GDF15 antagonism limits severe heart failure and prevents cardiac cachexia

Abstract

Aims: Heart failure and associated cachexia is an unresolved and important problem. This study aimed to determine the factors that contribute to cardiac cachexia in a new model of heart failure in mice that lack the integrated stress response (ISR) induced eIF2α phosphatase, PPP1R15A.

Methods and results: Mice were irradiated and reconstituted with bone marrow cells. Mice lacking functional PPP1R15A, exhibited dilated cardiomyopathy and severe weight loss following irradiation, whilst wild-type mice were unaffected. This was associated with increased expression of Gdf15 in the heart and increased levels of GDF15 in circulation. We provide evidence that the blockade of GDF15 activity prevents cachexia and slows the progression of heart failure. We also show the relevance of GDF15 to lean mass and protein intake in patients with heart failure.

Conclusion: Our data suggest that cardiac stress mediates a GDF15-dependent pathway that drives weight loss and worsens cardiac function. Blockade of GDF15 could constitute a novel therapeutic option to limit cardiac cachexia and improve clinical outcomes in patients with severe systolic heart failure.

Keywords: Cachexia; GDF15; Heart failure; Integrated stress response; PPP1R15A.

Figure 1

Mice lacking functional PPP1R15A exhibit severe heart failure and weight loss following whole-body irradiation. Mice lacking functional PPP1R15A (^ΔC/ΔC), male n = 5, or female n = 10, or wild type female (^+/+) n = 10 and male (^+/+) n = 10 controls were subjected to whole body irradiation (11 Gy) then bone marrow (BM) transfer and monitored over a time course (weeks 1–12) by (A) echocardiography for left ventricular fractional shortening (LVFS%) and (B) body weight (% of body weight compared to day 0 of irradiation). The number of mice decreased over time due to euthanasia due to reaching humane endpoints. Male (^ΔC/ΔC) and (^+/+) mice are littermates. Female mice (^ΔC/ΔC) n =10, shown from combined experiments, are not littermates. P values of data show comparison between genotypes at identical time points calculated by two-way ANOVA with Sidaks correction for multiple comparisons. Male and female mice were calculated independently (C) TD-NMR analysis of body composition of lean and fat mass using an independent cohort of mice using two-tailed unpaired t test (D) Non-irradiated and irradiated Ppp1r15a^+/ΔC littermates heart function (LVFS%) up to 22 weeks. Irradiated Ppp1r15a^+/+ and Ppp1r15a^ΔC/ΔC male mice reconstituted with either Ppp1r15a^+/+ or Ppp1r15a^ΔC/ΔC BM, and show (E) LVFS% and (F) body weight at 9 weeks post-irradiation. e,f using two-way ANOVA with Tukey’s correction. **P < 0.01, ***P < 0.001, ****P < 0.0001.

Figure 2

Heart failure is associated with local inflammation and activation of the integrated stress response. (A) Ppp1r15a transcript levels assessed by Single Molecule In Situ Hybridisation (SM-ISH) in heart tissue sections derived from non-irradiated, or irradiated female WT mice, at 13,5 and 7 weeks post-irradiation. H-score of Ppp1r15a expression was analysed by ANOVA and Dunnett’s post-test. (B) Analysis of tissue from Ppp1r15a^+/+ (^+/+) or Ppp1r15a^ΔC/ΔC(^ΔC/ΔC) male mice (n = 3 mice/group), 7 weeks post-irradiation with H&E staining (scale bar 200 μm). (C) Heatmap of top DEGs related to the selected gene ontology pathways. Comparison between week 4,7 post-irradiation Ppp1r15a^ΔC/ΔC (KO) and Ppp1r15a^+/+ (WT) log2FoldChanges. The Black cell colour represents genes not identified in the corresponding DEGs analysis. (D) Breakdown of immune system cell types found in none and post-irradiation of Ppp1r15a^+/+ (^+/+) or Ppp1r15a^ΔC/ΔC(^ΔC/ΔC) obtained through RNA-seq sample deconvolution. The number following each group name indicates the days post-irradiation (E) RT-qPCR for inflammatory and ISR gene expression in heart (^+/+n = 6, ^ΔC/ΔC n = 5), P values were determined by two-tailed Student’s t-test. Heart (H), quadricep (Q), kidney (K), and liver (L) tissue taken from male mice at 7 weeks from post-irradiation and analysed for expression for (F) Ddit3 and (G) Il6 (n = 3). Tissue taken from female mice at indicated time post-irradiation analysed for (H) Ddit3 (Heart, Lung, Liver. Kidney) and (I) Il6 gene expression (heart and lung). 7–9 week samples taken at the humane endpoint. Data analysed by two-way ANOVA and Tukey’s post-comparison All data expressed as mean +/-SD. *P < 0.05 **P < 0.01, ***P < 0.001, ****P < 0.0001.

Figure 3

Absence of functional PPP1R15A results in increased expression of Gdf15 following whole-body irradiation. Ppp1r15a^ΔC/ΔC(^ΔC/ΔC) and Ppp1r15a^+/+ (^+/+) mice were irradiated (11 Gy) and reconstituted with Ppp1r15a^+/+ bone marrow. (A) GDF15 in plasma from non-irradiated mice, and at 2 and 9 weeks post-irradiation. (B) Gdf15 expression, by RT-qPCR, in heart, lung, liver and kidney taken at 7 weeks post-irradiation. P values by two-way ANOVA with Tukey’s correction. Gdf15 expression in hearts, assessed by SM-ISH. (C) Representative spatial plots, (D) H-score (E) and the distribution of cells expressing Gdf15 transcripts, n = 3 hearts/group. % cells and H-score are mean ± SD (P values by students t test). ***P < 0.001, ****P < 0.0001.

Figure 4

The absence of Ddit3 does not prevent irradiation-induced heart failure or GDF15 expression. Ppp1r15a^ΔC/ΔC(^ΔC/ΔC) Ddit3^+/−, or Ppp1r15a^ΔC/ΔC(^ΔC/ΔC) Ddit3^−/− mice, n = 10 per group were irradiated (11 Gy) and reconstituted with Ppp1r15a^+/+ BM. Mice were culled at 8.5 weeks post-irradiation or earlier due to humane endpoints. % of body weight of (A) male and (B) female mice. (C) LV function (LVFS %) assessed by echocardiography. (D) GDF15 analysis of the plasma and comparison with irradiated Ppp1r15a^ΔC/ΔC(^ΔC/ΔC) Ddit3^+/+ mice. Gdf15 and Ddit3 expression in heart tissue assessed by SM-ISH and distribution was shown in (E) spatial plots and (F, G) H-score of cells expressing GDF15 and Ddit3 transcripts. H-score was analysed by ANOVA and Dunnett’s post-test. (H) Normalized gene expression from RNAseq analysis for P values determined by two-way ANOVA. Data show mean ± SD. **P < 0.01, ***P < 0.001, ****P < 0.0001.

Figure 5

GDF15 is required for weight-loss and severe heart failure in irradiated mice lacking functional PPP1R15A. Ppp1r15a^ΔC/ΔC or Ppp1r15a^+/+male and female mice were irradiated (11 Gy) and reconstituted with Ppp1r15a^+/+ BM. At 4 weeks (open arrow), Ppp1r15a^ΔC/ΔC mice were randomly assigned and given an antibody that blocks GDF15 activity (αGDF15), n = 11 or an isotype control antibody (Isotype Ab), n = 10, at 10 mg/kg, at 3-day intervals. Mice were culled at 9.5 weeks post-irradiation or earlier due to humane end points. (A) Body weights, % of body weight at 4 weeks post-irradiation. P values were determined by two-way ANOVA with Sidaks correction for multiple comparisons, P values indicated by asterisks, show comparison between treatment groups at identical time points. (B) Relative weight of fat pads, female mice indicated by half-filled symbols. P values were calculated by two-tailed Student’s t-test. (C) Survival plot, no mice were found dead but mice that approached humane endpoints were euthanized. Data analysed using Gehan-Breslow-Wilcoxon Test. (D) LV function (LVFS %) was assessed by echocardiography at 9 weeks post-irradiation. Data were analysed using ANOVA and Tukey’s post comparison. (E) Correlation of LV end diameter at diastole (LVEDd) with % body weight at 9 weeks post irradiation, r calculated by Pearson’s correlation coefficient. (F) Analysis of fibrosis in heart tissue sections. Blood taken at 9.5 weeks post-irradiation (or earlier due to humane endpoints) and plasma analysed for (G)) troponin and (H–L) metabolic parameters as indicated. Data (F–H, I, K, L) analysed using one way ANOVA and Tukey test post comparison. Data in (J) analysed using Kruskal-Wallis test with Dunn’s post test. All data show mean ± SD. *P < 0.05 **P < 0.01, ***P < 0.001, ****P < 0.0001.

Figure 6

GDF15 correlates with parameters of cachexia in patients with heart failure. Analysis of the BIOSTAT-CHF cohort of patients with heart failure, the correlation between Log-transformed plasma GDF-15 and (A) log-transformed estimated protein intake n = 2112 and (B), log-transformed urinary creatinine. n = 2143. (C) Dot plots of plasma GDF15 were compared in patient groups according to whether they fit the criteria for cachexia. No n = 2153, or Yes n = 90. Data analysed using Mann–Whitney U test *P < 0.05. Medians are shown.