Epidemiological and experimental evidence suggests that a suboptimal environment during perinatal life programs offspring susceptibility to the development of metabolic syndrome and Type 2 diabetes. We hypothesized that the lasting impact of perinatal protein deprivation on mitochondrial fuel oxidation and insulin sensitivity would depend on the time window of exposure. To improve our understanding of underlying mechanisms, an integrative approach was used, combining the assessment of insulin sensitivity and untargeted mass spectrometry-based metabolomics in the offspring. A hyperinsulinemic-euglycemic clamp was performed in adult male rats born from dams fed a low-protein diet during gestation and/or lactation, and subsequently exposed to a Western diet (WD) for 10 wk. Metabolomics was combined with targeted acylcarnitine profiling and analysis of liver gene expression to identify markers of adaptation to WD that influence the phenotype outcome evaluated by body composition analysis. At adulthood, offspring of protein-restricted dams had impaired insulin secretion when fed a standard diet. Moreover, rats who demonstrated catch-up growth at weaning displayed higher gluconeogenesis and branched-chain amino acid catabolism, and lower fatty acid β-oxidation compared with control rats. Postweaning exposure of intrauterine growth restriction-born rats to a WD exacerbated incomplete fatty acid β-oxidation and excess fat deposition. Control offspring nursed by protein-restricted mothers showed peculiar low-fat accretion through adulthood and preserved insulin sensitivity even after WD-exposure. Altogether, our findings suggest a testable hypothesis about how maternal diet might influence metabolic outcomes (insulin sensitivity) in the next generation such as mitochondrial overload and/or substrate oxidation inflexibility dependent on the time window of perinatal dietary manipulation.
- energy metabolism
- insulin clamp
- metabolic programming
- perinatal nutrition
type 2 diabetes mellitus (T2DM) is becoming a worldwide epidemic that may affect around 4% of the world population by 2030 (3). It results from a complex set of interactions between genetic—contributing less than 1% of disease risk (58)—and environmental factors (48), such as perinatal nutrition. Individuals born with a low birth weight due to an intrauterine growth restriction (IUGR) are exposed to a 6-fold increase in the risk of developing T2DM and are 18 times more likely to present metabolic syndrome in adult life than individuals with a normal birth weight (26). Such observations led Hales and Barker to propose the “Thrifty Phenotype” hypothesis, also known as developmental origins of health and disease or metabolic programming. This postulates that poor nutrition during critical periods early in development triggers adaptations that improve offspring survival but have adverse metabolic consequences in adulthood, including increased hepatic glucose output, insulin resistance (IR), and impaired insulin secretion, three hallmarks of T2DM. These adverse metabolic consequences are associated with altered regulation of glucose and lipid metabolism in several tissues such as liver, skeletal muscle, and pancreas (75, 82).
Although the exact mechanisms underlying these metabolic disorders are not yet fully understood, emerging evidence suggests that mitochondrial biogenesis and/or activity (1) and metabolic inflexibility, defined as an inappropriate adjustment of mitochondrial fuel selection in response to nutritional signals (69), might contribute to T2DM. Moreover, we recently reported alterations in fatty acid metabolism in adult male rats born from protein-restricted mothers (4, 51). Therefore, we postulated that perinatal protein deprivation could permanently impact mitochondrial fuel oxidation with specific effects on insulin sensitivity (IS) depending on the time window of exposure (i.e., gestation and/or lactation). To address that hypothesis, the current study used a rat model of IUGR induced by maternal dietary protein restriction, a programming model validated in several studies (52, 64).
More recent than other global genotyping/phenotyping approaches, metabolomics is emerging as a valuable tool for generating new descriptors and/or mechanistic information related to complex biological processes as T2DM development (5, 19, 33). Herein, an untargeted mass spectrometry-based metabolomic approach was combined with targeted measurements to gain a broader understanding of metabolic and physiological outcomes in programming (4). Fasting blood biochemical and acylcarnitine profile that reflect fatty acid and amino acid oxidation (63) were combined with insulin sensitivity and body composition analysis in rats fed a balanced or an unbalanced diet (60). Because the liver is a key organ in glucose and lipid metabolism and the only site of ketone body biosynthesis, we also analyzed the hepatic expression of genes involved in energy homeostasis.
To our knowledge, this work is the first integrative study using metabolomics, acylcarnitines pattern, and hyperinsulinemic-euglycemic clamp to address the long-term effects of perinatal “programming”. We found strong evidence for alterations in mitochondrial fatty acid β-oxidation that could explain the progressive impairment in energy homeostasis and insulin sensitivity elicited by a nutritional insult inflicted at various perinatal time windows.
MATERIALS AND METHODS
Animal Experiments, Housing, and Diets
The experimental protocol applied (CEEA.2011.4) and previously described in detail (42) was carried out in accordance with current European regulation regarding the protection of animals used for experiments and approved by the French Veterinary Department and the Institut National de la Recherche Agronomique (Paris, France). Briefly, virgin female Sprague Dawley rats were time mated and received either ad libitum control regular (20 g of protein/100 g of food: C, n = 25) diet or an isocaloric, low-protein (8 g of protein/100 g of food: R; n = 25) diet throughout gestation and lactation. Both diets, previously used with success (18, 64), were purchased from Arie Blok (Woerden, The Netherlands). Low-protein diet leads to the delivery of low-birth weight pups, as described previously (18). At birth, female pups were discarded, and litter size was adjusted to eight male pups per dam. Pups born from protein-restricted dams were randomly assigned to C (RC group) or to R dams (RR) for lactation (Fig. 1A). During suckling period, pups born from control dams were adopted by C (CC) or R dams (CR). At D164, rats were split in two groups, one maintained on standard diet and the second one fed a Western diet (WD) containing 17% protein, 43% carbohydrate (10.2% corn-starch, 89.8% sucrose/maltodextrin), and 41% fat (Research diet, New Brunswick, NJ) for 10 wk, to obtain eight experimental groups: CC, CCWD, RC, RCWD, RR, RRWD, CR, and CRWD. The experimental design is depicted in Fig. 1B.
Feeding and Anthropometric Parameters
Relative food intake (kcal/g body wt) and food efficiency (g of body wt gain/kcal) were calculated between D164 and D220 for each group (42). Body composition was determined by dual-energy X-ray absorptiometry (DEXA) (Atlantic Bone Screen Platform, Nantes, France) at D209 after 7.5 wk of high-energy diet for WD rats. Before scanning, animals were anesthetized with Dorbene vet (20 μg/kg; Pfizer-Centravet, Plancoët, France). DEXA subdivides body composition into bone mineral mass, fat mass, and lean body mass.
Insulin Sensitivity and β-Cell Secretory Capacity
Whole-body insulin sensitivity was assessed at D63 and D227. At D63, a noninvasive method was preferred to avoid the risk of mortality associated with surgery for catheter placement since both techniques were applied on the same animals.
Oral glucose tolerance test.
At D63, after a 15-h overnight fast, all rats received a 1 g/kg body wt dose of glucose by gavage. Blood samples were collected from tail vein to determine blood glucose and insulin levels. Peripheral IS index (ISI) during an oral glucose tolerance test (OGTT; ISI0,120) was estimated using the equation: ISI0,120 = m/MPG/log(MSI), where MPG is mean plasma glucose concentration (mg/l, mean of the 0 and 120 glucose values from the OGTT), MSI is the mean serum insulin concentration (mU/l, mean of the 0 and 120 insulin values from the OGTT), and m is glucose uptake rate (mg/min), calculated as follows: glucose load (mg) + (0 min glucose − 120 min glucose, mg/l) × 0.19 (glucose space, L) × BW (body weight, kg)/120 min, as previously described in humans (25) and rats (73). ISI0,120 is highly correlated to the insulin sensitivity index determined with hyperinsulinemic euglycemic clamp, the gold standard method (r = 0.63) (25), considered as a method of choice in animals (47).
At D227, after 9 wk of receiving either the control or the Western diet, and after a 15-h overnight fast, rats from all groups were anesthetized with 2% isoflurane/oxygen (1.5 l/min), which has been previously shown to have, among all available veterinary anesthetics, the least effect on glucose and insulin levels (8, 28, 83). Basal glucose concentration was determined from tail vein blood before surgery, and a venous catheter was inserted into a jugular vein. Rats then underwent a 150-min hyperinsulinemic-euglycemic clamp with a 72 mU/kg insulin bolus over 1 min, followed by a continuous, intravenous infusion of human insulin (Actrapid Novo Nordisk, Bagsyaerd, Denmark) at a constant rate of 18 mU·kg−1·min−1 (14), which has previously been shown to achieve maximal stimulation of glucose uptake and completely suppress hepatic glucose production in control rats (59). The rate of infusion of a 20% glucose solution (Braun Medical SAS, Boulogne, France) was adjusted according to blood glucose measured at 5-min intervals throughout the 150-min clamp, so as to maintain basal plasma glucose levels. Nonesterified fatty acids (NEFA) levels were measured during the last 20 min of the clamp. IS was calculated as the rate of infused glucose normalized to lean body mass (LBM) (mg·kg LBM−1·min−1) during the steady-state phase of the clamp (31).
Homeostasis model assessment of β-cells.
The insulin secretory capacity of β-cells was estimated before the OGTT (D63) and before death (D234) using homeostasis model assessment (HOMA)-β index calculated using the following equation: HOMA-β = 20 × fasting insulin (mU/l)/[fasting glucose (mmol/l) − 3.5] (43, 79). HOMA-β is considered the best index of β-cell function in nondiabetic humans (2) and was recently used in Zucker rats (44).
Targeted Metabolic Variables
At D234, after a 15-h overnight fast, blood glucose concentrations from vein tail were monitored on vigil animals using a glucometer (Accu-CheckH active, Roche Diagnostics). Animals were then killed using inhaled CO2 followed by decapitation. Cardiac blood samples were collected in EDTA tubes (Pfizer-Centravet, Plancoët, France), and freshly prepared plasma was snap frozen in liquid nitrogen and stored at −80°C until analyzed for free NEFA (Wako NEFA-HR, Neuss, Germany), triacylglycerol (TAG), cholesterol (Biomérieux TG PAP 1000_Cholestérol RTU, Lyon, France), ketone bodies (Wako Autokit total ketone bodies, Neuss, Germany), lactate (lactic acid kit; Biosentec, D/L Auzeville-Tolosane, France) and insulin (rat insulin ELISA kit; Eurobio, Courtaboeuf, France). Free carnitine and short- (C2 to C5), medium- (C6 to C10) and long- (C12 to C18) chain acylcarnitines were determined after derivatization to their respective butyl-ester forms using isotope dilution electrospray ionization tandem mass spectrometry (ESI-MS/MS), as described previously (46).
Nontargeted Plasma Phenotyping by LC-HRMS Metabolomics
The MS-based metabolomic fingerprinting workflow used in the current study was previously applied and described by our group (4, 10). At D234, fasted plasma samples from all rats, were submitted to a preliminary filtration step (cut off at 10 kDa) before analysis by reverse-phase liquid chromatography coupled to high-resolution MS (HPLC-ESI-HRMS) using an Orbital trap instrument operating in positive or negative electrospray ionization mode and in full-scan acquisition mode at a 30,000 resolution of the full-width at half maximum (FWHM). Annotation of the generated metabolomic profiles and subsequent identification of putative metabolites/biomarkers of interest were achieved using an in-house reference databank (11).
RNA Isolation and Real-Time RT-PCR
Total RNA was extracted from frozen liver in 234-day-old rats with TRIzol reagent, according to manufacturer's instructions (Ambion RNA; Life Technologies, Paris, France). mRNA of genes implicated in glucose metabolism [liver pyruvate kinase (LPK), glucose transport (GLUT2), phosphoenolpyruvate carboxykinase (PEPCK), lactate dehydrogenase (LDHb)], lipid metabolism [fatty acid synthase (FAS), acétyl-CoA-carboxylase I (ACC1), fatty acid binding protein (FABP1a), carnitine-palmityl-CoA-transférase I (CPT1a), malonyl-CoA decarboxylase (Mylcd), acetyl-CoA carboxylase beta (Acacb), 3-hydroxy-méthyl-glutaryl-CoA-synthase (HMGc1)], insulin signaling (InsR2), and transcription factors involved in the regulation of these biological pathways [carbohydrate-responsive-element-binding-protein (ChREBP), forkhead box protein O-1 (FOXO1), peroxisome proliferator-activated receptor-γ coactivator 1 alpha (PGC1-α), sirtuin 1 (SIRT1), peroxisome proliferator-activated receptor-α (PPARα), sterol regulatory element-binding transcription (Srebf1)] were quantified by real-time RT-PCR. Primers were designed on the basis of the sequences available at the National Center for Biology Information gene bank using the “Perlprimer” program (v.1.1.17) and synthesized by Eurogentec (San Diego, CA). mRNA expression was calculated using the 2−ΔΔCt method after normalization with β-actin used as the reference gene (40). Amplification efficiencies of the different transcripts, including β-actin were first determined to validate the use of the CT method. The specificities of the PCR products were confirmed by single dissociation curves of the PCR products. To analyze the effect of maternal perinatal diet on gene expression in the offspring, regardless of their diet at adulthood, Ct values obtained for PCR amplification on CC rats were used as the calibrator group. To analyze the effect of Western diet, Ct values obtained for PCR amplification on CC, RC, RR, and CR groups under standard diet were used as a calibrator for CCWD, RCWD, RRWD, and CRWD groups, respectively.
Results are presented as means ± SE in tables and figures. Metabolomic data were processed using the open-source XCMS, as previously described (4) for nonlinear alignment of the generated metabolomic data and for automatic integration and extraction of the peak intensities for each detected features (ions of given mass-retention time [m/z; rt]). In targeted and nontargeted data, differences among groups were analyzed by the nonparametric Mann-Whitney U-test using GraphPad Prism software version 5.00, due to the nonnormality of the data, explained by the small size of the groups (n = 8 observations per group) combined to its interest in metabolomics data, as recently detailed (61). Differences with P < 0.05 were considered significant.
Phenotyping Consequences of Perinatal Programming
We previously showed that compared with CC, IUGR animals had a significantly lower birth weight (∼7%), remained thinner when nursed by dams fed a low-protein diet (RR), and caught up with controls at D6, when cross-fostered by normally fed mothers (RC) (42). Control pups nursed by dams fed a low-protein diet (CR) had a lower body weight than CC pups at D6. Adult IUGR rats fed a standard diet had a lower body weight, lower adiposity, and higher lean body mass than CC, although their food intake was similar (Table 1). Adult CR rats exhibited lower body weight than CC but comparable lean body mass and adiposity. Fasting circulating levels were higher for glucose but similar for insulin, NEFA, and cholesterol in RC, RR, and CR compared with CC (Table 1). Both IUGR groups had triglyceride concentrations similar to those in CC rats, whereas levels were lower in CR. Circulating ketone body and lactate levels did not differ between CC, RR, and CR groups but were lower for RC than CC.
Chronic ingestion of a WD tended to increase body weight, whole body, and abdominal adiposity in all groups, although only the two IUGR groups displayed a significant increase of fat mass and demonstrated higher feeding efficiency, compared with their standard-fed counterparts (Table 1, Fig. 1C). RRWD rats maintained a lower body weight compared with CCWD. Surprisingly, no change in lean and fat mass or difference in feeding efficiency was observed between CR and CRWD, despite higher relative food intake. Neither perinatal nutrition nor adult diet had any impact on bone mineral density. Overall, WD-exposure did not alter fasting plasma glucose, insulin, lactate, or triglyceride concentrations, but produced higher cholesterol and ketone body concentrations in all groups of rats when compared with their standard-fed counterparts and sustained lower triglyceride concentrations in CRWD than CCWD. In addition, RCWD had significantly lower circulating lactate than CCWD and lower ketone body levels than CCWD and RRWD. Both IUGR groups on a WD presented higher fasting NEFA concentrations than CCWD.
Postnatal Nursing by a Protein-Restricted Mother Enhances Insulin Sensitivity
IS was assessed using OGTT in young adulthood (D63), using a hyperinsulinemic-euglycemic clamp during adulthood (D227) in rats on a standard diet or a WD. At D63, the three groups that underwent perinatal dietary manipulation showed improved peripheral IS, as evidenced by significantly increased ISI0,120 index compared with CC (Fig. 2A) despite similar fasting glucose and insulin levels (data not shown). At D227, the hyperinsulinemic-euglycemic clamp showed increased whole-body IS only in RR and CR rats, compared with CC (Fig. 2C). HOMA-β was significantly lower for RC, RR, and CR rats compared with CC (Fig. 2D).
Chronic exposure to a WD induced a significant decrease in β-cell insulin secretory capacity in CCWD and RCWD rats (Fig. 2D) while HOMA-β was persistently low for RRWD. This was associated with unchanged whole body IS for CCWD, RCWD, and RRWD compared with their standard-fed counterparts (Fig. 2C). However, plasma NEFA concentrations during the steady-state period of the clamp were significantly increased in CCWD and RCWD (Fig. 2E). Interestingly, CRWD displayed a significant increase in whole-body IS compared with CCWD, RCWD, and their standard-fed counterparts.
Perinatal Maternal Protein Restriction Is Associated With Impaired Fatty Acid and Amino Acid Oxidation
At D243, under standard diet, fasting free- and C2-carnitine (derived from carbohydrate catabolism and from the ultimate product of β-oxidation, acetyl-CoA) (Table 2), as well as total- (Table 2), short- (Fig. 3C), medium- (Fig. 3D) and long-chain (Fig. 3E) acylcarnitines concentrations were similar between all rats except for the long-chain acylcarnitines, which were lower in RC compared with CC and RR. Furthermore, RC presented lower levels of C3- and C4-carnitine, derived from amino acid and fatty acid catabolism, than CR and CC (Table 2). Additionally, lower acetyl-carnitine/free-carnitine (Fig. 3A) and free-carnitine/C18-carnitine (Fig. 3B) ratios were observed in RR compared with CC, RC, and CR. The latter two ratios are used to assess the completeness (or incompleteness) of long-chain fatty acid β-oxidation.
WD significantly reduced free-carnitine but increased the acetyl-carnitine/free-carnitine ratio (Fig. 3A) in RCWD compared with its standard-fed counterpart, whereas this ratio was decreased in CRWD. Total short-chain acylcarnitines levels were decreased in CCWD, RCWD, and CRWD (Fig. 3C) compared with their standard-fed counterparts. More specifically, CCWD presented lower levels of acetyl-carnitine than observed in RCWD and CRWD, lower levels of C3-carnitine than observed in CRWD, both groups reaching lower concentrations than RRWD, and of C4- and C5-carnitine, the latter resulting from leucine/isoleucine catabolism and being lower than IUGRWD rats. In contrast, total short-chain acylcarnitines remained constant in RRWD. Indeed, RRWD displayed increased C3DC- (malonyl by-product) and C4DC-carnitine levels compared with RR, RCWD, and CRWD. Additionally, RRWD group presented a rise in total medium-chain acylcarnitines, such as a twofold increase in C12-carnitine, compared with its standard-fed counterpart. Chronic exposure to a WD decreased free-carnitine/C18-carnitine (Fig. 3B) ratio in all groups and in CCWD compared with RCWD and increased total long-chain acylcarnitines levels in the three perinatally manipulated groups and in RRWD and CRWD, compared with RCWD (Fig. 3E). Finally, several hydroxyl acylcarnitines were increased in RRWD and CRWD, such as C16-OH-carnitine, compared with their standard-fed counterparts and to CCWD, C14-OH-carnitine compared with RCWD and particularly, the C4-OH-carnitine, derived from ketone body oxidation, which was higher in CRWD rats compared with its standard-fed counterpart and in RRWD compared with CCWD.
Specific Metabolic Signature Associated with IUGR and Postnatal Catch-Up Growth
An untargeted metabolomic analysis was applied to plasma samples collected in adult rats to identify the metabolic pathways altered by perinatal programming. Specific metabolic signatures in RC were highlighted by annotated metabolites involved in energy or nucleotide metabolism (Table 3). Under standard diet and compared with CC, RC rats had lower branched-chain amino acids (valine, leucine/isoleucine), aromatic amino acids (tyrosine and tryptophan), gluconeogenic amino acids (serine), and higher levels of succinate, without any alteration in other annotated TCA cycle intermediates, such as citrate and cis-aconitate.
Whatever the group, a significant decrease in cholic and hippuric acid was observed after WD exposure. Specifically, WD induced a twofold increase in circulating arginine levels for RRWD and in succinate for CCWD, RRWD and CRWD compared with their standard-fed counterparts (Table 3). RCWD presented higher citrate but lower asparagine/aspartate levels than its standard-fed counterpart. Furthermore, RCWD displayed lower circulating levels in asparagine/aspartate than CCWD, RRWD, and CRWD and in nicotinic acid, serine, glycine, and glutamate than CCWD. Lower taurine and higher taurochloric acid abundances were observed in RCWD, CRWD, and RRWD compared with CCWD.
Perinatal Maternal Protein Restriction Modifies Expression of Hepatic Genes Involved in Energy Metabolism
Compared with CC, perinatal maternal protein restriction reduced hepatic transcripts involved in glucose metabolism such as PEPCK in RC and GLUT2 in CR, lipid metabolism, such as ACC1 in CR and RC, Mlycd in RC, or transcription factors involved in the regulation of energy fuel metabolism, such as ChrEBP, SIRT1, FOXO1, and Srebf1 in CR (Fig. 4).
WD induced a significant decrease in hepatic gene expression of the gluconeogenic enzyme PEPCK, and an increase in the gene expression of the glycolytic enzyme LPK in CRWD (Fig. 5) compared with its standard-fed counterpart group. Genes regulating glucose transport (GLUT2), insulin signaling (InsR2) and transcription factors (ChREBP, PGC1α, FOXO1) involved in both pathways were also increased in CRWD. In contrast, CCWD had a decrease of SIRT1, FOXO1 and InsR2 gene expression and RRWD had a decrease of GLUT2 gene expression, compared with their standard-diet group. Similar to CRWD, RCWD presented increased of ChREBP, PGC1α, and FOXO1 gene expression. The expression of genes involved in liver lipogenesis (FASN, ACC1, FABP1) was decreased in CCWD. Only FAS gene expression was decreased in both IUGRWD rats (Fig. 5). Concerning β-oxidation, a significant increase was observed in hepatic gene expression of CPT1a in RRWD, compared with its standard-fed counterpart group (Fig. 5) and to RCWD (data not shown). An increase of Acacb gene expression in CRWD and HMGcl in RCWD was observed, conjointly to the gene expression of transcriptional factors, such as PPARα in RCWD and Srebf1 in RCWD and CRWD, compared with their standard-fed counterpart groups.
Although there is growing evidence that adult rats, with a low birth weight have impaired insulin sensitivity, and, therefore, are at a higher risk of developing T2DM (27, 55), very little is known about the mechanisms associated with the progressive loss of IS. To the best of our knowledge, this is the first attempt to study metabolic profiles in relation to insulin secretory capacity and insulin sensitivity in rats born from protein restricted mothers at various time windows (gestation/lactation) and exposed to a WD at adulthood. The data were obtained by a combination of integrative and targeted approaches on adult rats before they ever developed diabetes. Although, these data were obtained at a single time point at D234, the clusters of discriminant metabolites and their abundances allowed us to generate testable hypothesis about how maternal perinatal diet might influence metabolic outcomes in the male offspring, particularly their metabolic “inflexibility” under a high-energy diet. Our findings suggest that 1) in adult male rats born under IUGR and fed a standard diet, insulin secretion is impaired, whereas whole-body insulin sensitivity is preserved; 2) in the rats born under IUGR and that experience catch-up growth, the fasting substrate switch is impaired even when rats were fed a standard diet, and energy storage when exposed to high-energy diet is altered; 3) in IUGR rats with a slow postnatal growth, mitochondrial β-oxidation is impaired and associated with increased fat accretion when exposed to WD; and 4) normal-born rats with a slow postnatal growth rate are “protected” against fat accretion or insulin sensitivity loss when exposed to WD. Altogether, these data demonstrate that maternal protein restriction, depending on the timing of exposure, leads to specific changes in insulin sensitivity or metabolic alterations in the offspring at adulthood. Metabolomic profile and hepatic transcript analysis then bring further hypotheses on the various pathways impacted by the maternal diet.
Impaired Insulin Secretion, Fat Storage, and Fuel Oxidation Switch in IUGR Rats with Catch-Up Growth
The enhanced insulin sensitivity observed in the current study in young RC and RR adult rats compared with CC is consistent with previous reports for rats born under IUGR (37, 53, 55). Although insulin sensitivity no longer differed between RC and CC rats in late adulthood, regardless of diet, HOMA-β index decreased in RC group fed a standard diet, suggesting that catch-up growth impairs pancreatic β-cell function.
Under standard-fed conditions, mild elevation of blood glucose, lower circulating lactate, and unchanged levels in acetyl-carnitine (C2-CN), which originates mostly from carbohydrate catabolism, are consistent with higher rates of gluconeogenesis in RC in fasting state than in CC rats. Compared with CC, the RC group presented a specific decrease in the circulating concentrations of branched-chain amino acids (BCAA; valine, leucine, or isoleucine) and two by-products of isoleucine and valine catabolism, the propionyl (C3)- and the methylmalonyl (C4DC)-carnitine, and an increase in circulating succinate level, whereas levels of acetyl- and isovaleryl (C5)-carnitine, a by-product of leucine/isoleucine catabolism, were unchanged. This finding suggests that IUGR rats with catch-up growth fed a standard diet displayed increased BCAA catabolism, probably through the propionyl-CoA-methylmalonyl-CoA-succinyl-CoA pathway (49). That leads to enhanced production of succinate, a TCA intermediate. This enhanced fasting whole-body BCAA catabolism in RC rats takes place without acyl-CoA in excess, as reflected by the absence of short-chain acylcarnitines accumulation, and could reflect BCAA catabolism in fat tissue, which significantly contributes to the regulation of circulating BCAA (29). Additionally, the similar circulating levels of NEFA and acetyl-carnitine and a significant decrease in both C4- and total long-chain acyl-carnitines, combined with unchanged blood and hepatic TAG concentrations, in RC compared with CC rats reflect unchanged rates of lipogenesis along with a downregulation of fatty acid β-oxidation in intrauterine growth-restricted rats with catch-up growth and fed a standard diet. The latter could be mediated by α-ketoacids generated by BCAA catabolism, as recently suggested in the liver (7, 17). This drop in circulating BCAA is consistent with the lower insulin secretion (lower HOMA-β index) in RC rats, as BCAA stimulate insulin secretion (76). Lower BCAA may also impair insulin signaling through mTOR and SREBP-1c downregulation and AMPK upregulation as recently reported (36, 41).
Under WD, higher expression of ChREBP, PGC1α, and FOXO1, known to upregulate hepatic glucose production could be associated with higher gluconeogenesis capacity in RC rats (30, 56). Whereas CCWD (and CRWD) had decreased PEPCK transcript, no change was apparent in RCWD at this age in our study, but PEPCK activity was found to be increased in 11-mo-old IUGR rats in other studies (15). However, this finding was not associated with elevated blood glucose or a decrease in circulating BCAA in RCWD rats compared with CCWD, suggesting specific WD-induced metabolic adaptations in IUGR caught-up rats. Epigenetic silencing of gluconeogenic downregulators may be involved, as found in recent studies in IUGR rats with catch-up growth (77). Interestingly, the higher citrate concentrations, combined with a dramatic decrease in asparagine/aspartate observed in RCWD compared with RRWD and CRWD, suggests a higher rate of oxaloacetate-asparagine conversion to citrate in caught-up IUGR rats. Citrate accumulation under WD feeding could result from lesser utilization of citrate in the TCA cycle: this is supported by the high acetyl-carnitine/free-carnitine ratio combined with the lack of elevation in succinate in RCWD as observed in the other three WD groups compared with their standard-fed counterparts. Furthermore, the accumulated citrate could be diverted toward lipogenesis, as suggested by the higher fat mass gain measured in RCWD (61 g) compared with CCWD (44 g). Beside this higher WD-induced fat accretion, RCWD displayed lower hepatic transcript of FAS, increased blood fasting NEFA but unchanged circulating TAG concentrations than CCWD. Altogether, these results suggest that in RC group under WD, increased peripheral lipolysis, likely due to the loss of the antilipolytic insulin action in adipose tissue, as suggested by higher circulating NEFA during the hyperinsulinemic-euglycemic clamp, was not counterbalanced by TAG production. This could have adverse metabolic consequences since high NEFA are thought to produce glucose intolerance (66) and diabetes (6) through impaired insulin signal transduction (81) and lipotoxicity (75). These findings suggest that whereas a prenatal nutritional insult could have lasting effects on β-cell function and adipose tissue insulin sensitivity (52, 64), accelerated postnatal growth may constitute a “second hit” that worsens the initial insult. Concerning fatty acid metabolism, WD exposure induced an increase in total long-chain acylcarnitine in the RR, RC, and CR rats. However, IUGR rats with catch-up growth exposed to plethoric diet at adulthood displayed less long-chain fatty acid oxidation than RRWD and CCWD, as evidenced by the higher plasma free-carnitine/C18-carnitine ratio. In that group the first step of β-oxidation was probably impaired since the acetyl-carnitine/free-carnitine ratio was low despite unaltered total medium- and short-chain acylcarnitines levels. One explanation for this reduced level of long-chain acylcarnitines could be citrate conversion to malonyl-CoA, known to inhibit CPT1a; such inhibition would account for the lower rates of hepatic β-oxidation (78) but also ketogenesis (16). Lower ketogenesis, indeed, is sustained by lower plasma ketone levels in RCWD compared with CCWD and RRWD. A lesser availability of the precursor, the acetyl-coA, following β-oxidation decrease in RCWD is the most likely reason, given that the ability of RC to produce ketones was retained under WD as reflected by higher liver expression of PPARα, the regulator of “fasting-lipid oxidation-glucose sparing” (20), and of HMGcl compared with their standard-diet counterparts. Additionally, regarding bile acids, RCWD had lower circulating glycine and taurine levels associated with higher taurochloric acid. This discrepancy may also reflect an early liver dysfunction (13), as well as deregulation in glucose/lipid homeostasis in RCWD rats as bile acids are involved in lipid clearance and glucose/lipid metabolism (38). In agreement with our findings, alterations in bile acid pool (62) and lower glycine levels (80) were recently identified in diabetic humans and proposed as a prediabetes-specific marker in a human metabolomic study (80) and confirmed in the EPIC-cohort (22). Finally, RCWD rats were likely to present depletion in antioxidant defense system reflected by the decrease in several precursors of glutathione (serine, glycine, and glutamate). In summary, the metabolite profile detected in IUGR caught-up rats may reflect preferential oxidation of BCAA rather than glucose in the fasting state under a standard diet and a lesser fatty acid β-oxidation than in IUGR rats nursed by protein-restricted mothers in response to chronic WD exposure.
Prenatal and Postnatal Maternal Protein Restriction Could Increase Incomplete β-Oxidation
The embryonic phase is crucial for pancreatic β-cell differentiation and maturation (72). Impaired insulin secretory capacity, therefore, appears to be an early phenomenon in IUGR, leading to an increased susceptibility to develop T2DM later in life (68). The lower HOMA-β in adult IUGR rats observed in our study supports this view. However, in other studies, IUGR offspring nursed by protein-restricted mothers had impaired glucose tolerance by 15 mo of age (27), and developed frank diabetes at 17 mo (55) with a transient enhancement in glucose tolerance and whole body insulin sensitivity (37, 55) in young adulthood. This was not the case in the present work, in which the enhanced insulin sensitivity persisted in adult RR rats (53, 55) at 9 mo of age, probably due to the presence of adipocytes of smaller size than previously described (9), since small adipocytes are more insulin-sensitive than large ones (12). The lack of effect of WD exposure on HOMA-β and circulating NEFA during the clamp suggest lesser β-cell responsiveness, possibly due to lipotoxicity. However, the blood metabolomic fingerprint and liver gene pattern highlighted mechanisms that involve the same pathways (namely fatty acid oxidation and TCA cycle) in the two groups of IUGR rats but to different degrees. Fasting and high-energy intake have long been known to elicit a switch from glucose to fat utilization as a preferred fuel (35, 57). Perinatal nutritional insult may alter this mechanism. Indeed, lower plasma free-carnitine/C18-carnitine ratio and higher total long-chain acylcarnitine observed in RR both argue in favor of increased long-chain fatty acid β-oxidation in RR compared with RC rats. Chronic exposure to plethoric diet exacerbated mitochondrial incomplete fatty acid β-oxidation (45) in RRWD rats compared with their standard-fed counterparts or to CCWD, as evidenced by acylcarnitine patterns combined with unaltered triglyceride liver, increased hepatic CPT1a expression, and a twofold increase in plasma arginine, an amino acid thought to regulate multiple pathways involved in β-oxidation (71). Furthermore, the higher levels of total long-chain acylcarnitines despite unchanged even short-chain-length species (C2- and C4-carnitines), in RRWD compared with RCWD, are consistent with enhanced long-chain fatty acid flux into mitochondrial β-oxidation “stopped” in the initial rounds of β-oxidation for RRWD. This dysfunction of fatty acid metabolism in RRWD was also supported by higher levels of two other emergent biomarkers (C14-OH- and C16-OH-carnitine) (70). Besides, the concomitant increase in levels of methylmalonyl-carnitine (a precursor of succinyl-CoA, C4DC-carnitine) and succinate in RRWD compared with their standard-counterparts and RCWD suggest a dysfunction in TCA cycle at a latter step for RRWD (succinyl/succinate processing) than for RCWD (citrate processing). Interestingly, compared with RCWD, RRWD displayed higher levels of ketones and D3-OH-butyryl-carnitine (namely C4-OH-carnitine), a β-hydroxyl-acylcarnitine produced from the ketone body D3-hydroxybutyrate and considered as an emergent biomarker of insulin resistance induction in skeletal muscle (65). Although we could not exclude that the hydroxybutyryl moiety of C4-OH-carnitine represents an α-hydroxybutyryl moiety, perhaps reflecting α-hydroxybutyrate, an organic acid very recently positioned at a crossroad of glutathione biosynthesis and upstream to the TCA cycle through the propionyl-CoA pathway, and considered to be an emergent biomarker of Type 2 diabetes risk (19, 21, 23). Altogether, these findings are consistent with the mitochondrial overload concept recently proposed (34), associated with incomplete β-oxidation and likely due to combined glucotoxicity and lipotoxicity (45). Such an overload could contribute to impaired insulin signaling and sensitivity in skeletal muscle (34). Finally, the fact that both IUGR groups shared similarity in the bile acid pattern, such as the higher taurocholic acid compared with CCWD, combined with the lower cholic and hippuric acids levels in RRWD than their standard-fed counterparts, could be explained by changes in gut microbiota following IUGR (24).
Slow-Postnatal Growth Preserves Energy Homeostasis and Insulin Sensitivity
To the best of our knowledge, the peculiar metabolic response of CR rats to the Western diet observed in the current study has never been described before. CRWD rats showed improved insulin sensitivity, along with lesser fat accretion despite a hyperphagia driven by impaired satiety control, as described in our previous study (42). The effects of insulin and nutrients on fat accretion could be counteracted in CRWD by increased energy expenditure due to leptin-induced thermogenesis (50). This hypothesis is in agreement with the central leptin hypersensitivity (67) and the elevated thyroid function (54) described in offspring nursed by mothers fed a protein-restricted diet. Here, increased total long-chain acylcarnitine levels, decreased free-carnitine/C18-carnitine, and acetyl-carnitine/free-carnitine ratios combined with lower C2- but higher C4-carnitine levels in CRWD than in their standard-fed counterparts support the classical higher whole body long-chain fatty acid β-oxidation in CR rats in response to WD. Additionally, in CRWD, the enhanced ketogenesis in response to chronic WD exposure seems to be associated with an increase in ketone body oxidation, as evidenced by the combination of higher circulating ketone bodies and C4-OH-carnitine levels with unchanged BCAA catabolism features and lower plasma triglyceride levels. Moreover, compared with the other groups, liver adapted differently to Western diet in CRWD group with changes that may reflect higher glucose uptake (increased GLUT2 transcript) and glycolytic activity (increased LPK) associated with lower gluconeogenesis (decreased PEPCK, increased SIRT1). This hypothesis is supported by the recent demonstration that SIRT1, a nutrient sensor, suppresses hepatic gluconeogenesis in long-term fasting, whereas it enhances gluconeogenesis in short-term fasting (39). Surprisingly, WD induced an apparent decrease in hepatic β-oxidation capacity in CR rats, as suggested by the dramatic increase in the expression of Acacb, an enzyme that limits β-oxidation by inducing malonyl-coA production, compared with the expression of Mlcyd (involved in malonyl-coA conversion to acetyl-coA). This finding supports altered hepatic acetyl-CoA-malonyl-CoA conversion in CRWD on a high-energy diet, which suggests preferential β-oxidation in skeletal muscle rather than in liver in CRWD rats. Taken together, this mechanism could confer male offspring with a slow postnatal growth a higher metabolic flexibility (32) to cope with a nutritionally deleterious environment.
Perspectives and Significance
With the use of an integrative approach to study metabolic imprinting in a rodent model, our intention was to provide a deeper understanding of the alterations in metabolism that differ depending on the time window of the perinatal insult with long-term consequences on whole body insulin sensitivity. Metabolomic signatures combined with targeted acylcarnitine profiling and analysis of liver gene expression used here provide new insight into the contribution of BCAA, fatty acid, and ketone oxidation pathways combined with altered TCA activity, to metabolic imprinting. Our findings are consistent with the hypothesis that maternal protein restriction during gestation and/or lactation leads to metabolic inflexibility in the offspring. However, the mechanism linking these events remains to be elucidated. It is unclear whether metabolic inflexibility is the cause or the consequence of metabolic programming. Additionally, supplemental studies are necessary to determine whether metabolic inflexibility could result from differences in the relative abundance of various tissues, or in their relative metabolic activity, which will affect both metabolic clearance and production. From a pediatric nutrition standpoint, it would be of most interest to determine whether nutritional strategies to manipulate fatty acid and/or BCAA oxidation early in life would be able to prevent or attenuate insulin resistance in later life.
A. Martin Agnoux was the recipient of a doctoral fellowship from PONAN (Pôle Nantais Alimentation et Nutrition).
No conflicts of interest, financial or otherwise, are declared by the authors.
Author contributions: A.M.A., P.P., and M.-C.A.-G. conception and design of research; A.M.A. and G.P. performed experiments; A.M.A., J.-P.A., G.S., and M.-C.A.-G. analyzed data; A.M.A., D.D., and M.-C.A.-G. interpreted results of experiments; A.M.A. and M.-C.A.-G. prepared figures; A.M.A., J.-P.A., P.P., and M.-C.A.-G. drafted manuscript; A.M.A., J.-P.A., G.S., D.D., P.P., and M.-C.A.-G. edited and revised manuscript; A.M.A., J.-P.A., G.S., D.D., P.P., and M.-C.A.-G. approved final version of manuscript.
The authors thank Muhammad-Quaid Zaman and Véronique Leray (Food Science and Engineering, Nutrition and Endocrinology, L'Université Nantes Angers Le Mans, Oniris, Nantes, France) for their expertise in hyperinsulinemic-euglycemic clamp technique. The authors also acknowledge Frédérique Courant and Anne-Lise Royer (Laboratoire d'Etude des Résidus et Contaminants dans les Aliments, Oniris, Nantes, France) for standard characterization in LCMS method. The authors are grateful to Thomas Moyon (Unité Mixte de Recherche Physiologie des Adaptations Nutritionnelles (PhAN), Nantes, France) for helpful discussions of statistical tools.
- Copyright © 2014 the American Physiological Society