Acute stress induces long-term metabolic, functional, and structural remodeling of the heart

Abstract

Takotsubo cardiomyopathy is a stress-induced cardiovascular disease with symptoms comparable to those of an acute coronary syndrome but without coronary obstruction. Takotsubo was initially considered spontaneously reversible, but epidemiological studies revealed significant long-term morbidity and mortality, the reason for which is unknown. Here, we show in a female rodent model that a single pharmacological challenge creates a stress-induced cardiomyopathy similar to Takotsubo. The acute response involves changes in blood and tissue biomarkers and in cardiac in vivo imaging acquired with ultrasound, magnetic resonance and positron emission tomography. Longitudinal follow up using in vivo imaging, histochemistry, protein and proteomics analyses evidences a continued metabolic reprogramming of the heart towards metabolic malfunction, eventually leading to irreversible damage in cardiac function and structure. The results combat the supposed reversibility of Takotsubo, point to dysregulation of glucose metabolic pathways as a main cause of long-term cardiac disease and support early therapeutic management of Takotsubo.

Fig. 1. Cardiac strains in LV, RV, LA, and RA at baseline and 2 h, 7 d, 1 mo, and 3 mo post ISO.

Boxplots showing median, 25 and 75 percentiles, and extremes of values of n = 4 to 6 animals per time point. a Longitudinal strain is increased in the LV at 2 h and returns to normal in the remaining time-points (0 vs. 2 h: p = 0.0428, 2 h vs. 7d: p = 0.0379, 2 h vs. 1mo: p = 0.0175, and 2 h vs. 3mo: p = 0.0385). In contrast, longitudinal strain in the RV decreases at 2 h and then returns progressively to baseline values (0 vs. 2 h: p = 0.0002). b Longitudinal and circumferential strains in the apex and in the basal region. Hyperkinesia, the normal reaction to a catecholaminergic surge is observed at 2 h in the basal region highlighted by the increase in the basal longitudinal (0 vs. 2 h: p = 0.0081) and circumferential (0 vs. 2 h: p = 0.0107, and 2 h vs. 3mo: p = 0.0185) strains, while at the same time no substantial changes were found in the longitudinal and circumferential strains in the apex. c LA and RA strains and LA surface area. LA strain decreases progressively after the initial rise at 2 h, suggesting an irreversible LA dysfunction (0 vs. 3mo: p = 0.0098, 2 h vs. 7d: p = 0.0038, 2 h vs. 3mo: p < 0.0001, and 1mo vs. 3mo: p = 0.0203). It goes along with the increase of LA minimum (0 vs. 2 h: p = 0.0447, and 2 h vs. 3mo: p = 0.0357) and maximum (2 h vs. 7d: p = 0.0038 and 2 h vs. 3mo: p = 0.0031) surface areas over time. The strain that corresponds to the reservoir function decreases in the RA at 2 h (2 h vs. 1mo: p = 0.0310, and 2 h vs. 3mo: p = 0.0376). Unpaired comparison tests: *p < 0.05, **p < 0.01 and ***p < 0.001. All values are individually normalized to the baseline value for each animal. Statistical significance (p < 0.05) for each variable was estimated by one-way or two-way ANOVA when group variances were equal (Bartlett test); if not the non-parametric Kruskall–Wallis test, and the Holm multiple comparisons test was used to execute simultaneous t-tests. Source data are provided as a « SourceData_Figure1 » file.

Fig. 2. Progression of cardiac fibrosis after ISO stress.

a–d Representative section of Sirius red staining in the LV apex and base, right atrium and left atrium of the heart, at baseline (scale bars of a length of 100 μm) and 2 h (scale bars of a length of 500 μm in LV images and of 100 μm in LA and RA images), 7d (scale bars of a length of 100 μm) and 1-month (scale bars of a length of 500 μm in LV apical and 250 μm in LV basal images, and of 100 μm in LA and RA images,) and 3-months (scale bars of a length of 500 μm in LV images and of 100 μm in LA and RA images) post-ISO. e–h Quantitative analysis of staining using FIBER-ML in the LV apex and base, right atrium and left atrium of the heart at the corresponding post-ISO time points for the indicated number of animals represented as boxplots showing median, 25 and 75 percentiles, and extremes of values. The diffuse fibrosis that first appears in the LV apex at 7d post-ISO augments and extends into the LV base at 1 and 3 months (in the apex, 0 vs. 7d: p = 0.0353, 0 vs. 1mo: p = 0.0011, 0 vs. 3mo: p = 0.0001, 2 h vs. 1mo: p = 0.0073, and 2 h vs. 3mo: p = 0.0012; in the base, 0 vs. 1mo: p = 0.0059, 0 vs. 3mo: p = 0.0015, 2 h vs. 1mo: p = 0.0038, 2 h vs. 3mo: p = 0.0009, 7d vs. 1mo: p = 0.0043, and 7d vs. 3mo: p = 0.0005); unpaired comparison tests: *p < 0.05, **p < 0.01 and ***p < 0.001. Statistical significance (p < 0.05) for each variable was estimated by one-way or two-way ANOVA when group variances were equal (Bartlett test); if not the non-parametric Kruskall–Wallis test, and the Holm multiple comparisons test was used to execute simultaneous t-tests. Diffuse fibrosis is also apparent in the LA at 1- and 3-months post-ISO: ANOVA test, p = 0.0294. Source data are provided as a « SourceData_Figure5 » file.

Fig. 3. Longitudinal FDG PET before, during and after ISO stress.

a Representative images of FDG PET registered UUI (PETRUS) at the indicated time points post-ISO from diastolic phase to systolic phase. Images acquired 30 min after FDG injection of one section along the long axis of the LV. Color scale depicts the Standard Uptake Value (SUV) from 0 to 10. Note the increase in FDG uptake at 2 h and 7d respective to baseline. b Data are presented as mean values +/− SD of n = 9 animals per time point. Quantitative analysis of FDG uptake in the LV: SUV mean in whole LV, apex (7d vs. 1mo: p = 0.0337) and basal LV, normalized to baseline values; rate constant K1 reflecting the exchange of FDG from blood to tissue; rate constant k3 reflecting FDG phosphorylation, calculated using two-compartmental analysis. Note the global increase of SUV at 2 h and 7d post-ISO, the modest increase of K1 that is not statistically significant from baseline at any time point, and the significant decrease in k3 at 2 h followed by an increase at 7d post-ISO (in the apex, 2 h vs. 7d: p = 0.0004, 7d vs. 1mo: p = 0.0005, and 7d vs. 3mo: p = 0.0029; in the base, 0 vs. 2 h: p = 0.0312, 2 h vs. 7d: p = 0.0096, and 7d vs. 1mo: p = 0.0312). Paired comparison tests: *p < 0.05, **p < 0.01 and ***p < 0.001. Statistical significance (p < 0.05) for each variable was estimated by one-way or two-way ANOVA when group variances were equal (Bartlett test); if not the non-parametric Kruskall–Wallis test, and the Holm multiple comparisons test was used to execute simultaneous t-tests. Source data are provided as a « SourceData_Figure2 » file.

Fig. 4. Immunohistochemistry of Glut1, Glut4 and Cd68 expression at baseline, 2 h and 7d post-ISO.

a–f Representative fields of view of 4 µm cardiac sections stained for Glut1, Glut4 and Cd68. a–c apex, d–f base; the three stainings were performed in each section (OPAL® technology). g–l Quantification of staining densities (medians, 25 and 75 percentiles, and extremes of values of the percentage of section surface staining for each protein) of 4 to 7 animals for Glut1, Glut4 and Cd68. g, h apex, j-l: base. Note the significant increase in Glut1 expression at 2 h post-ISO followed by a return to baseline values in the apex (0 vs. 2 h: p = 0.0307) and the base (0 vs. 2 h: p = 0.0020, and 2 h vs. 7d: p = 0.0050). In contrast, Glut4 expression is unchanged from baseline at 2 h post-ISO but increases at 7d in the apex (0 vs. 7d: p = 0.0196, and 2 h vs. 7d: p = 0.0362). Disperse Cd68+ inflammatory cells are observed at 2 h (apex + basal, in the base: 0 vs. 2 h: p = 0.0063) and 7d (apex). Note that Glut1 and Glut4 stain essentially myocardial cells. Unpaired comparison tests: *p < 0.05, **p < 0.01 and ***p < 0.001. Statistical significance (p < 0.05) for each variable was estimated by one-way or two-way ANOVA when group variances were equal (Bartlett test); if not the non-parametric Kruskall–Wallis test, and the Holm multiple comparisons test was used to execute simultaneous t-tests. Indicated scale bars in the images correspond to a length of 40 μm. Source data are provided as a « SourceData_Figure3 » file.

Fig. 5. Expression of major proteins of the different pathways of glucose metabolism in the heart apex.

a Hierarchical clustering with all proteins of interest involved in different pathways of glucose metabolism in the heart apex, based on Pearson correlation on z-scored quantification values. Heatmap indicating proteins involved in glycolysis, glycogenesis, gluconeogenesis, oxidative phosphorylation, hexosamine biosynthetic pathway, and polyol pathway in the LV apical segment. Colors depict the z-scored changes in protein expression at 2 h and 7d post-ISO respective to baseline expression levels: red, overexpression higher than 1.3-fold, green: under-expression lower than −1.3-fold. All values are statistically significant with p < 0.05 using two-tailed Student’s t-test, n = 5 per time point. b Western blot analysis of Gfat1 (0 vs. 7d: p = 0.0168, and 2 h vs. 7d: p = 0.0238), Gfat2 (0 vs. 2 h: p = 0.0486, and 0 vs. 7d: p = 0.0112) represented in boxplots showing median, 25 and 75 percentiles, and extremes of values (a.u.: arbitrary units) depicted a significant increase of their expression at 2 h and 7d post ISO. The levels of O-GlcNAcylated proteins (i.e., the effect of the HBP overactivation), were significantly increased at 7d post-ISO compared to the control groups (p = 0.0118); unpaired comparison tests: *p < 0.05, **p < 0.01 and ***p < 0.001. Statistical significance (p < 0.05) for each variable was estimated by one-way or two-way ANOVA when group variances were equal (Bartlett test); if not the non-parametric Kruskall–Wallis test, and the Holm multiple comparisons test was used to execute simultaneous t-tests. Source data are provided as a « SourceData_Figure4_WB » file.

Fig. 6. Post-stress vascular remodeling of the LV.

a Representative 4 µm-section of the LV apex co-immunostained for Cd31 and alpha-Sma. The scale bar in the main image represents a length of 50 μm, while the scale bar in the two zoom-in images corresponds to a length of 20 μm. Quantitative analysis of Cd31 and alpha-Sma expression’s rate represented as boxplots showing median, 25 and 75 percentiles, and extremes of values. No dissociation between the Cd31-stained endothelial layer and the alpha-Sma-stained smooth muscle cell layer was found. b Increase in Cd31 expression (in the LV, 0 vs. 7d: p = 0.0013, and 2 h vs. 7d: p = 0.0318; in the apex, 0 vs. 7d: p = 0.0318, and 2 h vs. 7d: p = 0.0300; in the base, 0 vs. 7d: p = 0.0011, and 2 h vs. 7d: p = 0.0162) and in the ratio of capillaries per cardiomyocyte (2 h vs. 7d: p = 0.0003) and c in alpha-Sma (in the LV, 0 vs. 7d: p = 0.0004, and 2 h vs. 7d: p = 0.0209; in the apex, 0 vs. 7d: p = 0.0221, and 2 h vs. 7d: p = 0.0369) at 7d post-ISO indicates endothelial proliferation and angiogenesis. Unpaired comparison tests: *p < 0.05, **p < 0.01 and ***p < 0.001. Statistical significance (p < 0.05) for each variable was estimated by one-way or two-way ANOVA when group variances were equal (Bartlett test); if not the non-parametric Kruskall–Wallis test, and the Holm multiple comparisons test was used to execute simultaneous t-tests. Source data are provided as a « SourceData_Figure6 » file.