Background
Maternal obesity is associated with adverse neurodevelopmental outcomes in children, including autism spectrum disorders, developmental delay, and attention-deficit hyperactivity disorder. The underlying mechanisms remain unclear. We previously identified second-trimester amniotic fluid and term cord blood gene expression patterns suggesting dysregulated brain development in fetuses of obese compared with lean women.
Objective
We sought to investigate the biological significance of these findings in a mouse model of maternal diet–induced obesity. We evaluated sex-specific differences in fetal growth, brain gene expression signatures, and associated pathways.
Study Design
Female C57BL/6J mice were fed a 60% high-fat diet or 10% fat control diet for 12–14 weeks prior to mating. During pregnancy, obese dams continued on the high-fat diet or transitioned to the control diet. Lean dams stayed on the control diet. On embryonic day 17.5, embryos were weighed and fetal brains were snap frozen. RNA was extracted from male and female forebrains (10 per diet group per sex) and hybridized to whole-genome expression arrays. Significantly differentially expressed genes were identified using a Welch’s t test with the Benjamini-Hochberg correction. Functional analyses were performed using ingenuity pathways analysis and gene set enrichment analysis.
Results
Embryos of dams on the high-fat diet were significantly smaller than controls, with males more severely affected than females ( P = .01). Maternal obesity and maternal obesity with dietary change in pregnancy resulted in significantly more dysregulated genes in male vs female fetal brains (386 vs 66, P < .001). Maternal obesity with and without dietary change in pregnancy was associated with unique brain gene expression signatures for each sex, with an overlap of only 1 gene. Changing obese dams to a control diet in pregnancy resulted in more differentially expressed genes in the fetal brain than maternal obesity alone. Functional analyses identified common dysregulated pathways in both sexes, but maternal obesity and maternal dietary change affected different aspects of brain development in males compared with females.
Conclusion
Maternal obesity is associated with sex-specific differences in fetal size and fetal brain gene expression signatures. Male fetal growth and brain gene expression may be more sensitive to environmental influences during pregnancy. Maternal diet during pregnancy has a significant impact on the embryonic brain transcriptome. It is important to consider both fetal sex and maternal diet when evaluating the effects of maternal obesity on fetal neurodevelopment.
Click Supplemental Materials under article title in Contents at ajog.org
Maternal obesity has reached epidemic proportions in the United States. More than one third of reproductive-age women are obese at conception, with a 70% increase in prepregnancy obesity in recent decades. Data from large epidemiological studies suggest an association between maternal obesity and adverse neurodevelopmental outcomes in children, including lower general cognitive capabilities, increased prevalence of autism spectrum disorders, and increased prevalence of attention-deficit hyperactivity disorder. Yet the mechanisms by which maternal obesity results in adverse neurodevelopmental outcomes for offspring remain unclear.
In our prior functional genomic analyses of second-trimester amniotic fluid supernatant and term umbilical cord blood in fetuses of obese vs lean women, we found gene expression patterns consistent with dysregulated brain development and dysregulated inflammatory and metabolic signaling. We transitioned to a mouse model of maternal diet–induced obesity (DIO) to evaluate the biological significance of these findings via direct examination of the fetal brain.
In our analyses of human amniotic fluid supernatant and term cord blood, subjects were matched for fetal sex, and sex-specific effects of maternal obesity on the fetus were not evaluated. Given that some animal model studies suggest sex-specific effects of maternal obesity on offspring neurodevelopment and different susceptibilities of male and female offspring to maternal dietary interventions, here we examined both male and female embryonic brains.
Our objective was to determine whether maternal obesity is associated with a distinct pattern of fetal brain gene expression and whether maternal obesity has different effects on male and female fetal brain development by characterizing embryonic brain gene expression signatures and associated pathways in a mouse model of diet-induced obesity.
Materials and Methods
Animal model and genotyping
The Tufts Medical Center Institutional Animal Care and Use Committee approved this protocol (number B2013-82); all institutional guidelines for animal care and use were followed.
Five week old female C57BL/6J mice (Jackson Laboratory, Bar Harbor, ME) were fed ad libitum either a lard-based, high-fat diet (HFD) containing 60% calories from fat (n = 22; D12492l; Research Diets, New Brunswick, NJ) or a control diet (CD) containing 10% calories from fat (n = 10; D12450J; Research Diets). The diets were matched for protein, fiber, sucrose, and micronutrient content ( Supplemental File 1 ).
Female mice were weighed weekly during the 12–14 week feeding period. Maternal obesity was defined as at least a 30% increase in weight compared with age-matched controls. The male C57BL/6J mice were fed the CD. Prior to breeding, male and female body composition was characterized by EchoMRI (EchoMRI LLC, Houston, TX).
Obese and lean females were bred with control males. The presence of a vaginal plug was defined as embryonic day (e) 0.5. Females were weighed at pregnancy day (P) 0, P10 and P15. To separate the fetal effects of prepregnancy obesity from the effect of exposure to a high-fat diet during pregnancy, the following 3 study groups were formed:
(1) C57BL/6J female mice fed a CD throughout (CD group); (2) C57BL/6J female mice fed a HFD to induce a DIO phenotype and then continued on a HFD during gestation (HFD/HFD); and (3) C57BL/6J female mice fed a HFD to induce a DIO phenotype and then switched at e0.5 to a CD for gestation (HFD/CD).
At e17.5, pregnant mice were euthanized with isoflurane followed by decapitation. Embryos were rapidly dissected from the uterine horns and placed in ice-cold phosphate-buffered saline (1 time) containing an ribonucleic acid (RNA) preservative (RNALater; QIAGEN, Valencia, CA). Crown-rump lengths and weights were recorded. Sex genotyping was performed on tail snip DNA using real-time PCR with specific probes for the Sry gene (Transnetyx, Cordova, TN).
Colony statistics were performed using GraphPad Prism 6 (GraphPad Software, San Diego, CA). Male and female embryos were analyzed separately, with differences between the 3 diet groups determined using Kruskal-Wallis testing, followed by pairwise comparisons if significance was identified (Bonferroni-corrected P < .05).
Forebrain RNA isolation and microarray methods
Forebrains were rapidly dissected from skulls and snap frozen in liquid nitrogen. Total RNA was isolated using the NucleoSpin RNA/protein kit (Macherey-Nagel, Düren, Germany). RNA purity, integrity, and quantity were assessed using the NanoDrop ND-8000 (NanoDrop, Wilmington, DE) and the bioanalyzer system (Agilent 2100; Agilent Technologies Inc, Palo Alto, CA).
RNA samples were processed using the Affymetrix GeneChip WT PLUS kit and hybridized to Mouse Gene 1.0 ST arrays (Affymetrix, Santa Clara, CA). Ten arrays per sex per experimental group were used, with each array corresponding to 1 animal. Four to 6 litters were represented in each diet group to minimize litter effects. Quality control and normalization were performed using the pipeline ( www.arrayanalysis.org ; Maastricht University, The Netherlands).
Normalization was performed using the robust multichip average algorithm and the default Affymetrix chip description file for this chip. The 27,619 main probe sets were used for further analysis; probe sets corresponding to Affymetrix controls or unmapped sequences were discarded after normalization.
Bioinformatics analysis of microarray data
Statistical analyses were performed using R software version 3.1.2. Male and female gene expression data were analyzed separately. Welch’s t tests were used to identify differentially expressed genes (DEGs) between diet groups.
P values were corrected for multiple testing by calculating the Benjamini and Hochberg false discovery rate (FDR). Probe sets with an FDR < 20% were considered significantly differentially expressed. The number of DEGs per comparison was visualized as a Venn diagram.
Gene expression changes were further visualized as a heat map combined with hierarchical clustering analysis using Euclidean distance and Ward linkage. Principal component analyses were performed using R software.
Further in silico functional analysis was performed on the top 1% of up- or downregulated genes using ingenuity pathway analysis (IPA; Ingenuity Systems, Core Analysis build version 338830M, content version 23814503). Statistical significance within the IPA was determined according to recommended thresholds ( P < .05 or bias-corrected absolute Z score ≥ 2). Only pathways containing 3 or more genes were considered. The microRNA Target Filter tool on the IPA was used to predict genes and pathways affected by dysregulated microRNAs. Only high-confidence predictions were considered.
Whole-genome analysis of functional gene set regulation was determined by Gene Set Enrichment Analysis (GSEA), using gene sets from the Developmental Functional Annotation at Tufts (DFLAT) database on fetal development (parameters as previously described). Gene sets were considered significantly regulated if the FDR q value was < 0.20. Within each sex, analyses were performed for the following diet group comparisons: HFD/HFD vs CD; HFD/CD vs CD; HFD/CD vs HFD/HFD.
Results
Maternal DIO mouse model
Dam weight trajectories prior to pregnancy by diet group are depicted in Figure 1 . Of 22 dams fed the HFD, 18 became obese after 14 weeks of feeding. Four dams did not meet criteria for obesity and were excluded. HFD-fed dams had significantly higher percentage body fat and significantly lower percentage lean body weight than control females or males by EchoMRI analysis ( Figure 2 ). Descriptive data for the dams and embryos by sex and diet group are reported in the Table . There were no significant differences between study groups with respect to litter size. There were significant differences between groups with respect to dam weight gain in pregnancy. Control dams gained the most weight in pregnancy, whereas the HFD/CD dams gained the least ( Figure 3 ).
Characteristic | Sex | Diet group | P value a | ||
---|---|---|---|---|---|
HFD/HFD | CD/CD | HFD/CD | |||
Embryo weight, g (mean ± SEM) | Male | 0.85 (0.04) | 1.02 (0.04) | 0.95 (0.04) | .01 |
Female | 0.86 (0.04) | 0.95 (0.04) | 0.88 (0.03) | .26 | |
Combined | 0.85 (0.03) | 0.98 (0.03) | 0.91 (0.03) | .007 | |
Embryo length, mm (mean ± SEM) | Male | 19.58 (0.36) | 21.00 (0.31) | 21.02 (0.42) | .02 |
Female | 19.49 (0.36) | 20.53 (0.35) | 19.87 (0.35) | .28 | |
Combined | 19.54 (0.25) | 20.78 (0.23) | 20.37 (0.28) | .01 | |
Dam weight at breeding, g (mean ± SEM) | N/A | 28.63 (0.64) | 20.32 (0.35) | 28.03 (0.99) | .002 |
Dam weight gain in pregnancy, g (mean ± SEM) | N/A | 13.18 (2.2) | 15.68 (0.96) | 8.57 (1.35) | .01 |
Litter size, n (mean ± SEM) | N/A | 7.25 (1.03) | 7.83 (0.40) | 6.33 (1.12) | .44 |
Sex-specific differences in embryo size
Male embryos exposed to maternal obesity were significantly (approximately 17%) smaller in weight than their corresponding controls ( Table ; P = .01). Female embryos demonstrated milder embryo size reduction in the setting of maternal obesity, which did not achieve statistical significance (9.5%, P = .26).
Sex-specific differences in embryonic brain gene expression profiles
Maternal obesity and maternal obesity with dietary change in pregnancy (herein after referred to as dietary change) resulted in more dysregulated genes in male compared with female fetal brains (386 in males vs 66 in females, P < .001). These numbers reflect the total number of unique DEGs in males and females for each of the diet group comparisons, including HFD/HFD vs CD, HFD/CD vs CD, and HFD/HFD vs HFD/CD.
To consider only those genes differentially regulated by maternal obesity with or without dietary change in pregnancy, we determined the sum of the HFD/HFD vs CD and HFD/CD vs CD comparison, again finding a significant difference between the sexes in the number of dysregulated genes (223 in males vs 28 in females, P < .001).
Figures 4 , A–D, depict Venn diagrams for the number of unique and common differentially expressed genes in males vs females for each comparison of maternal obesity and dietary change in pregnancy. Maternal obesity and dietary change are associated with distinct brain gene expression signatures for each sex, with overlap of only 1 gene ( nuclear encoded rRNA 5S 170 , function unknown) across all comparisons ( Figure 4 D). Maternal obesity with dietary change during pregnancy resulted in more DEGs in the fetal brain than continuation on the HFD. Supplemental Files 2 and 3 show the list of significant DEGs for the males and females for each diet group comparison.
Gene expression changes in males and females compared with their corresponding controls were further visualized as a heat map ( Figure 5 A). The hierarchical clustering and expression intensity confirm stronger gene expression changes in males v females in the setting of maternal obesity and dietary change. Within these DEGs, a block of 37 differentially expressed stem-loop microRNAs (miRNAs) were identified ( Figure 5 B). For these miRNAs, the stronger effects of maternal obesity and dietary change on male fetuses are even more pronounced.
A principal component analysis was performed to identify the dominant sources of variation in the gene expression data. The most pronounced clustering was by sex ( Figure 6 ). Subtle clustering by diet group was observed in the male but not the female brain gene expression data. There were no significant differences in within-group variability between males and females or between any of the diet groups.
The average relative variation in gene expression for each diet group was as follows: female CD/CD, 14.9%; female HFD/CD, 15.3%; female HFD/HFD, 17.4%; male CD/CD, 15.9%; male HFD/CD, 14.9%; and male HFD/HFD, 14.9%. Principal component analysis also demonstrated that gene expression in the samples cannot be distinguished by litter within the male or female diet groups (ie, no evidence of litter effects).
Pathway analyses
Pathway analyses were performed using IPA and GSEA/DFLAT. The working files for the IPA analyses may be found in Supplemental File 4 . Dysregulated biological processes and pathways in males, females, and both are depicted for each diet group comparison in Figures 7-9 .