Maternal Obesity Promotes a Proinflammatory Signature in Rat Uterus and Blastocyst

Abstract

Maternal obesity at conception increases the risk of offspring obesity, thus propagating an intergenerational vicious cycle. Male offspring born to obese dams are hyperresponsive to high fat-diets, gaining greater body weight, fat mass, and additional metabolic sequelae compared to lean controls. In this report, we identify the impact of maternal obesity before conception, on the embryo, and intrauterine milieu during the periimplantation period. We conducted global transcriptomic profiling in the uterus and periimplantation blastocyst, gene/protein expression analyses of inflammatory pathways in conjunction with endocrine and metabolic characterization in the dams at implantation. Uterine gene expression profiles of lean and obese dams revealed distinct signatures for genes regulating inflammation and lipid metabolism. Both pathway and gene-set enrichment analysis revealed uterine nuclear factor-κB and c-Jun N-terminal kinase signaling to be up-regulated in the uterus of obese dams, which was confirmed via immunoblotting. Obese uteri also evidenced an inflammatory secretome with higher chemokine mRNA abundance (CCL2, CCL5, CCL7, and CxCL10) and related regulators (TLR2, CD14, and Ccr1). Increased inflammation in the uterus was associated with ectopic lipid accumulation and expression of lipid metabolic genes. Gene expression in sex-identified male periimplantation blastocyst at day postcoitum 4.5 was clearly influenced by maternal obesity (359 transcripts, ±1.4-fold), including changes in developmental and epigenetic regulators. Akin to the uterus, nuclear factor-κB-regulated proinflammatory genes (CCL4 and CCL5) increased and expression of antioxidant (GPx3) and mitochondrial (TFAM and NRF1) genes decreased in the obese embryos. Our results suggest that ectopic lipid and inflammation may link maternal obesity to increased predisposition of offspring to obesity later in life.

At present, nearly half of all pregnancies are in women who are either overweight or obese at conception (1, 2). Being overweight during pregnancy can adversely impact the immediate wellbeing of the mother, as well as the long-term health of the offspring (3). Pregravid obesity significantly increases the risk of preeclampsia, gestational diabetes, and other labor-related complications (3, 4). Maternal overweight also increases the odds of being large-for-gestational age (LGA) at birth, which substantially adds to the risk of obesity in adulthood (5). Moreover, evidence from experimental models clearly demonstrate that maternal obesity, independent of birth weight, leads to developmental programming of adiposity gain in the offspring (6–15). Although precise mechanisms of such programming remain to be fully elucidated, they are likely to be multifactorial, involving a number of interrelated targets (11, 16).

The intrauterine environment plays a critical role in shaping the long-term susceptibility to obesity. Perturbations during gestational development toward either nutritional deprivation or excess can permanently program a number of physiological responses (17). However, only a few studies have examined the early events leading to fetal programming (18–21). Most importantly, almost no information exists about the effect of pregravid maternal obesity on the uterus, at the time of implantation. During implantation, the trophectoderm cells of the blastocysts, in an intricately defined series of events, interact with the luminal epithelium of the uterus (22, 23). The goal of implantation is to anchor the developing embryo to the endometrial stroma and this is orchestrated by ovarian hormones and locally produced signaling molecules, including cytokines and growth factors (22). With the progression of gestation, the trophectoderm will eventually develop into a functional placenta and buffer the embryo from many unfavorable changes in the maternal circulation. However, at early stages, especially during implantation, the uterine environment is likely to exert a major influence on the development of the offspring. In the present study, we directly examine this important issue.

Using a model of preconceptional obesity in the rat, we previously demonstrated that maternal obesity programs increased sensitivity to weight gain in the offspring, in the absence of changes in birth weights (14, 15). In this model, exposure to maternal obesity is limited to the in utero period, and gestational weight gain is equalized. This results in no significant differences in birth weight but a marked increase in offspring weight gain after weaning to an obesogenic high-fat diet (HFD). Before conception, obese rat dams develop significant signs of metabolic syndrome, including markedly elevated serum insulin, leptin, free fatty acids, triglycerides, and insulin resistance in the absence of hyperglycemia (14). In this report, we focus on the effect of maternal obesity at conception on the uterine milieu and the blastocyst at implantation [4.5 d postcoitum (dpc)]. The majority of studies examining maternal obesity have used energy-dense (high fat) or highly palatable diets to produce increases in adiposity (6, 8, 10, 13, 24). Hence, relatively little is known about the effects of overconsumption of a nutritionally appropriate diet (i.e. excessive calories per se). Self-limiting consumption of diets due to satiety has been the primary limitation in development of obesity in animal models, necessitating use of energy-dense diets. By employing controlled feeding of liquid diets via total enteral nutrition (TEN), we overcome this limitation.

The present study had three objectives. First, we examined whether maternal obesity altered uterine gene expression during the implantation window. Specifically, we examined whether maternal obesity preconception led to changes in lipid metabolic genes, inflammatory mediators, and cytokines in the uterus. Second, we examined key signaling pathways known to link ectopic lipid accumulation and proinflammatory signaling in the uterus. Finally, we elucidated global transcriptomic changes in the periimplantation blastocyst using microarrays to identify the earliest changes associated with maternal obesity. Our data strongly suggest that both the blastocyst and the uterus of obese dams within the periimplantation window have metabolic signatures consistent with lipotoxicity.

Materials and Methods

Experimental design

All experimental treatments were conducted in accordance with the ethical guidelines established and approved by the Institutional Animal Care and Use Committee at the University of Arkansas for Medical Sciences (protocol 2971). Virgin female Sprague-Dawley rats (150–175 g; Charles River Laboratories, Wilmington, MA) were housed in an Association for Assessment and Accreditation of Laboratory Animal Care-approved animal facility. Animals were intragastrically cannulated and allowed to recover for 10 d as previously described (14, 15, 25–27). Rats were fed liquid diets at either 155 kcal/kg^3/4 · d (referred to as lean dams) or at 220 kcal/kg^3/4 · d (40% excess calories, referred to as obese dams). Caloric intake for the lean group was determined from preliminary studies and mimicked body weights and body composition of rats consuming standard commercial diets ad libitum (14, 15, 27). Diets met National Research Council nutrient recommendations, including essential fatty acids, and were 20% protein (casein), 75% carbohydrate (dextrose and maltodextrin), and 5% fat (corn oil) as percentage of total calories. Feeding was carried out 23 h/d using computer- controlled syringe pumps for 4 wk. Body weights were monitored three times a week. At the beginning and the end of 4 wk, body composition was noninvasively estimated using nuclear magnetic resonance (NMR) (Echo Medical Systems, Houston, TX) (14, 15, 27).

After 4 wk of overfeeding, lean and obese rats (n = 10 per group) were allowed to mate with control lean breeder male rats. Each female rat was housed with one male and allowed ad libitum access to AIN-93G diet for the duration. Successful mating was confirmed by the presence of sperm in the vaginal lavage the next morning (designated as dpc 0.5). After mating, all female rats (lean and obese) resumed receiving diets at 220 kcal/kg^3/4 · d (NRC-recommended caloric intake for pregnancy in rats) until dpc 4.5. Using this experimental paradigm, conception and early embryonic development occurred in the context of an obese intrauterine environment (14, 15). Between 1000 and 1130 h on dpc 4.5, dams were killed under nembutal anesthesia. Blood, liver, and adipose tissues (retroperitoneal and gonadal depots) were weighed and collected. From individual rats, the entire uterus was dissected out. Periimplantation blastocysts were flushed from both uterine horns with approximately 1 ml of DMEM (supplemented with 10% fetal bovine serum) into six-well cell-culture plates. After removal of the blastocyst, uterine samples along with other tissues (liver, muscle, and adipose tissues) were either fixed in neutral-buffered formalin for histological analyses (or optimal cutting temperature media). Remaining tissues were frozen in liquid nitrogen and stored at −70 C for RNA and protein analyses. Serum was obtained by centrifugation of blood samples and stored at −20 C for endocrine and metabolic assessments.

Serum parameters and uterine histology

Serum glucose, triglycerides, cholesterol, nonesterified fatty acid (NEFA), insulin, leptin, and adiponectin were measured in lean and obese dams (n = 7–9 per group) as described in Supplemental material, published on The Endocrine Society's Journals Online web site at http://endo.endojournals.org (14, 15, 27). Concentrations of 10 cytokines were assayed using custom-designed 10-plex Luminex bead assays (Millipore, Bedford, MA). Undiluted serum was used and assay performed according to manufacturer's recommendations. Uterine sections were stained with hematoxylin and eosin or Oil Red O as previously described (15, 26).

mRNA and genomic DNA isolation and gender determination in individual blastocysts

Individual blastocysts (n = 50–65 from each maternal group) were collected in lysis buffer and stored at −70 C until further analyses. Only mature blastocysts with a compact inner cell mass and visible zona pellucida were collected in this study. Poly-A RNA and genomic DNA were isolated from individual blastocysts using Dynabeads mRNA-Direct kit and procedures described previously by Kwong et al. (28). Details of the procedure are further described in Supplemental material. mRNA from individual blastocysts was stored at −70 C for RNA amplification and gene expression analyses. Total genomic DNA from individual blastocysts was isolated from the supernatant after poly-A RNA selection using phenol-chloroform-isoamyl alcohol (details provided in Supplemental material).

In previous studies, we observed significant hyperresponsiveness of male offspring born to obese dams challenged to a HFD. Therefore, in the present study, the effect of maternal obesity on male blastocyst was investigated to exclude any sex-associated confounding factors. Sex determination of blastocysts was performed by amplification of the Sry gene using nested PCR (primer sequences in Supplemental Table 1). Details of the PCR conditions are provided in Supplemental material. In preliminary experiments using liver genomic DNA from bona fide male and female adult rats, we ascertained that unequivocal gender discrimination was achieved (i.e. bands were observed only in male samples) even when input DNA was as low as 30 pg using these PCR conditions.

Uterine and blastocyst gene expression analyses

Uterine RNA isolation and microarray analyses

Total RNA was isolated from uterus of dams at dpc 4.5 (n = 7–9 per group) using a combination of TRI reagent and RNeasy-mini columns (QIAGEN, Valencia, CA), including on-column deoxyribonuclease digestion. For uterine gene expression, three microarrays were used for each group (Affymetrix GeneChip Rat 230 2.0). Pools of equal amounts of RNA from two to three rats were used for analyses per microarray. Thus, n = 7–9 rats per group were represented over the three microarrays. Amplified RNA synthesis, labeling, hybridization, and scanning were carried out as previously described (15, 27, 29). RNA pools containing poly-A RNA from six to eight male blastocysts from at least two to three separate rat dams from either lean or obese dams were used for microarray analysis. Four microarrays per group were used to assess gene expression representing approximately 30 blastocyst from each group (Affymetrix GeneChip Rat 230 2.0). Two-cycle RNA amplification was carried out using Message Amp II (Ambion, Foster City, CA) according to manufacturer's instructions, except a 15% aliquot of cDNA in the second cycle was saved and not used for biotinylated aRNA preparation. This cDNA was used for confirmation of selected genes later. All other procedures for RNA amplification and microarray analyses were carried out according to manufacturer's instructions.

Microarray data normalization and analysis

Microarray data analyses were carried out using GeneSpring version 11 software (Agilent Technologies, Santa Clara, CA) (15, 27). The .CEL files containing probe level intensities were processed using the robust multiarray analysis algorithm, for background adjustment, normalization, and log₂-transformation of perfect match values (30). Subsequently, the data were subjected to normalization by setting measurements less than 0.01–0.01 and by per-chip and per-gene normalization using GeneSpring. The normalized data were used to generate list of differentially expressed genes (between lean and obese groups) in the blastocyst and uterus. Genes were filtered based on minimum ±1.4-fold (or ±1.3-fold change for uterus) (obese vs. lean) and P ≤ 0.05 using Student's t test. A list of transcripts that were differentially expressed as a function of maternal obesity was generated, and correlation-based hierarchical clustering between treatment groups was performed. Three separate analyses of enrichment of biological function annotation were performed. Known biological functions of genes were queried using Affymetrix NetAffx and gene ontology (GO) analyses for biological/molecular function performed using GeneSpring using FDR corrected P < 0.1 (default for GeneSpring) (15, 27). Further, the list of genes affected in uterus and the blastocyst by maternal obesity was analyzed using Ingenuity pathway analysis (IPA). This analysis included identifying top interacting networks based on IPA-curated knowledge-base from the known literature. Finally, we carried out gene-set enrichment analysis (GSEA) to identify biological processes, cellular components, and motifs enriched by maternal obesity in the uterus and the blastocyst. GSEA does not rely on an arbitrary cutoff (such as fold change between groups) and is a computational method that determines whether an a priori-defined set of genes shows statistically significant, concordant differences between two biological states (31).

Real-time RT-PCR

Total RNA from uterus of dams (same as those used for microarray analyses) was isolated using RNeasy columns (QIAGEN). One microgram of total RNA was reverse transcribed (n = 7–9 per group) using IScript cDNA synthesis kit (Bio-Rad, Hercules, CA). cDNA aliquots prepared from blastocyst were diluted, and real-time PCR analysis was performed as described previously (15, 27). Gene-specific primers were designed using primer express software (Supplemental Table 1). Relative amounts of mRNA were quantitated using a standard curve and normalized to the expression of signal recognition particle 14kDa (SRP14) mRNA (15, 27).

Immunoblotting and immunohistochemistry

Immunoblotting was carried out using standard procedures (15, 27). A detailed description is provided in Supplemental material. Immunoblotting was performed for pAKT (Thr308), AKT, p-p38 MAPK (Tyr182), p38-MAPK, p-c-Jun N-terminal kinase (JNK)1/2 (Thr183/Tyr185), JNK, p-p65-nuclear factor (NF)-κB (Ser536), total p65-NF-κB, ERK1/2, or pERK1/2 (Thr202/Tyr204) proteins in total uterine lysates, or proteins from nuclear extracts. Immunohistochemistry was carried out on OCT-embedded 8-μm sections using Vectastain Elite reagents. Sections were incubated overnight with CCL5/regulated upon activation, normal T-cell expressed, and secreted (RANTES) antibody (Abcam, Cambridge, MA).

Statistical analysis

Data are expressed as means ± sem. Statistical differences between lean and obese rat dams before conception or dams during gestation were determined using two-tailed Student's t test. Similarly, differences between blastocyst of lean and obese dams at dpc 4.5 were determined using two-tailed Student's t test. Statistical significance was set at P < 0.05. Statistical analyses were performed using SigmaStat 3.3 software (Systat Software, Inc., San Jose, CA).

Results

Maternal overfeeding produces metabolic dysfunction at implantation

Overfeeding female rats via TEN for 4 wk resulted in approximately 20% greater body weight gain and 30% greater total body fat (P < 0.001) (assessed by NMR) in obese females compared with lean controls (Table 1 and Supplemental Fig. 1). Obese females developed elevated serum insulin (63% greater than lean controls, P < 0.05), leptin (100% greater than lean controls, P < 0.0005), triglycerides (50% higher), cholesterol (60% higher), NEFA (100% greater than lean controls), and 20% lower serum adiponectin levels (P = 0.06) at the end of the 4-wk period (Table 1). Serum glucose levels did not differ between the groups. At the end of the overfeeding period (4 wk from the beginning of diets), each female rat was housed with one male breeder rat and given ad libitum access to pelleted AIN-93G-based diets. After confirmation of successful mating, both lean and obese dams resumed receiving TEN feeding at 220 kcal/kg^3/4 · d (the NRC-recommended caloric intake for pregnancy). All animals were killed at dpc 4.5. Consistent with NMR analyses, percentage visceral adipose tissue weight in the obese dams was approximately 65% greater compared with their lean counterparts (P < 0.0001) (Table 1).

Table 1.

Maternal weight, body composition, and endocrine parameters at implantation after 4 wk of overfeeding

Parameter	Lean	Obese	P value
Body weight at beginning of TEN	224 ± 4.3	216 ± 5.7	0.29
Body weight at conception	278 ± 6.4	328 ± 7.6	<0.0005
% Liver Wt	4.1 ± 0.18	3.9 ± 0.26	0.48
% Visceral fat Wt	4.1 ± 0.28	6.7 ± 0.13	<0.0001
% Lean mass	61 ± 1.22	55 ± 1.15	<0.05
% Fat mass	24.9 ± 1.09	32.5 ± 0.66	<0.001
Glucose (mg/dl)	151 ± 14	146 ± 14	0.82
Triglyceride (mg/dl)	80 ± 5	120 ± 16	0.005
Total cholesterol (mg/dl)	75 ± 6	120 ± 9	0.001
NEFA (μm)	278 ± 32	559 ± 99	0.02
Insulin (ng/ml)	3.2 ± 0.53	5.2 ± 0.67	<0.05
Leptin (ng/ml)	14.5 ± 1.7	29.5 ± 2.9	<0.0005
Adiponectin (μg/ml)	29 ± 2.19	23 ± 1.88	0.06
IL-1β	12.7 ± 2.2	7.6 ± 2.4	0.18
IL-2	ND	ND
IL-4	19.3 ± 8.3	29.5 ± 13.9	0.59
IL-6	250 ± 77	405 ± 251	0.52
IL-10	ND	ND
IL-12p70	50 ± 9.4	33.2 ± 3.9	0.50
INF-γ	76.8 ± 20	144.8 ± 87	0.29
RANTES	ND	ND
TNF-α	ND	ND

Data were obtained from lean or obese dams at 4.5 dpc (n = 7–9 per group). Lean and obese dams were fed via TEN as described under Materials and Methods. Data are expressed as means ± sem. Indices of percentage lean and fat mass were determined using noninvasive quantitative NMR. Weights (Wt) of liver and visceral adipose tissues (retroperitoneal + gonadal fat depots) were assessed at sacrifice. Serum cytokines in lean and obese dams was assessed at dpc 4.5 using 10-plex multiplex immunoassays. Concentrations are in pg/ml. ND, Not detected. P values were determined using a Student's t test.

Maternal obesity alters uterine transcriptome at the time of implantation

To assess whether maternal obesity altered gene expression in the uterus at implantation (dpc 4.5), we performed expression profiling using microarrays. Unsupervised global condition clustering revealed clustering of expression profiles based on maternal body composition phenotypes, suggesting significant treatment effect on global gene expression (Supplemental Fig. 2A). After normalization, 403 transcripts were identified to be differentially expressed in obese dams (±1.3-fold, P ≤ 0.05) (Supplemental Table 2). Pearson's correlation-based hierarchical clustering of the genes affected by maternal obesity is depicted in Supplemental Fig. 2B. These transcripts were used for GO analyses based on molecular function, biological function, and pathway analyses. Altered genes possessed binding (58%), catalytic functions (23%), transporter activity (4%), or regulated transcription (3% each) functions (Supplemental Fig. 2C). Of the 407 transcripts altered, 122 were sequences were poorly annotated in NetAffx or GeneSpring. Seventy-two transcripts of these were resolved using DAVID bioinformatics resource 6.7 (32).

Of the genes affected by maternal obesity, we identified biological functions involved in immune response, inflammation, and cytokine/chemokine signaling, all of which were increased in obese dams (Supplemental Table 2). Suites of genes with known roles in lipid homeostasis, metabolic processes, cell adhesion/communication, and angiogenesis were altered in obese dams. Additionally, we found response to stimuli, transport, protein modification/translation, and proteolysis to be affected in obese rat uterus. GO analyses of biological processes using GeneSpring clearly revealed immune function, leukocyte and chemokine activity, and defense response as significantly enriched terms (Supplemental Table 3). We further used pathway analysis software (IPA) to identify common regulators of the altered genes. The top network included inflammatory chemokines, NF-κB complex, toll-like receptor (TLR), and JNK, again highlighting the predominant changes in immune response in the uterus (Supplemental Fig. 2D). Hierarchical clustering of immune response-related genes is shown in Fig. 1A. A uniform induction of 20 proinflammatory genes, including chemokines (CCL2/MCP-1, CCL5/RANTES, CCL7/MCP-3, and CXCL10), receptors (TLR2 and CCR1), and downstream NF-κB targets (ICAM-1), was evident in obese rats. Further, we used GSEA to examine transcription motifs that were overrepresented in the obese uterus at dpc 4.5 (31). Consistent with increased proinflammatory signatures and mRNA expression, genes containing the NF-κB motif in their promoters were significantly enriched (Fig. 1B). Our analyses identified 29 genes whose expression was increased in obese uterus, including several of the aforementioned genes.

Fig. 1.

Maternal obesity alters uterine gene expression at implantation toward an inflammatory profile. A, Correlation-based clustering of genes with known functions in inflammation derived from the list of genes altered by maternal obesity. Gene expression was assessed in rat dams using Rat Genome 230 2.0 microarrays (n = 3 microarrays per group representing pools of two to three separate dams, n = 7–9 per group). Genes were filtered based on minimum ±1.3-fold change (obese vs. lean) and P ≤ 0.05 using Student's t test. Heat-map colors red, white, and blue represent up-regulation, no relative effect, and down-regulation of transcripts, respectively. B, GSEA of transcription factor motifs enriched in uteri of obese dams. Twenty-nine transcripts containing NF-κB motifs were enriched in obese dams. C, Uterine mRNA expression of inflammatory genes in lean and obese dams at dpc 4.5 (n = 7–9 per group). Gene expression was assessed via real-time RT-PCR. D, Immunohistochemical staining for CCL5/RANTES (n = 7 per group) in formalin-fixed uterine sections. a and c, Lean dams. b and d, Obese dams; top panel, ×50; bottom panel, ×400. E, Serum CCL2/MCP-1 concentrations in lean and obese dams at dpc 4.5 (n = 7–9 per group). Cytokine levels were assessed using 10-plex multiplex assays. Statistical differences were determined using a Student's t test. *, P < 0.05.

We independently verified expression of inflammation-related genes using real-time RT-PCR. Expression of CCL2 (3-fold), CCL5 (1.9-fold), CCL7 (2.1-fold), CXCL10 (1.5-fold), CCR1 (2.4-fold), CD14 (1.4-fold), and TLR2 (1.7-fold) was significantly induced (P < 0.05) in uterus of obese dams (Fig. 1C). mRNA expression of TLR-4 was numerically increased in obese uterus but did not reach statistical significance. Next, we investigated whether changes in cytokine mRNA were translated to higher protein levels. Immunohistochemical staining of uterine sections showed increased staining for CCL5 in obese uterus (Fig. 1D). Staining for CCL5 was most prominent in the luminal and glandular epithelium. We also used 10-plex Multiplex immunoassays assays to simultaneously quantitate levels of 10 cytokines in serum from lean and obese rat dams. Although the titers of several cytokines in serum were at or below the detectable threshold, circulating CCL2/MCP-1 levels were significantly elevated (52%) in obese rats (Fig. 1E) as has been reported previously (33). Systemic levels of other cytokines, including IL-1β, IL-4, IL6, IL-10, IL-12, and interferon-γ, were not altered by obesity (Table 1) in this model.

Uterine inflammation is associated with local lipid accumulation, JNK, and NF-κB activation

Pathway analyses of microarray data also indicated that lipid metabolism-related genes were affected in the uterus. Because ectopic fat deposition common in obesity is associated with increased inflammatory signaling in other tissues, such as liver, skeletal muscle, and pancreas, we examined lipid accumulation in the uterus at dpc 4.5. Oil Red O staining of neutral lipids clearly showed accumulation of lipid droplets in the obese dams, particularly in the luminal epithelium, where implantation of the embryo occurs (Fig. 2A). We next examined mRNA expression of critical lipid metabolism associated genes using real-time PCR and observed a robust induction of fatty acid binding protein-4 (FABP4) (30-fold), CD36 (2.5-fold), and lipoprotein lipase (1.7-fold) (Fig. 2B). Along with greater expression of genes commonly associated with adipose tissue and ectopic fat, we observed a strong increase (∼10-fold) in the expression of the insulin resistance-inducing adipokine retinol-binding protein-4 (RBP-4) in uterus of obese dams (Fig. 2C).

Fig. 2.

Maternal obesity increases ectopic lipid accumulation in the uterus. A, Representative photomicrographs of uterine cross-sections stained with Oil Red O for lipid droplets. a and c, Lean dams. b and d, Obese dams; left panel, ×50; right panel, ×400. B and C, mRNA expression of lipid and insulin-resistance associated genes in uterus at dpc 4.5 (n = 7–9 per group). Gene expression was assessed via real-time RT-PCR. Statistical differences were determined using a Student's t test. *, P < 0.05.

Because both ectopic lipid accumulation and inflammation were elevated in the uterus of obese dams, reminiscent of lipotoxicity observed in other tissues, we examined whether components of NF-κB, JNK, and other MAPK signaling were affected. In total uterine lysates, phosphorylation of JNK1/2 was significantly elevated (45%) (Fig. 3, A and B) in obese dams. Despite increased circulating levels of insulin, no differences were observed in either phosphorylated or total Akt levels. To ascertain whether NF-κB signaling is activated in the obese uterus, we assessed nuclear levels of phosphorylated-p65 NF-κB. As shown in Fig. 3, A and B, nuclear levels of p-p65 NF-κB (relative to total p65 NF-κB) were significantly (∼2-fold, P < 0.05) increased in obese rats. Because other MAPK (p38 and ERK) are also known to play important roles in uterine development during implantation, we assessed the levels of these proteins. No differences were observed in either phosphorylated or total p38-MAPK levels between lean and obese rats. However, pERK levels were markedly up-regulated (6- and 2-fold in ERK1 and ERK2, respectively) (Fig. 3, C and D). These data strongly suggest that obese dams demonstrate increased proinflammatory gene and protein expression associated with increased lipotoxicity and JNK/NF-κB signaling.

Fig. 3.

Lipotoxic uterine inflammation is associated with increased JNK phosphorylation. A, Quantitation of proteins involved in inflammation in nuclear (p65-NF-κB, total NF-κB) and total uterine lysates (JNK1/2) from lean and obese dams (n = 6 per group). B, Densitometric quantitation of immunoblots from lean and obese dams. C, Quantitation of MAPK proteins (p38 and ERK1/2) and AKT proteins in total uterine lysates from lean and obese dams (n = 6 per group). D, Densitometric quantitation of immunoblots from lean and obese dams. Statistical differences were determined using a Student's t test. *, P < 0.05.

Blastocyst in obese intrauterine environment display inflammatory signature

To examine the impact of maternal obesity on early embryonic development in an obese intrauterine environment, we performed gene expression analyses. In the present study, we specifically examined male blastocyst from lean and obese dams, because sex of the offspring significantly impacts embryonic and placental gene expression (18, 34), and in previous studies, we examined male offspring from obese rat dams (at weaning and post-natal day 120). Hence, we reasoned that it would be important to determine gene expression patterns in the male embryos from lean and obese dams. Four microarrays per maternal phenotype were used to evaluate separate pools of six to eight male blastocyst per array. Global condition clustering and principal component analysis of samples is presented in Supplemental Fig. 3. Comparison of blastocyst from obese and lean dams revealed differential expression of approximately 359 transcripts (±1.4-fold, P ≤ 0.05) (Supplemental Table 4). Correlation-based hierarchical clustering of the genes affected by maternal obesity is depicted in Fig. 4A. Analysis of biological processes associated with these transcripts showed that genes involved in cell cycle and adhesion, embryonic development, and chromatin and epigenetic regulation were altered in blastocyst from obese dams. In addition, genes regulating transport, protein modification, and proteolysis were also affected in obese blastocyst. Most importantly, as observed in the uterus of obese dams, proinflammatory immune response-related transcripts were up-regulated in the blastocyst of obese rats (Supplemental Table 4). Pathway analyses confirmed the predominance of inflammatory processes, and IPA identified a network coordinated by NF-κB, CCL5, and the proinflammatory cytokine IL-18 (Fig. 4B). Using real-time PCR, we confirmed the increased expression of several genes, including CCL4/MIP-1β (2.3-fold), urokinase-type plasminogen activator (2.1-fold), and CCL5 (1.3-fold) in obese blastocysts (Fig. 5A). mRNA expression of the secreted glutathione peroxidase GPx3 was decreased approximately 5-fold in embryos from obese dams (Fig. 5B). Obese blastocyst also showed decreased mRNA expression of antiapoptotic, Bcl2 (Fig. 5B). However, unlike the uterus, mRNA expression of lipid storage/transport-related genes, CD36, FABP4, and perilipin was not altered in the blastocyst due to maternal obesity. GSEA analyses independently confirmed the enrichment of proinflammatory immune-response genes in obese embryos. In addition to genes identified by threshold cutoff, other important cytokines, such as IL-6, IL-6R, and CCL20, also appear to be up-regulated in obese blastocysts (Fig. 4D and Supplemental Fig. 4). Finally, we assessed the enrichment of genes based on their cellular localization using GSEA. Although transcripts in extracellular space were enriched in obese embryos, a set of 69 transcripts localizing to the mitochondria was remarkably decreased (Fig. 4E and Supplemental Fig. 4). Decreased mRNA expression of TFAM and NRF1 in obese blastocyst was confirmed using real-time PCR (Fig. 5B). The findings collectively suggest that exposure to maternal obesity produces a proinflammatory status in the periimplantation embryo along with global disruptions in mitochondrial gene expression.

Fig. 4.

Male blastocyst from obese dams display a proinflammatory transcriptomic signature. A, Hierarchical clustering of 359 transcripts altered by maternal obesity in the embryo. Gene expression was assessed in male blastocyst using Rat Genome 230 2.0 microarrays (n = 4 microarrays per group, six to eight blastocyst for each array). Genes were filtered based on minimum ±1.4-fold change (obese vs. lean) and P ≤ 0.05 using Student's t test. B, Network involving NF-κB-regulated inflammatory genes, CCL5/RANTES, and IL-18 from IPA increased in obese dam embryos. Colors green and red represent down-regulation and up-regulation, respectively. C, GO analyses of altered transcripts based on molecular function derived from GeneSpring Gx. Enrichment of genes in obese embryos involved in (D) immune function (biological process) and (E) mitochondrial localization (cellular component) after GSEA. Red and blue colors depict high and low expression, respectively.

Fig. 5.

mRNA expression of (A) proinflammatory, (B) antioxidant, and epigenetic regulator genes in male blastocyst from lean and obese dams (n = 4 pools per group, six to eight blastocyst per pool). Gene expression was assessed via real-time RT-PCR. Statistical differences were determined using a Student's t test. *, P < 0.05.

Discussion

A multitude of factors influence the development of obesity and related metabolic comorbidities. Of these, maternal diet and body composition are important determinants of the risk of obesity in the adulthood. Convincing evidence also suggests that obesity-related outcomes in the offspring may be “programmed” at least in part during gestation, including via epigenetic alterations (9, 16). However, the earliest events leading to long-term programming of obesity in the offspring remain enigmatic. Several novel findings are evident from the present study. Maternal obesity without gestational diabetes clearly led to: 1) changes in the uterine and embryonic transcriptome, revealing a distinct up-regulation of proinflammatory pathways; 2) ectopic lipid accumulation in the uterine endometrium associated with lipotoxic pathways, i.e. JNK and NF-κB; and 3) an up-regulation of NF-κB in the male blastocysts, in addition to global changes in developmental and mitochondrial genes. A schematic summarizing the findings in obese dams at implantation is presented in Fig. 6.

Fig. 6.

Schematic summarizing changes in uterine and blastocyst gene expression at dpc 4.5 in due to maternal obesity. Maternal obesity leads to ectopic lipid accumulation in the uterine endometrium and is associated with a distinct up-regulation of proinflammatory pathways in both the uterus and the blastocyst. Genes represented in the black box and gray boxes are altered in the uterus and blastocyst, respectively (directionality of change is represented by arrow signs). Overall, genes involved in lipid biosynthesis and inflammation (especially NF-κB) are up-regulated, and mitochondrial genes are down-regulated in the blastocyst.

A distinctive aspect of the present experimental design is the use of TEN to overfeed rats in a controlled manner while maintaining dietary composition (14, 15). Using this model of caloric overconsumption, we replicated metabolic and endocrine features of overweight individuals, including hyperinsulinemia, hyperleptinemia, insulin resistance, and increased concentrations of serum triglycerides and NEFA (14). Offspring from these dams are hyperresponsive to a HFD and gain excessive weight revealing fetal programming (14, 15). Our results implicate that maternal obesity per se is likely to affect early developmental changes.

Two central findings of our studies are: the identification of increased ectopic lipid accumulation and a coordinated appearance of proinflammatory chemokines in the uterus at implantation in obese dams. Although it may be intuitive that maternal adiposity would affect uterine gene expression, empirical data on the nature of these changes are currently unavailable. Hence, the identification of a lipotoxic intrauterine milieu at implantation in obese dams is novel and consistent with our overall understanding of obesity-driven metabolic changes. Obese ob/ob and db/db mice also accumulate lipids in the endometrial cells consistent with development of hyperinsulinemia (35). Similar ectopic lipids in the uterus have also been reported in diabetes (35) and in rats chronically fed alcohol (36). We are not aware of any data in nongenetically driven models of obesity demonstrating lipid accumulation in the uterus. Importantly, our results confirm these findings with increased expression of genes regulating lipid metabolism, including CD36, LPL, and FABP4, known to be induced by HFD (26) and classically associated with the adipocyte-like phenotype (37). Similar up-regulation of CD36 and FABP4 in the ovaries of HFD-fed mice has been previously reported (38).

A strong causal relationship exists between lipid accumulation and induction of inflammatory pathways in a number of tissues (39, 40). Studies in the liver, muscle, and pancreatic β-cells show that lipotoxicity is closely linked to insulin resistance and metabolic dysfunction through inflammation (41, 42). Long-chain saturated fatty acids, such as palmitic acid, interact with TLR2 and TLR4 on the cell surface and activate multiple signaling pathways (43). Mice lacking either TLR2 or TLR4 are protected from HFD-induced insulin resistance, adipose hypertrophy, hepatic steatosis, and reduced inflammatory cytokines (MCP-1 and TNF-α), indicating a bona fide connection between TLR and lipotoxic inflammation (44–47). Indeed, concomitant with increased lipid accumulation, our studies identified increased expression of CD14, TLR2, CCR1, and multiple downstream cytokines in obese uteri. Although this study did not detect differences in several systemically circulating cytokines, these data should be interpreted with caution. Use of high-sensitivity ELISA assays presumably is another approach to critically assess the levels of low abundance cytokines in serum. The critical link between fatty acid-induced TLR activation and inflammatory gene expression is regulation of JNK and NF-κB pathways (41, 48, 49). Elegant studies by the Hotamisligil and Karin groups have unequivocally established the central role of JNK in fatty acid-mediated insulin resistance. JNK1 deletion in vivo or in hepatocyte cultures abolishes HFD (or palmitate)-induced insulin resistance or inflammatory gene expression (50). Likewise, NF-κB activation via IkB kinase-β, downstream of JNK, is central to orchestrating inflammatory gene expression and insulin resistance (48, 49, 51). Our results not only highlight lipid accumulation in the obese uterus but also JNK and NF-κB activation and induction of a battery of downstream targets involved in inflammation. These findings indicate that lipotoxic changes associated with obesity may be central to inflammatory and oxidative stress changes involved in fetal programming. It is important to note that lipid-associated inflammation may also be mediated or amplified in part via recruitment of macrophages to the uterus. Although the present study did not examine this in detail, conditions such as such as steatosis and hypertrophic adipocytes lead to accumulation of activated macrophages to the site of ectopic lipids. In addition to JNK activation, phosphorylation of ERK was significantly increased in obese uterus. Although the precise significance of these results remain unclear, they are consistent with a recent report by Villavicencio et al. (52) demonstrating an 8-fold increase in ERK phosphorylation in endometrial tissue from obese women.

It is important to recognize that uterine and embryonic gene expression data in the present report were derived from tissues taken early in gestation (d 4.5). Although the uterus of obese rats has an inflammatory signature before implantation, so does the blastocyst. This suggests an important influence of the maternal intrauterine environment on the developing embryo within just 4.5 d of mating. Further, the inflammatory response in the blastocyst could also presumably be a consequence of ovarian effects even before conception. It is conceivable that by virtue of obesity-associated influences on oocyte development, obese dam oocytes express different gene expression profiles that make the embryos more inflamed and susceptible to oxidative stress. Maternal RNA accumulated in the mature oocyte critically determines the initial developmental events in the embryo and may be altered due to obesity. To date, global effects of maternal obesity on oocyte transcriptome remain unknown; however, recent studies have identified changes in the oocytes and blastocyst of HFD-fed rats and mice. A consensus appears evolving around the findings of altered redox state, oxidative stress, and mitochondrial changes in the oocytes and embryos from obese dams (19–21). Mitochondrial dysfunction and increased reactive oxygen species were observed in oocytes and zygotes of mice fed a HFD (20). Similarly, Jungheim et al. (19) showed that HFD-fed mice had greater follicular apoptosis and delayed maturation after superovulation. Most importantly, recent work by Sen and Simmons (21) suggests a key role for oxidative stress in programming of adiposity. Supplementation with antioxidants (vitamins A, E, C, and selenium) effectively prevented the adiposity in offspring at 2 months of age. Although oxidative changes in early development appear as a common theme, the initiating mechanisms remain unclear. Our results from GSEA analyses of blastocyst gene expression identified mitochondria as an important target of maternal obesity (Fig. 4). Furthermore, we identified a 5-fold decrease in the expression of Gpx3 in the blastocysts, along with a set of 69 transcripts localizing to the mitochondria that was significantly decreased. It is plausible that decreased Gpx3 underlies enhanced oxidative stress in embryos from obese dams. Thus, the present findings of ectopic uterine lipids, resulting in JNK/NF-κB activation and inflammation, may explain the genesis of important oxidative stress due to maternal obesity. This inflammatory uterine environment would serve as a powerful reinforcing factor in the metabolic programming of the embryo due to obesity.

In addition to early developmental effects, maternal obesity influences metabolic processes throughout gestation. Maternal obesity in sheep has been recently shown to induce inflammatory changes in fetal heart and intestine (53, 54). Consistent with our results, the sheep studies reported an increase in JNK phosphorylation and greater expression of proinflammatory genes, including MCP-1, and TLR2. Maternal HF consumption in nonhuman primates also triggers lipotoxicity associated with fetal hepatic triglyceride accumulation and JNK activation (6). The effects of maternal obesity and associated circulating factors are not limited to the embryo proper but clearly extend to the placenta. Obesity increases inflammation in sheep midgestation placental (55) and term placenta from obese women (56, 57). Pregravid obesity also results in a increase in circulating lipopolysaccharide in women (58), perhaps from an altered gut microflora known to occur in obesity (59). Our results suggest that uterine and embryonic inflammation precede placentation, and lipotoxic-inflammatory mechanisms may in fact be present during the earliest periods of development and hence be of consequence for fetal programming.

In conclusion, we have demonstrated that exposure to maternal obesity results in uterine lipid deposition, and extensive proinflammatory gene expression changes at implantation. Increased expression of a number of genes regulating inflammation is associated with coordinated NF-κB and JNK signaling and with increased circulating insulin and fatty acid levels. Male blastocyst from obese dams also display higher inflammatory gene expression and reduced mitochondria-associated genes, including antioxidant Gpx3. Several developmental and epigenetic regulators show altered expression in obese blastocyst consistent with long-term in utero programming. These results suggest that targeting inflammatory pathways early in gestation may be an effective strategy in mitigating fetal programming. Ongoing studies are focused on early programming of the oocyte and embryo, independently, as well as in conjunction with the uterine environment.