Introduction

Altered lipid metabolism is believed to be a key event to central nervous system (CNS) injuries [1, 2]. In fact, phospholipid bilayer and associated lipids provide not only a permeability barrier but also a structured environment that is essential for the proper functioning of membrane-bound proteins [3]. One of the most important regulators of lipid organization is cholesterol. Its structure allows filling interstitial spaces between hydrophobic fatty acid chains of phospholipids [3].

Alterations in membrane metabolism has been extensively described in Alzheimer’s disease (AD) mostly those caused by decrease in phospholipase A2 (PLA2) activity, the main enzymes accountable for membrane phospholipids metabolism [4,5,6]. Furthermore, disturbances in cholesterol metabolism have been associated with the major pathological signatures of AD, i.e. extracellular deposition of Aβ in senile plaques, intracellular accumulation of hyperphosphorylated aggregated Tau in neurofibrillary tangles and degeneration of neurons and their synapses [7, 8].

The major pool of cholesterol in the central nervous system (CNS) is in myelin although neurons and glial cells also contain cholesterol. Brain cells synthesize all cholesterol found in CNS, since the blood–brain barrier (BBB) hinders entry of cholesterol synthesized in the periphery [9]. To maintain brain cholesterol homeostasis, it is converted into a more polar metabolite, the 24(S)-hydroxycholesterol (24OHC) which can be driven through the BBB [10,11,12,13]. 24OHC is an endogenous regulator of the nuclear receptor Liver X Receptor (LXR) potentially regulating the cholesterol and fatty acid synthesis pathways in the brain. In the adult brain almost all the neuronal requirements of cholesterol are supplied by cholesterol transported by ApoE-containing lipoproteins released from astrocytes. Via LXR, 24OHC regulates the expression, synthesis and secretion of ApoE [14] and the expression of the sterol transporters ATP-binding cassette A1, G1 and G4, involved in the transporter of cholesterol from glia to ApoE particle [15]. Therefore, 24OHC participates in both cholesterol efflux pathways. Plasma levels are kept constant by liver clearance of 24OHC converted into bile acids or further conjugated for final elimination. Thus, all 24OHC present in blood plasma arises from CNS, reflecting the brain cholesterol catabolism [15].

In the periphery, the major cholesterol metabolite is 27-hydroxycholesterol (27OHC), representing an important mechanism for the daily elimination of cholesterol from the body. It is also able to cross the BBB [9, 16].

It is worth stressing the importance of ApoE in lipid homeostasis: it regulates cholesterol, triglyceride and phospholipid transport and metabolism [17]. APOE gene has three common alleles ε2, ε3 and ε4 and gives rise to three main structural isoforms protein that differ in their effectiveness: ApoE2, ApoE3 and ApoE4. 65–80% of AD patients carry at least one of ε4 [18]. The presence of this allele leads to a less effective protein highlighting 24OHC role in cholesterol efflux [19,20,21].

In fact several studies have already described a decrease in 24OHC and an increase of 27OHC in AD [22] in serum [23, 24] and CSF [25, 26], as a reflection of disease burden, the loss of metabolically active neurons and the degree of structural atrophy [23, 25, 27,28,29,30,31,32,33,34]. Findings from cell culture and animal studies suggest that 27OHC is associated with increased levels of APP as well as β-secretase (BACE1), the enzyme that cleaves APP to yield Aβ. On the other hand, 24OHC is associated with increased levels of sAPPα, suggesting that 24OHC facilitates the processing of APP by non-amyloidogenic pathway [35,36,37,38].

Donepezil hydrochloride, a selective acetylcholinesterase inhibitor (AChEi), is one of the widely used drugs in AD treatment [39,40,41]. These drugs delay the progression of brain atrophy, indicating a disease-modifying effect by attenuating neuronal death and a neuroprotective role and seem to affect cholesterol levels [42].

Since study from our group shows a significant effect of donepezil on PLA2 activity, increasing membrane phospholipid metabolism [4], the aim of this study was to evaluate if short- and long-term donepezil treatment also affects lipid metabolism in a broader way by assessing cholesterol and metabolites 24OHC and 27OHC in plasma of AD patients and in healthy volunteers. Moreover, we investigated a possible role for APOE genotypes on cholesterol metabolism.

Materials and methods

The sample comprised 30 patients with mild or moderate AD and 33 physically and cognitively healthy elderly controls. This open-label study was conducted at the Institute of Psychiatry, University of Sao Paulo, Brazil and participants were middle-income, community-dwelling individuals from the hospital catchment area. Local Ethical Committee approved this study, and all subjects signed a written informed consent prior to inclusion in the study. The diagnosis of probable AD was made according to the NINCDS–ADRDA diagnostic criteria [43]. Cognitive assessments were made with the Cambridge Cognitive Test (CAMCOG) [44] and the Mini-Mental State Examination (MMSE) [45] (Table 1). Table 2 shows the genotype frequencies of ApoE. In the last 6 months prior to enrolment in the study, patients and the control group did not use AChEi or other drugs (e.g. neuroleptics, antidiabetic, lithium, antidepressants and anticonvulsants). After the completion of baseline assessments, all patients were started on donepezil treatment (5 mg/day). Doses were administered orally and increased to 10 mg/day after 1 month. Patients were monitored monthly, the MMSE was evaluated after 3 and 6 months, and CAMCOG was evaluated 6 months after donepezil treatment.

Table 1 Sociodemographic characteristics of patients and controls at baseline assessment and MMSE and CAMCOG score diagnosis
Full size table
Table 2 Genotype frequencies of apolipoprotein E (ApoE)
Full size table

Sample collection

Blood samples were collected, in the morning after 8-h fasting, using Vacutainer tubes containing EDTA (Vacutainer, Becton Dickinson). Samples were centrifuged at 20 °C and 1400g for 10 min. Plasma was obtained, frozen and stored at −80 °C until analysis.

Plasma cholesterol concentrations were measured using commercially available enzymatic–colorimetric methods (Labtest Diagnostica, Brazil).

Measurements of oxysterols (24-hydroxycholesterol and 27-hydroxycholesterol) were performed according to Dzeletovic et al. [46]. Following is a brief description of the method.

Sample preparation

To plasma (1 mL) was added a mixture of deuterium-labelled internal standards followed by an alkaline hydrolysis with potassium hydroxide (0.4 M). After 2 h of incubation, the pH was adjusted to 7 with phosphoric acid. Extraction was performed with the addition of 20 mL of chloroform and 6 mL of 0.9 M NaCl followed by vigorous shaking. The organic phase was evaporated using a rotary evaporator at room temperature. The residue was resuspended with 1 mL of toluene. The next step was a solid-phase extraction using a silica cartridge (Supelclean LC-Si SPE tube) and isopropanol as eluent solution. The collected isopropanol was evaporated and the residue was treated with pyridine:trimethysilyl trifluoroacetamide (1:1, v/v) at 60 °C for 30 min.

Analysis by gas chromatography–mass spectrometry

The samples (1 μL) were injected in split mode into a gas chromatograph (GC) coupled to a mass spectrometer (MS) (Shimadzu GCMS-QP2010 Plus) equipped with the GCMS solution software version 2.5. The separation was performed with a Restek capillary column (30 m × 0.25 mm, 0.25 μm phase thickness). The oven temperature program was as follows: 240 °C for 7 min, 5 °C/min to 290 °C. Helium was used as carrier gas. The gas chromatograph was operated in constant flow mode with the flow rate set to 44.1 cm/s. The mass spectrometer was operated in single ion monitoring and the electron ionization energy was 70 eV. The ion source and interface temperature was 300 °C. [46].

APOE genotyping

DNA was extracted from peripheral blood by salting out [47]. Two single nucleotide polymorphisms (SNPs) (rs7412 and rs429358) were evaluated to determine the APOE genotype as previously described [48]. Amplification reactions were performed using real-time polymerase chain reaction (PCR) SNP genotyping systems (TaqMan Assays; Life Technologies, Carlsbad, CA, USA), as follows: TaqMan PCR Mastermix 1×/μL, TaqMan SNP genotyping assay 1×/μL, 10 ng/μL genomic DNA and ultrapure water to make up a volume of 7 μL were mixed in each well of an optical plate. Allelic discrimination was evaluated in a 7500 real-time PCR system (Life Technologies) comparing fluorescence levels before and after amplification (45 cycles of 15 s at 95 °C and 1 min at 60 °C).

Statistical analysis

Baseline comparison of sociodemographic characteristics of patients and controls were assessed with Fisher’s exact test for categorical variables and with Student’s t test for numerical ones. Longitudinal analysis of patients was carried out with a linear mixed-effect model (24OHC, 27OHC, cholesterol, APOE, MMSE and CAMCOG). Pairwise comparison of means by time was corrected using sequential Bonferroni method. Normality assumptions were checked inspecting the QQ plots. All statistical analyses were performed with the SPSS (Statistical Package for Social Sciences, for Windows, v. 14, Chicago, IL) and significance level was p < 0.05.

Results

At baseline, we found a decrease of plasma 24OHC (p = 0.003) in probable AD patients. No differences were observed regarding 27OHC and cholesterol levels at baseline comparing to controls. After 3 months with 10-mg donepezil treatment we did not find any change in cholesterol or 24OHC, which remained lower than controls but 27OHC levels presented a slight increase (p = 0.023). After 6 months of treatment we found a discrete increment in cholesterol levels (198.66 ± 54.49 mg/dL) (p = 0.04), but no change in 24OHC. 27OHC restored the baseline levels (Table 3; Fig. 1).

Table 3 Oxysterol and cholesterol levels according to diagnostic groups at baseline and post-treatment
Full size table
Fig. 1
figure 1

Cholesterol and oxysterols pre- and post-donepezil treatment

Full size image

The presence of an ApoE4 allele did not show an association with cholesterol (p = 0.051), 24OHC (p = 0.262) or 27OHC (p = 0.167). Education years and cognitive assessments (MMSE and CAMCOG) also show no association with the metabolites. In our samples there were no differences between probable AD and controls regarding the presence of APOE4 allele.

After 6 months on donepezil, 46% of the patients remained cognitively stable or slightly improved on CAMCOG (mean improvement 7.1 ± 3.5 points), while the remaining patients deteriorated (4.7 ± 3.2). These results did not show any association with the metabolite levels.

Discussion and conclusions

In the present sample, we found reduced plasma levels of 24OHC in probable AD patients, whereas no differences were found regarding 27OHC and total cholesterol levels when compared to controls.

Decreased 24OHC levels in AD were reported in several studies [49,50,51,52,53]. In fact, considerable evidences suggest a relationship between the degree of brain atrophy and the cholesterol metabolite 24OHC in plasma level of AD patients and in other neurodegenerative diseases [28, 54]. Famer and colleagues [36] addressed in vitro the effects of 24OHC on the activity of α- and β-secretases, the two key enzymes involved in APP metabolism. They found the α-secretase activity and the relative protein values of extracellular sAPPα and total sAPP were significantly increased, whereas the β-secretase activity was reduced by 24OHC, demonstrating that this oxysterol tends toward APP α-cleavage [36].

However, others studies report that plasma levels of 24OHC in AD patients were not significantly different or increased comparing to control [34, 55, 56]. Some authors suggest that demyelination may cause a transient increase in the flux of 24OHC from the brain to the blood. Conversely, a destruction of neurons in combination with a loss of myelin occurring during a long period would be expected to result in decreased circulating levels of 24OHC at a later stage of the disease [57]. Thus, we assume that at least part of the contradictory findings regarding 24OHC levels in probable AD may be explained by sample heterogeneity regarding the different stages of the disease.

Regarding 27OHC, some authors have reported that this oxysterol increases levels of Aβ and p-tau and induces endoplasmic reticulum stress, a cellular response that is implicated in AD [58]. Based on this, it is expected to find a decrease in plasma 27OHC and an increase in AD patients’ brain. In fact, brain levels of 27OHC were described to be increased in several areas of autopsied AD brains [57]. Likewise, Kolsch and colleagues [30] found significantly reduced 27OHC in plasma of patients with dementing disorders compared to non-demented subjects [17]. Despite all, our study found no differences in 27OHC plasma levels between AD patients and controls.

Many studies have shown that changes in membrane cholesterol levels influence APP processing [59]. Cholesterol in the brain is mainly located in neuronal cell membranes and myelin sheets. This lipid is essential for cell membrane structure and function, particularly in lipid rafts, specialized membrane microdomains where the last step in Aβ production [60] occurs. On the periphery, elevated cholesterol has been proposed to be a risk factor to developing AD by animal models and in vitro studies [61, 62]. However, in human studies, the results of cholesterol levels are controversial, both on the periphery and in the brain [63]. In our study, cholesterol levels at baseline were no different in plasma of probable AD patients as compared to controls.

APOE*4 allele was found to be associated with lowest levels of 24OHC, according to the hypothesis of an earlier faster and more intensive degenerative process in ε4 and the positive effect of ε4 allele on AD markers [64]. Despite this, the presence of an APOE4 allele did not show an association with cholesterol and oxysterol plasma levels in this study.

Regarding treatment with donepezil, we did not find an effect of 3- and 6-month treatment on 24OHC plasma levels. With respect to 27OHC, we observed a slight increase with 3 months of treatment, but this increment was not maintained during the 6 months of donepezil treatment. Numerous studies have revealed that AChEI could regulate cholesterol level through activating the LXR pathway [65]. LXRs act as a dominant supervisor in cholesterol metabolism including synthesis, uptake and trafficking. In studies with NPC mice, donepezil delayed the loss of cerebellar Purkinje cells and improved motor function via regulating the cholesterol-related homeostasis. These data suggest that donepezil reduced the cholesterol accumulation via inhibition of cholesterol synthesis through upregulation of the LXR pathway in NPC mice [66]. Contrasting, in our study we observed a significant increase in cholesterol levels as a result of donepezil treatment. Nonetheless, this increment of total cholesterol levels should not be overemphasized, as the mean values remained within the normal range (<200 mg/dL). It is, however, of interest that Adunsky and colleagues [67], in a cross-sectional study, found increased plasma cholesterol in AD patients using donepezil compared with AD patients who did not.

More studies are necessary in this field to confirm the effect of donepezil on cholesterol and metabolites. A relevant factor, to be taken into account, is the small size of our sample, follow-up length and the lack of lumbar punctures. Despite this, it is a valuable sample, since patients were free of AChEi treatment and solely one variable was introduced in the follow-up: donepezil treatment. However, these results confirm and extend previous studies demonstrating disturbed cholesterol turnover in Alzheimer’s disease.