Maternal Protein Restriction Leads to Pancreatic Failure in Offspring: Role of Misexpressed MicroRNA-375

Over the past 10 years the occurrence of type 2 diabetes (T2D) has increased at a frightening pace. While sedentary lifestyle and overfeeding undoubtedly contribute to this worldwide epidemic, the intrauterine environment of the fetus is an additional influential actor in long-term health. Indeed, several epidemiological studies of human populations highlighted a direct correlation between intrauterine growth retardation (IUGR) and the appearance of insulin resistance and T2D in adult life (1,2). Such observations have led to the concept of developmental origins of adult disease, which proposes that crucial programming of human disorders originates in early life (3,4).

To unravel the mechanisms underlying the programming of adult diseases, several animal models of IUGR have been generated (5). Despite differences in the nature, timing, and duration of the intrauterine insult, most of the animal IUGR models produce comparable outcomes. One of the most extensively studied models uses maternal protein restriction in rats. Importantly, this situation exhibits phenotypic similarities to the human IUGR, which is coupled to an age-dependent deterioration of glucose tolerance.

In the pathogenesis of T2D, insulin resistance associated with β-cell failure leads to chronic hyperglycemia, defining diabetes. In IUGR in animals and humans, functional disruption in multiple tissues including muscle, adipose tissue, liver, and pancreas occurs during adulthood. However, the endocrine pancreas seems to be affected most severely at early developmental stages, suggesting that, in the context of intrauterine modifications, T2D could originate from developmental defects in this organ.

Obviously, any disturbance in the environment of the endocrine cells within a given developmental time window may predispose to T2D, even in the next generation (6). Although the consequences of unfavorable intrauterine environment on fetal development have been documented, the molecular processes by which they occur are only starting to emerge (7).

An intriguing feature of the developmental origin of adult disease concerns the generational transmission of the health disorder. Such programming has been associated to changes in DNA methylation and/or histone modifications affecting gene expression and contributing to this cellular memory (8,9). Furthermore, nutrient-dependent modulation of microRNAs (miRNAs) may also trigger disease susceptibility and metabolic complications in offspring. Indeed, misexpressed miRNAs participate in the programming of adipose tissue in rats with retarded growth, causing lipotoxicity and insulin resistance, and hence they increase the susceptibility to metabolic disease during adulthood (10).

The mammalian genome codes for several hundreds of miRNAs that regulate gene expression through modulation of their target mRNAs. Most of these single-stranded 20– to 22-nucleotide-long RNAs interact with specific sequences in the 3′ untranslated region of the mRNA. By doing so, miRNAs induce mRNA degradation and/or translation inhibition (11). The biological importance of miRNAs is demonstrated by the diverse and profound phenotypic sequelae upon changes in their expression. These alterations are associated with perturbed development and pathological situations. Indeed, miRNAs seem to be major regulators of gene expression in many biological programs, including organ development and metabolism (12–17). Computational predictions of miRNA targets estimate that a single miRNA can affect a gamut of different mRNAs, suggesting that a large proportion of the transcriptome is subjected to miRNA modulation. Using a rat model of maternal malnutrition, we address here the hypothesis that perturbation of the programmed expression of key miRNAs in the endocrine pancreas of progeny contributes to the effects of early nutriture on the establishment of β-cell mass and, consequently, on the long-term health of the organism.

All procedures were performed in accordance with the guidelines for the care and use of laboratory animals of the French National Institute of Health and Medical Research. Nulliparous female Wistar rats weighing 200–250 g (Janvier, Le Genest-Saint-Isle, France) were mated overnight with male Wistar rats. The pregnant females were individually housed with free access to water. Dams were fed ad libitum during gestation and lactation with a control (20% w/w protein) or isocaloric low-protein (LP) diet (8% w/w protein; LP group) (Hope Farm, Woerden, the Netherlands) (18). After weaning, the progeny of both groups were fed standard chow ad libitum. A minimum of 3 litters per group were analyzed in each experiment.

Neoformed fetal rat islets were obtained as described previously (19). Briefly, pancreases of 21-day-old fetuses were removed aseptically, minced, and digested with collagenase (Sigma-Aldrich, St. Louis, MO). The digested pancreases were incubated for 7 days at 37°C in a humidified atmosphere. Islets of 3-month-old rats were obtained as described elsewhere (20). Following density-gradient centrifugation using a histopaque-1077 (Sigma-Aldrich), islets were washed and hand-picked under a stereomicroscope. Finally, they were either dissociated for transfection experiments (see Culture and Transfection of Dissociated Islet Cells) or cultured in RPMI 1640 Medium (Gibco, Grand Island, NY) supplemented with FBS (10% v/v) and Pen Strep (1% v/v).

Before transfection, rat islets were dissociated by trypsinization (0.5 mg/mL; Gibco) and seeded in 60 mm–diameter Petri dishes containing 5 mL RPMI 1640 medium (Gibco) containing FBS (10% v/v) and Pen Strep (1% v/v). Cells were incubated for 16 h at 37°C to allow full recovery. Then, dissociated islet cells were seeded at a density of 25 × 10³ cells/cm² on 35-mm well dishes coated with 804G-ECM (gift from Phillipe Halban, Geneva, Switzerland).

Using Lipofectamine 2000 (Invitrogen), 48 h after plating dissociated islet cells were transfected with 100 nmol/L of double-stranded RNA oligonucleotides corresponding to mature miR-375 or miRNA hairpin inhibitor to block endogenous miR-375 (Thermo Scientific, Gometz-le-Châtel, France). Cells were analyzed 72 h after transfection.

Pancreases from fetuses and from 3-month-old rats were fixed in 3.7% (w/v) formalin, dehydrated, and embedded in paraffin. Sections (5 μm) were collected, dewaxed, and hydrated in ethanol, and an antigen retrieval method was used. Tissues were permeabilized and incubated for 30 min with a blocking solution containing BSA (3% w/v) before overnight incubation with primary antibodies (see Supplementary Data). Next samples were incubated for 1 h at room temperature with secondary antibodies (see Supplementary Data).

All measurements were performed blindly to avoid influences from testers’ expectations. At fetal day 21, a minimum of 3 sections (every 150 µm) from each of 6 animals from 3 different litters per group were analyzed. A minimum of 6 sections (every 150 µm) from each of 6 animals per group of 3-month-old rats were analyzed.

Insulin-positive and glucagon-positive areas were measured to determine β- and α-cell fractions, respectively. These were measured as the ratio of the insulin-positive and glucagon-positive cell area over the total tissue area of the entire section.

Frequency distribution of various islet sizes was measured with ImageJ software. We considered as an islet a cluster of at least three insulin-positive cells (21). For their size distribution, islets were arbitrarily classified as small (<100 μm), medium (100 μm < diameter < 200 μm), or large (>200μm). The number of islets in each class was expressed as the percentage of total islets per group. Assuming that the cells are spheres, the diameter of individual β-cells was calculated with ImageJ software. To determine β-cell proliferation, pancreatic sections were stained with antibodies to phosphorylated histone H3 and to insulin and appropriate secondary antibodies.

RNA from fetal pancreas and islet cells was isolated using TRIzol reagent (Invitrogen). Total RNA (1 μg) was reverse-transcribed into complementary DNAs (cDNAs) and analyzed using SYBR Green (ABI PRISM 7000 Sequence Detector System). The amount of cDNA used in each reaction was normalized to the housekeeping gene cyclophilin A. The primers and quantitative PCR assay conditions are available upon request.

miRNA quantitative PCR array analysis was performed using the rno-miRNome miRNA profiling kit (System Biosciences, Mountain View, CA). RNA was extracted from individual pancreases of fetuses in the control and LP groups. Then, RNA from a minimum of four pancreases per litter was pooled for the array. Three litters per group were analyzed. Per litter, 1 μg of pooled RNA was reverse transcribed into first-strand cDNA using the QuantiMir RT Kit (System Biosciences). Profiling of mature miRNAs was performed on 3 litters per group by quantitative PCR in a 384-well plate format, including 380 miRNA-specific primers and using SYBR Green. Expression levels were normalized using U6 since its melting curve was more satisfactory compared with that obtained for the two other reference genes.

A heat map was created using open-source analysis software with a multiexperiment viewer (www.tm4.org). miR-375 expression was confirmed with the miRCURY LNA Universal RT MicroRNA PCR kit (Exiqon, Vedbaek, Denmark), according to the manufacturer’s instructions.

Fetal pancreas or dissociated islet cells were processed for protein isolation, as described by Dumortier et al. (22). For Western blotting, 10–50 µg of total proteins were separated by electrophoresis and transferred to polyvinylidene difluoride membranes (Immobilon-P; Millipore, Bedford, MA). Immunoreactive proteins were revealed by enhanced chemiluminescence (Millipore, Billerica, MA). Antibody to phosphoinositide-dependent kinase-1 (PDK-1) was from Cell Signaling Technology (Danvers, MA) and antibody to β-actin was from Sigma-Aldrich. Western blots were quantified by densitometry using ImageQuant software.

All experiments were performed with a modified Krebs-Ringer solution containing 2.8 mmol/L glucose (23). Dissociated islet cells or batches of 20 free-floating, size-matched islets were incubated at 37°C in 1 mL of Krebs-Ringer medium containing glucose at a concentration of 2.8 or 20 mmol/L. After 120 min, the incubation medium was removed so insulin could be measured. Islets or dissociated islet cells were collected and homogenized by sonification to extract insulin. To eliminate variations due to differences in individual islet batches, insulin secretion during incubation was expressed as a percentage of the islet insulin content at the start of incubation, which is referred to as fractional insulin release.

After an overnight fast, progeny of the rats in the control and LP groups were given an oral glucose bolus (2 g). Blood was collected from the tail vein at 0, 15, 30, 60, 90, and 120 min. Glycemia was measured using the OneTouch glucometer (LifeScan Inc., Milpitas, CA).

Results are shown as means ± SEM; n represents the number of litters analyzed. The Student t test was used to compare two conditions (paired or unpaired), and two-way ANOVA followed by post hoc Fisher test was used when more than two conditions were analyzed. A P value <0.05 was considered significant.

To investigate whether miRNA expression in the fetal pancreas is altered by a maternal LP diet, we compared the miRNA expression profile obtained from normal fetal pancreas and pancreas from fetuses in the LP group at day 21 of gestation. We used the quantitative PCR–based miRNome miRNA profiler to screen 380 mature rat miRNAs in the pancreas of 3 different litters of male rats in the control and LP groups (Fig. 1A). Of the 360 detectable (95%) miRNAs, 47 are differentially expressed between the control and LP groups (Fig. 1B). Remarkably, 43 are upregulated (28 with P < 0.05 and 15 with P < 0.09) in the LP group, while only 4 are downregulated (P < 0.05). Interestingly, we found that the expression of miR-375, which is highly expressed in endocrine pancreas and involved in β-cell proliferation and function, is increased at day 21 of gestation in the pancreas of fetuses in the LP group (1.4 ± 0.08 arbitrary units in LP group compared with 1 ± 0.028 arbitrary units in control group; see Supplementary Fig. 1).

PDK-1, one of the targets of miR-375 (24), is a key component of the insulin and growth factor signaling pathway, downstream of phosphoinositide 3-kinase in the protein kinase B cascade. Because miR-375 is upregulated in the pancreas of fetuses in the LP group, we measured PDK-1 protein levels in the pancreases of our experimental groups. Using Western blotting, we observed no difference in total PDK-1 in pancreases of the LP group compared with controls (Fig. 2A), whereas quantitative PCR analysis revealed a moderately decreased level of PDK-1 mRNA (Fig. 2B). However, immunohistochemical analyses showed that PDK-1 immunoreactivity is specifically reduced in the endocrine compartment of the pancreas (Fig. 2C).

PDK-1 is a key regulator of β-cell proliferation and growth in the developing pancreas (25). To investigate whether the reduction of PDK-1 protein observed in islets of animals in the LP group affects pancreatic development, immunohistochemical and morphometric analyses were performed. While in the progeny in the LP group the number of islets measured as insulin-positive aggregates per pancreas area is the same as in the control progeny, the islets are smaller than in the control group progeny (Table 1). This diminished islet size in pancreas of animals in the LP group can be accounted for by a 50% decrease in β-cell fraction area, whereas the α-cell fraction was preserved (Table 1). In fact, the maternal LP diet significantly increased the proportion of small islets and concomitantly reduced the proportion of large islets compared with the control progeny (Fig. 3A).

To gain more insight into the mechanism underlying reduced β-cell growth, the proliferative capacity of these cells was investigated. Co-immunostaining of pancreas sections with antibodies to insulin and phosphorylated histone H3 (Fig. 3B) or antibodies to Ki-67 (Supplementary Fig. 2A) showed a decreased mitotic index of β-cells in pancreases from animals in the LP group. Quantification revealed that β-cell proliferation is reduced by approximately 50% in fetuses in the LP compared with control groups. In contrast, apoptosis assessed using the TUNEL method (data not shown) and differentiation measured by β-cell markers seem to be intact (Supplementary Fig. 2B). In conclusion, protein restriction during pregnancy reduces in the progeny β-cell mass because of decreased β-cell expansion.

In addition to its role in β-cell proliferation, PDK-1 has been found to exert a major effect on β-cell size (26). To evaluate whether this was occurring in our experimental conditions, we took advantage of the fact that β-cells are the only islet cells that express glucose transporter-2 at the cell membrane. Morphometric measurements indicate that fetal β-cells in the LP group display a modest reduction in size (Fig. 3C). Thus, an LP diet during pregnancy diminishes the β-cell diameter of the fetuses by approximately 10%. When extrapolated to the islet volume, however, this decrease represents a reduction of approximately 25%. Note that the reduction in β-cell size was confirmed by E-cadherin/insulin co-immunostaining (Supplementary Fig. 2C).

Animal studies clearly demonstrated that poor nutrition during gestation irreversibly leads to reduced numbers of cells in tissues such as the pancreas (27). Here we evaluated whether the decreased cell proliferation observed in vivo persists in LP neoformed islets in culture. Indeed, our measurements revealed that the size of islets neoformed from fetal pancreas from the LP group is reduced compared with those from control pancreas (Fig. 4A). Further, insulin content per islet is diminished in neoformed islets in the LP group (Fig. 4B), whereas relative insulin mRNA expression is preserved (Fig. 4C). Together these data suggest that cell proliferation is decreased in LP neoformed islets. Because of the scant quantity of available material, we were unable to accurately measure levels of critical proteins in the cell cycle, and hence we used quantitative PCR to examine their mRNA expression (Fig. 4D). While cyclin D2, which accumulates during the G1 phase, was increased in neoformed islets derived from animals in the LP group, cyclin A2, which is necessary for progression into mitosis, was reduced. In addition, the expression of an important inhibitor of G1 cyclin/cdk complexes, p57, is increased in neoformed islets from the LP group (Fig. 4D). These data are compatible with an impaired β-cell proliferation of neoformed islets from the LP group. To determine whether miR-375 might be involved in perturbing the regulation of β-cell replication in neoformed islets from the LP group, we examined its expression. Notably, we found that miR-375 expression is significantly increased in neoformed islets from the LP group compared with controls (Fig. 4E). Taken together, our data strongly suggest that β-cell alterations induced by maternal protein restriction are not the result of direct regulation by the maternal milieu. Rather, they represent changes transmitted to the next generation of cells.

To investigate whether miR-375 interferes with cell proliferation and/or insulin secretion, we mimicked the changes in miR-375 expression observed in LP group progeny by transfecting dissociated primary rat islet cells with mature miR-375. Cell proliferation and insulin secretion were measured 72 h after transfection. Similar to our data from the INS-1E cell line (Supplementary Fig. 3), miR-375 expression in dissociated primary islet cells reduces levels of the PDK-1 protein (Fig. 5A and B) and inhibits cell proliferation by approximately 40% (Fig. 5C). Moreover, increased miR-375 expression decreases glucose-induced insulin secretion in cultured primary islet cells (Fig. 5D) without affecting insulin content (data not shown), confirming observations from the insulinoma MIN6 cell-line made by Poy and colleagues (28,29).

Since increased miR-375 expression and reduced proliferation are maintained in neoformed islets from the LP group, we evaluated whether this was also the case in adult 3-month-old male offspring. Importantly, we found that miR-375 expression is increased in isolated islets from 3-month-old animals in the LP group (Fig. 6A). Accordingly, mRNA levels of PDK-1 and myotrophin (MTPN), another miR-375 target (28), are diminished (Fig. 6B). As expected from the decrease in PDK-1, β-cell mass and proliferation are decreased (Fig. 6C and D). Consistent with MTPN and PDK-1 reduction, the β-cell functioning is deteriorated in 3-month-old animals in the LP group, as glucose-stimulated insulin secretion is severely blunted (Fig. 6E). Given the diminished β-cell mass and insulin secretion in LP group progeny, we studied glucose homeostasis in 3-month-old offspring. As shown in Fig. 6F, offspring from the LP group are glucose intolerant after an oral glucose challenge. However, insulin injection did not reveal hormone resistance in LP compared with control progeny at 3 months of age (data not shown), indicating that the glucose intolerance is due to insufficient insulin secretion.

To address whether miR-375 mediates the pancreatic alterations in LP group progeny, we used a loss-of-function approach with anti-miR-375 oligonucleotides to reverse the elevated expression of miR-375 in islet cells derived from 3-month-old rats in the LP group. miR-375 is upregulated in islet cells derived from LP animals (Supplementary Fig. 4), and glucose-induced insulin secretion and cell proliferation are reduced (Fig. 7A and B). Of significance, anti-miR-375-driven normalization of miR-375 levels (Supplementary Fig. 4) in dissociated islet cells derived from LP animals improved glucose-induced insulin secretion (Fig. 7A) without affecting insulin content (data not shown) and cell proliferation (Fig. 7B). The latter is accompanied by a return to normal levels of PDK-1 protein (Fig. 7B). Taken together, these data show that restoration of normal miR-375 levels in islets from LP animals tends to correct key deviations of the β-cell phenotype induced by maternal protein deficiency.

The biological importance of miRNAs has been placed in the limelight following the demonstration of diverse and profound phenotypic alterations upon changes in their expression. These modifications are associated with perturbed development and pathological situations (13). The data presented here demonstrate that an unfavorable environment during fetal development results in changes in programmed miRNA expression in the endocrine pancreas of the progeny. Among the miRNAs with aberrant expression, miR-375, which regulates the proliferation and the size of β-cells, contributes, at least in part, to the reduction in fetal β-cell mass observed at birth. Our results from adult 3-month-old animals reveal a sustained increase in miR-375 expression continuously affecting β-cell physiology and function. We show that early environmental factors, such as maternal nutrition, durably influence pancreatic expression of miRNAs, including miR-375, which affects β-cell physiology and controls long-term health.

miR-375 is a highly conserved miRNA that was cloned from a mouse insulinoma β-cell line (28). miR-375 is expressed in β-cells as well as in non–β-cells of the pancreas (30–32) and in the pituary gland (33). In human and mouse islets, miR-375 seems to be the miRNA with the most robust expression (30).

Using mice lacking miR-375 Poy et al. (32) reported that miR-375 deletion influences not only β-cell mass but also α-cell mass by regulating a cluster of genes controlling cellular growth and proliferation. Indeed, in normal situations, mice lacking miR-375 exhibit an increased number of α-cells, whereas the loss of miR-375 has little influence on β-cell mass. When the metabolic demand increases, however, genetic miR-375 deletion in mice counteracts the normally occurring β-cell hyperplasia (32). At first glance this result seems to vary from our observations. However, several studies have reported that miR-375 acts not as a positive regulator of cell proliferation but rather as a negative modulator (24,34–36). At least two scenarios could account for this apparent divergence. One possible explanation relates to differences among animal species. Using mice models, Tattikota and colleagues (37) reported that the effect of miR-375 on β-cell proliferation was mostly attributed to its target Cadm1 (cell adhesion molecule 1). As shown in Supplementary Fig. 5, Cadm1 expression is localized in all human and rat islet cells, while its expression is particularly perceptible at the periphery of mouse islets corresponding to α-cells. This could explain the robust influence of miR-375 deletion on α-cell mass under normal physiological conditions (37). When the metabolic demand augments (e.g., ob/ob mice), it is possible that the reduced β-cell mass results from α-cell dysfunction. Hence the nature of miR-375 action on β-cell proliferation could be different in mice compared with humans and rats.

Another scenario relates to recent views that miRNAs are responsive to cellular and extracellular stress and are used by cells to adjust to changes in their environment(38). A pioneering example in Drosophila revealed that miR-7 loss had no detectable effect on photoreceptor development under uniform laboratory conditions, but miR-7 becomes necessary under conditions of temperature fluctuation (39). It is thus conceivable that, under intensified stress conditions caused by increased metabolic pressure, miR-375 can negatively influence cell proliferation, whereas under physiological circumstances it fosters cell proliferation.

In addition to its role in β-cell proliferation, miR-375 is a regulator of chief β-cell functions. Indeed, forced miR-375 expression in insulinoma cells leads to reduced glucose-stimulated insulin secretion without interfering with the effect of glucose action on ATP production and intracellular calcium concentrations. To be more precise, miR-375 interacts with a series of gene products, including MTPN mRNA, governing insulin granule fusion with the plasma membrane, and thereby inhibits exocytosis (28,29). In neoformed islets from LP group fetuses, we failed to observe a difference in insulin secretion compared with the control group (data not shown). This is probably because of the variable ability of neoformed pancreatic islets to respond to glucose. Indeed, in fetal islets the acquisition of stimulus/secretion coupling of insulin in response to glucose occurs after birth. While some reports view neonatal islets as developmentally immature (40), others explain the weak responsiveness to glucose by a high activity at low glucose concentrations but a lower one at high glucose concentrations (41). In contrast to neoformed pancreatic islets, islets isolated from 3-month-old progeny in the LP group, with increased miR-375 expression, show a lower fractional insulin release in response to glucose. At the same time MTPN and PDK-1 mRNA are decreased. Importantly, normalization of miR-375 levels restores insulin secretion in islet cells of 3-month-old progeny in the LP group.

One of the most intriguing and important aspects of disease programming concerns the transmission of the disorder (42). Such phenomenon seems to occur in the animals in our LP group, where increased miR-375 expression observed in fetal islets persists in neoformed islets. Indeed, a similarly augmented level of miR-375 prevailed in neoformed islets from LP animals after 7 days of culture under the same conditions as control islets. Furthermore, in agreement with previous publications, cell proliferation seems to be impaired in neoformed islets from LP animals (43–45). Indeed, neoformed islets from progeny in the LP group were smaller than neoformed islets from control progeny, and a select set of genes involved in cell cycle regulation has altered expression, suggesting a reduction in cell proliferation similar to the in vivo situation. Most notably, 3-month-old progeny in the LP group display persistently increased miR-375 expression, reduced β-cell proliferation, and decreased glucose-induced insulin secretion. As a whole our data indicate that the elevated miR-375 level is unlikely to result from direct regulatory signals from the maternal milieu; rather, they reflect the occurrence of a developmental memory in the islet cells that recalls the intrauterine protein malnutrition. This induces long-lasting damaging effects on the pancreas, resulting in reduced β-cell proliferation and function.

Our results of miR-375 misexpression in endocrine pancreas from the LP animals are supported by the existence of epigenetic marks adjacent to the miR-375 coding region. Indeed, the miR-375 gene is epigenetically regulated by DNA methylation, consistent with the presence of two large CpG-rich regions (46). Moreover, the miR-375 promoter region contains several consensus binding sites for transcription factors; these have been implicated in the development and function of islets such as hepatocyte nuclear factor-6 and insulinoma-associated 1 (47). Pancreatic and duodenal homeobox-1 also was found to interact with the upstream region of the miR-375 gene (48). Most of these transcription factors are themselves regulated by epigenetic mechanisms (49).

In conclusion, we propose that under deleterious in utero conditions, the altered control exerted by miRNAs on gene regulation increases the postnatal risk of β-cell failure and hence of T2D. This augmented risk could reflect a reduced ability of β-cells to contend with stress because of a deficiency in adaptive β-cell mass and function. While biological systems are characterized by robustness against environmental changes to ensure proper development and health, failure of protective mechanisms sets the stage for exacerbated disease risk. A groundbreaking concept in this context is that miRNAs are instrumental in generating biological robustness (39). Dovetailing with this view, our data suggest that the level of miR-375 is key to establishing adequate β-cell mass and function, which are necessary for preserving the homeostatic health of the organism. Taking into account the fact that miR-375 has been identified as a diabetes-related circulating miRNA (50) and that miR-375 misexpression is observable before and after birth in animals with retarded growth, miR-375 seems to be a promising biomarker of β-cell failure.

Acknowledgments. The authors thank Phillipe Halban, University of Geneva, for providing the 804G matrix.

Funding. Our research was supported by INSERM, Université Nice Sophia Antipolis, Conseil Régional PACA and Conseil Général des Alpes-Maritimes; by the European Foundation for the Study of Diabetes (EFSD/Lilly, European Diabetes Research Programme 2011); and by the Agence Nationale de la Recherche (grant no. RPV12004AAA).

Duality of Interest. Our research was also supported by Aviesan/AstraZeneca’s “Diabetes and the Vessel Wall Injury” program. No other potential conflicts of interest relevant to this article were reported.

Author Contributions. O.D. and C.H. researched data, wrote the manuscript, and contributed to discussion. N.G. and V.C. researched data. S.P. researched data on human tissue shown in supplementary figures. E.V.O. wrote the manuscript and contributed to discussion. E.V.O. is the guarantor of this work and, as such, had full access to all the data and takes full responsibility for the integrity of data and the accuracy of data analysis.