Excess of glucocorticoids (GCs) during pregnancy is strongly associated with the programming of glucose intolerance in the offspring. However, the impact of high GC levels on maternal metabolism is not clearly documented. This study aimed to test the hypothesis that mothers exposed to elevated levels of GCs might also display long-term disturbances in glucose homeostasis. Dexamethasone (DEX) was administered noninvasively to the mothers via drinking water between the 14th and the 19th days of pregnancy. Mothers were subjected to glucose and insulin tolerance tests at 1, 2, 3, 6, and 12 mo postweaning. Pregnant rats not treated with DEX and age-matched virgin rats were used as controls. Pancreatic islets were isolated at the 20th day of pregnancy and 12 mo postweaning in order to evaluate glucose-stimulated insulin secretion. The expression of the miR-29 family was also studied due to its responsiveness to GCs and its well-documented role in the regulation of pancreatic β-cell function. Rats treated with DEX during pregnancy presented long-term glucose intolerance and impaired insulin secretion. These changes correlated with 1) increased expression of miR-29 and its regulator p53, 2) reduced expression of syntaxin-1a, a direct target of miR-29, and 3) altered expression of genes related to cellular senescence. Our data demonstrate that the use of DEX during pregnancy results in deleterious outcomes to the maternal metabolism, hallmarked by reduced insulin secretion and glucose intolerance. This maternal metabolic programming might be a consequence of time-sustained upregulation of miR-29s in maternal pancreatic islets.
- insulin secretion
- cellular senescence
glucocorticoid (gc) hormones are essential for proper fetal development (35) as well as for maternal homeostatic resetting after delivery (3, 32, 46). On the other hand, fetal exposure to excessive GCs leads to intrauterine growth restriction (IUGR), which increases the risk of chronic metabolic diseases later in life, such as obesity and type 2 diabetes, then reflecting the noxious action of GC excess on metabolic programming of the offspring. Glucose intolerance and impaired insulin secretion are usually found in both rodents and humans that had IUGR induced by prenatal GC excess (39, 47, 62).
Metabolic programming of the IUGR phenotype is strongly associated with an adverse fetal environment such as maternal nutritional deficiency, restricted placental blood supply, stress, and synthetic GC therapy (57, 9, 62). It is noteworthy that all of these states share excessive maternal GC levels as a common feature (63). Several studies have shown that epigenetic mechanisms such as DNA methylation and histone modifications are underlying mechanisms for the development of IUGR-linked metabolic diseases (reviewed in Ref. 67). The involvement of microRNAs (miRNAs) on IUGR outcomes is still elusive (6, 20, 66).
miRNAs are endogenous single-stranded small noncoding RNA molecules that control gene expression by binding to the 3′-untranslated region (UTR), thereby repressing translation or decreasing messenger RNA stability. Deregulation of miRNA expression in several tissues has been related to the pathogenesis of type 2 diabetes. Due to its key role in the control of glucose homeostasis, miRNA expression in pancreatic islets has drawn special attention (17). Altered expression of miR-375 (7, 14, 15, 28, 29, 34, 54), miR-7 (71), miR-30d (73), miR-29 (4, 5, 55, 60), and miR-338–3p (25), among others, has been demonstrated to correlate with pancreatic β-cell failure.
Numerous studies have been focusing on the metabolic programming induced by gestational GC excess, with special attention to the mechanisms leading to the late-onset development of metabolic syndrome in offspring subjected to IUGR. However, the long-term impact of gestational GC excess on maternal metabolism is not well documented. We therefore examined whether GC excess during pregnancy would alter maternal energy metabolism and pancreatic islet function later in life. We also sought correlative changes on miRNAs and their putative targets, particularly those that were demonstrated to play a key role on pancreatic β-cell function.
MATERIALS AND METHODS
Experimental design and animal treatment.
Nulliparous Wistar rats at 8 wk of age were housed in a temperature-controlled room (22 ± 2°C) with the lights on from 7:00 AM to 7:00 PM; standard chow and water were available ad libitum. Rats were allowed to acclimate for 2 wk in our animal facility prior to being randomized into four groups. After habituation, two groups of female rats were housed with male rats for 5 days (two female with one male per cage). The concomitant presence of spermatozoa and estrous cells in a vaginal lavage indicated day 0 of gestation. Pregnant rats were housed separately until weaning but remained in visual, olfactory, and auditory contact with other animals at all times. Daily food intake was assessed at the 13th and at the 20th days of pregnancy by calculating the variation in chow mass within a 24-h interval (measurements started at the moments of lights on); the same protocol was performed in age-matched virgin rats. Animals were allowed to deliver undisturbed. On postnatal day 21, offspring were weaned. The groups of age-matched virgin rats were housed in the same environment. Treatment consisted of administration of 0.2 mg·kg−1·day−1 water-soluble dexamethasone (Sigma-Aldrich, St. Louis, MO) diluted in the drinking water from the 14th to the 19th day of pregnancy or for 6 days for virgin rats. Random glucose levels were monitored during the treatment every morning. Body weights were measured during the treatment at weaning and at 1, 2, 3, 6, and 12 mo after weaning. All studies were performed according to the guidelines of the Brazilian College for Animal Experimentation (COBEA) and approved by the Ethics Committee on Animal Use at the Institute of Biomedical Sciences, University of São Paulo, Brazil.
Intraperitoneal glucose tolerance test.
Rats were fasted for 12 h prior to glucose injection (2 g/kg ip of a 20% solution of d-glucose). Blood samples were collected from the tail at 0, 10, 15, 30, 60, and 120 min for the measurement of blood glucose. The area under the curve (AUC) of blood glucose vs. time was calculated above each individual baseline (basal glycemia) to estimate glucose tolerance.
Intraperitoneal insulin tolerance test.
Fasted rats (12 h) were subjected to insulin injection (2 IU/kg ip). Blood samples were collected from the tail at 0, 5, 10, 15, 20, 25, and 30 min for the measurement of blood glucose. The constant rate for glucose disappearance (KITT) was calculated from the slope of the least squares analysis of the blood glucose concentrations during the linear phase of decay.
Pancreatic islet isolation and insulin secretion.
Rats were narcotized in a CO2 chamber and euthanized by decapitation. Trunk blood was collected to quantify circulating insulin levels, and islets were isolated after perfusion and digestion of the exocrine pancreas with collagenase solution (10). After isolation, groups of five islets were preincubated at 37°C for 30 min in 0.5 ml of Krebs-Henseleit buffer saturated with a mixture of O2 (95%) and CO2 (5%) and containing 0.2% BSA and 5.6 mM glucose. The medium was replaced with a fresh buffer with 0.2% BSA in the presence of 5.6, 8.3, 11.1, or 16.7 mM glucose and incubated at 37°C for 1 h. At the end of the experiment, the medium was collected and insulin measured by radioimmunoassay using rat insulin as the standard.
RNA extraction and qPCR.
Total RNA was extracted from a pool of ∼200 freshly isolated islets using TRIzol reagent, as previously described (10). Extracted RNA was eluted in RNase-free water, treated with Turbo DNA-free (Ambion, Austin, TX), and quantified by spectrophotometry.
The analysis of miRNA expression was adapted from a previously described method (64). For this, 250 ng of total RNA was polyadenylated using E. coli poly(A) polymerase (New England Biolabs, Hitchin, Hertfordshire, UK) according to the manufacturer's instructions at 37°C for 10 min. Then, poly(A) polymerse was heat inactivated (65°C, 5 min) and annealed to the adaptor [5′-GGCCACGCGTCGACTAGTAC(T)12-3′] by sequential incubations at 60°C (5 min) and then at 25°C (2 min). Annealed samples were chilled and used for standard reverse transcription (Improm-II; Promega, Madison, WI) before setting up of qPCR reactions.
For mRNA expression analysis, 1 μg of total RNA was reverse-transcribed using Improm-II reverse transcriptase (Promega) and random primers, according to the manufacturer's instructions. cDNAs were used for qPCR as follows.
Real-time amplifications were performed using Kapa SYBR Fast DNA polymerase (Kapa Biosystems, Boston, MA) and following standard procedures (32). The primer sequences for miRNA amplification were as follows: antisense (universal): 5′-GGCCACGCGTCGACTAGTAC-3′; miR-29a sense: 5′-TAGCACCATCTGAAATCGGTTA-3′; miR-29b sense: 5′-TAGCACCATTTGAAATCAGTGTT-3′; miR-29c 5′-TAGCACCATTTGAAATCGGTTA-3′; U1 snRNA sense: 5′-CCAGGGCGAGGCTTATCCATTGC-3′; 18S rRNA sense: 5′-GACTCAACACGGGAAACCTCACC-3′. All reactions were performed using 60°C for the annealing step of amplification.
For mRNA expression analysis, primer sequences, accession number, TM, and product lengths were as follows: p53 (NM_030989) sense: 5′-CGTTGCTCTGATGGTGACGG-3′ and antisense: 5′-AGCGTGATGATGGTAAGGATGG-3′ (57°C; 230 bp); Bbc3 (NM_173837) sense: 5′-CGGCGGAGACAAGAAGAGCAAC-3′ and antisense: 5′-GGCACCTAGTTGGGCTCCATTTC-3′ (58°C, 128 bp); Ercc6 (NM_001107296) sense: 5′-CATCATAGATGGGTCCAGTCCG-3′ and antisense: 5′-TGGGAACACCAGATGTTGCC-3′ (56°C, 128 bp); Gadd45a (L32591) sense: 5′-TCAACATCCTGCGGGTCAGC-3′ and antisense: 5′-TGTGGGTTCGTCACCAGCAC-3′ (58°C, 133 bp); Nfkb1 (NM_001276711) sense: 5′-CTCAAGAACAGCAAGGCAGCAC-3′ and antisense: 5′-AGAGGTGTCGTCCCATCGTAGG-3′ (59°C, 266 bp); Rpl37a (X14069) sense: 5′-CAAGAAGGTCGGGATCGTCG-3′ and antisense: 5′-ACCAGGCAAGTCTCAGGAGGTG-3′ (57°C, 290 bp).
Values of miRNA and mRNA expression were normalized using the geometric mean calculated from the internal control genes (snRNA U1 and 18S rRNA for miRNA; 18S rRNA and Rpl37a for mRNA). Fold changes were calculated by the 2−ΔΔCT method.
Protein extraction and immunoblotting.
Pools of ∼350 freshly isolated islets were processed for Western blotting as described elsewhere (10). The primary antibodies were anti-p53 (BD Pharmigen, cat. no. 554157), anti-syntaxin-1a (Stx-1a; Abcam, cat. no. ab78539, Cambridge, UK), and anti-α-tubulin (Invitrogen, cat. no. 32–2500, Carlsbad, CA). A secondary antibody conjugated with horseradish peroxidase (Bio-Rad, Hercules, CA) was used, followed by the chemiluminescent detection of the bands on X-ray-sensitive films. Optical densitometry was performed using the Scion Image analysis software (Scion, Frederick, MD).
Rat pancreata were carefully excised, cleared from lymph nodes and fat, and separated into a duodenal part (“head”) and a splenic part (“tail”). After being weighed, head and tail parts were fixed in Bouin's solution for 12 h, dehydrated in ethanol and embedded in paraffin. Sequential sections (5 μm) of the tail parts were obtained with the use of a rotary microtome and placed on individual silanized slides. Next, sections were deparaffinized in xylene and rehydrated. Endogenous peroxidase activity was quenched by incubating the slides with 3% H2O2 for 10 min. Sections were washed with 0.1 M phosphate buffer (pH 7.4), blocked with 5% nonfat dried milk (diluted in phosphate buffer containing 0.05% Tween 20), and incubated with guinea pig anti-insulin antibody (1:150) (Dako, cat. no. A0564, Glostrup, Denmark) at 4°C overnight. After a washing, the sections were incubated with biotinilated anti-guinea pig IgG at room temperature for 1 h. The streptavidin-biotin complexes were detected with diaminobenzidine (DAB) solution (0.1% DAB and 0.02% H2O2 in PBS). Finally, the sections were rapidly stained with Harris' hematoxylin and mounted for microscopic observation (LEICA DM 5000 B). Images were acquires using a CCD camera (LEICA CTR 5000) and the software LEICA Q Win Plus v. 3.2.0. Randomly 12–15 fields from tail parts of each animal were analyzed. Pancreatic islet mass (mg per pancreas) was calculated by multiplying relative pancreatic islet area (percentage of pancreatic islet in each sliced pancreas area) by pancreas weight. The number of pancreatic β-cell relative to number of islet cells was calculated by dividing the number of insulin-positive cells by the number of cells in the islet of a sequential section.
The results are presented as means ± SE. Comparisons were performed using one-way ANOVA followed by Tukey-Kramer post hoc testing (INStat GraphPad Software, San Diego, CA). P values of <0.05 indicate a significant difference.
Changes in body weight, random and fasting glycemia, insulinemia, and glucose tolerance induced by DEX treatment.
Six-day treatment with DEX led to a weight loss in virgin rats and avoided the weight gain found in pregnant rats between the 14th and 19th days of gestation. Untreated pregnant rats showed no changes in body weight throughout the 21 days of lactation. On the other hand, rats that received DEX during pregnancy showed a marked weight gain between the 1st and the 14th day of lactation. At the 14th postpartum day, the body weight of rats that received DEX during pregnancy was similar to that of untreated mothers (Fig. 1A).
Aside from changes in maternal body weight during the last week of pregnancy, pups from DEX-treated mothers had a significantly low birth weight (4.4 ± 0.4 vs. 7.2 ± 0.2 g in nontreated group, P < 0.0001).
A significant but transient increase in random glucose levels was seen in virgin rats treated with DEX between the 2nd and the 5th days of treatment. Glucose levels were lower in pregnant rats irrespective of the treatment (Fig. 1B). At the end of treatment, only virgin rats that received DEX showed glucose intolerance, as observed by timeline changes in glucose levels after a glucose load and the AUC values (Fig. 1, C and D, respectively).
Daily food intake of pregnant rats was higher than that of virgin rats prior to the beginning of treatment with DEX (day 13 of pregnancy). One day after the end of treatment, DEX induced a reduction in food intake of virgin but not in pregnant rats. Insulin levels were similar between virgin and pregnant rats before the beginning of DEX treatment. One day after the end of treatment, insulin levels of virgin+DEX were higher than in virgin rats. Pregnant+DEX rats also exhibited higher insulin levels than untreated pregnant rats. As expected, insulin levels of pregnant rats at day 20th of pregnancy were higher than in age-matched virgin rats. Also, at the end of treatment, fasting glucose levels were higher in virgin+DEX trhan in virgin rats. Fasting glucose levels were similar in pregnant and pregnant+DEX rats, and both were lower than in virgin rats (Table 1).
Late onset of glucose intolerance induced by DEX treatment at the end of pregnancy was associated with impaired glucose-stimulated insulin secretion (GSIS) but not with insulin resistance.
A glucose tolerance test was performed at 1, 2, 3, 6, and 12 mo after weaning and in age-matched virgin rats previously treated or not with DEX. At 1 and 2 mo after weaning, glucose tolerance was similar among all groups (Fig. 2, A and B). At 3, 6, and 12 mo after weaning, glucose intolerance was found in rats treated with DEX during pregnancy. These changes can be seen in timeline changes of glycemia and in the AUC (Fig. 2, C–E). In parallel, an increase in body weight was observed 3, 6, and 12 mo after weaning in mothers that were treated with DEX during gestation (Fig. 2F).
To determine the mechanisms underlying the long-term appearance of glucose intolerance in rats treated with DEX during pregnancy, we next assessed whole body insulin sensitivity. An insulin tolerance test was performed at 1, 2, 3, 6, and 12 mo after weaning and in age-matched virgin rats previously treated or not with DEX (Fig. 3). Although no differences were found in the KITT, the rapid increase in glucose levels after insulin injections (which was probably due to a stress response) was transitorily suppressed in virgin+DEX rats. In this way, glucose levels obtained 5 min after insulin injections were lower in virgin+DEX rats (vs. the other groups) specifically 3 and 6 mo after the end of the treatment.
Increased glucose responsiveness is a hallmark of the adaptive changes of pancreatic islets to pregnancy (3, 52). To explore both early- and late-onset effects of GC excess in β-cell function, we performed GSIS in islets isolated from rats at the end of pregnancy and 1 yr after. One day after the end treatment, 16.7 mM GSIS by islets from pregnant+DEX rats was lower than that from untreated pregnant rats. As expected, islets from untreated pregnant rats were more responsive to stimulatory glucose concentrations than age-matched virgin rats. DEX treatment of virgin rats did not change insulin secretion by pancreatic islets (Fig. 4A).
Twelve months after weaning, insulin secretion stimulated by 16.7 mM glucose was similar between rats that underwent pregnancy without treatment and age-matched virgin rats. Importantly, insulin secretion stimulated by both 11.1 and 16.7 mM glucose was found to be reduced 12 mo after weaning in rats that were treated with DEX during pregnancy (Fig. 4B).
Long-term changes in mir-29 expression in islets of rats treated with DEX during pregnancy correlates with impaired insulin secretion.
In a search for the mechanism responsible for the long-term impairment of insulin secretion we screened changes in islets' miRNA profile induced by DEX treatment. The contents of miR-29a and miR-29c were reduced in islets of virgin rats 1 day after the end of DEX treatment (Fig. 5, A and G). However, ∼8 and 12 mo after the end of DEX treatment, miR-29 expression showed different patterns. In DEX-treated virgin rats, miR-29a returned to values similar to that of control virgin rats (Fig. 5, B and C), whereas miR-29c significantly increased (Fig. 5, H and I) and miR-29b showed a transient increase specifically 8 mo after treatment (Fig. 5, E and F). On the other hand, when DEX treatment was performed during pregnancy, we observed an increase in miR-29s in islets 1 day after the end of the treatment. In that case, the modulation was characterized by a sustained response, as miR-29s remained elevated at 8 and 12 mo after DEX treatment (Fig. 5, A–I).
Long-term changes in expression of putative targets of mir-29 in islets of rats treated with DEX during pregnancy correlates with impaired insulin secretion.
As higher body weigh gain and glucose intolerance appeared at the 3rd month after weaning in pregnant+DEX rats, and increased miR-29 expressions were found in all periods studied, we performed the next experimental protocols at an intermediate time point (8 mo after weaning).
The transcriptional factor p53 has already been described to be involved in pancreatic β-cell viability (41). Importantly, p53 was also described as an indirect target that is upregulated by miR-29 (27). The content of p53 was increased at the mRNA and protein levels in islets of pregnant rats treated with DEX 1 day after the end of the treatment (Fig. 6, A and C, respectively). The levels of p53 mRNA remained elevated 8 mo after weaning in islets from DEX-pregnant rats (Fig. 6B). The content of p53 was increased later in life after DEX treatment independently of whether it occurred during pregnancy or not (Fig. 6D).
We next assessed the expression of Bbc3 [the gene that encodes p53 upregulated modulator of apoptosis (Puma)], a direct target of p53 transcriptional activity (45). The level of the Bbc3 mRNA was increased in pancreatic islets of rats treated with DEX during pregnancy. This upregulation was detected 1 day after the end of DEX treatment and 8 mo after weaning (Fig. 6, E and F, respectively).
Stx-1a is a t-SNARE protein that plays a crucial role in GSIS (44, 50). Stx-1a mRNA is directly targeted by miR-29a, thereby influencing insulin secretion (5). Accordingly to an increase in miR-29a, treatment with DEX during pregnancy resulted in reduced Stx-1a protein levels in maternal pancreatic islets 8 mo after weaning (Fig. 6H).
DEX treatment during pregnancy did not change islet morphology.
It is well established that during normal pregnancy β-cell mass increases, peaking at the end of the second gestational period, which corresponds in the rat to the 13th day of pregnancy (31, 52). To verify whether the late-onset metabolic disturbances found in DEX-treated dams are associated with morphological changes, we performed an immunohistochemical analysis of pancreatic islets at 8 mo after treatment.
The pancreatic islet architecture and insulin-positive β-cells were relatively conserved in all groups (Fig. 7A). β-Cells were always typically detected within the islet core, and islet size and number were unchanged 8 mo after weaning. Histomorphometric measurements showed that the mean islets' perimeters were less than 500 μm in the four groups [ranging from 414 ± 24 μm in pregnant+DEX to 488 ± 48 μm in virgin+DEX; not significant (NS)]. Pancreas mass relative to body weight was also similar among all groups (ranging from 0.36 ± 0.015% in pregnant to 0.48 ± 0.06% in pregnant+DEX; NS). Also, no significant changes were observed in average islet area, islet mass, or the percentage of β-cells per islet (Fig. 7, B–D, respectively).
Expression of cellular senescence-related genes exhibits long-term modulation in pancreatic islets of rats treated with DEX during pregnancy.
As miR-29s have been previously related to cellular senescence (38), and the decrease in GSIS is a feature of aged β-cells (58), we have searched for putative mechanisms that could strengthen the correlation between miR-29s and β-cell phenotype. The expression of Nfkb1, the gene that encodes the 50-kDa subunit of the NF-κB (nuclear factor κB DNA binding subunit), is upregulated in islets from untreated pregnant rats at gestational day 20. DEX treatment abrogated the Nfkb1 overexpression induced by pregnancy (Fig. 8A). One year after DEX treatment, the expression of Nfkb1 in islets from DEX-treated dams was significantly higher than from the other three groups (Fig. 8B).
Gadd45a (growth arrest and DNA damage protein 45a) and Ercc6 (excision repair cross-complementing rodent repair deficiency, complementation group 6) are two genes that have been associated with p53 signaling and cellular senescence (12, 18). One year after treatment, islets from DEX-treated dams showed an increase in Gadd45a and a decrease in Ercc6 expression (vs. the other three groups; Fig. 8C). At that time point, a significant increase in Ercc6 expression was found in islets isolated from untreated pregnant rats (Fig. 8C).
Numerous studies dealing with GC excess in pregnancy and metabolic programming have focused on late-onset metabolic syndrome in IUGR pups (39, 47, 57), but the impact of prenatal GC excess on maternal physiology has been neglected. Since mothers and fetuses are exposed to a similar gestational environment, we hypothesized that metabolic programming of maternal energy metabolism could also take place later in life. Here we demonstrate that, as already described in their offspring, mothers treated with DEX during the last third of pregnancy also develop long-term glucose intolerance associated with impaired GSIS. In addition, we show that miR-29 upregulation is an epigenetic process that is maintained until the late onset of glucose intolerance, thus being a putative mechanism for long-term pancreatic β-cell failure.
DEX treatment via drinking water reduced the body weight of virgin rats and abrogated maternal body weight gain during the last third of pregnancy, which attests to the effectiveness of our experimental protocol. As our protocol of treatment with DEX did not reduce food intake in pregnant rats, it is plausible to hypothesize that exogenous DEX administrated during the last week of pregnancy might avoid body weight gain by increasing energy expenditure. Also, glucose intolerance and random hyperglycemia observed in nonpregnant rats immediately after a chronic DEX treatment are phenotypic traits of DEX therapy (51). Finally, a significant reduction in pups' weight was observed, which is a hallmark of fetal metabolic programming due to maternal GC excess (40, 49).
Glucose homeostasis resetting of virgin+DEX rats later in life was not unexpected. It has been previously demonstrated that glucose intolerance and pancreatic β-cell dysfunction induced by DEX in male rats are reversed after discontinuation of the treatment (56). Importantly, either the interaction between DEX with some component of the gestational milieu or the changes in body weight gain induced by DEX in pregnant rats are possible causes for the long-term impacts of DEX treatment during pregnancy on pancreatic islet physiology. In accord with this hypothesis, a 65% food restriction during the last week of pregnancy in rats reduced body weight and caused maternal glucose intolerance (2).
Although we have shown that insulin secretion by pancreatic islets isolated at end of pregnancy from pregnant+DEX rats is reduced compared with pregnant rats, circulating insulin levels of pregnant+DEX dams were higher than in untreated dams. Similarly, others have described that DEX treatment of pregnant rats leads to hyperinsulinemia at the end of pregnancy with parallel reduction of GSIS by isolated pancreatic islets (23).
A plausible explanation for these in vivo and in vitro differences might rely on extrapancreatic adaptations induced by DEX during pregnancy that leads to increased in vivo insulinemia. It was already demonstrated that women treated with DEX during pregnancy display increased insulin levels due to increased autonomic inputs to the pancreatic islets (1). As we interpret it, this autonomic modulation is probably lost in experiments with isolated pancreatic islets.
Interestingly, the appearance of glucose intolerance and higher weight gain in pregnant+DEX rats was a delayed and a time-sustained phenomenon. Epigenetic processes are primary candidates for mechanisms that can stably modulate gene expression involved in metabolic programming. In the search for the underlying mechanism governing the late onset of glucose intolerance in pregnant-DEX rats, we hypothesized that alterations in miRNA expression could be involved. Our hypothesis was based on 1) previous work showing that synthetic GC can regulate miRNA expression in several cell types (65), 2) the pivotal role of miRNAs in pancreatic β-cell function (17, 26), and 3) our previous screening of DEX effects on miRNA expression in pancreatic islets (unpublished data). We chose the miR-29 family for further analysis due to its putative relevance to pancreatic β-cell function and maintenance (4, 5, 55, 60). Also, in our previous screening all members of miR-29 family were regulated by DEX.
The miR-29 family has been widely studied because of its deregulation in several processes such as tumorogenesis (72), organ/tissue fibrosis (22, 59), senescence (38, 69), and regulation of extracellular matrix (24). The plethora of miR-29 actions is due to a common characteristic of miRNAs: each miRNA can regulate hundreds of genes and each gene can be regulated by multiple miRNAs (30). This feature results in a complex combinatorial posttranscriptional regulation of gene expression and is reflected in a cell type- and context-specific response.
In pancreatic islets and insulinoma cells, miR-29 overexpression led to impairment in GSIS and β-cell apoptosis (4, 60). These reports showed that both proinflammatory cytokines and high glucose upregulated miR-29 expression, leading to pancreatic β-cell dysfunction. The long-term effects observed in DEX+pregnant rats are in agreement with these previous studies. The lack of significant β-cell dysfunction immediately after DEX treatment could be due to a weak direct effect of the synthetic GC on mothers' physiology, as seen in random glycemia and glucose tolerance tests. Of note, GCs inhibit inflammation by abrogating the activity of NF-κB, a central signaling molecule of proinflammatory cytokines (42), which could be counteracting an inflammatory dependent miR-29 network by the time of pregnancy.
In the nonpregnant state, DEX induced a short-term downregulation of miR-29a/c in pancreatic islets. DEX-induced miR-29a repression has already been described (70). On the other hand, miR-29c has been suggested as a signature miRNA in the diabetic environment (36). In kidney cells, knockdown of miR-29c prevented high-glucose-induced cell apoptosis, whereas forced expression of miR-29c induced cell death (36). Although not definitely evidenced by our study, we believe that the permanent differential modulation of the miR-29 family might influence the cellular senescence process of pancreatic β-cells. We consider this hypothesis because the miR-29 family was previously reported to mediate the senescence process in different cell types (38).
It has been suggested that the stress-activated p53 and NF-κB signaling pathways, acting in an antagonistic fashion, are key players in the regulation of cellular senescence (61). However, the tumor suppressor p53 is an important trigger of cellular senescence (11), and NF-κB signaling is involved in the induction of the senescence-associated secretory phenotype, a condition where senescent cells secrete proinflammatory factors (61). Indeed, it has been recently reported that both p53 and NF-κB can operate synergistically (33). Since elevated activation of NF-κB is related to a poor cellular redox state and the inflammatory process, it is possible that NF-κB could be accelerating the oxidative stress and/or an inflammatory response leading to β-cell dysfunction.
Gadd45a was the first well-defined p53 downstream gene that behaves like a DNA damage-inducible gene (12). In mammary tumor, Gadd45a activates the stress-induced JNK and p38 kinases, which participate in increased apoptosis and Ras-induced cellular senescence (68). The paralleled upregulation of Gadd45a and p53 in islets from rats exposed to GC excess during pregnancy reinforces the putative increase in p53 activity in DEX-treated mothers.
As mentioned above, p53 integrates different physiological signals. In response to different forms of cellular stress, p53 protein stabilizes and triggers multiple specific events to prevent damaged cells from undergoing tumor transformation. These events can lead to cellular senescence and apoptosis due to a permanent cell cycle arrest. It has been shown that the absence of the DNA-binding protein CSB (Cockayne syndrome protein) can stabilize p53 (18), which gives rise to pronounced cell fragility due the activation of p53 target genes related to the apoptotic commitment. CSB, an essential DNA-binding protein involved in the transcription-coupled DNA repair pathway is encoded by Ercc6, a gene that shows an inverse correlation to p53 expression in islets from DEX-treated mothers. Thus, it is possible that the downregulation of CBS and the upregulation of p53, as well as Gadd45a, could mediate premature β-cell senescence, as suggested in other cell types (reviewed in Ref. 19).
Importantly, aging of pancreatic islets displays two classic features, reduced secretory capacity combined with increased mass (58). As our data further demonstrate, reduced insulin secretion later in life by islets from pregnant+DEX rats is paralleled by a tendency in increase of pancreatic islet mass.
A mechanism that could explain the sustained miR-29 expression also involves the expression and activity of p53 (69). Many signal transduction pathways converge on p53 and result in differential regulation of its downstream targets. Because of its functional diversity and its importance to cell fate decisions, p53 levels and activity are tightly regulated through positive and negative feedback loops (27). Recent studies have shown that the cross talk between the p53 network and miRNAs involves the regulation of miRNA transcription/maturation by p53 as well as the role of miRNAs on the abundance and activity of p53 (69). Among them, it has been shown that miR-29 regulates the regulators of p53 to indirectly activate it (27).
Our present data do not allow us to determine whether p53 is upstream to the long-lasting miR-29 expression or if elevated miR-29 levels is maintaining increased p53 levels. However, Bbc3, the gene that encodes Puma, and Gadd45a mRNAs are directly induced by p53 (45, 12). Thus, the higher p53 content seen in DEX-treated mothers suggests increased activity of this transcription factor. We believe that positive feedback could be involved in the sustained p53/miR29 upregulation.
Increased miR-29 (60) as well as Puma levels (21, 41), had been associated with pancreatic β-cell apoptosis. Here, we did not further focused on pancreatic β-cell death, especially because we did not find any relevant change in pancreatic gross morphology. Instead, we focused on the exocytotic machinery, which is a well-recognized target of miRNAs (16, 37, 53).
Stx-1a is a t-SNAREs involved in insulin exocytosis. Both in rodents and humans, decreased expression of Stx-1a impairs GSIS (44, 50). Moreover, using a luciferase-based reporter gene assay, a recent study has demonstrated that miR-29a directly targets the Stx-1a gene in the 3′-UTR (5). In agreement with this study, we found that a decrease in Stx-1a protein levels paralleled GSIS impairment in pregnant+DEX rats later in life. Furthermore, our results show that increased miR-29 expression is associated with Stx-1a gene downregulation and decreased GSIS in an in vivo context.
Due to its intrinsic complexity, we cannot presently explain how the combination of pregnancy and GC therapy results in a long-lasting upregulation of miR-29s. However, given the late-onset nature of the impairment of pancreatic β-cell function in DEX-treated mothers, we believe that determining the changes in the gestational internal milieu responsible for this maternal programming requires further investigation. We can speculate that several factors are candidates for the disruption of maternal environment. For example, secretion of prolactin, the main regulator of pancreatic β-cell adjustment to pregnancy (8), is reduced by GCs (13). A fall in maternal plasma concentration of estradiol, a steroid known to increase insulin biosynthesis and to enhance GSIS (43), was also observed after antenatal DEX therapy (48).
All together, our data describe a long-term impairment of maternal pancreatic islet function caused by the use of DEX during pregnancy. DEX-induced metabolic programming accounts for weigh gain and glucose intolerance later in life. The present study also describes correlative data pointing to a putative mechanism underlying the late onset of inaccurate energy metabolism regulation, which involves an aberrant expression of miR-29s and p53 and a downregulation of Stx-1a in pancreatic islets.
This study was supported by the Research Foundation of the State of São Paulo (FAPESP) and National Council of Research (CNPq).
No conflicts of interest, financial or otherwise, are declared by the author(s).
Author contributions: P.R.G., M.F.G., L.A.V., M.Q.L., A.R.C., G.F.A., and S.B. conception and design of research; P.R.G., M.F.G., L.C.P., A.L.R., and S.C.R. performed experiments; P.R.G., L.A.V., M.Q.L., A.R.C., G.F.A., and S.B. analyzed data; P.R.G. and S.B. interpreted results of experiments; P.R.G. and S.B. prepared figures; P.R.G., G.F.A., and S.B. drafted manuscript; P.R.G., M.F.G., L.C.P., A.L.R., S.C.R., L.A.V., M.Q.L., A.R.C., G.F.A., and S.B. approved final version of manuscript; L.A.V., M.Q.L., A.R.C., and S.B. edited and revised manuscript.
We thank Marlene S. Rocha (University of Sao Paulo) for technical assistance.
- Copyright © 2014 the American Physiological Society