Activation of Asparaginyl Endopeptidase Leads to Tau Hyperphosphorylation in Alzheimer Disease*

Background: Abnormal hyperphosphorylation of the microtubule-associated protein Tau is a hallmark of Alzheimer disease. Results: Acidosis of the brain activates and translocates asparaginyl endopeptidase from neuronal lysosomes to the cytoplasm where it leads to Tau hyperphosphorylation by inhibition of protein phosphatase 2A through cleavage of its inhibitor 2 into two active fragments. Conclusion: Activated asparaginyl endopeptidase causes Tau pathology. Significance: Brain acidosis can trigger Tau hyperphosphorylation. Neurofibrillary pathology of abnormally hyperphosphorylated Tau is a key lesion of Alzheimer disease and other tauopathies, and its density in the brain directly correlates with dementia. The phosphorylation of Tau is regulated by protein phosphatase 2A, which in turn is regulated by inhibitor 2, I2PP2A. In acidic conditions such as generated by brain ischemia and hypoxia, especially in association with hyperglycemia as in diabetes, I2PP2A is cleaved by asparaginyl endopeptidase at Asn-175 into the N-terminal fragment (I2NTF) and the C-terminal fragment (I2CTF). Both I2NTF and I2CTF are known to bind to the catalytic subunit of protein phosphatase 2A and inhibit its activity. Here we show that the level of activated asparaginyl endopeptidase is significantly increased, and this enzyme and I2PP2A translocate, respectively, from neuronal lysosomes and nucleus to the cytoplasm where they interact and are associated with hyperphosphorylated Tau in Alzheimer disease brain. Asparaginyl endopeptidase from Alzheimer disease brain could cleave GST-I2PP2A, except when I2PP2A was mutated at the cleavage site Asn-175 to Gln. Finally, an induction of acidosis by treatment with kainic acid or pH 6.0 medium activated asparaginyl endopeptidase and consequently produced the cleavage of I2PP2A, inhibition of protein phosphatase 2A, and hyperphosphorylation of Tau, and the knockdown of asparaginyl endopeptidase with siRNA abolished this pathway in SH-SY5Y cells. These findings suggest the involvement of brain acidosis in the etiopathogenesis of Alzheimer disease, and asparaginyl endopeptidase-I2PP2A-protein phosphatase 2A-Tau hyperphosphorylation pathway as a therapeutic target.

In acidic conditions such as generated by brain ischemia and hypoxia, especially in association with hyperglycemia as in diabetes, I 2 PP2A is cleaved by asparaginyl endopeptidase at Asn-175 into the N-terminal fragment (I 2NTF ) and the C-terminal fragment (I 2CTF ). Both I 2NTF and I 2CTF are known to bind to the catalytic subunit of protein phosphatase 2A and inhibit its activity. Here we show that the level of activated asparaginyl endopeptidase is significantly increased, and this enzyme and I 2

PP2A
translocate, respectively, from neuronal lysosomes and nucleus to the cytoplasm where they interact and are associated with hyperphosphorylated Tau in Alzheimer disease brain. Asparaginyl endopeptidase from Alzheimer disease brain could cleave GST-I 2 PP2A , except when I 2 PP2A was mutated at the cleavage site Asn-175 to Gln. Finally, an induction of acidosis by treatment with kainic acid or pH 6.0 medium activated asparaginyl endopeptidase and consequently produced the cleavage of I 2 PP2A , inhibition of protein phosphatase 2A, and hyperphosphorylation of Tau, and the knockdown of asparaginyl endopeptidase with siRNA abolished this pathway in SH-SY5Y cells. These findings suggest the involvement of brain acidosis in the etiopathogenesis of Alzheimer disease, and asparaginyl endopeptidase-I 2 PP2A -protein phosphatase 2A-Tau hyperphosphorylation pathway as a therapeutic target.
Alzheimer disease (AD) 2 is characterized by neuronal degeneration associated with the abnormal hyperphosphorylation and aggregation of the microtubule associated protein Tau and extracellular deposits of A␤. Aggregation of both the altered Tau and amyloid ␤ have been shown to be associated with brain acidosis (1)(2)(3)(4). Furthermore, ischemia and hypoxia, especially in association with hyperglycemia, lower brain pH. The fall in pH may result in a decrease in cellular metabolic activity, accumulation of acidic metabolites in the cell, and modulation of lysosomal enzymes (5)(6)(7).
Lysosomal enzyme, asparaginyl endopeptidase (AEP, also known as legumain), is a cysteine proteinase that is synthesized as a zymogen (pro-AEP, 56 kDa) and autocatalytically processed into active AEP under acidic conditions. Processing of AEP requires sequential removal of the C-and N-terminal propeptides at different pH thresholds to generate 46-kDa (active pro-AEP) and 36-kDa (active AEP) active enzymes (8,9). AEP has multiple roles depending on substrate specificity such as initiator of invariant chain processing during MHC class IImediated antigen presentation (10), inducer of cell migration (11,12), and modulator of processes such as proliferation (13) and both pro-death and pro-survival functions using apoptotic mechanisms (14). The variety of roles of AEP is due to its ability to proteolyse multiple substrates, including among others the extracellular matrix protein fibronectin (15), progelatinase (16), and cathepsin H, B, and L (17). One protein shown to be cleaved by AEP during acidosis in both cellular and animal models of stroke is inhibitor 2 (I 2 PP2A ) of protein phosphatase 2A (PP2A) (18,19). I 2 PP2A has been described as a potent inhibitor of PP2A (20), a phosphatase that accounts for ϳ70% of the adult human brain phosphoserine/phosphothreonine Tau protein phosphatase activity (21). PP2A activity is compromised in AD and is believed to be a cause of the Tau neurofibrillary pathology (22)(23)(24) , also known as SET, TAF-1␤, or PHAP II, is a 39-kDa protein widely expressed in different tissues and mainly localized in the nucleus (25). Despite its nuclear location, I 2 PP2A is able to move freely to the cytoplasm to induce cell migration (26,27). Moreover, I 2 PP2A is involved in the control of cell cycling (28), gene transcription (29), and protection of histone from acetylation (30), and it is even suggested that I 2 PP2A is a part of the neuronal apoptotic pathway related to AD (31). I 2 PP2A was found to be selectively cleaved at Asn-175 in AD brain (32), and the PP2A activity is inhibited by the interaction of the two I 2 PP2A cleavage products: N-and C-terminal fragments (I 2NTF and I 2CTF ) with the PP2A catalytic subunit PP2Ac (33). Furthermore, expression of I 2NTF and I 2CTF in rat brain induced inhibition of PP2A activity, abnormal hyperphosphorylation and aggregation of Tau, intraneuronal accumulation of A␤, and neurodegeneration (34,35). If the PP2A activity is decreased in AD by its interaction with either the I 2NTF or I 2CTF fragment, then AEP can be a responsible enzyme that starts this pathological mechanism.
In this study, we show that AEP is highly activated in AD brain and is translocated from neuronal lysosomes to the cytoplasm where I 2 PP2A is also translocated from the nucleus, suggesting that the cleavage of I 2 PP2A at Asn-175 by AEP takes place mainly in this compartment. Furthermore, we show that the cleavage of I 2 PP2A by AEP can trigger one of the main pathways for the AD-type hyperphosphorylation of Tau.

EXPERIMENTAL PROCEDURES
Human Brain Tissue-Frozen tissue samples of frontal cortices from autopsied and histopathologically confirmed AD and control cases were obtained from the Sun Health Research Institute Brain Donation Program (Sun City, AZ) and from the Harvard Brain Bank of McLean Hospital (Belmont, MA). The tissue was homogenized to 10% (w/v) final concentration in cold buffer containing 50 mM Tris-HCl, pH 7.4, 8.5% sucrose, 2 mM EDTA, 2 mM EGTA, 10 mM ␤-mercaptoethanol, 5 mM benzamidine, 0.5 mM 4-(2-aminoethyl) benzenesulfonyl fluoride, 4 g/ml pepstatin A, and 10 g/ml each of aprotinin and leupeptin, 20 mM ␤-glycerolphosphate, 100 mM sodium fluoride, 1 mM sodium vanadate, and 100 nM okadaic acid and used for Western blots. The ages, genders, and postmortem delays of the AD and control cases are listed in Table 1. The use of frozen human brain tissue was in accordance with National Institutes of Health guidelines and was approved by the Institutional Review Board of the New York State Institute for Basic Research in Developmental Disabilities. Formalin-fixed AD and control brain paraffin-embedded sections (5-m thickness) were used for immunohistochemistry.
Subcellular Fractionation-Lysosomes, nuclei, and cytosol were isolated from human brain tissue as previously described (36). Briefly 5 g of brain tissue was homogenized in 3 volumes of cold homogenizing buffer containing 0.32 M sucrose, 1 mM EDTA disodium, 10 mM HEPES, pH 7.4, using Potter-Elvehjem homogenizer with 5-10 gentle strokes at 500 rpm. Homogenates were centrifuged at 750 ϫ g for 10 min in a swinging bucket rotor. The supernatant was kept, and the pellet was resuspended with 3-4 strokes in 10 ml of the buffer and centrifuged as above. The second supernatant was combined with the previous one, and the pellet was saved as the nuclear fraction. The pooled supernatant was centrifuged at 20,000 ϫ g for 10 min in a fixed angle rotor. The pellet was saved, the supernatant was centrifuged at 105,000 ϫ g for 1 h, and the resulting supernatant was saved as the cytosolic fraction. The 20,000 ϫ g pellet from above was resuspended with 3-4 strokes in 20 ml of the homogenizing buffer and centrifuged at 20,000 ϫ g for 10 min. The supernatant was discarded, and the pellet was resuspended with 3-4 strokes in 4 ml of the buffer and layered over 36 ml of 27% (v/v) Percoll and centrifuged at 20,000 ϫ g for 90 min. The lysosomal band in the bottom 1-2 ml was collected and centrifuged at 100,000 ϫ g for 1 h, and the turbid layer of lysosomes just above the pellet was collected. After protein quantification, the samples were resuspended in Laemmli buffer and used for Western blots.
Immunoprecipitation and Proteolysis Assay-AD and control brains were homogenized to 10% (w/v) final concentrations in cold buffer containing 50 mM sodium citrate, pH 5.5, 0.1 M NaCl, 1 mM EDTA, and 2 mM ␤-mercaptoethanol. The homogenate was subjected to three cycles of freezing and thawing to disrupt organelles, followed by centrifugation at 12,000 ϫ g for 15 min. Protein concentration was determined by modified Lowery, and 150 g of the 12,000 ϫ g extract was precleared with 45 l of protein G beads (Pierce) for 4 h. After low speed centrifugation, the beads were kept as a control for nonspecific binding, and the extract was incubated on a rotater with 1.9 g of anti-human legumain (AEP; 1:500; R & D Systems, Minneapolis, MN) and 45 l of protein G beads overnight at 4°C. The sample was centrifuged at low speed, the supernatant contained the immunodepleted fraction, and the beads were used as the IP fraction. To perform the proteolysis assay, 50 ng of either GST-I 2 PP2A -WT (wild-type) or GST-I 2 PP2A -M (Asn-175 mutated to Gln) were incubated in the presence of 10 g of extract, IP, immunodepleted, cytosolic, lysosomal, or nuclear fraction at 30°C for different times, and the reaction was stopped by boiling the samples in Laemmli buffer; GST-I 2 PP2A -WT treated identically but with buffer only served as a control.
Co-immunoprecipitation-Cos-7 cells were transfected with pTRE2hyg/HA-I 2 PP2A and lysed by sonication in 10 mM Tris-HCl, pH 7.6, 50 mM NaCl, 1 mM EDTA, 1 mM EGTA, 10 g/ml aprotonin, and 10 g/ml pepstatin. The lysate was centrifuged at 12,000 ϫ g for 15 min. The supernatant was incubated with anti-legumain (AEP) precoupled onto protein G-agarose (4 g of goat anti-AEP/500 g of cell lysate) overnight at 4°C. After extensive washing, the bound proteins were eluted from the beads by boiling in Laemmli sample buffer and subjected to Western blot analyses. The IP products were detected with goat anti-AEP (1:200; R & D Systems) and co-IP with mouse anti-HA (1:10,000; Sigma Clone HA-7).
Metabolically Active Rat Brain Slices-The protocol was as described previously (37). Briefly, male Wistar rats (Charles River Lab, Wilmington, MA) 2-3 months old were injected intraperitoneally with 50 mg/kg pentobarbital and decapitated when deeply anesthetized. The brains were immediately removed and cooled down in ice-cold artificial cerebrospinal fluid (aCSF) consisting of 126 mM NaCl, 3.5 mM KCl, 1.2 mM NaH 2 PO 4 , 1.3 mM MgCl 2 , 2.0 mM CaCl 2 , 11 mM D(ϩ)-glucose, 25 mM NaHCO 3 , pH 7.4 (control), 6.5, or 5.5, for 7-8 min. The hippocampus and surrounding cortex from each brain was dissected and chopped into 400-m-thick coronal slices with a Camden Vibraslicer (WP Inc., Sarasota, FL). The slices were transferred into a chamber containing the aCSF and incubated at 33°C for 2 h. The oxygenation of the aCSF was carried out by bubbling the solution with a mixture of 95% O 2 and 5% CO 2 during the entire procedure. At the end of incubation, the brain slices were homogenized with homogenizing buffer (described above) at a ratio of 9 ml of buffer/1.0-g tissue slice. After protein assays, the samples were prepared for SDS-PAGE and subjected to Western blots.
PP2A Activity Assay-PP2A activity was assayed as described previously (38). Briefly, an ELISA was used to assay PP2A activity in 12,000 ϫ g extract, of metabolically active rat brain slices. No phosphatase inhibitors were used in the homogenizing buffer for this protocol. A 96-well plate was coated for 8 h at room temperature with 60 l of coating buffer (35 mM NaHCO 3 ) containing 8.0 g/ml synthetic Tau phosphopeptide p17 corresponding to Tau amino acid residues 194 -207, to which three lysines were added at the C-terminal end (Ac-Arg-Ser-Gly-Tyr-Ser-Ser(OPO 3 2Ϫ )-Pro-Gly-Ser-Pro-Gly-Thr-Pro-Gly-Lys-Lys-Lys-NH 2 ) to enhance its binding to the microtiter plate. For blocking, the coating solution was removed, and the wells were blocked with 150 l of protein-free blocking buffer (Pierce) at 4°C overnight. The plates were then washed two times for 15 min with 150 l/well 50 mM Tris-HCl, pH 7. The enzymatic reaction was performed with 60 l/well of 0.15 g of extract protein in reaction buffer containing 50 mM Tris-HCl, pH 7, 20 mM ␤-mercaptoethanol, 2 mM EGTA, 2 mM MnCl 2 , and 0.01 mg/ml BSA in the presence or absence of 20 nM okadaic acid as a specific inhibitor of PP2A. The plates were incubated for 60 min at 30°C in a moist chamber. To stop the enzyme reaction, the mixture was removed, and the wells were washed two times for 10 min with blocking buffer containing 50 mM NaF and then with the buffer without NaF. Each well was then incubated with 75 l of the monoclonal antibody Tau-1, which recognizes Tau unphosphorylated at Ser-198/199/202 (1:25000) (39) overnight at 4°C. The plates were quickly washed three times and two times further for 10 min with 50 mM TBS, 0.05% Tween 20, followed by the same washes with only TBS. The plates were incubated with secondary antibody (peroxidase-linked goat anti-mouse IgG, 1:5000) for 1 h at room temperature. Finally, the plates were washed as indicated previously, and the color reaction was performed using 70 l of tetramethylbenzidine/well, and the development was monitored for 30 min at 650 nm in a microtiter plate reader. The values of the samples with okadaic acid were subtracted from the corresponding samples without okadaic acid to obtain the PP2A activity.
Western Blots-Western blots were carried out using 10 or 15% SDS-PAGE, PVDF membrane of 0.45 m pore size and blocking with 5% skim milk. The following primary antibodies were used for the blots: anti-human legumain ( , and anti-histone 3 (1:1000; Santa Cruz Biotechnology). Immunoblots were probed with the corresponding anti-mouse, anti-rabbit (1:5,000; Jackson Immuno Research), or anti-goat HRP secondary antibodies (1:1000; Invitrogen) and detected using the enhanced chemiluminescence reagents (Thermo Scientific, Rockford, IL). Multigauge V3 software (Fuji Photo Film, Tokyo, Japan) was used to quantify the protein bands in Western blots. Student's t test was performed for the statistical analysis between two groups.
For diaminobenzidine staining, AD and control brain paraffin sections were processed as above except that the sections were treated with 0.5% hydrogen peroxide in PBS (Fisher) for 15 min after the antigen retrieval. The tissue sections were incubated with anti-human legumain (1:100), followed by anti-goat HRP-conjugated secondary antibody (1:250). Thereafter, a solution of 0.01% hydrogen peroxide combined with 0.06% 3,3Ј-diaminobenzidine (Sigma-Aldrich) in PBS was used to develop the reaction. The samples were counterstained with hematoxylin (Sigma-Aldrich) to visualize the nuclei.
To induce AEP knockdown, SH-SY5Y cells were cultured in advanced DMEM F-12 medium and transfected with siRNA against human AEP (siAEP; legumain siRNA(h); Santa Cruz Biotechnology) using Lipofectamine 2000 (Invitrogen) according to the manufacturer's instructions. After 48 h post-transfection, the cells were either lysed in pH 7.4 for control or pH 6.0 lysis buffer (as above, but without pepstatin and phosphatase inhibitors) or treated with 150 M kainic acid (Tocris Bioscience, Minneapolis, MN) for 24 h (18) and then lysed as above at pH 7.4; SH-SY5Y cells treated identically but without siAEP were used as a control. The cell lysates were subjected to 10% SDS-PAGE, followed by Western blots.

PP2A
-WT I 2 PP2A construct was subcloned into N-terminal GST-tagged pGEX-6P-1 plasmid (GE Healthcare) and transformed into Escherichia coli BL21 host strain for expression. A 500-ml culture of E. coli harboring pGEX-6P-1/ GST-I 2 PP2A -WT plasmid was grown at 37°C to an A level of ϳ600 and induced with 0.4 mM isopropyl ␤-D-1-thiogalactopyranoside at room temperature for 4 h before harvesting the bacteria by centrifugation. The bacteria were resuspended in 25 ml of cold PBS with protease inhibitor mixture (Sigma-Aldrich) and lysed by sonication and mixed after adding Triton X-100 to 1% final concentration. All purification procedures were carried out at 4°C. The total soluble fraction was recovered by centrifugation at 8500 ϫ g for 10 min. DTT was added to the extract to a final concentration of 5 mM, followed by incubation for 4 h with 5 ml of glutathione-Sepharose beads. The beads were then transferred to a disposable column and washed with PBS. The GST-I 2 PP2A was then eluted from the beads with 10 mM GSH in 50 mM Tris HCl, pH 8.5. The eluted protein was dialyzed and stored in aliquots at Ϫ80°C. Mutant I 2

PP2A
(I 2 PP2A -M) was constructed by PCR amplification from wild type I 2 PP2A as a template. The mutation was generated by amplifying individual fragments that contain the mutated region replacing Asn-175 for Gln using the following primers: forward, 5Ј-CGT TCG AGT CAA ACG CAG CAA AAA GCC AGC AGG AAG AGG-3Ј; and reverse, 5Ј-CCT CTT CCT GCT GGC TTT TTG CTG CGT TTG ACT CGA ACG-3Ј, and the sequences were flanked by BamHI and EcoRI for insertion into pGEX-6P-1 plasmid vector with the following primers: forward, 5Ј-CG GGA TCC ATG TCG GCG CCG GCG GCC AAA GTC AGT AAA AAG-3Ј; and reverse, 5Ј-CG GAA TTC CTA GTC ATC TTC TCC TTC ATC CTC CTC TCC-3Ј. The expression and purification of pGEX-6P-1/GST-I 2 PP2A -M was carried out using the same protocol as above for the wild type fusion protein.

Activation of AEP and Cleavage of I 2
PP2A Are Increased in AD Brain-To determine the possible involvement of AEP in the cleavage of I 2 PP2A , we analyzed frontal cortex homogenates from eight AD and ten non-AD control cases for levels of activated AEP and the cleaved I 2 PP2A by Western blots. We found that the level of the 36-kDa active AEP was higher in AD than control brains (Fig. 1A), and the level of pro-AEP (56 kDa) was lower in AD than in control cases (Fig. 1B). The level of I 2NTF , a product of I 2 PP2A cleavage by AEP, was found to be increased in AD as compared with control cases. These data raised the possibility of the involvement of AEP in the cleavage of I 2
To learn whether activated AEP cleaves I 2 PP2A , we investigated the interaction between the two proteins as follows. We transfected Cos-7 cells with PTRE2hyg/HA-I 2 PP2A and then immunoprecipitated AEP with anti-legumain (AEP). We found that I 2 PP2A co-immunoprecipitated with activated AEP (Fig.  1C), suggesting an interaction between the two proteins.
Translocation of AEP and I 2 PP2A to Cytoplasm Is Increased in AD Brain-Because AEP and I 2 PP2A are lysosomal and nuclear proteins, respectively, we evaluated whether there is an interaction between these two proteins and determined in which subcellular compartment this interaction could take place. To identify the location of these proteins, we performed immunohistofluorescence, followed by confocal microscopy on paraffin sections of the hippocampus from AD and control brains. We found increased I 2 PP2A cytoplasmic staining in AD relative to control brains ( Fig. 2A), and the co-localization between I 2 PP2A and the nuclear protein histone 3 was higher in control than in AD brains. Using LAMP-2 as a lysosomal marker, we observed a low level of co-localization between lysosomes and AEP in AD brains; the AEP staining was mainly cytoplasmic and diffuse, and the LAMP-2 staining showed vesicle formations that were larger than normal lysosomes. In contrast, we observed a high level of co-localization between LAMP-2 and AEP, as well as increased number and definition of the lysosomal structures in control cases (Fig. 2B). These data suggest that both AEP and I 2 PP2A undergo translocation respectively from lysosomes and nucleus to cytoplasm and that the interaction between these proteins probably takes place mainly in this compartment in AD brain.
AEP Is Translocated from Neuronal Lysosomes to Cytoplasm and Co-exists with I 2 PP2A and Abnormally Hyperphosphorylated Tau in AD-I 2 PP2A was previously shown to undergo cleavage into I 2NTF and I 2CTF and translocate from the neuronal nucleus to the cytoplasm in AD brain (32). However, translocation of AEP from lysosome to cytosol was not understood. To address this question, we studied the level of active AEP in subcellular fractions from AD and control brains. We used LAMP-2 to distinguish lysosomal from nonlysosomal fractions (Fig. 3). Western blot analysis by densitometry showed that active AEP is mostly concentrated in the cytosol more in AD than control brains (Fig. 3, A and B). Consistent with these data, immunohistochemical staining of AD brain with anti-AEP showed a marked signal in the neuronal cytoplasm along with lysosomes and a weaker reaction in the nucleus (Fig. 3D, right  inset); the tangle-bearing neurons showed especially robust cytoplasmic immunostaining (Fig. 3D, left inset). The level of active AEP in the nuclear fraction showed a small increase in AD as compared with control brains, but this increase was not significant (Fig. 3C).
Because it is already known that I 2 PP2A plays an important role in Tau hyperphosphorylation (38), we evaluated whether the cytosolic distribution of AEP is associated with hyperphosphorylated Tau in AD brain. For this purpose, hippocampal sections were subjected to triple staining with anti-AEP, anti-I 2 PP2A , and anti-pTau. The analysis of these data by confocal microscopy revealed that cells without Tau pathology had AEP immunostaining in lysosomes (Fig. 3E, arrow, punctated staining), whereas hyperphosphorylated Tau-positive neurons showed diffuse AEP staining in the cytoplasm (Fig. 3E, arrowhead). Moreover, I 2 PP2A cytoplasmic staining was found in hyperphosphorylated Tau-bearing neurons. These data suggest that translocation of AEP from neuronal lysosomes to cytoplasm could be involved in Tau pathology in AD.
AEP from the Cytosolic Fraction of AD Brain Cleaves Recombinant I 2 PP2A -The above findings suggested that in AD, I 2

PP2A
and AEP are released to the cytoplasm where these proteins can interact and I 2 PP2A can be cleaved by AEP. However, there is a possibility that an enzyme other than AEP could also be responsible for this cleavage under the same conditions. Because previous reports indicated that the cleavage of I 2 PP2A occurs at Asn-175 (18, 32), first we investigated whether the proteolytic activity from AD brain is able to cleave I 2 PP2A at Asn-175 to FIGURE 1. AEP interacts with I 2 PP2A and in AD brain the levels of activated AEP and I 2NTF are selectively increased. A, AD and control (Ctrl) human brain homogenates were subjected to Western blots developed with antibodies to AEP and I 2NTF . In AD brain, increases in the levels of both activated AEP (36-kDa band) and I 2 PP2A cleavage product I 2NTF were found. B, quantitation of Western blots showed a decrease in the level of pro-AEP and an increase in the activation of AEP and the cleavage of I 2 PP2A into I 2NTF in AD brains. C, immunoprecipitation of AEP with anti-AEP (legumain) showed co-immunoprecipitation of I 2 PP2A from lysates of Cos-7 cells transfected with HA-I 2 PP2A .
generate I 2NTF . Second, we determined whether AEP is the enzyme responsible for this cleavage. Third, we investigated in which subcellular compartment this cleavage takes place. To address these questions, we purified GST-I 2 PP2A recombinant fusion protein and subjected it to proteolysis with AD brain extract. GST-I 2 PP2A was cleaved in a time-dependent manner generating I 2NTF (Fig. 4A). To confirm that this cleavage was catalyzed by AEP, we mutated Asn-175 of I 2 PP2A to Gln in GST-I 2 PP2A and studied its proteolysis with brain extract. We found that wild type GST-I 2 PP2A but not the mutated GST-I 2 PP2A was cleaved when treated with the brain extract (Fig. 4B). To verify that only AEP in the brain extract was responsible for the cleavage of I 2 PP2A at Asn-175, we IP AEP from AD brain extract and performed the proteolysis assay as above. GST-I 2 PP2A was cleaved by IP-AEP but not by AEP immunodepleted brain extract (Fig. 4C). Because, as described above, we observed that both I 2 PP2A and active AEP were located in the cytosol from AD brain, we evaluated where in the cell I 2 PP2A cleavage could take place. After subcellular fractionation of the brain, GST-I 2 PP2A was subjected to proteolysis by cytosolic, lysosomal, and and AEP are localized mainly in neuronal cytoplasm in AD brain. A, AD and control (Ctrl) hippocampal paraffin sections were subjected to double immunostaining with anti-histone 3 (a nuclear marker) and anti-I 2 PP2A and analyzed by laser scanning confocal microscopy. In AD brains, I 2 PP2A was localized mainly in neuronal cytoplasm, whereas in control cases it was seen in the nucleus. B, immunostaining with anti-LAMP-2 (a lysosomal marker) and anti-AEP showed a diffuse staining of AEP in neuronal cytoplasm in AD and lysosomal localization in control brain neurons. The insets show high magnification of the area indicated by the arrow. Scale bar, 40 m. nuclear fractions. The cytosolic fraction showed the highest concentration of the active AEP and was most effective in cleaving I 2 PP2A , generating I 2NTF (Fig. 4, D-F). Neither the lysosomal fraction nor the nuclear fraction showed any significant cleavage of I 2 PP2A (Fig. 4D). These data suggested that AEP is most likely the primary enzyme in the brain that cleaves I 2 PP2A at Asn-175, and this cleavage takes place mainly in the neuronal cytoplasm in AD.

I 2 PP2A Co-localizes with AEP in the Cytoplasm of Cos-7 Cells and Produces Both I 2NTF and I 2CTF under Acidic Conditions-
AEP is a lysosomal zymogen that undergoes autoproteolysis in acidic conditions during activation. We found that this enzyme is responsible for I 2 PP2A cleavage, but the generation of I 2NTF and I 2CTF fragments, both of which have been reported to inhibit (33), was not established. To confirm the generation of both I 2NTF and I 2CTF from the cleavage of I 2 PP2A by AEP, we incubated Cos-7 cells after 48 h of double transfection with HA-I 2 PP2A and AEP in acidic medium, pH 5.5, for 30 min, followed by immunocytochemistry with anti-HA and anti-AEP. In normal medium-treated cells (control), I 2 PP2A was located in the nucleus and co-localized with the nuclear marker TOPRO 3, and AEP staining had a vesicular appearance in the cytoplasm indicating lysosomal location. However, after incubation in acidic medium, both I 2 PP2A and AEP were localized diffused in the cytoplasm (Fig. 5A). To determine whether I 2NTF and I 2CTF were formed under acidic conditions, Cos-7 cells were either co-transfected or individually transfected with AEP and I 2 PP2A -Myc. I 2CTF was detected using anti-Myc, whereas I 2NTF was recognized by 10E7 antibody. We evaluated the levels of transfections and observed an increase in active AEP in AEP transfected cells, particularly under acidic conditions (Fig. 5B). I 2 PP2A was cleaved in the double-transfected cells generating two protein bands at 22 and 24 kDa (Fig. 5C). The band at 24 kDa, a larger N-terminal cleavage product than I 2NTF , was also present in the I 2 PP2A singly transfected cells, but this band and the 39-kDa band disappeared, whereas only 22-kDa I 2NTF was seen in double-transfected cells under acidic condition (Fig. 5C,  upper panel). The cleavage of I 2 PP2A by AEP in the double-transfected cells showed the appearance of a 17-kDa I 2CTF fragment under acidic conditions (Fig. 5C, lower panel). These data suggest that I 2 PP2A and AEP translocation to the cytoplasm requires acidic conditions to generate both I 2NTF and I 2CTF fragments. . Active AEP is selectively translocated from neuronal lysosomes to the cytoplasm and the nucleus and is associated with abnormally hyperphosphorylated Tau in AD. A, AD and control (Ctrl) human brain samples were subjected to subcellular fractionation, followed by Western blots to determine the distribution of AEP. B, the translocation of active AEP from the neuronal lysosomes to the cytoplasm was found. LAMP-2 was used as specific marker for lysosomal fraction. C, Western blot analysis showed a trend but no significant increase in translocation of active AEP from neuronal lysosomes to the nucleus in AD brain. D, immunohistochemical staining of AEP in AD hippocampal paraffin sections counterstained with hematoxylin showed its localization in neuronal lysosomes (arrowhead), cytoplasm (arrow), nuclei (asterisk), and neurofibrillary tangles (double arrows). E, AD and control hippocampal paraffin sections were subjected to triple immunohistofluorescent staining with anti-AEP, anti-I 2 PP2A , and anti-Ser(P)-262/356 (12E8) Tau and analyzed by laser scanning confocal microscopy. Diffuse cytosolic staining of AEP was observed in Tau pathology-bearing neurons (arrowheads), whereas in cells without hyperphosphorylated Tau AEP was localized in lysosomes (arrow). Scale bar, 10 m.

Treatment of Metabolically Active Rat Hippocampal Slices in
Acidic Conditions Leads to Activation of AEP, Cleavage of I 2 PP2A , Inhibition of PP2A, and Hyperphosphorylation of Tau-AEP activation has been previously reported in a stroke mouse model (18), which indicates that acidic conditions could play an important role for AEP activity, and we postulated that Tau hyperphosphorylation increases as a consequence of this activation. To assess this relation, we incubated metabolically active rat hippocampal slices in aCSF of different pH levels. After incubation, to evaluate AEP activation (36-kDa fragment) and Tau hyperphosphorylation, the tissue slices were lysed and used for Western blots and PP2A activity assays (Fig. 6). Densitometric analysis of the blots showed an increase in AEP activation (Fig. 6B) and, because of cleavage of I 2 PP2A , a reduction in the level of I 2FL at acidic pH (Fig. 6C). Corresponding to the cleavage of I 2 PP2A at acidic conditions, the PP2A activity was inhibited (Fig. 6D), and the abnormal hyperphosphorylation of Tau was increased at several sites (Fig. 6E). These changes increased with decreasing pH from pH 7.4 to 5.5. These data showed a clear pH dependence from neutral to acidic pH in activation of AEP and consequent cleavage of I 2 PP2A , a decrease in PP2A activity, and an increase in the abnormal hyperphosphorylation of Tau.
Abnormal Hyperphosphorylation of Tau Produced by Acidic Conditions Is Caused by the Activation of AEP-To learn whether the abnormal hyperphosphorylation of Tau produced during acidosis of the brain as in ischemia could be due to the activation of AEP, we treated the SH-SY5Y cells with siRNA to human AEP, followed by treatment with kainic acid or acidic pH. Transfection of SH-SY5Y cells with the siRNA successfully knocked down the expression of the active AEP; the level of the abnormal hyperphosphorylation of Tau at Ser-262/356 in cells treated with kainic acid was markedly increased as compared with the identically treated cells with siAEP or cells treated with neither siRNA nor kainic acid as a control (Fig. 6F). Furthermore, although pH 6.0 treatment increased the abnormal hyperphosphorylation of Tau at Ser-262/356, no such differences were seen between AEP knocked down SH-SY5Y cells lysed at pH 6.0 and the control treated cells lysed at the neutral pH (Fig. 6G).

DISCUSSION
AD is a multifactorial disease that involves several etiopathogenic mechanisms (43). Neurofibrillary pathology of hyperphosphorylated Tau is a histopathological hallmark of AD and related neurodegenerative disorders called tauopathies. Different factors are involved in the onset of this pathology, and the origin of the sporadic forms of these diseases remains unclear. However, a key mechanism that leads to neurofibrillary pathology is the abnormal hyperphosphorylation of Tau (39,45). In the present study, we evaluated a target that is upstream of Tau hyperphosphorylation in AD. We found that: (i) AEP, which is cleaved by AEP from the cytosolic fraction of AD brain. A, GST-I 2 PP2A recombinant fusion protein was cleaved by AD brain extract (Ext) in a proteolysis assay, generating GST-I 2NTF in a time-dependent manner. B, GST-I 2 PP2A (W) but not mutated N175Q (M) GST-I 2 PP2A could be cleaved with the brain extract. C, under the same proteolysis conditions GST-I 2 PP2A was cleaved by the IP AEP, but not by the immunodepleted extract (ID). D, the cytosolic fraction isolated from AD brains was markedly more robust than the lysosomal or the nuclear fraction in cleavage of GST-I 2 PP2A into GST-I 2NTF during 0 -90 min of incubation. E and F, quantitation of Western blots showed that the level of active AEP was the highest in the cytosolic fraction (E) and that GST-I 2 PP2A was most robustly cleaved by the cytosolic fraction (F). Actin, LAMP-2, and histone 3 were used as specific markers of cytoplasm, lysosomes, and nuclei, respectively. cleaves I 2 PP2A at Asn-175 into I 2NTF and I 2CTF , is selectively activated and translocated from the neuronal lysosomes to the cytoplasm and at a much smaller level to the nucleus in AD brain; (ii) AEP cleaves I 2 PP2A mainly in the neuronal cytoplasm; and (iii) acidic conditions such as caused by ischemia promote the activation and translocation of AEP as well as the cleavage and the translocation of I 2 PP2A to the neuronal cytoplasm and consequently inhibition of PP2A and the abnormal hyperphosphorylation of Tau.
AEP expression and activity have been linked to a number of pathological conditions including atherosclerosis, cancer, and inflammation (11,12). However, its biological role in these pathologies remains obscure. Here we found that active AEP is present in AD brain at a significantly higher level as compared with control brain. This finding is notable because this is the first study describing AEP involvement in AD pathology. The activation of AEP has been reported in experimental stroke and after transient middle cerebral artery occlusion in mice (18,19). Stroke elicits acidosis (47), and ischemic lesions (silent strokes) and acidic pH have been described in AD (2,4). Most lysosomal enzymes are synthesized as zymogens and can be autoactivated under acidic conditions, and it has been demonstrated that AEP requires acidic pH for its activation (9,48). In the present study, we found that active AEP is selectively translocated from the neuronal lysosomes to the cytoplasm in AD. Previous studies reported that AEP is synthesized in the endoplasmic reticulum as pro-AEP (56 kDa) and is routed via Golgi to the lysosomal system for processing. For this to happen, a pH-reduced environment is crucial (48). In AD brain, the pH is reduced (2). The present study shows that the acidic environment induces AEP activation. The presence of high level of pro-AEP found in control brains in the present study suggests its normal main storage in compartments other than lysosomes such as endoplasmic reticulum. An increase in the cleavage and translocation of FIGURE 5. I 2 PP2A is released from the cell nucleus to the cytoplasm, co-localized with AEP and cleaved into I 2NTF and I 2CTF under acidic conditions. A, Cos-7 cells co-transfected with AEP and HA-I 2 PP2A were subjected to immunocytochemical staining with anti-AEP and anti-HA, respectively. I 2 PP2A was seen mostly localized in the nucleus in Cos-7 cells but translocated to the cytoplasm following incubation with acid pH medium, pH 5.5, for 30 min. Under the acidic condition, AEP was seen diffused into cell cytoplasm and the nucleus. B, protein expressions evaluated by Western blots. Cos-7 cells were either co-transfected or individually transfected with AEP and I 2 PP2A -Myc and incubated with neutral pH or acid pH medium for 30 min. At acid pH, the active pro-AEP and active AEP signals were markedly increased. C, in the same conditions as in B, cells were lysed, followed by Western blots developed with anti-I 2NTF (10E7) or anti-I 2CTF (anti-c-Myc). In the presence of acidic conditions, cells co-transfected with AEP and I 2FL showed the generation of both I 2NTF and I 2CTF . NCP, N-terminal cleavage product of I 2 PP2A that neither requires AEP nor acidic pH; NSB, nonspecific band. I 2 PP2A in AD brain shown by us previously (32) is consistent with an increase in the level and translocation to neuronal cytoplasm of the activated AEP we found in the present study. In agreement with the present study, I 2 PP2A was previously reported to be selectively cleaved at Asn-175 in wild type, but not in AEP null mice (18).
A selective increase in the cleavage and translocation of I 2 PP2A from the neuronal nucleus to the cytoplasm in AD brain found in the present study is in agreement with similar findings reported by us previously (32). Although I 2 PP2A has been previously reported to translocate to cytoplasm under steady-state conditions in an apparently random fashion (27), during cell migration (26) and under oxidative stress, the marked translocation from the neuronal nucleus to the cytoplasm in AD brain observed in the present study is probably mainly due to its cleavage into I 2NTF and I 2CTF ; the cleavage products I 2NTF , a 175-amino acid-long fragment, and I 2CTF , a 102-amino acidlong fragment, because of their small sizes, can easily diffuse between the nucleus and the cytoplasm.
The immunohistochemical staining showed that although AEP was primarily co-localized with LAMP-2 in the lysosomal compartment in control brain neurons, it was found diffused in the neuronal cytoplasm in AD brains. Consistent with these findings, pro-AEP is known to be located in different compartments such as endoplasmic reticulum, Golgi, or endosomes (9) before the activation in the lysosomes (48). Furthermore, in kidney, AEP was previously shown to be co-localized with LAMP-2, suggesting its association with late endosomes (17).
We confirmed the translocation of AEP to the neuronal cytoplasm in AD brain by subcellular fractionation. We found that active AEP was located in the lysosomes as well as the cytosol, indicating that in AD, the pathological conditions are favorable in inducing translocation of AEP to the cell cytoplasm. The lysosomal system has been implicated in AD pathogenesis in which multiple alterations in neurons of AD patients were found (49). Lysosomal membrane permeability was previously reported in AD and was speculated to be induced by different factors such as the presence of A␤ (50) or oxidative stress (51). Thus, it is possible that in AD the damage to the lysosomes induces lysosomal membrane permeability, and thereby active AEP is released into the cytosol, where I 2 PP2A is also located and could be cleaved by this endopeptidase. In the present study, we observed that neurons with AEP punctate staining did not show Tau hyperphosphorylation in AD brain. However, when AEP was found diffused in the cytoplasm, Tau hyperphosphorylation was present, suggesting that the location of cytoplasmic AEP could play an important role in Tau hyperphosphorylation.
The possible mechanism involved is I 2 PP2A cleavage by activated AEP, followed by PP2A inhibition by I 2NTF and I 2CTF and Tau hyperphosphorylation (Fig. 7). This proposed mechanism is supported by the findings that: (i) AEP cleaved I 2 PP2A into I 2NTF and I 2CTF; (ii) AEP in the brain caused this cleavage; (iii) the brain level of activated AEP was increased in AD; (iv) AEP was mainly in the cytosolic fraction; and (v) neurons in which AEP was found diffused in the cytoplasm showed hyperphosphorylated Tau, whereas in neurons in which AEP was mainly localized in lysosomes, Tau was not hyperphosphorylated. The knockdown of AEP with siRNA, followed by acidosis of the SH-SY5Y cells by treatment with kainic acid or acid pH medium, confirmed the role of AEP and the molecular pathway by which acidosis of the brain could lead to Tau pathology.
In mammalian brain PP2A and PP1 account for ϳ70 and 10% of the total Tau phosphatase activity, respectively (21). Although I 2 PP2A inhibits PP2A (20,52), it activates PP1 (53). Thus, the abnormal hyperphosphorylation of Tau observed in rat hippocampal slices at acidic pH in the present study is most likely primarily due to the inhibition of PP2A caused by activation of AEP and the consequent cleavage of I 2 PP2A into I 2NTF and I 2CTF . The attenuation of the abnormal hyperphosphorylation of Tau by siAEP and at acid pH or kainic acid treatment in SY5Y cells is most likely through inhibition of activation of AEP and consequent reduction of I 2 PP2A cleavage and of inhibition of PP2A activity; kainic acid is not a direct PP2A inhibitor. The abnormal hyperphosphorylation of Tau is known to lead to its aggregation into paired helical filaments (45), and our previous studies (34,35) showed that the expression of I 2NTF and I 2CTF causes abnormal hyperphosphorylation and aggregation of Tau, as well as neurodegeneration and cognitive impairment in rats.
It is known that lysosomal hydrolase activation requires an optimal acidic pH, but it is not known how AEP is able to function in the cytosol. However, in vitro studies have shown that some lysosomal proteases can function at neutral pH (54,55), confirming their potential for activation outside the lysosomes. Acidic conditions in AD brains increase the lysosomal protease activity and lysosomal membrane permeability (51,55). Lysosomal enzyme activity in the cytosol has been previously described, such as cytosolic cathepsins in both pathological (56,57) and physiological (44) apoptotic mechanisms. AEP is predominantly localized in lysosomes but is overexpressed in many types of tumors and associated with enhanced tissue invasion and metastasis. In these cases, AEP can also be found on the cell surface, co-localized with integrin ␤1 (11), and in blastocysts displaying a pro-survival role (14), indicating that lysosomal location of AEP is not required for its activity. We observed in cultured cells under acidic conditions that both I 2 PP2A and AEP translocated to the cytosol. However, the cell morphology and nuclear condensation indicate that these cells undergo patho- FIGURE 7. Proposed mechanism of I 2 PP2A cleavage and abnormal hyperphosphorylation of Tau by activated AEP in AD. In AD different etiopathogenic mechanisms activate the translocation of active AEP from the neuronal lysosomes to the cytoplasm and the nucleus where it interacts with I 2 PP2A and cleaves it into I 2NTF and I 2CTF , which bind to PP2A and inhibit its activity. The inhibition of PP2A leads to abnormal hyperphosphorylation of Tau. logical states, which could be due to impairment in lysosomal stability caused by acidic conditions in cell culture, as previously reported (46).
It is known that PP2A plays an important role in AD (24). The acidosis associated with ischemic changes could cause PP2A inhibition through activation of AEP and consequent cleavage and translocation of I 2 PP2A from the neuronal nucleus to the cytoplasm in AD. AEP could be considered an important trigger and hence a therapeutic target for sporadic AD and related conditions characterized by Tau pathology.