Single-cell RNA transcriptomics in mice reveals embryonic origin of fibrosis due to maternal obesity

Abstract

Background: Over 40% of pregnant women in the USA are obese which negatively affects fetal development and offspring health. Maternal obesity (MO) leads to fibrotic infiltration in multiple tissues and organs of offspring during their adulthood although the origin and mechanisms are unclear.

Methods: C57BL/6J female mice were fed a control and high-fat diet to mimic MO condition. Embryonic somatic tissues were obtained at E9.5, E11.5, and E13.5 (equivalent to 6 weeks of human pregnancy) from control (CON) and MO mice for single-cell RNA-sequencing (scRNA-seq). To explore the role of AMP-activated protein kinase (AMPK), AMPK was activated by metformin and A769662, and knocked out in embryonic mesenchymal cells (EMC) using AMPKα1 floxed mice.

Findings: Using unsupervised clustering, we identified three major cell populations with fibrogenic capacity. Compared to CON, the population of fibrogenic cells increased dramatically (by ∼125%) due to MO, supporting an embryonic origin of fibrosis in the offspring. MO induced inflammatory response and elevated expression of transforming growth factor β (TGFβ) signalling and fibrogenic genes in embryos. MO inhibited AMPK and its activation by metformin and A769662 inhibited TGFβ signalling and fibrogenesis.

Interpretation: MO profoundly enhances embryonic fibrogenesis, explaining the origin of fibrosis in the offspring of mothers living with obesity. Our data underscore the importance of early intervention, before 5-6 weeks of pregnancy, in improving embryonic development, and AMPK is an amiable target for suppressing excessive fibrogenesis in MO embryos to assist increasing populations of obese mothers having healthy children.

Funding: This work was funded by National Institutes of Health Grant R01HD067449.

Keywords: AMPK; Embryo; Fibrogenesis; Maternal obesity; PRRX1; TGFβ; scRNA-seq.

Fig. 1

Single-cell transcriptomic profile of mouse embryos at embryonic days E9.5, 11.5 and 13.5 from Control (CON) and Maternal obesity (MO) group. (a) Uniform manifold approximation and projection (UMAP) plot showing 18 major clusters from integrated data sets from three different time points. Names of individual clusters are mentioned next to the figure along with their corresponding identification number. (b) UMAP showing clusters of cells from CON and MO groups. A major differential area between two groups is indicated by dotted circles. (c) Bar plots showing comparison between the numbers of each cell type in CON and MO groups. (d) Bar plot showing numbers of differentially expressed genes (DEGs) induced by MO in each cell type. Coral and Aqua bar indicates upregulated and downregulated DEGs respectively (obtained from edgeR, P < 0.05). (e) Representative enriched KEGG pathways of upregulated genes from integrated dataset due to MO. (f) Representative biological processes from Gene set enrichment analysis (GSEA) related to inflammatory response, transforming growth factor β (TGFβ) production, response to TGFβ and fibroblast proliferation enriched in upregulated genes due to MO. CON, Control group and MO, Maternal Obesity group. (See also Supplementary Tables S1, S2, and S4).

Fig. 2

Identification of unique cell clusters with fibrogenic lineage cells and in vivo re-construction of the developmental trajectory from single-cell profiling in early mouse embryos. (a) Three major cell clusters identified by their top marker genes expression containing cells with fibrogenic lineages. (b) Distribution of fibroblast marker genes and other common extracellular matrix (ECM) synthesis genes in the fibrogenic lineage cell clusters. (c) Expression and density of fibroblast marker and ECM synthesis genes across three clusters of the fibrogenic lineage. (d) Monocle 3 trajectory plot of fibrogenic cell lineage. Figure on top right showing the location of each cluster in pseudo-time trajectory. Figures on bottom showing individual lineage tree projected of each of the EMC and EFA cluster to CT cluster. (e) Distribution of cells according to their types and according to their developmental stages (E9.5, E11.5, and E13.5) in the pseudo developmental trajectory. (f) Expression of fibroblast marker genes Pdgfra in fibrogenic cell lineage clusters with respect to pseudo-time coordinates. (g) Relative expression of Tgfb2 and Prrx1 in pseudo-time trajectory. (h) Relative expression of extracellular matrix synthesis genes in fibrogenic cell lineage clusters with respect to pseudo-time coordinates.

Fig. 3

Maternal obesity enhances the number of cells with fibrogenic potential in early mouse embryo. (a) Integrated UMAP of three fibrogenic clusters showing relative distribution of cells in CON and MO groups. (b) Relative percentages of cells and their distribution in the UMAP from each of the three fibrogenic lineage clusters from CON and MO groups in the integrated data set of E9.5, E11.5 and E13.5. Numbers on the top right corner of each UMAP denote the cell population for the respective cluster. (c) Relative percentages and distribution of cells belong to each fibrogenic lineage cell clusters in the UMAP during three embryonic time points E9.5–E13.5 in CON and MO groups. The number in the right corner of each UMAP denote the percentage of that specific cell types in whole embryo and the line graph below represents the pattern of changes in cell number with the progression of embryonic days. CON, Control group; MO, Maternal Obesity group.

Fig. 4

Maternal obesity enhances the expression of fibrogenic and extracellular matrix (ECM) genes in early mouse embryo. (a) Volcano plot of differentially expressed genes in Maternal obesity (MO) group relative to control (CON) group in the fibrogenic cluster. Significance (adjusted P-value) was calculated in Seurat using the non-parametric Wilcoxon rank-sum test. (b) Violin plot showing relative expression of fibroblast marker genes Pdgfra and Pdgfrb in embryos from CON and MO groups in fibrogenic lineage clusters. (c) Violin plot showing relative expression of major extracellular matrix (ECM) synthesis genes, between CON and MO embryos. (d) Relative expression of Tgfb2 in CON and MO groups and in three different embryonic time points: E9.5, E11.5 and E13.5 fibrogenic lineage clusters. (e) Relative expression of Prrx1 embryos from CON and MO groups. (f) Adjust P-value of upregulated genes in MO group involved in TGFβ signalling and regulation of SMAD across all clusters. CON, Control group; MO, Maternal Obesity group. (See also Supplementary Table S7 and Supplementary Table S8).

Fig. 5

Maternal Obesity induces differential cell to cell communication and enhances Ligand-Receptor (L-R) activity of TGFβ signalling pathway in early embryos. (a) Cell–cell communication networks identify significant communications across all cell clusters. (b) Circle plot showing the number of differential interactions among three fibrogenic cell lineage clusters and other clusters due to MO. (c) Incoming signalling pathways in all cell clusters in the integrated dataset from embryonic time points E9.5, E11.5 and E13.5. Red box showing overall TGFβ activity. (d) Heatmap showing comparison of major signalling pathways between CON and MO groups. (e) Heatmap showing the differential role of TGFβ signalling pathways between CON and MO groups. (f) Bar chart showing contribution of each Ligand-Receptor pair involved in TGFβ signalling. (g) Receptors expression involved in TGFβ signalling pathways between CON and MO groups. CON, Control group; MO, Maternal Obesity group.

Fig. 6

MO enhances fibrogenesis via upregulation of PRRX1 and inhibition of AMPK in embryos at E 13.5. (a) Relative mRNA expression of fibroblast marker genes Pdgfra and Tgfb2, and common ECM synthesis genes Col1a1, Col3a1, Col5a2 and Col6a3, and Prrx1 in CON (n = 5 mice per group) and MO (n = 5 mice per group) groups. (b) Cropped Western Blots of pP65, P65, pSMAD3, SMAD3, and PRRX1 (β-Tubulin as loading control) from CON (n = 5 maternal mice per group) and MO (n = 5 maternal mice per group) embryos. (c) Cropped Western Blots of pAMPK, AMPK and TGFβ2 (β-Tubulin as loading control) from CON (n = 5 maternal mice per group) and MO (n = 5 maternal mice per group) embryos. (d) Breeding of PDGFRα^EGFP male mice with CON and MO female mice. (e) In vivo fluorescence image and measurement of fluorescence signal from CON and MO embryos at E13.5. (f) Sections (Left: paraxial mesoderm; right: adjacent area) from E13.5 CON and MO embryos visualizing PDGFRα+ fibrogenic cells distribution. CON: Control group; MO: Maternal Obesity group. Data are shown as mean ± SEM, and each dot represents one litter. ∗P < 0.05, ∗∗P < 0.01, and ∗∗∗P < 0.001 in CON versus MO by one-tailed unpaired Student's t test (a–e).

Fig. 7

Activation of AMPK suppresses fibrogenesis through inhibiting TGFβ signalling and Prrx1 expression. (a) Initiation of fibrogenesis in C3H/10T1/2 cells with TGFβ and treatment with AMPK activators metformin and A-769662. (b) Relative mRNA expression of fibroblast marker gene Pdgfra, common extracellular matrix (ECM) synthesis genes Col1a1, Col3a1 and Prrx1 and (c) Protein content of pAMPK, AMPK, pSMAD3, SMAD3 and PRRX1 in C3H/10T1/2 cell lines after incubation with 10 ng/ml TGFβ in presence and absence 500 μM AMPK activator MET (Metformin) and 50 μM of A76 (A769662) for 48 h. β-Tubulin as loading control. (d) Sirius red staining of collagen (visualized by red) in C3H/10T1/2 cells after incubation with 10 ng/ml TGFβ in the presence and absence of 500 μM AMPK activator MET (Metformin) and 50 μM of A76 (A769662) for 96 h. Scale bars, 400 μM. Data are presented as mean ± SEM, and each dot represents one independent experiment. ∗P < 0.05, ∗∗P < 0.01, and ∗∗∗P < 0.001 groups by one-way ANOVA (n = 3). (e) AMPK activation in MO mice by metformin (f) Relative mRNA expression of fibroblast marker genes Pdgfra, Prrx1, and Tgfb2, and common ECM synthesis genes Col1a1, Col1a2, Col3a1 and Col5a2 and (g) Cropped WB image showing protein content of pAMPK, AMPK, COL1A1, TGFβ2 and PRRX1 from CON (n = 5 maternal mice per group), MO (n = 5 maternal mice per group), and MO-MET (n = 5 maternal mice per group) embryos after treatment with metformin during pregnancy. β-Tubulin was used as loading control (Loading control used for COL1A1, TGFβ2 and PRRX1 was also used as loading control for Supplementary Fig. S7d). CON: Control groups; MO: Maternal Obesity groups; MO-MET: MO mice treated with metformin. Data are presented as mean ± SEM, and each dot represents one litter. ∗P < 0.05, ∗∗P < 0.01, and ∗∗∗P < 0.001 groups by one-way ANOVA (n = 5).

Fig. 8

AMPK KO enhances fibrogenesis in embryonic mesenchymal cells (EMCs). (a) Isolation of EMC cells from AMPKα1^fl/fl-CreER and AMPKα1^fl/fl embryos. (b) Relative mRNA expression of fibroblast marker gene Pdgfra and common extracellular matrix (ECM) synthesis genes Col1a1, Col3a1 and Prrx1, (c) Protein contents of AMPKα, pSMAD3/SMAD3 and PRRX1 in EMCs after incubation in presence and absence of 10 ng/ml TGFβ for 48 h (β-Tubulin was used as loading control). (d) Sirius red staining of collagen (visualized by red) after incubation in the presence and absence with 10 ng/ml TGFβ for 4 days (Scale bars, 200 μM). Sirius red was recovered, and absorbance was measured at 550 nm against a blank. Data are presented as mean ± SEM, and each dot represents one independent experiment. (e) Generation of AMPK-WT and AMPK knockout (KO) embryos. (f) Relative mRNA expression of fibroblast marker genes Pdgfra, Prrx1, and Tgfb2, and common ECM synthesis genes Col1a1, Col1a2, Col3a1, and Col5a2. (g) Cropped WB image showing protein contents of pAMPK, AMPK, COL1A1, TGFβ2, and PRRX1 from AMPK-WT and AMPK-KO embryos. Data are presented as mean ± SEM, and each dot represents one litter. ∗P < 0.05, ∗∗P < 0.01, and ∗∗∗P < 0.001 by one-way ANOVA.