Obesity and High-Fat Diet Induce Distinct Changes in Placental Gene Expression and Pregnancy Outcome

Abstract

Obese women are at high risk of pregnancy complications, including preeclampsia, miscarriage, preterm birth, stillbirth, and neonatal death. In the current study, we aimed to determine the effects of obesity on pregnancy outcome and placental gene expression in preclinical mouse models of genetic and nutritional obesity. The leptin receptor (LepR) null-reactivatable (LepR^loxTB), LepR-deficient (Lepr^db/+), and high-fat diet (HFD)–fed mice were assessed for fertility, pregnancy outcome, placental morphology, and placental transcriptome using standard quantitative polymerase chain reaction (qPCR) and qPCR arrays. The restoration of fertility of LepR^loxTB was performed by stereotaxic delivery of adeno-associated virus-Cre into the hypothalamic ventral premammillary nucleus. Fertile LepR^loxTB females were morbidly obese, whereas the wild-type mice-fed HFD showed only a mild increase in body weight. Approximately 80% of the LepR^loxTB females had embryo resorptions (∼40% of the embryos). In HFD mice, the number of resorptions was not different from controls fed a regular diet. Placentas of resorbed embryos from obese mice displayed necrosis and inflammatory infiltrate in the labyrinth and changes in the expression of genes associated with angiogenesis and inflammation (e.g., Vegfa, Hif1a, Nfkbia, Tlr3, Tlr4). In contrast, placentas from embryos of females on HFD showed changes in a different set of genes, mostly associated with cellular growth and response to stress (e.g., Plg, Ang, Igf1, Igfbp1, Fgf2, Tgfb2, Serpinf1). Sexual dimorphism in gene expression was only apparent in placentas from obese LepR^loxTB mice. Our findings indicate that an obese environment and HFD have distinct effects on pregnancy outcome and the placental transcriptome.

Obese women have increased risk of poor pregnancy outcome. With the use of mouse models, we found that obesity and high fat consumption have distinct effects on placental transcriptome and morphology.

Obesity is a worldwide epidemic that affects individuals of both sexes and all ages (1). In the United States, it is estimated that nearly 60% of women of reproductive age (15 to 49 years) are overweight or obese (2). Obesity is associated with reproductive deficits, but a high percentage of obese women is able to conceive. These women are exposed to increased risk of pregnancy complications and an increased probability of maternal and fetal morbidity and mortality. They are more susceptible to recurrent pregnancy loss, and they have twice the risk of preterm birth, stillbirth, and neonatal death (3–6). Congenital abnormalities, including neural tube and cardiac defects, orofacial malformations, and intestinal anomalies (e.g., anorectal atresia and omphalocele), are also more common in children of obese mothers (4, 6–8). However, the exact causes or mechanisms behind these deleterious effects are not completely known.

Obesity is a condition of chronic low-grade inflammation, characterized by sustained secretion of proinflammatory cytokines by hypertrophic adipocytes (9–11). Among them, leptin (Lep), tumor necrosis factor, and interleukin 6 are well described (8, 9, 12–14). High adiposity also induces the expression of proinflammatory mediators in the placenta, increasing the chances of placental inflammation (10, 11, 15). This outcome favors the installation of an adverse intrauterine milieu—or fetoplacental dysfunction—impairing placental blood flow and nutrient exchange (8). In nonhuman primates, an obesogenic diet increases the risk of placental infarction and stillbirth (16), and excess nutrition in pregnant ewes is associated with a reduction in placental vascularity and fetal growth restriction (17).

Obesity is also a major risk factor for the development of type 2 (pregestational) diabetes (14, 18, 19). Hyperglycemia, in the early stages of pregnancy, is associated with inadequate oocyte maturation and blastocyst development, increasing the chances of deficient implantation (20). Compromised preimplantation development may explain the higher rates of early, spontaneous miscarriages observed in pregestational diabetic women with poor glycemic control (21, 22). If pregnancy develops, hyperglycemia may alter placental angiogenesis, affecting fetoplacental circulation and normal embryo development (22–24).

The effects of obesity in placental differentiation and function are controversial, probably a result of the inherent variability of human cohorts and the high prevalence of comorbidities (e.g., diabetes, hypertension) in human populations (18, 25). Moreover, the causative role of either high adiposity or obesogenic diets on the progression of placental dysfunction has been difficult to determine as a result of limited availability of adequate preclinical models. For example, it is still unclear if the dysfunctional placenta and poor pregnancy outcome reported in animal models of diet-induced obesity are caused by the adiposity of the mother or by the components of the diet.

In the current study, we performed a comparative analysis of genetic obese mice fed a regular chow diet (RD) and wild-type mice fed a high-fat diet (HFD) to unravel differences in fertility, pregnancy outcome, placental transcriptome, and placental morphology. We hypothesized that obesity in mice recapitulates the poor pregnancy outcome observed in obese women (miscarriages and stillbirths) and that high adiposity and high fat consumption perturb distinct subsets of placental genes required for the maintenance of pregnancy.

Materials and Methods

Mice

The Lep receptor (LepR)^loxTB (kindly provided by Dr. Joel Elmquist, The University of Texas Southwestern Medical Center, Dallas, TX; available in JAX^® mice, stock no. 018989) (26), Lepr^db/+ (JAX^® mice, stock no. 000697), and C57BL/6 mice (JAX^® mice, stock no. 000664) were housed in the University of Michigan Unit for Laboratory Animal Medicine facility with 12-hour light/dark cycles. They were fed phytoestrogen-reduced standard chow (16% protein/4% fat, 2016 Teklad Global Diet; Envigo) until breeding, at which point, they were fed phytoestrogen-reduced breeder’s chow (19% protein/9% fat, 2019 Teklad Global Diet; Envigo). Phytoestrogen-reduced diets were used to avoid the effects of exogenous estrogens on the physiology of the mice. Another group of mice was fed an HFD (42%, TD.88137, or 60% fat, TD.06414, Teklad Global Diet; Envigo) for 12 or 16 weeks.

Stereotaxic injections and breeding

Stereotaxic injection of adeno-associated virus (AAV) vectors, expressing Cre and/or green fluorescent protein (GFP; AAV-Cre/GFP or AAV-GFP; Vector Core, University of North Carolina) unilaterally into the ventral premammillary nucleus (PMV) of female LepR^loxTB/loxTB mice at 7 to 10 weeks of age, was performed. Cre-mediated excision of the loxP sites induces endogenous expression of the Lepr gene. As the LepR^loxTB mice are infertile at baseline, we crossed mice heterozygous for the LepR mutation to generate our experimental group. Genotypes were defined using DNA extracted from the tail and the following primers: forward (F) 5′ CAG TCT GGA CCG AAG GTG TT 3′ and reverse (R) 5′ TAG GGC CAA ACC CAC ATT TA 3′. Mice homozygous for loxTB were used as the experimental group (AAV-Cre/GFP injections), and mice homozygous for the wild-type allele were used as controls (AAV-GFP injections). Mendelian ratios were obtained; 25% of the offspring were experimental, 50% were heterozygous, and 25% were control. Heterozygous mice were used as breeders, as mentioned previously. Females were housed together postoperatively. External signs of pubertal maturation were evaluated daily by vaginal opening for 2 to 3 weeks. Experimental and control mice were housed with wild-type adult male mice of proven fertility to breed. The females were weighed every other day, and when the slope of the weight-gain trajectory increased, the female mice were euthanized. This strategy was used, because in pilot experiments, we found that copulatory plugs were not a reliable indication of successful fertilization in obese mice (27).

Histology and immunohistochemistry

Before euthanasia, mice were deeply anesthetized with isoflurane (2% to 4%, Fluriso; Vet One). Brain, placentas, and embryos were harvested, and blood samples were collected directly from the heart. Brains were snap frozen on dry ice and stored at −80°C. Reproductive organs and embryos were postfixed in buffered formalin for histological analyses. The placentas were microdissected under a dissecting stereoscope (Zeiss), and maternal tissue was removed. They were divided into two groups, one for gene expression analyses, which was snap frozen on dry ice, and the other for histological analyses, which was placed in buffered formalin. Brains were cut on a cryostat (30 μm sections, five series frontal plane) and stored at −20°C. Uteri, ovaries, embryos, and placentas were processed for standard paraffin-embedded sectioning and hematoxylin and eosin staining. Placental tissue was also processed for markers of endothelial [angiogenesis, platelet and endothelial cell adhesion molecule 1 (Pecam1), 1:50, catalog no. DIA-310; Dianova, Research Resource Identifier (RRID): AB_2631039] (28) and immune cells [neutrophils, lymphocyte antigen 6 complex locus G6D (Ly6G), 1:200, catalog no. 551459; BD Biosciences, RRID: AB_394206] (29) using standard immunoperoxidase and diaminobenzidine as chromogen. Brain sections were processed for the detection of GFP immunoreactivity (1:10,000, catalog no. GFP-1010; Aves Laboratories, RRID: AB_2307313) (30, 31) using standard immunoperoxidase and diaminobenzidine as chromogen.

In situ hybridization

Adult female wild-type mice (n = 4) were perfused transcardially with 10% buffered formalin; brains were dissected out, cryopreserved overnight in 20% sucrose, and sectioned in a freezing microtome (30 μm sections, five series frontal plane). To compare the pattern of GFP labeling and LepR messenger RNA (mRNA) distribution in the PMV, the mouse LepRb was used as described previously (32). The following primers and T3/T7 polymerases were used to amplify a 400–base pair (bp) fragment: F 5′ AAA GAG CTC CTT CTC TGG GTC TCA GAG CAC 3′ and R 5′ AAA AAG CTT CTC ACC AGT CAA AAG CAC ACC AC 3′. Tissue sections were mounted onto SuperFrost plus slides (Fisher Scientific) and microwaved in sodium citrate buffer, pH 6.0, for 10 minutes. The ³⁵S-labeled LepRb riboprobe was diluted to 10⁶ cpm/mL in a hybridization solution containing 50% formamide, 10% dextran sulfate, and 1× Denhardt solution. The riboprobe was applied to each slide following incubation overnight at 57°C. Slides were treated with 0.02% RNase A (Roche) and subjected to stringency washes in saline-sodium citrate buffer. Slides were placed in X-ray film cassettes with BMR-2 film (Kodak) for 2 days, dipped in NTB-2 autoradiographic emulsion (Kodak) for 2 weeks. Slides were developed with D-19 developer (Kodak), dehydrated in graded ethanol, cleared in xylenes, and coverslipped with DPX (Sigma-Aldrich).

Quantitative polymerase chain reaction

Placentas were initially submitted to a quantitative polymerase chain reaction (qPCR) array, and the validation was performed using standard qPCR. RNA was extracted using the Qiazol lysis reagent (RNeasy Mini Kit; Qiagen). The cDNA was generated using 1 μg total RNA from each placenta (RT² First Strand Kit; Qiagen). The mRNA quantification was carried out using a mouse angiogenesis qPCR array with 84 genes per plate and 12 control wells (RT² Profiler PCR Array; Qiagen). The genes were normalized to five housekeeping genes, and the resulting values were expressed as fold change, above or below control levels. Data from the qPCR array were represented using Prism 7 software (GraphPad Software). For validation, qPCR was performed on a CFX-384 reverse transcription PCR detection system (Bio-Rad) using SYBR^® Green Gene Expression Assays and pairs of primers designed by Sigma-Aldrich or Integrated DNA Technologies. The housekeeping gene 18S ribosomal RNA (18s) was used as an internal reference. The amplifications were performed in a total volume of 10 μL and included 5 μL of 2× SYBR Green Master Mix Reagent (Qiagen), 1 μL cDNA, and 0.5 μL each primer (500 nmol/L). All primer sequences are listed in Table 1. To define the sex of the embryos, genomic expression of Sry and Ddx3y genes was used as specific male markers. All reactions were run in triplicate and included negative controls with no template. Relative fold expression of each gene was calculated using the comparative cycle threshold method, normalizing to the internal 18s reference and comparing with the control group.

Table 1.

Sequence of Primers Used for qPCR

Genes	Forward	Reverse
18s	F: 5′ TGACTCAACACGGGAAACCT 3′	R: 5′ AACCAGACAAATCGCTCCAC 3′
Ang	F: 5′ CAGCTTTGGAATCTCTGTTG 3′	R: 5′ GCTTCTTCTCTTCATCATAGG 3′
Angpt2	F: 5′ AATAAGTAGCATCAGCCAAC 3′	R: 5′ AGTAGTACCACTTGATACCG 3′
Bai1	F: 5′ TCATGCTGGTCATCATCTAC 3′	R: 5′ AAGGATGAGAGCATTAGAGG 3′
Bhlhe40	F: 5′ CAGGCGGGGAATAAAACGGA 3′	R: 5′ GGGCACAAGTCTGGAAACCT 3′
Cd36	F: 5′ AATTTGTCCTATTGGCCAAGCT3′	R: 5′ AGCGTAGATAGACCTGCAAATG 3′
Cpt1b	F: 5′ CTTAGCCTCTACGGCAAAGC 3′	R: 5′ CCACGAGTGTTCGGTGTTGA 3′
Crh	F: 5′ AGGCATCCTGAGAGAAGTCC 3′	R: 5′ ACGACAGAGCCACCAGCA 3′
Csf3	F: 5′ CAGATGGAAAACCTAGGGG 3′	R: 5′ TGGAAGGCAGAAGTGAAG 3′
Ctgf	F: 5′ GAGGAAAACATTAAGAAGGGC 3′	R: 5′ AGAAAGCTCAAACTTGACAG 3′
Edn1	F: 5′ GAAGTGTATCTATCAGCAGC 3′	R: 5′ GCACTATATAAGGGATGACTTC 3′
Egf	F: 5′ GTACTCTTGGGTGTGAAAAC 3′	R: 5′ CAAGTTCGTGACATTGTTTC 3′
F2	F: 5′ AACATCAATGAGATACAGCC 3′	R: 5′ GCTCTTCATGACAAAGGTC 3′
Fgf2	F: 5′ CTGGCTTCTAAGTGTGTTAC 3′	R: 5′ GAAGAAACAGTATGGCCTTC 3′
Figf	F: 5′ CTCTTTGAGATATCAGTGCC 3′	R: 5′ GAGGACATTCATCTTCTTCTG 3′
Flt1	F: 5′ CAAGAGCGATGTGTGGTCCT 3′	R: 5′ TCCCATCCTGTTGGACGTTG 3′
Hif1a	F: 5′ ACCTTCATCGGAAACTCCAAAG 3′	R: 5′ ACTGTTAGGCTCAGGTGAACT 3′
Ifng	F: 5′ TGAGTATTGCCAAGTTTGAG 3′	R: 5′ CTTATTGGGACAATCTCTTCC 3′
Igf1	F: 5′ GACAAACAAGAAAACGAAGC 3′	R: 5′ ATTTGGTAGGTGTTTCGATG 3′
Igfbp1	F: 5′ CCCTGCCAACGAGAACTCTA 3′	R: 5′ TCTCCATCCAGGGATGTCTC 3′
Il1b	F: 5′ GGATGATGATGATAACCTGC 3′	R: 5′ CATGGAGAATATCACTTGTTGG 3′
Il6	F: 5′ AAGAAATGATGGATGCTACC 3′	R: 5′ GAGTTTCTGTATCTCTCTGAAG 3′
Insl3	F: 5′ AAGAAGCCCCATCATGACCT 3′	R: 5′ TTATTTAGACTTTTTGGGACACAGG 3′
Lep	F: 5′ TCTCCGAGACCTCCTCCATCT 3′	R: 5′ TTCCAGGACGCCATCCAG 3′
Lepr	F: 5′ CCTCTTGTGTCCTACTGCTCG 3′	R: 5′ GAAATTCAGTCCTTGTGCCCAG 3′
Mdk	F: 5′ CAAAGCCAAGAAAGGAAAG 3′	R: 5′ CACTGGTGGGTTATATCTTG 3′
Nfkbia	F: 5′ CCACTCCACTTGGCTGTGAT 3′	R: 5′ GACACGTGTGGCCATTGTAG 3′
Nos3	F: 5′ AAAGCTGCAGGTATTTGATG 3′	R: 5′ AGATTGCCTCTATTTGTTGC 3′
Pecam1	F: 5′ CATCGCCACCTTAATAGTTG 3′	R: 5′ CCAGAAACATCATCATAACCG 3′
Plg	F: 5′ ATTTCCCCTGCAAAAATCTG 3′	R: 5′ ATGGAATCTCACAGTACTCC 3′
Serpinf1	F: 5′ CCAAGTTTGACTCGAGAAAG 3′	R: 5′ CAATCTTGCAGTTGAGATCAG 3′
Slc2a1	F: 5′ GGGAGAGGTGTCACCTACAGC 3′	R: 5′ ATTGCCCATGATGGAGTCTAA 3′
Slc2a3	F: 5′ GGAGCAGGCGTGGTCAATAC 3′	R: 5′ TGAAAACGGAGCAAACAGCC 3′
Sphk1	F: 5′ GGTACTCTCATCTCGACTTC 3′	R: 5′ GCCAGATTTTTAGCTTCCAG 3′
Tgfb2	F: 5′ GAGATTTGCAGGTATTGATGG 3′	R: 5′ CAACAACATTAGCAGGAGATG 3′
Timp2	F: 5′ GGATTCAGTATGAGATCAAGC 3′	R: 5′ GCCTTTCCTGCAATTAGATAC 3′
Tlr3	F: 5′ GTCTTCTGCACGAACCTGAC 3′	R: 5′ GAGCAGTTCTTGGAGGTTCTC 3′
Tlr4	F: 5′ AATCCCTGCATAGAGGTAGTTCCTA 3′	R: 5′ GTCTCCACAGCCACCAGATT 3′
Tnf	F: 5′ CTATGTCTCAGCCTCTTCTC 3′	R: 5′ CATTTGGGAACTTCTCATCC 3′
Tymp	F: 5′ GCAACTGGAGTGGCCCAAAG 3′	R: 5′ GATCCCATCAGCAGGAACCA 3′
Vegfa	F: 5′ TAGAGTACATCTTCAAGCCG 3′	R: 5′ TCTTTCTTTGGTCTGCATTC 3′

Abbreviations: Ang, angiogenin; Angpt2, angiopoietin 2; Bai1, brain-specific angiogenesis inhibitor 1; Bhlhe40, basic helix-loop-helix family member E40; CD36, cluster of differentiation 36; Cpt1b, carnitine palmitoyltransferase 1b; Crh, corticotropin-releasing hormone; Csf3, colony-stimulating factor 3; Ctgf, connective tissue growth factor; Edn1, endothelin 1; Egf, epidermal growth factor; F2, coagulation factor 2; Fgf2, fibroblast growth factor 2; Flt1, fms-related tyrosine kinase 1 (VegfR1); Hif1a, hypoxia-inducible factor-1 α subunit; Ifng, interferon γ, Igf1, insulinlike growth factor 1; Igfbp1, insulinlike growth factor-binding protein 1; Il1b, interleukin 1b; Il6, interleukin 6; Insl3, insulinlike 3; Mdk, midkine; Nfkbia, NF κB inhibitor α; Nos3, nitric oxide synthase 3; Plg, plasminogen; Serpinf1, serpin family F member 1; Slc2a1, glucose transporter 1 (Glut1); Slc2a3, glucose transporter 3 (Glut3); Sphk1, sphingosine kinase 1; Tgfb2, transforming growth factor β 2; Timp2, tissue inhibitor of metalloproteinases; Tlr3, Toll-like receptor 3; Tlr4, Toll-like receptor 4; Tymp, thymidine phosphorylase; Vegfa, vascular endothelial growth factor a.

Data analysis, statistics, and production of photomicrographs

Sections of brain, uterus, ovary, placenta, and embryo were analyzed using an Axio Imager M2 microscope (Zeiss). Photomicrographs were produced by capturing images with a digital camera (Axiocam; Zeiss), mounted directly on the microscope using Zen software. Image-editing software (CS6; Adobe Photoshop) was used to combine photomicrographs into plates. Only sharpness, contrast, and brightness were adjusted. Data are expressed as means ± standard error of the mean. Comparisons between two groups were carried out using the unpaired two-tailed Student t test. One-way analysis of variance (ANOVA), followed by the pairwise Tukey test, was used to compare three or more groups simultaneously. Statistical analyses were performed using Prism 7 software (GraphPad Software), and an α value (P) of 0.05 was considered in all analyses.

Study approval

All animal studies were carried out in accordance with the guidelines established by the National Institutes of Health Guide for the Care and Use of Laboratory Animals and approved by the University of Michigan Institutional Animal Care and Use Committee (PRO00004380; Ann Arbor, MI).

Results

Mouse models of obesity show poor pregnancy outcome

In previous studies from our laboratory (33), we observed that the obese LepR-null mice show high rates of embryo resorptions and fetal death. These mice recapitulate the phenotype of the LepR-deficient db/db (Lepr^db) mice (34). They are obese, diabetic, and infertile. Following endogenous LepR reactivation in PMV neurons, we observed an improvement in the fertility of female LepR-null mice but a low rate of successful pregnancies at term. Of five pregnant females, three showed resorptions, and one delivered pups with malformations (Fig. 1A). The only female to deliver apparently healthy pups was the leanest of all (37.3 vs 45.6 ± 1.4 g compared with 18.10 ± 0.7 g in wild-type mice). This observation suggested that the extreme obesity of the LepR-null females had a negative impact on the progression of pregnancy.

Figure 1.

Poor pregnancy outcome in LepR^loxTB obese mice with re-expression of LepR in the PMV. (A) Embryos with malformations delivered by a PMV-hit LepR-null (LepR^neo/neo) female mouse. (B and C) Implantations in (B) wild-type and (C) PMV-hit LepR^loxTB mice, euthanized at midgestation [embryonic day (E)10.5)]. (C) Note the high number of resorptions (arrows) and high adiposity. (D) Darkfield images showing the distribution of LepR mRNA in the PMV of a wild-type female mouse. (E) Brightfield image showing injection of AAV-Cre/GFP targeting the PMV (example of a PMV-hit). Image shows GFP immunoreactive (GFP-ir) cells. (F) Brightfield image showing a LepR^loxTB mouse with AAV-Cre/GFP outside the PMV (PMV-miss). (D–F) Original scale bar, 300 μm. 3V, third ventricle.

To test this hypothesis, we used a similar mouse model functionally null for LepR—the LepR^loxTB mouse that has loxP sites flanking a transcription blocking cassette, inserted between exons 16 and 17 of the Lepr gene (26, 31, 35). This mouse line is also obese, diabetic, and infertile. One cohort of LepR^loxTB (n = 18) was injected with AAV-Cre/GFP into the PMV to re-express endogenous LepR and restore sexual maturation and fertility. Of the 18 injected LepR^loxTB mice, 11 correctly targeted the PMV (named PMV-hits). All 11 showed sexual maturation (vaginal opening), but only five became pregnant in 4 weeks of fertility testing. Of the five pregnant mice, three were euthanized at midgestation owing to a sudden decrease in body weight. All three females had a normal number of implantations (8.6 ± 0.6) but a high percentage of resorptions (54.2% ± 4.2; Fig. 1B and 1C). The other two pregnant mice delivered pups at term. One female delivered eight normal-appearing pups; the other delivered three pups and was euthanized shortly after as a result of delivery complications. Seven fetuses were retained, six of which were dead. The three pups that were delivered had no signs of developmental abnormalities. Control littermates (n = 8) had no pregnancy complications.

Metabolic phenotype and pregnancy outcome of the obese fertile and infertile LepR^loxTB mice

Two more cohorts of LepR^loxTB mice (first cohort n = 21 females; second cohort n = 8 females) and wild-type littermates (first cohort n = 13 females, six with AAV-GFP injection; second cohort n = 7 without injections) were evaluated. At the time of euthanasia, brains were collected and sections processed for immunohistochemistry to detect GFP immunoreactivity and confirm the injection sites (Fig. 1D–1F). Of the 29 injected mice, 19 had injections targeting the PMV (PMV-hits). All PMV-hits showed a vaginal opening (external sign of puberty onset) between postoperative days 6 and 14 (average 9.9 ± 0.5 days). The mice were then bred with wild-type males of proven fertility; 10 females became pregnant after 6 weeks of fertility testing. Pregnant (n = 10) and nonpregnant (n = 9) PMV-hit mice showed no differences in body weight on the day of surgery, but by 12 weeks of age (before pregnancy), the nonpregnant group was significantly heavier (Fig. 2A and 2B). As expected, as a result of sexual maturation and increase in sex steroids, PMV-hits showed decreased weight gain compared with those outside the PMV (PMV-misses; at 12 weeks of age; Fig. 2C). Increased latency to pregnancy was observed compared with wild-types (Fig. 2D). Females were euthanized at midpregnancy [embryonic day (E)11.5 to E15.5]. At this time window, the number of implantations was similar to controls (P = 0.26), but a substantially higher percentage of resorptions was apparent in PMV-hit mice (P = 0.01; Fig. 2E).

Figure 2.

Metabolic phenotype and pregnancy outcome of LepR^loxTB mice with re-expression of LepR in the PMV. (A–C) Bar graphs showing differences in body weight and weight gain of LepR^loxTB mice following reactivation of LepR in the PMV (PMV-hits). (B) Note that nonpregnant females have higher body weight following surgery. (D–F) Bar graphs showing (D) increase in latency to pregnancy, (E) percentage of resorptions, and (F) normalization of uterus weight to wild-type (WT) levels in PMV-hits. (G–O) Brightfield images showing sexual maturation [increased uterus size and presence of corpora lutea (CL)] in PMV-hit compared with PMV-miss mice. (H, K, and N) Higher magnification of G, J, and M (boxes), respectively. Note development and the myometrium (Myo) and endometrium (Endo) layers in PMV-hit mice. *P < 0.05; **P < 0.01; ***P < 0.001; ****P < 0.0001. Unpaired two-tailed Student t test for comparisons between two groups; one-way ANOVA, followed by the pairwise Tukey test for comparisons of three groups; and (B) two-way ANOVA, followed by Sidak multiple comparison test. Original scale bar, (G, I, J, L, M, and O) 3 mm; (H, K, and N) 200 μm. Fol, follicles; Lu, lumen.

As all PMV-hits showed vaginal opening, but approximately one-half of them had no signs of pregnancy, we assessed uterine and ovarian morphology. Nonpregnant PMV-hit mice had a comparable uterine size to wild-type diestrous mice, whereas the uterus of PMV-misses had no signs of pubertal development (Fig. 2F–2I). The ovaries of PMV-hits had corpora lutea, which was not observed in the ovaries of PMV-misses (Fig. 2J–2L). These data indicate that although re-expression of LepR in PMV neurons had restored sexual maturation and ovulation, a subset of these mice was unable to become pregnant or sustain pregnancy.

As expected (26), LepR^loxTB mice had higher fasting glucose levels before surgery compared with wild-type littermate controls (Fig. 3A). Random glucose and insulin levels were obtained at the time of euthanasia. Fasting glucose levels and glucose or insulin tolerance were not assessed to avoid interference with pregnancy progression and gene-expression analysis. The obese PMV-miss mice had very high glucose and insulin levels compared with wild-type pregnant mice. PMV-hits also had higher glucose levels compared with wild-type controls, but no difference in insulin levels was detected as a result of high variability in obese females. Glucose levels were lower in PMV-hit pregnant compared with PMV-hit nonpregnant mice (Fig. 3B and 3C).

Figure 3.

LepR^loxTB mice have high glucose and insulin levels but no difference in the expression of the Lepr gene in the placentas. (A) Bar graphs showing fasting glucose levels of wild-type and LepR^loxTB mice before surgery. Note the high glucose levels of LepR-null mice. (B and C) Bar graphs showing glucose levels of wild-type and LepR^loxTB mice before euthanasia. Note that although an improvement in glucose levels was observed in PMV-hits, they have not been normalized to wild-type levels. Furthermore, note the higher glucose levels in nonpregnant vs pregnant PMV-hits. (D) Image of a gel showing the genotyping of eight embryos (e1 to e8) from a LepR-null PMV-hit dam. All embryos are heterozygous for the Lepr allele (wild-type allele = 540 bp, LepR-null allele = 300 bp). (E and F) Bar graphs showing relative mRNA expression of Lepr and Lep genes in placentas of male and female embryos from wild-type (white bars) and PMV-hit LepR^loxTB (black bars) dams. No difference was noticed between genotypes. (G–J) Bar graphs showing (G) body weight, (H) blood glucose, (I) number of implantations, and (J) percentage of resorptions comparing wild-type with Lepr^db/+ mice. A small difference in body weight between genotypes was observed. Unpaired two-tailed Student t test for comparisons between two groups and one-way ANOVA, followed by the pairwise Tukey test for comparisons of three or more groups. Different letters represent differences (P < 0.05) comparing groups. *P < 0.05; ***P < 0.001.

Expression of the Lepr gene is not altered in placentas of embryos from LepR^loxTB mice

The genotypes of one litter of a PMV-hit female (n = 8 embryos) were assessed for validation. As expected, all embryos were heterozygous for the LepR-null mutation (LepR^loxTB/+; Fig. 3D); therefore, the corresponding placentas had one functional copy of the Lepr gene. To assess if the poor pregnancy outcome of the PMV-hit LepR^loxTB mice was caused by a decrease in the expression of Lepr gene, we performed qPCR analysis on the placentas from PMV-hit and from wild-type male and female embryos. No differences in the expression of either Lepr or Lep genes were observed between the genotypes in both sexes (Fig. 3E and 3F).

To assess if lack of one copy of the Lepr gene predisposes mice to embryo resorption, we bred mice heterozygous for the Lepr gene mutation (Lepr^db/+) with wild-type males to generate Lepr^db/+ and wild-type (Lepr^+/+) embryos in the same litter. Compared with wild-type females, Lepr^db/+ had slightly higher body weight on the first mating day (10 weeks of age, difference between means = 3.7 ± 1.6 g, P = 0.048), but glucose levels were comparable (Fig. 3G and 3H). No differences in latency to pregnancy (P = 0.89), number of implantations, or percentage of resorptions were observed (Fig. 3I and 3J). The genotyping of the placentas of all resorbed embryos from Lepr^db/+ mice (n = 11) showed that 45.5% of embryos were wild-types (Lepr^+/+, n = 5), and 55.5% were Lepr^db/+ (n = 6), suggesting that lack of one functional Lepr allele in the fetoplacental unit cannot explain the increased rate of resorptions in LepR^loxTB mice.

Placentas of embryos from obese LepR^loxTB mice show necrosis and defective angiogenesis

Placentas from wild-type and LepR^loxTB pregnant mice at midgestation were submitted for histological analyses (n = 4 placentas from wild-type, and n = 7 placentas from LepR^loxTB mice). Three out of four placentas from wild-type embryos showed small, focal areas of necrosis (Fig. 4A), although overall, the tissue architecture was preserved. Moderate-to-severe necrosis was observed in all seven LepR^loxTB resorbed placentas (Fig. 4A–4C). Large zones of placental necrosis centered on the vasculature, with influx of inflammatory cells into the vascular walls were also evident (Fig. 4D–4F). Necrosis within the placental labyrinth and/or the syncytiotrophoblast layer within the basal zone and the decidua basalis was often associated with fibrin deposition and inflammatory infiltrate. In affected areas, we observed dilation of labyrinth blood vessels and altered expression of angiogenesis markers, Pecam1, a.k.a. as CD31 (Fig. 4G–4I). Pecam1 is a membrane glycoprotein associated with cell-to-cell adhesion and angiogenesis (36). There were also multifocal areas of low-to-mild mineralization, randomly scattered within the labyrinth in all resorbed placentas, and mild cystic structures in the basal zone.

Figure 4.

Placentas of embryos from obese LepR^loxTB mice show necrosis, inflammatory infiltrate, and defective angiogenesis. (A–C) Brightfield images showing sections of placentas from (A and B) LepR^loxTB and (C) wild-type for comparison. Note the zone of necrosis within labyrinth zone (asterisk), bordered by spongiotrophoblasts at the margin of necrosis (arrowheads) and multifocal foci of neutrophilic inflammation (arrow). (D) Higher magnification of region denoted by arrow in (B), showing focal infiltration of neutrophils and fibrin adjacent to zone of necrosis. (E and F) Ly6G immunohistochemistry for neutrophils in region shown in (C). Several inflammatory cells show diffuse immunoreactivity to antibody for Ly6G (Ly6G-ir), indicative of neutrophil lineage. (G) Higher magnification of the spongiotrophoblasts layer at the margin of necrosis denoted by the arrowhead in (B). (H and I) Pecam1 immunohistochemistry for endothelial cells in the region shown in (G). Positive immunoreactivity (Pecam1-ir) in the labyrinth zone and spongiotrophoblast layer consistent with endothelial cells. Note the disorganized structure in (H) compared with (I). Original scale bar, (A) 400 μm; (B and C) 200 μm; (D–I) 100 μm. BZ, basal zone; DB, decidua basalis; GC, giant cell layer; LZ, labyrinth zone.

Differentially expressed genes in the placentas of embryos from obese LepR^loxTB mice

Because of the histopathological findings, we initially ran a mouse angiogenesis qPCR array for gene-expression analyses. We compared placentas of resorbed embryos from PMV-hit LepR^loxTB mice with placentas of normal developing embryos from wild-type mice at midgestation (Fig. 5A). Of 10 obese LepR^loxTB pregnant females, eight had resorptions (80%). In two females, resorptions were at advanced stages and were not used for analysis. Of the remaining six obese females with resorbed placentas, we obtained five placentas from female embryos and one from a male embryo. For consistency, only placentas from female embryos were used in the array. Genes within 2.5-fold difference in expression were reassessed using standard qPCR for data validation (one placenta per dam). Expression of the housekeeping gene 18s was not different between groups (mean comparative cycle threshold values: 11.86 ± 0.4 in controls vs. 12.05 ± 0.3 in obese, P = 0.91). The differentially expressed gene (DEG) clustered into four functional categories: angiogenesis (Vegfa, Hif1a), inflammation (Nfkbia, Il6, Il1b, Nos3, Pecam1, Tlr3, Tlr4), cellular growth (Mdk, Ctgf), and response to stress (Edn1, Crh; Fig. 5B–5E).

Figure 5.

Gene-expression analysis of resorbed placentas from obese LepR^loxTB mice and normal placentas from wild-type mice at midgestation. (A) Heat map showing the relative expression levels of genes associated with angiogenesis in mice. (B–E) Bar graphs showing expression levels of genes found on the qPCR array to be 2.5× differentially expressed compared with controls. The genes were clustered in four functional categories, i.e., (B) angiogenesis, (C) inflammation, (D) cellular growth, and (E) response to stress. Unpaired two-tailed Student t test for comparisons between two groups. *P < 0.05; **P < 0.01; ***P < 0.001. R, resorbed; Wt, wild-type.

We further assessed gene expression in placentas with no morphological signs of resorption (classified as normal) from embryos of obese LepR^loxTB and control wild-type females. Placentas from male and female embryos were evaluated independently (n = 4 to 6 per genotype and sex). Striking sex differences were observed. Increased expression of genes associated with angiogenesis (Vegfa, Bai1, Angpt2, Bhlhe40, Hif1a, Flt1), inflammation (Nfkbia, Pecam1, Tlr3, Tlr4), cellular growth (Egf, Sphk1, Ctgf), and response to stress (Tgfb2, Edn1, Crh) was observed in placentas from female embryos of LepR^loxTB dams (Fig. 6A–6D). In contrast, only Insl3 expression was augmented in normal placentas from male embryos of obese dams (P = 0.02, Supplemental Table 1).

Figure 6.

Gene-expression analysis of placentas from obese LepR^loxTB and from mice on an HFD at midgestation. (A–H) Bar graphs showing relative expression of genes associated with (A and E) angiogenesis, (B and F) inflammation, (C and G) cellular growth, and (D and G) response to stress (A–D) in placentas from female embryos of obese LepR^loxTB mice and (E–H) in placentas from male and female embryos of mice fed an HFD. Unpaired two-tailed Student t test for comparisons between two groups. *P < 0.05; **P < 0.01; ***P < 0.001; ****P < 0.0001; ^#P < 0.1 (trend).

Because Lep levels are increased in the obese mice (26), and placental Lepr is not altered, we assessed the expression of Lep-associated downstream signaling molecules in the placentas of male and female embryos. We found an increase in the expression of Stat3 and Foxo1 (P < 0.05) and a trend to increase in Socs3 (P = 0.08) only in placentas from female embryos. No difference was detected in placentas from male embryos. Glucose transporter 1 (Slc2a1) and lipid transporters (Cd36 and Cpt1b) were decreased (P < 0.05) in resorbed and nonresorbed placentas of embryos from obese LepR^loxTB mice. No difference in glucose transporter 3 (Slc2a3) was observed (Supplemental Table 1).

Pregnancy outcome in mice fed an HFD

To assess if the findings in LepR^loxTB mice are recapitulated in mice on an obesogenic diet, two cohorts of females fed an HFD were evaluated. The first cohort consisted of Lepr^db/+ mice fed an HFD (42%, n = 7) for 12 weeks, and the second cohort consisted of wild-type mice fed an HFD (60%, n = 7) for 16 weeks.

In the first cohort, the Lepr^db/+ mice were used to assess the potential effect of a lack of one Lepr allele in an obesogenic environment. Only a small, nonsignificant weight gain (29.61 ± 1.03 g in Lepr^db/+ vs 27.27 ± 1.2 g in control mice fed RD, n = 6 to 7, P = 0.17) and no change in glucose levels (168.4 ± 6.7 mg/dL in Lepr^db/+ vs 170.5 ± 8.8 mg/dL in control mice fed a RD, n = 6 to 7, P = 0.85) were observed. The percentage of resorptions was not different between groups (16.2 ± 8.8% in Lepr^db/+ vs 10.9 ± 6.3% in control mice, n = 6 to 7, P = 0.65).

A second cohort consisting of wild-type (C57BL/6J) mice fed an HFD (60% fat) for 16 weeks was analyzed. In this cohort, an increase in body weight (28.7 ± 0.28 g in HFD mice vs 20.3 ± 1.12 g in control mice fed a RD, n = 7/group, P < 0.01) but no difference in blood glucose levels (184.7 ± 6.4 mg/dL in HFD mice vs 181 ± 3.7 mg/dL in control mice, n = 7/group, P = 0.63) was observed. All control wild-type mice on a RD were fertile, and latency to pregnancy spanned from 2 to 4 days. Of the seven females on 60% HFD, one did not become pregnant (no implantations) in 2 months of the fertility trial, and one had one dead embryo (one implantation) at the time of euthanasia (∼E18.5). In the remaining five females on an HFD, no difference in the percentage of implantations or resorptions was observed compared with control mice (P = 0.52 and P = 0.68, respectively).

Placental DEG of embryos from mice on an HFD are distinct from those of obese LepR^loxTB mice on an RD

Wild-type females fed an HFD showed a mild increase in body weight and a small impairment in pregnancy outcome. To assess the potential differences in gene expression caused by the diet, we compared the expression levels of genes identified in the qPCR array (2.5-fold difference in obese LepR^loxTB mice). In contrast to findings in obese PMV-hit LepR^loxTB mice, differences in gene expression were observed in placentas from embryos of both sexes. In HFD mice, we found that the DEG were associated with cellular growth and response to stress in female and male placentas (Plg, Ang, Ctgf, Igf1, Igfbp1, Fgf2, Tgfb2, Serpinf1, F2, Csf3, n = 6/group; Fig. 6, Supplemental Fig. 1, and Supplemental Table 1). Virtually no sexual dimorphism was apparent.

Discussion

We observed poor pregnancy outcome following restoration of the reproductive function in the obese LepR^loxTB mice. The obese mice had increased latency to pregnancy and a high percentage of embryonic resorptions per litter. These effects are not a result of the lack of one copy of the Lepr gene in the placentas, as roughly equal numbers of wild-type and heterozygous resorptions were observed in the Lepr^db/+ litter. The histological analyses revealed necrosis in the labyrinth (the site of maternal fetal exchange) and inflammatory infiltrate centered on blood vessels. The differences in gene expression between resorbed or normal placentas from obese mice and normal placentas from wild-type mice also support a mechanism of defective angiogenesis and inflammation. Notably, placentas from mice on an HFD showed a distinct set of DEG associated with cellular growth and response to stress (Fig. 7), suggesting that an HFD may alter the genetic program in placental function independent of obesity.

Figure 7.

Schematic illustration of genes altered in nonresorbed placentas of embryos from obese LepR^loxTB (PMV-hit) and wild-type mice fed an HFD. Ang, angiogenin; Angpt2, angiopoietin 2; Bai1, brain-specific angiogenesis inhibitor 1; Bhlhe40, basic helix-loop-helix family member E40; CD36, cluster of differentiation 36; Cpt1b, carnitine palmitoyltransferase 1b; Crh, corticotropin-releasing hormone; Csf3, colony-stimulating factor 3; Ctgf, connective tissue growth factor; Edn1, endothelin 1; Egf, epidermal growth factor; F2, coagulation factor 2; Fgf2, fibroblast growth factor 2; Flt1, fms-related tyrosine kinase 1 (VegfR1); Foxo1, forkhead box o1; Hif1a, hypoxia-inducible factor 1 α subunit; Igf1, insulinlike growth factor 1; Igfbp1, insulinlike growth factor–binding protein 1; Nfkbia, NF κ B inhibitor alpha; Plg, plasminogen; Serpinf1, serpin family F member 1; Slc2a1, glucose transporter 1 (Glut1); Shpk1, sphingosine kinase 1; Stat3, signal transducer and activator of transcription 3; Tgfb2, transforming growth factor β 2; Timp2, tissue inhibitor of metalloproteinases; Tlr3, Toll-like receptor 3; Tlr4, Toll-like receptor 4; Vegfa, vascular endothelial growth factor a.

Pregnancy is a physiological condition of high energy demands. Endocrine and metabolic adaptations develop throughout the different phases of pregnancy to ensure adequate nutrition of mother and offspring. An increase in white adipose tissue mass, changes in adipokine production and secretion, development of mild insulin resistance, and virtually no change in lean mass are expected in women of normal body mass index and uncomplicated pregnancies (37, 38). In obese women, however, pregnancy starts in a very different metabolic state; i.e., increased adipose tissue mass associated with low-grade inflammation, hyperleptinemia, and increased insulin resistance (1, 18, 39–41). The deleterious consequences of obesity in pregnancy are well described by different groups using epidemiological approaches in large population studies (3, 4, 7, 11, 42–44). Whereas the outcome is well known, the causes and associated mechanisms are poorly defined.

Rodents are relevant preclinical models for translational research in women’s health because of the physiological similarities during pregnancy. Placentation in both humans and rodents is defined as hemochorial, which means that maternal blood comes in direct contact with the fetal chorion (45, 46). Other examples of similarities are the analogous function of the labyrinth layer of mice and the chorionic villi of human placentas, as well as the effects of hypoxia on placental development in both species (45, 47, 48). Moreover, a substantial advancement in the understanding of the genetic control of placental development has been achieved using targeted mutations in mice (45). Although differences should be noted, short generation times, large litter size, and recent development of mouse genetics and related molecular tools are all advantages of use of the mouse as a preclinical model (49).

The LepR^loxTB mouse is a model of monogenic mutation causing obesity and infertility (26, 31, 35). In previous and present studies, we showed that re-expression of endogenous Lepr in the PMV of LepR-null mice improved the fertility but had no effect on metabolism; mice were still morbidly obese (33). Females were able to conceive but showed high rates of resorptions. The lack of Lep signaling in trophoblast or placental tissue is unlikely to cause this phenotype, as the placentas express one copy of functional Lepr gene and are responsive to Lep. Whether the placental dysfunction and resorptions were caused by a gene dosage effect was further tested in the Lepr^db/+ mice. As there was a similar frequency of resorptions between wild-type and Lepr^db/+ embryos and also, given that Mendelian ratios of offspring were seen in all litters when crossbreeding the heterozygous LepR^loxTB/+ mice, the lack of one copy of Lepr gene in the placentas is not the cause of pregnancy loss. Our findings are also in agreement with previous studies showing that Lep signaling is required for fertility but is not necessary for pregnancy progression in mice (50, 51). However, Lep does exert a regulatory role in placental perfusion and nutrient transport in both mice and humans (52–55). In this regard, it is important to mention that obese LepR^loxTB fertile mice are still diabetic and show decreased expression of placental glucose and lipid transporters. Further studies are necessary to assess the contribution of these factors to the pregnancy outcome. Moreover, as the obese LepR^loxTB mice are hyperleptinemic (26), excess Lep signaling may be associated with the placental dysfunction. In fact, a growing body of literature has demonstrated that Lep levels are increased in preeclamptic pregnancies (56–59). Dysregulation of genes associated with preeclampsia in humans (e.g., Vegfa, Hif1a, Bhlhe40, Pecam1, Flt1, Il6, Il1b) (60–65) was also observed in placentas from obese LepR^loxTB mice. However, Lep signaling-deficient mice, albeit obese, are normotensive (66). Whether expression of one copy of Lepr in placentas is sufficient to trigger preeclamptic responses requires further investigation.

In contrast to our genetic model of obesity, mice on an HFD showed disruption of a different set of placental genes compared with obesity alone. This is particularly relevant for a translational perspective, as most of the studies using animal models of obesity rely on obesogenic diets (16, 67). As female rodents are usually resistant to an HFD as a result of the protective effects of estradiol (68), a diet high in fat and sometimes in sugar content is commonly used as an experimental approach to induce obesity (67, 69). However, the toxicity of this artificial manipulation is not always clear. Our findings argue in favor of dissociated effects between high adiposity and the consumption of a diet high in fat content. In obese LepR^loxTB female mice, high resorption rates were associated with defective angiogenesis and inflammation observed in histological analyses and changes in placental transcriptome (e.g., Vegfa, Flt1, Tlr3, Tlr4). On the other hand, mice fed an HFD (mildly overweight compared with the LepR^loxTB line) had small resorption rates at midgestation but showed changes in genes associated with cellular growth and response to stress (e.g., Igf1, Igfbp, Fgf2, Edn1).

The use of genetic rodent models of obesity on RD has become an important experimental strategy to dissociate the effects of adiposity/low-grade inflammation from an HFD. For example, loss-of-function mutation of the melanocortin 4 receptor (Mc4r/MC4R) gene causes obesity in rodents and humans (70–72). The MC4R^+/− rats were overweight (∼7.5% above control) but had a very small number of resorptions (73). The LepR^loxTB mice, however, were morbidly obese (twice the weight of control mice attributed to an increase in fat mass) (26) and showed ∼40% of embryo resorptions per litter. In addition, approximately one-half of the obese females had no successful pregnancies during the fertility testing. These mice were significantly more obese and hyperglycemic than the fertile LepR^loxTB mice, suggesting that the differences in metabolic conditions may have an important impact on the infertility phenotype. Together, our findings and those from other laboratories (73) suggest that pregnancy loss is correlated with the degree of adiposity in rodents. The LepR^loxTB mouse may open opportunities for understanding the idiopathic infertility observed in morbidly obese women.

One of the strengths of our study is the effects of obesity on placental morphology. Scant data on the histopathological changes of unsuccessful pregnancies are currently available in the scientific literature. Although some degree of necrosis is expected at term, the embryonic resorptions in our study had massive necrosis of the labyrinth at midgestation, likely causing uteroplacental insufficiency and acute death of the embryos. The expression of genes associated with vascular growth and remodeling was also dysregulated, reinforcing the histological findings. Of those, the homologs of Vegfa, Flt1, Hif1a, and Bhlhe40 genes have been associated with preeclampsia and recurrent pregnancy loss in humans (61, 62, 74–77). Similar data have also been described for Nfkbia, Tlr3, and Tlr4 genes associated with inflammation (78–81). Notably, when resorbed and nonresorbed placentas from obese mice were evaluated, opposite changes (i.e., decreased and increased expression, respectively) were observed. Further studies will be necessary to inform the significance of these findings, but these apparently opposing effects are suggestive of a compensatory or adaptive process in the face of lipotoxicity (43). This is in line with findings showing that female physiology and placental tissue undergo various adaptations and compensatory responses before a poor pregnancy outcome ensues (17, 43).

The effects of obesity on poor pregnancy outcome have been described in a series of epidemiological studies, with an important impact for populations worldwide. However, the causative players, as well as the development of intervention strategies, have been difficult to determine as a result of lack of experimental models and tools. Our findings indicate that the degree of obesity disrupts the ability of the placenta to overcome the metabolic insults—ultimately leading to defective angiogenesis, tissue necrosis, and fetal death. This mouse model highlights the striking similarities of obesity-induced placental dysfunction between rodents and humans. It also emphasizes the necessity of a clear dissociation between the effects of high adiposity and excess calorie or fat consumption for pregnancy outcome.

Glossary

Abbreviations:

18s

18S ribosomal RNA

AAV

adeno-associated virus

ANOVA

analysis of variance

bp

base pair

DEG

differentially expressed gene

E

embryonic day

F

forward

GFP

green fluorescent protein

HFD

high-fat diet

Lep

leptin

LepR

leptin receptor

Ly6G

lymphocyte antigen 6 complex locus G6D

MC4R

melanocortin 4 receptor

mRNA

messenger RNA

Pecam1

platelet and endothelial cell adhesion molecule 1

PMV

ventral premammillary nucleus

qPCR

quantitative polymerase chain reaction

R

reverse

RD

regular chow diet

RRID

Research Resource Identifier