Study Design and Participants
The study included 18 participants between 33 and 76 years old, selected from a project about new biomarkers of insulin resistance (34). Exclusion criteria were known diabetes mellitus or use of anti-diabetic medications, other endocrine disorders, diseases of the exocrine pancreas, pregnancy or use of oral anticoagulants or lipid-lowering drugs. Individuals who had symptoms of an acute viral or bacterial infection, or with plasma high-sensitivity C-reactive protein (hsCRP) concentrations higher than 10 mg/L, were also excluded. Blood samples were drawn in EDTA (ethylenediaminetetraacetic acid) tubes after an 8-hour fast. Plasma was promptly separated, supplemented with a preserving cocktail (benzamidine, phenylmethylsulphonyl fluoride and gentamicin), aliquoted and stored at -80°C for later analyses. Plasma concentrations of apoA-I were measured using a nephelometric method (35) and fasting plasma glucose (FPG), plasma lipids and creatinine using conventional colorimetric assays (Biosystems, Barcelona, Spain). Glycated hemoglobin A1c (HbA1c) was determined using a National Glycohemoglobin Standardization Program-certified boronate affinity technique (NycoCardTM Reader II, Alere Technologies, Oslo, Norway).
HDL isolation and separation
Plasma was passed through Acrodisc® 5µm filters (Pall Corporation, Port Washington, NY, USA), in order to remove fibrin impurities. Then, HDL was purified from plasma using immunoaffinity chromatography as follows: Sepharose 4B™ resin bound to goat polyclonal anti-human apoA-I antibodies (Academy Bio-Medical, Houston, Texas, USA) was loaded into 10mL Poly-Prep® chromatography columns (Bio-Rad, Hercules, California, USA), and 1 mL of plasma was incubated overnight in the column. The unbound fraction was collected by gravity flow and stored at -80°C. The bound fraction was eluted using 3 consecutive washes of 3M sodium thiocyanate (NaSCN) and one last phosphate-buffered saline (PBS) wash followed by concentration and desalting in 10 KDa molecular weight cutoff Amicon® filters (Merck Millipore, Billerica, MA, USA) until 1 mL of fraction was achieved. Then two PBS washes were performed to remove any NaSCN left over in the column before another use. The same process was carried out with columns containing goat polyclonal anti-human apoC-III and anti-human apo-E antibodies, in that order. At the end, total HDL from each participant was divided in 4 subfractions: HDL without apoE or apoC-III (E-C-), HDL with apoE but without apoC-III (E+C-), HDL without apoE but with apoC-III (E-C+) and HDL with both apoE and apoC-III (E+C+). The efficiency of the immunoaffinity columns was 95-98% for all study subjects. The final buffer for all HDL subfractions was PBS.
Determination of enzymatic content and activity in HDL subfractions
LCAT concentration was determined using an immunoenzymatic, double-sandwich assay (ALPCO Diagnostics, Salem, NH, USA), in which capture is performed by a first monoclonal antibody against LCAT (MoAb 36486) and detection is performed with a different, horseradish peroxidase (HRP)-labeled anti-LCAT monoclonal antibody (MoAb 36487). After incubation with a substrate solution and termination with a stop reagent, the intensity of absorbance at 492 nm was read in a Synergy HT microplate reader using Gen5 software (BioTek, Winooski, VT, USA). CETP concentration was determined using an immunoenzymatic sandwich assay (ALPCO Diagnostics, Salem, NH, USA), with MoAb 3-11D as capture antibody and HRP-labeled MoAb 14-8F as detection antibody.
Measurement of LCAT activity was done using a fluorometric assay (Calbiochem, Darmstadt, Germany). The assay is based on the incubation of a substrate that fluoresces at 470 nm with the study samples. After LCAT in the samples removes a fatty acid from the substrate and transfers it to cholesterol, the substrate loses fluorescence at 470 nm and gains fluorescence at 390 nm. Thus, LCAT activity was measured as change in 470/390 emission intensity by comparison against a calibration curve based on dilutions of a plasma pool. CETP was also measured with a fluorometric assay (Abcam, Cambridge, UK). The assay principle involves incubating the sample with a mixture that contains a self-quenched fluorescent neutral lipid and an acceptor molecule. Action of CETP in the sample results in transfer of the neutral lipid to the acceptor molecule and an increase in its fluorescence (excitation at 465 nm, emission at 535 nm). For CETP too, activity was measured by comparison against a calibration curve based on dilutions of a plasma pool. Enzymatic activity was expressed as percent of activity in the plasma pool, and was therefore expressed in Arbitrary Units (AU). All measurements were performed in duplicate.
The cholesteryl ester content of plasma and HDL subfractions was measured using the Abcam cholesteryl ester kit (ab65359, Abcam, Cambridge, UK). The assay is based on two simultaneous reactions for cholesterol determination, one of which includes cholesterol esterase as part of the reaction mixture, while the other does not. Cholesteryl esters are then calculated by subtraction as (total cholesterol - free cholesterol). Phosphatidylcholine in plasma and HDL subfractions was measured using the Abcam fluorometric assay (ab83377, Abcam, Cambridge, UK). The test is based on an enzyme-coupled reaction that hydrolyzes phosphatidylcholine and releases choline, which in turn reacts with the OxiRed probe and generates fluorescence at 587nm wavelength.
Determination of HDL size distribution
HDL contained in each of the four apoC-III and apoE-defined fractions were separated by size using non-denaturing polyacrylamide gradient gel electrophoresis (NDPAGGE). Twenty-five microliters of each immunofraction plus 25 microliters of sample buffer were loaded into a 4-30% polyacrylamide gradient gel (Jule Inc., Milford, CT, USA). Wells 1 and 10 were loaded with molecular size standards (Amersham HMW Native Marker Kit, GE Healthcare, Little Chalfont, UK) and gels were run for 24 hours at constant 70V. Then, contents of the gel were transferred to a 0.45 micrometer pore size polyvinylidene fluoride (PVDF) membrane (Pall Corporation, New York, NY, USA) in a wet transfer apparatus at 30V for 24 hours. The lanes containing the MW markers were cut from the rest of the membrane, stained in 0.2% amido black solution for 20 minutes and stored for later photographing. The rest of the membrane was blocked with 5% powder low-fat milk, incubated with an HRP-conjugated goat anti-human apoA-I antibody (Academy Bio-medical, Houston, TX, USA), and revealed using 3,3′,5,5′-tetramethylbenzidine as substrate. Later, the marker lanes and the rest of the membrane were placed side to side and photographed in a Bio-Rad ChemiDocTM MP gel documenter. Using the molecular size standards as reference, the intensity of the bands/smears in each size range fractions was quantitated in Image LabTM software. Size fractions were defined as follows: prebeta HDL: <7.1 nm, alpha 3 HDL: 7.1-8.2 nm, alpha 2 HDL: 8.2-9.5 nm and alpha 1 HDL: 9.5-12.2 nm. The concentration of apoA-I in each HDL size subfraction was estimated by multiplying the proportion of apoA-I within that fraction by the directly measured total plasma apoA-I. Laboratory procedures were executed at the Diabetes, Lipids and Metabolism laboratory of Universidad de Los Andes, following current institutional biosafety protocols.
Statistical analyses
A sample size of eighteen participants provided 87% power to detect a true difference of at least 8% in LCAT activity between two HDL subtypes, assuming a 5% variation coefficient in LCAT activity (18), at a 5% significance level. The distribution of plasma apoA-I across HDL subfractions was compared in a 2-way analysis of variance (ANOVA) model in which apoA-I concentration in each subfraction was the dependent variable, and HDL size and immunofraction were fixed factors. Enzyme concentrations and activities were compared among HDL subtypes using a 1-way ANOVA in which HDL immunofraction was the only fixed factor. When global ANOVA was significant, post-hoc comparisons against the reference fraction (E-C-) were done using Scheffé´s method. Comparisons of numeric variables between groups of participants were performed using Mann-Whitney´s U test. The significance of differences in proportions between groups of participants was assessed using Fisher´s exact test. All tests were done at a 0.05 significance level and all reported p-values are 2-sided.