Altered Expression Profile of Phosphatidylinositols in Erythrocytes of Alzheimer’s Disease and Amnestic Mild Cognitive Impairment Patients

Abstract

Background:

Studies have demonstrated that the levels of phospholipids, including phosphatidylinositols (PIs), were decreased in Alzheimer’s disease (AD) brain, presenting as a potential biomarker for AD. The plasma phospholipids levels have also been discovered to predict the conversion of cognitively normal elderly adults to amnestic mild cognitive impairment (aMCI) or demented patients.

Objective:

To investigate the expression profile of PIs in erythrocytes of AD and aMCI patients, which would serve as a blood-based method to distinguish AD and aMCI patients from normal controls (NC).

Methods:

In this study, we used anion-exchange high-performance liquid chromatography to analyze PIs alterations in erythrocytes from a total of 86 prospectively recruited subjects (including 24 NC, 21 aMCI patients, and 41 AD patients).

Results:

We found that the levels of PI40 : 4, PI3/5P, and PI(3,4)P2 in aMCI patients, and the levels of PI4P, PI(3,4)P2, and PI3/5P in AD patients were significantly decreased compared to NC. The changed expression profile of PIs could effectively discriminate AD and aMCI patients from NC (AUC = 0.964, 0.938, respectively).

Conclusion:

The altered expression profile of erythrocytes PIs might be a potential blood-based biomarker for AD and aMCI. This alteration of PIs probably reflected the impaired deformability and oxygen-carrying capacity of erythrocytes in AD and aMCI patients.

Keywords

Alzheimer’s disease biomarker erythrocytes mild cognitive impairment phosphatidylinositols

INTRODUCTION

Alzheimer’s disease (AD) is clinically characterized by the progressive memory loss along with impaired social and occupational function. Amyloid plaques and neurofibrillary tangles are the main pathological changes in AD brain. In the past decades, extensive efforts have been devoted to identifying biomarkers that can assist diagnosis and monitor disease severity. Considerable advances in AD diagnostics tools have been obtained. AD cerebrospinal fluid (CSF) profile (decreased amyloid-β (Aβ)₄₂ together with increased P-tau), volumetric magnetic resonance imaging (MRI), and positron emission tomography (PET), constitute the current main biomarker framework [1]. However, brain imaging, especially PET, is costly and only available in a very limited number of hospitals. Lumbar puncture, on the other hand, is invasive with relatively low acceptability among Chinese patients. Therefore, blood-based biomarkers, which are easily obtained, may be a more attractive alternative option.

Phosphatidylinositols (PIs) are crucial components of cell membrane. Skeletal structure of PI contains an inositol ring and a glycerol backbone with two fatty acids. Following the recruitment of certain phosphoinositide kinases and phosphatases, different positions of the inositol ring can undergo reversible phosphorylation, which turns PIs into a family with seven members, including PI4P, PI(4,5)P2, PI(3,4,5)P3, PI(3,4)P2, PI(3,5)P2, PI3P, and PI5P [2]. PIs, PIPs family, and the involved enzymes have pivotal roles in synaptic modification and neurotransmission in brain [3, 4]. A recent study revealed that PI is 1 of the 10 serum lipids that can predict cognitively normal older adults converting to amnestic mild cognitive impairment (aMCI) or AD patients with over 90% accuracy [5]. However, it is not known whether the PIs were changed in blood cells, since PI is a minor component locating on the cell membranes.

With the aim of finding out whether PIs or PIPs would change in blood cells of AD and aMCI patients, we utilized anion-exchange high-performance liquid chromatography (HPLC) to detect PIs metabolites of erythrocytes in AD, aMCI patients, and normal controls (NC). We set up a profile of PIs metabolites, which may serve as a blood-based method to distinguish AD and aMCI from NC.

METHODS

Subjects

We analyzed data from our prospectively collected subjects including 41 AD patients, 21 aMCI patients, and 24 NC. All AD and aMCI patients were enrolled from the Department of Neurology, the Second Affiliated Hospital, Zhejiang University School of Medicine from September 2015 to January 2017. Healthy control subjects were age-matched individuals from patients’ caregivers without consanguinity. They had Clinical Dementia Rating (CDR) scores of 0 and Mini-Mental State Examination (MMSE) scores between 24 and 30, without significant memory complaints.

All AD patients were diagnosed as probable AD according to the Diagnostic and Statistical Manual of Mental Disorders-IV (DSM-IV) [6] and the guidelines of National Institute for Neurological Diseases and Stroke/Alzheimer’s Disease and Related Disorders Association (NINCDS-ADRDA) [7]. All AD patients included had CDR scores of 1 or above. MMSE scores≤24 for patients with education level of junior school or above,≤20 for education level of primary school, and≤17 for illiteracy, were employed for the study [8].

The aMCI patients were diagnosed based on Petersen’s criteria [9]: 1) memory complaint, 2) normal activities of daily living, 3) normal general cognitive function, 4) abnormal memory for age, and 5) not demented. They had a CDR score of 0.5 and a minimal MMSE score of 24.

All patients were screened by two experienced neurologists (YC and BZ) and underwent appropriate evaluation procedure. Brain MRI and blood tests were carried out to exclude other causes of cognitive deficits. None of them were found to have other neurologic or psychological diseases. The control subjects were also evaluated by detailed examinations to exclude cognitive dysfunction or other neuropsychological diseases. We exclude subjects with 1) metabolic diseases like diabetes, obesity (body mass index (BMI)>28) or marasmus (BMI < 18), hyperlipidemia, arthrolithiasis; 2) consumptive disease such as malignancy, tuberculosis, hyperthyroidism, hepatitis, and colitis. All subjects were firstly diagnosed, denying receiving any acetylcholinesterase inhibitors, memantine, antipsychotics, or any concomitant medication that can influence plasma lipid levels. The blood samples were collected during their first visit.

This study was approved by the Research Ethics Committee, the Second Affiliated Hospital, Zhejiang University School of Medicine, China. Written informed consent was obtained from each participant or their care givers.

Blood collection and pretreatment

Fasting venous blood was collected in ethylenediaminetetraacetic acid (EDTA) anticoagulant tubes during morning hours in all cases. Erythrocytes separation was carried out within 1 h after blood collection as follows: 1) 50μl blood was centrifuged at 500 g for 10 min at 4°C; 2) After removing the plasma and buffy coat, the RBC pellet was washed 3 times by re-suspending the cells in 500μl D-Hank’s solution (NaCl 8.01 g/L, KCl 0.4 g/L, Na2HPO4•H 2O 0.06 g/L, NaHCO3 0.35 g/L, KH2PO4 0.06 g/L, Glucose 0.34 g/L, ddH2O 1000 ml, pH 7.2–7.4) and centrifuging at 500 g for 10 min at 4°C; 3) A total of 1×10⁹ red blood cells (RBC) pellet was flash frozen in dry ice and deposited at –80°C until use.

Apolipoprotein E genotyping

APOE genotype was measured by using DNA isolated from peripheral blood according to the standard phenol/chloroform extraction method. The PCR reactions and recycling program was performed according to a previously described method [10].

Phosphatidylinositol extraction procedure

All experiments were carried out under acidic conditions to extract PIs and PIPs efficiently [11]. After being ground with 10 volumes of extraction solvent (ice-cold chloroform: Methanol: Concentrated HCl = 100 : 100 : 0.7), the samples were centrifuged, then supernatant was separated. The second extraction was performed with 4 volumes of the extractant and re-centrifuged to recover the supernatant. Two supernatant fractions were mixed-up with 2 volumes of 0.6 N HCl, centrifuged, and the upper phase was discarded. The lower phase was washed 3 times with 4 volumes of washing solvent (chloroform: Methanol:0.6 N HCl = 3 : 48 : 47), dried. The extract was stored at –80°C until further analysis.

Ion chromatography analysis

PIs and PIPs was quantified by Dionex Ion Chromatography-5000 (Dionex, Sunnyvale, CA, USA) [11]. The extract mentioned above was dissolved with 0.5 mL methylamine reagent (methanol: 40% methylamine aqueous solution: 1-butanol aqueous solution (47 : 36 : 9:8)) and deacetylated at 50°C for 45 min. The aqueous phase was dried and re-dissolved in 0.5 mL 1-butanol: Petroleum ether: Ethyl formate (20 : 40 : 1), and then extracted twice with an equal volume of water. The aqueous phase extracts were dried again and re-dissolved in water for anion-exchange high-performance liquid chromatography (HPLC) on an Ionpac AS11-HC column (Dionex, Sunnyvale, CA, USA).

Statistical analysis

The data obtained from chromatographic analysis was imported into R 3.5 software. Pearson’s χ² assessed differences in the distributions of genders and APOE ɛ4 incidence. If meeting the assumptions of normality, one-way ANOVA was used to assess the difference among the three classifications (AD, aMCI, and NC), followed by Bonferroni test. For those who do not meet the normal distribution, Kruskal-Wallis test was performed followed by Dunn’s test for pairwise comparison. Principal component analysis (PCA) was a dimension-reduction algorithm in an unsupervised way to reveal any clustering of the three study groups. The orthogonal partial least squares discriminant analysis (OPLS-DA) was utilized to establish a predictive model between patients and healthy controls (AD versus NC; MCI versus NC) in a supervised manner. Significance was established by p < 0.05.

RESULTS

The clinical characteristics of all participants were listed in Table 1. No significant differences in age, gender, years of education, and serum lipid levels were found among the three groups. As expected, patients of AD and aMCI showed lower MMSE scores when compared to NC (p < 0.001). The APO E ɛ4 genotype prevalence was higher in AD and aMCI patients (p = 0.02). Significant differences were found across the three clinical groups for mean corpuscular hemoglobin concentration (MCHC) (p = 0.004).

Table 1

Characteristics and blood tests of subjects

	AD (n = 41)	aMCI (n = 21)	NC (n = 24)	p
Age (mean years±SD)	68.44±9.82	67.81±8.04	64.46±8.87	0.230
Education (mean years±SD)	8.39±4.26	8.29±5.10	7.88±4.38	0.903
Gender (male/female)	23/13	9/12	11/13	0.547
MMSE (mean scores±SD)	11.8±5.29	25.14.19±1.74	28.17±2.01	<0.001^*
APOEɛ4 carrier (%)	17.4	10.5	2.3	0.02^*
Cholesterol (mean mmol/L±SD)	4.71±0.83	4.69±0.65	5.40±0.91	0.554
Triglyceride (mean mmol/L±SD)	1.23±0.64	1.85±0.77	1.70±0.88	0.059
HDL (mean mmol/L±SD)	1.44±0.46	1.23±0.33	1.43±0.33	0.483
LDL (mean mmol/L±SD)	2.37±0.52	2.44±0.58	2.47±0.70	0.898
RCC (mean×10¹²/L±SD))	4.31±0.52	4.53±0.52	4.59±0.45	0.075
HGB (mean g/L±SD)	130.90±17.03	137.10±15.01	138.42±13.7	0.127
HCT (mean % ±SD)	40.74±5.35	42.43±4.77	41.80±4.26	0.413
MCV (mean fl±SD)	93.34±7.21	91.30±3.33	91.15±4.22	0.233
MCH (mean pg±SD)	30.37±2.66	29.98±1.06	30.22±1.64	0.118
MCHC (mean g/L±SD)	324.37±9.46	328.95±8.14	331.42±6.29	0.004*
RDW (mean % ±SD)	13.00±1.10	12.65±0.60	13.15±0.87	0.166

AD, Alzheimer’s disease; aMCI, amnestic mild cognitive impairment; MMSE, Mini-Mental State Examination; HDL, high density lipoprotein; LDL, low density lipoprotein; SD, standard deviation; RCC, red cell count; HGB hemoglobin; HCT, hematocrit; MCV, mean corpuscular volume; MCH, mean corpuscular hemoglobin; MCHC, mean corpuscular hemoglobin concentration; RDW, red cell distribution width; ^*p < 0.05.

A total of 22 PIs, PIPs, and (Lysophosphatidyl inositol) LPI were identified in erythrocytes by HPLC (Supplementary Table 1). To have a more comprehensive view of the expression profiles, PCA was conducted. No clear separation was seen among three groups. The PCA score plots showed Dim 1 = 23.4% and Dim 2 = 16.6% (Fig. 1). The contributions of variables to Dim-1 and Dim-2 were shown in Supplementary Figure 1.

Fig.1

PCA score plot shows the mild separation in AD, aMCI, and Control group. All samples are labeled by serial number.

Further analysis with OPLS-DA clearly distinguished AD from NC (R²X: 0.401[cum], R²Y: 0.611[cum] and Q²Y: 0.34[cum]; Fig. 2A), as well as MCI from NC (R²X: 0.297[cum], R²Y: 0.553[cum] and Q²Y: 0.327[cum]; Fig. 2B). However, OPLS-DA analysis could not distinguish MCI from AD. Permutation tests were conducted to validate the models (Supplementary Figure 2).

Fig.2

OPLS-DA score plot shows a clear discrimination between AD and NC (A), as well as between aMCI and NC (B). All samples are labeled by serial number.

The PIs and PIPs levels in aMCI and AD patients that were significantly different from those in NC were shown in spot plots (Fig. 3). Compared to NC, PI40 : 4 (PI40 : 4 refers to a phosphatidylinositol species with 40 carbons and 4 double bonds), PI3/5P (In ion-chromatography, the peaks for PI3P and PI5P cannot be separated, so we denote results as a mixture of PI3P and/or PI5P, i.e., PI3/5P), PI(3,4)P2 were significantly decreased in MCI (p = 0.021, p = 0.008, and p = 0.003, respectively), while PI4P, PI3/5P, and PI(3,4)P2 and were significantly decreased in AD patients (p = 0.002, p < 0.001, and p < 0.001, respectively). No significant differences in PI or PIPs levels were found between APOE ɛ4 carrier and non-carrier, except for PI4P. Lower PI4P level was found in APOE ɛ4 carriers (p = 0.04, Supplementary Table 2).

Fig.3

PIs and PIPs were decreased in erythrocytes from aMCI and AD patients. PI40 : 4 (A), PI(3,4)P2 (C), and PI3/5P (D) were significantly lower in aMCI compared to NC (p = 0.021, p = 0.003, and p = 0.008, respectively). PI4P (B), PI(3,4)P2 (C), and PI3/5P (D) were significantly lower in AD compared to NC (p = 0.002, p < 0.001, and p < 0.001, respectively). *p < 0.05 versus normal control.

Receiver operating characteristic (ROC) analysis was performed to evaluate the predictive power of the algorithm (Fig. 4). An area under the curve (AUC) is 0.964 for AD versus NC, with the sensitivity of 93% and specificity of 88%. For aMCI versus NC, the AUC is 0.938 with the sensitivity of 83% and specificity of 90%.

Fig.4

ROC curves of the composite biomarker panel for AD and aMCI. A) Diagnosis accuracy of the OPLS-DA algorithm to distinguish AD from NC: 0.946. B) Diagnosis accuracy of the OPLS-DA algorithm to distinguish aMCI from NC: 0.938.

DISCUSSION

Abnormal lipids metabolism has been widely reported in AD [12 –14]. In 1987, Stokes et al. discovered significantly decreased concentration of PIs in the hippocampus of AD brain compared to the control [15]. Considering the rapid postmortem breakdown of PIs and PIPs, further studies focusing on PI kinases were conducted. Decreased PI kinases activities confirmed the alteration of PI metabolism in AD brains [16 –18]. A wealth of evidence suggests that disturbed PIPs levels can affect amyloid protein processing and probably promote amyloid plaque formation in AD brain [19 –21]. However, studies focusing on the peripheral levels of PIs in AD or aMCI patients are very limited.

In the present study, we identified a changed profile of PIs in erythrocytes of AD and aMCI patients. As far as we know, this is the first study utilizing the HPLC method to detect the expression profile of PIs in erythrocytes, which may serve as a potential peripheral biomarker for AD/aMCI. We paid special attention to the distribution changes of these molecules, not to one single PI, because PIs are in dynamic changes. OPLS-DA models were built to specifically distinguish AD/aMCI from NC, indicating that the expression profiles of the PIs in erythrocytes reflect the specific changes in AD and aMCI patients. We found the levels of PI40 : 4, PI3/5P, and PI(3,4)P2 were decreased in aMCI patients, while AD group showed lower levels of PI4P, PI(3,4)P2, and PI3/5P compared to NC. It has been found that the changes in serum PI started 2–3 years before the onset of dementia symptoms [5]. We also observed decreased levels of PIs in erythrocytes of aMCI patients, indicating that disturbed levels of PI in erythrocytes occurred in the early stage of the disease, even in prodromal phase. However, no obvious differences of PI were observed between aMCI and AD group, indicating that the expression of PI in erythrocytes may not reflect the severity of the disease.

Erythrocytes are the most important medium for the transportation of oxygen in vertebrates. Mammalian erythrocytes are remarkably flexible and deformable so as to squeeze through tiny capillaries and release their oxygen load efficiently. PI and PIPs family constitutes to 2%–5% of membrane phospholipids [22]. PIs of erythrocytes participate in regulating Ca^2 + concentration in cytoplasm, mediating membrane fusion events, maintaining discoid morphology and so on [23]. The putative role of PI40 : 4 in erythrocytes is unknown and needs further explore. In this study, we found decreased level of PI4P in erythrocytes from AD and aMCI patients, which may be associated with the impaired erythrocytes deformability. The rapid turnover of PI4P and PI(4,5)P2 groups has been demonstrated in erythrocytes [24]. This conversion was associated with the shape control of erythrocytes [25]. Microscopic morphological analysis showed that over 15% erythrocytes from AD patients were elongated as compared to 5.9% in control erythrocytes [26]. The morphological changes of erythrocytes in AD lead to a decrease in the excess surface area and deficiency in erythrocytes deformability. Aβ deposition in erythrocytes has been generally demonstrated [27]. The binding of Aβ to erythrocytes can lead to an increased oxidative stress, and subsequently, compromised deformability of erythrocytes [28]. Besides, Aβ oxidizes red cell hemoglobin and contributes to high risk of anemia in AD patients [29]. Our study also observed a significantly decreased MCHC and a descending trend of hemoglobin in AD patients.

2,3-Bisphosphoglycerate (2,3-BPG) is one of the crucial metabolites that regulates the ability of hemoglobin binding to oxygen [30]. When 2,3-BPG binds to deoxyhemoglobin, it stabilizes the spatial conformation of deoxyhemoglobin, thus reducing the affinity of hemoglobin to oxygen and promoting the dissociation oxygen from hemoglobin to tissues. In erythrocytes from AD patients, the concentration of 2,3-BPG is decreased [31], resulting in increased hemoglobin affinity to oxygen and inefficient oxygen release. 2,3-BPG is a structural analog of inositol triphosphate (IP₃) and is involved in the PIs cascade [32], which may also affect the deformability of erythrocytes and their capacity of releasing oxygen. Thus, the altered PI and PIPs in erythrocytes of AD/aMCI patients might be associated with the reduced cell surface area, impaired deformability and increased affinity for oxygen, resulting in deficient oxygenation delivery and hypoxia status in brain tissue, and finally impaired cognitive function. In the future studies, it is necessary to simultaneously detect the levels of 2,3-BPG, PIPs and oxygen carrying ability of erythrocytes to confirm their relationships.

There are several limitations of our study. First, all subjects enrolled in this study were clinically diagnosed without the acquisition of PET or CSF biomarkers. Because PET is expensive and lumber puncture is an invasive procedure, the acceptance of which are very low in China. Second, the sample size is not large enough which may have an impact on the power of test. However, this was a pilot study, the sample size was relatively small. Third, we did not validate the results in another cohort, and this study was conducted at one single dementia clinic. Therefore, further studies conducted in multicenter cohorts with lager sample size and biomarkers (PET or CSF biomarkers) confirmed cases are needed to validate our results.

In conclusion, we employed HPLC technique to measure PIs levels in erythrocytes. We identified changed expression profiles of PIs in erythrocytes from AD and aMCI patients, which may distinguish AD and aMCI patients from healthy elderly. However, this finding requires verification in a larger patient cohort and the role of PIs in the occurrence and development of AD needs further investigation.

Footnotes

ACKNOWLEDGMENTS

This study was supported by the National Natural Science Foundation of China Major International (Regional) Cooperation and Exchange Projects (81520108010), National Natural Science Foundation of China (81870826), and Zhejiang Provincial Natural Science Foundation of China (LY18H090004 and LY17H090003).

Authors’ disclosures available online ().

The supplementary material is available in the electronic version of this article: .

References

Silverberg

, Elliott

, Ryan

, Masliah

, Hodes

(2018) NIA commentary on the NIA-AA Research Framework: Towards a biological definition of Alzheimer’s disease. Alzheimers Dement 14, 576–578.

Balla

(2013) Phosphoinositides: Tiny lipids with giant impact on cell regulation. Physiol Rev 93, 1019–1137.

Ueda

(2014) The role of phosphoinositides in synapse function. Mol Neurobiol 50, 821–838.

Frere

, Chang-Ileto

, Di Paolo

(2012) Role of phosphoinositides at the neuronal synapse. Subcell Biochem 59, 131–175.

Mapstone

, Cheema

, Fiandaca

, Zhong

, Mhyre

, MacArthur

, Hall

, Fisher

, Peterson

, Haley

, Nazar

, Rich

, Berlau

, Peltz

, Tan

, Kawas

, Federoff

(2014) Plasma phospholipids identify antecedent memory impairment in older adults. Nat Med 20, 415–418.

American Psychiatric Association (1994) Committee on nomenclature and Statistics:Diagnostic and Statistical Manual of mental Disorders Fourth Edition (DSM-IV), American Psychiatric Association.

Dubois

, Feldman

, Jacova

, Dekosky

, Barberger-Gateau

, Cummings

, Delacourte

, Galasko

, Gauthier

, Jicha

, Meguro

, O’Brien

, Pasquier

, Robert

, Rossor

, Salloway

, Stern

, Visser

, Scheltens

(2007) Research criteria for the diagnosis of Alzheimer’s disease: Revising the NINCDS-ADRDA criteria. Lancet Neurol 6, 734–746.

Guo

, Liu

, Wong

, Liao

, Yan

, Lin

, Chang

, Hsu

(1988) Chinese version and norms of the Mini-Mental State Examination. J Rehabil Med Assoc 16, e59.

Petersen

, Smith

, Waring

, Ivnik

, Tangalos

, Kokmen

(1999) Mild cognitive impairment: Clinical characterization and outcome. Arch Neurol 56, 303–308.

10.

Zivelin

, Rosenberg

, Peretz

, Amit

, Kornbrot

, Seligsohn

(1997) Improved method for genotyping apolipoprotein E polymorphisms by a PCR-based assay simultaneously utilizing two distinct restriction enzymes. Clin Chem 43, 1657–1659.

11.

Kanehara

, Yu

, Cho

, Cheong

, Torta

, Shui

, Wenk

, Nakamura

(2015) Arabidopsis AtPLC2 is a primary phosphoinositide-specific phospholipase C in phosphoinositide metabolism and the endoplasmic reticulum stress response. PLoS Genet 11, e1005511.

12.

Pettegrew

, Panchalingam

, Hamilton

, McClure

(2001) Brain membrane phospholipid alterations in Alzheimer’s disease. Neurochem Res 26, 771–782.

13.

, Huang

, Li

, Gong

C-X

, Schuchman

(2010) Deregulation of sphingolipid metabolism in Alzheimer’s disease. Neurobiol Aging 31, 398–408.

14.

Wong

, Braidy

, Poljak

, Pickford

, Thambisetty

, Sachdev

(2017) Dysregulation of lipids in Alzheimer’s disease and their role as potential biomarkers. Alzheimers Dement 13, 810–827.

15.

Stokes

, Hawthorne

(1987) Reduced phosphoinositide concentrations in anterior temporal cortex of Alzheimer-diseased brains. J Neurochem 48, 1018–1021.

16.

Jolles

, Bothmer

, Markerink

, Ravid

(1993) Reduced phosphatidylinositol kinase activity in Alzheimer’s disease: Effects of age and onset. Dementia 4, 81–86.

17.

Jolles

, Bothmer

, Markerink

, Ravid

(1992) Phosphatidylinositol kinase is reduced in Alzheimer’s disease. J Neurochem 58, 2326–2329.

18.

Zubenko

, Stiffler

, Hughes

, Martinez

(1999) Reductions in brain phosphatidylinositol kinase activities in Alzheimer’s disease. Biol Psychiatry 45, 731–736.

19.

Waugh

(2015) PIPs in neurological diseases. Biochim Biophys Acta 1851, 1066–1082.

20.

McLaurin

, Franklin

, Chakrabartty

, Fraser

(1998) Phosphatidylinositol and inositol involvement in alzheimer amyloid-beta fibril growth and arrest. J Mol Biol 278, 183–194.

21.

Morel

, Chamoun

, Lasiecka

, Chan

, Williamson

, Vetanovetz

, Dall’Armi

, Simoes

, Point Du Jour

, McCabe

, Small

, Di Paolo

(2013) Phosphatidylinositol-3-phosphate regulates sorting and processing of amyloid precursor protein through the endosomal system. Nat Commun 4, 2250.

22.

Minetti

, Low

(1997) Erythrocyte signal transduction pathways and their possible functions. Curr Opin Hematol 4, 116–121.

23.

Allan

(1982) Inositol lipids and membrane function in erythrocytes. Cell Calcium 3, 451–465.

24.

Muller

, Hegewald

, Jaroszewicz

, Cumme

, Hoppe

, Frunder

(1986) Turnover of phosphomonoester groups and compartmentation of polyphosphoinositides in human erythrocytes. Biochem J 235, 775–783.

25.

Ferrell

Jr. , Huestis

(1984) Phosphoinositide metabolism and the morphology of human erythrocytes. J Cell Biol 98, 1992–1998.

26.

Mohanty

, Eckley

, Williamson

, Launer

, Rifkind

(2008) Do red blood cell-beta-amyloid interactions alter oxygen delivery in Alzheimer’s disease? Adv Exp Med Biol 614, 29–35.

27.

Kiko

, Nakagawa

, Satoh

, Tsuduki

, Furukawa

, Arai

, Miyazawa

(2012) Amyloid beta levels in human red blood cells. PLoS One 7, e49620.

28.

Jayakumar

, Kusiak

, Chrest

, Demehin

, Murali

, Wersto

, Nagababu

, Ravi

, Rifkind

(2003) Red cell perturbations by amyloid beta-protein. Biochim Biophys Acta 1622, 20–28.

29.

Faux

, Rembach

, Wiley

, Ellis

, Ames

, Fowler

, Martins

, Pertile

, Rumble

, Trounson

, Masters

, The

ARG

, Bush

(2014) An anemia of Alzheimer’s disease. Mol Psychiatry 19, 1227.

30.

Barvitenko

, Adragna

, Weber

(2005) Erythrocyte signal transduction pathways, their oxygenation dependence and functional significance. Cell Physiol Biochem 15, 1–18.

31.

Kosenko

, Aliev

, Kaminsky

(2016) Relationship between chronic disturbance of 2,3-diphosphoglycerate metabolism in erythrocytes and Alzheimer disease. CNS Neurol Disord Drug Targets 15, 113–123.

32.

Guillemette

, Favreau

, Lamontagne

, Boulay

(1990) 2,3-Diphosphoglycerate is a nonselective inhibitor of inositol 1,4,5-trisphosphate action and metabolism. Eur J Pharmacol 188, 251–260.