Increased intracellular miR-142 in adults with Type 1 diabetes

Abstract

microRNAs (miRs), including miR-142, modulate gene expression and processes implicated in vascular damage and may serve as therapeutic targets and agents, including in Type 1 diabetes (T1D). The project aimed to assess whether miR-142 levels differ between people with and without T1D, and to analyse miR-142 associations with cardiovascular (CVD) risk factors. Intracellular miRs were isolated from whole blood cell pellets using TRIzol-based methodology. In a cross-sectional study in 102 adults cellular miR-142 levels were significantly higher (on unadjusted and adjusted analyses) in 69 adults with Type 1 diabetes relative to 33 non-diabetic subjects: mean±SD, 3.53±3.66 vs. 1.25±0.78, p<0.0002, but was not related to HbA1c levels. Further miR-142 research, including longitudinal and intervention studies and basic science are of interest. miR-142 may be valuable in clinical practice for predicting health and as a treatment target.

Introduction

Type 1 diabetes (T1D) is associated with microvascular complications (retinopathy, nephropathy and neuropathy) and cardiovascular disease (CVD) ¹. Novel biomarkers are of interest, as even with well-controlled traditional risk factors, such as hyperglycaemia, hypertension and dyslipidaemia, residual risk remains ². Biomarkers suggest mechanisms of tissue damage, protection and therapeutic targets. microRNAs (miRs) are non-coding RNAs which modulate multiple genes and alter cell signalling and processes in immune diseases and atherosclerosis, including cell survival and proliferation, inflammation, clotting, fibrosis and angiogenesis. miRs form intracellularly and are usually present at much lower levels in extracellular fluids.

In human studies miRs, including miR-142, have been associated with T1D immune processes ^3,4, platelet dysfunction ⁵, CVD ⁶, and coronary artery stent restenosis ⁷. To our knowledge miR-142 levels in people with T1D and its complications and risk factors have not been reported, hence we conducted a cross-sectional study in adults with and without T1D.

Subjects and methods

Participants (n=102) provided written informed consent (St Vincent’s Ethics Committee HREC-A-024/00). Participants were 33 healthy non-diabetic and 69 T1D adults (35 without complications (T1D-No Cx) and 34 with microvascular complications (T1D-Cx); 26 with proliferative diabetic retinopathy, 28 with nephropathy (increased albuminuria in ≥2/3 12- or 24-hour urine collections or estimated glomerular filtration rate (eGFR)<60 ml/min/1.73m²) and 20 with both). None had known CVD. Lipids, kidney function, glycated haemoglobin (HbA1c) and blood counts were analysed (St. Vincent’s Pathology) on fasted venous blood and albumin/creatinine on a spot urine sample. Systemic inflammation (high sensitivity assay C-reactive protein (hsCRP) (Roche, Basel, Switzerland)); vascular inflammation (soluble cell adhesion molecules) (R&D, Minneapolis, MN), oxidative stress (myeloperoxidase, oxidised low-density lipoprotein (oxLDL)) (Mercodia, Upsala, Sweden) markers were analysed by ELISAs, short-term glycaemia marker - 1,5-anhydroglucitol (1,5AG) (GlycoMark, New York, NY), all with intra- and inter-assay CVs of <5% and <10% respectively. The estimated glucose disposal rate (eGDR), a measure of insulin sensitivity, was calculated by the Williams et al equation ⁸.

Cellular miR-142

Approximately 1ml packed cells from centrifuged EDTA blood (3,000g, 15min) was frozen (−80°C) until analysis. Total RNA, including miR, from the blood cells was isolated using 10 ml of TRIzol LS reagent (ThermoFisher, Waltham, MA) and further purified using a miRNeasy kit (QIAGEN, Germantown, MD) as per manufacturer’s instructions. RNA concentration and purity were determined with a NanoDrop spectrophotometer ND1000 (ThermoFisher, Waltham, MA) and stored at −80 °C until further processing.

cDNA was synthesised using a TaqMan Advanced miR cDNA Synthesis Kit (ThermoFisher, Waltham, MA) and the qRT-PCR reactions were carried out with iQ SuperMix (Bio-Rad, Hercules, CA) and TaqMan Advanced miR Assay (ThermoFisher, Waltham, MA) using a CFX96 Touch System (Bio-Rad, Hercules, CA). miR-142 levels were measured in the 102 study participants as described ⁹ and normalised with the geometric mean of two abundant stable endogenous miRs (miR-25-3p and miR-26a-5p) across the three groups selected with TaqMan Advanced miR Human Endogenous Controls (ThermoFisher, Waltham, MA). The primer efficiency was calculated with an online qPCR Efficiency Calculator ¹⁰. Relative expression levels were calculated with Microsoft Excel 2003 using a Pfaffl method ¹¹.

Statistical analysis

Statistica for Windows (Tibco, Palo Alto, CA) and XLStat (Addinsoft, Paris, France) were used. Missing values (<10%) were imputed (Markov Chain Monte Carlo multiple imputation method). Best models were selected using exhaustive search with adjusted R² as selection criterion. Correlations between variables were assessed using Spearman rank correlation test. Significance was taken at p<0.05.

Results

Subjects

Table 1 shows subject demographics of non-diabetic, all T1D and T1D subjects by complication status. Relative to non-diabetic subjects, T1D subjects had significantly higher systolic blood pressure, HbA1c, total- and LDL-cholesterol (LDL-C), eGFR, hsCRP and lower eGDR (reflecting insulin resistance). Apart from higher HbA1c and hsCRP levels there were no significant differences between non-diabetic and complication-free T1D subjects. HbA1c, body habitus, total cholesterol, smoking and eGDR were significantly more adverse in T1D subjects with vs. without complications.

Table 1.

Subject characteristics and miR-142 levels, including by Type 1 diabetes complications status. Data showed: mean ± SD, median (LQ, UQ) or mean (95% CI).

	Non-diabetic	All T1D	T1D-No Cx	T1D-Cx	Pb
N (M/F)	33 (16/17)	69 (34/35)	35 (16/19)	34 (18/16)	0.77
Age (yrs)	43 ± 13	39 ± 13	37 ± 13	42 ± 13	0.10
T1D duration (yrs)	–	24 ± 13	20 ± 14	28 ± 11‡	0.01
WHR (cm/cm)	0.86 ± 0.10	0.87 ± 0.08	0.82 ± 0.06	0.91 ± 0.08‡	0.0005
BMI (kg/m2)	26.7 ± 4.8†	27.8 ± 6.5	24.8 ± 2.7	31.0 ± 7.7‡	0.0001
BP (mmHg)	121†/69† ± 12/9	132/73 ±19/10	125/70 ± 11/7	140‡/75 ± 22/12	<0.0001/0.04
HbAlc (%)	5.2 ± 0.4†,‡	8.1 ± 1.5	7.9 ± 1.2	8.3 ± 1.7	<0.0001
HbAlc (mmol/mol)	33.3 ± 4.7†,‡	65.4 ± 16.4	63.2 ± 12.8	67.4 ± 19.1	<0.0001
1,5AG (μg/mL)	18.9 ± 9.2†,‡	4.8 ± 3.2	4.1 ± 2.0	4.1 ± 3.9	0.0005
Total Cholesterol (TC) (mM)	5.2 ± 1.1‡	4.7 ± 1.3	4.3 ± 0.8	5.1 ± 1.7‡	0.01
Fasting triglycerides (TG) (mM)	1 (0.7, 1.4)	0.8 (0.6, 1.2)	0.7 (0.6, 1.0)	1.0 (0.8, 2.2)‡	0.01
HDL-C (mM)	1.5 ± 0.4	1.5 ± 0.4	1.5 ± 0.4	1.4 ± 0.5	0.81
LDL-C (mM)	3.2 ± 1.1‡	2.7 ± 1.1	2.4 ± 0.7	3.0 ± 1.4	0.02
eGFR (ml/min/1.73 m2)	111 (102, 120)	100 (84, 120)	104 (93, 124)	96 (78, 116)	0.06
hsCRP (mg/L)	0.7 (0.4, 1.2)†	2.0 (0.5, 5.4)	1.2 (0.4, 3.2)	2.5 (1.2, 6.4)	0.003
eGDR (mg/kg/min)	10.5 ± 2.1†,‡	7.4 ± 2.5	8.9 ± 1.7	5.7 ± 2.1‡	<0.0001
Smoking (current/ex/never)	3/10/19a	12/15/39a	10/6/18a	2/9/21a,‡	0.07
Retinopathy (Y/N/only)	–	28/41/8	–	28/6/8	–
Nephropathy (Y/N/only)	–	26/43/6	–	26/8/6	–
miR-142	Non-diabetic	All T1D	T1D-No Cx	T1D-Cx	Pb
miR-142 (raw) (mean ± SD)	1.25 ± 0.78†	3.53 ± 3.66	2.68 ± 2.14	4.40 ± 4.62‡	0.0002
miR-142 (raw) (median (LQ, UQ))	1.03 (0.77, 1.68)†,‡	2.61 (0.93, 5.19)	2.21 (0.80, 3.57)	3.00 (1.16, 6.05)	0.0002
Adjusted(1) miR-142 (mean, 95% CI)	0.67 (−0.89, 2.24)†	3.99 (2.79, 5.19)	3.12 (1.80, 4.44)	4.44 (3.02, 5.85)	0.01
Adjusted(2) miR-142 (mean, 95% CI)	0.97 (−0.43, 2.37)†	3.94 (2.80, 5.08)	2.95 (1.73, 4.16)	4.85 (3.58, 6.12)	0.0006

BOLD p < 0.05 Non-diabetic vs. All T1D.

Adjusted (1): BP, HbA1c, TC, LDL, eGFR, hsCRP, eGDR, Adjusted (2): Age, Sex, HbA1c.

^aSome data missing.

^bNon-diabetic vs. T1D-No Cx vs. T1D-Cx: ANOVA, Kruskal-Wallis ANOVA or χ².

^‡p < 0.05 vs. T1D-No Cx.

^†p < 0.05 vs. T1D-Cx.

miR-142

Higher in T1D.

On unadjusted analyses intracellular miR-42 levels were significantly higher in T1D vs. non-diabetic adults, with no significant difference by T1D complication status. (Table 1).

Using Bonferroni tests miR-142 levels remained significantly higher in T1D vs. non-diabetic adults when adjusted for age, sex and HbA1c; and also, when adjusted for HbA1c, BP, total and LDL-C, eGFR, CRP and eGDR.

On receiver operating characteristic analysis intracellular miR-142 levels were a significant determinant of T1D presence / absence, with area under the curve (AUC) of 0.736, sensitivity 0.45, specificity 1.00 and p<0.0001; but was not for T1D complication status (Supplementary material Figure 1).

Correlates (Spearman) of miR-142

Non-diabetic group: BMI, r_S=0.42, p=0.02; sICAM-1, r_S=−0.57, p=0.003; 1,5AG r_S=−0.76, p=0.01.

All T1D: serum creatinine, r_S=0.25, p=0.046.

T1D-No Cx: urinary ACR, r_S=−0.40, p=0.03; sICAM-1, r_S=−0.49, p=0.01.

T1D-Cx: Nil.

In T1D miR-142 levels were not statistically significantly associated with age, sex, body habitus, smoking, T1D duration, HbA1c, lipids, eGFR or full blood count metrics.

Determinants of miR-142

In all T1D and non-diabetic subjects combined, in a 10-factor model (from Table 1) using exhaustive search, T1D presence or absence and total cholesterol were the only significant determinants of miR-142 levels (adjusted R²=0.168, p=0.002).

In non-diabetic subjects in a 5-variable model log triglycerides and HDL-C were significant determinants of miR-142 (adjusted R²=0.250, p=0.023).

In T1D the best (7-variables) model TC, eGFR and sex were significant independent predictors (adjusted R²=0.096, p=0.066.)

Discussion

We report higher blood cell miR-142 levels in adults with vs. without T1D, and no differences by T1D complication status. There were links with lipids and kidney function, but not glycaemia reflected by HbA1c levels. In keeping, high-glucose-exposed human monocyte cell-line THP-1 cells overproduced miR-142 in extracellular vehicles (EVs), which induced dysfunction in exposed human umbilical vein endothelial cells ¹². Another basic science study showed increased miR-142 in EVs with diabetes ¹³. In contrast, miR-142-5p was downregulated in diabetic rat retina and in human retinal endothelial cells vs. non-diabetic controls ¹³. The discrepancy may relate to differences in tissues and in animal or cultured cells vs. humans.

miR-142 can activate the PPARα-linked Wnt pathway, and we demonstrated in T1D animals that the PPARα agonist fenofibrate is a Wnt pathway inhibitor ^14,15. In Type 2 diabetes FIELD and ACCORD Lipid trials fenofibrate reduced: CVD in dyslipidaemic adults ¹⁶, retinopathy ^17,18 and nephropathy ^19,20. miR-142 levels were not evaluated in these trials.

Study strengths include novelty and detailed subject characterisation. Limitations include the cross-sectional design and lack of plasma miR-142 levels.

Research merited includes mirR-142 studies in specific cell types, plasma and urine, longitudinal studies, and intervention effects. miR-142 may predict health in clinical practice and be a treatment target.

Highlights for review:

1. Evaluation of cellular miR-142 in Type 1 diabetes and adults without diabetes.

2. Well characterised study patients.

3. Association of miR-142 levels with kidney function (eGFR).

4. No difference in miR-142 levels in the presence of vascular complications.