Endothelial oestrogen-myocardial cyclic guanosine monophosphate axis critically determines angiogenesis and cardiac performance during pressure overload

Abstract

Aims: Oestrogen exerts beneficial cardiovascular effects by binding to specific receptors on various cells to activate nuclear and non-nuclear actions. Oestrogen receptor α (ERα) non-nuclear signalling confers protection against heart failure remodelling, involving myocardial cyclic guanosine monophosphate (cGMP)-cGMP-dependent protein kinase G (PKG) activation; however, its tissue-specific role remains elusive. Herein, we examine the cell type-specific role of ERα non-nuclear signalling in oestrogen-conferred protection against heart failure.

Methods and results: We first assessed the tissue-specific impacts of ERα on the cardiac benefits derived from oestrogen, utilizing endothelial ERα deletion (ERαf/f/Tie2Cre+) and myocyte ERα deletion (ERαf/f/αMHCCre+) female mice. Female mice were ovariectomized and the effect of estradiol (E2) was assessed in hearts exposed to 3 weeks of pressure overload [transverse aortic constriction (TAC)]. E2 failed to improve cardiac function in ERαf/f/Tie2Cre+ TAC hearts but provided benefits in ERαf/f/αMHCCre+ TAC hearts, indicating that endothelial ERα is essential. We next assessed the role of non-nuclear signalling in endothelial cells (ECs), employing animals with endothelial-specific inactivation of ERα non-nuclear signalling (ERαKI/KI/Tie2Cre+). Female ovariectomized mice were supplemented with E2 and subjected to 3-week TAC. ERαKI/KI/Tie2Cre+TAC hearts revealed exacerbated cardiac dysfunction and reduced myocardial PKG activity as compared to littermate TAC hearts, which were associated with attenuated myocardial induction of vascular endothelial growth factor (VEGF) and angiogenesis as assessed by CD31-stained capillary density. This phenotype of ERαKI/KI/Tie2Cre+was rescued by myocardial PKG activation from chronic treatment with a soluble guanylate cyclase (sGC) stimulator. We performed co-culture experiments to determine endothelial-cardiomyocyte interactions. VEGF induction by E2 in cardiac myocytes required a co-existence of intact endothelial ERα signalling in a nitric oxide synthase-dependent manner. On the other hand, VEGF was induced in myocytes directly with an sGC stimulator in the absence of ECs.

Conclusion: An endothelial oestrogen-myocardial cGMP axis stimulates angiogenic response and improves cardiac performance during pressure overload.

Keywords: Angiogenesis; Cyclic GMP; Heart failure; Mice; Non-nuclear signalling; Oestrogen; Transgenic.

Graphical Abstract

Figure 1

Cardiac phenotype of tissue-specific ERα-knockout mice (OVX ± E2) exposed to LV pressure overload for 3 weeks. (A) Representative M mode images of LV by echocardiography at 3 weeks after TAC surgery, time stamp: 100 ms, vertical bar: 2 mm. (B) LV FS (%) time course (pre, 1 week, and 3 weeks; n = 5–7 per group). (C) LV FS (%) at 3 weeks after TAC surgery (n = 5–7 per group). (D) Representative PV loops of LV at 3 weeks after TAC surgery from ERα^f/f/Cre−, ERα^f/f/αMHC^Cre+, and ERα^f/f/Tie2^Cre+ mice. PV loops during pre-load reduction and ESPVR (upper left straight lines) are shown. (E) dP/dt max/IP (/s), ESPVR slope (mmHg/μL), and EF (%) from PV loop analyses at 3 weeks after TAC (n = 5–8 per group); *P < 0.05 vs. sham and ^†P < 0.05 vs. TAC E2- were determined by using Tukey’s honest significant difference test following two-way ANOVA; P_int, interaction P value as determined by two-way ANOVA; scatter dot plots with bars show individual values and mean ± S.E.M. E2, estradiol; OVX, ovariectomy; dP/dt max/IP, dP/dt max normalized to instantaneous pressure; ESPVR, end-systolic pressure–volume relationship; ERα^f/f/αMHC^Cre+, mice lacking the ERα in CMs; ERα^f/f/Tie2^Cre+, mice lacking the ERα in ECs; ERα^f/f/Cre−, wild-type littermates.

Figure 2

Cardiac phenotype of animals with endothelial-specific ablation of ERα non-nuclear signalling (OVX + E2) exposed to LV pressure overload for 3weeks. (A) Representative M mode images of LV by echocardiography at 3 weeks after TAC surgery, time stamp: 100 msec, vertical bar: 2 mm. (B) LV FS (%) time course (pre, 1 week, and 3 weeks; upper panel) and LV FS (%) comparison at 3 weeks after TAC surgery (lower panel), time stamp: 100 msec, vertical bar: 2 mm (n = 5–11 per group). (C) Left ventricular mass (LVM) (mg) by echocardiography (upper panel) and post-mortal HW (mg) normalized to TL (mm) (lower panel) at 3 weeks after TAC surgery (n = 5–11 per group). (D) Representative PV loops during pre-load reduction (n = 5–7 per group) with corresponding ESPVR (upper left straight line) and EDPVR (lower right exponential curve). (E) dP/dt max/IP (/sec), EF (%), ESPVR slope (mmHg/μL) and co-efficient β of EDPVR from PV loop analyses at 3 weeks after TAC (n = 5–7 per group); *P < 0.05 vs. sham and ^†P < 0.05 vs. Cre− TAC were determined by using Tukey’s HSD test following two-way ANOVA; P_int, interaction P value as determined by two-way ANOVA; scatter dot plots with bars show individual values and mean ± S.E.M.; ERα^KI/KI/Tie2^Cre+: Cre+, mice lacking ERα non-nuclear signalling in ECs; ERα ^KI/KI/Tie2^Cre−: Cre−, wild-type littermates; EDPVR: end-diastolic pressure–volume relationship; other abbreviations as in Figure 1.

Figure 3

Gene expression and PKG activity in ERα^KI/KI/Tie2^Cre− and ERα^KI/KI/Tie2^Cre+ hearts. (A) Gene expression in the LV myocardium at 3 weeks after TAC surgery normalized to Gapdh (n = 5–9 per group). (B) PKG1 activity in myocardium (LV free wall) at 3 weeks after TAC surgery (n = 5–6 per group); *P < 0.05 vs. sham and ^†P < 0.05 were determined by using Tukey’s HSD test following two-way ANOVA; P_int, interaction P value as determined by two-way ANOVA; scatter dot plots with bars show individual values and mean ± S.E.M.; abbreviations are as in Figures 1 and 2.

Figure 4

Angiogenesis in ERα^KI/KI/Tie2^Cre− and ERα^KI/KI/Tie2^Cre+ OVX hearts exposed to 3-week pressure overload with or without E2 supplementation. (A) Representative histological images of left ventricles stained with CD31 at 3 weeks after TAC surgery (arrow heads indicate CD31-positive micro-vessels; scale bar: 32 μm). (B) Quantification results of capillary density (counts/mm²) at 3 weeks after TAC surgery in ERα^KI/KI/Tie2^Cre− and ERα^KI/KI/Tie2^Cre+ hearts (n = 5–6 per group). (C) Myocardial expression of Vegfa at 3 weeks after TAC surgery with E2 supplementation (n = 5–6 per group); *P < 0.05 vs. sham and ^†P < 0.05 were determined by using Tukey’s HSD test following two-way ANOVA; P_int, interaction P value as determined by two-way ANOVA; scatter dot plots with bars show individual values and mean ± S.E.M.; abbreviations are as in Figures 1 and 2.

Figure 5

The effect of sGC stimulation on ERα^KI/KI/Tie2^Cre− and ERα^KI/KI/Tie2^Cre+ OVX hearts subjected to pressure overload. (A) Representative M mode images of LV by echocardiography at 3 weeks after TAC surgery, time stamp: 100 msec, vertical bar: 2 mm. (B) LV FS (%) time course for ERα^KI/KI/Tie2^Cre− mice after TAC surgery (pre, 1 week, and 3 weeks; n = 8 per group). (C) LV FS (%) time course for ERα^KI/KI/Tie2^Cre+ mice after TAC surgery (pre, 1 week, and 3 weeks; n = 7 per group). (D) LV FS (%) and HW (mg) normalized to TL (mm) of ERα^KI/KI/Tie2^Cre− mice at 3 weeks after TAC surgery (n = 8 per group). (E) Nppb expression in the LV myocardium of ERα^KI/KI/Tie2^Cre− mice at 3 weeks after TAC surgery (n = 5–7 per group). (F) LV FS (%) and HW (mg) normalized to TL (mm) of ERα^KI/KI/Tie2^Cre+ mice at 3 weeks after TAC surgery (n = 7 per group). (G) Nppb expression in the LV myocardium of ERα^KI/KI/Tie2^Cre+ mice at 3 weeks after TAC surgery (n = 7 per group); *P < 0.05 vs. sham and ^†P < 0.05 were determined by using Tukey's HSD test following one-way ANOVA; scatter dot plots with bars show individual values and mean ± S.E.M.; abbreviations are as in Figures 1 and 2.

Figure 6

Angiogenesis in E2-deprived ERα^KI/KI/Tie2^Cre 3-week TAC hearts treated with an sGC stimulator. (A) Representative histological images of ERα^KI/KI/Tie2^Cre− left ventricle stained with CD31 at 3 weeks after TAC surgery with concomitant treatment with sGC stimulation (arrow heads indicate CD31-positive micro-vessels; scale bars: 64 μm for left panels and 32 μm for right panels). (B) Quantification results on capillary density (counts/mm²) (upper panel) and capillary/cardiomyocyte ratio (lower panel) at 3 weeks after TAC surgery for OVX ERα^KI/KI/Tie2^Cre− mice (n = 5–6 per group). (C) Vegfa expression in the LV myocardium at 3 weeks after TAC surgery with concomitant treatment with sGC stimulation in OVX ERα^KI/KI/Tie2^Cre− mice (n = 5–8 per group). (D) Representative histological images of ERα^KI/KI/Tie2^Cre− left ventricle stained with CD31 at 3 weeks after TAC surgery with concomitant treatment with sGC stimulation (arrow heads indicate CD31-positive micro-vessels; scale bars: 64 μm for left panels and 32 μm for right panels). (E) Quantification results on capillary density (counts/mm²) (upper panel) and capillary/cardiomyocyte ratio (lower panel) at 3 weeks after TAC surgery for OVX ERα^KI/KI/Tie2^Cre+ mice (n = 5–8 per group). (F) Vegfa expression in the LV myocardium at 3 weeks after TAC surgery with concomitant treatment with sGC stimulation in OVX ERα^KI/KI/Tie2^Cre+ mice (n = 5–6 per group); *P < 0.05 vs. sham and ^†P < 0.05 were determined by using Tukey’s HSD test following one-way ANOVA; scatter dot plots with bars show individual values and mean ± S.E.M.; abbreviations are as in Figures 1 and 2.

Figure 7

Interactive mechanism between ECs and CMs via oestrogen-cGMP signalling pathways. Vegfa expression in cultured CMs isolated from wild-type adult mice in the presence of an sGC stimulator (stimulated for 24 h) (A) and the amount of VEGFa protein (normalized to cardiomyocyte protein amount) in the culture medium (B). Vegfa expression in wild-type CMs co-cultured with wild-type (ERα^KI/KI/Tie2^Cre−) ECs in the presence or absence of E2/L-NAME (C) and the amount of VEGFa protein (normalized to cardiomyocyte protein amount) in the culture medium (D). Vegfa expression in wild-type CMs co-cultured with ERα^KI/KI/Tie2^Cre− (Cre−) or ERα^KI/KI/Tie2^Cre+ (Cre+) ECs in the presence or absence of E2 (E) and the protein amount of VEGFa (normalized cardiomyocyte protein amount) in the culture medium (F). (G) Representative images of E2-stimulated HUVEC (EC) migration with or without co-culture of wild-type (ERα^KI/KI/Tie2^Cre−) CMs (left panels). The ratio of migrated ECs to total seeded ECs (%) shown in the right bar graphs (n = 6–8 wells per group; scale bar: 300 μm). (H) Representative images of HUVEC (EC) migration in the presence or absence of E2 when EC and wild-type CMs were co-cultured. The ratio of migrated ECs to total seeded ECs (%) shown in the right bar graphs (n = 6–8 wells per group; scale bar: 300 μm); *P < 0.05 in Figure 7A, B, G, and H was determined by using Student’s t-test. P values in Figure 7C and D were determined by using Tukey’s HSD test following one-way ANOVA. *P < 0.05 vs. E2- and ^†P < 0.05 in Figure 7E and F were determined by using Tukey’s HSD test following two-way ANOVA; P_int, interaction P value as determined by two-way ANOVA; scatter dot plots with bars show individual values and mean ± S.E.M. HUVEC, human umbilical vein endothelial cells; ECs, endothelial cells; CMs, cardiomyocytes; other abbreviations are as in Figures 1 and 2.