Prostaglandin A1 inhibits the phosphorylation of tau via activating protein phosphotase 2A in a michael addition mechanism at the site of cysteine 377


 Background: Prostaglandin (PG) A1 is a metabolic product of cyclooxygenase 2 (COX-2), which potentially involved in regulating the development and progression of Alzheimer’s disease (AD). As a cyclopentenone (cy) PG, PGA1 is characterized by the presence of a chemically reactive α, β-unsaturated carbonyl. Although PGA1 is potentially involved in regulating multiple biological processes via michael addition, its specific roles in AD remained unclear.Methods: The tauP301S transgenic (Tg) mice were employed as in vivo AD models and neuroblastoma (N) 2a cells as in vitro neuronal models. By intracerebroventricular injected (i.c.v) with PGA1, the binding proteins to PGA1 are analyzed by HPLC-MS-MS. In addition, western blots are used to determine the phosphorylation of tau in PGA1 treated Tg mice in the absence or presence of okadaic acid (OA), an inhibitor of protein phosphotase (PP) 2A. Combining a synthesis of pull down assay, immunoprecipitation, western blots and HPLC-MS-MS, PP2A scaffold subunit A alpha (PPP2R1A) was identified to be activated by directly binding on PGA1 in cysteine 377-dependent manner. Via inhibiting the hyperphosphorylation of tau, morris maze test was employed to determine the inhibitory effects of PGA1 on cognitive decline of tauP301S Tg mice.Results: By incubation with neuroblastoma (n)2a cells and pull down assay, mass spectra (MS) analysis revealed that PGA1 binds with more than 1000 proteins, among which contains the proteins of AD, especially tau protein. Moreover, short-term administration of PGA1 to tauP301S Tg mice significantly decreased the phosphorylation of tau at the sites of Thr181, Ser202 and Ser404 in a dose-dependent manner. To the reason, it’s caused by activating PPP2R1A in tauP301S Tg mice. More importantly, PGA1 has the ability to form michael adduct with PPP2R1A via its cysteine 377 motif, which is critical for the enzymatic activity of PP2A. By activating PP2A, long-term application of PGA1 to tauP301S Tg mice significantly reduced the phosphorylation of tau, which results in improving the cognitive decline of tauP301S Tg mice.Conclusion: Our data provided the first insights needed to decipher the mechanisms underlying the ameliorating effects of PGA1 on cognitive decline of tauP301S Tg mice via activating PP2A in a PPP2R1AC377-dependent Michael adducting mechanisms.


Introduction
Alzheimer's disease (AD) is an occult and progressive neurodegenerative disease in central nervous system. The main clinical syndrome is the cognitive dysfunction. Although the etiology and pathogenesis of AD have not been thoroughly elucidated, the deposition of β-amyloid protein (Aβ) in β-amyloid plaques (APs) and hyperphosphorylated tau in neuro bribillary tangles (NFTs) are still regarded as the pathological characteristics of AD (1). There are at least 50 million AD patients in the world as a chronic disease and the number of the patients will increase dramatically with the increase of age, which results in heavy burden to the family and society.
Although the pathogenesis of AD is not thoroughly clear, some clinical and experimental studies have shown that the occurrence and development of AD are related to metabolic disorders. For instances, the risk of AD in patients will increase signi cantly with the metabolic disturbance in glucose and cholesterol (2,3). Interestingly, metabolic disorders in either glucose or cholesterol will result in the activation of cyclooxygenase-2 (COX-2) (4,5). As a metabolic enzyme, the activation of COX-2 can signi cantly accelerate the occurrence and development of AD, including the production and deposition of Aβ and phosphorylated tau proteins, which impaired the learning ability and memory (6). In this process, the in ammatory mechanisms of COX-2 play the key roles in AD (7). The administration of COX-2 speci c inhibitor can improve the learning and memory ability of APP/PS1 transgenic (Tg) mice by inhibiting the production and aggregation of Aβ and phosphorylated tau protein, which result in reducing the apoptosis of neurons and promoting the neurogenesis (8). As an important synthases of PGs, the activation of COX-2 will inevitably affect the metabolism of PGs. The over accumulation of PGE 2 in the brain can activate presenilin (PS) 1/2 by upregulating tumor necrosis factor (TNF)-α, leading to promote the production and aggregation of Aβ and ultimately damage the learning and memory ability of APP/PS1 Tg mice (9). PGI 2 promotes the amyloid metabolism of amyloid precursor protein (APP) by activating APH-1α and − 1β and increases the production and deposition of Aβ, which impaired the learning and memory ability of AD mice (10).
Apart from 15d-PGJ 2 , PGA1 is another important member of cyPGs. Compared to 15d-PGJ 2 , the roles of PGA1 in AD are highly overlooked. Since cyPG is different from the classical PGs, PGA1 contains α-and β-unsaturated ketone groups and an electrophilic center, which makes it more likely to have michael addition reaction with other proteins, leading to different physiological functions, including antiin ammatory, oxidative stress and transcriptional activity (18). As an important member of cyPGs, PGA1 has shown neuroprotective effect (19). PGA1 can also be directly combined with free sulfhydryl group of cysteine residue through michael addition, which results in strong anti-in ammatory effect (20). In SH-SY5Y cells, PGA1 can reduce rotenone-induced cytotoxicity (21). In C6 glioma cells, PGA1 can up regulate the expression of neurotrophic factors (22). These neuroprotective effects of PGA1 are likely to improve the pathological characteristics of AD. Based on this hypothesis, our prior works have found that PGA1 can reduce the production and deposition of Aβ by inhibiting the activity of presenilin enhancer 2 (PEN2) (23). During this process, PGA1 induces the e ux of cholesterol in ATP-binding cassette transporter A1 (ABCA1)-and PPAR γ-dependent mechanism, which deactivates γ-secretase subunit, PEN2 in vitro and in vivo (23).
Although PGA1 has shown its effects on inhibiting the production and aggregation of Aβ, there is no evidence to elucidate the mechanisms of PGA1 in the phosphorylation of tau during the course of AD development and progression. As a dephosphorylating enzyame, protein phosphotase (PP) 2A has been reported to be downregulated in AD (24). For a trimeric serine (Ser)/threonine (Thr) protein phosphatase, PP2A is composed of a 65 kDa scaffolding subunit (A), 36 kDa catalytic subunit (C) and more variable regulatory subunit (B), leading to its diverse functions in in ammation and neurodegeneration (25).
On the basis of these potential clues, we aimed to elucidate the mechanisms of PGA1 in suppressing the phosphorylation of tau via regulating the activity of PP2A in tau P301S Tg mice. Furthermore, the motifs for the activity of PP2A in mediating the effects of PGA1 on inhibiting the phosphorylation of tau were identi ed by IP and HPLC-MS-MS. Finally, PGA1 shows its effects on suppressing the cognitive decline of tau P301S Tg mice. On the basis of these observations, PGA1 might be a useful chemical candidate for treating AD.
(Piscataway, NJ, USA Tg Mice and treatments-Tau P301S Tg mice (Stock No. 008169) were obtained from the Jackson Laboratory (Bar Harbor, ME, USA), as a model of AD. Genotyping was performed at 1 month after birth. Mice were housed in a standard environment. 3-month-old Tg mice were intranasally instilled with PGA1 (2.5 mg/kg/day) for 6 months.
The learning ability of the mice were determined by Morris water maze test or nest construction assay. After behavior assay, the mice were anesthetized, perfused and xed before collecting the brains for further physiological and biochemical test.
Morris water maze test- The mice were trained and tested in a Morris water maze after 6-month treatment with PGA1. Brie y, the mice were pretrained in a circular water maze with a visible platform for 2 d. The mice were then allowed to swim freely in a hidden platform pool by adding milk for 6 d. On the last day, the platform was removed and the times that the mice passed through the memorized platform were recorded.

Transfection-
The retroviral vectors encoding the sequences of PPP2R1A, PPP2R1A C377A and PPP2R1A C390A were cotransfected to HEK293T cells with the lentivirus packaging vectors pLP/VSVG, pRSV-Rev and pMDLg/pRRE by a calcium phosphate transfection kit (AMOGENE, Fujian, China). The ag-tagged proteins were used for pull-down (PD) or immunoprecipitation (IP) assay. Pull down assay-The proteins were extracted from cells or brains, which were further incubated with str-resin. Brie y, 100 µg of protein were incubated with str-resin in the buffer containing 1% NP-40 and 0.1% SDS. The unbound proteins were removed by extensive washes with 0.1 mM EDTA, 50 mM Tris-HCl (pH 7.6), 10 min. The supernatant was used for western blots. The bio-PGA1 and PPP2R1A were visualized by a str-HRP (1:5000, 1 µg/µl) or mouse monoclonal antibody against ag-tag (1:2000, 1 µg/µl).
Mass spectrum-1 µg His-tagged PPP2R1A was incubated with 1 µM PGA1 at 37 °C for 1 h. The reacting solution was centrifuged for collecting supernatant, which was then subjected to HPLC-MS-MS analysis. As previously described (32), unassigned ions or those with a charge < 1 + or > 7 + were rejected by MS/MS. Raw data was processed by Thermo Proteome Discoverer (version 2.1) against the SwissProt proteome database using an extracted FASTA le speci ed for "mouse" taxonomy. The searches were carried out with the maximum 10 ppm and 0.02 Da error tolerances for the survey scan and MS/MS analysis, respectively. The protein identi cations were based on at least six amino acids in length and < 1% of the false discovery rate (FDR). On the basis of this analysis, enriched pathways were further identi ed in the Kyoto Encyclopedia of Genes and Genomes (KEGG) or gene ontology (GO) analysis. Statistical Analysis-All data are presented as the means ± S.E. of independent experiments. The statistical signi cance of the differences between the means was determined using Student's t test or one-way analysis of variance, where appropriate. If the means were signi cantly different, multiple pair wise comparisons were performed using Tukey's post hoc test.

PGA1 suppresses the phosphorylation of tau in tau P301S Tg mice
Prior works have demonstrated that PGA1 is closely associated with AD (23), we thereby continue to elucidate the inherent mechanisms. First of all, n2a cells were incubated with biotin conjugated PGA1 (bio-PGA1) for 2 h. By addition of Streptavidin-555, the distribution of PGA1 was determined under immuno uorescence confocal microscopy. The results demonstrated that PGA1 primarily located in the cytoplasm, but less in the nucleus of neurons (Fig. 1a). As early as the 1970s, PGA1 has been found to be responsible for regulating the physiological and biochemical functions through michael addition (33).
After that, a series of proteins including thioredoxin reductase and heat shock protein (Hsp) 90 were fond to form michael adducts with PGA1 (34,35). However, the binding proteins with PGA1 in AD are quite limited. Therefore, we are promoted to identify the proteins to bind with PGA1 in tau P301S Tg mice. By pull down assay, we found that PGA1 interacted with a variety of proteins, whose molecular weight ranges from less than 26 kDa to more than 180 kDa (Fig. 1b). Moreover, most of proteins concentrated between 43 and 72 kDa in the gel (Fig. 1b). To distinguish these proteins, the lanes of western blots were cut for MS analysis. As a consequence, 1060 proteins were identi ed to be signi cantly modi cated by PGA1 (10 µM) in tau P301S Tg mice. The proteins were then functionally annotated by Cluster Pro ler for analyzing the pathways and biological processes in PGA1-stimulated tau P301S Tg mice. The results of KEGG analysis revealed that the PGA1-modi cated proteins are associated with Parkinson, Huntington and Alzheimer's diseases, etc (Fig. 1c). In addition, the results of GO analysis showed that PGA1 has the ability to bind on many proteins, among which tau were identi ed in PGA1 treated tau P301S Tg mice (Fig. 1d).
PP2A mediates the effects of PGA1 on dephosphorylating tau.
As PP2A has been regarded as a canonical enzyme for dephosphorylation (36), we next determine its roles in tau phosphorylation. By treatment of n2a cells with PGA1 (10 µM) in the absence or presence of PP2A inhibitor, OA (20 nM), the activity of PP2A was initially determined by malachite green kit according to the manufacturer's instructions. The results demonstrated that OA treatment thoroughly blocked the effects of PGA1 on stimulating the activity of PP2A (Fig. 3a). In addition, the inhibitory effects of PGA1 on the phosphorylation of tau were reversed by the addition of OA (Fig. 3d). In line with these in vitro data, injection (i.c.v.) of 5 µl OA (2 ng/µl) to the ventricles of tau P301S Tg mice also blocked the suppressive effects of 5 µl PGA1 (1 µg/µl) on the phosphorylation of tau via modulating the activity of PP2A (Figs. 3b,  c, e, f). These observations demonstrated that PGA1 suppressed the phosphorylation of tau at the sites of Thr181, Ser202 and Ser404 via activating PP2A.
PPP2R1A is identi ed as the downstream target of PGA1 in regulating the phosphorylation of tau.
Since PP2A is composed by a series of subunits including A, B and C and the minimal structure of PP2A to dephosphorylate tau is the complex composed of A and C subunits (37), we thereby continue to explore the effects of different subunits of PP2A on the phosphorylation of tau. By MS analysis, PPP2R1A was identi ed as a potential target of PGA1. In addition, knocking down the expression of PPP2R1A obviously impaired the enzymatic activity of PP2A in PGA1-stimulated n2a cells (Fig. 4a). Once the activity of PP2A was attenuated by knocking down the expression of PPP2R1A, the phosphorylation of tau was restored to the levels of untreated controls (Fig. 4b). Therefore, PPP2R1A is an essential subunit for regulating the activity of PP2A, which mediates the effects of PGA1 on regulating the phosphorylation of tau during the course of AD development and progression. For the reason, we rstly determined if PGA1 has the ability to upregulate the expression of PPP2R1A in vitro and in vivo. Unfortunately, the mRNA and protein expression of PPP2R1A are not regulated in PGA1-treated n2a cells and tau P301S Tg mice (Figs. 4c, d).
PGA1 usually exerts its biological functions by forming michael adducts with different proteins (38), we therefore speculate if PGA1 can bind with PPP2R1A via michael addition. To this purpose, PPP2R1A with ag-tag was ectopically expressed in HEK293 cells. By pull down assay, we found that PPP2R1A overexpression enhanced the interaction between bio-PGA1 and proteins (Fig. 4e). This result indicated that PGA1 and PPP2R1A might form the michael adducts. To con rm this hypothesis, the PPP2R1A was immunoprecipitated with anti-ag antibody and the combined bio-PGA1 was visualized by western blots. The results demonstrated that PGA1 has the ability to form michael adducts with PPP2R1A (Fig. 4f).
To identify the binding sites of PPP2R1A, HPLC-MS-MS was carried out to analyze the in vitro incubated complex of PGA1 and puri ed PPP2R1A. The results of fragment ions showed that a 336 Da mass shifts were found by analyzing a continuous series of b-and y-type ions, which is corresponding to the adduction of PGA1 on the Cys motif of PPP2R1A (Fig. 5a). By analyzing the protein sequences of PPP2R1A, Cys 377 and Cys 390 are the potential binding sites of PPP2R1A for PGA1. By amino acid mutation, we nd that Cys 377, but not Cys390 mutation clearly impaired the binding between PGA1 and PPP2R1A (Figs. 5b, c). Without combination, PGA1 could not affect the enzymatic activity of PP2A anymore (Figs. 5d, e). On the basis of these observations, forming the michael adducts between PGA1 and PPP2R1A is the critical process for affecting the enzymatic activity of PP2A.
Long-term administration of PGA1 improved the cognitive decline of tau P301S Tg mice via decreasing the phosphorylation of tau in a PGA1-activating mechanism.
Since short-term administration of PGA1 to tau P301S Tg mice has shown its effects on inhibiting the phosphorylation of tau, 3-month-old tau P301S Tg mice were further intranasally instilled with PGA1 (2.5 mg/kg/d) for 6 m. By measuring the enzymatic activity of PP2A and the phosphorylation of tau, we found that PGA1 stimulated the activity of PP2A and inhibited the phosphorylation of tau in either WT or tau P301S Tg mice (Figs. 6a-d). On the basis of these observations, we are promoted to determine if PGA1 has the ability to affect the learning ability of tau P301S Tg mice. By nest construction and morris maze test, we found that PGA1 treatment obviously prevented the cognitive decline of tau P301S Tg mice (Figs. 6e-h). All these observations revealed that long-term administration of PGA1 improved the cognitive decline of tau P301S Tg mice via decreasing the phosphorylation of tau in a PP2A-activating mechanism.
Discussion AD is associated with the production and deposition of Aβ and phosphorylated tau in APs and NFTs (1). The overloading of aggregated Aβ and phosphorylated tau will induce excessive toxicity to the neurons, which results in impairing the brains of AD patients (39,40). For neuroprotection, PGA1 has been reported to blunt N-methyl-D-aspartate receptor (NMDAR)-mediated neuronal apoptosis and inhibits enhancement of the intracellular calcium concentration, TXA 2 production, and platelet activation in rodent models of stroke (41). In addition, PGA1 protects striatal neurons against excitotoxic injury in rat striatum (42). In rat models of permanent focal cerebral ischemia, PGA1 also plays neuroprotective effects by inhibiting the activity of NF-κB and upregulating PPARγ (43). In C6 glioma cells, PGA1 was postulated to upregulate the expression of neurotrophic factors, which is bene cial the neurons (44). All these evidences indicated the bene cial effects of PGA1 on AD. By this hypothesis, our previous studies have demonstrated that PGA1 has the ability to inhibit the production and aggregation of Aβ via inducing the e ux of cholesterol from the neurons (23). To extend the prior works, the current investigation studied the mechanisms of PGA1 in regulating the phosphorylation of tau via PP2A-activating mechanisms in tau P301S Tg mice.
As an important member of cyPGs, PGA1 usually modulates biological processes via binding with other functional proteins by michael addition. By MS analysis, PGA1 has been found to form michael adducts with a series of proteins. For example, PGA1 deactivates the activity of the enzyme by binding covalently to Cys299 motif of AKB1B10 (38). Using MALDI-TOF MS, H-Ras Cys118 was identi ed to be the critical site for PGA1 binding by michael addition (45). Although the residues involved in the modi cation have not been identi ed, covalent modi cation of Hsp90 by biotinylated PGA1 (PGA1-B) has been demonstrated in vitro (46). Even though the functions of PGA1 adducts have yet to be explored, numerous in vitro and in vivo studies have revealed the biological effects of cyPG adducts on cellular processes. For instance, the anti-in ammatory roles of cyPG have been revealed by inhibiting the expression of iNOS or COX-2 in PPARγ-or Nrf2-activating and NF-κB-or AP-1-deactivating mechanisms (47,48). Apart from its antiin ammatory effects, cyPGs also have the ability to regulate the cellular redox status (49), proliferation and apoptosis (50).
On the basis of these clues, we extended the prior works to nd that bio-PGA1 primarily distributed in the cytosol of n2a cells (Fig. 1a). In line with our observation, PGA1 was also found to be located in the cytosol of NIH3T3 cells (46). These data indicated the potential ability of PGA1 to bind on the cytosolic proteins in neurons. By this hypothesis, the bio-PGA1 conjugated proteins were analyzed by MS. KEGG and GO analysis depicted the PGA1-regulated the physio-or pathological processes, among which includes AD and tau protein (Figs. 1b-d). More importantly, PGA1 was rstly identi ed to suppress the phosphorylation of tau in tau P301S Tg mice (Fig. 2). Even though there is no direct evidence to support our results, the anti-in ammatory effects of PGA1 might be critical for dephosphorylating tau. For example, IL-1β expression in APP/PS1 Tg mice markedly induced the phosphorylation of tau (51). Blocking the signaling cascade of IL-1 inhibited the tauopathy, which results in improving the learning ability of APP/PS1 Tg mice (52). These in vivo observations were further supported by the in vitro data showing that IL-1β stimulated tau phosphorylation in cultured microglia, astrocytes, and neurons (53). Apart from IL-1β, IL-18 and IL-6 also has the ability to phosphorylate tau in human SH-SY5Y cells (54) and cultured hippocampal neurons (55). As a potential anti-in ammatory factor (47,48), PGA1 might inhibit the phosphorylation of tau via suppressing the expression of pro-in ammatory cytokines. As another important member of cyPGs, 15d-PGJ 2 has shown its effects on suppressing the proin ammatory cytokines (56), leading to the phosphorylation of tau in vitro and in vivo (51)(52)(53)(54)(55). Along these lines, these anti-in ammatory pathways might be the mechanisms for mediating the effects of PGA1 on suppressing the phosphorylation of tau.
Even though in ammatory pathways had not been elucidated in our experiments, we found that PGA1 has the ability to activate PP2A (Fig. 3). More importantly, Cys377 is the key motif for PP2A subunit, PPP2R1A to bind with PGA1, which regulates the activity of PP2A (Figs. 5). In agreement with our results, Cys377 on the scaffolding subunit, PPP2R1A was reported to be responsible for the activity of PP2A (57). In addition, PGA1 has the ability to bind on cytosol proteins in NIH3T3 cells (58). Based on the crystal structures of PP2A, Cys377 resides at an interface between three subunits of the core PP2A complex (59), which provides the position advantage of PPP2R1A to regulate the activity of other subunits.
Since PGA1 has the ability to activate PP2A, the question is easily raised if PP2A has the ability to regulate the phosphorylation of tau during the course of AD development and progression. Recently, more and more studies established the relationship between PP2A and AD (60,61). More accurately, the activity and expression of PP2A including its A, B and C subunits are decreased in the brains of AD patients (61)(62)(63). More closely, the expression of PPP2R1A was downregulated in the brains of AD patients compared to that of corresponding normal subjects (63,64). Speci cally in the high aluminum AD model, the activity of PP2A decreased by 59% (65). Further studies showed that the accumulation of zinc ions and the expression of APOE were the pathological factors for decreasing the activity of PP2A (66,67). Due to the decrease of PP2A, the phosphorylation of tau is increased by activating GSK3β (68).
The use of endogenous PP2A inhibitor obviously induces the phosphorylation of tau (69). Consistently, OA, an inhibitor of PP2A blocked the effects of PGA1 on suppressing the phosphorylation of tau in vitro and in vivo (Fig. 3). These observations reinforced the critical roles of PP2A in mediating the effects of PGA1 on suppressing the phosphorylation of tau during the course of AD development and progression.
As tau experiences phosphorylation, aggregation and deposition during the course of AD development and progression, it will lose its ability to maintain the stability of microtubes (70). Moreover, the aggregated form of tau will induce neurotoxicity (39), leading to the apoptosis of neurons (71). Long-term potentiation (LTP) was obviously suppressed in tau P301S Tg mice, which potentially affected the memory of the mice (72). Not surprisingly, the results of nest construction and morris maze test demonstrated that the learning ability was obviously impaired in tau P301S Tg mice (Fig. 6). More importantly, PGA1 treatment ameliorated the cognitive decline of tau P301S Tg mice (Fig. 6).

Con icts of Interest
The authors declare no con icts of interests.
Availability of data and materials Page 14/26 All data generated or analyzed in this study are included in the current published article.
Ethics approval and consent to participate All animals were handled according to the guidelines for the care and use of medical laboratory animals (Ministry of Health, Peoples Republic of China, 1998)    further analyzed by the Image J software. The data represent means ± S.E. of independent experiments. ***p < 0.001 compared to vehicle-treated WT mice. ###p < 0.001 compared to vehicle-treated tauP301S Tg mice.   activity was detected by malachite green kit. The data represent means ± S.E. of independent experiments. *p < 0.05, **p < 0.01 compared to vehicle-treated control. Long-term administration of PGA1 decreased the phosphorylation of tau via activating PP2A, which in turn ameliorates the cognitive decline of tauP301S Tg mice. 3-month-old tauP301S Tg mice were intranasally treated with or without PGA1 (2.5 mg/kg/d) for 6 months. (a, b) The activity of PP2A was determined by malachite green kit. (c, d) The phosphorylation of tau at the sites of Thr181, Ser202 and Ser404 were determined by western blots. The total tau and β-actin served as internal control. (e-h) The learning ability of tauP301S Tg mice was determined by morris maze test and nest construction assay. The data represent means ± S.E. of independent experiments. **p < 0.01, ***p < 0.001, compared to vehicle-treated control.