Maternal diet-induced obesity during pregnancy alters lipid supply to fetuses and changes the cardiac tissue lipidome in a sex-dependent manner

Maternal obesity during pregnancy has immediate and long-term detrimental effects on the offspring heart. In this study, we characterized the cardiac and circulatory lipid profiles in fetuses of diet-induced obese pregnant mice and established the changes in lipid abundance and fetal cardiac transcriptomics. We used untargeted and targeted lipidomics and transcriptomics to define changes in the serum and cardiac lipid composition and fatty acid metabolism in male and female fetuses. From these analyses we observed: (1) maternal obesity affects the maternal and fetal serum lipidome distinctly; (2) female heart lipidomes are more sensitive to maternal obesity than male fetuses; (3) changes in lipid supply might contribute to early expression of lipolytic genes in mouse hearts exposed to maternal obesity. These results highlight the existence of sexually dimorphic responses of the fetal heart to the same in utero obesogenic environment and identify lipids species that might mediate programming of cardiovascular health.


INTRODUCTION
. Fetal characteristics at gestational day 18.5. (A) Body weight of male and female fetuses from healthy control (CTL) and obese (OB) mouse dams at gestational day 18.5. (B-C) Heart weight and heart weight/body weight ratio of male and female fetuses from CTL and OB dams at gestational day 18.5. Male CTL n=8, male OB n=7, female CTL n=9, female OB n=8. (D) Representative images and quantification of cell nuclei count per µm 2 . Histological sections stained with haematoxylin and eosin of male and female fetuses from CTL and OB dams at gestational day 18.5. Male CTL n=6, male OB n=6, female CTL n=5, female OB n=7. Scale bar indicates 20 µm.
p-value calculated by Student t-test.
We then explored if a subset of lipids was transported from the maternal to the fetal circulation 112 using linear regression. We showed that only a few maternal phospholipids and lyso-   phospholipids comprising three and two fatty acids respectively, that are covalently bound to 131 glycerol. As we observed differences in the abundance of individual lipid isoforms in maternal 132 and fetal serum, we sought to elucidate whether the distinct signatures observed in fetuses from 133 obese dams would also translate into an imbalance in the distribution of fatty acid residues in 134 phospholipids. When clustered according to the number of double bonds as either saturated, 135 monounsaturated or polyunsaturated residues, we saw that the relative abundances of these 136 were not significantly affected by maternal diet either in maternal or fetal serum ( Figure 3A). 137 We also noted that fatty acids from phospholipids with a chain length shorter than 18 carbons 138 were generally more abundant in obese serum, whereas the abundance of longer molecules 139 tended to be lower ( Figure 3B).

143
In line with the lower polyunsaturated phospholipid levels, those containing n-3 144 docosapentaenoic acid (DPA) (22:5) were less abundant, and those with saturated fatty acids 145 myristic (14:0) and margaric (17:0), and monounsaturated myristoleic acid (14:1) were more 146 abundant in the fetal sera of both sexes in response to maternal obesity. In contrast, oleic acid 147 (18:1) from phospholipids was more abundant in the serum of obese dams only. Despite the 148 differences observed, the fetal levels of a few residues were significantly correlated with the 149 maternal ones, and an overall trend for positive correlation between fatty acids levels in 150 maternal and fetal sera was observed ( Figure 3C). By generating correlation matrices with 151 group independent variables using Euclidian clustering, we also observed that several saturated negatively correlated to polyunsaturated species.

156
Maternal obesity sex-specifically affects the fetal heart lipidome 157 We next sought to determine whether the lipid composition of fetal hearts was influenced by  Through PCAs, we observed a weaker degree of separation between control and obese 167 individual lipidomes in males ( Figure 4C) when compared to females ( Figure 4D). In contrast, 168 female heart lipidomes showed clear distinction between control and obese hearts, with more      Figure 5B). Figure 5C shows expression of genes regulated by PPAR-alpha transcriptional 204 activity, and expression of genes mapped to the main predicted IPA pathways. We later 205 validated the expression of key genes associated with lipid metabolism in the hearts of both 206 male and female offspring from a completely new cohort using qPCR ( Figure 5D). 208 Having observed changes in transcriptional activity indicating increased lipid metabolism in 209 fetal hearts in response to maternal obesity, we conducted a final experiment to investigate 210 whether acyl-carnitine species were also affected by maternal obesity. These comprise fatty 211 acid residues produced during beta-oxidation and are markers of mitochondrial and 212 peroxisomal lipid metabolism. Regardless of the lack of significant differences between sex-213 matched obese and control offspring ( Figure 6A), we observed increased levels of total 214 hydroxylated acyl-carnitines in response to maternal obesity by factorial ANOVA ( Figure 6B).

215
At the individual level, we found the hydroxylated acyl-carnitine C05-OH to be more abundant 216 in both males and females ( Figure 6C, 6D, 6E), and C16-OH to be less abundant in females 217 only ( Figure 6E). C12:0 and C12:1 were also more abundant, whereas C20:0 and C22:5 were  might also indicate that the fetal myocardium exposed to maternal-obesity-induced stress 284 maintains its ability to uptake circulating fatty acids. 285 We observed changes in the fetal cardiac transcriptome that were induced by maternal obesity,  In conclusion, we have carried out a comprehensive study of how obesity during pregnancy 314 influences lipid availability to the fetus and consequently affects the fetal heart lipidome. From 315 our findings, three main principles emerge: (1) There is a discrepancy between how the 316 maternal metabolic status affects the maternal and fetal serum lipidome, an outcome likely 317 related both to the placental actions as a selective barrier and to changes in fetal metabolism;

318
(2) Despite being exposed to the same maternal metabolic milieu, male and female fetal hearts 319 show distinct responses to maternal obesity, mainly at the level of fatty acid residues and greater changes than males. Cardiac morphological changes were also uniquely observed in    with an air gap (1·5 μL). The tip was pressed against a fresh nozzle and the sample was 412 dispensed using 0·2 psi (N2 (g)). Ionisation was achieved at a 1·2 kV. The Exactive was set to 413 start acquiring data 20 s after sample aspiration began. The data were collected with a scan rate  Theoretical lists of known species (by m/z) were used for both positive ion and negative ion 424 mode (~8·5k species including different adducts and fragmentations). Variables whose mass 425 deviated by more than 9 ppm from the expected value, had a signal-to-noise ratio of <3 and had signals for fewer than 20% of samples were discarded. The correlation of signal intensity to 427 concentration of lipid variables found in pooled mouse heart homogenate, pooled mouse liver 428 homogenate, and pooled human serum samples (0·25, 0·5, 1·0×) was used to identify the lipid 429 signals the strength of which was linearly proportional to abundance (r > 0·75) in samples.  (Nelson et al., 1996) to impute values and populate the dataset.   514 Due to the untargeted high-throughput aspect of this study and to the scarcity of available data 515 into fetal lipidomics, the use of a power analysis accounting for changes in fetal cardiac lipids 516 to predict sample size was challenging. The number of animals used in the present study was 517 therefore predicted using previously obtained data on histological assessment of the ratio left 518 ventricle:lumen in the male fetal heart, and on the extensive track record of published studies 519 from our research group using the same maternal obesity model employed in the current study.

520
According to an a priori unpaired t-test power calculation, an n equal or greater than 5 would 521 be required to achieve significance set at α < 0.05, 80% power. Also, according to the resource 522 equation, an n equal to or greater than 6 results in more than 10 degrees of freedom, and is 523 therefore adequate. We then concluded that a sample size greater than 6 would be necessary to 524 show any significant changes in our study.            Figure 6-source data 1. Acyl-Carnitines abundance obtained by spectrometry of the heart 618 (negative mode). Raw data and statistical significance.