Influence of high density lipoprotein cholesterol levels on circulating monocytic angiogenic cells functions in individuals with type 2 diabetes mellitus

A total of 119 men and women with type 2 diabetes for more than 12 months and on stable anti-diabetic treatment regimen (oral anti-diabetic medications and/or long acting or short acting insulin) for the prior 3 months were consecutively recruited at the Diabetes Unit of the Department of Clinical and Experimental Medicine of the University of Pisa. Pregnant or breastfeeding women, individuals of non European descent, subjects with type 1 diabetes, those with advanced, sight-threatening diabetic retinopathy or chronic kidney disease (CKD) stages ≥ 3b, as well those with poor glycemic control, defined as HbA1c ≥ 9.0% as measured at the enrollment visit, were excluded. Further exclusion criteria were body mass index (BMI) ≥ 35 kg/m², severe uncontrolled hypertension defined as systolic blood pressure (BP) ≥ 180 mmHg and/or diastolic BP ≥ 110 mmHg, sign of acute illness or infection, significant hepatic disease, chronic inflammatory disease, immunological and myeloproliferative diseases or other malignancy, and any major cardiovascular event within 3 months of entering in the study. All patients underwent a structured interview in order to collect information about the onset of diabetes, its duration, smoking habits, current anti-diabetic treatment, and BP- and lipid-lowering therapies [22]. Body weight and height were assessed and BMI calculated; waist circumference was measured at midway between the costal margins and the iliac crests. BP was measured after 5-min rest while seated and the average of two consecutive measurements obtained about 5-min apart was calculated. Hypertension was defined as systolic BP > 140 mmHg and/or diastolic BP > 80 mmHg and/or use of any antihypertensive medication.

In all subjects a venous blood specimen was drawn after an overnight fast for determination of HbA1c (high-performance liquid chromatography using DCCT-aligned methods) [23], serum creatinine, glucose, total-, HDL-, and LDL-cholesterol (Friedewald formula), triacylglycerol, alanine aminotransferase (ALT), aspartate aminotransferase (AST), gamma-glutamyltransferase (GGT), uric acid, fibrinogen and blood cell count using standard laboratory methods. A first-voided urine sample was also collected for determination albumin (BNII; Dade Behring Diagnostic, Marburg, Germany) and creatinine (modified Jaffé reaction). Urinary samples with abnormal sediments on routine analysis were excluded. Finally, all patients underwent a screening for diabetic complications as detailed elsewhere [22]. Briefly, diabetic retinopathy was assessed by retinal photography and its severity classified according to the Global Diabetic Retinopathy Project Group criteria [24]; patients with advanced, sight-threatening retinopathy (severe non-proliferative, proliferative, maculopathy or blindness) were excluded. Based on ACR (albumin-to-creatinine ratio) values, UAE (urinary albumin excretion) categories were defined as normoalbuminuria (< 30 mg/g, < 3.4 mg/mmol), microalbuminuria or “moderately increased albuminuria” (30–299 mg/g, 3.4–34 mg/mmol) and macroalbuminuria or “severely increased albuminuria” (≥ 300 mg/g, ≥ 34 mg/mmol). Estimated glomerular filtration rate (eGFR) was calculated by the CKD-EPI (Chronic Kidney Disease Epidemiology Collaboration) equation [25] and subjects with eGFR < 45 ml/min/1.73 m² (CKD stages ≥ 3b) were excluded. Diabetic neuropathy was assessed by a validated questionnaire, by knee and ankle reflexes and by measurement of vibration perception threshold (VPT) [22]. Finally, previous cardiovascular diseases (CVD) were determined based on medical history by recording all documented major acute CVD events and revascularization procedures. Peripheral vascular disease was also assessed by search of femoral and foot pulses, and measurement of ankle/brachial pressure ratio [22].

Patients were divided in subjects with low [≤ 40/50 mg/dl (≤ 1.034/1.293 mmol/l) in men/women] and high HDL-C levels [> 40/50 mg/dl (> 1.034/1.293 mmol/l) in men/women]. The main characteristics of the two study groups are given in Table 1. In all these subjects, a 50 ml fasting blood sample was drawn in EDTA-containing tubes and processed within 2 h from collection for culture and assessment of MACs functional properties (see below).

The Ethics Committee of the University of Pisa approved the study protocol and written informed consent was obtained from all participants before any study procedure.

MACs were cultured according to previously described techniques [26]. PBMCs were isolated from blood of each individual with type 2 diabetes by Biocoll (Biochrom AG; density = 1.077 g/ml) density-gradient centrifugation. Total PBMCs were seeded on 2 µg/cm² fibronectin coated culture dishes (BD Falcon™) or on Lab-Tek ^®II chamber slides system (Sigma-Aldrich Ltd, Poole, Dorset, UK) after red cell lyses. After isolation, cells were cultured in endothelial basal medium (EBM-2, Lonza Sales AG, Basel, Switzerland) supplemented with EGM-2-MV-SingleQuots containing human endothelial growth factor (EGF), hydrocortisone, insulin-like growth factor-1 (IGF-1), fibroblast growth factor (FGF), vascular endothelium growth factor (VEGF), ascorbic acid, antibiotics and 5% fetal bovine serum (FBS, Lonza Sales AG). After 3 days of culture, non-adherent cells were discarded by washing with phosphate buffered saline (PBS) and the culture medium was replenished daily. On day 5, adherent cells displaying an elongated spindle-shaped morphology were identified as MACs.

MACs (“early” EPCs) were characterized for the uptake of 1,10-dioctadecyl-3,3,3,3-tetramethylindocarbo-cyanine-labeled acetylated low-density lipoprotein (DiI-ac-LDL) and for lectin binding. The staining was performed incubating cells with 10 µg/ml of DiI-ac-LDL (Invitrogen, Life Technologies Ltd, Paislet, UK) for 2 h at 37 °C in dark conditions. Cells were fixed in 4% paraformaldehyde for 30 min and counterstained with 1 mg/ml FITC-labelled lectin from Ulex europaeus (Sigma-Aldrich Ltd) for 2 h at 37 °C in dark conditions. Stained cells were observed by a fluorescence microscope and double positive DiI-ac-LDL/Lectin cells were identified as MACs. To evaluate the immunophenotype of MACs, adherent cells were detached with trypsin–EDTA and 5 × 10⁵ cell/tube were incubated with anti-human CD34-PE (BD Biosciences), CD133-PE (Miltenyi Biotec), VEGFR-2-Alexa Fluor 647 (BioLegend), CD31-FITC (BD Biosciences), CD45-FITC (BD Biosciences) and CD42-PE (BD Biosciences) for 30 min in dark conditions at 4 °C. Isotype control antibodies were used to set baseline fluorescence levels. The labeled cells were analyzed on a FACS-Calibur Instrument (BD Biosciences), acquiring 2 × 10⁴ events for each analysis. The flow cytometric analysis was repeated six times. After a 5 days culture under standard conditions, MACs resulted in an adherent population consisting of cells that showed elongated with a spindle-like shape and exhibited double positivity for Ac-LDL and lectin binding (UEA-1-FITC) as established by fluorescent microscope analysis (Fig. 1a). MACs phenotype was confirmed by the expression of main endothelial cell surface markers: CD14 (99.06 ± 0.55%), CD31 (42.73 ± 25.24%), CD34 (44.24 ± 34.74%), CD42 (1.56 ± 1.77%), CD45 (96.09 ± 4.07%), CD133 (9.27 ± 6.93%) and VEGFR-2 (44.15 ± 21.78%) (Fig. 1b).

Viability of the cultured MACs was determined in cells obtained from 48 low-HDL and 40 high-HDL subjects by using the MTT assay. MTT (3-(4,5-Dimethylthiazol-2-yl)-2,5-diphenyltetrazolium bromide) measures mitochondrial activity in living cells. Briefly, after 5 days of culture MACs were incubated with MTT (Sigma, St. Louis, USA) (1 mg/ml) for 3 h at 37 °C, 5%/95% CO₂/O₂. Upon incubation, the medium was removed and the cells were solubilized in 10% DMSO/90% Isopropanol. Then, the amount of the dye released from the cells was quantified by measuring the optical density at 540 nm (reference wavelength: 620 nm) with a multiplate reader (Multiskan EX, THERMO). The optical density is directly correlated with the amount of metabolically active cells. To test the effect of an oxidative stress condition on MACs viability, H₂O₂ (1 mM, 1 h) was added to the culture medium.

Adhesion capacity was determined in MACs from 21 low-HDL and 24 high-HDL. To this purpose MACs were washed with PBS, and gently detached with 0.25% trypsin/EDTA. After centrifugation and re-suspension, equal cell numbers (50,000 cells/well) were seeded on fibronectin coated 96-well microplates, and incubated for 30 min at 37 °C, 5% CO₂. The cells were fixed in 4% paraformaldehyde and then incubated with 0.25% crystal violet for 30 min; therefore, excess dye was removed by several washes with PBS and the dye absorbed by adherent cell nuclei was extracted with 33% AcOH. The amount of the dye released from the cells was quantified by measuring the optical density at 540 nm.

Senescence was evaluated in MACs from 21 low-HDL and 19 high-HDL. Senescent cells were identified using the Senescence Cells Histochemical Staining kit (Sigma-Aldrich Ltd, Poole, Dorset, UK). Briefly, MACs (300,000 cells/well) were washed in PBS, fixed for 7 min at room temperature, washed again and incubated for 16–18 h at 37 °C without CO₂ and with X-gal chromogenic substrate. After that, cells were washed with PBS, and DMSO was added to dissolve the stain at 37 °C for 30 min; absorbance was measured at 620 nm. To test the effect of an oxidative stress condition on MACs senescence, H₂O₂ (1 mM, 1 h) was added to culture medium.

MACs migration was determined in cells from 25 low-HDL and 23 high-HDL using the 5 µM QCM™ Chemotaxis Assay (Millipore, USA), based on the Boyden chamber principle. After 5 days in culture, the cells were detached using Trypsin/EDTA and harvested by centrifugation; hence, 200,000 cells/well were added to the upper part of a modified Boyden chamber placed in a 24-well culture dish containing EGM-2 and EGM-2 enriched with VEGF (50 ng/mL) (Sigma-Aldrich Ltd, Poole, Dorset, UK) and incubated for 24 h. Cells in the insert were stained and then placed in a well containing a stain extraction buffer after washing in PBS. Migrating cells were counted by colorimetric measurement (optical density at 540 nm).

Reactive oxygen species production was evaluated in cells from 31 low-HDL and 20 high-HDL using ROS-sensitive fluorescent probe 5-(and-6)-chloromethyl-2′,7′-dichloro-di-hydro-fluorescein diacetate, acetyl ester (CM-H₂DCFDA) (Invitrogen, Life Technologies Ltd). Briefly, MACs (200,000 cells/well) were incubated with CM-H₂DCFDA (10 µM/well) for 30 min at 37 °C in dark conditions and ROS production was detected as an increase in fluorescence, by a fluorescence microplate reader, at 495 nm excitation and at 527 nm emission.

Statistical analyses were carried out using the SPSS 13.0 software (SPSS Inc., Chicago, I, USA) for Mac OS X. Data are expressed as median (interquartile range) and/or mean ± SD for continuous variables, and number of cases and percentage for categorical variables. Continuous variables were compared by unpaired Student’s t test or by one-way ANOVA (with Scheffe post hoc multiple comparisons) for normally distributed variables and by the Wilcoxon Sum-of-Ranks (Mann–Whitney) U test or the Kruskal–Wallis test for variables with skewed distribution. The general linear model (GLM), as an extension of the linear multiple regression for a single dependent variable (each parameter exploring MACs function), has been employed to verify whether HDL levels (the categorical independent factor) still has an effect, beyond the effects of covariates for which significant (or marginally significant) differences have been observed in low-HDL as compared to high-HDL individuals (gender, age, diabetes duration, BMI, waist circumference, non-HDL cholesterol, triacylglycerol, eGFR, presence of hypertension and smoking habits). Estimated marginal means (ESM) have been reported where appropriate. Pearson χ² or the Fisher exact probability tests were applied to categorical variables. Logistic regression analysis with backward stepwise variables elimination was used to assess the independent impact of predictors on each criterion variable stratified by the median level. p values of 0.05 or less were considered statistically significant.

The main anthropometric and clinical characteristics, and the pharmacological treatments of the study cohort are shown in Table 1. Compared with individuals with low HDL-C those with high HDL-C were more frequently females, were older, and had longer diabetes duration and higher total cholesterol levels. Low-HDL group had higher BMI, waist circumference and prevalence of obesity, higher diastolic BP and prevalence of hypertension, higher non-HDL cholesterol, triacylglycerol, uric acid and creatinine levels, with lower eGFR. Subjects with low-HDL were more frequently current smokers and on blood pressure-lowering treatments. No differences were found for fasting plasma glucose levels, HbA1c, systolic BP, LDL cholesterol, treatment with RAAS (Renin Angiotensin Aldosterone System) inhibitors, lipid-lowering agents or glucose-lowering drugs. Finally, no differences were observed in the prevalence of non-advanced retinopathy, previous major acute cardiovascular events or distribution of UAE categories. MAC studies were conducted in subgroups out of the whole cohort as specified in the “Methods” section.

To further evaluate the relationship between HDL-C levels and MACs functional properties, the study population was stratified into quartiles (Q) of HDL-C levels yielding quartiles thresholds of 34, 42 and 71 mg/dl, respectively (0.88, 1.09 and 1.84 mmol/l). Viability increased from Q1 to Q4 (p = 0.001) and was significantly higher in Q4 (124.8 ± 27.1%) than in Q1 (86.9 ± 24.5%, p < 0.001), Q2 (95.0 ± 33.2%, p = 0.020) and, marginally, than in Q3 (98.8 ± 32.7%, p = 0.055); this remained by comparing Q4 and Q1–Q3 (124.8 ± 27.1 vs. 93.3 ± 30.1%, p < 0.0001, Fig. 3). Effect of HDL quartiles on viability was still significant (p = 0.008) beyond the effects of covariates and with a marginal role for eGFR (p = 0.094). ESM was 128.6 ± 39.3 in Q4 vs. 88.0 ± 36.6 (p < 0.001), 93.2 ± 35.7 (p = 0.006) and 97.1 ± 39.2% (p = 0.004) in Q1–Q3, respectively. This difference persisted comparing Q4 and Q1–Q3 (126.7 ± 36.9 vs. 92.4 ± 42.7%, p < 0.001), with marginal effects for both eGFR (p = 0.092) and diabetes duration (p = 0.090). Consistently, adhesion was higher in Q4 than in Q1–Q3 (115.5 ± 24.1 vs. 95.6 ± 30.4%, p = 0.044; Fig. 3). This remained marginally significant (ESM 114.4 ± 32.0 vs. 96.0 ± 30.6%, p = 0.057) beyond the effects of covariates. No differences by HDL-C quartiles were observed for senescence, migration or ROS production.

HbA1c and HDL-C resulted independent covariates of MACs viability: indeed, viability increased with the improvement of HbA1c (OR 0.789) as well with the increase in HDL-C (OR 1.034) (Table 2).

When HDL quartiles were included in the regression instead of HDL-C levels, Q4 was strongly associated with EPCs viability (OR 8.343; 95% CI 2.219–31.369, p = 0.002).

Independent predictors of the adhesion capacity were HDL-C and BMI, i.e. adhesion was directly related to HDL levels (OR 1.029) and inversely with BMI (OR 0.949). When HDL quartiles were included in the regression instead of HDL-C levels, Q4 was strongly associated with MACs adhesion capacity (OR 6.492; 95% CI 1.217–31.635, p = 0.011). Finally, HbA1c (OR 1.683, directly) and BMI (OR 0.878, inversely) emerged as independent correlates of MACs senescence.