Delta-Secretase Phosphorylation by SRPK2 Enhances Its Enzymatic Activity, Provoking Pathogenesis in Alzheimer s Disease

Transcription

1 Article Delta-Secretase Phosphorylation by SRPK2 Enhances Its Enzymatic Activity, Provoking Pathogenesis in Alzheimer s Disease Graphical Abstract Authors Zhi-Hao Wang, Pai Liu, Xia Liu,..., Shan Ping Yu, Jian-Zhi Wang, Keqiang Ye Correspondence wangjz@mails.tjmu.edu.cn (J.-Z.W.), kye@emory.edu (K.Y.) In Brief Wang et al. show that delta-secretase phosphorylation by SRPK2 promotes its cytoplasmic translocation and enzymatic activities. The unphosphorylated mutant rescues AD-like pathologies, implicating its potential therapeutic role in AD. Highlights d SRPK2 phosphorylates serine 226 on delta-secretase in AD brains d d d Delta-secretase phosphorylation escalates its enzymatic activities Phosphorylated delta-secretase provokes AD-like pathology Unphosphorylated delta-secretase restores cognitive functions Wang et al., 217, Molecular Cell 67, September 7, 217 ª 217 Elsevier Inc.

2 Molecular Cell Article Delta-Secretase Phosphorylation by SRPK2 Enhances Its Enzymatic Activity, Provoking Pathogenesis in Alzheimer s Disease Zhi-Hao Wang, 1,5 Pai Liu, 4 Xia Liu, 1 Fredric P. Manfredsson, 2 Ivette M. Sandoval, 2 Shan Ping Yu, 3 Jian-Zhi Wang, 5,6, and Keqiang Ye 1,7,8, 1 Department of Pathology and Laboratory Medicine, Emory University School of Medicine, Atlanta, GA 3322, USA 2 Department of Translational Science and Molecular Medicine, Michigan State University, 333 Bostwick Avenue NE, Grand Rapids, MI 493, USA 3 Department of Anesthesiology, Emory University School of Medicine, Atlanta, GA 3322, USA 4 Emory College of Arts and Sciences, Emory University, Atlanta, GA 3322, USA 5 Department of Pathophysiology, Key Laboratory of Ministry of Education of Neurological Diseases, Tongji Medical College, Huazhong University of Science and Technology, Wuhan, China 6 Co-innovation Center of Neuroregeneration, Nantong 2261, China 7 Translational Center for Stem Cell Research, Tongji Hospital, Tongji University School of Medicine, Shanghai 265, China 8 Lead Contact Correspondence: wangjz@mails.tjmu.edu.cn (J.-Z.W.), kye@emory.edu (K.Y.) SUMMARY Delta-secretase, a lysosomal asparagine endopeptidase (AEP), simultaneously cleaves both APP and tau, controlling the onset of pathogenesis of Alzheimer s disease (AD). However, how this protease is post-translationally regulated remains unclear. Here we report that serine-arginine protein kinase 2 (SRPK2) phosphorylates delta-secretase and enhances its enzymatic activity. SRPK2 phosphorylates serine 226 on delta-secretase and accelerates its autocatalytic cleavage, leading to its cytoplasmic translocation and escalated enzymatic activities. Delta-secretase is highly phosphorylated in human AD brains, tightly correlated with SRPK2 activity. Overexpression of a phosphorylation mimetic (S226D) in young 3xTg mice strongly promotes APP and tau fragmentation and facilitates amyloid plaque deposits and neurofibrillary tangle (NFT) formation, resulting in cognitive impairment. Conversely, viral injection of the non-phosphorylatable mutant (S226A) into 5XFAD mice decreases APP and tau proteolytic cleavage, attenuates AD pathologies, and reverses cognitive defects. Our findings support that delta-secretase phosphorylation by SRPK2 plays a critical role in aggravating AD pathogenesis. INTRODUCTION Serine-arginine protein kinase 2 (SRPK2) belongs to a small protein kinase family that specifically phosphorylates serine residues of the serine-arginine (SR)-rich motif on substrates (Wang et al., 1998, 1999). It is a cell-cycle-regulated kinase that modulates the activity of the pre-mrna alternative splicing machinery (Ngo et al., 25; Zhong et al., 29; Zhou et al., 212). A cellcycle signal induces nuclear translocation of SRPK2 at the G2/M boundary (Ding et al., 26). Previously, we reported that Akt-mediated phosphorylation of SRPK2 at T492 controls the expression of cell-cycle regulators, thus coupling cell-cycle regulation and cell death machinery in the nervous system. Consequently, activation of SRPK2 in neurons leads to cell-cycle progression, DNA synthesis, and, eventually, apoptosis (Jang et al., 29). Moreover, we showed that SRPK2 is cleaved by caspase 3 to induce apoptosis (Hong et al., 211). Interestingly, SRPK2 can be activated by b-amyloid peptide (Ab), and activated SRPK2 phosphorylates tau on S214, suppressing taudependent microtubule polymerization and inhibiting axonal elongation in neurons. SRPK2 activity is augmented in the neurons of Alzheimer s disease (AD) mice and patients (Hong et al., 212), strongly implicating a role for SRPK2 in AD etiology and pathogenesis. Our previous work also showed that SRPK2 phosphorylates SC35, an SR splicing factor located in nuclear speckles (Jang et al., 29), and it has recently been reported that dysregulated RNA processing with accumulation of unspliced RNA species occurred in human AD and MCI (mild cognitive impairment) patients (Bai et al., 213). AD is characterized by the accumulation of amyloid beta (Ab) within the brain, along with hyperphosphorylated and cleaved forms of the microtubule-associated protein tau. The physiopathology of AD has not yet been totally established. Nevertheless, it is known that a dysfunction in the metabolism of the Ab precursor protein (APP) leads to the formation of neuritic plaques, and abnormal tau phosphorylation results in neurofibrillary tangles (NFTs) (Mattson, 24). Mammalian asparaginyl endopeptidase (AEP, gene name LGMN) is a lysosomal cysteine protease that cleaves peptide bonds C-terminally to asparagine residues. AEP is synthesized as a zymogen (pro-aep, 56 kda) and is autocatalytically processed into active AEP under acidic conditions (Li et al., 23). We recently reported that AEP cleaves both 812 Molecular Cell 67, , September 7, 217 ª 217 Elsevier Inc.

3 APP and tau in human AD brains. Interestingly, AEP expression level and activity are escalated in aged mice and the AD brain (Zhang et al., 214, 215). These findings indicate that AEP might act as a novel delta-secretase in AD progression. AEP (referred to as delta-secretase hereafter) cuts APP at both the N3 and N585 residues in the ectodomain and facilitates Ab production by decreasing the steric hindrance for BACE1. Depletion of delta-secretase significantly reduces Ab production and senile plaque formation in the 5XFAD mouse brain, leading to restoration of synaptic activity and cognitive function (Zhang et al., 215). Additionally, delta-secretase also cleaves tau at N255 and N368 and abolishes its microtubule assembly activity, resulting in its aggregation and NFT formation. Moreover, the cleaved tau fragment is neurotoxic. Deletion of delta-secretase from tau P31 mice reverses the synaptic plasticity defect and cognitive dysfunction (Zhang et al., 214). Accordingly, inhibition of delta-secretase using a small-molecule inhibitor potently decreases amyloid plaque deposits and NFT in both the 5XFAD and tau P31S mouse models, improving cognitive function (Zhang et al., 216, 217). However, how this critical protease is regulated post-translationally during aging and AD pathogenesis remains elusive. In the current report, we show that SRPK2 selectively phosphorylates delta-secretase on serine 226 and facilitates its activation, elevating its enzymatic activity in human AD brains. Expression of a non-phosphorylatable mutant (S226A) in 5XFAD mice attenuates the AD pathology, rescuing cognitive function. Conversely, expression of the phosphorylation mimetic S226D in young triple-transgenic (3xTg) mice promotes an earlier onset of senile plaque deposits and NFT formation, accelerating cognitive dysfunction. RESULTS SRPK2 Phosphorylates Serine 226 of Delta-Secretase Both SRPK2 and delta-secretase are activated in an age-dependent manner in AD mouse models and in human AD brains, triggering deleterious effects in this neurodegenerative disease. Moreover, both proteins regulate tau pathological functions via either phosphorylation or cleavage (Hong et al., 212; Zhang et al., 214, 215). Accordingly, we hypothesize that these two pathways are linked in that SRPK2 might regulate delta-secretase via phosphorylation. To test this hypothesis, we conducted an in vitro kinase assay with purified His-tagged AEP recombinant protein in the presence of g- 32 P-[ATP]. We observed that His-AEP recombinant protein was strongly phosphorylated by wild-type () SRPK2, but not kinase-dead (KD) SRPK2 (Figure 1A). As a positive control, we utilized glutathione S-transferase (GST)-tau recombinant protein. To further explore whether SRPK2 phosphorylates delta-secretase in intact cells, we co-transfected mammalian GST-tagged (mgst) delta-secretase with or KD SRPK2 into HEK293 cells. Transfected delta-secretase was pulled down and examined by immunoblotting with anti-p-s/t antibody. We found that GST-delta-secretase associated with SRPK2 and that both proteins in this complex were phosphorylated (Figure 1B). Delta-secretase activation requires sequential removal of its N- and C-terminal peptides at residues D25 and N323 (Li et al., 23). Accordingly, we assessed the extent of phosphorylation of the mature version of delta-secretase by SRPK2. We found that the delta-secretase amino acids (aa) were phosphorylated to a higher degree than full-length delta-secretase (Figure 1C), indicating that the phosphorylation sites are located within this mature domain. Because SRPK2 predominantly phosphorylates the substrate at an SR dipeptide (Wang et al., 1998), we identified two domains (aa and aa ; Figure 1D, top) within delta-secretase. To test the ability of SRPK2 to phosphorylate these motifs, we generated point mutants (S226A, S328A) and a dual mutant (S226A/S328A) and performed an in vitro kinase assay. We found that both the S226A and S328A mutations evidently reduced delta-secretase phosphorylation and that removing both motifs largely eliminated phosphorylation (Figure 1D, center). Moreover, a GST pull-down assay in intact cells demonstrated that S226A mutation completely abolished delta-secretase phosphorylation by SRPK2, whereas the S328A mutation did so to a much lesser extent. This suggests that S226 might be the major SRPK2 phosphorylation residue on delta-secretase (Figure 1E, first panel). To determine whether delta-secretase is a physiological substrate of SRPK2, we generated a phospho- S226-specific antibody. Indeed, delta-secretase S226 phosphorylation tightly correlated with our anti-phospho-s/t antibody results (Figure 1E, second panel). To validate the specificity of anti-p-s226 antibody, we co-transfected or the non-phosphorylatable S226A mutant into HEK293 cells with or KD SRPK2. delta-secretase was selectively phosphorylated by, but not KD SRPK2, which was visualized by anti-p- S226 antibody. Importantly, p-delta-secretase signals were completely abolished by co-incubation with the phosphopeptide antigen, demonstrating the specificity of detection of this phosphorylation (Figure S1A). As expected, the p-s226 antibody specifically recognized phosphorylated delta-secretase in control, but not in delta-secretase-depleted human SY5Y cells. Noticeably, the truncated mature form was more strongly phosphorylated than full-length delta-secretase (Figure S1B). Thus, these results suggest that S226 is a major SRPK2 phosphorylation target on delta-secretase. SRPK2 Phosphorylates Delta-Secretase in Human AD Brains Our previous study shows that SRPK2 is activated and phosphorylated in AD brains (Hong et al., 212). To assess whether Ab activates SRPK2 to phosphorylate AEP in primary neurons, we performed immunofluorescence (IF) staining to find that Ab induced SRPK2 T492 and delta-secretase S226 phosphorylation in a dose-dependent way (Figures 2A and 2B), and Ab treatment triggered the subcellular co-localization of these two proteins as well (Figures 2C and 2D). To test whether SRPK2 is indeed implicated in phosphorylating delta-secretase in neurons, we knocked down SRPK2 with lentivirus-mediated expression of short hairpin RNA (shrna), followed by Ab treatment overnight. We found that Ab triggered SRPK2 T492 phosphorylation, which is a marker for its activation (Jang et al., 29). Knockdown of SRPK2 eliminated T492 phosphorylation by Ab. As expected, incubation with Ab also resulted in delta-secretase fragmentation and S226 phosphorylation, which was inhibited by SRPK2 knockdown (Figure 2E, first through fourth panels). As a result, we saw an increase in the cleavage of downstream Molecular Cell 67, , September 7,

4 A 1 C 1 D 1 Control GST-SRPK2 GST-SRPK2 KD Control GST-SRPK2 GST-SRPK2 KD Autoradiography Coomassie blue GFP-AEP FL KD KD paep IgG IP: anti-gfp WB: anti-phospho-serine/threonine IP: anti-gfp WB: anti-gfp Cell lysate WB: anti-srpk2 AEP FL ptau (GST-Tau) paep (His-AEP) GST-SRPK2 GST-Tau His-AEP (Myc-SRPK2 ) AEP IgG 181 kmvfyieace sgsmmnhlpd ninvyattaa npressyacy ydekrstylg 31 hldltpspdv pltimkrklm ntndleesrq lteeiqrhld arhlieksvr S226A S328A S226AS328A paep (GST-AEP) Autoradiography 1 AEP GST-AEP Coomassie blue Tau Phosphorylation ratio (% of AEP FL) Autophosphorylation (GST-SRPK2) GST-SRPK2 GST-AEP FL Phosphorylation-to-protein level (% of Tau) 1 1 delta-secretase targets, including APP and tau (APP N585 and tau N368). The extent of cleavage was mitigated by SRPK2 knockdown (Figure 2E, fifth through seventh panels). This suggests that Ab triggers SRPK2 activation in primary neurons, which subsequently phosphorylates delta-secretase, inducing its proteolytic activity. We virtually made the same observations in human SY5Y cells (Figure S2A). Next, we extended our studies to human AD brains, where we found a high level of S226 phosphorylation of delta-secretase (associated with its active fragments). In contrast, delta-secretase was barely cleaved or phosphorylated in healthy control brains. Delta-secretase S226 phosphorylation correlated well with SRPK2 phosphorylation as well as with the phosphorylation of SRPK2 substrates such as p-tau S214 and p-acinus S422 (Jang et al., 28; Figure 2F). IF staining with anti-phosphor-/total delta-secretase and SRPK2 1 1 AEP Tau E B KD 1 1 GST-AEP Control Myc-SRPK2 Myc-SRPK2 KD psrpk2 paep (GST-AEP) GST pull-down WB: anti-phospho-serine/threonine GST pull-down WB: anti-gst Cell lysate WB: anti-myc GST-AEP S226A S328A S226AS328A (Myc-SRPK2 ) GST pull-down WB: anti-phospho-serine/threonine GST pull-down WB: anti-gst Cell lysate WB: anti-myc KD GST pull-down WB: anti-aep ps226 KD KD psrpk2 (Myc-SRPK2) paep (GST-AEP) Figure 1. SRPK2 Phosphorylates Delta- Secretase In Vitro and In Vivo (A) In vitro kinase assay. His-AEP was incubated with immunoprecipitated GST-SRPK2 or GST- SRPK2 KD in the presence of [g 32 P]-ATP. The reaction products were separated in SDS-PAGE and analyzed by autoradiography. Purified GST-tau was used as a positive control. The protein inputs are shown (bottom left). The phosphorylation-toprotein level ratio is presented for AEP and tau (right). (B) SRPK2 phosphorylates delta-secretase in intact cells. mgst-delta-secretase was co-transfected with SRPK2 or KD, respectively, into HEK293 cells. After 48 hr, mgst-delta-secretase was pulled down and analyzed with anti-p-serine/ threonine antibody (top). Expression of pulleddown GST-AEP (center) and Myc-SRPK2 (bottom) in cell lysates is also shown. (C) Western blot showing phosphorylation of fulllength and mature delta-secretase by SRPK2 and KD in HEK293 cells. The phosphorylation ratio is presented (right). (D) S226 in delta-secretase is the major phosphorylation residue. Shown are schematics of the potential SRPK2 phosphorylation sites in delta-secretase (top). Also shown is an in vitro kinase assay showing phosphorylation of and mutated delta-secretase (S-to-A, unphosphorylated mutant) by SRPK2 (top blot). The inputs are also shown (bottom blot). (E) Western blot analysis of the phosphorylation level of and mutated delta-secretase after cotransfection with SRPK2 or KD. See also Figure S1. showed prominent co-localization between SRPK2 and delta-secretase and increased expression of SRPK2 pt492, delta-secretase, and delta-secretase ps226 in human AD hippocampcal CA1 compared with healthy controls (Figures S2B and S2C). Together, these results suggest that delta-secretase S226 is phosphorylated by SRPK2 in human AD brains. AEP (delta-secretase) is a lysosomal cysteine protease that is activated in the AD brain, where it is translocated from neuronal lysosomes to the cytoplasm (Basurto-Islas et al., 213). To determine whether S226 phosphorylation is involved in this process, we conducted subcellular fractionation. Interestingly, we found that mature and active delta-secretase in AD patients were more abundant than in healthy controls in both cytosolic and lysosomal fractions, and the active truncate was highly phosphorylated on S226 in AD (Figure 2G). To further test whether S226 phosphorylation dictates delta-secretase cytoplasm translocation from the lysosomes, we transfected HEK293 cells with GST-tagged, S226A, or S226D deltasecretase and performed co-staining with fluorescein isothiocyanate (FITC)-conjugated anti-gst antibody and LAMP1, a specific marker for lysosomes. Interestingly, we found that the 814 Molecular Cell 67, , September 7, 217

5 A Aβ 2 μm Control SRPK2 pt492 AEP DAPI Merge C Colocalization SRPK2 pt492-aep (Pearson s coefficient) Control Aβ 2 μm Aβ 2 μm D Colocalization SRPK2-AEP ps226 (Pearson s coefficient).4.2 Control Aβ 2 μm Aβ 2 μm Aβ 2 μm E Cortical neuron DIV 13 Ctrl shsrpk WB: anti-aep ps226 Aβ (μm ) B SRPK2 AEP ps226 DAPI Merge Control 1 1 WB: anti-aep WB: anti-srpk2 pt492 1 WB: anti-srpk2 Aβ 2 μm 1 WB: anti-app N585 WB: anti-tau N368 Aβ 2 μm F WB: anti-tubulin Human cortex Control AD G 1 Lysosomal Cytosolic Control AD Control AD WB: anti-aep ps226 WB: anti-aep WB: anti-lamp1 Human brain WB: anti-aep ps226 WB: anti-aep WB: anti-srpk2 pt492 WB: anti-srpk2 WB: anti-acinus ps422 WB: anti-tubulin WB: anti-tau ps214 WB: anti-tubulin Figure 2. Delta-Secretase Phosphorylation by SRPK2 in Neurons and AD Brains (A and B) Confocal pictures showing co-localization. Scale bars, 5 mm. (A) Co-localization of SRPK2 pt492 and delta-secretase in rat primary cortical neurons (day in vitro [DIV] 15) treated with 2 or 2 mm of pre-aggregated Ab for 16 hr. (legend continued on next page) Molecular Cell 67, , September 7,

6 phosphorylation mimetic mutant S226D was localized to the cytoplasm, whereas both and S226A predominantly colocalized with anti-lamp1, indicating that and non-phosphorylatable delta-secretase mainly reside in the lysosomes, whereas phosphorylated delta-secretase translocates into the cytoplasm (Figures 3A and 3C). IF staining with anti-p-s226 in primary neurons showed that the Ab-mediated augmentation of S226 phosphorylation (e.g., Figure 2B) correlated with increased cytosolic localization (Figures 3B and 3D). To further explore whether delta-secretase cytoplasmic translocation is associated with its p-s226 activity, we conducted subcellular fractionation with Ab peptide-treated SY5Y cells. Notably, Ab strongly escalated delta-secretase S226 phosphorylation and cytosolic translocation. Again, the mature form in the cytosolic fraction was more pronouncedly phosphorylated than those in the lysosomes (Figure 3E). To further interrogate where and when SRPK2 might phosphorylate AEP, we transfected SY5Y cells with small interfering RNA (sirna) against SRPK2, followed by Ab (1 mm) treatment at different time points. Subcellular fractionation showed that Ab treatment elicited SRPK2 to phosphorylate delta-secretase in the lysosomes; subsequently, the phosphorylated and truncated (mature form) delta-secretase translocated into the cytosol. Moreover, Ab incurred potent AEP S226 and SRPK2 T492 phosphorylation at 8 hr in both the lysosome and the cytoplasm fractions, and it was augmented in a time-dependent manner. Depletion of SRPK2 substantially abolished Ab-induced delta-secretase S226 phosphorylation (Figure 3F). Remarkably, the total protein levels of SRPK2 started to decrease at 24 hr in both fractions, indicating that SRPK2 might be degraded by Ab-activated proteases like caspases. Immunofluorescent co-staining also supported that p-s226/ SRPK2 co-localized with LAMP1 (Figures 3G 3I). Because SRPK2 is a cell-cycle-regulated kinase, we wondered whether delta-secretase phosphorylation-mediated translation is related to the cell cycle. Accordingly, we employed a well-characterized double thymidine synchronization method (Gui et al., 1994) to arrest the cell cycle in HeLa cells. Noticeably, both SRPK2 T492 and AEP-S226 phosphorylation increased as the cell cycle progressed. Delta-secretase translocation from the lysosome to the cytosol escalated from G to G1/S to G2 and M phases, which was repressed when SRPK2 was depleted. It is worth noting that the total protein levels of SRPK2 decreased as the cell cycle proceeded, indicating that SRPK2 may be degraded or proteolytically truncated (Figure S3A). Moreover, the unphosphorylated S226A mutant remained constant in the lysosome fraction during the cell cycle (Figure S3B). In alignment with these findings, Ab-triggered cytoplasmic translocation was abolished for the S226A mutant (Figures S3C and S3D). Clearly, blockade of delta-secretase phosphorylation on S226 reduces its cytosolic translocation, supporting that p-s226 leads to its cytoplasmic translocation. Together, these data argue that Ab stimulates SRPK2 to phosphorylate delta-secretase on S226, eliciting p-delta-secretase translocation from the lysosomes into the cytoplasm. Delta-Secretase S226 Phosphorylation Enhances Its Autocatalytic Process and Enzymatic Activity To assess the biological effect of S226 phosphorylation on deltasecretase enzymatic activity, we co-transfected mgst-srpk2 ( or KD) and Myc-delta-secretase into HEK293 cells and conducted a delta-secretase enzymatic assay. We found that SRPK2 overexpression increased delta-secretase enzymatic activity compared with the control. By contrast, expression of KD SRPK2 decreased delta-secretase enzymatic activity (Figure 4A). Next, we analyzed endogenous delta-secretase enzymatic activity in the presence of SRPK2. As expected, compared with the KD and vector control, SRPK2 expression also enhanced endogenous delta-secretase activity (Figure 4B). We also extended our studies to the delta-secretase S226D and S226A mutants and found that the phosphorylation mimetic S226D displayed higher enzymatic activity than the and that the S226A mutant possessed lower enzymatic activity compared with the (Figure 4C). Next, we measured deltasecretase autocatalytic fragmentation following different time points of incubation at C. Immunoblotting revealed that delta-secretase proteolytic processing rates were the same as its enzymatic efficacy, where S226D > > S226A (Figure 4D). To further test the effect of SRPK2 phosphorylation on delta-secretase proteolytic activity, we co-transfected Myc-SRPK2 ( or KD) with mgst-delta-secretase in HEK293 cells and incubated the cell lysates with GST-tau at different time points. Delta-secretase was phosphorylated by, but not KD SRPK2, fitting with its profile of enzymatic activity (Figure S4A, top right). Immunoblotting showed that phosphorylated delta-secretase cleaved tau in a time-dependent fashion, whereas non-phosphorylatable delta-secretase exhibited an extremely low degree of tau cleavage (Figure S4A, left). Quantitative analysis of delta-secretase phosphorylation is shown at the bottom right in Figure S4A. We made similar observations with tau proteolytic degradation by delta-secretase S226D and S226A, which aligned with these mutants enzymatic activities (Figure 4E). Our data support that p-s226 on delta-secretase facilitates its autoprocessing into the mature form. To directly test this notion in vitro, we purified mgst-aep and S226A mutant recombinant proteins, performed in vitro phosphorylation with isolated Myc-SRPK2 or KD in ph 7.4 buffer for 1 hr, and then we initiated autocleavage at ph 6. at different time points. We found that SRPK2 phosphorylation elicited time-dependent (B) Co-localization of SRPK2 and delta-secretase ps226 in rat primary cortical neurons (DIV 15) treated with 2 or 2 mm of pre-aggregated Ab for 16 hr. (C and D) Quantification of SRPK2 pt492 with delta-secretase (C) and SRPK2 with delta-secretase ps226 (D) co-localization. Data are shown as mean ± SEM (n = 2 28 cells/group, p <.5, Student s t test). (E) Knockdown of SRPK2 reduces delta-secretase phosphorylation on S226 induced by Ab. The western blot was conducted with primary cultures infected with a virus expressing control or SRPK2 shrna. (F) Western blot analysis of phosphorylated delta-secretase and SRPK2 in AD brains compared with age-matched healthy controls. (G) Phosphorylated AEP selectively translocated from lysosomes to the cytoplasm in AD brains. The western blot data in (E) (G) are representative of three independent experiments. See also Figure S Molecular Cell 67, , September 7, 217

7 A GST-AEP GST-AEP S226A GST-AEP S226D E 1 G AEP ps226 LAMP1 SRPK2 -biotin Merge GST LAMP1 DAPI Merge SH-SY5Y cells Lysosomal Cytosolic WB: anti-aep ps226 WB: anti-aep WB: anti-lamp1 WB: anti-tubulin HEK293 cells Aβ (μm) F Control Aβ 2 μm Aβ 2 μm AEP ps226 LAMP1 DAPI Merge SH-SY5Y cells Lysosomal Cytosolic Aβ Ctrl sisrpk2 Ctrl sisrpk2 (1 μm) hours WB: anti-aep ps226 WB: anti-aep WB: anti-srpk2 pt492 WB: anti-srpk2 WB: anti-lamp1 WB: anti-tubulin SH-SY5Y cells Ctrl sisrpk2 Aβ (1 μm) hours B Rat neuron H I C D hour) A EP ps226 (1% of of Intensity Colocalization AEP ps226-lam P1 Colocalization GST-LAMP1 (Pearson s coefficient) Colocalization AEP ps226-lamp1 (Pearson s coefficient) S226A S226D Control Aβ 2μM Aβ 2μM hr hr (legend on next page) Molecular Cell 67, , September 7,

8 AEP autocleavage, which was substantially reduced with the S226A or KD mutant. Remarkably, the slowest autoprocessing occurred in the S226A + KD group, indicating that AEP phosphorylation on S226 plays a key role in dictating AEP autocleavage and maturation (Figure S4B, top). Moreover, we analyzed the in vitrophosphorylated delta-secretase recombinant protein s enzymatic activity using well-characterized mgst-tau as a substrate. Again, we found that tau fragmentation displayed a similar cleavage pattern as delta-secretase autoprocessing (Figures S4B and S4C, bottom). As expected, employing the mature and truncated delta-secretase (aa ) fragment and its S226A mutant, we found stronger enzymatic activity with mature delta-secretase. Consequently, we monitored tau fragmentation with a shorter period of incubation (from full-length s 3 6 min to truncate s 15 3 min) and discovered an even more robust tau fragment (Figures S4D and S4E), supporting that in vitro phosphorylation of truncated delta-secretase enhances its protease activity. Together, these data strongly support that SRPK2 phosphorylates delta-secretase and triggers its autocatalytic activation. Delta-Secretase S226D Facilitates AD Pathologies in Young 3xTg Mice, Stimulating Cognitive Dysfunction A triple-transgenic mouse model (3xTg-AD) harboring PS1 (M146V), APP(Swe), and tau (P31L) transgenes has been developed to study the interaction between Ab and tau and their effect on synaptic functions. 3xTg-AD mice progressively develop amyloid plaques in the hippocampus at 6 months of age and NFTs at 12 months (Oddo et al., 23). To investigate whether expression of phosphorylated delta-secretase may facilitate the early onset of AD pathology, we injected the hippocampus of 2-month-old mice and age-matched 3xTg mice with a lentivirus expressing control delta-secretase or the S226D mutant (Figure S5A). Three months following vector delivery, the animals were sacrificed and analyzed for delta-secretase activity. Immunoblotting analysis revealed that animals overexpressing or S226D delta-secretase displayed a mature active fragment, and this cleavage was observed at a higher degree with S226D overexpression as opposed to the. Moreover, the level of active delta-secretase truncates was higher in 3xTg mice than in mice (Figure 5A, top). Accordingly, we also found a corresponding increase in tau N368 and APP N585 fragments with the pattern of S226D > > control, fitting with the levels of active delta-secretase (Figure 5A, second through fourth panels). An enzymatic assay showed the same pattern of activity, and 3xTg mice exhibited higher delta-secretase activity than mice (Figure 5B). Thioflavin S staining and immunohistochemistry (IHC) staining with an anti-ab showed that S226D overexpression resulted in more amyloid plaque deposits compared with treatment. Very few plaques were found in control vector-treated animals in the same hippocampal region (Figures 5C and S5B). Accordingly, Ab ELISA demonstrated that S226D overexpression produced significantly more Ab peptide than that of deltasecretase, followed by control virus in 3xTg mice. However, Ab levels remained comparable in mice regardless of deltasecretase or S226D expression (Figures S5C and S5D). Interestingly, AT8 and AT1 staining revealed that S226D overexpression induced tau phosphorylation in the hippocampus compared with and control-treated animals (Figure 5D), which normally only occurs at 12 months of age or later (Oddo et al., 23). Synaptic loss is believed to be the basis of cognitive impairment in the early phase of AD (Shankar and Walsh, 29). To assess the physiological effect of p-s226 on delta-secretase in this process, we performed electronic microscopy (EM) analysis, which demonstrated that the synapses were gradually decreased when or S226D delta-secretase was overexpressed in 3xTg mice. Notably, 3xTg/S226D mice possessed significantly fewer synapses than 3xTg/control or /S226D mice. In contrast, no change in synapse number was observed in mice, regardless of or S226D (Figure S5E). Dendritic spine density of hippocampal neurons corresponded to the EM analysis (Figure 5E). Electrophysiological measurements demonstrated that long-term potentiation (LTP) was pronouncedly impaired in 3xTg mice compared with mice. Again, S226D overexpression substantially reduced LTP in 3xTg compared with the control or delta-secretase groups (Figures 5F and S5F). Next, we performed a Morris water maze (MWM) behavioral test to evaluate learning and memory capability. During the training phase, 3xTg mice displayed a much longer latency to reach the platform than mice, and delta-secretase overexpression exacerbated this phenotype; 3xTg/S226D mice performed the worst, followed by 3xTg/ and 3xTg/control. A probe trail found that S226D overexpression significantly Figure 3. Phosphorylated Delta-Secretase on S226 Is Selectively Translocated from Neuronal Lysosomes to the Cytoplasm (A) Immunofluorescent staining. Shown is co-localization of GST and LAMP1 in HEK293 cells transfected with GST-tagged and mutated delta-secretase. Confocal pictures show co-localization between GST and LAMP1. Scale bars, 5 mm. (B) Co-localization of delta-secretase ps226 and LAMP1 in rat primary cortical neurons (DIV 15) treated with 2 or 2 mm of pre-aggregated Ab for 16 hr. Confocal pictures show co-localization of delta-secretase ps226 and LAMP1. Scale bars, 5 mm. (C and D) Quantification of GST and LAMP1 (C) and ps226 and LAMP1 (D) co-localization. Data are shown as mean ± SEM (n = cells/group, p <.5, Student s t test). (E) Phospho-S226 mediates delta-secretase cytosolic translocation. SH-SY5Y cells were treated with different concentrations of Ab and then subjected to subcellular fractionation followed by western blot. LAMP1 was used as a specific marker for the lysosomal fraction. Tubulin was applied as a loading control for the cytoplasmic fraction. (F) SRPK2-phosphorylated delta-secretase translocates into the cytoplasm from the lysosomes. SH-SY5Y cells were transfected with control sirna or si-srpk2, followed by Ab (1 mm) treatment at different time points. The lysosomal and cytosolic fractions were prepared and analyzed by immunoblotting. (G) Ab triggers delta-secretase S226 phosphorylation escalation and cytoplasm translocation. Shown is immunofluorescent staining of SH-SY5Y cells with antip-s226/srpk2/lamp1 after Ab treatment. Scale bar, 1 mm. (H and I) Quantification of AEP ps226 intensity (H) and quantification of AEP ps226 and LAMP1 co-localization (I) is shown as Pearson s coefficient. Data are shown as mean ± SEM (n = cells/group, p <.5, Student s t test). See also Figure S Molecular Cell 67, , September 7, 217

9 A AEP activity (AFU) C AEP activity (AFU) 1 Myc-AEP and Vector Myc-AEP and GST-SRPK2 KD Myc-AEP and GST-SRPK Time (min) GST-AEP GST-AEP S226A GST-AEP S226D Time (min) B AEP activity (AFU) Vector GST-SRPK2 KD GST-SRPK Time (min) WB: anti-gst D GST-AEP GST-AEP S226A GST-AEP S226D WB: anti-tubulin Incubation time (min) FL Figure 4. Delta-Secretase Phosphorylation on S226 by SRPK2 Increases Its Enzymatic Activity (A C) Delta-secretase activity assay of HEK293 cell lysates. HEK293 cells were co-transfected with delta-secretase and SRPK2 /KD (A) or transfected with SPRK2 /KD (B) or and mutated delta-secretase (C). Data are shown as mean ± SEM of three independent experiments. (D and E) Western blot analysis of autocatalytic process of /mutated delta-secretase (D) and cleavage of tau by /mutated delta-secretase (E) after incubation in ph 5.5 reaction buffer at C. Truncated delta-secretase or tau bands were quantified using densitometry. After being normalized with total delta-secretase or tau levels, the relative levels were plotted against reaction times. Each curve (D, bottom; E, right) was fitted with GraphPad Prism 5 using the Boltzmann sigmoidal function. Data are representatives of four independent experiments. See also Figure S4. E Incubation time (min) FL GST-Tau cotransfected with AEP WB: anti-tau5 FL GST-Tau cotransfected with AEP S226A WB: anti-tau5 FL GST-Tau cotransfected with AEP S226D WB: anti-tau impaired memory in 3xTg mice compared with delta-secretase- or control virus-treated mice (Figures 5G, 5H, and S5G). The swim speed was similar among the groups, indicating that delta-secretase expression does not affect motor function (Figure S5H). A cued fear conditioning assay showed that 3xTg/ S226D mice exhibited less freezing compared with 3XTg/control and 3xTg/ (Figure S5I). Together, these results strongly support that SRPK2 phosphorylation of delta-secretase stimulates its activation and enzymatic activity, leading to early onset of AD pathology and acceleration of cognitive dysfunction in young 3xTg AD mice. Delta-Secretase S226A Reduces AD Pathologies in 5XFAD Mice, Alleviating Cognitive Impairment Delta-secretase activity is elevated in the 5XFAD mouse model and in human AD brains (Basurto-Islas et al., 213; Zhang et al., 215). To examine the pathological role of S226 phosphor- Relative level of AEP cleavge (%) Relative level of Tau cleavge (%) Ti me (mi n) Ti me (mi n) GST-AEP GST-AEP S226A GST-AEP S226D AEP AEP S226A AEP S226D ylation in AD pathogenesis, we injected a lentivirus expressing delta-secretase or the S226A mutant into the hippocampus of 2-month-old mice (Figure S6A). Three months after the viral injection, we measured delta-secretase expression and activity by immunoblotting. In 5XFAD mice, overexpression of delta-secretase resulted in a higher level of formation of the active mature form of delta-secretase compared with S226A overexpression. Consistent with these findings, the levels of both tau N368 and APP N585 fragments were increased in delta-secretase-overexpressing animals compared with controls. In contrast, S226A overexpression reduced both tau and APP proteolytic cleavage by deltasecretase. On the other hand, overexpression of delta-secretase, but not S226A, in mice resulted in the formation of active mature delta-secretase, but to a lesser extent than that seen in 5XFAD mice. Consequently, tau N368 and APP N585 were increased in delta-secretase-treated animals compared with control and S226A-expressing mice (Figure 6A). Enzymatic activity was significantly increased in delta-secretase-overexpressing brains, and no increase in enzymatic activity was seen with S226A expression. In line with the delta-secretase cleavage, enzymatic activity was higher in 5XFAD mice than in the (Figure 6B). Ab concentrations tightly correlate with APP N585 levels in the hippocampus of 5XFAD mice. However, no significant increase in Ab peptides was seen in mice (Figures S6B and S6C). Thioflavin S staining and IHC staining with anti-ab showed that amyloid plaque deposits were enhanced in delta-secretase-injected hippocampi compared with control samples. S226A mutant overexpression resulted in decreased amyloid plaque formation Molecular Cell 67, , September 7,

10 A hippocampus Ct AEP AEP S226D WB: anti-aep WB: anti-tau N368 1 WB: anti-app N585 C D WB: anti-tubulin Thioflavin S staining 3xTg hippocampus Ct AEP AEP S226D Human Mouse 3xTg-Ct 3xTg-AEP 3xTg-AEP S226D 3xTg- Ct 3xTg- AEP 3xTg- AEP S226D B E AEP activity (AFU) Average number of plaques in hippocampus Ct -AEP -AEP S226D 3xTg-Ct 3xTg-AEP 3xTg-AEP S226D 3xTg-Ct 3xTg-AEP 3xTg-AEP S226D Thioflavin S+ average pixel intensity/ mm IHC: anti-at8 IHC: anti-at1 AT8 immunoreactivity (% of control) Ct -AEP -AEP S226D AT1 immunoreactivity (% of control) xTg-Ct 3xTg-AEP 3xTg-AEP S226D 3xTg-Ct 3xTg-AEP 3xTg-AEP S226D # Pulse interval (1ms) Time (min) Latency to mount platform (s) Ct -AEP AEP S226D -Ct -AEP -AEP S226D 3xTg-Ct 3xTg-AEP 3xTg-AEP S226D 5 -Ct F G H Paired pulse ratio (%) fepsp slope (% of baseline) Spine number (/1μm) -AEP -AEP S226D 3xTg-Ct 3xTg-AEP 3xTg-AEP S226D Training day 3xTg-Ct Primary latency (s) 3xTg-AEP xTg-AEP S226D -Ct -AEP -AEP S226D 3xTg-Ct 3xTg-AEP 3xTg-AE P S226D Figure 5. Delta-Secretase S226D in Young 3xTg Mouse Hippocampal CA1 Accelerates the Onset of AD Pathogenesis and Worsens Cognitive Dysfunction (A) Delta-secretase S226D escalates delta-secretase activity and induces APP and tau cleavage. Hippocampal CA1 tissues from lentivirus-injected (control, delta-secretase, or S226D) 3xTg mice were analyzed by western blot (n = 4 mice/group). The molecular weight of non-tagged exogenous delta-secretase is similar to endogenous delta-secretase. (legend continued on next page) 82 Molecular Cell 67, , September 7, 217

11 in 5XFAD mice (Figures 6C and 6D). EM analysis revealed that delta-secretase overexpression substantially reduced synapse numbers in the hippocampus compared with the control. In contrast, S226A mutant overexpression increased the number of synapses (Figure S6D). These observations were confirmed via quantification of dendritic spines, where we found that S226A mutant expression greatly increased spine density compared with the control, whereas overexpression of delta-secretase resulted in decreased spine density (Figure 6E). 5XFAD mice exhibit impaired LTP at Schaffer collateral-ca1 pathways (Kimura and Ohno, 29). Delta-secretase overexpression reduced LTP compared with the control, whereas S226A expression rescued LTP. The paired-pulse ratios were lower in 5XFAD/ compared with 5XFAD/control, and 5XFAD/S226A subjects reversed the deficit, suggesting that inactivation of delta-secretase rescues the synaptic dysfunctions (Figure 6F). Next, we evaluated spatial memory in 5-month-old 5XFAD, 5XFAD/AEP, and 5XFAD/S226A mice using the MWM. Compared with the control, overexpression of delta-secretase significantly increased the latency to platform, whereas S226A reduced it (Figure 6G). In the probe trial, the S226A group spent more time in the target quadrant than the control group, indicating better memory. By contrast, the AEP expression group spent less time in the quadrant (Figure 6H). All groups of mice exhibited a comparable swim speed (Figure S6E). Similarly, the cued fear conditioning test also demonstrated that S226A expression can rescue the cognitive defects in 5XFAD mice (Figure 6I). Together, these data support that expression of a non-phosphorylatable delta-secretase mutant ameliorates AD pathology in 5XFAD mice, rescuing the cognitive dysfunction associated with this model. DISCUSSION In the present study, we show that SRPK2 phosphorylates deltasecretase on residue S226 and accelerates its autocatalytic process, resulting in enhanced delta-secretase enzymatic activity. S226 phosphorylation enables delta-secretase to proteolytically cleave its substrates more efficiently, which, in turn, facilitates AD pathogenesis, including amyloid plaque deposits and NFT formation. As a result, this phosphorylation substantially exacerbates synaptic loss, which leads to LTP impairment and cognitive dysfunction. On the other hand, inhibition of this phosphorylation by expression of the non-phosphorylatable mutant S226A reverses these biological and pathological actions, rescuing the cognitive deficits in AD mouse models. It is worth noting that overexpressing delta-secretase or the phosphomimetic S226D induces both tau and APP proteolytic cleavage in mice; however, we failed to observe any prominent pathological or electrophysiological effects in these mice. Presumably, the truncated tau N368 and APP N585 produced by delta-secretase are efficiently cleared from brains, whereas there are many more abundant APP and tau substrates in 3xTg mice. Thus, the production rate of APP N585 (and subsequent Ab peptides) and tau N368 is higher than the rate of clearance, leading to senile plaques and NFT accumulation in 3xTg mice. Accumulating evidence demonstrates that delta-secretase (AEP) plays a critical role in a variety of human diseases, including cancers and neurodegenerative diseases (Basurto- Islas et al., 213; Lin et al., 214; Liu et al., 23; Ohno et al., 213; Zhang et al., 216). However, how this seemingly deleterious protease is post-translationally regulated remains incompletely understood. Previous studies reveal delta-secretase modifications, like glycosylation (Chen et al., 1997) and ubiquitination by TRAF6 (Lin et al., 214). This is, however, the first study investigating the role of delta-secretase phosphorylation in protein function and disease. Delta-secretase is synthesized as a zymogen (pro-aep, 56 kda) and autocatalytically processed into active delta-secretase under acidic conditions. Processing of delta-secretase 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 (Halfon et al., 1998; Li et al., 23). It remains unknown whether glycosylation or ubiquitination plays any role in this activation process. However, in the current study, we provide extensive evidence demonstrating that Ab-activated SRPK2 selectively phosphorylates delta-secretase on S226, within an SR dipeptide motif, and accelerates its autocatalytic process. Importantly, we found that delta-secretase is highly phosphorylated in human AD brains, tightly coupled to the activation of its upstream kinase SRPK2 (Figure 2). Moreover, because deltasecretase is greatly upregulated in numerous human cancers (B) Delta-secretase S226D increases its activity. Data represent mean ± SEM of 3 mice/group (p <.5, one-way ANOVA). (C) Delta-secretase S226D stimulates the early formation of amyloid plaques in 3xTg mouse CA1. Shown is thioflavin S staining detection of senile plaques in CA1 (left; scale bar, mm) and quantitative analysis of plaque number and size (right). Plaque analysis data represent mean ± SEM of sections from 3 mice in each group (p <.5, one-way ANOVA). (D) Tau pathology is promoted by delta-secretase S226D. Shown is IHC staining with anti-p-tau antibodies. The brain sections were immunostained with AT8 (top) and AT1 (center); scale bars, mm. Also shown is AT8 and AT1 immuno-reactivity quantification (bottom, mean ± SEM, sections from 3 mice, p <.5, one-way ANOVA). (E) Delta-secretase S226D decreases dendritic spine density. Top: Golgi staining was conducted on brain sections from the control lentivirus (LV-Ct)/deltasecretase /AEP S226D-treated apical dendritic layer of the CA1 region. Scale bar, 5 mm. Bottom: quantification of spine density represents mean ± SEM of 9 12 sections from 3 mice in each group. (p <.5, one-way ANOVA). (F) Electrophysiology analysis. Delta-secretase S226D expression in CA1 worsened the LTP defects in 3xTg mice. Shown is the ratio of paired pulses in different groups (mean ± SEM, n = 6 in each group, p <.5 compared with 3xTg-Ct, one-way ANOVA) (left). Also shown is the LTP of field excitatory postsynaptic potentials (fepsps) (mean ± SEM, n = 6 in each group, p <.5 3xTg-S226D versus 3xTg-Ct, # p <.5 3xTg-S226D versus 3xTg-AEP, one-way ANOVA) (right). (G and H) MWM analysis of cognitive functions. Delta-secretase S226D expression in the CA1 exacerbated the learning (G, training) and memory (H, probe trial) dysfunctions in 3xTg mice (mean ± SEM, n = 8 1 mice/group, p <.5, one-way ANOVA). See also Figure S5. Molecular Cell 67, , September 7,

12 1 mv A C D E 1 1 IHC:anti-Aβ hippocampus Ct AEP WB: anti-aep WB: anti-tau N368 WB: anti-app N585 (long-time exposure) WB: anti-tubulin AEP S226A 5XFAD hippocampus Ct AEP WB: anti-app N585 (short-time exposure) AEP S226A 5XFAD hippocampus Ct AEP AEP S226A Thioflavin S staining 5XFAD hippocampus Ct AEP AEP S226A 5XFAD CA1 Ct S226A Spine number (/1μm) 1 5 Ct S226A F Paired pulse ratio (%) 2 1 Plaques (% of control) Aβ immunoreactivity (% of control) Pulse interval (1ms) 5XFAD-Ct 5XFAD -AEP 5XFAD -AEP S226A # B AEP activity (AFU) XFAD-Ct 5XFAD -AEP 5XFAD -AEP S226A Count Average size fepsp slope (% of baseline) 1 -Ct -AEP -AEP S226A 5XFAD-Ct 5XFAD -AEP 5XFAD -AEP S226A Total area 5XFAD-Ct 5XFAD -AEP 5XFAD -AEP S226A Ct AEP AEP S226A Time (min) 2 # 1 ms G Latency to mount platform (s) XFAD-Ct 5XFAD -AEP 5XFAD -AEP S226A Training day H % of time spent in platform quadrant 5XFAD-Ct 5XFAD -AEP 5XFAD -AEP S226A I Freezing (%) XFAD-Ct 5XFAD -AEP 5XFAD -AEP S226A Figure 6. Delta-Secretase S226A Ameliorates the Pathologies and Behavioral Defects in 5XFAD Mice (A) The S226A mutation reduces delta-secretase and APP and tau fragments cleaved by delta-secretase. Hippocampal CA1 tissues from lentivirus-injected (control,, or delta-secretase S226A) 5XFAD mice were analyzed by immunoblotting with various antibodies (n = 4 mice/group). The molecular weight of nontagged exogenous delta-secretase is similar to endogenous delta-secretase. (legend continued on next page) 822 Molecular Cell 67, , September 7, 217

13 (Gawenda et al., 27; Li et al., 23; Zhu et al., 216), one may postulate that active delta-secretase may also be robustly phosphorylated on S226 by SRPK2, mediating cancer cell-cycle progression, cell proliferation, and invasion (Andrade et al., 211). Terminally differentiated neurons are unable to re-enter the cell cycle, but accumulating evidence has demonstrated the upregulation of cell-cycle regulatory proteins in degenerating neurons of AD brains (Husseman et al., 2). Moreover, SRPK2 is regulated by the cell cycle. Remarkably, we also found that SRPK2 T492 phosphorylation escalated during cell-cycle progression, correlating with delta-secretase S226 phosphorylation, which mediated delta-secretase cytoplasmic translocation from the lysosome (Figure 3). Our earlier study showed that SRPK2 translocates from the cytoplasm into the nucleus and phosphorylates SC35, an SR domain containing a splicing factor, triggering cell-cycle transition in neurons, thus inducing apoptosis through regulation of nuclear cyclin D1. Interestingly, SRPK2 is phosphorylated on T492 by activated Akt in the ischemia-damaged brain (Jang et al., 29). Moreover, we have also reported that SRPK2 binds and phosphorylates acinus, an SR protein essential for RNA splicing and apoptosis (Sahara et al., 1999; Schwerk et al., 23), and redistributes it from the nuclear speckles to the nucleoplasm, resulting in cyclin A1, but not A2, upregulation. Hence, SRPK2 phosphorylates the SR splicing machinery, regulates the cell cycle, and contributes to leukemia tumorigenesis. Recently, we reported that SRPK2 phosphorylates tau and suppresses tau-dependent microtubule polymerization, inhibiting axonal elongation in neurons. Depletion of SRPK2 in the dentate gyrus inhibits tau phosphorylation in APP/PS1 mice and rescues LTP, alleviating cognitive impairment. Moreover, active SRPK2 is increased in the APP/PS1 mouse cortex and in the human AD brain (Hong et al., 212). In AD, cell-cycle defects and associated aneuploidy have been described. The aneuploid and hyperploid neurons that arise in AD are particularly prone to degeneration and could account for 9% of the neuronal loss that characterizes late-stage AD (Arendt et al., 21). Individuals with the tau P31L mutation, which is associated with frontotemporal dementia, have several chromosome aberrations, such as aneuploidies in their fibroblasts and lymphocytes (Rossi et al., 28). Consequently, increased levels of aneuploidy in splenic lymphocytes of tauopathy transgenic mouse models have recently been reported (Rossi et al., 214). In Drosophila, human tau excess impairs mitosis, leading to aneuploidy and cell death (Bougé and Parmentier, 216). Plausibly, abnormal activation of SRPK2 may phosphorylate numerous substrates (including delta-secretase, tau, and SR domain-containing splicing factors), coordinating the AD pathologies and coupling defective cell-cycle machinery to neuronal cell death. Selective blockade of SRPK2 or delta-secretase by a small molecular inhibitor may provide an innovative therapeutic treatment for the devastating neurodegenerative disease. STAR+METHODS Detailed methods are provided in the online version of this paper and include the following: d KEY RESOURCES TABLE d CONTACT FOR REAGENT AND RESOURCE SHARING d EXPERIMENTAL MODEL AND SUBJECT DETAILS B Mice, primary cultured rat neurons, cell lines, and human tissue samples d METHOD DETAILS B Transfection and infection of the cells B In Vitro Kinase Assay B Subcellular Fractionation B AEP activity assay B AEP cleavage assay B Cleavage of Tau by SRPK2-phosphorylated AEP B HeLa cell synchrony B Immunoprecipitation and western blot analysis B Immunohistochemistry B Quantification of co-localization B Ab Plaque Staining B Electron Microscopy B Stereotaxic injection of lentivirus in mouse CA1 B Golgi stain B Electrophysiology B Morris Water maze B Contextual fear conditioning d QUANTIFICATION AND STATISTICAL ANALYSIS d DATA AND SOFTWARE AVAILABILITY SUPPLEMENTAL INFORMATION Supplemental Information includes six figures and can be found with this article online at (B) The S226A mutation reduces delta-secretase enzymatic activity. Data represent mean ± SEM of 3 mice/group (p <.5, p <.1, one-way ANOVA). (C and D) The S226A mutant decreases amyloid plaques and Ab in 5XFAD mouse CA1. (C) Quantification of the number and surface area of amyloid plaques (right). Plaque analysis data represent mean ± SEM of sections from 3 mice in each group (p <.5, p <.1, one-way ANOVA). Scale bar, mm. (D) IHC staining of Ab in the hippocampus (left; scale bar, mm). Also shown is quantitative analysis of Ab immunoreactivity (right). Data represent mean ± SEM of sections from 3 mice in each group (p <.5, one-way ANOVA). (E) The S226A mutant increases the spine density. Scale bar, 5 mm. Golgi staining was conducted on brain sections from CA1 regions of 5XFAD mice (mean ± SEM, n = 5, p <.5, one-way ANOVA). (F) Electrophysiology analysis. The S226A mutant increased the ratio of paired pulses (left) and rescued the LTP defects in 5XFAD mice (right) (mean ± SEM, n = 6 in each group, p <.5 5XFAD-S226A versus 5XFAD-Ct, # p <.5 5XFAD-S226A versus 5XFAD-, one-way ANOVA). The traces are representative fepsps recorded before (black) and 6 min after (red) theta-burst stimulation (TBS). (G I) MWM (G and H) and conditional fear conditioning (CFC) analysis of cognitive functions. The S226A mutant in CA1 rescues the learning and memory impairments in 5XFAD mice (mean ± SEM, n = 9 1 mice/group, p <.5, one-way ANOVA). See also Figure S6. Molecular Cell 67, , September 7,