Excess Maternal Fructose Consumption Increases Fetal Loss and Impairs Endometrial Decidualization in Mice

Abstract

The most significant increase in metabolic syndrome over the previous decade occurred in women of reproductive age, which is alarming given that metabolic syndrome is associated with reproductive problems including subfertility and early pregnancy loss. Individuals with metabolic syndrome often consume excess fructose, and several studies have concluded that excess fructose intake contributes to metabolic syndrome development. Here, we examined the effects of increased fructose consumption on pregnancy outcomes in mice. Female mice fed a high-fructose diet (HFrD) for 6 weeks developed glucose intolerance and mild fatty liver but did not develop other prominent features of metabolic syndrome such as weight gain, hyperglycemia, and hyperinsulinemia. Upon mating, HFrD-exposed mice had lower pregnancy rates and smaller litters at midgestation than chow-fed controls. To explain this phenomenon, we performed artificial decidualization experiments and found that HFrD consumption impaired decidualization. This appeared to be due to decreased circulating progesterone as exogenous progesterone administration rescued decidualization. Furthermore, HFrD intake was associated with decreased bone morphogenetic protein 2 expression and signaling, both of which were restored by exogenous progesterone. Finally, expression of forkhead box O1 and superoxide dismutase 2 [Mn] proteins were decreased in the uteri of HFrD-fed mice, suggesting that HFrD consumption promotes a prooxidative environment in the endometrium. In summary, these data suggest that excess fructose consumption impairs murine fertility by decreasing steroid hormone synthesis and promoting an adverse uterine environment.

According to the Centers for Disease Control, roughly one-third of United States adults suffer from metabolic syndrome, which is defined as having 3 of the next: abdominal obesity, high triglycerides, low high-density lipoprotein cholesterol, high blood pressure, and high fasting blood glucose (1). Several studies have concluded that excess intake of fructose contributes to development of metabolic syndrome (2). Consumption of fructose-sweetened beverages and foods has increased significantly in recent decades (3), and individuals with metabolic syndrome often consume excess fructose. Increased fructose intake may contribute to metabolic disease in 2 ways. First, unlike glucose, fructose bypasses phosphofructokinase and instead forms fructose-1-phosphate, which enters glycolysis and can be used directly for the synthesis of triglycerides. As a result, excess lipids are generated. Second, metabolism of fructose to fructose-1-phosphate reduces intracellular ATP levels. As ATP levels fall, AMP deaminase is activated, catalyzing the degradation of AMP to inosine monophosphate and, eventually, uric acid (4). Under normal conditions, uric acid can act as a potent antioxidant, but excess uric acid can lead to oxidative stress and cellular dysfunction (4–6).

Metabolic syndrome is especially problematic in reproductive-age women because it is associated with reproductive problems including subfertility, increased rates of early pregnancy loss (7, 8), and preeclampsia (9). Furthermore, obesity and associated metabolic syndrome decrease the odds that a woman undergoing fertility treatment will become pregnant and increase her risk of miscarriage (10, 11). Miscarriage can occur if the uterus fails to become receptive to embryo implantation. Implantation requires the stromal fibroblasts of the uterus to differentiate into specialized secreting cells in a process termed endometrial decidualization (12). Aberrant decidualization has been linked to numerous pregnancy complications including infertility, recurrent miscarriage, preeclampsia, intrauterine growth restriction, and preterm delivery (13–17).

Endometrial decidualization is regulated by the hormones estradiol (E₂) and progesterone (P4), which prepare the uterus for implantation of the blastocyst. In humans, decidualization occurs to some extent with every menstrual cycle when ovulation occurs and the corpus luteum produces P4. In mice, decidualization is initiated at the time of blastocyst attachment to the uterine lining, which occurs on day 4.5 of pregnancy. During decidualization, the endometrial stromal cells neighboring the implanting blastocyst proliferate and differentiate into epithelioid-like cells (12). The decidual cells support pregnancy by producing hormones and cytokines that are required for embryo development, controlling trophoblast invasion (12), producing antioxidants, and regulating immune cell function (18).

Here, we examined the effect of a high-fructose diet (HFrD) on reproduction in mice. We found that mice fed a diet in which the complex carbohydrates were replaced with fructose had lower pregnancy rates, smaller litters, poorer decidualization, and greater fetal loss than their control chow (Con)-fed counterparts. Additionally, our analysis indicates that high-fructose consumption promotes a prooxidative environment in the uterus. Finally, we report that mice on a HFrD had decreased P4 production and that exogenous P4 administration rescued decidualization.

Materials and Methods

Animal care and use

All procedures in this study were approved by the Animal Studies Committee at Washington University School of Medicine and conformed to National Institutes of Health guidelines. Six-week-old female C57Bl/6 mice were fed either a HFrD (66.8% fructose, 20.2% protein, 12.9% fat; Harlan Teklad) or standard chow (Con) (28.5% protein, 58% complex carbohydrate, 13.5% fat; LabDiet) for 6 weeks. Body weights were measured weekly. After 5 weeks of dietary exposure, an oral glucose tolerance test was administered after a 6-hour fast (19). Con and HFrD-fed mice were then either mated with chow-fed male mice to determine pregnancy rate and litter size or mated with vasectomized chow-fed ICR male mice (Harlan) to be used for embryo transfers or induced deciduoma experiments. Successful mating was determined by visualization of a copulatory plug. Litter size was determined at killing on day postcoital (dpc)5.5, dpc14.5, and dpc18.5. To visualize embryo implantation on dpc5.5, dams were injected iv with Chicago Blue dye before killing. To determine pregnancy success for mice killed on dpc18.5, body weights were measured throughout pregnancy. A lack of weight gain or weight loss indicated absence or loss of pregnancy, respectively, and mice were mated a second time. For mice killed on dpc5.5 and dpc14.5, absence of embryos indicated absence or loss of pregnancy.

P4 administration

HFrD-fed mice were sc injected with 1-mg P4 in 100 μL of sunflower seed oil (20) (controls received vehicle alone) beginning on the day on which a plug was first observed. Injections were repeated daily until the day of killing for the induced deciduoma experiments (dpc7) or until the time of placental development (dpc10.5) for the litter size experiments.

Serum analysis

Triglycerides, cholesterol, nonesterified free fatty acids, insulin, and uric acid were measured in serum from fasted (6 h) pseudopregnant (dpc7.0) mice (described below) by the Infinity Triglyceride and Cholesterol Reagent kits (Fisher Diagnostics), the nonesterified fatty acids-HR Assay kit (Wako Diagnostics), the rat/mouse insulin ELISA kit (Crystal Chem), and the Amplex Red Uric Acid Assay kit (Invitrogen), respectively, according to the manufacturers' instructions. The Ligand Assay and Analysis Core Laboratory at University of Virginia performed ELISAs to determine serum levels of E₂ and P4. A glucometer was used to measure glucose in whole blood.

In vitro fertilization, embryo culture, and transfers

To induce superovulation, female ICR mice (which produce higher quality embryos than do C57BL/6 mice) were injected with pregnant mare's serum gonadotropin and human chorionic gonadotropin (Sigma) (20). Preparation of cauda epididymal sperm from male ICR mice, insemination, and embryo culture were performed as described previously (20). Fifteen blastocyst-stage embryos were nonsurgically transferred into each recipient pseudopregnant Con- or HFrD-fed female mouse by using the NSET devise (ParaTechs) according to the manufacturer's instructions. The mice were killed on embryonic day 14.5, and the numbers of total implantation sites (resorption sites plus normal-for-gestational-age embryos) were recorded. Implantation rate was calculated as total number of implantation sites divided by number of embryos transferred, resorption rate was calculated as the number of resorbed embryos divided by the number of implantation sites, and litter sizes were calculated as the number of viable normal-for-gestational-age embryos.

Artificially induced deciduomas

After 6 weeks of dietary exposure, estrus cycles were checked daily via vaginal lavage with sterile saline solution and cytology analysis. On the day of proestrus/estrus, female mice were mated to vasectomized ICR male mice. Successful mating was determined by the presence of a vaginal plug; the morning it appeared was considered pseudopregnancy day 1. Decidualization of the uterine stroma was stimulated on pseudopregnancy day 4 by placing a silk thread in one of the 2 uterine horns. The untouched uterine horn served as an internal control. Seventy-two hours later (pseudopregnancy d 7), the body composition of the mice was analyzed by quantitative magnetic resonance (EchoMRI-900 Whole Body Composition Analyzer; Echo Medical Systems). Mice were conscious during the procedure. Mice were fasted for 6 hours and then killed (pseudopregnancy d 7.5). Tissues were collected, and the degree of decidualization was determined by the relative weight of the stimulated and unstimulated uterine horns.

Histology, immunohistochemistry, and imaging

Uteri and livers were fixed with 10% neutral buffered formalin solution (Sigma), dehydrated in ethanol, and embedded in paraffin. For morphologic analysis, 5-μm tissue sections were deparaffinized, rehydrated, and stained with hematoxylin and eosin. Tissues were visualized by light microscopy (Nikon Eclipse E800). For assessment of cell proliferation, deparaffinized and rehydrated sections were probed with mouse antiphospho-histone H3 (Ser10) antibody (see Table 1 for details) overnight at 4°C, washed, and probed with Alexa Fluor 488-conjugated secondary antibody (1:500; Molecular Probes) and 49,6-diamidino-2-phenylindole (Sigma). Controls excluding the primary antibody were performed for all staining procedures. Tile scan images were obtained on a Leica TCS SPE Confocal Microscope with a dry ×10 APO Objective, 0.3 NA, at a ×1.5 zoom, and 1 airy pinhole. Sequential scans were done to avoid any overlap between fluorescence; sequence 1: 405-nm excitation laser, 402- to 479-nm emission window, 913 gain, and −1 offset; sequence 2: 488-nm excitation laser, 499- to 551-nm emission window, 802 gain, and −7 offset. The numbers of pH3-positive cells per section were counted in images from 6 mice per group.

Table 1.

Antibody Table

Peptide/Protein Target	Antigen Sequence (If Known)	Name of Antibody	Manufacturer, Catalog Number, and/or Name of Individual Providing the Antibody	Species Raised in; Monoclonal or Polyclonal	Dilution
Phospho-histone H3 (Ser10)	Unknown	pH3	Cell Signaling, 9706S	Mouse; monoclonal	1:1000
Bone morphogenic protein 2	Unknown	Bmp2	Abcam, ab6285	Mouse; monoclonal	1:1000
Cyclophillin B	Residues 150 to C terminus	CycloB	Abcam, ab16045	Rabbit; polyclonal	1:1000
Superoxide dismutase 1 [Cu-Zn]	Amino acids 120–146 at C terminus	Sod1	Santa Cruz Biotechnology, Inc, sc-271014	Mouse; monoclonal	1:1000
Superoxide dismutase 2 [Mn]	Amino acids 1–222 full-length of human	Sod2	Santa Cruz Biotechnology, Inc, sc-133254	Mouse; monoclonal	1:1000
Forkhead box protein O1	Unknown	FoxO1	Cell Signaling, 2880P	Rabbit; monoclonal	1:1000

Protein isolation and Western blotting

Tissue lysates were prepared in radioimmunoprecipitation assay buffer (25mM Tris·HCl, 150mM NaCl, 1.0% nonyl phenoxypolyethoxylethanol-40, 1.0% deoxycholic acid, 0.1% SDS, and 2mM EDTA) containing 1mM phenylmethylsulfonyl fluoride and protease cocktail. Proteins were resolved by Sodium Dodecyl Sulfate-PAGE, and immunoblotting followed standard procedures (21). Membranes were incubated with primary antibodies against bone morphogenic protein 2 (BMP2), cyclophillin B (cycloB), superoxide dismutase 1 [Cu-Zn] (SOD1), superoxide dismutase 2 [Mn] (SOD2) and forkhead box protein O1 (FoxO1) for 16 hours at 4°C (see Table 1 for details). Blots were washed and then probed with horseradish peroxidase conjugated antirabbit and antimouse IgG (Santa Cruz Biotechnology, Inc). The SuperSignal West Pico kit was used to detect signal, and ImageJ software was used to quantitate band intensities on film. To control for protein loading, band intensities for the proteins of interest were normalized to the band intensities of cycloB, which was probed for on each Western blot.

RNA isolation and quantitative RT-PCR

Total RNA was isolated by using TRI reagent (Molecular Research Center) and quantified via NanoDrop (Thermo Scientific). Total RNA (1 μg) was reverse transcribed with the QuantiTect Reverse Transcription kit (QIAGEN), and subsequent real-time PCR analysis was performed on an ABI Prism 7500 sequence detection system (Applied Biosystems). Primer Express Software (Applied Biosystems) was used to design gene-specific primers (see Table 2). The relative amounts of mRNA were quantified by using a standard curve and normalized to the level of 18s rRNA.

Table 2.

Primer sequences for real-time RT-PCR

Gene	Forward	Reverse
Bmp2	CTACATGCTAGACCTGTATGC	CCCACTCGTTTCTGGTAGTTC
Fst1	TGCCAGTGACAATGCCACAT	GCAGGCAGCTTCCTTCAT
Id1	GAACGTCCTGCTCTACGACATG	GGGCACCAGCTCCTTGAG
Prp	TCCTGGCCAATAATGCTGCCATTG	AGCAGCCATTCTCTCCTGTTTGAC
Wnt4	GTGCCAGTACCAGTTCCG	CACACCTGCCGAAGAGAT

Statistical analyses

For pregnancy rate, a χ² analysis was used to determine statistical significance of differences. All other data are expressed as means ± SEM. Real-time RT-PCR data are expressed as mean fold change from control ± SEM, and Western blotting data are expressed as means ± SEM. Comparisons between Con and HFrD or between unstimulated and stimulated were determined a priori, and a 2-tailed Student's t test was used to determine statistical significance, which was defined as P < .05. GraphPad Prism 6 software was used for all statistical analyses.

Results

Six weeks of high-fructose consumption induces mild metabolic dysfunction

We first examined the metabolic consequences of exposure (beginning at 6 wk of age) to a HFrD (see Materials and Methods for composition of the diets). Body weights (Figure 1A) and body compositions (Figure 1B) were not significantly different between control (Con)- and HFrD-fed mice after 6 weeks of dietary exposure. However, HFrD-fed mice showed impaired glucose tolerance (Figure 1C), suggesting that HFrD exposure induced some metabolic abnormalities. Additionally, the livers of HFrD-fed mice had more lipid droplets than those of Con-fed mice (Figure 1D), but we did not observe any significant differences in fasting serum levels of lipids (triglycerides, nonesterified free fatty acids, or cholesterol), uric acid, glucose, or insulin (Figure 1E). Together, these results suggest that 6 weeks of HFrD exposure induced some metabolic dysfunction but did not induce metabolic syndrome in female mice.

Figure 1.

Six weeks of high-fructose consumption induces fatty liver and glucose intolerance. Six-week-old C57BL/6 mice were fed either Con or 60% fructose diet (HFrD) for 6 weeks, and then their metabolic profile was characterized. A, Body weights of HFrD- and Con-fed mice during the feeding paradigm. B, Body composition presented as percentages of fat mass (FM) and fat-free mass (FFM). C, Glucose tolerance test performed over 120 minutes. D, H&E staining of livers from Con- and HFrD-fed mice. Yellow arrows indicate lipid droplets. E, Serum measures in fasted mice: triglyceride (TG), nonesterified fatty acids (NEFA), cholesterol, uric acid, glucose, and insulin. All data are expressed as mean ± SEM. *, P < .05 by Student's t test.

HFrD consumption leads to pregnancy loss and reduced litter size

We next examined the effect of 6 weeks of HFrD exposure on pregnancy by mating the HFrD- and Con-fed females to Con-fed males. We followed the first pregnancies of all mice in which we detected a copulatory plug until dpc5.5, dpc14.5, or dpc18.5. Although 83% of Con-fed mice were pregnant at one of these time points, 53% of the HFrD-fed mice were pregnant at these time points, indicating a significant decrease in pregnancy rate in the HFrD-fed mice. Litter sizes did not significantly differ between HFrD- and Con-fed mice shortly after implantation (dpc5.5), but litters were significantly smaller at midpregnancy (dpc14.5) and late pregnancy (dpc18.5) in HFrD-fed mice than in Con-fed mice (Figure 2B), suggesting that HFrD consumption promoted fetal loss.

Figure 2.

Pregnancy rate and litter size are decreased in HFrD-fed mice. A, Pregnancy rates determined as the percent of pregnancies per group (Con or HFrD) after detection of a vaginal plug. *, P < .05 by χ² test. B, Litter size determined as the number of gestationally appropriate viable embryos during early pregnancy (dpc5.5), midpregnancy (dpc14.5), and term pregnancy (dpc18.5). Data are expressed as mean ± SEM. *, P < .05 by Student's t test.

To determine whether the HFrD uterine environment was responsible for the observed decrease in litter size, in vitro-fertilized embryos were generated with oocytes and sperm from Con-fed mice, transferred into the uteri of Con- and HFrD-fed mice, and allowed to develop until embryonic day 14.5. Although implantation rates did not differ between HFrD- and Con-fed recipient mice, we observed a nonsignificant 25% increase in embryo resorption in the HFrD-fed mice (Figure 3A). Similarly, the number of viable embryos at embryonic day 14.5 was smaller in HFrD-fed than in Con-fed recipient mice (Figure 3B). Although this difference was not statistically significant, this result suggests that the uterus is, at least in part, responsible for the decreased litter size observed in HFrD-fed mice. We also examined the placentas and found that those in HFrD-fed mice weighed significantly less than placentas in Con-fed mice (Figure 3C), suggesting that the uterine environment in HFrD-fed mice impaired placentation. However, crown-rump lengths of embryos were identical at this time point, indicating that embryonic development was not impaired by the uterine environment in HFrD-fed mice (Figure 3D).

Figure 3.

Control embryos transferred into HFrD-fed mice have decreased survival and smaller placentas. Blastocyst-stage embryos were generated by in vitro fertilization (IVF) of oocytes from chow-fed control mice and transferred into pseudopregnant Con- and HFrD-fed mice on dpc2.5. Embryos were analyzed on embryonic day 14.5. A, Implantation rate and resorption rates expressed as a percentage of total embryos transferred into Con- and HFrD-fed mice. B, Litter size expressed as the total number of gestationally appropriate viable embryos. C, Placental weight averaged per litter. D, Crown-rump (C/R) length averaged per litter. All data are expressed as mean ± SEM. *, P < .05 by Student's t test.

Decidualization is impaired in HFrD-fed mice

Decidualization of endometrial stromal cells is essential for placentation and maintenance of pregnancy. Therefore, we hypothesized that the fetal loss observed in HFrD-fed mice was due to impaired decidualization. To test this hypothesis, we used the induced deciduoma system, in which we artificially stimulated decidualization in pseudopregnant HFrD- and Con-fed mice by placing a silk thread into one of the 2 uterine horns. The other uterine horn served as an unstimulated control (Figure 4A), and the degree of decidualization was assessed by comparing uterine weights in stimulated and unstimulated uterine horns 72 hours later. Although decidualization occurred in both the Con- and HFrD-fed mice, stimulated uteri from HFrD-fed mice weighed 48% less than those from Con-fed mice (Figure 4B), suggesting that HFrD intake impaired endometrial stromal cell decidualization. Morphologic analysis of deciduomas revealed that Con-fed mice had enlarged and rounded endometrial stromal cells that were often binucleated, consistent with the proliferation that occurs during decidualization. In contrast, endometrial stromal cells from HFrD-fed mice remained small and mono-nucleated (Figure 4C). Additionally, staining with an antibody that recognizes phosphorylated histone H3, a marker of mitotic activity (22), confirmed that proliferation was significantly lower in the deciduomas from HFrD-fed mice than in those from Con-fed mice (Figure 4D). Murine endometrial stromal cell decidualization is also marked by induction of a number of P4 receptor (PR)-regulated genes such as prolactin-related protein (Prp). We found that expression of Prp mRNA was significantly higher in the stimulated uteri than in the unstimulated uteri from both Con- and HFrD-fed mice (Figure 4E). However, induction of Prp expression was significantly lower in the deciduomas from HFrD-fed mice than in those from Con-fed mice, confirming that decidualization was impaired in HFrD-fed mice.

Figure 4.

Decidualization is impaired in HFrD-fed mice. Uterine horns from pseudopregnant mice were either artificially stimulated (S) to decidualize or remained unstimulated (US) as an internal control and collected on pseudopregnancy day 7. A, Representative images of US and S deciduomas from Con- and HFrD-fed mice. Ruler size markings are millimeters. B, Mean uterine weights. C, Representative images of H&E-stained uteri from Con- and HFrD-fed mice. Yellow arrows indicate multinucleated decidual cells. D, Representative images and quantitation of deciduomas immunofluorescently labeled with antiphospho-histone H3 (pH3) antibody (green) and stained with 49,6-diamidino-2-phenylindole (DAPI) (blue). E, mRNA expression of the murine decidualization marker PrP. A priori comparisons were made between dietary treatments (Con vs HFrD) or the induction of artificial decidualization (US vs S). Student's t tests were used to determine statistical differences between groups; *, P < .05 for Con vs HFrD; $, P < .05 for US vs S.

Bmp2 signaling is impaired in decidua of HFrD-fed mice

The phenotype in HFrD-fed mice was similar to that found in mice in which Bmp2 was knocked out in the uterus (24, 25). BMP2 drives endometrial stromal cell proliferation and differentiation in response to PR signaling. Thus, we explored the BMP2 pathway in the uterus in HFrD-fed mice. First, we confirmed that levels of Bmp2 mRNA were equivalent (P = .5) between unstimulated uteri in Con- and HFrD-fed mice. We found that although Bmp2 mRNA expression was higher in the stimulated than in the unstimulated uteri from both Con- and HFrD-fed mice, Bmp2 induction was significantly lower in the deciduomas from HFrD-fed mice than in those from than Con-fed mice (Figure 5A). Consistent with gene expression changes, BMP2 protein levels were significantly higher in stimulated than in unstimulated uteri from Con-fed mice but not in those from HFrD-fed mice (Figure 5B). Next, we measured the mRNA levels of transcriptional targets of Bmp2: Id1, Fst1, and Wnt4. We found that mRNA expression of all 3 genes was equivalent in unstimulated uteri from Con- and HFrD-fed mice (P > .4) and was significantly higher in the stimulated uteri than in the unstimulated uteri from both Con- and HFrD-fed mice. However, mRNA expression of these genes was lower in the stimulated uteri of HFrD-fed mice than in those from Con-fed mice (Figure 5C). Taken together, these data suggest that impaired BMP2 signaling contributes to poor decidualization in HFrD-fed mice.

Figure 5.

The BMP2/Wnt-4 pathway is impaired in decidua of HFrD-fed mice. A, mRNA expression of Bmp2 in unstimulated (US) or artificially stimulated (S) uteri in Con- and HFrD-fed mice. Data are expressed as the fold change from the US levels. B, Western blot analysis of BMP2 and cycloB in US and S uteri. Densitometry for BMP2 is normalized to cycloB levels. C, mRNA expression of Id1, Fst1, and Wnt4. Data are expressed as the fold change from US in deciduomas from either Con- or HFrD-fed mice. Data are expressed as mean ± SEM. A priori comparisons were made between dietary treatments (Con vs HFrD) or the induction of artificial decidualization (US vs S). Student's t tests were then used to determine statistical differences between groups; *, P < .05 for Con vs HFrD; $, P < .05 for US vs S.

Foxo1 and Sod2 levels are decreased in uteri from HFrD-fed mice

In other models of increased fructose consumption, HFrD induces oxidative stress (26), so we wondered whether this occurs in the uteri. Thus, we examined uterine expression of the transcription factor FOXO1, which may play a role in regulating antioxidant defenses by transcriptionally activating superoxide dismutase expression (CuSod, Sod1 and MnSod, Sod2) during implantation and placentation to ensure successful pregnancy (23). Additionally, FOXO1 acts downstream of BMP2 signaling (24) and is essential for decidualization in human endometrial stromal cells (though it is not as important in murine endometrial stromal cell decidualization) (25). Western blot analysis showed that FOXO1 expression was significantly lower in both unstimulated and stimulated uterine horns from HFrD-fed mice than in those from Con-fed mice (Figure 6). Furthermore, although levels of SOD1 protein were similar between uteri from HFrD- and Con-fed mice, expression of SOD2 was significantly lower in unstimulated and stimulated uterine horns from HFrD-fed mice than in those from Con-fed mice (Figure 6). These data suggest that HFrD consumption promoted a prooxidative uterine environment, which may have contributed to the increased fetal loss in HFrD-fed mice.

Figure 6.

Foxo1 and Sod2 levels are decreased in decidua from HFrD-fed mice. Western blot analysis was used to determine protein levels of antioxidants in artificially stimulated (S) and unstimulated (US) uteri from Con- and HFrD-fed mice. A, Representative images of Western blottings for Foxo1, CuSod (Sod1), MnSod (Sod2), and cycloB in US and S uterine horns from Con- and HFrD-fed mice. B, Densitometric analysis of Western blottings normalized to CycloB. All data are expressed as mean ± SEM. *, P < .05 by Student's t tests; n = 6 per group.

HFrD consumption causes decreased circulating P4

The ovary-produced hormones P4 and estrogen (E₂) prime the uterus for embryo attachment and implantation and are necessary for decidualization. Therefore, we measured P4 and E₂ in pseudopregnant mice at the time of deciduoma collection at dpc7.5 and found that HFrD-fed mice had significantly lower P4 levels than Con-fed mice (Figure 7A). However, levels of E₂ did not differ between the 2 groups (Figure 7A). One possibility was that HFrD consumption impaired ovulation and thus led to formation of fewer corpus lutea, which produce P4. However, we found no difference in the number of corpus lutea per ovary in pseudopregnant HFrD- and Con-fed mice on dpc7.5 (4.2 ± 0.5 and 4.2 ± 0.8). This result suggests that impaired P4 synthesis, rather than impaired ovulation, contributed to the observed decrease in P4 levels in HFrD-fed mice.

Figure 7.

Exogenous P4 rescues decidualization in HFrD-fed mice. A, Circulating ovarian hormones P4 and E₂ were measured in pseudopregnant Con- and HFrD-fed mice on the day of deciduoma collection (dpc7). B, Representative images of unstimulated (US) and artificially stimulated (S) uterine horns from Con- and HFrD-fed mice and HFrD-fed mice treated with exogenous P4 (HFrD + P4). Ruler size markings are millimeters. C, Uterine weights measured on day 7 of pseudopregnancy. D, Litter size on dpc18.5 after daily exogenous P4 administration from dpc0.5 through dpc10.5. E–H, RT-qPCR analysis for Bmp2 (E) and BMP2 targets Id1 (F), Fst (G), and Wnt4 (H) in US and S uterine horns from Con, HFrD, and HFrD + P4 mice. All data are expressed as mean ± SEM. A priori comparisons were made between dietary treatment groups or between US and S uterine horns. Student's t tests were used to determine significance; different letters indicate significantly different (P < .05) values between groups.

Exogenous P4 rescues the decidualization defect in HFrD-fed mice

Finally, to determine whether the impaired decidualization observed in HFrD-fed mice resulted from decreased P4, we attempted to rescue the phenotype by providing pseudopregnant HFrD-fed mice with exogenous P4. We sc injected P4 into HFrD-fed mice daily starting on day 1 of pseudopregnancy and throughout the entire induced deciduoma protocol. As expected, HFrD-fed mice had significantly lower (P < .02) serum P4 levels (2.3 ± 0.9 ng/mL) than Con (7.8 ± 2.0 ng/mL) and HFrD + P4 (78.0 ± 22.9 ng/mL) mice. Elevating P4 levels qualitatively (Figure 7B) and quantitatively (Figure 7C) rescued the decidualization defect observed in HFrD-fed mice. Furthermore, when exogenous P4 was administered to pregnant HFrD-fed mice from dpc0.5 through dpc10.5 to mimic ovarian P4 production, litter size was restored to Con levels (Figure 7D), suggesting that by rescuing the decidualization defect, fetal survival was improved. Finally, exogenous P4 rescued the HFrD-induced decrease in expression of Bmp2 (Figure 7E) and the BMP2 target genes Id1 (Figure 7F), Fst (Figure 7G), and Wnt4 (Figure 7H) in artificially decidualized uteri. Taken together, these data indicate that reduced ovarian P4 production in HFrD-fed mice resulted in impaired decidualization and decreased BMP2 signaling.

Discussion

An increasing body of both experimental and clinical evidence suggests that maternal metabolic syndrome is associated with poor reproductive outcomes (26–28). Although increased fructose consumption has been linked to metabolic syndrome (4), its effects on fertility and early pregnancy have not been explored. Here, we addressed this question by feeding mice a HFrD before and during pregnancy. Several lines of evidence support our conclusion that consumption of excess fructose leads to pregnancy loss as a result of reduced P4 production and impaired endometrial stromal cell decidualization. First, HFrD-fed mice had lower pregnancy rates and smaller litters by midgestation than their Con-fed counterparts. Additionally, decidualization was impaired in HFrD-fed mice, and the uterine cells of these mice had decreased BMP2 signaling and reduced expression of antioxidant proteins. Together, these effects create a uterine environment that is not able to fully support embryo development. Finally, HFrD-fed mice had reduced levels of P4, and decidualization was rescued with exogenous P4.

An important feature of our HFrD model is that the mice did not develop most of the features of metabolic syndrome. Although they were slightly glucose intolerant and had increased lipids in their livers, they did not become overweight, have altered body composition, or have altered serum levels of lipids, uric acid, glucose, or insulin. Thus, this model allows us to study the effects of high-fructose consumption separately from the effects of metabolic dysfunction. This has not been possible in other studies. For example, rats fed a HFrD during pregnancy and lactation deliver pups with metabolic dysfunction (see Sloboda et al [29] for an excellent review of this literature) and are themselves hyperglycemic (30, 31), hyperinsulinemic (30, 32), and have elevated plasma triglycerides (30, 33) often associated with altered liver lipid metabolism (30, 31, 34). Because fructose-induced insulin resistance is likely mediated through fructose-induced hyperuricemia (5), the absence of hyperuricemia may explain the lack of insulin resistance in our mice. Thus, our model may be more representative of acute exposure to fructose than of metabolic syndrome. This may be especially relevant to the United States population; Americans consume 26.8 pounds of high-fructose corn syrup per capita per year (http://www.ers.usda.gov/data-products/sugar-and-sweeteners-yearbook-tables.aspx). This high level of fructose consumption may be a significant contributor to reproductive dysfunction in humans and should be explored further.

In addition to defects in decidualization, we found that placentas from control embryos developing in uteri from HFrD-fed mice were smaller than those from Con-fed mice. This finding is reminiscent of studies in rats showing that maternal fructose consumption during pregnancy impairs fetal and placental growth (30, 32, 35). Although the underlying mechanisms have not been fully established, Alzamendi et al showed that total blood vessel area in placentas from fructose-fed dams was significantly lower than that in placentas from Con-fed dams (35). This could be a consequence of impaired decidualization as placental vascular development and maternal vascular remodeling depend on invasion of trophoblast cells into the decidua at implantation. Inadequate preparation of the uterine lining in fructose-fed rodents may impede trophoblast invasion and thus impair placental development. This could explain the increased resorption rates and pregnancy loss we observed in HFrD-fed mice. This conclusion is further supported by our finding that P4 administration improved decidualization and rescued litter size in HFrD-fed mice.

We observed that early implantation was not affected by HFrD exposure, as litter sizes were identical in HFrD- and Con-fed mice at dpc5.5. This may reflect the fact that the HFrD-fed mice had normal levels of E₂, which contributes to 2 aspects of decidualization and implantation. Early in pregnancy, E₂ stimulates the luminal epithelial cells to proliferate, but the increase in P4 levels after corpus luteum formation causes proliferation to stop, allowing embryo attachment to occur (36). E₂ also regulates embryo attachment by stimulating the uterine glands to produce leukemia inhibitory factor (37).

We found that HFrD-fed mice had impaired decidualization and lower levels of P4 and that exogenous P4 could rescue the decidualization defect. This finding is consistent with previous literature documenting a key role of P4 in decidualization and implantation. In mice lacking the PR, the proproliferative (38) and proinflammatory (39) effects of E₂ were unchecked, thus impairing endometrial stromal cell decidualization and implantation. During implantation, uterine stromal cells become responsive to P4, and P4 signaling via the PR induces a transcriptional program that drives proliferation and differentiation into epithelioid-like decidual cells (reviewed in Ref. 40). Our finding that HFrD-fed mice had normal-sized litters at dpc5.5 but smaller litters at dpc14.5 indicated that pups were resorbed. This is similar to the observation that administration of PR antagonist on days 5–6 of rat pregnancy caused complete resorption of implantation sites (41). Thus, the resorptions we observed were likely a result of decreased P4 levels.

The transcriptional program induced during decidualization by PR signaling is dynamic and includes the transcription factors FOXO1, CCAAT/enhancer-binding protein b, signal transducers and activators of transcription 5, and homeobox A10, which drive differentiation of the stromal cells and can physically interact with and thereby modulate PR transcriptional capabilities (40). In humans and rodents, PR also regulates expression of Bmp2, which plays an indispensable role in regulating endometrial decidualization (24, 41–43). In the uterus of pregnant mice, Bmp2 expression is tightly spatiotemporally correlated with implantation (44), and mice lacking Bmp2 expression in the uterus have impaired decidualization and are completely infertile (42). Importantly, these mice have impaired stromal cell differentiation and decidual cell proliferation but have normal embryo attachment, preimplantation stromal cell proliferation, and endometrial vascularization (42). Consistent with these findings, we found that deciduomas of HFrD-fed mice had impaired Bmp2 signaling and decreased stromal cell differentiation and decidual proliferation, but the HFrD-fed mice had normal embryo attachment. Thus, impaired Bmp2 signaling explains many of the fructose-induced uterine phenotypes we observed.

Oxidative cell death at the maternal-fetal interface occurs in early pregnancy loss and in pregnancy disorders associated with altered implantation (preeclampsia and fetal growth restriction) (45), suggesting that decidualization is sensitive to oxidative stressors. Although decidualizing endometrial stromal cells acquire a heightened defense system against oxidative stress (23), when these defenses are defective, oxidative cell death occurs and decidualization is impaired (46, 47). The FOXO subfamily of forkhead box transcription factors activates genes involved in cell cycle inhibition, apoptosis, defense against oxidative stress, and DNA repair (reviewed in Ref. 18). In human endometrial stromal cells, FOXO1 is expressed during decidualization as a result of PR and Bmp2 activation (24). FOXO regulates stromal cell differentiation (48–51) and induces expression of the mitochondrial antioxidant MnSOD during decidualization, contributing to the heightened defense against oxidative stress during implantation (23). Our observation that FOXO1 expression decreased during decidualization is consistent with earlier reports that FOXO1 does not play a critical role in decidualization in mice (25). Although it is unclear why FOXO1 has such an important role in human but not mouse decidualization, we suspect that other FOXO family members are important for murine decidualization. However, whether FOXO1 contributes to regulating oxidative stress at the site of implantation has not been explored. Our finding that expression of both FOXO1 and MnSOD decreased in the uteri of HFrD-fed mice suggests that this pathway may be important during pregnancy and impaired by increased fructose consumption. Studies in the rodent liver and in endothelial cells suggest that fructose can drive oxidative stress through increased triglyceride synthesis and uric acid production (4). Although we did not measure oxidative stress in this study, our data suggest that impaired antioxidant defenses contribute to a prooxidative environment within the uterus that stresses the implanting embryo. This could then promote fetal loss and contribute to smaller litters in the HFrD-fed mice. Future studies will be focused on the prooxidative potential of fructose during pregnancy.

Our work in mice highlights the notion that we must understand the biological consequences of high-fructose consumption in the United States and around the world. Fructose consumption is elevated in persons with metabolic syndrome, and metabolic syndrome has been linked to an increased risk of early pregnancy loss and obstetric complications. Thus, elevated fructose intake may promote reproductive dysfunction and should be explored further in future clinical studies.