Decreased Levels of Blood AMPK α 1 but not AMPK α 2 Isoform in Patients with Mild Cognitive Impairment and Alzheimer’s Disease: A Pilot Study

Abstract

Background:

There is an urgent need to develop feasible biomarkers for diagnosis and prognosis of Alzheimer’s disease (AD). Mounting evidence implicates that dysregulation of energy metabolism is a key and early event in AD pathogenesis. AMP-activated protein kinase (AMPK) is a central molecular sensor that plays a critical role in maintaining cellular energy homeostasis, and aberrant brain AMPK activities are linked to AD pathophysiology.

Objective:

We aimed to investigated protein levels of AMPKα isoforms, AMPKα1 and AMPKα2, in plasma samples from patients clinically diagnosed with mild cognitive impairment (MCI) or AD, along with age-matched healthy controls.

Methods:

30 participants (10 MCI, 10 AD, and 10 controls) were included in our pilot study. Plasma levels of AMPKα1 and AMPKα2 were determined by ELISA. Receiver operating characteristic (ROC) analysis was used to assess sensitivity and specificity. Linear regression was used to assess the correlation between levels of AMPKα isoforms and other biomarkers.

Results:

Plasma levels of AMPKα1 were decreased in MCI and AD patients, while levels of AMPKα2 were unaltered as compared to controls. ROC analysis showed relatively high sensitivity and specificity for AMPKα1 to distinguish MCI and AD from controls. Linear regression analysis showed that plasma levels of AMPKα1 were correlated with a brain imaging biomarker (AD signature cortical thicknesses).

Conclusion:

Plasma levels of AMPKα1 were decreased in MCI and AD patients. Future endeavor to explore whether blood AMPKα1 protein expression has the value as a potential biomarker for AD and MCI diagnosis shall be encouraged.

Keywords

Alzheimer’s disease AMPK biomarker isoform mild cognitive impairment plasma

INTRODUCTION

Alzheimer’s disease (AD) is the most common dementia syndrome, and one of the leading causes of disability and casualty in elderly [1, 2]. Aging is the most known risk factor for AD. Concomitant with the aging population worldwide, the prevalence and incidence of AD have increased dramatically in recent years. It was predicted that there will be more than 13.8 million patients with AD in the United States alone by 2050 [3]. Currently neither cure nor effective intervention to slow the disease progress exists for AD. Multiple recently finished clinical trials for disease-modifying therapies, most targeting at the late stage AD, have not been successful [4, 5]. It has thus been proposed that early diagnosis and accordingly treatment at early state of the disease might be critical [5, 6]. To date there are no validated biomarkers in clinical practice for AD diagnosis in patients. Substantial evidence indicates that people with mild cognitive impairment (MCI) are more likely to develop AD [7]. Notably, many biomarkers that are currently recommended for diagnostic criteria (e.g., amyloid-β (Aβ)₄₂, total tau (t-tau), and phospho-tau (p-tau) in cerebrospinal fluid (CSF), Pittsburgh compound B imaging, and magnetic resonance imaging (MRI) [8]) are not so sensitive in patients with MCI as compared to their performance in AD patients [9].

Energy metabolic dysfunction is linked to multiple neurodegenerative diseases including AD [10 –12]. Disruption of energy metabolism takes place in the early stage of AD, thus molecules involved in signaling pathways controlling energy metabolism could potentially be used as biomarkers for the early diagnosis of AD [13, 14]. One of the key molecules that play an important role in metabolism is AMP-activated Protein Kinase (AMPK). AMPK is a master energy sensor of cells. It senses the cellular energy status and accordingly regulates catabolic and anabolic processes to adjust the energy metabolism to equilibrium [15, 16]. Moreover, AMPK functions as a nexus for several signaling pathways controlling protein synthesis, autophagy, and lipid metabolism; dysregulations of all these processes are implicated in AD pathogenesis [14 , 17–19]. Mounting evidence also indicates that AMPK plays an important role in generation of Aβ and tau phosphorylation, two hallmarks of AD brain pathology [20 –24]. Moreover, aberrant AMPK activities are linked to multiple AD pathophysiologies [14 , 26]. Structurally, mammalian AMPK is a heterotrimer composed of a catalytic subunit (α subunit) and two regulatory subunits (β and γ subunits). The α subunit has 2 isoforms, α1 and α2 [27]. AMPKα1 and α2 are encoded by different genes and co-expressed in most tissues to regulate energy metabolism. The specific targets and exact roles of the two AMPKα isoforms in the central nervous system have not been determined yet [14 , 28–30]. Notably, a recent study indicates critical roles of AMPKα isoforms, particularly AMPKα1, in AD pathophysiology [31]. Here, we examined protein levels of AMPKα isoforms in plasma samples from patients clinically diagnosed with MCI or AD, along with age-matched healthy controls.

MATERIALS AND METHODS

Subjects

All the subjects were enrolled from clinical studies that were conducted in Wake Forest Baptist hospital, North Carolina, USA. MCI and AD patients were diagnosed according to the recommendation from the National Institute on Aging-Alzheimer’s Association workgroups on diagnostic guidelines for AD [8]. Patients who had major diseases like cancer, stroke, heart disease, psychiatric disease, etc., were excluded from our study. Controls are age-matched elderly without complaints in memory loss and history of major diseases. Demographic information was collected by trained staffs. Neuropsychological assessment batteries, including Mini-Mental State Examination (MMSE), Logical Memory test II and Category Fluency test, were conducted by trained nurses. Fast bloods were drawn at the time of enrollment with anticoagulant and then centrifuged to separate plasma. Plasmas were stored at –80°C. CSFs were collected and levels of Aβ₄₂, tau, and p-tau in CSF were determined at the same time. MRI scanning was carried out to measure the thicknesses of related corti and volumes of hippocampus and entorhino cortex using the program FreeSurfer [32]. All the subjects and/or their guardians were fully informed of the study and signed the consent form.

Protocol

10 MCI patients, 10 AD patients and 10 controls were enrolled in our study. Levels of AMPKα1 and AMPKα2 in plasma were determined by ELISA kits following the manufacture’s instruction (AMPKα1 ELISA kits were purchased from abcam, Cambridge, MA, USA (ab181422), AMPKα2 ELISA kits were purchased from LSBio, Seattle, WA, USA (LS-F10706)). Briefly, plasma was diluted by 10 times (for AMPKα1) or 4 times (AMPKα2) with diluent provided by the kits for each sample. 50μl or 100μl of diluted plasma was added to the wells and incubated with antibody cocktails at 37°C for 1 or 2 h based on the protocols. Then liquid in each well was aspirated and washed with wash buffer for 3 times. After incubated with TMB substrate for 10 min, stop solution was added to each well. Plates were read by a microplate reader at 450 nm. All the samples were duplicated. Intra-assay coefficient of variability (CV) for AMPKα1 is 7.7%, intra-assay CV for AMPKα2 is 11.2%. Values of protein levels were analyzed by Grubbs’ test and outliers determined by statistical test were dropped from further analysis.

Ethics statement

This study was approved by the Institutional Review Board (IRB) at the Wake Forest University Health Sciences. All participants provided written informed consent.

Statistics

Category variables were presented as frequency (percentage) and continuous variables were presented as mean±standard deviation. Fisher exact test was performed for category variables. 1-way ANOVA with post-hoc Tukey test or Kruskal-Wallis test was performed for continuous variables based on the distribution of the variables if there were more than two groups and t test was performed for continuous variables if there were two groups. Grubbs’ test was performed to determine outlier. ROC analysis was performed to calculate AUC, cut-off point, sensitivity and specificity. Linear regression was performed to determine the correlation between levels of AMPKα and other biomarkers. Due to loss of follow-up, data of neuropsychological test, CSF biomarker and MRI scanning were incomplete. Those without the follow-up data were excluded from linear regression analysis. All the statistical analyses were conducted using GraphPad 5.0. p < 0.05 was considered significant.

RESULTS

Demographic and clinical characteristics of subjects

Demographic and clinical characteristics of those subjects are shown in Table 1. No differences were found in gender, age, year of education or diabetes incidence among groups. Apolipoprotein (APOE) ɛ4 frequency is significantly higher in AD patients. Cognitive status of subjects was evaluated by neuropsychological test batteries, including MMSE, Logical Memory test II, and Category Fluency test. Logical Memory test II is a subtest from Wechsler Memory Scale IV. Subjects are read a short passage and then asked to recall as many details as possible both immediately and after a 30 min delay [33]. In Category Fluency test, subjects are asked to produce as many words as possible from a category in 1 min [34]. Scores of MMSE and Category Fluency test are significantly lower in AD patients as compared to control groups. Performance of both MCI and AD patients in Logical Memory test II is significantly worse compared to control group. Aβ levels in CSF are decreased in MCI and AD patients, but not significantly lower compared to controls. p-tau and t-tau levels in CSF are significantly increased in AD patients, compared to control and MCI. AD signature cortical thickness (ADSCT), an unweighted average of thicknesses of cortex regions that specifically atrophy in AD [32], including bilateral entorhinal, inferior temporal, middle temporal, inferior parietal, fusiform, and precuneus regions, are not significantly different among groups. No difference is found between groups in hippocampus and entorhinal cortex volume that has been corrected to the whole cerebral volume. In brief, results from neuropsychological tests, CSF and imaging biomarkers assays are consistent with MCI and AD characteristics, though there are variations in some of the indices.

Table 1

Demographic and clinical characteristics of the subjects

	Control (n = 10)	MCI (n = 10)	AD (n = 10)	Statistics
Male (%)	50.0	50.0	50.0	N.S.
Age (y)	66.5±9.1	68.0±10.5	67.5±7.3	N.S.
Education (y)	16.0±2.3	14.6±2.5	14.1±2.0	N.S.
Diabetes (%)	70.0	80.0	40.0	N.S.
APOE ɛ4 (%)	22.2	5.6	40.0	p = 0.0467
MMSE	28.8±1.3	28.5±1.7	23.7±4.4	p = 0.0093
LMII	26.0±6.0	17.0±6.5	6.0±6.3	p < 0.0001
Category Fluency	26.1±3.8	20.3±5.0	15.4±6.2	p = 0.0013
Aβ₄₂ in CSF	543.0±121.5	511.7±193.2	389.0±175.6	N.S.
p-tau in CSF	33.3±32.2	29.1±10.6	70.8±35.5	p = 0.0065
t-tau in CSF	58.2±34.9	44.5±11.2	109.6±51.2	p = 0.0209
ADSCT	2.66±0.07	2.54±0.20	2.51±0.21	N.S.
HC Volume_corrected	0.0027±0.0002	0.0028±0.0005	0.0025±0.0005	N.S.
ERC Volume_corrected	0.0011±0.0001	0.0011±0.0002	0.0010±0.0002	N.S.

APOE, Apolipoprotein E; MMSE, Mini-Mental State Examination; LMII, Logical Memory test II; ADSCT, Alzheimer’s disease signature cortical thicknesses; HC, hippocampus; ERC, entorhinal cortex.

Expression levels of AMPKα isoforms in plasma

We performed ELISA to determine levels of AMPKα1 and AMPKα2 in the aforementioned groups. Results are shown in Fig. 1. Levels of AMPKα1 in plasma are significantly decreased in MCI patients as compared to those in control patients (Fig. 1A). There is also a clear trend of decrease in plasma levels of AMPKα1 in AD patients as compared to those in control group (p = 0.058) (Fig. 1B). When MCI and AD groups are combined, levels of AMPKα1 are significantly decreased compared to those in control group (Fig. 1C). We also measured plasma levels of AMPKα2 using ELISA, and did not observe significant differences between control and MCI or AD group (Fig. 1D–F).

Fig. 1

Levels of plasma AMPKα1 and AMPKα2 determined by ELISA in MCI, AD, and control subjects. A) Levels of AMPKα1 are significantly decreased in MCI patients (p = 0.0195). B) Levels of AMPKα1 show a trend of decrease in AD patients (p = 0.0580). C) Compared to controls, levels of AMPKα1 in MCI and AD combined group are significantly decreased (p = 0.0098). D–F) Levels of AMPKα2 are unaltered between control, MCI, AD groups (p > 0.05).

Receiver operating characteristic (ROC) curve analysis for plasma AMPKα1 expression

To evaluate the efficacy and determine cut-off point of AMPKα1 as a potential biomarker, we performed ROC analysis for plasma AMPKα1 expression in MCI and AD patients. ROC analysis can give corresponding sensitivity and specificity at a specific point and can determine the best cut-off point for a biomarker. The efficacy of a biomarker is represented by area under the curve (AUC). ROC analysis in MCI patients shows AUC is 0.8000 and the best cut-off point is 172.1 mg/ml with the corresponding sensitivity of 88.89% and specificity of 70.00% to differentiate MCI patients from controls (Fig. 2A). ROC analysis in AD patients shows AUC is 0.7444 and the best cut-off point is 193.2 mg/ml with the corresponding sensitivity of 77.78% and specificity of 70.00% to differentiate AD patients from controls (Fig. 2B). When MCI and AD patients are combined as one group, ROC analysis shows AUC is 0.7722 and the best cut-off point is 193.2 mg/ml with the corresponding sensitivity of 83.33% and specificity of 70.00% to differentiate patients from controls (Fig. 2C).

Fig. 2

Receiver operating characteristic (ROC) cure analysis for plasma AMPKα1 expression. A) ROC analysis in MCI patients. B) ROC analysis in AD patients. C) ROC analysis in MCI and AD patients combined.

Correlation analysis of plasma AMPKα1 expression levels with AD-related CSF biomarkers

CSF levels of Aβ₄₂, p-tau, and t-tau are established biomarkers for AD [9]. Levels of Aβ₄₂ are decreased while levels of p-tau and t-tau are increased in the CSF of AD patients. Here we explored the correlation between levels of AMPKα1 and these CSF biomarkers by performing linear regression analysis. Our analysis showed levels of AMPKα1 are not correlated with CSF levels of Aβ₄₂ (Fig. 3A, p > 0.05), p-tau (Fig. 3B, p > 0.05), or t-tau (Fig. 3C, p > 0.05).

Fig. 3

Correlation analysis between levels of AMPKα1 and AD-related CSF biomarkers. Levels of AMPKα1 are not correlated with CSF levels of Aβ₄₂ (A, p > 0.05), p-tau (B, p > 0.05), or t-tau (C, p > 0.05).

Correlation analysis of plasma AMPKα1 expression levels with measures from brain imaging studies

We also explored the correlation between levels of AMPKα1 and imaging biomarkers, including ADSCT, hippocampus volume, and entorhinal cortex volume, whose atrophy was relatively specific in AD patients. Hippocampus volume and entorhinal cortex volume are corrected to the whole cerebral volume. Our analysis revealed that levels of AMPKα1 are correlated with ADSCT (Fig. 4A, p < 0.05), but not correlated with hippocampus volume (Fig. 4B, p > 0.05) or entorhinal cortex volume (Fig. 4C, p > 0.05).

Fig. 4

Correlation analysis between levels of AMPKα1 and brain imaging biomarkers. Levels of AMPKα1 are correlated with AD signature cortical thicknesses (ADSCT) (A, p < 0.05), but not correlated with hippocampus (HC) volume (B, p > 0.05) or entorhinal cortex volume (ERC) (C, p > 0.05).

Correlation analysis of plasma AMPKα1 expression levels with performance in cognitive tests

Lastly, we explored the correlation between levels of AMPKα1 and performance on neuropsychological tests, including MMSE, logical memory test II and category fluency test. Our analysis showed that levels of AMPKα1 are not correlated with MMSE (Fig. 5A, p > 0.05), logical memory test II (Fig. 5B, p > 0.05), or category fluency test (Fig. 5C, p > 0.05).

Fig. 5

Correlation analysis between AMPKα1 levels and Neuropsychological test scores. Levels of AMPKα1 are not correlated with Mini-Mental State Examination (MMSE) (A, p > 0.05), logical memory test II (B, p > 0.05) or category fluency test (C, p > 0.05).

DISCUSSION

Early diagnosis AD remains a challenge, and many biomarkers that are intensively studied in late-stage of AD are not sensitive in early stage of AD or MCI [9]. For example, levels of Aβ₄₂, p-tau, and t-tau in CSF are not significantly altered in MCI patients compared to controls [9]. Meanwhile, most finished clinical trials were conducted in late-stage AD patients and have not been successful, raising the notion to intervene early on the disease process since the AD pathologies might be irreversible in its late stage [4]. Many even tried to define pre-clinical stage of AD, the stage before patients show any symptoms, and it is speculated that AD could only be cured in this stage [35]. These discussions highlight the importance of biomarkers that can define the early even pre-clinical stage of AD. Mounting evidences indicate that energy metabolic dysfunction might be an early and key event in AD pathogenesis, which makes those molecules that are involved in metabolic process such as AMPK attractive candidates as biomarkers for the early diagnosis of AD [14 , 36]. In the current study, we observed decreased levels of AMPKα1 isoform in plasma of MCI and AD patients compared to the healthy control group (Fig. 1). Moreover, biomarkers derived from blood are more accessible and have advantages over CSF or brain imaging biomarkers. Taken together, findings from the current pilot study provide insights into future investigations on whether plasma AMPKα1 might be a feasible biomarker for the early diagnosis of AD.

Previous studies demonstrated AD-related hyper-activities of AMPK (mainly assessed by AMPKα phosphorylation) [14 , 24]. Interestingly, we found increased expression levels of AMPKα1 in brain tissue from postmortem human AD and AD model mice, and brain-specific repression of AMPKα1 but not α2 isoform is able to alleviate cognitive deficits in aged AD model mice [31]. Future studies are warranted to elucidate the relationship between peripheral and brain AMPK isoform dysregulations and the underlying mechanisms. For example, how peripheral regulation of AMPK isoforms is related to AD pathology such as Aβ accumulation and tau phosphorylation in blood or CSF. Notably, here we show that the levels of plasma AMPKα1 do not correlated with CSF Aβ₄₂ or tau (Fig. 3). One possible explanation for such “dissociation” could be that measurement of CSF Aβ₄₂ or tau is not sensitive (as biomarkers) in MCI patients [9]. It is worth mentioning that we also tried examining levels of AMPKα isoforms in CSF samples from MCI/AD and control patients. Unfortunately, no CSF AMPKα1/α2 could be detected with the currently available ELISA kit.

There are also some limitations in our study. First, the number of subjects is relatively small for clinical studies. But even with 10 subjects in each group, levels of AMPKα1 in MCI patients are already significantly lower as compared to controls. In AD patients, the trend of decease is clear but not significant (p = 0.058), which might be attributed to the small sample size. In fact, when MCI and AD patients are combined, the significance of difference between patients and controls becomes even more distinct. Notably, carriers of APOE ɛ4, a well-known genetic risk factor for non-familial AD [3], are much lower in MCI group (5.6%) compared to control (22.2%) and AD (40%) groups. How/whether AMPKα dysregulation is linked to APOE ɛ4 status is unknown. Second, our data did not reveal a significant correlation between plasma AMPKα1 levels and cognitive performance or AD biomarkers (tau/Aβ). In the future, more comprehensive studies are required to determine whether reduction of plasma AMPKα1 represents more peripheral metabolism abnormality in AD/MCI. Finally, the sensitivity and specificity of AMPKα1 as a biomarker is not so high, which could also be due to the sample size. Nevertheless, considering the sensitivity of AMPKα1 in MCI is almost 90% and corresponding specificity is 70%, AMPKα1 in plasma is still a potential valuable biomarker candidate that deserves further study.

Footnotes

ACKNOWLEDGMENTS

We thank Ms. Karlyn Donohoe and Xueyan Zhou for administrative support with human samples. This work was supported by National Institutes of Health grants R01 AG055581, R01 AG056622 (T.M.), the Alzheimer’s Association grant NIRG-15-362799 (T.M.), the BrightFocus Foundation grant A2017457S (T.M.), Wake Forest Alzheimer’s Disease Research Center (ADRC) grant P30AG049638 (S.C.), Wake Forest Clinical and Translational Science Institute (CTSI) pilot grant (T.M.).

Authors’ disclosures available online ().

References

Mayeux

, Stern

(2012) Epidemiology of Alzheimer’s disease. Cold Spring Harb Perspect Med 2, a006239.

Lane

, Hardy

, Schott

(2018) Alzheimer’s disease. Eur J Neurol 25, 59–70.

Alzheimer’s Association (2019) Alzheimer’s disease facts and figures. Alzheimers Dement 15, 321–387.

Guell-Bosch

, Montoliu-Gaya

, Esquerda-Canals

, Villegas

(2016) Aβ immunotherapy for Alzheimer’s disease: Where are we? Neurodegener Dis Manag 6, 179–181.

Briggs

, Kennelly

, O’Neill

(2016) Drug treatments in Alzheimer’s disease. Clin Med 16, 247–253.

Piton

, Hirtz

, Desmetz

, Milhau

, Lajoix

, Bennys

, Lehmann

, Gabelle

(2018) Alzheimer’s disease: Advances in drug development. J Alzheimers Dis 65, 3–13.

Knopman

, Petersen

(2014) Mild cognitive impairment and mild dementia: A clinical perspective. Mayo Clin Proc 89, 1452–1459.

McKhann

, Knopman

, Chertkow

, Hyman

, Jack

, Kawas

, Klunk

, Koroshetz

, Manly

, Mayeux

, Mohs

, Morris

, Rossor

, Scheltens

, Carrillo

, Thies

, Weintraub

, Phelps

(2011) The diagnosis of dementia due to Alzheimer’s disease: Recommendations from the National Institute on Aging –Alzheimer’s Association Workgroups on diagnostic guidelines for Alzheimer’s disease. Alzheimers Dement 7, 263–269.

Olsson

, Lautner

, Andreasson

, Ohrfelt

, Portelius

, Bjerke

, Holtta

, Rosen

, Olsson

, Strobel

, Wu

, Dakin

, Petzold

, Blennow

, Zetterberg

(2016) CSF and blood biomarkers for the diagnosis of Alzheimer’s disease: A systematic review and meta-analysis. Lancet Neurol 15, 673–684.

10.

Yaffe

, Blackwell

, Kanaya

, Davidowitz

, Barrett-Connor

, Krueger

(2004) Diabetes, impaired fasting glucose, and development of cognitive impairment in older women. Neurology 63, 658–663.

11.

Lin

, Beal

(2006) Mitochondrial dysfunction and oxidative stress in neurodegenerative diseases. Nature 443, 787–795.

12.

, Klann

(2012) Amyloid β: Linking synaptic plasticity failure to memory disruption in Alzheimer’s disease. J Neurochem 120, 140–148.

13.

Pedros

, Petrov

, Allgaier

, Sureda

, Barroso

, Beas-Zarate

, Auladell

, Pallas

, Vazquez-Carrera

, Casadesus

, Folch

, Camins

(2014) Early alterations in energy metabolism in the hippocampus of App^swe/PS1^de9 mouse model of Alzheimer’s disease. Biochim Biophys Acta 1842, 1556–1566.

14.

Wang

, Zimmermann

, Ma

(2019) Therapeutic potential of AMP-activated protein kinase in Alzheimer’s disease. J Alzheimers Dis 68, 33–38.

15.

Hardie

, Carling

(1997) The AMP-activated protein kinase-fuel gauge of the mammalian cell? Eur J Biochem 246, 259–273.

16.

Hardie

(2014) AMPK-sensing energy while talking to other signaling pathways. Cell Metab 20, 939–952.

17.

Krishan

, Richardson

, Sahni

(2014) AMP kinase (PRKAA1). J Clin Pathol 67, 758–763.

18.

Salminen

, Kaarniranta

, Haapasalo

, Soininen

, Hiltunen

(2011) AMP-activated protein kinase: A potential player in Alzheimer’s disease. J Neurochem 118, 460–474.

19.

, Chen

, Vingtdeux

, Zhao

, Viollet

, Marambaud

, Klann

(2014) Inhibition of AMP-activated protein kinase signaling alleviates impairments in hippocampal synaptic plasticity induced by amyloid β. J Neurosci 34, 12230–12238.

20.

Thornton

, Bright

, Sastre

, Muckett

, Carling

(2011) AMP-activated protein kinase (AMPK) is a tau kinase, activated in response to amyloid β peptide exposure. Biochem J 434, 503–512.

21.

Vingtdeux

, Davies

, Dickson

, Marambaud

(2011) AMPK is abnormally activated in tangle- and pre-tangle-bearing neurons in Alzheimer’s disease and other tauopathies. Acta Neuropathol 121, 337–349.

22.

Domise

, Didier

, Marinangeli

, Zhao

, Chandakkar

, Buee

, Viollet

, Davies

, Marambaud

, Vingtdeux

(2016) AMP-activated protein kinase modulates tau phosphorylation and tau pathology in vivo . Sci Rep 6, 26758.

23.

Won

, Im

, Kim

, Singh

(2010) Involvement of AMP-activated protein kinase (AMPK) in neuronal amyloidogenesis. Biochem Biophys Res Commun 399, 487–491.

24.

Vingtdeux

, Giliberto

, Zhao

, Chandakkar

, Wu

, Simon

, Janle

, Lobo

, Ferruzzi

, Davies

, Marambaud

(2010) AMP-activated protein kinase signaling activation by resveratrol modulates amyloid-β peptide metabolism. J Biol Chem 285, 9100–9113.

25.

, Chai

, Zhao

, Li

, Zhang

, Liu

, Wu

, Liu

, Wang

, Zhou

(2015) AMPK activation ameliorates Alzheimer’s disease-like pathology and spatial memory impairment in a streptozotocin-induced Alzheimer’s disease model in rats. J Alzheimers Dis 43, 775–784.

26.

Wang

, Zhang

, Liang

, Xu

, Xie

, Liu

, Viollet

, Yan

, Zou

(2010) AMPKa2 deletion causes aberrant expression and activation of NAD(P)H oxidase and consequent endothelial dysfunction in vivo: Role of 26S proteasomes. Circ Res 106, 1117–1128.

27.

Carling

, Mayer

, Sanders

, Gamblin

(2011) AMP-activated protein kinase: Nature’s energy sensor. Nat Chem Biol 7, 512–518.

28.

Michell

, Stapleton

, Mitchelhill

, House

, Katsis

, Witters

, Kemp

(1996) Isoform-specific purification and substrate specificity of the 5’-AMP-activated protein kinase. J Biol Chem 271, 28445–28450.

29.

Turnley

, Stapleton

, Mann

, Witters

, Kemp

, Bartlett

(1999) Cellular distribution and developmental expression of AMP-activated protein kinase isoforms in mouse central nervous system. J Neurochem 72, 1707–1716.

30.

Steinberg

, Kemp

(2009) AMPK in health and disease. Physiol Rev 89, 1025–1078.

31.

Zimmermann

, Yang

, Kasica

, Zhou

, Wang

, Beckelman

, Lee

, Furdui

, Keene

, Ma

(2020) Brain-specific repression of AMPKα1 alleviates pathophysiology in Alzheimer’s model mice. J Clin Invest, doi: 10.1172/JCI133982.

32.

Schwarz

, Gunter

, Wiste

, Przybelski

, Weigand

, Ward

, Senjem

, Vemuri

, Murray

, Dickson

, Parisi

, Kantarci

, Weiner

, Petersen

, Jack

(2016) A large-scale comparison of cortical thickness and volume methods for measuring Alzheimer’s disease severity. Neuroimage Clin 11, 802–812.

33.

Wechsler

, Holdnack

, Drozdick

(2009) Wechsler Memory Scale: Fourth Edition. Technical and Interpretive Manual. Pearson, San Antonio.

34.

Lezak

(1995) Neuropsychological Assessment. Oxford University Press.

35.

Sperling

, Aisen

, Beckett

, Bennett

, Craft

, Fagan

, Iwatsubo

, Jack

, Kaye

, Montine

, Park

, Reiman

, Rowe

, Siemers

, Stern

, Yaffe

, Carrillo

, Thies

, Morrison-Bogorad

, Wagster

, Phelps

(2011) Toward defining the preclinical stages of Alzheimer’s disease: Recommendation from the national institute of aging –Alzheimer Association Workgroups on diagnostic guidelines for Alzheimer’s disease. Alzheimers Dement 7, 280–292.

36.

Kandimalla

, Thirumala

, Reddy

(2017) Is Alzheimer’s disease a Type 3 diabetes? A critical appraisal. Biochim Biophys Acta Mol Basis Dis 1863, 1078–1089.