Maternal genetics and diet modulate vitamin A homeostasis of the offspring and affect the susceptibility to obesity in adulthood in mice

Abstract

Perinatal nutrition exerts a profound influence on adult metabolic health. This study aimed to investigate whether increased maternal vitamin A (VA) supply can lead to beneficial metabolic phenotypes in the offspring. The researchers utilized mice deficient in the intestine-specific homeobox (ISX) transcription factor, which exhibits increased intestinal VA retinoid production from dietary β-carotene (BC). ISX-deficient dams were fed a VA-sufficient or a BC-enriched diet during the last week of gestation and the whole lactation period. Total retinol levels in milk and weanling livers were 2- to 2.5-fold higher in the offspring of BC-fed dams (BC offspring), indicating increased VA supplies during late gestation and lactation. The corresponding VA-sufficient and BC offspring (males and females) were compared at weaning and adulthood after being fed either a standard or high-fat diet (HFD) with regular VA content for 13 weeks from weaning. HFD-induced increases in adiposity metrics, such as fat depot mass and adipocyte diameter, were more pronounced in males than females and were attenuated or suppressed in the BC offspring. Notably, the BC offspring were protected from HFD-induced increases in circulating triacylglycerol levels and hepatic steatosis. These protective effects were associated with reduced food efficiency, enhanced capacity for thermogenesis and mitochondrial oxidative metabolism in adipose tissues, and increased adipocyte hyperplasia rather than hypertrophy in the BC offspring. In conclusion, maternal VA nutrition influenced by genetics may confer metabolic benefits to the offspring, with mild increases in late gestation and lactation protecting against obesity and metabolic dysregulation in adulthood.

NEW & NOTEWORTHY A genetic mouse model, deficient in intestine-specific homeobox (ISX) transcription factor, is used to show that a mildly increased maternal vitamin A supply from β-carotene feeding during late gestation and lactation programs energy and lipid metabolism in tissues and protects the offspring from diet-induced hypertrophic obesity and hepatic steatosis. This knowledge may have implications for human populations where polymorphisms in ISX and ISX target genes involved in vitamin A homeostasis are prevalent.

INTRODUCTION

Nutrition in critical periods of prenatal development and early postnatal life influences metabolic health and susceptibility to obesity in adulthood through programming effects on central and peripheral (1, 2). Knowledge in this area has important public health implications, as it suggests prevention strategies for the rising burden of obesity and metabolic disease (1).

Vitamin A (VA) is a potent diet-derived modulator of metabolism. As retinoic acid, VA impacts lipid and energy metabolism and adipogenesis in mammals (3–5). Dietary VA supplementation (6, 7) or retinoic acid treatment (8–10) counteracts diet-induced (6, 10) and genetic (7) obesity in murine models, decreases body fat in lean mice (8), and reduces obesity in mice with established obesity (9). The antiadiposity action of retinoids (VA and its metabolites) is attributed to stimulating brown fat thermogenesis (11–13), white fat browning (14, 15), and substrate oxidation in nonadipose tissues like skeletal muscle (16, 17) and liver (18). These effects involve the interaction of retinoids with transcription factors and nutrient sensors such as retinoic acid receptor (RAR), peroxisome proliferator-activated receptor (PPAR) beta/delta, AMPK, or p38MAPK (3–5).

Despite these findings in adult animals, the role of VA as a metabolic programming factor in early life stages remains understudied or has yielded disparate results depending on the study design and animal model (19–25). For instance, supplementing suckling rat pups with VA increased adiposity gain (but not weight gain) when the animals were fed a high-fat diet after weaning (20). Conversely, maternal VA supplementation during gestation and lactation protected mouse offspring from diet-induced obesity in adulthood (22). These studies indicate significant actions of dietary retinoids in early life on programming adiposity. Still, the effects seem to depend on factors such as the developmental stage, route of administration (mother vs. pups), dose, and the mother’s diet and metabolic conditions.

In this study, we aimed to gain further insight into the activity of VA in early life in imprinting complex metabolic phenotypes. We took advantage of intestine-specific homeobox (ISX)-knockout mice, which exhibit increased intestinal absorption and conversion of dietary β-carotene (BC) to retinoids due to the lack of ISX, a transcriptional repressor of BC metabolism (26, 27). Offspring of ISX-deficient dams have previously shown increased maternal VA supplies, indicated by increased hepatic VA content and altered gastrointestinal immunity when dams were fed a BC diet (28). Notably, the ISX gene and its target genes encoding β-carotene-oxygenase-1 (BCO1; which cleaves BC to produce VA) and scavenger receptor class B type 1 (SRB1; involved in intestinal BC absorption) are affected by common genetic polymorphisms (29). Therefore, our study outcomes may have implications for how maternal genetics and diet affect metabolic health in the general population.

MATERIALS AND METHODS

Animal Experiment

The animal study protocols were reviewed and approved by the Institutional Animal Care and Use Committee (IACUC) of Case Western Reserve University (Protocol No. 2014-0106). Mice were maintained on a standard chow diet (Prolab RMH 3000; LabDiet, St. Louis, MO). Isx^−/− dams were mated to Isx^+/+ males to obtain a homogeneous Isx^+/− progeny. During the last week of gestation and the entire lactation period, the dams were subjected to feeding a VA-sufficient diet (VAS diet; 4,000 IU VA/kg diet) or a BC-enriched purified mouse diet devoid of preformed VA (BC diet; 25 mg BC/kg diet). Milk was collected following a previously published protocol (30) on day 21 of lactation. The offspring of dams fed the VAS and the BC diet (from here on, VAS and BC offspring) were compared. A cohort was euthanized at weaning [postnatal day (PND) 21] and a second cohort at adult age, following 13 wk of feeding a standard fat diet (NFD; 10% energy as fat) or a high-fat diet (HFD; 45% energy as fat) with regular VA content (4 IU × g⁻¹). All diets were produced by Research Diets (New Brunswick, NJ), and BC beadlets were a generous gift from Adrian Wyss (DSM Ltd., Sisseln, Switzerland). Body weight and food intake were recorded periodically. Energy intake was estimated per cage (at least 2 per group) from the food consumed and its caloric equivalence. Animals were euthanized by decapitation, under fed conditions, within the first 2 h of the light cycle. Liver, interscapular brown adipose tissue (BAT), and the white adipose tissue (WAT) depots inguinal, gonadal, and retroperitoneal (iWAT, gWAT, and rWAT, respectively) were dissected entirely, weighed, snap-frozen in liquid nitrogen, and stored at –80°C. Biopsies of iWAT and BAT were fixed for histological studies. Blood collected from the heart was used to prepare serum, stored at −20°C. The adiposity index was defined as the sum of the mass of all WAT depots dissected expressed as a percentage of body weight. The study was conducted with equal numbers of female and male mice. Energy efficiency (over a period) was defined as body weight gained per calories eaten.

Blood General Parameters

Blood glucose was measured at euthanization using a glucometer (EasyMax Self-Monitoring Blood Glucose System Meter Kit). Serum insulin, leptin, and triacylglycerol levels were quantified using the following commercial kits: Mouse Insulin ELISA kit (Mercodia AB, Uppsala, Sweden), Mouse/Rat Leptin Quantikine ELISA Kit (R&D Systems, Minneapolis, MN), and the QCA colorimetric kit (Química Clínica Aplicada S.A., Tarragona, Spain), respectively, following the manufacturer’s instructions.

Retinoid Analyses

HPLC analysis for nonpolar retinoids was performed as previously described (31, 32). Briefly, solid tissue samples (∼10 mg) were homogenized in 200 μL of phosphate-buffered saline (PBS), and serum (∼50 μL) and milk (∼30 μL) samples were diluted to 200 μL with PBS. Retinoids were extracted twice using a mixture of methanol, acetone, and hexane (1:2:2.5, vol:vol:vol). HPLC analysis was performed on a normal-phase Zorbax Sil (5 μm, 4.6 × 150 mm) column. Chromatographic separation was achieved by isocratic flow of 10% ethyl acetate/90% hexanes using an Agilent 1260 Infinity series HPLC system. To quantify the molar amounts of retinoids, the HPLC was previously scaled with synthesized an all-trans-retinol standard (Toronto Research Chemicals, Toronto, Canada).

Histology

Tissue samples were fixed by immersion in 4% paraformaldehyde in 0.1 M sodium phosphate buffer, pH 7.4, overnight at 4°C, dehydrated in a graded series of ethanol, cleared in xylene, and embedded in paraffin blocks for light microscopy. Five-micrometer-thick sections of tissues were cut with a microtome, mounted on slides, and stained with hematoxylin-eosin. Morphometric analysis of iWAT sections was performed by the digital acquisition of adipose tissue areas using AxioVision 40 V 4.6.3.0 software and a Zeiss Axioskop 2 microscope equipped with an AxioCam ICc3 digital camera (Carl Zeiss S.A., Barcelona, Spain). Distributions of adipocyte size were obtained from individual data on cell sizes.

Liver Total Protein and Triacylglycerol Content

Approximately 20 mg of liver were weighed and homogenized with PBS (1/20 dilution) in ice. For total protein determination, the homogenate was centrifuged (10,000 g, 1 min, at room temperature). The resulting supernatant was used for the protein quantification following the Pierce BCA Protein Assay Kit (Thermo Scientific, Rockford, IL) protocol. Liver triacylglycerol quantification was performed by the QCA colorimetric kit (Química Clínica Aplicada S.A., Tarragona, Spain).

Tissue DNA Content

DNA was isolated from iWAT samples using the E.Z.N.A Tissue DNA Kit Protocol (Omega Bio-tek, Norcross, GA) following the manufacturer’s protocol. Once isolated, the DNA concentration was determined using the NanoDrop ND-1000 spectrophotometer (Nano-Drop Technologies, Wilmington, DE). The total DNA content in the iWAT depot was calculated from the DNA content per milligram of tissue and the iWAT depot mass.

RNA Isolation, Retrotranscription, and Real-Time PCR

Total RNA was isolated from the liver, BAT, and rWAT using the E.Z.N.A. Total RNA Kit I (Omega Bio-Tek, Norcross, GA) and the phenol-based TriPure Reagent (Roche Diagnostics GmbH, Mannheim, Germany), respectively. Once isolated, RNA integrity was checked by 1% agarose gel electrophoresis stained with SYBR Safe (ThermoFisher Scientific, Waltham, MA), and the isolated RNA quantification and quality were assessed using the NanoDrop ND-1000 spectrophotometer (Nano-Drop Technologies, Wilmington, DE). Then, isolated RNA was reverse transcribed to complementary DNA (cDNA) using the highly sensitive cDNA synthesis kit (iScript cDNA; Bio-Rad Laboratories, Madrid, Spain) and an Applied Biosystems 2720 Thermal Cycler (Applied Biosystems, Madrid, Spain). Gene expression was measured by specific cDNA amplification and semiquantification following the StepOnePlus protocol (Applied Biosystems, Madrid, Spain) and using the Power SYBR Green PCR Master Mix (Applied Biosystems, Madrid, Spain) as previously described (33). All primers used were purchased from Sigma-Aldrich (Sigma-Aldrich S.A., Madrid, Spain), and the sequences are available upon request. Data were normalized against 18S ribosomal RNA gene expression.

Immunoblotting

Uncoupling protein 1 (UCP1) protein levels were determined in BAT by Western blotting. In detail, 40 (±15) mg of BAT were homogenized in phosphate-buffered saline (PBS), with EDTA and protease inhibitor, solution in a 1:10 weight/volume proportion. The homogenate was centrifuged at 7,500 g for 2 min at 4°C, and the supernatant was used for total protein quantification using the Pierce BCA protein assay kit (Thermo Scientific, Rockford, IL). A weight of 10 μg of total protein was solubilized and boiled for 5 min in 4× Laemmli sample buffer (Bio-Rad Laboratories, Madrid, Spain) containing 10% 2-beta-mercaptoethanol (Sigma-Aldrich, Madrid, Spain). Proteins were then separated by an Any kD precast polyacrylamide gel electrophoresis and electrotransferred onto a nitrocellulose membrane (both from Bio-Rad Laboratories, Madrid, Spain). The “Precision Plus Protein Dual Color Standard” (Bio-Rad Laboratories, Madrid, Spain) was used as a protein weight marker. Electroblotting was carried out with the Trans-Blot Turbo Transfer System (Bio-Rad). Membranes were then blocked and incubated overnight at room temperature with the primary antibody diluted at 1:1,000 in Tris-buffered saline Tween 20 (TBS-T). The next day, the membranes were incubated with the secondary antibody at a dilution of 1:20,000 in TBS-T for 1 h at room temperature. The primary antibodies used were anti-UCP1 (GTX10983, GeneTex, Irvine, CA) and anti-ACTB (β-actin; no. 3700; Cell Signaling Technology, Danvers, MA), and the secondary antibodies used for UCP1 and ACTB, respectively, were IRDye 800CW Goat Anti-Rabbit (no. 926-32211) and IRDye 680RD Goat Anti-Mouse (no. 926-68070), both purchased from LI‐COR (LI‐COR Biosciences, Lincoln, NE). After UCP1 determination, a stripping step was carried out using Stripping Buffer 5× (LI‐COR Biosciences, Lincoln, NE) before the ACTB antibody labeling. For infrared detection, membranes were scanned in the Odyssey Infrared Imaging System (LI‐COR Biosciences, Lincoln, NE), and the bands were quantified using the software Odyssey v3.0. The UCP1 signal was normalized to the ACTB signal.

Statistical Analysis

All data are expressed as the means ± SE. Differences among the groups based on the maternal diet (MD: BC vs. VAS) and postweaning diet (D: NFD vs. HFD) and interaction effects (MDxD) between these two factors on the dependent variable were assessed by two-way ANOVA. The analyses were conducted separately by sex unless otherwise indicated. Two-tailed Student’s t test was used for binary comparisons. The significance threshold for all statistics was set at P < 0.05. P values between 0.05 and 0.10 were considered nonsignificant tendencies. Analyses were performed with SPSS for Windows (SPSS version 27.0, Chicago, IL).

RESULTS

BC Supplementation to Isx^−/− Dams Increased Milk VA Levels and the Weanlings’ VA Status and Did Not Affect the Weanling Adipose Phenotype

Isx^−/− mice were used as a model to increase maternal VA supply to the newborn through BC feeding. For this purpose, Isx^−/− dams and Isx^+/+ males were raised on a standard chow diet. After mating, pregnant Isx^−/− dams were fed either a VA sufficient or a BC-enriched purified rodent diet from gestational day 14 until the end of lactation. Total ROL levels were 2.5-fold higher in the milk of the dams on the BC diet (Fig. 1A). The BC concentration was below the detection limit of our HPLC system. Consistent with the increased VA supply during the suckling phase, the offspring of BC-fed dams displayed approximately double the amount of retinyl esters (RE) and total retinol (ROL) in the liver at weaning (PND 21) as compared to the offspring of VAS-fed dams (Fig. 1B). Serum ROL levels in the weanlings were unaffected by the maternal diet, reflecting the well-known homeostatic control of serum VA over a wide range of liver reserves (Fig. 1C). RE and ROL content in (gonadal) WAT was also comparable in the two groups of weanlings (Fig. 1D). There were no significant differences in body weight, liver mass, fat depots mass, adiposity index, and iWAT adipocyte area between the offspring of BC- and VAS-fed dams at weaning (Fig. 1, E–G, and results not shown). The histological appearance of the liver and (inguinal and gonadal) WAT was also similar between the two groups of weanlings (not shown).

Figure 1.

Retinoids in milk and weanlings and biometric parameters in weanlings. Retinoid levels at lactation day 21 in milk of Isx^+/− dams fed vitamin A-sufficient (VAS) or β-carotene (BC) diet (3 dams per group) (A) and retinoid levels in liver (B), serum (C), and gonadal white adipose tissue (gWAT) (D) and body weight (E), liver mass (F), and gWAT mass (G) in their corresponding 21-day-old Isx^+/− offspring. Dams were fed the experimental diets during the last week of gestation and the whole lactation period; bw, body weight. Data are expressed as means ± SE of combined male and female weanlings. Circles correspond to individual data. **P < 0.01, *P < 0.05, BC vs. VAS (two-tailed Student’s t test).

The Offspring of BC-Supplemented Isx^−/− Dams Were Protected against HFD-Induced Obesity

To study the potential influence of maternal VA supply on adiposity gain in the offspring later in life, 21-day-old offspring of BC- and VAS-fed Isx^−/− dams were subjected to NFD or HFD feeding for 13 wk. The study was conducted in parallel with female and male offspring. The corresponding growth curves are shown in Fig. 2A, with body weights at the experimental end point shown in Fig. 2B. Male mice gained more weight on the HFD than female mice, and in both sexes, the BC progeny showed an attenuated body weight response to the HFD when compared with the VAS progeny. HFD feeding increased body weight in the VAS female offspring but not the BC female offspring, indicating a significant maternal diet × postweaning diet interaction (MDxD; Fig. 2, A and B). In males, HFD feeding led to increased body weight regardless of the maternal diet (Fig. 2B). However, the BC male progeny had lower body weight than the VAS male progeny at multiple time points on the HFD (Fig. 2A) and displayed a trend to lower body weight at the end point (P = 0.079; Fig. 2B). Similar to body weight, the HFD-induced increases in fat depot mass and adiposity index were higher in the male than the female mice, and they were suppressed or attenuated in the BC progeny as compared with the VAS progeny (Fig. 2C and Supplemental Fig. S1). Such a profile was identifiable for all WAT depots dissected and the interscapular BAT depot (Supplemental Fig. S1). In good concordance with the adiposity results, HFD feeding increased blood leptin levels significantly in the male mice but not in the female mice and to a lower extent in the BC male progeny (3-fold increase) than the VAS male progeny (6-fold increase; Fig. 2D). Cumulative food intake was comparable in the female VAS and BC progeny on either diet (NF or HF) and even higher in the male BC progeny than its VAS counterpart on the HFD (Fig. 2E). Energy efficiency increased with the HFD in the VAS female progeny but not the BC female progeny. In males, energy efficiency increased upon HFD to a greater extent in the VAS progeny than in the BC progeny. Consequently, energy efficiency was lower in BC males as compared to VAS males on the HFD (Fig. 2F).

Figure 2.

Biometric parameters in the adult offspring. Male and female Isx^+/- mice born from Isx^-/- dams fed fed vitamin A-sufficient (VAS) or β-carotene (BC) diet during the last week of gestation and the lactation period were subjected to a 13-week normal fat (NF)/high-fat (HF) diet challenge from weaning. Growth curves during the NF/HF challenge (A), body weight (B), adipose index (C), serum leptin (D), and cumulative food intake (E) at the end of the experiment and energy efficiency [body weight (bw) gained per calories eaten] over the last 25 days of the NF/HF challenge (F). Data are the means ± SE of 2–4 animals per group. Circles correspond to individual data. MD, effect of mother diet; D, effect of postweaning diet; MDxD, mother diet per diet interaction (two-way ANOVA). **P < 0.01, *P < 0.05, BC vs. VAS; ^##P < 0.01, ^#P < 0.05, HF vs. NF (two-tailed Student’s t test).

Under conditions of NFD feeding, no significant differences in biometric parameters were observed between the adult BC and VAS progeny when the sexes were analyzed separately, except for BAT mass, which was significantly higher in the BC male progeny (Fig. 2 and Supplemental Fig. S1). When the NFD-fed animals of both sexes were pooled, a higher BAT mass in the BC progeny persisted (0.085 ± 0.009 vs. 0.060 ± 0.006 g, P = 0.045, n = 7, Students t test), and a higher visceral fat mass in the BC progeny became statistically significant (pooled BC vs. VAS progeny data on the NFD: gWAT mass, 0.426 ± 0.054 vs. 0.241 ± 0.020 g, P = 0.008, n = 7, Students t test; rWAT mass, 0.131 ± 0.028 vs. 0.050 ± 0.009 g, P = 0.017, n = 7, Students t test).

Histological analysis of subcutaneous iWAT showed that the mean adipocyte area significantly enlarged upon HFD feeding in the VAS female progeny (by 124%) but not the BC female progeny. In males, adipocyte enlargement was significant in both groups, yet more pronounced in the VAS progeny (340% increase in mean adipocyte area) than in the BC progeny (105% increase) (representative images in Fig. 3, A and B). The total iWAT depot DNA content significantly increased upon HFD feeding in both the VAS and BC female progeny (by 168% and 273%, respectively), indicative of adipocyte hyperplasia. Meanwhile, in males, the total iWAT DNA content decreased (by 77%) in the VAS progeny and increased (by 117%) in the BC progeny upon HFD feeding. This observation indicated a significant maternal diet x postweaning diet interaction effect in males (Fig. 3C). The frequency distribution of adipocyte area showed that, after HFD feeding, the BC female progeny had a higher proportion of small adipocytes and a lower proportion of large adipocytes in iWAT when compared with VAS female progeny. The BC and VAS male progenies displayed a similar distribution. Under conditions of NFD feeding, the adult BC progeny of both sexes had a higher proportion of large adipocytes and a lower proportion of small adipocytes in iWAT compared with sex-matched VAS progeny (Fig. 3D). Thus pooled data of mice fed with NFD revealed a significant greater iWAT mean adipocyte area in the BC offspring (3,291 ± 419 vs. 1,684 ± 238 µm², P = 0.012, n = 5/6, Students t test). No histological signs of iWAT browning were observed in any group during microscopic examination.

Figure 3.

Histology and morphometry in white adipose tissue of adult offspring. Representative microphotographs of hematoxylin-eosin-stained tissue sections illustrating adipocyte size (A), mean adipocyte area (B), total depot DNA content (C), and distribution of adipocytes by size (D) in inguinal white adipose tissue (iWAT) at the end of the experiment. Male and female Isx^+/- mice born from Isx^-/- dams fed vitamin A-sufficient (VAS) or β-carotene (BC) diet during the last week of gestation and the lactation period were subjected to a 13-week normal fat (NF)/high-fat (HF) diet challenge from weaning. Data are the means ± SE of 2–4 animals per group. Circles correspond to individual data. MD, effect of mother diet; D, effect of postweaning diet; MDxD, mother diet per diet interaction (two-way ANOVA). **P < 0.01, *P < 0.05. BC vs. VAS; ^##P < 0.01, ^#P < 0.05, HF vs. NF (two-tailed t Student). In D, between 200 and 300 cells per animal were included in the analysis of the distribution of adipocytes size. The area of individual adipocytes was measured using a quantitative morphometric method at ×20 magnification with the assistance of Axio Vision software. Adipocyte size distribution was statistically different between groups according to the Kolmogorov-Smirnov test (shown at right). For each sex and dietary condition (NF or HF), the distribution of adipocytes size of the BC group (in orange) is shown overlapping the one of the VAS group.

The results of gene expression analyses conducted at the mRNA level in visceral (retroperitoneal) WAT are shown in Fig. 4. Pooled data in female and male mice are shown since similar results were obtained in both sexes. HFD feeding induced the expression in rWAT of the white adipocyte marker genes Lep (encoding leptin) and Fabp4 [encoding fatty acid binding protein 4, also known as adipocyte protein 2 (aP2)], as expected (Fig. 4A). The induction of these two genes upon HFD feeding was less prominent in the BC offspring, in which it did not reach statistical significance, consistent with their reduced development of diet-induced obesity compared with the VAS offspring. On the NFD, however, adipose expression levels of Fabp4 were higher in the BC offspring, consistent with the higher visceral fat depot mass and the increased proportion of large adipocytes in iWAT observed in the BC offspring on this diet. HFD feeding led to decreased adipose expression levels of the Pcna gene (encoding proliferating cell nuclear antigen) in the VAS progeny but not the BC progeny (Fig. 4B). This result is consistent with the observed increased hyperplasic component of HFD-induced WAT expansion in the BC progeny and the iWAT total DNA content results. Adipose mRNA levels of Lep and Fabp4 correlated positively with iWAT mean adipocyte area (0.767 and 0.632 Spearman’s ρ, respectively, P < 0.05, n = 22) and serum leptin levels (0.528 and 0.901 Spearman’s ρ, respectively, P < 0.05, n = 20/21) and negatively with iWAT total DNA content (−0.599 and −0.699 Spearman’s ρ, respectively, P < 0.05, n = 24/26), as expected for markers of adipocyte hypertrophy. Adipose mRNA levels of Pcna showed the opposite correlations. i.e., negative with adipocyte area and serum leptin and positive with adipose total DNA content (−0.67, −0.644, and 0.452 Spearman’s ρ, respectively, P < 0.05, n = 19/24), as is to be predicted for a marker of adipocyte hyperplasia.

Figure 4.

Gene expression in white and brown adipose tissues of adult offspring. A–D: mRNA levels of the indicated genes related to white adipocyte markers (A), cell proliferative status (B), brown adipocyte markers and antioxidant defenses (C), and Pparg (D) in retroperitoneal white adipose tissue (rWAT) at the end of the experiment. E and F: mRNA levels of genes related to thermogenesis (E) and total UCP1 in the interscapular brown adipose tissue (BAT) depot and representative UCP1 Western blot (F) at the end of the experiment. Isx^+/- mice born from Isx^-/- dams fed fed vitamin A-sufficient (VAS) or β-carotene (BC) diet during the last week of gestation and the lactation period were subjected to a 13-week normal fat (NF)/high-fat (HF) diet challenge from weaning. Data are the means ± SE of 4–7 combined male and female animals per group and are expressed relative to the mean value of the NF-VAS group, which was set to 100. Circles correspond to individual data. MD, effect of mother diet; D, effect of postweaning diet; MDxD, mother diet per diet interaction (two-way ANOVA). *P < 0.05, BC vs. VAS; ^##P < 0.01, ^#P < 0.05) HF vs. NF (two-tailed Student’s t test).

Other genes assayed for mRNA expression in rWAT included Ucp1 (encoding UCP1, the essential effector of brown fat thermogenesis), Cd137 [a marker gene of beige adipocytes (34)], and Cox2 (a mitochondrial DNA gene encoding cytochrome c oxidase subunit 2), which were used as molecular indicators of WAT browning and WAT mitochondrial oxidative capacity; Sod2, encoding the antioxidant enzyme manganese superoxide dismutase; Slc2a4, encoding the insulin-regulated glucose transporter GLUT4; and Pparg, encoding the master transcription factor for adipose tissue biology PPARγ. Following HFD feeding, the expression levels of these genes were downregulated or tended to be so in the VAS progeny but not in the BC progeny (Fig. 4, C and D). However, the BC progeny showed decreased expression compared to the VAS progeny under NFD feeding conditions. This indicated maternal diet × postweaning diet interactions that were of statistical significance for Ucp1, Cd137, Cox2, Sod2, and Slc2a4 (Fig. 4C) and nearly significant for Pparg (Fig. 4D, P = 0.068 for the interactive effect in two-way ANOVA).

In BAT, the mRNA expression levels of Ucp1 increased significantly with the HFD, especially in the BC offspring (Fig. 4E). The profile was similar for the UCP1 protein levels, which tended to increase with the HFD (P = 0.074 for the postweaning diet effect in two-way ANOVA), and more clearly in the BC progeny than the VAS progeny (3.5-fold increment, P = 0.089 vs. 1.4-fold increment, P = 0.534, n = 5/6, Student’s t test; Fig. 4F). Other thermogenesis-related genes assayed in BAT increased upon the HFD only in the BC progeny (case of Cidea) or were expressed to higher levels in the BC progeny on the HFD (case of Ppara and Fgf21; Fig. 4E). Thus the lower energy efficiency of the BC progeny on the HFD associated with an increased expression of thermogenesis- and substrate oxidative metabolism-related genes in both visceral WAT and BAT.

The Offspring of BC-Supplemented Isx^−/− Dams Were Protected against HFD-Induced Metabolic Complications

Under the conditions of our experiment, HFD feeding resulted in a significant increase in fed blood triacylglycerol levels, and apparent but nonsignificant increases in fed blood glucose and insulin levels, in the male mice. In contrast, increases in these parameters after the HFD were not apparent in the female mice (Supplemental Fig. S2). These results are in keeping with previous reports indicating the resistance of female mice to develop HFD-induced metabolic complications as compared to males (35–37). Interestingly, the HFD-induced increase in circulating triacylglycerols was largely blunted in the BC male progeny, which had decreased fed blood triacylglycerol levels compared to the VAS progeny under the two postweaning diets assayed.

Liver weight increased significantly following HFD feeding in the VAS male progeny but not the BC male progeny or the female animals (Fig. 5A). Liver triacylglycerol content measured by biochemical analysis (as mg per g of liver protein) increased approximately twofold with the HFD, regardless of sex and maternal diet. However, the BC male progeny had lower liver triacylglycerol content compared to the VAS male progeny under the NFD (Fig. 5B). Semiquantitation of liver steatosis from histological sections suggested a higher degree of steatosis in the HFD-fed mice when compared with the NFD-fed mice and a lower degree of steatosis in the BC progeny when compared with the VAS progeny, both on the HFD and the NFD (representative images in Fig. 5, C and D).

Figure 5.

Liver parameters in the adult offspring. Liver weight (A) and triacylglycerol content (B), representative micrographs (×10 magnification) of hematoxylin-eosin-stained liver sections (C), heat map of hepatic steatosis grade in liver biopsies of 3 mice (D) of each experimental group, and retinoid levels in liver (E) and serum (F) at the end of the experiment. Male and female Isx^+/- mice born from Isx^-/- dams fed fed vitamin A-sufficient (VAS) or β-carotene (BC) diet during the last week of gestation and the lactation period were subjected to a 13-week normal fat (NF)/high-fat (HF) diet challenge from weaning. Data are the means ± SE of 2–4 animals per group. When the sex is not specified, the animals are pooled (4–7 animals per group). Circles correspond to individual data. MD, effect of mother diet; D, effect of postweaning diet (two-way ANOVA). *P < 0.05, BC vs. VAS; ^##P < 0.01, ^#P < 0.05, HF vs. NF (two-tailed Student’s t test).

The hepatic mRNA levels of lipid metabolism-related genes were compared in BC and VAS adult offspring to assess possible links with the histological and biochemical results. The results suggested different responses in the two sexes (Supplemental Table S1). The lipogenesis-related genes Srebp1, Cd36, Fasn, and Scd1 were expressed at lower levels in the BC male progeny than in the VAS male progeny under the HFD. Further, following HFD feeding, hepatic mRNA levels of Fasn, Scd1, and Plin2 tended to increase in the VAS male progeny and to decrease in the BC male progeny, indicating maternal diet × postweaning diet interactions that were close to statistical significance (P < 0.1). Additionally, the BC male progeny displayed a more pronounced HFD-induced increase in Cox2 mRNA levels and higher expression of Pgc1a in the liver compared with the VAS male progeny. In female mice, the downregulatory effects of the maternal BC diet on lipogenesis-related genes were less evident. Still, the hepatic mRNA levels of Plin5 were induced with HFD feeding selectively in the BC females, and similar trends were observed for other oxidative metabolism-related genes, such as Ppara and Pgc1a. The Plin5 product is unique among the perilipins because it anchors the lipid droplet to mitochondria, augmenting mitochondrial respiratory function (38). Hepatic expression levels of Cox2 also tended to be higher in the BC females. Overall, these results align with the liver histological results and are consistent with the lower degree of steatosis observed in the BC progeny of both sexes.

The Offspring of BC-Supplemented Isx^−/− Dams Were Protected against HFD-Induced Depletion of Hepatic Vitamin A Stores

To study if the BC intervention designed to increase VA supply to the newborn could affect VA homeostasis in adulthood, liver total ROL and RE (Fig. 5E) and serum ROL levels (Fig. 5F) were analyzed in the adult mice after the NFD/HFD challenge. Results are presented for both sexes combined, as the profile was similar in male and female animals. Serum ROL levels were not affected by the dam’s diet or the diet after weaning in either sex, whereas ROL and RE levels in the liver were decreased by the HFD more consistently in the VAS offspring than the BC offspring. Previous studies reported that HFD feeding is associated with reduced VA concentrations in tissue, but not serum, in rodents (25, 39).

DISCUSSION

The fat-soluble VA affects many aspects of metabolic health and energy homeostasis. We here showed in ISX-deficient dams that this transcription factor controls the metabolic flow of VA retinoids from mother to child during late gestation and lactation and that the VA supply received at these stages can alter the responses to nutritional challenges later in life. Thus, our data demonstrate that maternal VA nutrition as influenced by the mother’s genetic makeup may have long-lasting effects on the metabolic health of the offspring. The main findings and their implications for the general population, in which genetic variability in ISX and its target genes is prevalent (29), are discussed in the following paragraphs.

Compared to the Isx^−/− dams fed the VAS diet, the Isx^−/− dams fed the BC diet showed increased VA supply to the newborns, as indicated by increased VA levels in milk and higher VA status in the weanlings. This finding is consistent with previous observations (28). BC was undetectable in the milk of the BC-fed dams, as expected, indicating that almost all absorbed BC was readily converted in enterocytes to VA as previously observed (32). There were no differences between VAS and BC offspring regarding adipose phenotype and other studied parameters at weaning, suggesting that maternal BC feeding had no major impact on gross milk composition, although subtle changes cannot be discarded. The key finding was that the progeny of BC-fed dams was protected against subsequent HFD-induced obesity compared to the VAS progeny. This was evident from reduced adiposity gain, reduced adipocyte enlargement, and blunted increases in blood triacylglycerol levels (in males) and liver steatosis in the BC offspring after 13 wk of an obesity-inducing diet extending into adulthood. Moreover, our results suggest that a mild excess VA supply through the mother during late gestation and lactation (∼2.5-fold excess according to measurements at weaning) reduces fat accretion and can potentiate adipocyte hyperplasia over hypertrophy under an HFD. The effect of promoting adipocyte hyperplasia over hypertrophy was more evident in female progeny but was also suggested in male progeny. Expanding WAT through an increase in adipocyte number (hyperplasia) is considered a healthier form of expansion that may prevent metabolic dysfunction compared to an increase in adipocyte size (hypertrophy) (40).

Collectively, the observed changes in both adipocyte size and total DNA content in inguinal WAT (iWAT) following HFD feeding suggest that female sex and a mild excess perinatal VA supply are two factors that can favor a healthier, hyperplastic fat expansion under obesogenic conditions. Increased adipocyte hyperplasia in females compared to males has previously been reported in both rodents (41) and humans (Ref. 42 and references therein). The prohyperplastic effect of mild VA supplementation during early development aligns with previous reports (19, 20, 22). Mechanistically, this effect has been related to an increased pool of progenitor PDGFRα⁺ cells (capable of white and beige adipogenesis) in adipose tissues through angiogenesis following maternal VA or retinoic acid supplementation (22). It has also been related to the hypomethylation of the Pcna promoter and a higher PCNA content in the WAT of VA-supplemented pups at weaning (20, 43).

Resistance to the development of diet-induced obesity in the BC progeny cannot be attributed to lowered food intake. In fact, food intake was higher in the male BC progeny than the VAS progeny on the HFD. Instead, it was associated with reduced energy efficiency (weight gained per calories eaten) on the HFD in the BC offspring, pointing to the involvement of a metabolic mechanism. In this context, the BC offspring exhibited increased BAT thermogenic capacity, as indicated by higher gene expression levels of Ucp1 and other thermogenesis-related genes, particularly on the HFD. Further, the BC progeny was protected against the HFD-induced downregulation of the expression in visceral WAT of genes related to browning and mitochondrial oxidative capacity (Ucp1, Cd137, and Cox2) observed in the VAS progeny. Downregulation of genes in these functional categories, including Ucp1 and Cd137, upon HFD feeding, as found in the VAS progeny, was previously reported in mice (44–46) and could be part of an adaptative program to efficiently store energy in WAT under conditions of dietary surplus (47). This program would be less effective in the BC progeny. Furthermore, the BC progeny was protected against detrimental molecular responses in WAT to the HFD observed in the VAS progeny, such as the downregulation of Sod2, Glut4, and Pparg expression. HFD causes depletion of antioxidant defenses in tissues, leading to decreased gene expression and activity of manganese superoxide dismutase, encoded by Sod2 (48). Downregulation of Glut4 expression in WAT following HFD has been described as related to HFD-induced insulin resistance and hyperglycemia (49, 50). A decreased PPARγ mRNA and activity in WAT following obesogenic diet feeding and obesity has been described in mice (51) and humans (52, 53) and linked to adiposopathy and conditions such as dyslipidemia, hypoadiponectinemia, insulin resistance/diabetes, and chronic inflammation, among others (51). The lack of downregulation of Sod2, Glut4, and Pparg expression in WAT is thus consistent with a healthier WAT expansion and metabolic phenotype under an HFD in the BC progeny.

Interestingly, despite their relative resistance to developing obesity on the HFD, the BC offspring exhibited a slightly increased (∼2-fold) mass of visceral adipose depots (gonadal and retroperitoneal) and subcutaneous adipocyte size under the NFD. The increased adiposity under NFD became evident when the data of animals of both sexes were pooled. Gene expression results aligned with the adipose morphometric results. Thus, paralleling the increased retroperitoneal fat depot mass, the BC progeny had higher expression levels of adipose marker genes (Fabp4) and lower of Pparg and genes related to thermogenesis/mitochondria oxidative metabolism (Ucp1, Cd137, Cox2), insulin responsiveness (Glut4), and antioxidant capacity (Sod2) in the rWAT depot as compared to the VAS progeny on the NFD. Whether the observed mildly increased adiposity under nonobesogenic diet feeding conditions constitutes a metabolic burden to the BC offspring is unclear. In fact, the BC male offspring displayed significantly lower fed serum triacylglycerol and liver triacylglycerol levels compared to sex-matched VAS offspring on the NFD.

Sexually dimorphic responses to early life programming factors have been described (45, 54–56), which underlines the importance of studying both sexes. In the present study, male animals showed a more pronounced development of hypertrophic obesity and metabolic derangements following an HFD than female animals, which is consistent with previous studies. However, attenuation of the obesogenic response to the HFD was apparent in the BC offspring of both sexes. Thus, similar effects were observed in both sexes regarding whole body adiposity index and tissue (e.g., liver steatosis), cell (e.g., degree of adipocyte size enlargement), and molecular variables (e.g., gene expression in WAT) examined. The most different response between sexes concerned the expression of metabolism-related genes in the liver. Both female and male BC offspring showed decreased HFD-induced liver steatosis. Still, this response appeared more related to suppressing lipogenesis-related genes in the male offspring and the induction of substrate oxidation-related genes in the female offspring. Sex-specific differences in hepatic fat oxidation and synthesis have been previously described. In particular, lipogenesis-related genes are more effectively induced in males than in female mice upon refeeding after a fast (57) or HFD feeding (58). In this context, the finding of a reduced hepatic induction of lipogenesis-related genes with the HFD selectively in the BC male (but not BC female) offspring is noteworthy.

The conversion of dietary BC to VA retinoids driven by BCO1 decreases adiposity in rodents (59, 60). Despite this and other evidence of retinoic acid’s antiobesogenic and metabolic health benefits in adult mammals, studies addressing the metabolic programming action of VA supplementation in early life have shown contradictory results. We previously assessed mild supplementation as retinyl palmitate to newborn rats throughout the suckling period (leading to 3.6 excess liver VA stores at weaning) and found unexpectedly that it favored later HFD-induced hyperplasic obesity (20). In the present study, we sought a physiological increase of VA supply through maternal milk during the entire lactation. To avoid possible delays, Isx^−/− dams were fed a BC-enriched diet starting from gestational day 14 to increase their VA status. We did not intervene during the first 2 weeks of gestation to avoid effects on early organogenesis, considering the critical regulatory role VA plays in this process (61). While we cannot exclude that ISX and BC may affect development through unknown mechanisms (62), we interpret our results as indicating that mild (2 times) increased maternal VA supply programmed obesity resistance in the F1 offspring owing partly to increased energy metabolism in adipose tissues. Our results align with those of Wang et al. (22), who mildly (2 times) supplemented wild-type mouse dams with retinyl acetate during the entire gestation and lactation periods, although WAT browning was less pronounced in our F1 animals. This difference could relate to our dietary BC intervention not covering the first part of gestation when many de novo angiogenesis events occur in mice (63). These early angiogenesis events, stimulated by maternal retinoids, determine the size of the PDGFRα+ progenitor population capable of beige adipogenesis toward adipocytes with thermogenic and enhanced substrate oxidation capacities (22). This adipose-centered, VA-stimulated antiobesity programming mechanism, albeit attenuated, may operate in our experimental setting. Additional mechanisms should explain the distinct responses to the HFD found in BC versus VAS progeny regarding liver metabolism genes and liver VA stores. A lower reduction in tissue VA stores with HFD as observed in the liver of the BC progeny could help explain the resistance to dietary obesity, considering the proven cell-autonomous effects of retinoic acid on liver cells and other cell types relevant to energy homeostasis (17, 64, 65). Mechanisms behind the early programming of obesity resistance by VA may thus involve epigenetic effects on VA metabolism-related genes, which deserve further investigation.

Perspectives and Significance

In summary, the primary outcome of this study is that increased VA supply through the mother during late gestation and the suckling period can impact the offspring’s metabolism, conferring protection from HFD-induced obesity in adulthood. On a broader scale, this study may help better understand the potential physiological consequences of genetic variability in the human ISX gene. We speculate that the combination of genetics in the ISX-BCO1-SRB1 axis and VA nutrition in early life stages may impact the metabolic programming of offspring. Knowledge of these interactions could be valuable in preventive strategies against obesity and metabolic syndrome. Provitamin A carotenoids such as BC have lower toxicity and teratogenicity than preformed VA (66). A maternal diet enriched in BC during late gestation and lactation may, therefore, represent a safer way to sustain increased VA supply and confer obesity resistance to the offspring, particularly among mothers bearing genetic polymorphisms resulting in decreased ISX activity, increased BCO1 or SRB1 activity, or combinations thereof. Thus our findings may have implications for human populations where polymorphisms in the corresponding genes are prevalent (29).

DATA AVAILABILITY

The data supporting this study’s findings are available on request from the corresponding authors.

SUPPLEMENTAL MATERIAL

Supplemental Figs. S1 and S2 and Supplemental Table S1: https://doi.org/10.6084/m9.figshare.26234540.v2.

GRANTS

This work was supported by the Spanish Government (MICIU, AEI, Fondo FEDER/EU) under Grant PGC2018-097436-B-I00 (to A.P.) and National Eye Institute Grants EY028121 and EY020551 (to J.v.L).

DISCLOSURES

No conflicts of interest, financial or otherwise, are declared by the authors.

AUTHOR CONTRIBUTIONS

J.R., J.v.L., and M.L.B. conceived and designed research; R.S., S.G., D.V., P.R., and J.A. performed experiments; R.S., S.G., and J.R. analyzed data; J.R. and M.L.B. interpreted results of experiments; J.R. prepared figures; M.L.B. drafted manuscript; A.P., J.A., J.v.L., and M.L.B. edited and revised manuscript; J.R., J.v.L., and M.L.B. approved final version of manuscript.