Phosphoproteomic analysis identified tau-related kinases linking diabetes mellites and Alzheimer’s disease-related tau phosphorylation

Abstract

Background: Diabetes mellitus (DM) is a risk factor for both vascular dementia and Alzheimer’s disease (AD). We previously reported that tau phosphorylation is increased in the brain of a diabetic AD mouse model, which can be an underlying mechanism linking DM and AD. However, the key molecules mediating the tau phosphorylation by DM remain unknown. Here, using a phosphoproteomic approach, we aimed to identify candidate kinases that regulate tau phosphorylation by DM.

Method and Results: We generated diabetic tau-transgenic (TauTg) mice by high-fat diet (HFD) feeding and conducted a phosphoproteomic analysis of their brain tissues. Phosphoproteome profiling of the brains from the diabetic and non-diabetic TauTg mice were compared. We identified 30 kinases whose levels of phosphorylation were significantly altered in the brains of the diabetic TauTg mice. Molecular networking analysis found that six of the 30 kinases were associated with tau phosphorylation. Among the six kinases, the decrease in the phosphorylation levels of CaMK2α, which reflected the severity of diabetic conditions, was associated with the exacerbation of accumulation of phosphorylated tau in the brains of TauTg mice.

Conclusions: Tau-related kinases that potentially mediate DM-induced tau phosphorylation were identified using a phosphoproteomic approach. These kinases could be therapeutic targets for preventing AD in patients with DM.

 Introduction

Diabetes mellitus (DM) is a well-established risk factor for stroke and vascular dementia (VaD)^1,2). Cognitive decline associated with DM is also known as diabetic dementia; it results from changes occurring in the brain’s insulin signaling and energy metabolism due to DM and impairs neuronal function³⁾. Recurrent hypoglycemia caused by insulin treatment can result in DM-related neuronal damage and eventually develop into dementia⁴⁾. Recent epidemiological studies have highlighted that DM is not only a risk factor for VaD but also for Alzheimer’s disease (AD)^2,5–8). However, the underlying mechanisms by which DM increases the risk of developing AD are largely unknown.

Tau protein is a major component of neurofibrillary tangles (NFTs), one of the neuropathological features of AD^9,10). Tau in the NFTs is highly phosphorylated and aggregated¹¹⁾. The severities of neuronal loss and cognitive decline in patients with AD correlate well with their number of NFTs¹²⁾, indicating the crucial role of tau in neurodegeneration. Tau undergoes various post-translational modifications, including phosphorylation, in both the physiological and pathological conditions¹³⁾ associated with its biological functions, such as stabilizing the microtubules in neurons. The hyperphosphorylation of tau has been reported to accelerate self-aggregation and impair its physiological function^14,15) and is thought to play a critical role in the formation of NFTs and neuronal dysfunction¹⁶⁾.

DM has been reported to increase NFTs in post-mortem brains^6,17) and to promote the phosphorylation of tau in the brains of animal models^18–23). We previously demonstrated that DM induces a unique phosphorylation pattern of tau in the brains of diabetic TauTg mice²³⁾. These results suggested that the phosphorylation of tau is one of the key mechanisms linking DM and AD.

The phosphorylation of tau is regulated by multiple kinases and phosphatases²⁴⁾. In pathological conditions such as AD, the dysregulation of the enzymatic activities of the tau-related kinases and phosphatases induces the hyperphosphorylation of tau, leading to the formation of NFTs and neurodegeneration. For example, GSK3β, which is associated with insulin signaling, has been shown to play a role in developing aggregations of hyperphosphorylated tau^25,26). Tau-related kinases are currently one of the major targets for therapeutic development for AD and other tauopathies. Identifying tau-related kinases that potentially link DM and AD may provide a novel therapeutic avenue for AD prevention, especially for individuals with DM.

In this study, using a phosphoproteomic approach, we aimed to identify candidate kinases that are associated with the tau phosphorylation induced by diabetic conditions. We generated a diabetic AD mouse model by feeding TauTg mice a high-fat diet (HFD) and then performed phosphoproteomic analyses of their brain tissues. Phosphoproteome profiles of the brains of diabetic and non-diabetic TauTg mice were compared to explore kinases whose levels were significantly changed in the diabetic conditions. Molecular network analysis was used to identify candidate tau-related kinases. Finally, the correlations between the levels of the candidate tau-related kinases and the severity of DM, and the amount of phosphorylated tau in the brain, were evaluated.

 Materials and methods

 Animal models

TauTg (PS19) mice were purchased from Charles River Laboratories Japan, Inc. (Kanagawa, Japan). These express the P301S mutant form of human microtubule-associated protein tau, under the direction of mouse prion protein promoter, at a level five-fold greater than the endogenous mouse protein²⁷⁾. Mice were genotyped via DNA analysis from tail clips. Female TauTg littermates were maintained in a temperature-controlled room (25°C ± 2°C) with a 12-hour light–dark cycle and provided food and water ad libitum.

Mice were divided into two groups and fed either a HFD (60% kcal fat, HFD-60, Oriental Yeast Co., Ltd., Japan) or normal chow diet (NCD) (12% kcal fat, MF, Oriental Yeast Co., Ltd., Japan) from 1.5 to 9 months of age. For tissue collection, mice were anesthetized (0.75 mg medetomidine hydrochloride, 4.0 mg midazolam, 5.0 mg butorphanol tartrate/kg body weight, intraperitoneal administration) and intracardially perfused with ice-cold phosphate buffered saline (PBS). The right hemispheres were immediately frozen and stored at –80°C.

We performed all procedures at Osaka University School of Medicine after gaining approval of the study protocol from the Animal Experiments Committee of Osaka University, Osaka, Japan (decision reference number 29-046-015). This study protocol was reviewed and approved by the Osaka University Living Modified Organisms (LMO) Research Safety Committee (approval number 04232). All animal experiments complied with ARRIVE guidelines and were carried out in accordance with the National Institutes of Health Guide for the Care and Use of Laboratory Animals.

 Metabolic measurements

Blood glucose and plasma insulin determinations were made in fasting (16 hours) states at 9 months of age. We measured blood glucose levels (mg/dl) using the Glutest Neo alpha glucose oxidase method (Sanwa Kagaku Kenkyusho Co., Ltd., Japan) and plasma insulin using an enzyme-linked immunosorbent assay kit (M1104, Morinaga Institute of Biological Science, Inc., Japan). Abdominal fat pads were collected and weighed with a microbalance.

 Data-independent acquisition (DIA)-based phosphoproteomic analysis

A DIA phosphoproteomic analysis was performed in accordance with a previously reported protocol (Kazusa Genome Technologies Inc. (Promega))²⁸⁾. The cortex was homogenized in 5 × (w/v) PBS (14249-95, Nacalai Tesque) containing a protease inhibitor cocktail (PIC) (5871, Cell Signaling Technology). The brain homogenate was incubated to 8 volumes of cold acetone for 2 hours at –20°C. The sample was precipitated in acetonitrile containing 0.1% trifluoroacetic acid and then extracted in 0.5% sodium dodecanoate and 100 mM Tris-HCl (pH 8.5) for treatment with 10 mM dithiothreitol and alkylation with 30 mM iodoacetamide. The mixture was digested by adding 5 μg of Trypsin/Lys-C mix and centrifuged at 15,000 g at room temperature for 5 minutes. The supernatant was desalted using C18-StageTips, followed by drying with a centrifugal evaporator. The dried peptides were redissolved in 3% acetonitrile and 0.1% formic acid and then measured.

Peptides were separated with 90 main gradients of Solvent A (0.1% formic acid in water) and Solvent B (0.1% formic acid in 80% acetonitrile), comprising 1% B from 0 minutes and 70% B from 90 minutes. Peptides eluted from the column were analyzed on a Q Exactive HF-X (Thermo Fisher Scientific) for both data-dependent acquisition (DDA) and data-independent acquisition (DIA) MS analyses. LC–MS/MS analyses were performed with an UltiMate 3000 RSLC nanoLC system (Thermo Fisher Scientific). Peptides eluted from the column were analyzed on a Q Exactive HF-X system (Thermo Fisher Scientific) for both data-dependent acquisition (DDA) and DIA-MS analyses.

MS files were searched against the dataset containing the UniProtKB/Swiss-Prot database of Mus musculus and the additional sequence data for the following isoforms of human tau with the P301S mutation and mouse tau: isoform human fatal-tau (UniProt ID P10636-2), isoform human tau-E with P301S (UniProt ID P10636-7 with P301S), isoform human tau-F with P301S (UniProt ID P10636-8 with P301S), isoform mouse tau-A (UniProt ID: P10637-2), modified mouse isoform tau-A (UniProt ID P10637-2 with aa 63–91 deletion), isoform mouse tau-C (UniProt ID P10637-4), and isoform mouse tau-E (UniProt ID P10637-6). We used Proteome Discoverer v2.3 (Thermo Fisher Scientific) with Sequest HT for DDA-MS files and Scaffold DIA v3.0 (Proteome Software Inc., Portland, OR, USA) for DIA-MS files.

A chromatogram library was generated by searching MS data in the library against the dataset containing the UniProtKB/Swiss-Prot database of Mus musculus and the additional sequence data using Scaffold DIA. The peptide identification threshold was a peptide false discovery rate of <1%. Data were analyzed using the Scaffold DIA search engine. The protein identification threshold was a peptide or protein false discovery rate of <1% in both the protein sequence database and chromatogram library. Peptide quantification was calculated using the EncyclopeDIA algorithm²⁹⁾ in Scaffold DIA.

 Bioinformatics analysis

The Institute of Medicinal Molecular Design (IMMD) developed KeyMolnet³⁰⁾, a comprehensive and stand-alone database about relationships among human genes and proteins, small molecules, diseases, pathways, and drugs. Expert biologists regularly update the knowledgebase. We input the DIA phosphorylated proteome data into the KeyMolnet software (version 5.6 I, IMMD Inc., Tokyo, Japan). The tau-related kinases were screened from all kinases detected by DIA phosphoproteomics analysis, based on KeyMolnet information for kinases reported to interact with tau.

 Preparation of 1% sarkosyl-insoluble brain extracts

Protein extraction was performed on tissues dissected from the right hemisphere. Briefly, the brain homogenate was centrifuged at 4°C and 10,000 × g for 15 minutes. The pellet was resuspended in the same volume of PBS containing 1% sarkosyl (20135-14, Nacalai Tesque) and PIC and was incubated at 37°C for 30 minutes. The samples were centrifuged at 4°C and 100,000 × g for 30 minutes. The pellet was then resuspended in 100 μl PBS containing PIC and sonicated (SONIFIER 250, BRANSON) for 2 min at room temperature. This suspension was analyzed as the sarkosyl-insoluble fraction. Protein concentrations were determined using a BCA assay kit (T9300A, TaKaRa Bio Inc., Japan).

 Tau ELISA

The concentrations of phosphorylated tau in the sarkosyl-insoluble fraction of TauTg mouse brains were determined using a Human Tau (Phospho) [pT181] ELISA Kit (Invitrogen, KHO0631), according to the manufacturer’s instructions. Samples were diluted at 1:200 using a sample diluent buffer provided by the ELISA kit.

 Statistical analysis

All data are expressed as mean ± standard error of the mean. Two-group comparisons were performed using Welch’s t-tests. Spearman’s rank correlation was applied to determine the correlation between two parameters. Statistical analyses were performed using Statcel 4 (OMS Publishing Inc., Tokorozawa, Japan). P-values of <0.05 were considered significant.

 Results

 Phosphoproteomic analyses identified tau-related kinases in the brains of diabetic TauTg mice

Fig. 1A shows the schematic diagram of the experimental design. Brain tissues from diabetic TauTg mice fed with HFD and non-diabetic TauTg mice fed with NCD were collected and used for phosphoproteomic analysis. Among 9,064 phosphorylated peptides identified in the brain extracts, 627 peptides originated from 190 different protein kinases. Among them, the levels of 36 peptides were significantly altered (18 upregulated and 18 downregulated) in the brains of diabetic TauTg-HFD mice, compared to non-diabetic TauTg-NCD mice (p < 0.05, Fig. 1B, Supplementary fig. 1). These 36 phosphorylated peptides originated from 30 different protein kinases (Fig. 1C, Supplementary fig. 1). Among the 30 kinases, molecular network analysis identified six kinases that were biologically associated with tau, including calcium/calmodulin-dependent protein kinase type II subunit alpha (CaMK2α), protein kinase C gamma (PKCγ), tau-tubulin kinase 1 (TTBK1), RAC-gamma serine/threonine kinase (AKT3), brain-specific serine/threonine kinase 2 (BRSK2), and serine/threonine kinase TAO1 (TAO1) (Fig. 1D). The levels of phosphorylated CaMK2α, PKCγ, TTBK1, AKT3, and TAO1 were significantly downregulated in the brains of diabetic TauTg-HFD mice, compared to non-diabetic TauTg-NCD mice (Supplementary fig. 2). BRSK2 showed either upregulation or downregulation of phosphorylation, depending on the phospho-sites in the brains of diabetic TauTg-HFD mice (Supplementary fig. 2).

Fig. 1  Phosphoproteomic analyses identified tau-related kinases in the brains of diabetic TauTg-HFD mice

(A) Schematic representation of phosphoproteomic analyses using the brain tissues from diabetic TauTg-HFD and non-diabetic TauTg-NCD mice. (B) Volcano plots showing fold changes (log2 transformation on the x-axis) versus the significance of altered phosphorylated peptides (–log10 transformed p-value on the y-axis). Significance level is indicated with a horizontal dotted line (p < 0.05). Significantly altered peptides were determined by Welch’s t-test (p < 0.05) and shown as red plots. Peptides that did not reach statistical significance are shown as black plots (n = 4 TauTg-NCD, n = 6 TauTg-HFD). (C) Schematic representation of screening for tau-related kinases based on the results of phosphoproteomic analyses. The colored lines represent fragmented peptides originating from distinct protein kinases. Tau-related kinases were identified using molecular networking analysis. (D) List of six tau-related kinases whose levels were significantly altered in the brains of TauTg-HFD mice compared to TauTg-NCD mice. Kinases were listed in the order of their accession numbers in the UniProt knowledgebase.

Abbreviations: HFD, high-fat diet; NCD, normal chow diet

 Correlations between the amounts of candidate tau-related kinases and the severity of DM

Next, we examined the correlations between the amounts of phosphorylated tau-related kinases and the severity of metabolic abnormalities (Fig. 2). The strongest correlation was found between the brain levels of phosphorylated CaMK2α (p234) and abdominal fat weight (r = –0.88, p < 0.01). The amount of p234-CaMK2α was also correlated with plasma insulin levels (r = –0.77, p < 0.05) and blood glucose levels (r = –0.72, p < 0.05). Other kinases showed mild to moderate correlations with each metabolic parameter. These results suggest that CaMK2α is the most relevant tau-related kinase associated with diabetic conditions.

Fig. 2  Correlations between the amounts of phosphorylated tau-related kinases and the severities of diabetic conditions

Spearman’s rank correlation coefficients between the levels of phosphorylated peptides originating from each tau-related kinase and the severities of each diabetic parameter are presented. We listed each kinase in the order of its accession number in the UniProt knowledgebase. The correlations that showed significant alterations are highlighted with light pink (r = 0.65–0.85), light blue (r = –0.65 – –0.85), and dark blue (r = 0.85–0.99) backgrounds (n = 4 TauTg-NCD, n = 6 TauTg-HFD). *p < 0.05, **p < 0.01, Spearman’s rank test.

 Phosphorylated CaMK2α levels were correlated with the amount of phosphorylated tau in the brain

Finally, we examined the correlation between the amount of phosphorylated CaMK2α and the levels of phosphorylated tau accumulated in the brains of TauTg mice (Fig. 3). We used the insoluble fraction of brain extracts for the quantification of phosphorylated tau (p181-tau). The amount of phosphorylated CaMK2α showed a significant negative correlation with the levels of phosphorylated tau (r = –0.72, p < 0.05).

Fig. 3  Correlation between the amounts of phosphorylated CaMK2α and phosphorylated tau in the brains of TauTg mice.

Correlation between the amounts of p234-CaMK2α peptide and insoluble form of p181 tau (n = 4 TauTg-NCD, n = 6 TauTg-HFD). p < 0.05, Spearman’s rank test.

Abbreviations: HFD, high-fat diet; NCD, normal chow diet

 Discussion

Vascular risk factors such as DM increase the risk of developing AD²⁾; however, the mechanisms underlying this pathological interplay remain largely unknown. Here, we explored key molecules that mediate DM-induced tau phosphorylation using a phosphoproteomic approach combined with molecular network analysis. Six tau-related kinases were identified in the brain tissues of a diabetic AD mouse model as key candidate molecules (Fig. 1). Among them, the CaMK2α levels were most significantly associated with diabetic conditions (Fig. 2) and the cerebral accumulation of phosphorylated tau (Fig. 3).

The six tau-related kinases showed significant alterations in the amount of phosphorylated peptides in the brains of TauTg-HFD mice (Fig.1D). Among the six tau-related kinases identified, four kinases (all but AKT3 and TAO1) are reported to be highly expressed in neuronal cells, compared to astrocytes and oligodendrocytes^31–34). CaMK2α³⁵⁾ and PKCγ³⁶⁾ play important roles in synaptic plasticity and hippocampal long-term potentiation. BRSK2 is reported to be involved in the polarization of neurons and axonogenesis³⁷⁾. These findings imply that the changes in the phosphorylation levels of these kinases may affect neuronal functions and be involved in the pathogenesis of DM-induced cognitive impairment.

The enzymatic activity of protein kinases can be upregulated³⁸⁾ or downregulated³⁹⁾, depending on their phosphorylation state. For example, the activity of CaMK2 is increased by phosphorylation at Thr286³⁸⁾, while phosphorylation at Thr306 of CaMK2 causes the inactivation of its enzymatic activity³⁹⁾. Phosphorylation at Thr286 of CaMK2 was reported to promote tau phosphorylation⁴⁰⁾. To the best of our knowledge, the effect of the phosphorylation at Ser234 of CaMK2 on its enzymatic activity, especially in relation to tau phosphorylation, has not yet been reported. Our results showed that the decrease in the levels of phosphorylation of CaMK2α at Ser234 was associated with the increased amounts of phosphorylated tau in the brain (Fig. 3). This suggests that the phosphorylation of CaMK2α at this site inhibits its activity as a protein kinase, resulting in the lower amount of phosphorylated tau. It should be noted that this study is a correlation analysis and does not imply causation between the changes in the phosphorylation levels of CaMK2α at Ser234 and tau.

In conclusion, a phosphoproteomic approach combined with molecular network analysis allowed us to identify tau-related kinases that potentially mediate DM-induced tau phosphorylation in the brains of a diabetic AD mouse model. The tau-related kinases may be key molecules linking DM and AD. These findings may provide a novel therapeutic avenue for AD prevention.

 Disclosures

The authors have no potential conflicts of interest to declare.

Supplementary fig. 1. List of 30 kinases whose levels of phosphorylation were significantly altered in the brains of TauTg-HFD mice.

The list is arranged in alphabetical order.

Abbreviations: HFD, high-fat diet

Supplementary fig. 2. List of six tau-related kinases whose levels of phosphorylation were significantly altered in the brains of TauTg-HFD mice.

We have listed each kinase in order of its accession number in the UniProt knowledgebase.

Acknowledgements

This study was supported by the Center for Medical Research and Education, Graduate School of Medicine, Osaka University and by grants from the Japan Society for the Promotion of Science (JSPS) KAKENHI, including a Grant-in-Aid for Young Scientists (A) (17H05080) and a Grant-in-Aid for Scientific Research (B) (21H02828), awarded to S.T., and a research grant from the Cell Science Research Foundation awarded to S.T.

References

• 1)   Gudala  K,  Bansal  D,  Schifano  F, et al. Diabetes mellitus and risk of dementia: a meta-analysis of prospective observational studies. J Diabetes Investig 2013; 4: 640–650. doi: 10.1111/jdi.12087.
• 2)   Livingston  G,  Sommerlad  A,  Orgeta  V, et al. Dementia prevention, intervention, and care. Lancet 2017; 390: 2673–2734. doi: 10.1016/S0140-6736(17)31363-6.
• 3)   Hanyu  H. Diabetes-related dementia. Adv Exp Med Biol 2019; 1128: 147–160. doi: 10.1007/978-981-13-3540-2_8.
• 4)   Husain  KH,  Sarhan  SF,  AlKhalifa  HKAA, et al. Dementia in diabetes: the role of hypoglycemia. Int J Mol Sci 2023; 24: 9846. doi: 10.3390/ijms24129846.
• 5)   Ohara  T,  Doi  Y,  Ninomiya  T, et al. Glucose tolerance status and risk of dementia in the community: the Hisayama Study. Neurology 2011; 77: 1126–1134. doi: 10.1212/WNL.0b013e31822f0435.
• 6)   Ott  A,  Stolk  RP,  van Harskamp  F, et al. Diabetes mellitus and the risk of dementia: the Rotterdam Study. Neurology 1999; 53: 1937–1942. doi: 10.1212/wnl.53.9.1937.
• 7)   Barbiellini Amidei  CB,  Fayosse  A,  Dumurgier  J, et al. Association between age at diabetes onset and subsequent risk of dementia. JAMA 2021; 325: 1640–1649. doi: 10.1001/jama.2021.4001.
• 8)   Takeda  S,  Sato  N,  Uchio-Yamada  K, et al. Diabetes-accelerated memory dysfunction via cerebrovascular inflammation and Abeta deposition in an Alzheimer mouse model with diabetes. Proc Natl Acad Sci USA 2010; 107: 7036–7041. doi: 10.1073/pnas.1000645107.
• 9)   Takeda  S,  Wegmann  S,  Cho  H, et al. Neuronal uptake and propagation of a rare phosphorylated high-molecular-weight tau derived from Alzheimer’s disease brain. Nat Commun 2015; 6: 8490. doi: 10.1038/ncomms9490.
• 10)   Takeda  S. Tau propagation as a diagnostic and therapeutic target for dementia: potentials and unanswered questions. Front Neurosci 2019; 13: 1274. doi: 10.3389/fnins.2019.01274.
• 11)   Grundke-Iqbal  I,  Iqbal  K,  Quinlan  M, et al. Microtubule-associated protein tau. A component of Alzheimer paired helical filaments. J Biol Chem 1986; 261: 6084–6089. https://pubmed.ncbi.nlm.nih.gov/3084478/ (accessed 23 January 2025).
• 12)   Gómez-Isla  T,  Hollister  R,  West  H, et al. Neuronal loss correlates with but exceeds neurofibrillary tangles in Alzheimer’s disease. Ann Neurol 1997; 41: 17–24. doi: 10.1002/ana.410410106.
• 13)   Wesseling  H,  Mair  W,  Kumar  M, et al. Tau PTM profiles identify patient heterogeneity and stages of Alzheimer’s disease. Cell 2020; 183: 1699–1713.e1613. doi: 10.1016/j.cell.2020.10.029.
• 14)   Alonso  A,  Zaidi  T,  Novak  M, et al. Hyperphosphorylation induces self-assembly of tau into tangles of paired helical filaments/straight filaments. Proc Natl Acad Sci USA 2011; 98: 6923–6928. doi: 10.1073/pnas.121119298.
• 15)   Scott  CW,  Spreen  RC,  Herman  JL, et al. Phosphorylation of recombinant tau by cAMP-dependent protein kinase. Identification of phosphorylation sites and effect on microtubule assembly. J Biol Chem 1993; 268: 1166–1173. https://pubmed.ncbi.nlm.nih.gov/8419321/ (accessed 23 January 2025).
• 16)   Trojanowski  JQ,  Lee  VM. Paired helical filament tau in Alzheimer’s disease. The kinase connection. Am J Pathol 1994; 144: 449–453. https://pmc.ncbi.nlm.nih.gov/articles/PMC1887092/ (accessed 23 January 2025).
• 17)   Matsuzaki  T,  Sasaki  K,  Tanizaki  Y, et al. Insulin resistance is associated with the pathology of Alzheimer disease: the Hisayama Study. Neurology 2010; 75: 764–770. doi: 10.1212/WNL.0b013e3181eee25f.
• 18)   Elahi  M,  Motoi  Y,  Shimonaka  S, et al. High-fat diet-induced activation of SGK1 promotes Alzheimer’s disease-associated tau pathology. Hum Mol Genet 2021; 30: 1693–1710. doi: 10.1093/hmg/ddab115.
• 19)   Leboucher  A,  Laurent  C,  Fernandez-Gomez  F-J, et al. Detrimental effects of diet-induced obesity on τ pathology are independent of insulin resistance in τ transgenic mice. Diabetes 2013; 62: 1681–1688. doi: 10.2337/db12-0866.
• 20)   Wu  Z,  Chen  C,  Kang  SS, et al. Neurotrophic signaling deficiency exacerbates environmental risks for Alzheimer’s disease pathogenesis. Proc Natl Acad Sci USA 2021; 118: e2100986118. doi: 10.1073/pnas.2100986118.
• 21)   Chen  JL,  Luo  C,  Pu  D, et al. Metformin attenuates diabetes-induced tau hyperphosphorylation in vitro and in vivo by enhancing autophagic clearance. Exp Neurol 2019; 311: 44–56. doi: 10.1016/j.expneurol.2018.09.008.
• 22)   Niedowicz  DM,  Reeves  VL,  Platt  TL, et al. Obesity and diabetes cause cognitive dysfunction in the absence of accelerated β-amyloid deposition in a novel murine model of mixed or vascular dementia. Acta Neuropathol Commun 2014; 2: 64. doi: 10.1186/2051-5960-2-64.
• 23)   Ito  Y,  Takeda  S,  Nakajima  T, et al. High-fat diet-induced diabetic conditions exacerbate cognitive impairment in a mouse model of Alzheimer’s disease via a specific tau phosphorylation pattern. J Prev Alzheimers Dis 2024; 11: 138–148. doi: 10.14283/jpad.2023.85.
• 24)   Congdon  EE,  Sigurdsson  EM. Tau-targeting therapies for Alzheimer disease. Nat Rev Neurol 2018; 14: 399–415. doi: 10.1038/s41582-018-0013-z.
• 25)   Bhat  RV,  Budd Haeberlein  SL,  Avila  J. Glycogen synthase kinase 3: a drug target for CNS therapies. J Neurochem 2004; 89: 1313–1317. doi: 10.1111/j.1471-4159.2004.02422.x.
• 26)   Eldar-Finkelman  H. Glycogen synthase kinase 3: an emerging therapeutic target. Trends Mol Med 2022; 8: 126–132. doi: 10.1016/s1471-4914(01)02266-3.
• 27)   Yoshiyama  Y,  Higuchi  M,  Zhang  B, et al. Synapse loss and microglial activation precede tangles in a P301S tauopathy mouse model. Neuron 2007; 53: 337–351. doi: 10.1016/j.neuron.2007.01.010.
• 28)   Kawashima  Y,  Watanabe  E,  Umeyama  T, et al. Optimization of data-independent acquisition mass spectrometry for deep and highly sensitive proteomic analysis. Int J Mol Sci 2019; 20. doi: 10.3390/ijms20235932.
• 29)   Searle  BC,  Pino  LK,  Egertson  JD, et al. Chromatogram libraries improve peptide detection and quantification by data independent acquisition mass spectrometry. Nat Commun 2018; 9: 5128. doi: 10.1038/s41467-018-07454-w.
• 30)   Sato  H,  Ishida  S,  Toda  K, et al. New approaches to mechanism analysis for drug discovery using DNA microarray data combined with KeyMolnet. Curr Drug Discov Technol 2005; 2: 89–98. doi: 10.2174/1570163054064701.
• 31)   Burgin  KE,  Waxham  MN,  Rickling  S, et al. In situ hybridization histochemistry of Ca2+/calmodulin-dependent protein kinase in developing rat brain. J Neurosci 1990; 10: 1788–1798. doi: 10.1523/JNEUROSCI.10-06-01788.1990.
• 32)   Kishi  M,  Pan  YA,  Crump  JG, et al. Mammalian SAD kinases are required for neuronal polarization. Science 2005; 307: 929–932. doi: 10.1126/science.1107403.
• 33)   Sato  S,  Cerny  RL,  Buescher  JL, et al. Tau-tubulin kinase 1 (TTBK1), a neuron-specific tau kinase candidate, is involved in tau phosphorylation and aggregation. J Neurochem 2006; 98: 1573–1584. doi: 10.1111/j.1471-4159.2006.04059.x.
• 34)   Tanaka  C,  Saito  N. Localization of subspecies of protein kinase C in the mammalian central nervous system. Neurochem Int 1992; 21: 499–512. doi: 10.1016/0197-0186(92)90081-2.
• 35)   Zalcman  G,  Federman  N,  Romano  A. CaMKII isoforms in learning and memory: localization and function. Front Mol Neurosci 2018; 11: 445. doi: 10.3389/fnmol.2018.00445.
• 36)   Saito  N,  Shirai  Y. Protein kinase C gamma (PKC gamma): function of neuron specific isotype. J Biochem 2002; 132: 683–687. doi: 10.1093/oxfordjournals.jbchem.a003274.
• 37)   Sample  V,  Ramamurthy  S,  Gorshkov  K, et al. Polarized activities of AMPK and BRSK in primary hippocampal neurons. Mol Biol Cell 2015; 26: 1935–1946. doi: 10.1091/mbc.E14-02-0764.
• 38)   Yoshimura  Y,  Sogawa  Y,  Yamauchi  T. Protein phosphatase 1 is involved in the dissociation of Ca2+/calmodulin-dependent protein kinase II from postsynaptic densities. FEBS Lett 1999; 446: 239–242. doi: 10.1016/s0014-5793(99)00226-4.
• 39)   Colbran  RJ. Inactivation of Ca2+/calmodulin-dependent protein kinase II by basal autophosphorylation. J Biol Chem 1993; 268: 7163–7170. https://pubmed.ncbi.nlm.nih.gov/8385100/ (accessed 23 January 2025).
• 40)   Bennecib  M,  Gong  CX,  Grundke-Iqbal  I, et al. Inhibition of PP-2A upregulates CaMKII in rat forebrain and induces hyperphosphorylation of tau at Ser 262/356. FEBS Lett 2001; 490: 15–22. doi: 10.1016/s0014-5793(01)02127-5.