Compartmentalized metabolism supports midgestation mammalian development

Abstract

Mammalian embryogenesis requires rapid growth and proper metabolic regulation¹. Midgestation features increasing oxygen and nutrient availability concomitant with fetal organ development^2,3. Understanding how metabolism supports development requires approaches to observe metabolism directly in model organisms in utero. Here we used isotope tracing and metabolomics to identify evolving metabolic programmes in the placenta and embryo during midgestation in mice. These tissues differ metabolically throughout midgestation, but we pinpointed gestational days (GD) 10.5-11.5 as a transition period for both placenta and embryo. Isotope tracing revealed differences in carbohydrate metabolism between the tissues and rapid glucose-dependent purine synthesis, especially in the embryo. Glucose's contribution to the tricarboxylic acid (TCA) cycle rises throughout midgestation in the embryo but not in the placenta. By GD12.5, compartmentalized metabolic programmes are apparent within the embryo, including different nutrient contributions to the TCA cycle in different organs. To contextualize developmental anomalies associated with Mendelian metabolic defects, we analysed mice deficient in LIPT1, the enzyme that activates 2-ketoacid dehydrogenases related to the TCA cycle^4,5. LIPT1 deficiency suppresses TCA cycle metabolism during the GD10.5-GD11.5 transition, perturbs brain, heart and erythrocyte development and leads to embryonic demise by GD11.5. These data document individualized metabolic programmes in developing organs in utero.

Fig. 1. Metabolic transition at GD10.5–GD11.5.

a, Midgestation is a dynamic period of development. b, Tissue weights from pregnant dams, aged 13.6 ± 3.8 weeks. c, Group average heat map of metabolomics data. d, Metabolites with P < 0.05 and fold change (FC) > 1.2 or < 0.8 between GD10.5 and GD11.5. e, Heat map of purines and pyrimidines, plotted as fold change relative to GD10.5. Statistical tests: straight-line least-squares fitting followed by the extra sum-of-squares F-test (b); Student’s t-tests (d). Data are mean ± s.d. Statistical tests were two-sided. Guanidine Ac, guanidine acetate; IMP, inosine monophosphate (additional abbreviations, Supplementary Table 1). Source data

Fig. 2. Carbohydrate metabolism in midgestation.

a, Serial caesarian-section procedure. b, c, Time-dependent enrichment of [U-¹³C]glucose (b) and [U-¹³C]glutamine (c). d, Major glucose-6-phosphate isotopologues during serial caesarian-section infusion. e, Total enrichment (1 − unlabelled) of purines from [U-¹³C]glucose. f, M+1 enrichment in purines from [γ-¹⁵N]glutamine. ¹⁵N-glutamine enrichments are normalized to glutamine m+1 to account for differences among compartments (see Fig. 2c). Statistical tests: plateau followed by one-phase decay least-squares fitting followed by the Holm–Sidak’s multiple-comparisons adjustment (b–d) (b: embryo vs placenta P = 0.09, embryo vs blood P = 0.0001, placenta vs blood P = 0.003; c: embryo vs placenta P < 0.0001, embryo vs blood P < 0.0001, placenta vs blood P < 0.0001); paired t-tests or Wilcoxon matched-pairs signed-rank tests followed by Holm–Sidak’s multiple-comparisons adjustment (e); log₂ paired t-tests followed by Holm–Sidak’s multiple-comparisons adjustment (f). Data are mean ± s.d. Statistical tests were two-sided. p/e, placenta/embryo. Source data

Fig. 3. Evolving labelling during midgestation.

a, Enrichments normalized to glucose m+6. b, Labelling from [U-¹³C]glucose between GD9.5 and GD12.5. c, d, Organ-specific enrichments at GD12.5. Statistical tests: paired t-tests followed by Holm–Sidak’s multiple-comparisons adjustment (a); Kruskal-Wallis test followed by Dunn’s multiple-comparisons adjustment or one-way ANOVA followed by Tukey’s multiple-comparisons adjustment (b); linear mixed-effects analysis followed by Holm-Sidak’s multiple-comparisons adjustment (c, between-tissue comparisons); or Welch’s one-way ANOVA followed by the Dunnett’s T3 multiple-comparisons adjustment or Kruskal–Wallis test followed by the Dunn’s multiple comparisons adjustment (d). Data are mean ± s.d. Statistical tests were two-sided. Asp, aspartate; Cit, citrate; Glc, glucose; Lac, lactate; Mal, malate; Pyr, pyruvate; Suc, succinate. Source data

Fig. 4. Lipt1 deficiency impairs embryo metabolism, growth and erythropoiesis.

a, Relevant metabolites in embryos of the indicated genotypes. b, c, Labelling from [U-¹³C]glucose. d, Endothelial cells in the head stained for PECAM1 and endomucin (1 and 1′; scale bar, 200 μm), and endothelial cells in the heart stained for PECAM1 and endomucin (magenta) and connexin 40 (cyan) (2 and 2′; scale bar, 100 μm). Images are representative of n = 3 dams, and the following numbers of embryos: Lipt1^WT/WT (n = 5), Lipt1^WT/N44S (n = 4), Lipt1^N44S/N44S (n = 8). e, Quantification of cells from dissociated GD10.5 whole embryos stained with antibodies against CD71, TER119, CD41 and c-Kit. Erythrocytes express CD71 and TER119, and myeloid–erythroid progenitors (MEP) express CD41 and c-Kit. '% parent' indicates the proportion of CD71^–TER119^– cells that stained CD41⁺c-Kit⁺. f, Longitudinal red blood cell (RBC) measurements in a LIPT1-deficient patient. Statistical tests: Student’s t-tests (a); log₂-transformation followed by Holm-Sidak’s multiple-comparisons adjustment (b); Mann–Whitney tests followed by Holm-Sidak’s multiple-comparisons adjustment (e); Kruskal–Wallis tests followed by the Dunn’s multiple-comparisons adjustment (c). Data are ± s.d. Statistical tests were two-sided. α-KG, α-ketoglutarate; Aco, aconitase; Fum, fumarate; Lys, lysine; KIV, a-ketoisovalerate; KIC, α-ketoisocaproate; KMV, α-keto-β-methylvalerate. Source data

Extended Data Fig. 1. Dynamic metabolism in developmental tissues during midgestation.

(a) Principal component analysis and (b) Dendrogram analysis generated using Metaboanalyst 5.0 (https://www.metaboanalyst.ca/).

Extended Data Fig. 2. Tissue specific metabolic changes from gd10.5-gd11.5.

(a) Heatmap of top 50 metabolites significantly different between embryo and placenta. (b) Top 25 metabolites that contribute to the separation of gd10.5 embryo and placenta metabolic profiles. (c-d) Metabolic set overrepresentation analysis using metabolites in embryos (c) and placentas (d) differing in abundance (p < 0.05) between gd10.5-gd11.5. (e) Metabolites associated with the urea cycle transiently increase at gd11.5 and decrease the following two days. Statistical significance was determined using Student’s t-tests. All data represent mean ± s.d. Statistical tests were two-sided.

Extended Data Fig. 3. Glucose feeds purine synthesis in developmental tissues.

(a) Possible pathways contributing to labeling from [U-¹³C]glucose. (b-f) Isotopologues of G6P (b), R5P (c), IMP (d), GMP (e) and AMP (f) labeled with [U-¹³C]glucose. (g) Total enrichment (1-unlabeled fraction) of serine and glycine relative to total glucose enrichment. (h) Total enrichment in R5P (isobar with Ri5P/X5P), UMP and CMP labeled with [U-¹³C]glucose. (i) UMP and CMP m+1 enrichment from [γ-¹⁵N]glutamine. ¹⁵N-glutamine enrichment is normalized to glutamine m+1 to account for differences among compartments (see Fig. 2c). Statistical significance was determined using paired t-tests (b, c, h, and i) or Wilcoxon matched-pairs signed rank tests (b-g) followed by the Holm-Sidak’s multiple comparisons adjustment (b-i). All data represent mean ± s.d. Statistical tests were two-sided. Source data

Extended Data Fig. 4. Tissue-specific TCA cycle metabolism in midgestation.

(a) Labeling from [U-¹³C]glucose between gd9.5-gd12.5 in placenta. (b) Enrichments normalized to glucose m+6 on gd12.5. (c-d) Daily citrate m+2/pyruvate m+3 (c) and citrate m+3/pyruvate m+3 (d) enrichment ratios. (e) ETC-related transcript counts (n = 2) in fetal tissues normalized to gd10.5 (gd11.5 in liver). (f) Daily total citrate/pyruvate enrichment ratio in fetal heart. (g) Placental ETC complex gene expression. (h) Enrichment ratio of glutamate m+5/glutamine m+5 in fetal tissues infused with [U-¹³C]glutamine. Statistical significance was determined using Mann-Whitney tests (a, g), paired t-tests (b), or straight line least squares fitting (h) followed by the Holm-Sidak’s multiple comparisons adjustment (a, b, g, and h), linear mixed-effects analysis (c,d) followed by the Sidak’s (c-d; between-tissue comparisons) or Tukey’s (c-d; between-time comparisons) multiple comparisons adjustment, or one-way ANOVA followed by the Tukey’s multiple comparisons adjustment (f). All data represent mean ± s.d. Statistical tests were two-sided. Source data

Extended Data Fig. 5. LIPT1 activity is critical for transition from gd10.5-gd11.5.

(a) PCA plots of metabolomics data and (b) tissue weights in litters arising from Lipt1^WT/N44S intercrosses. (c) Relevant metabolites in placentas of the indicated genotypes. (d-e) Placental uptake (left) and embryo transfer (right) of [U-¹³C]glucose (d) and [U-¹³C]glutamine (e). (f) Expression of placental markers. (g) Labeling from [U-¹³C]glucose in placentas of various Lipt1 genotypes. Statistical significance was determined using two-way repeated measures ANOVA followed by the Sidak’s multiple comparisons adjustment (b), Student’s t-tests (c) or Mann-Whitney tests (d-g) followed by Holm-Sidak’s multiple comparisons adjustment. All data represent mean ± s.d. Statistical tests were two-sided. Source data

Extended Data Fig. 6. LIPT1 deficiency hinders organogenesis and erythropoiesis.

(a) Somite counts in embryos from litters arising from Lipt1^WT/N44S intercrosses at gd9.5. (b) Brightfield whole mount images, scale bar = 500μm (1,1’); Dorsal Aortae stained with Connexin 40, scale bar = 100μm (2,2’); PE staining of whole hearts, scale bar = 300μm (3,3’); H&E staining of hearts, scale bar = 50μM (4,4’). All images from gd9.5 embryos. (c-d) Gd10.5 whole embryo cells stained with antibodies against the erythroid lineage markers CD71 and TER119 (c) and myeloid/erythroid progenitor markers, cKIT and CD41 (d). Flow cytometry was performed in 24 individual embryos (Healthy n = 11, Mutant n = 7) White blood cell (WBC) (e) and platelet (f) counts from a LIPT1-deficient patient. Statistical significance was determined using Student’s t-tests followed by Holm-Sidak’s multiple comparisons adjustment. All data represent mean ± s.d. Statistical tests were two-sided. Source data

Extended Data Fig. 7. Flow cytometry gating strategies.

(a) Single cell suspensions from whole embryos were gated by forward and side scatter area (P1) then by forward scatter height and width (P2), then by side scatter height and width (P3). Cells had been stained with DAPI and live cells were gated as DAPI negative (P4), and then by CD117 (cKIT) negative (P5). Fetal erythrocytes were identified as CD71⁺/TER119⁺ (red box – P5:Q2), and myeloid/erythroid progenitors were gated as CD71^-/TER119^- (blue box - P5:Q3) and also gated as cKIT⁺/CD41⁺ (P6).