Multi-organ transcriptome atlas of a mouse model of relative energy deficiency in sport

Abstract

Insufficient energy intake to meet energy expenditure demands of physical activity can result in systemic neuroendocrine and metabolic abnormalities in activity-dependent anorexia and relative energy deficiency in sport (REDs). REDs affects >40% of athletes, yet the lack of underlying molecular changes has been a hurdle to have a better understanding of REDs and its treatment. To assess the molecular changes in response to energy deficiency, we implemented the "exercise-for-food" paradigm, in which food reward size is determined by wheel-running activity. By using this paradigm, we replicated several aspects of REDs in female and male mice with high physical activity and gradually reduced food intake, which results in weight loss, compromised bone health, organ-specific mass changes, and altered rest-activity patterns. By integrating transcriptomics of 19 different organs, we provide a comprehensive dataset that will guide future understanding of REDs and may provide important implications for metabolic health and (athletic) performance.

Keywords: REDs; energy intake; exercise-for-food; health and performance; mouse model; negative energy balance; physical activity; relative energy deficiency in sport; transcriptome.

Figure 1.. Physiology and behavior of relative energy deficiency in a mouse model

(A) Experimental set-up and design. (B) List of the 20 tissues collected. Red arrows indicate the timing of tissue collection (ZT5 and ZT17). The shaded area indicates the dark phase (ZT12-ZT24). Daily changes in (C) workload (m/J), (D) cumulative running-wheel activity, (E) cumulative energy intake, and (F) body weight in control (open symbols) and energy-deficient (closed symbols) female (red) and male (green) mice (n = 20/group). The dashed lines indicate when female (red) and male (green) mice significantly lost body weight. (G) Average activity profiles are depicted as 3-hour running averages ±SD for the last 5 days of the protocol (n = 20/group). (H) Average number of active bins, number of active bouts, and fragmentation (active bouts/active bins x 100) over the last 5 days of the protocol. Body composition measured by echoMRI and fat mass (% of body weight) after 3 weeks of intervention in (I) female and (J) male mice (n = 8/group). Fresh organ mass relative to body mass (%) is depicted for (K) females (red) and (L) males (green) (n = 20/group). One-way ANOVA was used to analyze the effect of an energy deficit on different physiological parameters between experimental groups. All data are presented as mean ± SEM, with *p < 0.05, ^**p < 0.01, ^***p < 0.001. See also Figure S1 and S2.

Figure 2.. Effects of energy deficiency on the estrous cycle and bone morphology

(A) Images of vaginal smear cytology stained with crystal violet staining reflecting the different stages of the estrous cycle: proestrus (primarily nucleated epithelial cells), estrus (primarily cornified epithelial cells), metestrus (50% cornified epithelial cells and leukocytes), and diestrus (primarily leukocytes). (B) Frequency plots for identified estrous stages during the last 7 days in control (top) and energy-deficient mice (bottom) (n = 10–11/group). (C) Examples of estrous cycle plots for individual control (top) and deficient mice (bottom). (D) Correlation between frequency entering metestrus and weight loss. (E) Bar graphs showing estrous cycle length and maximal consecutive days in one stage in control (white) and energy-deficient mice (red). (F) Trabecular and cortical bone morphology were analyzed in the regions 350 – 850 μm and 1500 – 2500 μm distal to the proximal GP, respectively. The region of analysis is indicated by the dashed yellow box. Trabecular bone in the (I,i) sagittal, (II, ii) transverse, and (III,iii) coronal view of control (I-III) and energy-deficient (i-iii) female tibia ~0.5 mm distal to the proximal growth plate. (G) Bar graphs showing trabecular bone mineral density and (H) bone volume / total volume (BV/VT) stage in control (white) and energy-deficient mice (red) (n = 5–6/group). (I) Significant correlations between trabecular parameters and weight change. (J) Definitions of morphologic parameters and p-values and R² of correlations with weight change. One-way ANOVA was used to analyze the effect of an energy deficit on different parameters between experimental groups. all data are presented as mean ± SEM, with *p < 0.05, ^**p < 0.01, ^***p < 0.001. Linear regression models were fit by using the lm function in R to assess the correlation between different parameters. See also Figure S2.

Figure 3.. Energy deficiency affects gene expression across endocrine and metabolic organs

(A) Number of differentially expressed (DE) genes from time-independent analyses for all tissues in female and male mice. Genes downregulated in energy-deficient mice are shown in blue and upregulated are shown in red. (B) Distribution of DE genes and the number of tissues in which they are downregulated or upregulated in females (top) and males (bottom). Normalized expression for a selection of (C) downregulated genes and (D) upregulated genes in control (open symbols) and energy-deficient (closed symbols) females (red) and males (green) across tissues. Connecting dots represent the same tissues. See also Figure S3, Tables S2–S3.

Figure 4.. HPA-axis and HPG-axis gene expression changes in energy-deficient mice

Normalized expression for a selection of candidate genes involved in (A) hypothalamic (PVN; paraventricular nuclei, POA; preoptic area, PVZ; periventricular zone) from a previously published dataset, (B) pituitary, (C) adrenal/gonadal-axis signaling. (D) Energy deficit induced log2 fold changes (log2FC±SE) of receptors in target organs in control (open symbols) and energy-deficient (closed symbols) females (red) and males (green). Data from all time points were combined in this analysis. All data are presented as mean ± SEM, with *p < 0.05, ^**p < 0.01, ^***p < 0.001. See also Figure S3 and Table S2.

Figure 5.. Energy deficiency affects gene expression in multiple muscles and tendons

(A) Number of differentially expressed (DE) genes from time-independent analyses in energy deficiency for all tissues in females and males. Female DE genes are shown in red, male DE genes in green, and overlapping genes between females and males in grey. Distribution of DE genes and the number of muscles and tendons in which they are (B) downregulated (blue) or (C) upregulated (red). (D) Principal component analysis (PCA) was performed on all muscle and tendon tissues. Different colors represent tissue type, open symbols represent control, closed symbols represent energy-deficient, triangles represent females and circles represent males. Bar graph of common Metascape-annotated GO biological processes pathways for DE genes in (E) >1 muscle or (G) >1 tendon. Energy deficit induced log fold changes (log2FC±SE) for a selection of genes in females (red) and males (green) across (F) muscles and (H) tendons. All data are presented as mean ± SEM. See also Figures S4, S5, S6 and Tables S2–3.

Figure 6.. The transcriptome of the uterus, ovaries, and male kidneys are most affected by energy deficiency.

Volcano plots of differentially expressed (DE) genes from time-independent analyses in the (A) uterus, (D) ovaries, and (G) male kidney. Downregulated genes in energy-deficient mice (blue), upregulated genes (red). Bar graphs of the top 8 GO biological processes for DE genes in (B) uterus, (E) ovaries, and (H) male kidney. The number of downregulated genes (blue) and upregulated genes (red) for each GO term are depicted in bar graphs. Heatmaps showing log2FoldChanges for a subset of genes for each GO term. Energy deficit induced changes for a subset of genes in (C) uterus, (F) ovaries, and (I) male kidney. All data are presented as mean ± SEM, with *p < 0.05, ^**p < 0.01, ^***p < 0.001. See also Table S2–3.

Figure 7.. Plasma proteomics and metabolomics in energy-deficient mice.

Volcano plots of differentially abundant proteins in (A) female and (C) male plasma. Proteins downregulated in energy-deficient mice (blue), upregulated (red). Top 10 differentially expressed plasma proteins are shown for (B) females and (D) males. (E) Examples of other plasma proteins that changed in response to energy deficiency. (F) Summary of metabolite changes in plasma. The number of changed metabolites between any two of the 4 different conditions is shown in the right panel. (G) Plasma corticosterone levels. (H) Plasma triglyceride and (I) cholesterol levels. Statistics: (B, D, E, F, G) Unpaired t test. (H, I) Two-way ANOVA (factors: group and ZT) and Tukey’s multiple comparisons test. Data are presented as mean ± SEM, with *p < 0.05, ^**p < 0.01, ^***p < 0.001. See also Figure S7 and Table S6–7.