Increased expression of microRNA-15a and microRNA-15b in skeletal muscle from adult offspring of women with diabetes in pregnancy

Abstract

Offspring of women with diabetes in pregnancy exhibit skeletal muscle insulin resistance and are at increased risk of developing type 2 diabetes, potentially mediated by epigenetic mechanisms or changes in the expression of small non-coding microRNAs. Members of the miR-15 family can alter the expression or function of important proteins in the insulin signalling pathway, affecting insulin sensitivity and secretion. We hypothesized that exposure to maternal diabetes may cause altered expression of these microRNAs in offspring skeletal muscle, representing a potential underlying mechanism by which exposure to maternal diabetes leads to increased risk of cardiometabolic disease in offspring. We measured microRNA expression in skeletal muscle biopsies of 26- to 35-year-old offspring of women with either gestational diabetes (O-GDM, n = 82) or type 1 diabetes (O-T1DM, n = 67) in pregnancy, compared with a control group of offspring from the background population (O-BP, n = 57) from an observational follow-up study. Expression of both miR-15a and miR-15b was increased in skeletal muscle obtained from O-GDM (both P < 0.001) and O-T1DM (P = 0.024, P = 0.005, respectively) compared with O-BP. Maternal 2 h post OGTT glucose levels were positively associated with miR-15a expression (P = 0.041) in O-GDM after adjustment for confounders and mediators. In all groups collectively, miRNA expression was significantly positively associated with fasting plasma glucose, 2 h plasma glucose and HbA1c. We conclude that fetal exposure to maternal diabetes is associated with increased skeletal muscle expression of miR-15a and miR-15b and that this may contribute to development of metabolic disease in these subjects.

Introduction

Fetal exposure to maternal diabetes predisposes offspring to diabetes and associated cardiometabolic diseases (1,2). The increased risk associated with exposure to maternal diabetes is greater than the risk associated with genetic predisposition alone (3,4), and is thought to be mediated, at least in part, by exposure to a detrimental intrauterine environment.

Although extensive epidemiological evidence indicates that the fetal environment plays an important role in determining disease risk in adulthood (5–7), the underlying molecular mechanisms are not well defined. Evidence suggests that a detrimental intrauterine environment leads to changes in the expression and function of genes in the offspring, without accompanying changes in the underlying DNA sequence (8–11). These changes include changes in expression of microRNAs (miRNAs)—small (approximately 22 nucleotides long), non-coding RNAs that serve as posttranscriptional regulators of gene expression (12–14). miRNAs are believed to regulate between 60% and 90% of genes in the human genome (15–17), and are involved in a range of biological functions including glucose and lipid metabolism, maintenance of glucose homeostasis and development of diabetes (18–21). miRNAs belong to different families, with members of the same family predicted to have similar mRNA targets. miR-15a and miR-15b belong to the miR-15 family, known to be involved in regulation of insulin signalling and synthesis (22–24).

miR-15a was downregulated in skeletal muscle from hyperglycemic T2DM patients (25) and in plasma of subjects with prediabetes and T2DM (26,27). Short-term glucose exposure (1 h) led to increased miR-15a expression, insulin expression and biosynthesis in mouse pancreatic beta cells, while long-term glucose exposure (3 days) resulted in decreased miR-15a expression, insulin expression and biosynthesis (24). Taken together, these findings indicate a role for miR-15a in insulin synthesis and a general decrease in miR-15a levels in subjects with prediabetes and T2DM.

Elevated miR-15b levels may contribute to insulin resistance through downregulation of the insulin receptor and decreased activity of important proteins in the insulin signaling cascade (22,28,29). Levels of miR-15b were increased in skeletal muscle of low birth weight subjects predisposed to type 2 diabetes (T2DM), and in line with this, low birth weight offspring of rats fed a low protein diet during pregnancy had increased skeletal muscle expression of miR-15b compared with offspring of rats fed a control diet (22). These findings suggest a role for the intrauterine environment in the control of skeletal muscle miR-15b expression.

miR-15b was also upregulated in skeletal muscle from hyperglycemic T2DM patients (25), while it was downregulated in monozygotic twins with overt T2DM compared with their non-diabetic co-twin (22).

Skeletal muscle is responsible for the majority of insulin-mediated glucose uptake in the body, and plays a key role in the pathophysiology of metabolic disease. Studies have demonstrated that peripheral insulin resistance can be present for many years before onset of disease in predisposed individuals (30), making skeletal muscle an ideal tissue to study when examining potential risk of metabolic disease.

Using a candidate gene approach, we examined skeletal muscle miR-15a and miR-15b expression in adult offspring exposed to maternal diabetes in fetal life, with a known susceptibility to T2DM and associated cardio-metabolic diseases. We hypothesized that exposure to maternal diabetes leads to altered offspring skeletal muscle miR-15a and miR-15b expression, and that these miRNAs may be associated with measures of glucose metabolism, representing a possible link between fetal environment and metabolic changes in adulthood.

Results

Clinical characteristics

Baseline clinical characteristics of the study population are shown in Table 1, and have been partially described previously (31). Exposed offspring had significantly higher 2 h OGTT plasma glucose values, and for O-GDM, significantly higher 30 min OGTT plasma glucose and borderline higher HbA1C levels. There was no difference in HOMA-IR levels between groups.

Table 1.

Baseline clinical characteristics of the cohort at follow-up

	O-GDM	O-T1DM	O-BP	O-GDM versus O-BP P-value	O-T1DM vs. O-BP P-value
N (total= 206)	82	67	57
Maternal data (1978–1985)
Age at delivery (years)	30.4 (5.2)	26.4 (4.7)	26.8 (4.6)	<0.001	0.645
Pregestational BMI (kg/m2)	24.3 (5.6)	21.7 (1.9)	21.2 (3.5)	<0.001	0.301
Family history of diabetes (yes versus no)	26% (21/82)	25% (17/67)	16% (9/57)	0.166	0.191
Smoking status (yes versus no)a	32% (22/69)	63% (37/59)	58% (26/45)	0.006	0.610
Offspring data (2012–2013)Anthropometric data
Age (year)	30.2 (2.1)	30.8 (2.4)	30.8 (2.4)	0.183	0.879
Gender (male)	52% (43/82)	46% (31/67)	46% (26/57)	0.429	0.942
Height (m)	1.76 (0.10)	1.74 (0.10)	1.74 (0.10)	0.481	0.676
Weight (kg)	77.8 (17.4)	78.3 (17.9)	75.3 (16.5)	0.398	0.331
Total body fat (%)	29.8% (0.1)	31.4% (0.1)	28.7% (0.1)	0.428	0.093
BMI (kg/m2)	25.2 (5.1)	26.0 (5.9)	24.6 (3.9)	0.493	0.113
Obese (BMI ≥ 30 kg/m2)	15% (12/82)	16% (11/67)	7% (4/57)	0.166	0.110
Results of OGTTb
Fasting plasma glucose (mmol/l)	5.0 (0.7)	4.9 (0.4)	4.9 (0.3)	0.245	0.381
30 min plasma glucose (mmol/l)	8.2 (1.7)	7.8 (1.7)	7.3 (1.6)	0.006	0.125
2 h plasma glucose (mmol/l)	6.0 (1.8)	6.3 (1.7)	5.3 (1.2)	0.016	0.001
Fasting insulin (pmol/l)	49 (43–55)	54 (48–61)	49 (42–56)	0.953	0.255
30 min insulin (pmol/l)	402 (347–467)	356 (303–418)	351 (296–415)	0.224	0.903
120 min insulin (pmol/l)	252 (210–304)	281 (240–329)	223 (185–268)	0.350	0.056
Fasting C-peptide (pmol/l)	621 (577–668)	647 (598–700)	591 (549–637)	0.369	0.106
30 min C-peptide (pmol/l)	2149 (1980–2331)	2034 (1857–2228)	2040 (1851–2249)	0.418	0.965
120 min C-peptide (pmol/l)	2357 (2156–2575)	2538 (2344–2748)	2170 (1973–2388)	0.218	0.012
HbA1C_DCCT (%)	5.4 (0.3)	5.3 (0.3)	5.3 (0.3)	0.079	0.569
Abnormal glucose tolerance (IFG, IGT or T2DM)	13% (11/82)	13% (9/67)	5% (3/57)	0.116	0.125
T2DM (diagnosed at follow-up)	2% (2/82)	1.5% (1/67)	0% (0/57)	0.235	0.354
Pre-diabetes (IFG and/or IGT)	10% (8/82)	10% (7/67)	5% (3/57)	0.334	0.291
IFG	1% (1/82)	0% (0/67)	0% (0/57)	0.403	NA
IGT	7% (6/82)	10% (7/67)	5% (3/57)	0.628	0.291
Both IFG and IGT	1% (1/82)	0% (0/67)	0% (0/57)	0.403	NA
T2DM (previously known)	1% (1/82)	1.5% (1/67)	0% (0/57)	0.403	0.354
HOMA-IRc	1.77 (1.56–2.02)	1.95 (1.71–2.22)	1.72 (1.47–2.02)	0.784	0.222

BMI, body mass index; HOMA-IR, homeostatic model assessment insulin resistance [HOMA-IR = G₀ (mmol/l) × 10 (pmol/l)/135); Hs-CRP, high sensitivity C-reactive protein; IFG, impaired fasting glucose; IGT, impaired glucose tolerance; O-BP, offspring of women from the background population; OGTT, oral glucose tolerance test; O-GDM, offspring of women with gestational diabetes; O-T1D, offspring of women with type 1 diabetes in pregnancy; T2DM, type 2 diabetes. Data are provided as mean (SD), median (25th–75th percentiles) or percentage (number), unless or otherwise indicated. All comparisons are to the O-BP control group. Analysis of differences (means or proportions) between groups was performed by independent samples t-test, Mann–Whitney U-test, or χ² test, respectively. P-values < 0.05 are in bold. Total body fat percent=  (total fat mass/total body mass) × 100.

a

Data on smoking status were missing for 33 subjects.

b

Based on 2 h 75 g OGTT and evaluated according to WHO criteria of 2006 [37].

c

Data are presented as geometric mean and 95% confidence intervals.

miR-15a and miR-15b expression levels in offspring exposed to maternal diabetes

Skeletal muscle expression of miR-15a and miR-15b was significantly increased in O-GDM (P < 0.001 for both) and O-T1DM (P = 0.024; P = 0.005, respectively) compared with O-BP in crude analyses as well as in models adjusted for potential confounders and mediators (Table 2).

Table 2.

Skeletal muscle miRNA-15a and miRNA-15b expression levels in offspring of women with gestational diabetes (O-GDM) or type 1 diabetes (O-T1DM) compared to offspring of women from the background population (O-BP) in univariate and multivariate analyses

	O-GDM		O-T1D		O-BP		O-GDM versus O-BP P-value	O-T1D veruss O-BP P-value
	N	Geometric mean (95% CI)	N	Geometric mean (95% CI)	N	Geometric mean (95% CI)
miRNA 15a	76	0.82 (0.75–0.89)	63	0.62 (0.50–0.77)	42	0.44 (0.35–0.50)	Unadjusted	Unadjusted
							<0.001	0.024
							Model 1	Model 1
							<0.001	0.013
							Model 2	Model 2
							<0.001	0.016
miRNA 15b	76	0.51 (0.45–0.59)	63	0.44 (0.37–0.52)	42	0.30 (0.24–0.37)	Unadjusted	Unadjusted
							<0.001	0.005
							Model 1	Model 1
							<0.001	0.003
							Model 2	Model 2
							<0.001	0.004

O-BP, offspring of women from the background population; O-GDM, offspring of women with gestational diabetes; O-T1D, offspring of women with type 1 diabetes in pregnancy. Expression levels are relative to the RNU48 control gene. All comparisons are to O-BP control group, performed by independent samples t-test on log transformed values. Model 1: maternal age at delivery, smoking status (yes/no), pre-pregnancy BMI, family history of diabetes (yes/no), offspring gender and age at follow-up. Model 2: model 1 with addition of offspring total body fat percent.

Correlations between miR-15a and miR-15b expression and parameters of offspring glucose metabolism in the cohort as a whole

There was a positive association between expression of both miRNAs and fasting- and 2 h plasma glucose as well as HbA1c. A positive association was also found between miR-15a expression and OGTT 120 min insulin and C-peptide levels (Table 3).

Table 3.

Correlations between miR15a and miR15b expression and clinical parameters of glucose and insulin sensitivity in the cohort as a whole

Offspring data	miR-15a expressionR (P-value)	miR-15b expressionR (P-value)
Fasting plasma glucose (mmol/l)	0.172 (0.021)	0.185 (0.013)
2 h plasma glucose (mmol/l)	0.259 (0.001)	0.196 (0.010)
Fasting insulin (pmol/l)	0.076 (0.337)	0.054 (0.497)
30 min insulin (pmol/l)	0.142 (0.076)	0.001 (0.989)
120 min insulin (pmol/l)	0.216 (0.006)	0.123 (0.122)
Fasting C-peptide (pmol/l)	0.097 (0.196)	0.036 (0.628)
30 min C-peptide (pmol/l)	0.073 (0.337)	−0.062 (0.420)
120 min C-peptide (pmol/l)	0.237 (0.002)	0.109 (0.154)
HbA1c—DCCT (%)	0.157 (0.035)	0.253 (0.001)
HOMA-IR	0.111 (0.162)	0.092 (0.246)
Total body fat percent	0.073 (0.331)	0.016 (0.834)

Correlations are presented as Spearmans rank coefficient R (P-value). P-values <0.05 are in bold.

In general the pattern of correlations between miR-15a and offspring parameters of metabolic disease were similar when examining offspring groups separately, although attenuated (Supplementary Material, Table S1).

Effect of maternal blood glucose values on miR-15a and miR-15b expression

First, we performed univariate regression analyses examining the effect of maternal blood glucose values (fasting and 2 h post OGTT values for GDM mothers, mean glucose values in first and third trimester for T1DM mothers) on miR-15a or miR-15b expression. We then performed multivariate regression analyses using model 1 and model 2.

In univariate regression analyses, there were no significant associations with miR-15a or miR-15b expression, although maternal 2 h post OGTT glucose level in GDM mothers was borderline significantly positively associated with miR-15a expression (P = 0.066). After adjustment for confounders in model 1, 2 h maternal post OGTT glucose levels became significantly associated with miR-15a expression (P = 0.039) and this association remained significant after adjustment for offspring total body fat percent in model 2 (P = 0.041). There were no significant associations between maternal mean blood glucose levels in first and third trimester and miR-15a and miR-15b expression.

Discussion

Skeletal muscle expression of miR-15a and miR-15b was significantly upregulated in offspring exposed to maternal GDM or T1DM, and the association between exposure to maternal diabetes and increased miRNA expression remained significant after adjustment for confounders. In support of hyperglycaemia in pregnancy as a potential cause of these changes, maternal 2 h OGTT glucose levels were significantly positively associated with miR-15a expression levels after adjustments for confounders and one mediator. These results indicate that maternal diabetes and hyperglycaemia can increase offspring miR-15a and miR-15b expression, in support of our original hypothesis.

Offspring exposed to diabetes in pregnancy are typically heavier at birth. Our finding of increased skeletal muscle miR-15b expression in offspring exposed to diabetes in pregnancy is similar to the situation seen in low birth weight offspring (22). This lends support to the theory that low birth weight and maternal diabetes, representing opposite ends of the scale in terms of intrauterine nutrient availability, exhibit similar potentially detrimental changes in offspring miR-15b expression levels, which could be one of the mediators of altered offspring metabolism via fetal programming (32).

Pathway analyses implicate miR-15b in insulin signalling pathways and miR-15b may contribute to development of insulin resistance by downregulation of the insulin receptor (22), and upregulation of the phosphoinositol 3-kinase regulatory subunit 1 (gene: PI3KR1, protein: p85α), leading to impaired insulin signalling (22,28,29). Previous results have been conflicting. miR-15b is upregulated in skeletal muscle of twins with low birth weight, who are predisposed to T2DM (22), and in subjects with overt T2DM compared with individuals with normal glucose tolerance (NGT) (25), but downregulated in monozygotic twins with T2DM compared with their non-diabetic co-twin (22). However, these studies have had large differences in study populations. In the study of monozygotic twins discordant for diabetes (22), 9 out of 11 of the non-diabetic co-twins in the monozygotic twin cohort had impaired glucose tolerance (IGT), and the lower miR-15b levels in the twins with T2DM could be a reflection of the comparison with IGT subjects. This is supported by findings of increased miR-15b levels in subjects with IGT compared with both those with T2DM and NGT (25). Taken together, these results indicate that miR-15b levels are higher in subjects with IGT (and low birth weight) compared with those with NGT. As miR-15b is involved in downregulation of the insulin receptor and impaired insulin signalling, and miR-15b levels are increased in exposed subjects in our cohort, and subjects with IGT and T2DM in previous studies mentioned, this could reflect the increased insulin resistance in these subjects. The findings of higher miR-15b levels in IGT subjects compared with T2DM subjects in the study by Gallagher et al. could indicate that increased miR-15b expression is actually a compensatory mechanism to protect against hyperinsulinema, which subsequently fails when overt T2DM develops. A similar differential regulation of several genes involved in the insulin signalling pathway has previously been described, with downregulation of these genes in subjects with T2DM but upregulation in their first degree relatives (33). The nature of these conflicting results, and different potential mechanisms for regulation warrants further research into the action of the miR-15 family in different populations.

The differences in miR-15b expression could also be related to differences with respect to degree and duration of diabetes in the cohorts studied, and discrepancies regarding the direction of change in specific miRNAs in insulin resistance and T2DM also exist between other studies (25,34).

While miR-15b is involved in insulin resistance, studies indicate that miR-15a is involved in controlling insulin synthesis and secretion. One hour of high glucose exposure led to upregulation of miR-15a, while prolonged exposure to glucose for three days led downregulation of miR-15a expression in mouse beta cells. Changes in miR-15a in these cells corresponded to changes in insulin secretion and miR-15a caused inhibition of uncoupling protein 2, a negative regulator of insulin secretion. In the same study, miR-15a was found to promote insulin gene expression in mouse insulinoma cells (24). Our findings of increased miR-15a expression in skeletal muscle in offspring exposed to maternal diabetes cannot be directly linked to insulin secretion, due to the different tissues involved. However, interestingly, we found a positive correlation between miR-15a expression and 120 min glucose, insulin and C-peptide levels during the OGTT in adult offspring in our cohort.

In contrast to our findings in skeletal muscle, studies have found decreased levels of miR-15a in plasma of subjects with both prediabetes and T2DM (26,27) as well as in skeletal muscle of subjects with T2DM (25). However, again the latter study indicated differential regulation of miR-15a in subjects with NGT, IGT and T2DM, similar to the studies in miR-15b. A likely reason for the discrepancy between our findings in skeletal muscle, a key metabolic tissue and those in studies of miR-15a in plasma from subjects with diabetes and prediabetes could be the different tissues studied. Indeed, previous studies have shown different miRNA expression patterns in tissues, cells and plasma (35,36).

We also found significant associations between expression of both miRNAs and fasting- and 2 h plasma glucose levels as well as HbA1c levels in the cohort as a whole, implying a role for these two miRNAs in regulation of glucose homeostasis. A previous study reported a negative association between miR-15b and HbA1c levels between twin pairs with and without T2DM (22), in contrast to our findings of a positive association between miR-15b and HbA1c in our primarily non-diabetic cohort, implying differential regulation in subjects predisposed to disease compared with those with manifest T2DM. Although T1DM mothers usually have higher levels of hyperglycaemia, as also evidenced by the greater proportion of O-T1DM born large for gestational age (31), the differences in miRNA expression in our study were greater in O-GDM versus O-BP. This implies that factors besides maternal hyperglycaemia alone are involved in controlling offspring miR-15a and miR-15b expression and thereby fetal programming of T2DM. The strength of our study lies in the relatively large number of subjects studied, access to skeletal muscle tissue in these subjects, as well as the availability of information regarding maternal blood glucose values. The study investigates changes in miRNA expression in offspring exposed to maternal diabetes and at increased risk of T2DM due to fetal programming, and as far as we are aware this has not been investigated before.

It needs mentioning that our study design does not allow us to establish causality, and we cannot exclude that the observed changes in miRNA expression could be due, at least in part, to genes or lifestyle and environmental exposures later in life, and not caused by exposure to diabetes in pregnancy alone. In this respect, genotyping genetic variants involved in insulin signalling presents a direction for future research to elucidate the complex interactions between genetic expression and posttranscriptional regulation. Genetic risk of developing diabetes is influenced by the maternal and the paternal genome. However, robust data on paternal diabetes are not available in this cohort. As such, paternal diabetes is not accounted for in the present study. Finally, some subjects already diagnosed with T2DM or metabolic syndrome in the first round of follow-up declined participation in the second round (31), leading to a potential source of selection bias—although this bias will likely result in underestimation of our findings.

Studies show that transcriptional repression of individual proteins by one miRNA is typically mild (37,38). Individual miRNAs can repress hundreds of different genes, and one single gene can be repressed by several different miRNAs (37–39), so that miRNAs essentially act in concert as rheostats to fine tune the expression of a single gene transcript. Thus, one weakness of our study is the limited number of miRNAs studied, with demonstration of collective changes in many miRNAs perhaps being of greater potential physiological relevance. On the other hand, the fact that we a priori hypothesized that the two miRNAs tested might be altered in offspring of women with diabetes in pregnancy based on knowledge of the role of these miRNAs in insulin biosynthesis and resistance as well as knowledge of changes in these miRNAs with exposure to a detrimental intrauterine environment, and subsequently went on to demonstrate changes in exposed offspring, provides strong support for our hypothesis and results.

Global miRNA expression studies have detected between 170 and 216 miRNAs in human skeletal muscle biopsies (25,34), and many of these may be altered in insulin resistance and T2DM and be involved in the fetal programming of adult metabolic disease. A global array would have provided some insight into these, but by using a targeted approach, we avoided the complicated and costly statistical and bioinformatic analyses associated with miRNA array approaches.

Finally, this study did not investigate potential target mRNA expression which would have helped show the function of the selected miRNAs.

In conclusion, we have shown in this study that the skeletal muscle expression of miR-15a and -15b is increased in adult offspring exposed to maternal diabetes in fetal life, even after adjustment for various maternal and offspring confounding factors, in support of our original hypothesis. We also showed a positive association between maternal 2 h post OGTT glucose levels and offspring miR-15a expression. These findings indicate that increased miR-15a and miR-15b expression could be a potential mechanism behind the fetal programming of T2DM and associated cardiometabolic disease. However, changes in skeletal muscle metabolism associated with fetal programming of T2DM most likely have multiple origins, and further studies are needed to shed light on the complex interactions between miRNA expression and metabolic phenotype.

Materials and methods

Study design

The study was a second follow-up of a cohort of adult offspring born to women with diabetes in pregnancy, and the cohort and study design have previously been described in detail (31). An overview of the study design is shown in Figure 1. A total of 206 out of 456 (45%) potentially eligible offspring, all born between 1978 and 1985 at Rigshospitalet in Copenhagen, Denmark, participated in the current round of follow-up. The offspring belonged to one of three groups depending on maternal diabetes status: Offspring of women with gestational diabetes (O-GDM, n = 82), offspring of women with type 1 diabetes (O-T1DM, n = 67) and a control group consisting offspring of women from the background population (O-BP, n = 57). Of the 456 eligible offspring 250 (55%) were lost to follow-up for various reasons, as described previously (31,40).

Figure 1.

Study design flowchart. Subjects participating and lost to follow-up. O-BP, offspring of women from the background population; O-GDM, offspring of women with gestational diabetes; O-NoGDM, offspring of women with risk factors for gestational diabetes but NGT in pregnancy (not invited to participate in second round of follow-up); O-T1DM, offspring of women with type 1 diabetes in pregnancy. *In the original cohort, 812 offspring were from the three groups O-GDM, O-T1DM and O-BP, due to the fact that O-NoGDM (n = 254) were not invited for the second follow-up (1066 − 254 =812).

Maternal selection criteria and diabetes in pregnancy in Denmark during the baseline period of 1978–1985

Maternal selection criteria have been published previously (1,31,40). In the period of 1978–1985, routine screening for GDM in Denmark was based on the presence of risk factors: a family history of diabetes, pre-pregnancy overweight, previous GDM, previous delivery of a baby weighing ≥4500 g and glycosuria. An oral glucose tolerance test (OGTT) was performed in the presence of one or more risk factors together with two consecutive fasting blood glucose measurements ≥4.1 mmol/l (41–43). Only diet-treated women with GDM were included in the cohort, and insulin-treated women with GDM were excluded in order to minimize risk of misclassification.

Mothers with type 1 diabetes fulfilled three criteria: onset of diabetes before or at 40 years of age, a classical disease history with symptoms of hyperglycaemia before diagnosis and initiation of insulin treatment no more than 6 months after diagnosis. HbA1c and self-monitored blood glucose measurements were not introduced into clinical practice at that time, and the routine care for pregnant women with T1DM in Denmark during the baseline period involved hospitalization for three days in the first and third trimester of pregnancy, amongst other things for the purpose of measuring 7-point blood glucose profiles daily. The average maternal blood glucose values in first and third trimester were calculated from these 3-day, 7-point blood glucose profiles.

Mothers from the background population were women referred to the same hospital for antenatal care and delivery, sampled consecutively by date of birth. The baseline prevalence of GDM in the background population at that time was estimated to be around 1–2% (41). Thus, the vast majority of women in this group are expected to have had NGT, although in accordance with the screening procedures at that time, not all of them were tested with an OGTT.

Offspring examination at follow-up

Participants were recruited and examined from May 2012 to September 2013, and coupling between mother and child was possible through the Danish Civil Registration System (44). Only singleton pregnancies were included, and in cases where more than one sibling met the inclusion criteria, only the oldest sibling was invited.

Participants were examined in the morning after an overnight fast and underwent tissue biopsies (skeletal muscle and subcutaneous adipose tissue, only skeletal muscle was used for this study), and a 2 h 75 g OGTT with venous blood sampling at 0, 30 and 120 min where glucose tolerance status was assessed according to World Health Organisation criteria (45). Dual X-ray absorptiometry (DEXA) whole-body scan (GE Medical Systems, Lunar Prodigy Advance, Fairfield Connecticut, USA), and anthropometric measurements including waist and hip circumference, blood pressure, height and weight were also performed. Details regarding blood sampling have previously been described in detail (31).

Skeletal muscle biopsies were obtained from the vastus lateralis muscle of the thigh using a Bergstöm needle after application of local anaesthesia to the skin and fascia of the biopsy site. All biopsies were frozen and stored at −80°C until analysis. The muscle biopsies have been used for another study from our group as well (31).

The study complied with the Declaration of Helsinki, was approved by the regional ethical committee, and written consent was obtained from all participants before inclusion.

Exclusion criteria

Offspring with type 1 diabetes, maturity-onset diabetes of the young (MODY), severe chronic disease or those who were pregnant, were excluded from participation.

Outcome variables

Outcomes of interest were expression of miR-15a and miR-15b in skeletal muscle of offspring exposed to maternal diabetes compared with unexposed offspring. In addition, we evaluated associations between maternal glucose estimates during pregnancy and expression of miR-15a and miR-15b in offspring skeletal muscle, as well as correlations between offspring miR-15a and miR-15b skeletal muscle expression levels and parameters of offspring glucose metabolism.

Exposure variables

The primary exposure variable was exposure to maternal GDM or T1DM, measured by offspring group. Maternal blood glucose values (fasting and 2 h post OGTT values for GDM mothers and first and third trimester blood glucose values for T1DM mothers) were also used as exposure variables in regression analyses.

Confounders

Potential confounders were chosen based on theoretical considerations as well as on findings from other studies.

We included maternal age at delivery, smoking status (yes/no), pre-pregnancy BMI and family history of diabetes (yes/no), as well as offspring age and gender, as potential confounders in model 1.

Mediators

Adiposity is known to be a causal factor in the development of insulin resistance (46), and total body fat percent could be a mediator of the effect of maternal diabetes on offspring risk of metabolic disease. We therefore included offspring total body fat percent as a potential mediator in model 2.

RNA isolation

A modified guanidinium thyiocyanate-phenol-chloroform extraction was used to extract total RNA from the muscle biopsies. Exactly 1 ml of cold Trizol (Thermo Fisher Scientific) and a −20°C cold steel bead (5 mm Qiagen) were added to each sample, the samples were then placed into −20°C cold racks and the tissue homogenized with a Qiagen Tissue Lyser (6 min, frequency 25). The homogenate was centrifuged at 13 200 rpm for 10 min at 4°C (Eppendorf 5415 R), 200 µl chloroform added to the supernatant, after 10 min at room temperature, the samples were centrifuged at 13 200 rpm for 15 min at 4°C. The RNA was precipitated from the aqueous phase by addition of an equal volume 2-propanol, vortexing and incubation for 5 min at RT. RNA was pelleted by centrifugation at 13 200 rpm for 15 min at 4°C and the pellet washed twice with 900 µl 75% ethanol. The pellets were resuspended in nuclease-free water. Samples were stored at −80°C until further use.

miRNA assays

To determine the amount of miR-15b and miR-15a in skeletal muscle biopsies, hsa-miR-15b-5p (miRBase accession number MI0000438) and hsa-miR-15a-5p (miRBase accession number MI0000069) Taqman miRNA assays (ThermoFisher Scientific) were used according to the manufacturer’s instructions. First, 10 ng of total RNA was reverse transcribed using the miRNA-specific RT primers and the TaqMan^® MicroRNA Reverse Transcription Kit (ThermoFisher Scientific). Quantitative real-time PCR was carried out using the TaqMan^® Universal PCR Master Mix and Taqman specific probes. Samples were run in triplicates on a ViiA™ 7 Real-Time PCR System (Applied Biosystems), cycler programme: 2 min 50°C, 95°C 10 min, followed by 40 cycles of: 15 s 95°C, 60 s 60°C. Expression levels were normalized to a standard curve and to the small-nucleolar RNA RNU48 (NCBI accession number NR_002745), determined using the Taqman miRNA control assay (ThermoFisher Scientific) following the same procedure as for the miRNAs. The TaqMAN miRNA assays used in this study have been tested and found to be highly precise, specific and sensitive for miRNA quantification (47). There was no cross-amplification between assays. In this study, the coefficient of variation between replicates was less than 2%, and a standard dilution series was performed to take interplate variations into account.

Statistical analysis

All statistical analyses were performed using IBM SPSS Statistics version 22. Normally distributed data are presented as mean (SD), while non-parametric data are presented as geometric mean (95% confidence intervals, CI). Differences between groups were analysed using independent samples Student’s t-test (after log-transformation for non-parametric data) or Chi-square test as appropriate. All comparisons were to the O-BP control group. Correlations were performed using Spearman’s rank correlation for non-parametric data. Forced entry multiple linear regression analysis was performed on log-transformed outcome variables in order to meet assumption of homoscedasticity. List-wise deletion was used in regression analyses while pairwise deletion was used in correlation analyses. A two-sided P-value <0.05 was considered significant.

Supplementary Material

Supplementary Material is available at HMG online.

Acknowledgements

We wish to thank and acknowledge the study participants, without whom the study would not have been possible. We also would like to express our gratitude to Malan Egholm, research nurse, MPH at the Department of Endocrinology at Rigshospitalet, for competent and skilful assistance during data collection.

Conflicts of Interest Statement. A.V. is currently employed by AstraZeneca, Mölndal, Sweden and AHO is currently employed by Novo Nordisk A/S. E.R.M. and P.D. are participating in a multinational study in collaboration with Novo Nordisk A/S and E.R.M. has received honorarium from Novo Nordisk A/S for talks. M.S., L.H., L.K., N.S.H., A.T., C.B. and T.D.C. have nothing to disclose.

Funding

This work was supported by the Danish Council for Strategic Research, the Novo Nordisk Foundation, Danish Diabetes Academy, Augustinus Foundation, Danish Diabetes Association, the A.P. Møller Foundation for the Advancement of Medical Science, European Foundation for the Study of Diabetes (EFSD) and the Rigshospitalet Research Fund.