The level of 24-Hydroxycholesteryl Esters is an Early Marker of Alzheimer’s Disease

Materials

24OH-C and (25,26,26,26,27,27,27-²H)-24OH-C (here mentioned as ²H-24OH-C) were purchased from Avanti Polar Lipids (Alabaster, AL, USA). Human serum albumin (HSA), rabbit anti-human albumin IgG, goat anti-rabbit horseradish peroxidase-conjugated IgG (GAR-HRP), and chemicals of the highest purity were purchased from Sigma-Aldrich (St. Louis, MO, USA). Columns for solid phase extraction (SPE; MFC18 ec, 200 mg) were from Macherey-Nagel (Duren, Germany). Organic solvents were purchased from Romil (Cambridge, UK). Dye reagent for protein titration was from Bio-Rad (Hercules, CA, USA). The columns Luna C18(2)-HST (250×2.1 mm, 5 μm) and BEH C18 (250×0.075 mm, 2.5 μm) for UHPLC were from Phenomenex (Torrance, CA, USA) and Waters (Milford, MA, USA), respectively.

Study participants

CSF and homologous serum samples were collected at the Neurology Clinic of the Second University of Naples from AD patients (n = 13) or subjects without any sign of neurological pathology as controls (n = 8), comparable for age and sex to patients (age, mean±SD: CTR = 51.25±15.05 years; AD = 63.08±11.81 years, p > 0.05; % Female: CTR = 75%; AD = 54%, p > 0.05). Clear and colorless CSF was obtained by lumbar puncture, and processed for routine analysis according to current guidelines [21]. Controls were subjected to lumbar puncture for anesthetic interventions or for clinical analyses, that provided negative results for any neurological diagnosis. The integrity of BBB was evaluated by measuring the concentration of albumin in the CSF and homologous serum, as described below, and by calculating the ratio between the albumin concentration in the CSF and in the blood (Q_alb) [22]. Albumin concentration was 213±83 μg/mL in CSF and 51±3 mg/mL in serum. Subjects with Q_alb > 0.008 were excluded. Further microscopic evaluation revealed no red blood cells in CSF.

Three groups of subjects were selected for plasma study: 30 AD patients, 74 subjects with mild cognitive impairment (MCI), and 30 individuals with normal cognitive function (controls). The general characteristics of the study participants are depicted in Table 1. Patients underwent clinical and neurological examination at the Memory Clinic of the IRCCS Fatebenefratelli, Brescia, Italy. All AD patients met the criteria for probable AD [2, 23]. MCI subjects met the Petersen criteria [24]. All MCI subjects were selected based on the presence of follow-up evaluation (range: 1–6 years). Thirty-four at follow-up presented with AD (MCI conv-AD) and forty did not have dementia (MCI non-conv). The severity of the dementia was assessed by the Mini-Mental State Examination (MMSE). Biologic samples, isolated according to standard procedures, were stored at Fatebenefratelli Biobank (Brescia, Italy). APOE genotyping was carried out by PCR and enzyme digestion [25]. Patients signed an informed consent (approval No. 26/2014). The study was approved by the Local Ethical committee (approval No. 22/2016).

Titration of albumin

Albumin concentration was measured by ELISA. Samples were diluted (CSF, 1:100, 1:200, 1:400, 1:800; serum 1:50000, 1:100000, 1:300000, 1:600000, 1:1000000) with coating buffer (7 mM Na₂CO₃, 17 mM NaHCO₃, 1.5 mM NaN₃, pH 9.6), and aliquots (50 μl) were then incubated in a microtiter plate overnight at 4°C. After four washes by T-TBS (130 mM NaCl, 20 mM Tris-HCl, 0.05% Tween 20, pH 7.4) and four by high salt TBS (500 mM NaCl in 20 mM Tris-HCl at pH 7.4), wells were blocked with T-TBS containing 5% non-fat milk (2 h, 37°C). After washing, the wells were incubated (1 h, 37°C) with rabbit anti-human Albumin (1:1500 dilution in T-TBS supplemented with 0.25% non-fat milk) followed by GAR-HRP IgG (1:9000 dilution). Peroxidase-catalyzed color development from o-phenylenediamine was measured at 492 nm. A calibration curve was obtained using 50, 20, 10, 5, 2, 1, and 0.5 ng commercial human albumin. The experimental data were obtained in the TRASE laboratory.

Analysis of 24OH-CE levels

Each sample (1 mL) was supplemented by ²H-24OH-C as internal standard, and then divided into two equal aliquots of 0.5 mL. The internal standard was 5 or 200 ng/mL in CSF or blood samples, respectively. EDTA and butylated hydroxytoluene were added to each aliquot at 80 μM and 0.5 mg/mL final concentration, respectively. One aliquot (namely S) was processed by alkaline hydrolysis with three volumes of 1 M KOH in ethanol, whereas the other one (namely NS) was mixed with three volumes of ethanol. After 1 h at 37°C, KOH was neutralized by 50% phosphoric acid until pH 7, and lipids recovered from both aliquots by liquid-liquid and liquid-solid extraction. In detail, six sequential 1 mL extracts in hexane were obtained from each aliquot, the extracts were pooled and dried by centrifugation under vacuum. The lipid residue was dissolved in 50% ethanol, and loaded on SPE column equilibrated in 50% ethanol. The column was washed by 3 mL of 50% ethanol, the oxysterol fraction was eluted by 3 mL of 70% ethanol, and dried by centrifugation under vacuum. The oxysterols from samples were dissolved in 50 μL of acetonitrile and fractionated by UHPLC (Accela 1250; Thermo Fisher Scientific, Cambridge, MA, USA) with the column Luna C18(2)-HST operating at 0.3 mL/min in mixtures of 0.1% formic acid (solution A) and 0.1% formic acid in acetonitrile (solution B) as follows: 75% B from 0 to 7 min, linear gradient (75 to 90% B) from 7 to 8 min, 90% B from 8 to 15 min (Fig. 1). The eluate was analyzed on a Orbitrap Elite mass spectrometer equipped by an electrospray ion source (Thermo Fisher Scientific; Cambridge, MA, USA) in SIM and in MRM mode, monitoring ion at m/z 385.29 or transition 385–367 for 24OH-C (Fig. 1) and ion at m/z 391.28 or transition 391–373 for the internal standard ²H-24OH-C. For the oxysterols from CSF samples, chromatographic separation was performed by a nanoUHPLC system (nanoAquity, Waters, Milford, MA, USA) on the BEH C18 capillary column operating at 300 nL/min. For these samples, the elution of oxysterols was achieved using 70% B from 0 to 5 min, then linear gradient from 70% to 90% B in 5 min, and finally 90% B for 10 min. MS analyses were carried out on the same spectrometer described above, using a nanospray ion source. The amounts of 24OH-C from the aliquots S and NS represented the total (i.e., esterified plus unesterified forms) level and the level of unesterified forms, respectively, in the CSF or plasma sample (Fig. 2). The level of 24OH-CE was calculated as the difference between the 24OH-C levels in aliquots S and NS, and expressed as percentage of total 24OH-C. This value was used as index of LCAT-mediated conversion of 24OH-C into esters forms.

Statistics

In all the experiments, the data were expressed as means±SD from three independent experiments. The SPSS 20.0 software for Windows (IBM) was used for statistical analysis. The Kolmogorov-Smirnov test was performed in all continuous variables to define the presence of normality. One way ANOVA with post-hoc test with Sidak correction was used for three group comparisons of normally distributed continuous variables. The Kruskal Wallis test and pair wise comparison with Bonferroni correction was used for three group comparisons of skewed variables. Two tailed p value equal or less than 0.05 was considered statistically significant. Disease free curves (Kaplan Meier statistics) were used to compare the age at disease onset by levels of % 24OH-CE.

Correlation between 24OH-CE levels in CSF and in serum

The concentration of unesterified 24OH-C (NS) and total 24OH-C (sum of unesterified 24OH-C with 24OH-CE, i.e., samples S) were measured in CSF and homologous serum from a pilot cohort of AD patients (n = 13) and controls (n = 8). Total 24OH-C was higher in AD than in control CSF (4.10±1.03 versus 1.60±0.59 ng/mL respectively, p < 0.0001, Mann-Whitney test) whereas levels in serum were not significantly different between controls and AD (56.94±18.15 versus 45.33±10.78 ng/mL, respectively, p > 0.05) (data not shown). To measure the LCAT-mediated conversion of 24OH-C into esters, the amount of 24OH-CE was then calculated and expressed in terms of percentage of esters in the total 24OH-C population (% 24OH-CE). Values of % 24OH-CE were significantly lower in AD than in control both in CSF (37.12±10.19% versus 68.01±3.22%; p < 0.001, Mann-Whitney test), as well as in serum (44.40±7.36% versus 76.88±4.74 %; p < 0.001, Mann-Whitney test) (Fig. 3A, B). Measures of these dispersions, expressed as inter-quartile ranges (IQR), resulted to be 34.41, 26.11, 9.31, and 15.73% for AD CSF, AD serum, control CSF, and control serum, respectively. Good correlation between the % 24OH-CE in CSF and homologous serum was found for controls (r = 0.8816, p = 0.0038) or AD patients (r = 0.8622, p = 0.0028) when analyzed separately, and a very good correlation resulted when the samples were analyzed as a whole (r = 0.9707, p = 1.1×10^;- 10) (Fig. 3C).

Decreased levels of 24OH-CE in AD and MCI subjects

The positive correlation found between 24OH-CE levels in CSF and serum suggested that disease-associated decrease of LCAT-mediated conversion of 24OH-C into esters forms in the brain might be detected by analyzing the 24OH-CE level in the blood. Thus, we further analyzed LCAT-mediated conversion of 24OH-C into esters in plasma from four groups: AD (n = 30), controls (n = 30), and n = 74 MCI, both MCI conv-AD (n = 34) and MCI non-conv (n = 40). The characteristics of the participants are depicted in Table 1. The study groups did not differ by age, education, age at onset of memory impairment; female % was higher in AD and controls as compared to MCI groups. As expected, there were significant intergroup differences in MMSE scores and the prevalence of APOE ɛ4 allele carriers was significantly higher in AD patients and in MCI conv-AD than in the other study groups. There was a significant difference in plasma % of 24OH-CE among the four study groups (Table 1, Fig. 4A).

As expected from preliminary results, we found a decrease in % 24OH-CE levels in AD with respect to controls (Fig. 4A, p < 0.001, post-hoc Dunn’s test). The ROC analysis fixed the best result for the discrimination between AD and controls at the % 24OH-CE cutoff level of 47.8: values≤47.8 gave a specificity of 96.7% and a sensitivity of 81.8% (data not shown). As a whole, the MCI subjects group had intermediate % 24OH-CE (median, range: 52.66, 20.80–92.20) that was significantly different from AD (p < 0.001, post-hoc Dunn’s test) and from controls (p = 0.038, post-hoc Dunn’s test) (data not shown). Comparing the two MCI subgroups, we found that the % of 24OH-CE was lower in MCI conv-AD as compared to MCI-nc (p < 0.001, post-hoc Dunn’s test, Fig. 4A). The ROC analysis fixed the best result for the discrimination between AD and MCI conv-AD from controls and MCI-nc at the % 24OH-CE cutoff level of 55.7: values≤55.7 gave a specificity of 90.0% and a sensitivity of 73.0% (data not shown). In order to evaluate the influence of the % 24OH-CE in plasma on disease onset and progression, we generated Kaplan-Meier survival curves depicting age at onset (in AD patients) or age at clinical conversion (in MCI subjects) stratified on the basis of their % 24OH-CE value in plasma. We observed a significant anticipation of disease onset (68.88±1.11 years as compared to 77.08±0.90 years) in subjects with a % of 24OH-CE below the median value, 55.7 (Breslow test p < 0.001) (Fig. 4B).