n = 21) and hSOD1 ^G93A ;AT-1 ^S113R/+ (167 days; n = 18) mice. ( B ) Median onset of symptoms in hSOD1 ^G93A (121 days; n = 20) and hSOD1 ^G93A ;AT-1 ^S113R/+ (121 days; n = 17) mice. ( C ) Hindlimb grip-strength in wild-type ( n = 15), hSOD1 ^G93A ( n = 20), and hSOD1 ^G93A ;AT-1 ^S113R/+ ( n = 16) at 60 and 120 days. No significant difference was observed between hSOD1 ^G93A and hSOD1 ^G93A ;AT-1 ^S113R/+ mice. Average grip-strength in wild-type mice was 100.0 ± 9.8 gf for males and 93.1 ± 16.2 gf for females at 60 days; and 103.2 ± 11.2 gf for males and 101.9 ± 9.1 gf for females at 120 days. Data are mean ± SD. ^#P < 0.0005 (from wild-type). ( D ) Western blot against hSOD1 in the spinal cord of early symptomatic hSOD1 ^G93A and hSOD1 ^G93A ;AT-1 ^S113R/+ mice after differential detergent extraction. Non-ionic detergent insoluble hSOD1 (insoluble; P2 pellet) is shown in the upper panel. The lower panel shows hSOD1 in the NP-40 soluble fraction (soluble; S1 supernatant).

Figure 4.

Increased autophagy in AT-1 ^S113R/+ mice rescued the phenotype of APP _695/swe mice. ( A ) Lifespan of indicated mice (Kaplan-Meier analysis). Numbers used: females APP _695/swe , n = 18; females APP _695/swe ;AT-1 ^S113R/+ , n = 18; males APP _695/swe , n = 25; males APP _695/swe ;AT-1 ^S113R/+ , n = 26. * P < 0.05; ** P < 0.005. Lifespan of APP _695/swe mice was similar to already published data (reviewed in Lalonde et al. , 2012 ). ( B ) Immunohistochemistry for intracellular amyloid-β aggregates. Two anti-amyloid-β antibodies were used (6E10 and 4E12). High magnification of indicated areas is shown. Animals (males) were 8 months old when analysed. ( C ) Western blot of extracellular amyloid-β oligomers in brain homogenate. Indicated bands correspond to already characterized amyloid-β oligomers ( Lesne et al. , 2006 ; Pehar et al. , 2010 ). soluble APP (sAPP) is also indicated. Animals (males) were 8 months old when analysed. ( D ) Quantification of major amyloid-β reactive species shown in C . Values are mean ( n = 3) ± SD. * P < 0.05. ( E ) Long-term potentiation induction in hippocampal slices. APP _695/swe mice lack the late component of long-term potentiation; these deficits were rescued by the AT-1 haploinsufficiency. Typical long-term potentiation of wild-type/non-transgenic animals is shown in Fig. 6 C. Values are mean ± SD. ^#P < 0.0005. ( F ) Western blot showing levels of BACE1, APP, NEP, IDE, C99 and C83.

In conclusion, the above results indicate that the increased activation of autophagy observed in AT-1 ^S113R/+ mice can selectively rescue the accumulation of amyloid-β toxic protein aggregates and the phenotype of APP _695/swe mice. Together with Figs 1–3 , they support the conclusion that a more efficient autophagy, as induced by targeting the ER-acetylation machinery, can resolve the accumulation of toxic protein aggregates that form within the secretory pathway but not those that form or accumulate in other compartments. Interestingly, AT-1 ^S113R/+ mice also show activation of unfolded protein response markers ( Supplementary Fig. 2 ) supporting the general conclusion of improved proteostatic mechanisms acting in the secretory pathway.

Biochemical inhibition of ATase1 and ATase2 in the mouse rescues the Alzheimer’s disease-like phenotype

To confirm the results obtained with AT-1 ^S113R/+ mice, we decided to target the ER-based acetyltransferases (ATase1 or ATase2), which act downstream of AT-1 ( Ko and Puglielli, 2009 ; Ding et al. , 2012 ). Specifically, we used recently identified and highly selective ATase1/ATase2 biochemical inhibitors ( Ding et al. , 2012 ). The biochemical properties as well as mechanism of action of ATase1/ATase2 inhibitors (Compounds 9 and 19) have already been described ( Ding et al. , 2012 ). The molecular characteristics of both compounds predicted excellent drug-like properties ( Supplementary Table 1 ). Although Compound 19 displayed increased solubility in aqueous systems ( Supplementary Table 2 ), Compound 9 had higher cLog P and was predicted to cross the blood–brain barrier with higher efficiency. As such, we decided to treat the animals with Compound 9. The highest concentration of the compound into the solid diet that could be reached without altering evident physical characteristics of the diet was 1.25 mg/g with multiple ethanol coating of sugar pellets and 2 mg/g with dustless extrusion of regular rodent pellets. For our studies we decided to use the dustless extrusion process ( Supplementary Table 3 ). The compound was administered at the final dose of 50 mg/kg/day.

To assess whether the compound was indeed able to reach the CSF, an initial group of mice received the compound for 1 week prior to collection of the CSF. Treatment was limited to 1 week, which is usually sufficient to reach equilibrium in biological fluids ( Ito et al. , 1998 ; Singh, 2006 ; Houston and Galetin, 2008 ). Concentration and duration of the treatment was based on previous studies with drug-like compounds having similar mass and solubility properties ( Ito et al. , 1998 ; Singh, 2006 ; Houston and Galetin, 2008 ). Assessment of the CSF by mass spectrometry identified the compound in all treated animals but not in control (untreated) animals, confirming our early prediction ( Supplementary Fig. 3 ).

In light of these results we decided to begin a long-term study with APP _695/swe mice. The animals received Compound 9 throughout the entire duration of the study. They develop Alzheimer’s disease-like neuropathology in an age-dependent fashion (reviewed in Duyckaerts et al. , 2008 ; Lalonde et al. , 2012 ); therefore, different disease-relevant manifestations were studied at different time points ( Fig. 5 A). When assessed at 5 months of age, APP _695/swe mice treated with Compound 9 displayed reduced levels of BACE1 ( Fig. 5 B) and soluble amyloid-β ( Fig. 5 C). BACE1 is the rate-limiting enzyme for the generation of amyloid-β from APP and a well characterized substrate of the ATases ( Ko and Puglielli, 2009 ; Ding et al. , 2012 , 2014 ). Thus, the reduced levels of BACE1 indicate successful inhibition of the ATases in the brain. APP _695/swe mice treated with Compound 9 also displayed reduced levels of p62, reduced LC3βI/LC3βII ratio, and increased LC3βII/actin ratio in the brain ( Fig. 5 D and E). The induction of autophagy is normally accompanied by reduced levels of p62, an autophagosome-associated protein that is degraded as a result of the autophagic process, as well as conversion of LC3βI into the autophagosome-bound LC3βII ( Mizushima et al. , 2010 ). As such, the results displayed in Fig. 5 D and E suggest that, similar to mice with reduced AT-1 activity (AT-1 ^S113R/+ ) ( Peng et al. , 2014 ), Compound 9-treated animals also display increased autophagy. This conclusion was further confirmed by the identification of LC3β staining in neurons of APP _695/swe mice treated with Compound 9 ( Fig. 5 F). Importantly, LC3β autophagy puncta were observed throughout the brain and showed complete overlap with NeuN ( Fig. 5 F). No LC3β puncta were observed in mice fed the control diet. We previously published that the induction of autophagy that results from the inhibition of ER acetylation depends on the acetylation status of ATG9A; specifically, reduced acetylation stimulates while increased acetylation blocks the induction of autophagy ( Pehar and Puglielli, 2013 ; Peng et al. , 2014 ). To confirm that Compound 9 acts through the same molecular pathway, we determined the acetylation status of ATG9A in cells treated with Compound 9. We observed a significant reduction in acetylated ATG9A ( Fig. 5 G), thus confirming our overarching conclusions.

Figure 5.

Biochemical inhibition of ATase1 and ATase2 rescued the phenotype of APP _695/swe mice (males) displayed at 5 months of age. ( A ) Schematic view of the study plan. Description is in the text. ( B ) Western blot analysis of BACE1 levels. Control = control diet; Cmpd 9 = Compound 9. ( C ) ELISA determination of amyloid-β levels. Values are mean ( n = 7) ± SD. ( D ) Western blot assessment of commonly-used autophagy markers. ( E ) Quantification of results shown in D . Values are mean ( n = 6) ± SD. * P < 0.05. ( F ) Immunolabelling showing LC3β puncta in Compound 9-treated animals (brain cortex). LC3β positive labelling co-localizes with NeuN staining. ( G ) Western blot showing the acetylation status of ATG9A following Compound 9 treatment (10 µM; 3 days; n = 4) of ATG9A overexpressing H4 cells (H4 _Atg9A ). Immunoprecipitation (IP) was performed with an anti-acetylated lysine antibody (AcK). INPUT is also shown. Levels of acetylated ATG9A (Atg9A-Ac) were normalized to the INPUT.

At 12 months of age, APP _695/swe mice display high molecular mass amyloid-β species (oligomers); they originate from the aggregation of the monomeric peptide and are highly toxic ( Cleary et al. , 2005 ; Lesne et al. , 2006 ; Pehar et al. , 2010 ). Treatment with Compound 9 resulted in a marked decrease in the levels of amyloid-β oligomers observed ( Fig. 6 A and B). As expected, APP _695/swe mice displayed a significant defect in the late phase of long-term potentiation, which is an indication of impaired synaptic plasticity; however, this defect was completely rescued by Compound 9 ( Fig. 6 C). Changes in long-term potentiation were not observed when the compound was administered to control (non-transgenic) mice ( Supplementary Fig. 4C ) indicating that treatment does not affect intrinsic synaptic activities but only rescues disease-relevant features.

Figure 6.

Biochemical inhibition of ATase1 and ATase2 rescued the phenotype of APP _695/swe mice (males) displayed at 12 months of age. ( A ) Western blot of extracellular amyloid-β oligomers in brain homogenate. Indicated bands correspond to already characterized amyloid-β oligomers ( Lesne et al. , 2006 ; Pehar et al. , 2010 ). Band specificity and loading controls are shown in Supplementary Fig. 4A and Supplementary Data . ( B ) Quantification of major amyloid-β reactive species shown in ( A ). ** P < 0.005; ^#P < 0.0005. ( C ) Long-term potentiation induction in hippocampal slices of indicated animals. Values are mean ± SD. ^#P < 0.0005.

When assessed at 16 months of age, APP _695/swe mice displayed severe amyloid-β pathology, as indicated by the high number of plaque formation throughout the brain parenchyma; this phenotype, which is typical of Alzheimer’s disease, was rescued by Compound 9 ( Fig. 7 A). Histological assessment also revealed reduced phospho-tau immunostaining, which was paralleled by increased synaptophysin labelling ( Fig. 7 B). Hyper-phosphorylation of tau and loss of the ‘synaptic mesh’ are typical features of Alzheimer’s disease. The drastic effect on levels of tau phosphorylation was also observed by immunoblotting ( Fig. 7 C and D). In addition to reduced levels of phospho-tau, western blot assessment of brain homogenates also detected a slight decrease in the C-terminal fragments of APP, C99 and C83 ( Fig. 7 C and D). Untreated APP _695/swe mice displayed reduced survival with 50% lethality at ∼14 months of age. However, this early lethality was completely rescued by Compound 9 treatment resulting in a normal lifespan ( Fig. 7 E and F). Routine histological assessment of peripheral tissues ( Supplementary Fig. 5 ) detected no evidence of toxicity associated with the treatment.

Figure 7.

Biochemical inhibition of ATase1 and ATase2 rescued the phenotype of APP _695/swe mice (males) displayed at 16 months of age. ( A ) Immunohistochemistry to visualize amyloid plaques. High magnification of indicated areas is shown. ( B ) Immunolabelling for phosphorylated tau (pTau) and synaptophysin in the CA3 region of the hippocampus. ( C ) Brain homogenate (cortex) of indicated animals were analysed by western blot. Levels of phosphorylated tau (pTau) were determined with two different phospho-specific antibodies (AT8 and 8EC6/C11). Total Tau and full-length APP (APP f.l.) served as internal loading controls. Only levels of pTau and C99/C83 showed significant changes ( D ). ( D ) Quantification of pTau, C99, and C83 levels shown in C . Values are mean ± SD. * P < 0.05; ** P < 0.005; ^#P < 0.0005. ( E and F ) Lifespan of wild-type/non-transgenic (Non-Tg) and APP _695/swe mice (males) fed a control diet ( E ) or a diet containing Compound 9 ( F ). Kaplan-Meier analysis is shown. Numbers used were as follows: Non-Tg, control diet, n = 30; Compound 9, n = 39); APP _695/swe (control diet, n = 26; Compound 9, n = 30). ^#P < 0.0005.

When taken together, the above results indicate that biochemical inhibition of ATase1/ATase2 can rescue the Alzheimer’s disease-like phenotype displayed by APP _695/swe mice in the absence of evident toxicity. This effect is likely due to a combination of events, among which reduced generation and increased disposal of toxic amyloid-β aggregates through autophagy.

Discussion

Here, we show that Nε-lysine acetylation in the ER lumen regulates normal proteostasis of the secretory pathway. We also report that inhibition of ER-acetylation can rescue diseases associated with accumulation of toxic protein aggregates that form within the secretory pathway. These conclusions were reached by using ex vivo and in vivo models. The latter included mice deficient in AT-1 transport activity as well as mice treated with ATase1/ATase2 specific inhibitors.

The major difference between AT-1 ^S113R/+ and Compound 9-treated animals was in BACE1 and APP metabolism. Indeed, both mouse models displayed increased autophagy (see Figs 4 and 5 and Peng et al. , 2014 ), but only Compound 9-treated animals displayed reduced levels of BACE1 and reduced processing of APP (compare Fig. 4 with Figs 5 and 7 ). The likely explanation is in the kinetics of acetylation of individual substrates. Specifically, the levels of acetylCoA influx into the ER of AT-1 ^S113R/+ mice are sufficiently low to affect the acetylation of ATG9A and stimulate autophagy ( Peng et al. , 2014 ), but not low enough to affect the levels of BACE1. Direct inhibition of the transferases, instead, is able to affect a larger number of substrates, thus reducing the generation of amyloid-β but also stimulating the autophagy-mediated disposal of amyloid-β aggregates. The existence of different substrate-saturation kinetics for ER acetylation is supported by published studies ( Costantini et al. , 2007 ; Mak et al. , 2014 ). Although it is impossible to dissect the specific contribution of the above mechanisms in the Compound 9-treated model, the results displayed in Figs 1 and 4 clearly suggest that the autophagy-mediated disposal mechanism is an important component. Obviously, it is also possible that additional and not yet characterized effects of Compound 9 (such as regulation of proteasome activity) might contribute in the phenotypic correction.

One of the functions of the ER is to ensure that nascent membrane and secreted polypeptides fold correctly. Incorrectly folded polypeptides, which failed quality control, must be sorted and disposed of. For this purpose, transient modifications have been designed to select correctly folded and unfolded/misfolded polypeptides. Studies conducted with two well-characterized substrates of the ATases, BACE1 and CD133, suggest that Nε-lysine acetylation might be part of normal quality control to select correctly folded polypeptides ( Costantini et al. , 2007 ; Ko and Puglielli, 2009 ; Ding et al. , 2014 ; Mak et al. , 2014 ). Indeed, a block in the acetylation of both BACE1 and CD133 resulted in the nascent protein being retained and disposed of in the early secretory pathway ( Costantini et al. , 2007 ; Mak et al. , 2014 ). Recognition features that control the activity of the acetyltransferases as well as the identity of all ATase1- and ATase2-specific substrates still remain to be determined.

Another important function of the ER is to dispose of unfolded/misfolded polypeptides as part of ER-associated degradation. Monomeric proteins that can be retro-translocated to the cytosol across the ER membrane are preferentially degraded by the proteasome ( Trombetta and Parodi, 2003 ). In contrast, large protein aggregates are mostly dealt with by expanding the ER and activating autophagy ( Bernales et al. , 2006 ; Ogata et al. , 2006 ; Ding et al. , 2007 ; Axe et al. , 2008 ). Therefore, autophagy is as an essential cellular function that ensures disposal of unwanted material. This is particularly important for neurons ( Klionsky, 2006 ; Komatsu et al. , 2006 , 2007 ; Lee, 2009 ; Pehar and Puglielli, 2013 ). If unchecked, autophagy can become terminal. Indeed, aberrant induction of autophagy in AT-1 ^S113R/+ mice resulted in a severe phenotype ( Peng et al. , 2014 ). However, compelling data also indicate that the autophagy machinery can be manipulated to improve the disposal of toxic protein aggregates. The same data also indicate that increased levels of autophagy, which are pathogenic in wild-type mice in the absence of toxic protein aggregates, can be beneficial in mouse models of diseases characterized by increased accumulation of toxic protein aggregates ( van Dellen et al. , 2000 ; Pickford et al. , 2008 ; Hetz et al. , 2009 ; Madeo et al. , 2009 ; Bhuiyan et al. , 2013 ).

Autophagy can be induced through different mechanisms, and to respond to different insults ( Klionsky, 2006 ; Lee, 2009 ; Pehar and Puglielli, 2013 ). As such, it is likely that a certain degree of specificity exists. The studies performed with α-synuclein ( Fig. 1 ) indicate that the increased levels of autophagy observed in AT-1 ^S113R/+ mice affect the disposal of protein aggregates that form within the secretory pathway but not in other compartments. The same conclusions were reached by using mouse models of disease ( Figs 2–4 ). Whether there are more precise spatial restrictions within the secretory pathway is unclear. Previous data have shown that the acetylation status of ATG9A is crucial for the induction of autophagy downstream of AT-1 ( Pehar et al. , 2012 b ; Peng et al. , 2014 ). ATG9A is the only membrane-bound autophagy protein and can be recruited to LC3β-positive autophagosomes from different locations, including the ER, the Golgi apparatus, and even the plasma membrane ( Young et al. , 2006 ; Ohashi and Munro, 2010 ; Tamura et al. , 2010 ; Puri et al. , 2013 ; Bejarano et al. , 2014 ). Therefore, it will be difficult to delineate possible spatial restrictions by targeting ATG9A. A similar limitation exists with amyloid-β, which can be generated in different cellular compartments, including the ER and the Golgi apparatus ( Haass et al. , 1995 ; Cook et al. , 1997 ; Takami et al. , 2009 ). Interestingly, autophagy deficiency, as caused by genetic disruption of ATG7, leads to aberrant accumulation of intracellular amyloid-β in the early secretory pathway and results in an exacerbated neurodegenerative phenotype ( Nilsson et al. , 2013 , 2015 ).

It is also worth noting that the role of autophagy in neurodegenerative mouse models is not completely straightforward. Indeed, alterations of specific regulatory steps of the autophagy process, including impaired fusion of autophagosomes with lysosomes, inefficient degradation of the cargo, or defective cytosolic cargo recognition, have been described in certain models causing autophagy induction to be detrimental ( Marino et al. , 2011 ; Vidal et al. , 2014 ). Although activation of autophagy has been shown to be protective in models of Alzheimer’s disease, Huntington’s disease, and amyotrophic lateral sclerosis (reviewed in Vidal et al. , 2014 ), the final outcome on disease progression appears to depend on the specific regulatory step being targeted and the pathological context. Accordingly, in mutant hSOD1 amyotrophic lateral sclerosis mouse models, activation of autophagy by rapamycin has detrimental or no effect, while activation of autophagy by trehalose treatment or downregulation of XBP1 decreases hSOD1 aggregates and enhances motor neuron survival ( Zhang et al. , 2011 ; Bhattacharya et al. , 2012 ; Vidal et al. , 2012 ; Castillo et al. , 2013 ). On the other hand, haploinsufficiency of beclin 1, a key player in the initiation steps of autophagy, appears to be beneficial by reversing autophagy alterations induced by an abnormal interaction of mutant hSOD1 with the beclin 1/BCL-X _L complex ( Nassif et al. , 2014 ). Thus, we cannot exclude the possibility that the presence of defects in other autophagy regulatory steps counteract the protective effect of increasing ER proteostasis in the Huntington’s disease and amyotrophic lateral sclerosis mouse models used here. It is also important to consider that, although hSOD1 lacks a signal peptide, it can translocate into the secretory pathway through a not well characterized mechanism that involves ATP consumption ( Urushitani et al. , 2008 ). Indeed, hSOD1 aggregates have been described within the secretory pathway ( Urushitani et al. , 2008 ). Thus, in the case of hSOD1 ^G93A , the disposal of hSOD1 aggregates within the secretory pathway might account for the small reduction in hSOD1 aggregates observed in the spinal cord of early symptomatic hSOD1 ^G93A ;AT-1 ^S113R/+ mice ( Fig. 3 ). If this is true, then we could assume that the toxicity of mutant hSOD1 in other cellular compartments prevented rescue of the phenotype. In the case of Huntington’s disease models, mutant huntingtin has been reported to alter the activity of the ubiquitin-proteasome system, thus interfering with both cytosolic protein degradation and ER-associated degradation ( Duennwald and Lindquist, 2008 ; Leitman et al. , 2013 ). The inhibition of ER-associated degradation promotes the accumulation of misfolded proteins in the ER and the subsequent activation of the unfolded protein response. However, we observed that the increased autophagy-dependent ER-associated degradation [ERAD(II)] associated with haploinsufficiency of AT-1 in the double transgenic mHtt ^Q160 ;AT-1 ^S113R/+ mice is not sufficient to revert the phenotype.

In conclusion, our results indicate that there is significant specificity in the induction of autophagy; they also support therapies targeting ER acetyltransferases, ATase1 and ATase2, for a specific subset of chronic degenerative diseases.

Glossary

Abbreviations

AT-1

acetyl CoA transporter 1

ATase1

acetyltransferase 1

ATase2

acetyltransferase 2

Atg

autophagy protein

ER

endoplasmic reticulum

MEF

mouse embryo fibroblast