To unravel the influence of the fetal genome and sex at the feto-maternal interface, the expression of 42 genes (Additional file 1) was assessed at the maternal interface (endometrium) during late gestation (D90/D110). Eighty-eight endometrial samples were collected from reciprocal crosses between LW or MS sows inseminated with mixed LW and MS semen. The experimental design is described in Additional file 2. In this condition, pure and crossed fetuses developed together in the same uterine environment. The gene set includes 27 imprinted genes selected from the 489 imprinted genes present in mammalian species (https://www.geneimprint.com/site/genes-by-name, June 2023). Twelve genes were selected because their expression was influenced by one of the parental genomes in our previous study using the same experimental design [10]. Finally, CBX4 (Chromobox Homolog 4) and SLC16A3 (Solute Carrier Family 16, Member 3) genes were selected as epigenetic regulator and target, respectively, in the mouse species, from the study of Goa et al. [29]. The gene encoding SLC38A3 (Solute Carrier Family 38, Member 3) was selected for its protein interaction with IGF1 (Insulin Like Growth Factor 2) [30].
Endometrial gene expression profile of pregnant sows with extreme phenotypes for neonatal survey
The imprinted genes are known to play a critical role in placental development and their expression profiles, in contrast to the endometrium, are well documented. Thus, changes in the expression levels of these genes may influence nutrient transport. We first investigate whether the expression levels of these genes vary in the endometrium at the end of gestation and between sows of extreme breeds for neonatal survival.
Of the 42 genes, 39 were found to be expressed and profiled in the endometrium. The expression of ASCL2 (Achaete-Scute Family BHLH Transcription Factor 2), PHLDA2 (Pleckstrin Homology Like Domain Family A Member 2) and MESP1 (Mesoderm Posterior BHLH Transcription Factor 1) genes were not detected. Table 1 describes the 22 genes that showed global significant differences between breeds and days of gestation.
Table 1
– Global effects of breeds and days of gestation on endometrial gene expression
GENE | Full model | | Additive model | | D: Breed | | D (D110/D90) | | Breed (MS/LW) |
| FDR | | FDR | | FDR | | FDR | FC | | FDR | FC |
GRB10 | 3.10E-03 | | 2.20E-03 | | | | | | | 1.40E-03 | 1.33 |
DHCR7 | 2.70E-03 | | 2.80E-03 | | | | 9.60E-04 | 0.45 | | | |
MEST | 8.10E-04 | | 1.20E-03 | | | | 4.80E-04 | 0.34 | | | |
PEG10 | 1.40E-02 | | 1.00E-02 | | | | 4.50E-03 | 0.47 | | | |
TFPI2 | 3.60E-02 | | 1.70E-02 | | | | 3.60E-02 | 1.30 | | | |
SLC38A4 | 2.70E-02 | | 1.20E-02 | | | | | | | 3.30E-02 | 0.833 |
SLC22A3 | 3.60E-02 | | 1.90E-02 | | | | 1.30E-02 | 0.73 | | | |
IFG2R | 2.50E-02 | | | | 0.095 | | | | | | |
NNAT | 2.30E-02 | | 1.20E-02 | | | | | | | 9.61E-03 | 1.29 |
DCN | 1.00E-02 | | 3.80E-03 | | | | | | | 2.40E-03 | 1.42 |
QPCT | 1.80E-02 | | 4.40E-02 | | | | 2.80E-02 | 1.26 | | | |
AMPD3 | 2.30E-04 | | 1.00E-04 | | | | 5.20E-03 | 0.83 | | 3.60E-04 | 1.3 |
NDP | 9.00E-04 | | 1.00E-02 | | 0.055 | | 5.20E-03 | 0.61 | | | |
NES | 2.50E-02 | | 1.10E-02 | | | | 5.20E-03 | 0.79 | | 2.40E-02 | 1.21 |
ASNS | | | 4.50E-02 | | | | 5.20E-03 | 1.1 | | | |
CITED1 | 9.10E-04 | | 2.20E-03 | | | | 5.20E-03 | 0.38 | | | |
TSPAN33 | 3.50E-03 | | 2.40E-03 | | | | | | | 1.60E-03 | 0.72 |
CDKN1C | 1.40E-02 | | 8.80E-03 | | | | 3.40E-03 | 0.70 | | | |
CREM | 3.00E-02 | | 1.40E-02 | | | | | | | 2.70E-02 | 0.86 |
ASPSCR1 | 2.70E-02 | | 1.20E-02 | | | | 2.00E-02 | 0.76 | | | |
H19 | 2.20E-03 | | 2.80E-03 | | | | 9.60E-04 | 0.45 | | | |
SLC38A2 | 3.20E-02 | | 1.40E-02 | | | | | | | 2.80E-02 | 0.83 |
Twenty-two genes are listed according to whether they are significant for the full, additive, interaction, days of gestation and breed models.
The fold change and FDR are given for each significant differential gene expression.
D: day of gestation; Breed (MS/LW): comparison between Meishan vs Large Whitebreed; D (D110/D90): comparison between 110 and 90 days of gestation; D: Breed: interaction between days of gestation and breeds.
The statistical analyses identified 21 genes with significant differences (FDR < 0.05) that fit the full linear model including day of gestation (D), breed, and their interactions. Twenty-one genes showed differential expression (DEGs) that fit the additive linear model. Finally, the analysis identified NDP (Norrin Cystine Knot Growth Factor NDP) and IGF2R (Insulin Like Growth Factor 2 Receptor) genes that tended to show relative expression depending on days of gestation and breeds (FDR = 0.055 and 0.095, respectively, Table 1). For these two genes, we observed a significant lower expression at D90 for IGF2R and a significant higher expression at D110 for NDP in MS compared to LW (Table 2, Additional file 3).
Table 2
– Differential gene expression according to gestation days and breed subsets
GENE | MS/LW D90 | | MS /LW D110 | | D110/D90 LW | | D110/D90 MS |
| FDR | FC | | FDR | FC | | FDR | FC | | FDR | FC |
GRB10 | 1.300E-02 | 1.3 | | 6.260E-02 | 1.37 | | | | | | |
DHCR7 | | | | | | | 3.000E-03 | 0.36 | | | |
MEST | | | | | | | 1.100E-03 | 0.25 | | | |
PEG10 | | | | | | | 1.600E-02 | 0.44 | | | |
TFPI2 | | | | | | | 4.800E-02 | 1.33 | | | |
SLC38A4 | | | | | | | | | | | |
SLC22A3 | | | | | | | 3.900E-02 | 0.66 | | | |
IFG2R | 1.300E-02 | 0.78 | | | | | 1.800E-02 | 0.83 | | | |
NNAT | | | | 3.200E-02 | 1.41 | | | | | | |
DCN | 1.300E-02 | 1.43 | | 8.800E-02 | 1.40 | | | | | | |
QPCT | 2.400E-02 | 0.84 | | | | | | | | 8.500E-03 | 1.46 |
AMPD3 | 2.900E-03 | 1.34 | | 7.600E-02 | 1.25 | | | | | 3.400E-02 | 0.81 |
NDP | | | | 4.000E-02 | 1.84 | | 1.100E-03 | 0.36 | | | |
NES | | | | | | | | | | | |
ASNS | | | | | | | | | | | |
CITED1 | | | | | | | 1.100E-03 | 0.29 | | | |
TSPAN33 | 1.260E-02 | 0.69 | | 7.900E-02 | 0.74 | | | | | | |
CDKN1C | | | | | | | 1.600E-02 | 0.66 | | | |
CREM | | | | 7.800E-02 | 0.83 | | | | | | |
ASPSCR1 | | | | | | | 4.000E-02 | 0.73 | | | |
H19 | | | | | | | 2.200E-03 | 0.34 | | | |
SLC38A2 | 3.400E-02 | 0.82 | | 9.200E-02 | 0.84 | | | | | | |
Twenty-two genes are listed according to whether they are significant between breeds at each days of gestation and between days of gestation for each breed.
The fold changes are given for each significant differential gene expression (FDR < 0.1).
The expression of 14 genes varied during the end of gestation (FDR < 0.05; Table 1). We observed a significant decrease in the expression at D110 compared to D90 for 11 genes (e.g. H19 (H19 Imprinted Maternally Expressed Transcript), DHCR7 (7-Dehydrocholesterol Reductase), MEST (Mesoderm Specific Transcript) or SLC22A3 (Solute Carrier Family 22 Member 3)) and the over-expression of three genes (TFPI2 (Tissue Factor Pathway Inhibitor 2), QPCT (Glutaminyl-Peptide Cyclotransferase) and ASNS (Asparagine Synthetase (Glutamine-Hydrolyzing))). The expression of nine genes differed between the two breeds regardless of the day of gestation (FDR < 0.05; Table 1). Five genes were up-regulated in MS compared to LW (GRB10 (Growth Factor Receptor Bound Protein 10), NNAT (Neuronatin), DCN (Decorin), AMPD3 (Adenosine Monophosphate Deaminase 3), NES (Nestin)) and four genes were down-regulated (SLC38A2 (Solute Carrier Family 38 Member 2), SCL38A4 (Solute Carrier Family 38 Member 4), TSPAN33 (Tetraspanin 33) and CREM (CAMP Responsive Element Modulator)).
Subsequent analyses focused on the 22 differentially expressed genes from the full and additive models (FDR < 0.05). Table 2 shows that at D90, the expression of four genes (IGF2R, QPCT, TSPAN33 (Tetraspanin 33), SLC38A2) decreased and the expression of three genes (GRB10, DCN and AMPD3) increased in MS compared to LW. At D110, three genes were down-regulated (TSPAN33, CREM and SLC38A2) and five genes were up-regulated in MS compared to LW (GRB10, NNAT, DCN, AMPD3 and NDP). The differential expression of five genes between breeds was maintained at each day of gestation (GRB10, DCN, AMPD3, TSPAN33 and SLC38A2). Finally, the change in gene expression at the end of gestation (D110/D90) mainly affected the LW breed, i.e. 11 genes in LW versus 2 in MS (QPCT, AMPD3). Figure 1 illustrates these results with the expression profiles of DCN, QPCT, IGF2R, TSPAN33, GRB10 and NNAT. At D90, we visualized a significant increase in DCN and GRB10 gene expression and a decrease in gene expression for QPCT, IGF2R and TSPAN33 in MS endometrium compared to LW. At D110, TSPAN33 gene tends to be less expressed in MS endometrium than in LW, but DCN, GRB10 and NNAT are more expressed in MS endometrium.
We carried out principal component analysis from the 22 regulated genes to confirm the relevance of the results. This descriptive analysis showed a clear separation between gestation days and sow breeds (Fig. 2A). The first axis explained 31.5% of the expression variability and discriminated the endometrium according to gestation days (D90/D110). The second axis explained 16.4% of the expression variability and discriminated the endometrium according to sow breed (LW/MS). Finally, the PCA plot visualized a greater variability of expression in the LW endometrium than in the MS endometrium. The correlation circle shows the genes that contribute most to the determination of the axis. The first axis of the correlation circle (Fig. 2B) confirmed the opposite regulation pattern reported in Table 1 with MEST, DHCR7 (7-Dehydrocholesterol Reductase), CITED1 (Cbp/P300 Interacting Transactivator With Glu/Asp Rich Carboxy-Terminal Domain 1), PEG10 and H19 over-expressed genes at D90 and TFPI2, ASNS, QPCT, CREM and SLC38A2 over-expressed genes at D110. The second axis of the correlation plot separated TSPAN33, IGF2R and SLC38A2 over-expressed genes in LW to NNAT, DCN and GRB10 over-expressed genes in MS. Finally, in this study LW endometrium can be characterized by an over-expression at D90 of TSPAN33, DHCR7 and MEST genes and at D110 of SLC38A2 and CREM genes. MS endometrium may be characterized by a strong over-expression of DCN and GRB10 genes compared to LW endometrium.
Associations between endometrium gene expression and fetal phenotypes.
We selected three sets of fetal phenotypes (plasma parameters and placental and fetal biometric measures) known to be associated with maturity traits to identify relevant DEGs that correlated with these phenotypic variables. Six of the fourteen fetal plasma parameters previously analyzed by Yao et al. [31] were selected as indicators of metabolic changes that occur late in gestation (glucose, lactate, fructose, albumin, IGF1, cortisol). Four fetal (body weight, body length, body mass index (BMI) and ponderal index (PI) and four placental (weight, area, width and efficiency) biometric measures were also selected as indicators of fetal growth and maturity. Given the large changes in gene expression at the end of gestation, analyses were performed at each stage of gestation to emphasize stage specificity.
First, we carried out a correlation analysis to assess the association between the expression of the 22 DEGs in endometrium and the phenotypes of their respective fetuses. The correlation matrices are shown in Fig. 3. At D90, the analysis identified six (GRB10, SLC22A3, DCN, QPCT, AMPD3, NDP), one (TSPAN33) and five genes (GRB10, SLC22A3, NES, TSPAN33, H19) with significant correlations (p < 0.01) for fetal plasma parameters, placental and fetal biometric measurements, respectively. At D110, the significant correlations concerned the expression of eight (e.g. SLC38A4, SLC38A2, AMPD3), seven (e.g. GRB10, TSPAN33) and six (e.g. GRB10, NES) genes, respectively. In particular, this analysis highlighted three phenotypes with the highest number of gene expression correlations. Lactate concentration in fetal plasma weight correlated with the expression of four genes at D90 (Fig. 3A; GRB10, DCN, QPCT, AMPD3). Placental weight correlated with the expression of six genes at D110 (Fig. 3B; GRB10, DCN, AMPD3, NES, TSPAN33, CREM). Fetal body weight correlated with the expression of five genes at D110 (Fig. 3C; GRB10, PEG10, NDP, NES, CDKN1C (Cyclin Dependent Kinase Inhibitor 1C)). Taken together, SCL22A3 and TSPAN33 had the highest number of phenotypic correlations. The expression of the SLC22A3 gene correlates with 3 fetal biometric measures (body length, BMI, PI) (Fig. 3C, D90) and the TSPAN33 gene correlates with six phenotypic variables derived from placental biometric measures (area, weight and efficiency) at D110 (Fig. 3B) and fetal biometric measures (weight, length and PI) at D90 (Fig. 3C). Figure 4 illustrates these correlations for endometrial DCN expression level and fetal plasma lactate at D90 (Fig. 4A) and glucose at D110 (Fig. 4B). Figures 4A and 4B show a progressive separation of the two maternal genotypes along the regression line. For example, at D90 endometrial DCN expression and plasma lactate levels were lower in LW than in MS.
Partial least squares regression analyses were used to explore the best correlation networks and attempt to explain the relationship between gene expression and the three groups of fetal phenotypes. At D90, lactate (Fig. 5A, correlation threshold |0.38|), placental efficiency (Fig. 6A D90, correlation threshold |0.32|), body length, and PI (Fig. 6B D90, correlation threshold |0.37|) were the phenotypic variables that had the highest correlations with DEG expression. At D110, fetal plasma metabolites generated two networks with four genes correlated with fructose and glucose (DCN, GRB10, NNAT, AMPD3) and two genes correlated with albumin (Fig. 5B; SLC38A2, CREM). Placental biometric measures generated a network of seven genes (Fig. 6A D110) that confirmed and complemented the previous analysis. Particularly, placental weight was associated with the expression of the seven genes with positive (CREM (r = 0.45) and TSPAN33 (r = 0.44)) or negative (GRB10 (r=-0.57), NES (r=-0.36), DCN (r=-0.48), AMPD3 (r=-0.37), NNAT (r=-0.36)) correlations. The fetal biometric measures had more correlations with DEG expression at D90 than at D110, generating a network of seven genes (Fig. 6B). At D90, body length was the phenotype most highly correlated with gene expression (GRB10, CITED1, MEST, SLC22A3, CDKN1C, TSPAN33). At D110, fetal weight is associated with the expression of four genes (Fig. 6B; GRB10, NES, CREM, SLC22A3).