Fetal maturity at the feto-maternal interface: contribution of fetal and maternal genomes and tissue metabolism perturbations – CO-LOcATION

Metabolome, lipidome and transcriptome analyses were performed on 224 fetuses: 7 biological replicates for the 2 maturity status, 2 sexes, 4 genotypes, 2 gestational stages of endometrium and placenta. The fetuses were selected from the 407 fetuses of the PORCINET project. For maturity, the selection of fetuses was carried out using biometric, plasmatic and metabolomic indicators.
The 448 samples (224 fetuses; 2 tissues) were extracted to separate total lipids and aqueous fractions. Metabolome of the aqueous fraction was performed using 1H-NMR at 600 MHz. Spectra were processed using ASICS package (R Bioconductor) for identification and quantification of metabolites.
The lipidic fraction was derivatized and analyzed using Gaz Chromatography coupled with a FID detector. The quantifications of the neutral lipid classes and total fatty acids are expressed relatively to internal standards and in mg/proteins.
RNA sequencing was performed using nine lines of Illumina Novaseq. Bio-informatic analyses were performed using the Nextflow nf-core RNAseq pipeline to produce data count files for transcripts, genes.
Statistical analyses are conduced upon R environment. Exploratory analyses (Principal Component Analysis, …) described the three datasets in placentas and endometria separately. Differential analysis relative to day of gestation, breeds, and their interactions is performing on each tissue and dataset, using linear mixed models or generalized mixed models followed by multiple testing corrections. Lists of differential variables will be analyzed for enrichment in biological processes, molecular functions, cellular components, KEGG pathways.
Unsupervised methods as Partial Least Square (PLS), Diablo from MixOmics, … will be performed to explore interactions between dataset features to highlight modules of interest in each tissue. The description of the crosstalk at the feto-maternal interface will be performed by establishing bipartite multi-tissue networks from (s)PLS and (r)CCC analyses… and identifying molecular features coevolving between endometrium and placenta.
We will use linear mixed models including factors and interactions as fixed effects and the sow as random effect to unravel the impact of the maternal, paternal genome and sex on the variables (transcriptome, metabolome and lipidome) in the two tissues.
Supervised (to predict the maturity status at the tissue level with minimal set of variables to predict) and unsupervised (to identify groups of fetuses with trait differences) methods will be then realized on each dataset with machine learning methods for classification. Integrated approaches will be undergone to combine several datasets or both tissues to more specifically target impaired crosstalk between endometrium and placenta. Machine learning methods may identify key variables for further studies.
Rules for the traceability and storage of data and analyzes have been established.

Objective 1 : Establish a first tissue co-expression network of the feto-maternal interface at the end of gestation
1-1 Caracterization of the biological processes involved at the end of gestation

A first multivariate exploratory analysis of the phenotypic data (45 variables) by principal component analysis shows that the 1st axis explains 52.7% of the total variability and separates the two gestation days (Figure 1A). Interestingly, we can note that some fetuses of less maturity at 110d of gestation join the fetuses at 90d. For each gestation period, axis 2 of the principal component analysis explains between 11.5 and 13% of the total variability and separates the fetuses according to their four genotypes.
The first statistical analyzes of multi-omics data show a significant metabolic and transcriptomic change in the endometrium between 90 and 110 days of gestation (DG) as well as between maternal (matG) and fetal (fetG) genotypes. Out of 46 metabolites present in the endometrium, the concentrations of 15 and 14 metabolites discriminate respectively between DG and matG (OPLS-DA, VIP>1). Transcriptomic analysis identifies a large number of differential transcripts (61% of expressed transcripts (ET); FDR < 0.01) with greater differences between matG (52% of ET) than between DG (34.3% of ET). In the placenta, of the 48 metabolites present, the concentrations of 20 and 32 metabolites discriminate respectively DG and fetG (PLS-DA, VIP>1). A large number of transcripts vary according to the fetG at fixed DG (32% of ET, FDR<0.01) or according to the DG at fixed fetG (33.9% of ET, FDR<0.01) (Figure 1B).

The results of this study will provide major issue with an integrated view of feto-maternal dialog (mother, fetus and placenta). The study will provide relevant genes, metabolites, lipids and molecular processes associated with variability in fetal maturity to unlock scientific barriers and promote better survival and robustness of piglets at older age. They will give crucial keys to stimulate the development of new genetic and/or nutritional strategies (e.g., better metabolic trade-off between the sow and its piglets, genomic variability of new candidate regions from dams and sires).

Not concerned

Neonatal mortality is mostly due to a reduced piglet maturity. An essential question for adaptation to extra-uterine environment lies in the optimization of the crosstalk between maternal and fetal genomes for a distribution of resources that allows fetuses to express healthier phenotypes at birth. The CO-LOcATION project proposes an integrative biology approach combining transcriptomics, metabolomics, lipidomics and maturity phenotypes to answer this crucial question for both research and pig production sustainability. The project will use samples from the former ANR PORCINET project, which produced pure genotypes and reciprocal crossbreed genotypes with contrasted robustness at birth. The complexity of the maturity phenotypes will be addressed by studying the trade-off between growth and survival with a focus on endometrial-placental interactions using whole-genome approaches. The CO-LOCATION project will propose new genetic and (or) nutritional strategies or hypothesis towards an increased piglet maturity and survival.