The animal study was reviewed and approved under N°11/037 by the local ethical committee for animal experimentation COMETHEA (“Comité d’Ethique en Expérimentation Animale du Centre INRAE de Jouy en Josas et AgroParisTech”), referenced as N°C2EA-45 in the French National registry CNREEA (“Comité National de Réflexion Ethique sur l’Expérimentation Animale”).

Female New Zealand white rabbits of INRA 1077 line were used. They were housed individually in the same facilities with temperature and light controlled environment. At 10 weeks of age, the does were divided into two groups, H and C that were fed ad libitum with either an 8%-lipid (including 6% of soybean oil)-cholesterol (0.2%)-enriched diet (high fat-high cholesterol diet, named to simplify HD as high-fat diet) or a control diet (CD, 2% of lipids), respectively. The nutrient and chemical compositions of CD and HD have been described previously (Picone et al., 2011; Tarrade et al., 2013). Briefly, HD contained quantitatively more fatty acids (FAs), mainly n-6 polyunsaturated fatty acids (PUFAs), namely linoleic acid (Supplementary Table 1 and Supplementary Figure 1), and provided 16% more energy than the control diet. At 18 weeks of age, the does (H: n = 20 females, C: n = 20 females) were superovulated by five subcutaneous administrations of pFSH (Stimufol^®; Merial, Boehringer Ingelhein, Germany) at 12-h intervals, as described previously (Mamo et al., 2008; Reis e Silva et al., 2012), followed 12 h later by an intravenous administration of 30 IU HCG (Chorulon; Intervet MSD Santé Animale, Beaucouze, France) at the time of natural mating. One day post coitum (dpc), the does were euthanized (named embryo donors), and 1-day embryos were recovered from oviducts by flushing and collected in HEPES M199 at 37°C. In parallel, recipient H (rH, n = 10) or C (rC, n = 10) female rabbits previously synchronized with hCG (0.8 μg Receptal^®; Intervet, France) were sedated with ketamin (Imalgen 1000^®, 10 mg/kg) and anesthetized by inhalation using a mixture of isoflurane/oxygen (Rousseau-Ralliard et al., 2019), and embryo transfers were performed on female oviducts after laparotomy. Female rabbits are characterized by a duplex uterus, i.e., two separate functional uteri and cervices, opening into a vagina simplex. To avoid any recipient-induced effect, five to six embryos in the same developmental stage from H or C donors (dH and dC, respectively) were transferred, respectively to the right and left oviducts of H and C female recipients (rH and rC, respectively), leading to four groups of feto-placental units: dHrH, dHrC, dCrH, and dCrC, named to simplify HH (6 females and 6 males), HC(6 females and 9 males), CH (10 females and 10 males), and CC(5 females and 10 males), respectively; the initial letter refers to embryo condition.

Subsequently in the text, feto-placental samples from C or H embryos transferred in C recipient females are designated by CC and HC, and those from C or H embryos transferred in H recipient females are designated by CH and HH, respectively (Figure 1).

Twenty-eight dpc, i.e., 3 days before the end of the gestation, and following overnight fasting, after 30-min premedication with an analgesic (Torbugesic), laparotomy was performed on anesthetized (isoflurane) six H and five C recipient does, and four combinations of collected feto-placental units were obtained: CC (n = 15), CH (n = 15), HC (n = 20), and HH (n = 12). Fetal blood was obtained by ultrasound-guided intra-cardiac puncture, through the uterine wall of the mother prior to the collection of feto-placental units. A drop of fetal blood was immediately used to measure glycaemia using a glucose-strip-test (KM2902, FreeStyle Optium Xceed Medisense; ABBOTT, Chicago, United States). Blood samples were collected in EDTA-coated Vacutainers, then centrifuged, and plasma was stored at −20°C until analysis. The does were euthanized by lethal injection of Dolethal (5 ml; Vetoquinol, Lure, France). Fetuses, placentas, and deciduae were then immediately collected and weighed. Placentas from each group were dissected from the deciduae. Decidua is formed in a process called decidualization under the influence of progesterone. Decidualization is a process that occurs in early pregnancy, indeed, the maternal endometrial changes occur initially at the site of and following blastocyst implantation. Histologically, the placenta of rabbits is composed of the labyrinth zone, junctional zone, decidua zone of necrosis, decidua zone of separation, and mesometrium (Furukawa et al., 2014). The rabbit placental junctional zone is in contact with the maternal decidua. Indeed, because decidua and junctional area (fetal origin) are nested, it is, therefore, not possible to completely separate them by dissection (Lopez-Tello et al., 2019). The fetuses were euthanized by lethal injection of Dolethal and dissected. Kidneys, liver, heart, and the labyrinthine area of the placenta, and decidua, including the junctional zone, were weighed and either snap-frozen and stored at −80°C or transferred in chloroform-methanol (2/1, v/v) and stored at −20°C until further analyses.

Deoxyribonucleic acid was extracted from fetal liver, and PCR for determination of Y chromosome was performed as described elsewhere (Diaz-Hernandez et al., 2008).

One hundred μl of plasma and samples of placental labyrinthine area (400 mg) and liver (200 mg) randomly chosen from each litter (to reach n = 6–8 per group and sex) were used for the analyses. Lipids were extracted with chloroform/methanol (2/1, V/V) (Folch et al., 1957). For placental and hepatic lipid extracts, phospholipid fraction, mostly representative of the membrane lipids, and neutral lipid fraction, representative of the storage lipids, were separated from non-phosphorous lipids on silica acid cartridges (Juaneda and Rocquelin, 1985). The fatty acids (FAs) from plasma lipids and tissue phospholipids and neutral lipids were transmethylated with Boron trifluoride methanol 7% (Sigma-Aldrich, Saint louis, United States) (Morrison and Smith, 1964). FA methyl esters were analyzed by gas chromatography coupled to FID (Gas Chromatograph 3900; Varian, Courtaboeurf, France) on an Econo-Cap EC-WAX capillary column (Tarrade et al., 2013). The FA profile was established for each sample with FA expressed as a percentage of total FAs. The analyses identified FA from C14:0 to C24:1n-9. In result tables, only a few relevant FAs are presented, but all of the identified FAs were considered in the calculation of total FAs. Heptadecanoic acid (C17:0, margaric acid), introduced prior to transmethylation, was used as an internal standard to obtain the relative quantification of total FAs, expressed either as mg/ml of plasma or mg/g of placenta or liver. The desaturase activity indices (DAIs) were estimated as the product:precursor ratios of individual FAs according to the following equations: Δ9 = [16:1ω7/16:0] and [18:1ω9/18:0], Δ6 = [18:3ω6)/18:2ω6)], and Δ5 = [20:4ω6/20:3ω6] (Warensjo et al., 2006; Robbez Masson et al., 2008).

To compare the four groups (CC, CH, HC, and HH), non-parametric Kruskal-Wallis tests were performed, followed by a pairwise comparison by Dunn’s multiple comparisons test (GraphPad Prism 7). P < 0.05 was considered statistically significant. Data are expressed as: median (Q1; Q3). The first quartile (Q1) and the third quartile (Q3) correspond to 25 and 75% percent of scores, respectively.

For lipid analyses on placenta, liver, and plasma, principal component analyses (PCAs) were additionally performed using the R statistical software (R Core Team, 2013).¹ The FactomineR package (dedicated to multivariate data analysis; Lê et al., 2008) was included in R commander to analyze the biological relevance of observed FA changes, namely to extract important information and express this information as a set of new variables called principal components (Rousseau-Ralliard et al., 2019). These principal components can be used to visualize graphically the data. The individual’s factor map is a plot of the principal component scores for individuals on the first two principal components. The correlation circles that do not overlap sign significant differences between groups (significant v.test when an absolute value greater than 2). The variables factor map presents a view of the projection of the observed variables projected into the plane spanned by the first two principal components. Still named correlation circle, it can help to visualize the most correlated variables (i.e., variables that group together) and to show the structural relationship between the variables and the components. The projection of a variable vector onto the component axis allows us to directly read the correlation between the variable and the component. The cos² values (and the v.test) are used to estimate the quality of the representation. The closer a variable is to the circle of correlations, the better its representation on the factor map (and the more important it is to interpret these components). Variables that are closed to the center of the plot are less important for the first components.

Multiple factor analysis is a multivariate data analysis method for summarizing and visualizing a complex data table in which individuals are described by several sets of variables (quantitative and/or qualitative) structured into groups, i.e., multiple factor analysis (MFA) is a global analysis where multiple sets of variables are simultaneously considered (De Tayrac et al., 2009). It takes into account the contribution of all active groups of variables to define the distance between individuals. The number of variables in each group may differ, and the nature of the variables (qualitative or quantitative) can vary from one group to the other, but the variables should be of the same nature in a given group. MFAs were performed on placenta data sets and liver data sets using the R statistical software and the package FactomineR.

Data are expressed in Table 1. The HH fetuses were significantly growth retarded with decreased placental efficiency, compared to controls. They also had significantly lighter liver and kidneys compared to controls. Only fetal length was significantly increased in the CH fetuses compared to controls, but CH placental efficiency was decreased, with significantly heavier placenta and decidua. In contrast, absolute kidney weight was significantly reduced in CH compared to controls. There was no significant difference for any measurement between HC and controls.

The objective of this study was to discriminate the effects of maternal lipid over-nutrition between effects due to a pre-conceptional period of exposure and those due to gestational exposure only. The rabbit model was used because of its metabolic (sensitivity to lipids) and anatomical (hemodichorial placenta) particularities similar to those of humans (Fischer et al., 2012). This study is the follow-up of a previous one, where dams fed with a high-fat diet (HD) started shortly after weaning were shown to have precocious puberty and increased follicular atresia (Cordier et al., 2013). In addition, embryos produced by these dams after mating with control males were characterized by deregulation of gene expression at the onset of embryonic genome activation (8–16 cells) (Picone et al., 2011) and in the blastocyst stage were observed near to term (Tarrade et al., 2013). Offspring postnatal phenotype was characterized by predisposition to develop metabolic syndrome, such as excess adiposity and hypertension (Picone et al., 2011; Dupont et al., 2014). However, since maternal exposure to high-fat diet extended from puberty to the end of gestation, it was not possible to dissociate the effect of each period of exposure on the feto-placental phenotype, i.e., an early effect, affecting the gametogenesis of rabbits from a pure gestational effect of this exposure. In this study, the fetoplacental effects of maternal exposure to a high-fat diet were investigated for each window of vulnerability and a combination of the two. Thereby, data from C embryos or H embryos transferred in C recipient females are designated by CC and HC, respectively, and those from C embryos or H embryos transferred in H recipient females are designated by CH and HH, respectively.

As a programming agent of offspring phenotype, the placental phenotype was investigated at the end of gestation. Both placenta and decidua were targeted to explore placental adaptations in response to the maternal environment and flexibility of the maternal decidua in response to embryo origin. The CC and HH groups were used the negative and positive controls, respectively, to confirm the fetal hypotrophy with significantly reduced placental efficiency previously observed in fetuses from high-fat diet-fed dams (Tarrade et al., 2013), and to exclude any significant effect of reproductive biotechnologies related to embryo transfers. Without any post-natal end point, in this study, possible alterations of offspring phenotype were investigated in the fetal stage, on fetal plasma and liver. Maternal high-fat diet during the preconception period and gestation periods, i.e., the HH group led to growth-retarded fetuses in rabbit. In the literature, the effects of a high-fat maternal diet on birth weight are variable, sometimes leading to no effect or higher birth weight, for example in rats, depending on the period of exposure (during the entire gestation or the last third of the gestation) and qualitative composition of dietary lipids, especially when the diet is saturated (Cerf et al., 2015; Seet et al., 2015).

While epigenetic marks are currently described as major candidates for predicting the long-term health of individuals (Tarrade et al., 2015), the importance of other biochemical parameters might have been overlooked. Among these, fatty acid (FA) profiling makes it possible to obtain a fine phenotype at the end of gestation to highlight the differences between groups according to the windows of exposure to excessive lipid intake in the mother, i.e., before and/or during pregnancy. FAs are described as one of the most essential substances in intrauterine human growth, through their involvement in a number of energetic and metabolic processes, such as the growth of cell membranes (Bobinski and Mikulska, 2015; Shrestha et al., 2020b). Essential FAs (EFAs), namely, linoleic acid (LA, ω6 series) and alpha-linolenic (ALA, ω3 series), and their derivatives, namely, arachidonic acid (AA, C20:4ω6) and docosahexaenoic acid (DHA, C22:6ω3), are among the most important FAs during the intrauterine growth period.

Linoleic acid is highly required during gestation as a precursor of arachidonic acid (and metabolites, i.e., prostaglandins) strongly involved in the development of embryos and placentation, with a positive association with fetal growth when combined with ω3 PUFA (Badart-Smook et al., 1997). However, excessive consumption of ω6 PUFA, i.e., excessive LA, leads to a negative association to fetal growth PUFA (Badart-Smook et al., 1997), by altering cellular metabolism through inflammation or modulation of mitochondrial function (Cetin et al., 2009). Fetal LA is obtained from the maternal diet, and FAs are transported to the fetus through placental FA transporters (FATPs) and binding proteins (FABPs). Moreover, it was recently reported that consumption of ω6 PUFA, i.e., LA, has increased gradually worldwide (Choque et al., 2014), with a potential harmful effect on human health, especially before or during pregnancy, leading to detrimental effects on fetal development that could influence offspring health at adulthood (Shrestha et al., 2020b). LA is a direct precursor of bioactive oxidized linoleic acid metabolites described to promote a strong pro-inflammatory response in rats (Choque et al., 2014). LA is also a precursor of arachidonic acid, which is the substrate of several enzymes leading to the production of pro-inflammatory eicosanoids (prostaglandin E2, thromboxane, and 4 series-leukotrienes). Therefore, excessive consumption of LA during pregnancy may, thus, improve its conversion to AA and unbalance eicosanoid synthesis (Elmes et al., 2004; Choque et al., 2014).

Maternal dietary-induced overweight and adiposity are usually characterized by maternal dyslipidemia. In fact, our previous studies have demonstrated abnormalities in lipid storage in term placentas (Tarrade et al., 2013). However, there is little information on the effect of pre-gestational overweight with excessive fat on placental lipid metabolism. The transfer of both C and H embryos in H females led to a reduction in placental efficiency. Nevertheless, the feto-placental phenotype also differed according to the origin of the embryo, as summarized in Figure 9. Compared to controls (CC), placentas were heavier in the CH group, without significant impairment of fetal or organ weight. On the other hand, compared to the CC group, placentas, and fetuses and their organs (liver and kidneys) were lighter in the HH group. The significant decrease in kidney weight, both in raw and relative to body weight values, could explain the hypertension previously observed in adult male offspring born to HH dams (Picone et al., 2011; Dupont et al., 2014).

In this study, LA-enriched HD significantly increased the fetal liver NL and plasma AA contents in the CH and HH groups, as already shown in ewes where LA supplementation in late pregnancy was shown to enhance placental PG production by increasing placental AA supply (Elmes et al., 2004). Moreover, exposure to HD throughout gestation led to the downexpression of genes involved in nutrient transport, namely glucose transporters (SLC2A1 and 3) or amino-acid transporters (SLC38A1, 2, and 3) as well as FATP4, whose role is to enhance FA internalization for transplacental transfers (Bobinski and Mikulska, 2015). Similar observations were reported on rats where a high maternal LA diet altered the placental FA composition (increased ω6 PUFA and decreased DHA), and increased the placental gene expression of inflammatory proteins and nutrient transporters, with a downregulation of fatty acid transport protein 4 (Fatp4) and glucose transporter 1 (Slc2a1) (Shrestha et al., 2020a).

Most studies show downregulation of key placental transporters for amino acids, lipids, and ions in human IUGR (Duttaroy and Basak, 2020), as seen here in the HH hypotrophic rabbit fetuses. These observations underlie the placental nutrient sensor model and fetal demand model hypotheses stipulating that the placenta integrates maternal and fetal nutritional cues with information from intrinsic nutrient sensors (Gaccioli et al., 2013); thus the placenta would play a critical role in modulating maternal-fetal resource allocation, thereby affecting fetal growth and the long-term health of the offspring. This hypothesis is corroborated here by the response of decidua membrane to embryo origin (increased FA concentration in deciduous PL of HH and, more slightly, in HC). Here, maternal HD during gestation induced increased FA concentration inside the labyrinthine area (in NL lipid fraction). In H recipients, the proportion of ω6 PUFA, such as LA and ALA (ω3 PUFA), in placental lipids (both PL and NL lipid fractions), along with increased PUFA-to-SFA ratio, was greater than in C recipients.

Pre-conceptional exposure to high fat-diet had a major effect on fetal biometry and organogenesis (lighter liver and kidneys) in HH. It also altered FA concentrations of placenta phospholipids, involved in membrane synthesis, as well as increased the MUFA contents in the NL lipid fraction, as revealed by placental MFA. This phenotype, i.e., the accumulation of lipids, was already observed at the embryo level (Picone et al., 2011) and in trophoblastic cells (Tarrade et al., 2013). Thus, an H embryo transferred to a C recipient (HC) evolved favorably in terms of biometrical development, which seemed similar to controls (CC) at term, with, however, a small but significant effect on placental profiles of fatty acids. This MFA analysis shows notably that the maternal diet during gestation plays a dominant role on the feto-placental phenotype (biometry and lipid signature) but that pre-conception maternal environment undeniably remains very important in term of programming.

In tissues, LA and ALA, which make up for the major part of dietary EFA, are converted into FA of longer and more unsaturated chain by alternate desaturation (delta 6, delta 5)-elongation reactions. The liver is one of the most active organs, and its role is critical in providing less active tissues, particularly the brain, with long-chain PUFA secreted and transported in VLDL (very low density lipoprotein) (Bezard et al., 1994). In this study, maternal HD had different impacts on fetal liver phenotype depending on time windows of exposure, as revealed by the fetal liver MFA.

Excessive ω6 PUFA consumption, together with high ω6/ω3 PUFA ratio (i.e., low intake of ω3 PUFA), is highly associated with the pathogenesis of many modern diet-related chronic diseases (Mariamenatu and Abdu, 2021) and could lead to a pro-inflammatory environment. In this study, the linoleic acid-rich HD significantly increased the AA content of fetal plasma and fetal liver NL in the CH and HH groups, as already shown in ewes in which LA supplementation in late pregnancy enhanced placental PG production by increasing the supply of AA (Elmes et al., 2004). In comparison to C recipients, in the H recipient groups (CH and HH), fetal liver lipids (both PL and NL lipid fractions), as for placenta, had a greater proportion of ω6 PUFA, such as LA, and of ALA with an increase in the PUFA-to-SFA and ω6-to-ω3 ratios, whereas DHA and long-chain MUFAs (C18:1, C20:1, and C24:1) were reduced in liver membranes (PL fraction). In rats, in the context of high-fat diet (HFD), high consumption of LA during pregnancy was reported to affect liver FA profiles of offspring in a similar manner, with increased ω6 PUFA and decreased MUFA (Marchix et al., 2020a). Surprisingly, however, and as previously observed in rabbits (Tarrade et al., 2013) and rats (Shrestha et al., 2021) fed with a high-fat diet rich in LA, FA concentrations in liver neutral lipids were similar among groups.

In parallel, the expression of several genes, among which IRS1, IR, PCTP, SCD5, and PPARα, as well as those involved in lipid metabolism was downregulated in CH and HH in this study. This is in agreement with data reported in rats, where a maternal high-LA diet was shown to lead to reduced hepatic mRNA expression of several genes involved in lipogenic pathway, srebp-1c (de novo lipogenesis), Scd-1 (conversion of SFA to MUFA), Fasn (FA synthase), and Elovl6 (FA elongase 6), and to affect the desaturation index (Δ5 desaturase) (Marchix et al., 2020b). Thus, in rats, maternal HF diet during pregnancy was shown to alter the liver C16 desaturase pathway (Seet et al., 2015), as in this study with an increased Δ9 C16 desaturase index. The strong decrease in the hepatic expression of PCTP could be the cause of a possible degradation of glucose homeostasis in fetuses of H recipients (CH and HH), since the increase in circulating PLTP (phospholipid transfer protein) is described as an inter-organ (brown adipose tissue and liver) mediator to improve glucose tolerance and insulin sensitivity (Sponton et al., 2020). Furthermore, in rats, dietary-induced maternal obesity has been shown to program hepatic metabolic anomalies that lead to the development of fatty liver in the offspring through alterations involving PPARα-activated mechanisms (Heinecke et al., 2020).

The expression of genes involved in lipid metabolism, such as FAS, was repressed in the liver of fetuses carried by the H recipients (i.e., CH and HH groups); however, the overexpression of albumin mRNA was also observed in these groups. Albumin reversibly binds fatty acids with high affinity (Naval et al., 1992) and might be involved in the high materno-fetal FA flux across the placenta. In rats, PUFA in fetal plasma and liver are significantly correlated to PUFA in maternal plasma, which is also closely related to the fatty acid composition in the maternal diet (Amusquivar and Herrera, 2003). In addition, rat fetuses exposed throughout gestation to a maternal high-fat diet were shown to have plasma FA profiles rich in ω6 PUFA (Cerf et al., 2015). In the present study, similar effects were observed in fetal plasma from CH and HC groups, with increased LA and DGLA contents, as well as in fetal liver, besides increased LA, eicosatrienoic acid (C20:2ω6) and AA whereas DPA decreased compared to controls.

Again, exposure to HD during the pre-conception period had a major effect on fetal biometry and organogenesis, with repercussions on the fatty acid concentration of fetal liver PL and NL, as reported in a previous study (Higham et al., 1984); it increased the content of MUFA in both PL and NL lipid fractions (in HC vs. CC and in HH vs. CH), and decreased the content of SFA and long-chain (C22) ω6 PUFAs. Thus, an H embryo transferred to a C recipient (HC) evolved favorably in terms of biometrical development, which seems similar to controls (CC) at term, with, however, a small but significant effect on fetal liver fatty acid concentration and biometry data. This MFA analysis also shows that the maternal diet during gestation played a dominant role in fetal liver phenotype (gene expression and lipid signatures).