OUP user menu

Unfavourable consequences of chronic cardiac HIF-1α stabilization

Marion Hölscher, Katrin Schäfer, Sabine Krull, Katja Farhat, Amke Hesse, Monique Silter, Yun Lin, Bernd J. Pichler, Patricia Thistlethwaite, Ali El-Armouche, Lars. S. Maier, Dörthe M. Katschinski, Anke Zieseniss
DOI: http://dx.doi.org/10.1093/cvr/cvs014 77-86 First published online: 18 January 2012

Abstract

Aims The hypoxia-inducible factor-1 (HIF-1) is the master modulator of hypoxic gene expression. The effects of chronically stabilized cardiac HIF-1α and its role in the diseased heart are not precisely known. The aims of this study were as follows: (i) to elucidate consequences of HIF-1α stabilization in the heart; (ii) to analyse long-term effects of HIF-1α stabilization with ageing and the ability of the HIF-1α overexpressing hearts to respond to increased mechanical load; and (iii) to analyse HIF-1α protein levels in failing heart samples.

Methods and results In a cardiac-specific HIF-1α transgenic mouse model, constitutive expression of HIF-1α leads to changes in capillary area and shifts the cardiac metabolism towards glycolysis with a net increase in glucose uptake. Furthermore, Ca2+ handling is altered, with increased Ca2+ transients and faster intracellular [Ca2+] decline. These changes are associated with decreased expression of sarcoplasmic/endoplasmic reticulum calcium ATPase 2a but elevated phosphorylation of phospholamban. HIF-1α transgenic mice subjected to transverse aortic constriction exhibited profound cardiac decompensation. Moreover, cardiomyopathy was also seen in ageing transgenic mice. In parallel, we found an increased stabilization of HIF-1α in heart samples of patients with end-stage heart failure.

Conclusion Changes induced with transgenic cardiac HIF-1α possibly mediate beneficial effects in the short term; however, with increased mechanical load and ageing they become detrimental for cardiac function. Together with the finding of increased HIF-1α protein levels in samples from human patients with cardiomyopathy, these data indicate that chronic HIF-1α stabilization drives autonomous pathways that add to disease progression.

  • Hypoxia
  • Prolyl-4-hydroxylase domain enzymes
  • Hypoxia-inducible factor-1
  • Metabolism
  • Ischaemia
  • Cardiomyopathy

1. Introduction

Coronary artery disease and increased mechanical load are two major leading causes for cardiomyopathy.1 Myocardial ischaemia, in the case of coronary artery disease, and cardiac hypertrophy, in the case of increased mechanical load, are both associated with a dysfunctional oxygen supply to the heart. This is due to vessel occlusion and capillary rarefaction, respectively.2 Hypoxia is the initial trigger during myocardial ischaemia, and during increased mechanical load Hypoxia is the initial trigger during the process of myocardial remodelling.

The hypoxia-inducible factor (HIF) plays a pivotal role in the transcriptional response to changes in oxygen availability.3 HIF comprises two subunits, the α-subunit, which is regulated in an oxygen-dependent manner, and the constitutively expressed β-subunit. The HIFα subunit has an exceptionally short half-life and low steady-state levels in normoxic conditions.4 The regulation of HIFα half-life is mediated by three prolyl-4-hydroxylase domain (PHD) enzymes, which hydroxylate two prolyl residues by the use of molecular oxygen.5,6 Hydroxylation of HIFα allows binding of the von Hippel-Lindau tumour (pVHL) suppressor, which targets HIFα for proteasomal degradation.7,8 Hypoxia impairs the hydroxylation, which results in HIFα stabilization, nuclear accumulation, heterodimerization with HIF-1β, and subsequent hypoxia-inducible gene expression. HIF-1 is known to control the expression of a myriad of genes that regulate cell survival, cell metabolism, and angiogenesis in hypoxic conditions.9 In line with these documented functions, increasing HIF-1α stabilization has been reported to be protective against acute cardiac ischaemia.1012 The consequences of HIF-1α stabilization in the heart, which may form the basis for ischaemic cardioprotection, are not fully characterized.13

In particular, the consequences of long-term stabilization of HIF-1α in the heart are yet to be described. To clarify the outcome of chronically increased HIF-1α protein levels in the heart, we analysed the influence of cardiac transgenic overexpression of HIF-1α in mice (Hif-1αtg). It has been shown previously that cardiac HIF-1α stabilization has tissue protective effects when these mice are challenged by myocardial infarction at a young age (12-week-old mice);14 however, the long-term consequences of HIF-1α stabilization in older mice and their response to mechanical load are unknown.

2. Methods

2.1 Animals and surgical intervention

All protocols regarding animal experimentation were approved by the Niedersächsische Landesamt für Verbraucherschutz und Lebensmittelsicherheit (33.9.42502-04-10/0024 and 33.9.42502-04-10/0069) and conform with the Directive 2010/63/EU of the European Parliament. Surgical interventions were performed with littermate mice that were either wild-type (Hif-1αwt) or heterozygous for the HIF-1α transgene (Hif-1αtg) as described previously;14 see Supplemental Methods for further details.

2.2 Echocardiography

Echocardiography and measurements of posterior wall thickness (PWT), septum thickness (ST), left ventricular end-systolic diameter (LVESD), left ventricular end-diastolic diameter (LVEDD), and fractional shortening (FS) of ageing mice were performed as described before by our group.15 Assessment of fractional area shortening (FAS), PWT, ST, LVESD, and LVEDD after transverse aortic constriction (TAC) was performed using a Vevo2100 system (VisualSonics, Toronto, ON, Canada). See Supplemental Methods for further details.

2.3 Positron emission tomography analysis

The [18F]FTHA and [18F]FDG positron emission tomography (PET) measurements were performed on a dedicated small animal Inveon PET scanner (Siemens Healthcare, Knoxville, TN, USA) with 2 days in between to monitor the energy alteration of fatty acid oxidation and glucose metabolism. Experimental details of the PET analysis and the radiotracer production can be found in the Supplemental Methods section.

2.4 Histology

For Massons's Trichrome and periodic acid-Schiff staining, heart tissue was fixed in 4% paraformaldehyde in phosphate-buffered saline (PBS) and embedded in paraffin before sectioning. To visualize triglycerides and lipids, 10-μm-thick cryosections were prepared and stained with Sudan III. Nuclei were stained with Haematoxylin.

2.5 Immunohistochemistry

Paraffin-embedded sections were immunostained using the CSA II System (Dako, Carpinteria, CA, USA) as described previously by our group.11 The anti-HIF-1α primary antibody (Novus NB100-123) was used at a 1:1000 dilution.

2.6 Cryosections and immunofluorescence labelling

Freshly isolated hearts were treated and stained as described before.16 Details can be found in the Supplemental Methods section.

2.7 RNA extraction and RT-PCR analysis

After RNA extraction, reverse transcription was performed with 2 μg of RNA and a first strand complementary DNA (cDNA) synthesis kit (Fermentas GmbH, St Leon-Rot, Germany). mRNA levels were quantified by using 0.5 μL of the cDNA reaction and the SyBR Green qPCR reaction kit (Agilent, Santa Clara, CA, USA) in combination with the MX3005P light cycler (Stratagene, Santa Clara, CA, USA). The initial template concentration of each sample was calculated by comparison with serial dilutions of a calibrated standard. Primer sequences can be found in the Supplemental Methods section.

2.8 Western blot

Heart tissue was rapidly homogenized in a buffer containing 4 M urea, 140 mM Tris (pH 6.8), 1% SDS, 2% NP-40 and protease inhibitors (Roche, Grenzach, Germany). For immunoblot analysis, protein samples were resolved by SDS-PAGE and transferred onto nitrocellulose membranes (Amersham Biosciences, München, Germany) by semi-dry blotting (PeqLab, Erlangen, Germany). Membranes were probed with specific antibodies (see Supplemental Methods).

2.9 Isolation of cardiomyocytes and measurement of cardiomyocyte shortening and intracellular Ca2+

Hearts were excised from mice that were anaesthetized in an anaesthetic induction chamber with isoflurane and mounted on a Langendorff perfusion apparatus. The ventricular tissue was digested enzymatically, and single myocytes were isolated. After Ca2+ reintroduction (stepwise increase to 0.8 mM), myocytes were then plated onto superfusion chambers and used for immediate measurements. Myocytes were loaded with fluo-3. The dye was excited and emitted fluorescence measured. From the raw fluorescence, ΔF/F0 (increase in fluorescence compared to baseline) was calculated. Myocytes were field stimulated (voltage 25% above threshold) at 0.5, 1, 2, and 4 Hz at 37°C until a steady state was achieved; only those cells exhibiting stable steady-state contractions were included in the study. At least 25 myocytes from four mice per group were studied. For the measurements of cardiomyocyte shortening and Ca2+ transients, the investigator was blinded regarding genotypes. For additional details see Supplemental Methods.

2.10 Flow cytometry analysis

Seven days after sham surgery or TAC surgery, left ventricles were isolated and sliced into small pieces. Tissue digestion, staining procedure, and subsequent flow cytometry analysis are described in the Supplemental Methods.

2.11 Human myocardial tissue

Failing left ventricular (LV) tissues were obtained from patients with end-stage heart failure [dilated cardiomyopathy (DCM, n = 9) and ischaemic cardiomyopathy (ICM, n = 10)] undergoing heart transplantation. The LV ejection fraction was <30%, cardiac index 1.5–3.0 L/min × m2. Eight non-failing LV tissues (NF) from organ donors that could not be transplanted for technical reasons were used as controls. Donor patient histories or echocardiography revealed no signs of heart disease. Details have been published previously.17 The study conforms to the principles outlined in the Declaration of Helsinki and was reviewed and approved by the Ethical Committee of the University Hospital Hamburg (Az. 532/116/9.7.1991).

2.12 Statistical analyses

Data are presented as means ± SEM. We determined statistical differences by two-tailed Student's t-test (Figures 1B, C, and G; 2A, C, and D; 3A, B, D, and E; 4A, D, and I; and 5B; see Supplementary material online, Figure S2CE and S3), one-way ANOVA (see Supplementary material online, Figure S2A and B; and Figure 5H) and two-way ANOVA (Figures 4FH; 5A; and 5CG). ANOVAs were followed by Bonferroni post hoc tests. A P-value less than 0.05 was considered statistically significant.

Figure 1

HIF-1α accumulates in human DCM samples. Cardiac-specific HIF-1α transgenic mice show no gross structural changes and have normal heart function in the resting state. (A) HIF-1α protein was detected by western blot analysis. Tissue homogenates were prepared from normal, non-failing hearts (NF), and hearts of patients with end-stage dilated cardiomyopathy (DCM) and ischaemic cardiomyopathy (ICM). Samples were probed for HIF-1α and calsequestrin (CSQ). (B) Bar graph shows mean values (±SEM) from densitometric analysis of all hearts analysed in A. Protein levels in DCM and ICM hearts are expressed as fold change relative to protein levels in NF, which was set to 1 (*P< 0.05). (C) RT-PCR analysis confirmed the significant increase of HIF-1α mRNA transcripts in the left ventricles of Hif-1αtg mice compared with Hif-1αwt littermates (***P< 0.001). (D) Protein extracts prepared from left ventricles of Hif-1αwt and Hif-1αtg mice were analysed with anti-HIF-1α antibodies, confirming stabilization of HIF-1α protein in transgenic animals. (E) Immunohistochemical analyses of hearts from Hif-1αwt and Hif-1αtg mice. (F) Paraffin-embedded tissue sections of Hif-1αwt and Hif-1αtg hearts stained with Trichrome. (G) Hearts of 8-week-old female Hif-1αwt and Hif-1αtg mice were excised, and the ratios of heart weight (HW) to body weight (BW) were determined. Septum thickness (ST), posterior wall thickness (PWT) and fractional shortening (FS) were analysed by echocardiography.

Figure 2

Vascular changes in the hearts of Hif-1αtg mice. (A) Quantitative real-time RT-PCR analysis of left ventricles from 8-week-old Hif-1αwt and Hif-1αtg mice was performed. Transcript levels of genes involved in angiogenesis and vasotonus regulation were analysed (*P< 0.05; **P< 0.01; ***P< 0.001). (B) Capillaries of hearts of Hif-1αwt and Hif-1αtg mice littermates were analysed by anti-CD31 staining. Sections were co-stained for vinculin and DNA (Hoechst). The number of capillaries were counted (C), and the capillary area (D) was determined by measuring the area of CD31-stained capillaries using ImageJ (*P< 0.05).

Figure 3

Metabolic changes in Hif-1αtg mice. (A) Quantitative real-time RT-PCR analysis of left ventricles from 8-week-old Hif-1αwt and Hif-1αtg mice was performed. Transcript levels of genes involved in glucose metabolism and pH regulation were analysed (***P< 0.001). (B) [18F]FDG uptake in Hif-1αwt and Hif-1αtg mice. Bar graph shows cardiac standard uptake values (SUV) at 60 min (*P< 0.05). Data represent mean values ± SEM. (C) Periodic acid–Schiff staining of paraffin-embedded tissue sections of Hif-1αwt and Hif-1αtg hearts. (D) RNA expression levels of PPARα, PPARγ, and FAT in left ventricles from 8-week-old Hif-1αwt and Hif-1αtg mice as determined by quantitative real-time RT-PCR. (E) [18F]FTHA uptake in Hif-1αwt and Hif-1αtg mice. Bar graph shows cardiac SUV at 60 min. (F) Sudan III staining of frozen sections of Hif-1αwt and Hif-1αtg mice.

Figure 4

Changes in calcium handling in cardiomyocytes of Hif-1αtg mice. (A) RNA expression of SERCA2a and phospholamban (PLB) in left ventricles of 8-week-old Hif-1αwt and Hif-1αtg mice was determined in quantitative real-time RT-PCR analysis (***P< 0.001). (B) SERCA2a protein levels were decreased and RyR2 protein levels unchanged in the left ventricles of Hif-1αtg mice compared with Hif-1αwt mice, as determined by western blot analysis. (C) Protein expression of calcium handling proteins was detected by western blot analysis of protein extracts from left ventricles of Hif-1αwt and Hif-1αtg mice. (D) Quantification of SERCA2a, PLB, PLB T-17 and PLB-S16 protein levels as shown in A, B and C was determined by western blot analysis. (E) CaMKII and phosphorylated CaMKII protein levels were determined by western blot analysis. (FI) Ca2+ homeostasis in isolated Hif-1αwt and Hif-1αtg cardiomyocytes. Cells were stimulated at 0.5–4 Hz and analysed for Ca2+ transient amplitude (F; **P< 0.01, ***P< 0.001; data represent mean values ± SEM), RT 80% intracellular [Ca2+] decline (G; ***P< 0.001; data represent mean values ± SEM), and RT 80% twitch relaxation (H;  **P< 0.01, ***P< 0.001). (I) The sarcoplasmic reticulum Ca2+ content is unaltered in cardiomyocytes from Hif-1αtg mice measured by caffeine-induced Ca2+ transients.

Figure 5

Hif-1αtg mice develop signs of heart failure with ageing and after transverse aortic constriction. (A) Echocardiographic measurements were performed serially on a cohort of male Hif-1αwt (n = 8) and Hif-1αtg mice (n = 8) between the age of 2 and 8 months. Fractional shortening (FS), septum thickness (ST), posterior wall thickness (PWT), left ventricular end-diastolic diameter (LVEDD), and left ventricular end-systolic diameter (LVESD) were analysed (*P< 0.05, **P < 0.01, ***P < 0.001, Hif-1αtg mice vs. Hif-1αwt control mice; #P < 0.05, ##P < 0.01, ###P < 0.001, aged Hif-1αtg mice vs. 2-month-old Hif-1αtg mice). Sustained pressure overload was induced in female Hif-1αwt and Hif-1αtg mice by TAC [TAC treatment, Hif-1αwt (n = 5) and Hif-1αtg (n = 6); sham treatment, Hif-1αwt (n = 3) and Hif-1αtg (n = 4)]. (B) The TAC-induced pressure gradient was measured 1 day after surgery. Numbers in the bars indicate the number of animals in each group. ST (C), PWT (D), LVESD (E), fractional area shortening (FAS; F), and LVEDD (G) were analysed 2, 4, and 8 weeks after TAC by echocardiography. (H) The heart weight-to-tibia length ratios were determined 8 weeks after TAC (*P< 0.05).

3. Results

3.1 Cardiomyopathy is associated with increased protein levels of HIF-1α

In acute ischaemia, HIF-1α is stabilized in the heart.18 To determine whether the HIF system is still affected in prolonged heart failure, we analysed HIF-1α protein levels in ventricular tissue samples of patients with ICM and DCM and samples of non-failing hearts. In DCM samples, we found increased HIF-1α protein levels compared with NF heart samples. In ICM samples, the increase was subtle and not significant (Figure 1A and B). Taken together with the previously described increased HIF-1α protein levels in HCM patients,19 these data indicate a chronic activation of the HIF pathway in some heart failure patients.

3.2 Cardiac-specific Hif-1αtg mice have normal heart function in the resting state

To dissect the impact of elevated HIF-1α expression in the heart, we analysed the consequences of transgenic cardiac overexpression of HIF-1α in a mouse model. Hif-1αtg mice demonstrate significantly increased HIF-1α RNA and protein levels in the heart, as demonstrated by quantitative real-time RT-PCR, western blot and immunohistochemistry analyses (Figure 1CE). Transverse Masson`s Trichrome- and Sirius Red stained heart sections of 2-month-old Hif-1αtg mice did not reveal any obvious signs of gross malformation or fibrosis (Figure 1F and Supplementary material online, Figure S1). Likewise, heart weight related to body weight did not show any difference when Hif-1αtg mice were compared with their wild-type littermates (Figure 1G). This was also verified by the analysis of septum thickness and posterior wall thickness by echocardiography. In addition, there was no significant change in fractional shortening, heart rate (see Supplementary material online, Figure S2A), or blood pressure (see Supplementary material online, Figure S2B).

3.3 Cardiac-specific Hif-1αtg mice show differences in the expression of genes related to vasotonus, anaerobic metabolism and Ca2+ handling

HIF-1 affects diverse pathways associated with cellular and systemic adaptation to hypoxic conditions, including oxygen supply and metabolism, which are important for HIF-mediated tissue-protective effects.9 Oxygen supply via the bloodstream is affected by vessel density and vessel diameter. Most interestingly, we found a significant up-regulation of the vascular endothelial growth factor in the hearts of the Hif-1αtg mice only, whereas the expression of angiopoietin 1 and 2 was unaffected (Figure 2A). In contrast, we detected several HIF-1 target genes involved in regulating vascular tone, i.e. adrenomedullin, apelin, CD73, and inducible nitric oxide synthase to be up-regulated in the hearts of the Hif-1αtg mice. Correspondingly, we found no change in the number of capillaries when analysing cryosections of left ventricles obtained from Hif-1αtg and Hif-1αwt mice by CD31 staining (Figure 2B and C); however, when analysing capillary area we found a significant increase in the Hif-1αtg mice (Figure 2D). The expected improvement in oxygen supply might be at least one reason for the cardioprotective effect following a myocardial infarction that was previously described in Hif-1αtg mice.14

Cellular metabolism is changed dramatically in response to hypoxia, with the net effect of increased anaerobic glycolysis. This has been attributed partly to HIF-mediated transcription of glucose transporter-1 (Glut-1) and glycolytic genes accompanied by pH buffering through the hypoxia-inducible carboanhydrase IX (CAIX).3 In line with those findings, we observed an increased mRNA expression of Glut-1, phosphofructokinase I (PfkI) and CAIX in the Hif-1αtg hearts (Figure 3A). To determine whether this was accompanied by changes in the metabolic footprint, we performed cardiac [18F]FDG PET analysis. Compared with their wild-type littermates, hearts of Hif-1αtg mice used more glucose (Figure 3B), without changes in blood glucose and insulin levels (see Supplementary material online, Figure S2C and D). When exposing tumour cells to hypoxia, increased anaerobic glucose metabolism is paradoxically associated with increased glycogen storage.20 Unlike the situation in tumour tissue, we could not find an increased expression of glycogen synthase 1 (Gys1) and only a 1.5-fold increase in glycogen branching enzyme 1 (GBE1; Figure 3A). Moreover, there were no signs of increased glycogen storage in the hearts of Hif-1αtg animals (Figure 3C). When analysing lipid metabolism as determined by quantifying the expression of peroxisome proliferator-activated receptor α (PPARα)  and PPARγ and the expression of the PPARγ target gene product, fatty acid translocase (FAT), we found a weak yet significant down-regulation of PPARα, but no transcriptional changes in PPARγ and FAT (Figure 3D). A slight down-regulation in fatty acid metabolism in Hif-1αtg mice was seen when performing cardiac [18F]FTHA PET analysis (Figure 3E). This was not significant and exhibited much lower standard uptake values (SUV) than [18F]FDG PET images of the myocardium. Sudan III staining revealed no signs of lipid accumulation in Hif-1αtg mice (Figure 3F). Taking these results together, the Hif-1αtg hearts demonstrate metabolic remodelling with an increase in glucose utilization.

Heart function relies critically on proper calcium handling. There is a recent report demonstrating that HIF-1α impairs the promoter activity of the sarcoplasmic calcium pump, sarcoplasmic/endoplasmic reticulum calcium ATPase 2a (SERCA2a).21 These in vitro data were verified in our in vivo mouse model, because we observed a reduction of RNA and protein levels of SERCA2a in the Hif-1αtg hearts when compared with their littermates (Figure 4A, B and D). Besides SERCA2a, calcium handling critically involves the function of ryanodine receptor 2 (RyR2) and phospholamban (PLB); thus, we also investigated possible changes of these calcium handling proteins. RyR2 expression (Figure 4B) and PLB protein levels (Figure 4C) were unchanged. However, the phosphorylation level of PLB at the protein kinase A (PKA) phosphorylation site (Ser16) and the Ca2+/calmodulin-dependent protein kinase II (CaMKII) phosphorylation site (Thr17) was elevated in the Hif-1αtg hearts by 60 and 61%, respectively (Figure 4C and D). These changes were not associated with increased CaMKII protein levels or increased CaMKII activity as determined by analysing the phosphorylation state of CaMKII (Figure 4E).

To investigate whether the observed changes in SERCA2a protein level and PLB activity result in abnormal Ca2+ handling, we determined Ca2+ dynamics of isolated cardiomyocytes. Hif-1αtg cells demonstrated significantly increased intracellular Ca2+ transients when stimulated at 0.5–4 Hz (Figure 4F). Twitch relaxation and intracellular [Ca2+] decline, which are approximately 90% dependent on SERCA function in mice,22 were significantly increased and thus faster in cardiomyoctes isolated from transgenic vs. wild-type mice (Figure 4G and H). Yet, the sarcoplasmic reticulum Ca2+ load did not differ (P > 0.05) when comparing isolated cardiomyocytes from both genotypes (Figure 4I). Overall, these data indicate that chronic overexpression of HIF-1α alters Ca2+ handling properties of the myocardium, and that alterations in PLB phosphorylation may functionally override a reduction in SERCA2a protein levels.

3.4 Hif-1αtg transgenic mice develop heart failure with advanced age, or at a young age when challenged by increased mechanical load

The alterations identified in the Hif-1αtg mice regarding vessel area, metabolism, and calcium handling are at least partly protective adaptive mechanisms, which can be interpreted as improved oxygen supply and energy-saving mechanisms in response to ischaemia. Indeed, similar mechanisms do occur as compensatory mechanisms during the development of ischaemic heart failure.23,24 Despite those changes, Hif-1αtg mice demonstrated no decrease in heart function at the age of 3months, as described above. To determine whether chronic activation of the HIF-induced adaptive pathways alters heart function over time, we followed Hif-1αtg and Hif-1αwt littermates over a period of 8 months (Figure 5A). With age, Hif-1αtg mice developed a spontaneous increase in septum wall thickness and a decrease in fractional shortening, indicating that over time the HIF-1α-mediated changes result in cardiomyopathy.

Cardiac decompensation can be triggered by increased mechanical load. Initially, pressure overload leads to a compensatory response and myocardial remodelling, with a development of cardiac hypertrophy and ventricular dilatation. Ultimately, these changes have a detrimental effect on ventricular function, resulting in heart failure. We investigated the response of 2-month-old Hif-1αtg mice and Hif-1αwt littermates to increased mechanical load after TAC. Constriction of the aorta resulted in a significantly increased pressure afterload, which was not different between Hif-1αtg mice and Hif-1αwt groups (Figure 5B). Cardiac HIF-1α protein levels were not obviously changed 8 weeks after TAC treatment in Hif-1αtg mice and Hif-1αwt, as detected by western blotting (see Supplementary material online, Figure S3A). Myocardial inflammation in sham-operated and TAC-treated animals did not show any differences when Hif-1αtg mice were compared with their wild-type littermates, as determined by quantifying the number of CD45-positive cells (see Supplementary material online, Figure S3B). Left ventricular hypertrophy was indicated by a significant increase in ST and PWT in TAC-treated Hif-1αtg mice (Figure 5C and D) 2weeks after surgery, as determined by echocardiography. While Hif-1αwt mice did not show any signs of decompensation compared with sham-treated mice, the Hif-1αtg mice demonstrated a significant increase in LVESD 4 and 8 weeks after surgery (Figure 5E), together with a decrease in FAS (Figure 5F). The LVEDD (Figure 5G) was not significantly affected in TAC-treated transgenic mice compared with wild-type animals. In support of the TAC-induced hypertrophy and heart failure in Hif-1αtg mice, the heart-to-tibia length ratio was also increased (Figure 5H). Taking these results together, chronic activation of HIF-1 triggers intrinsic pathways in the cardiomyocytes, which impair long-term preservation of heart function and weaken adaption to increased mechanical load.

4. Discussion

There is ample evidence that HIF-1 protects the heart from an acute ischaemic insult. Based on the characterization of the Hif-1αtg hearts, as demonstrated in our study, this may be attributed to an increased capillary area in the heart and metabolic preconditioning via increasing gene expression and glucose metabolism that is necessary for the anaerobic metabolic switch. Moreover, the changes in calcium handling in Hif-1αtg hearts demonstrate energy-saving mechanisms regarding the depressed SERCA2a expression. The decrease in SERCA2a might be counter-regulated by an increased PLB phosphorylation and thus, a net increase in calcium release and shorter relaxation time. Although the HIF-mediated adaptation pathways regarding metabolic reprogramming and altered calcium handling are beneficial to withstand acute ischaemia, these same pathways are indicative of defective cardiac homeostasis in the failing heart.11,24,25 Accordingly, the tissue-protective properties of HIF-1 in the young Hif-1αtg mice were accompanied by development of heart failure with ageing or after increased mechanical load. Thus, it is tempting to speculate that chronic activation of HIF-1 and its associated adaptive pathways alone are sufficient to induce cardiomyopathy. We found increased protein levels of HIF-1α in heart samples of cardiomyopathy patients, indicating that the HIF pathway is activated during disease progression. According to the data obtained with our mouse model, induction of this regulatory pathway may be beneficial initially; however, if sustained it may actively contribute to heart failure.

Bao et al.26 showed in a rat model that treatment with GSK360A, and thus chronic stabilization of HIF-1α signalling, after myocardial infarction improved cardiac remodelling and ventricular performance in the ischaemic myocardium without affecting infarct size. This is contrary to the observed cardioprotective effect in the Hif-1αtg mice, which goes along with a decreased infarct size. In the study of Bao et al. no long-term maladaptive effects were observed 3 months after treatment. Nevertheless, it should be noted that the animals were treated systemically, whereas in the Hif-1αtg mice HIF-1α is stabilized in cardiomyocytes only. Furthermore, treatment with the PHD-inhibiting compound GSK360A lasted for only 4 weeks, after which the compound was withdrawn and washed out.

Likewise, the cardiac short hairpin RNA PHD2 knockdown after myocardial infarction mediates HIF-1α up-regulation and improved cardiac function27 without any chronic adverse effects. Again, however, in that mouse model the HIF-1α protein levels returned to baseline only 4 weeks after short hairpin RNA injection, whereas in the present Hif-1αtg mice we saw the first signs of cardiomyopathy at 3 months of age.

A detrimental effect associated with long-term stabilization of HIF-1α is similar to ventricular dysfunction of the heart in cardiac-specific pVHL−/− mice.28 It is important to note that a detrimental cardiac effect was also reported in tetracycline-inducible cardiac-specific HIF-1α transgenic mice and cardiac-specific HIF-2α transgenic mice.29,30 In both mouse models, the authors demonstrated a spontaneous cardiac decompensation of the mice at a young age, whereas the animals in our mouse model demonstrated normal heart function in non-challenged conditions but aggravated myocardial dysfunction over time and after TAC. This development of cardiomyopathy over time enabled us to determine the HIF-dependent adaptive changes in the heart, as described above, which occur before the onset of heart failure. While Bekeredjian et al. have created tetracycline-inducible HIF-1α overexpressing mice, we analysed an unregulated HIF-1α transgenic mouse model, which most probably presents a lower transgenic HIF-1α expression compared with the acutely inducible model.29 This assumption is additionally supported by the fact that in the tetracycline-inducible model, an HIF-1α variant was used that substituted amino acids critical to HIF-1α degradation (i.e. Pro402Ala, Pro564Ala, and Asn803Ala), whereas in the present model the wild-type HIF-1α protein was overexpressed. Likewise, Moslehi et al. have created a cardiac-specific HIF-2α transgenic mouse model, which overexpresses a non-hydroxylatable HIF-2α variant.30 The different degrees of HIFα levels over time may explain the varying onset of heart failure in our model compared with the tetracycline-inducible HIF-1α model and the HIF-2α transgenic mice. Interestingly, we previously showed that mice with reduced HIF-1α levels (HIF-1α+/− mice) also develop heart failure after TAC.15 This might imply that HIF-1α protein levels need to be tightly regulated, whereby neither decreased nor elevated HIF-1α protein levels can be tolerated.

While ischaemia develops rapidly in myocardial infarction, tissue hypoxia supposedly develops over a longer period of time during cardiac hypertrophy. The heart is particularly susceptible to hypoxia/ischaemia, because only limited reserves of high-energy phosphates are maintained.31 Quick metabolic reprogramming to anaerobic metabolism is therefore of critical importance to the heart's survival of an acute ischaemic insult.32 The Hif-1αtg hearts demonstrated signs of energy metabolism remodelling, with a shift towards glucose consumption. The ratio of ATP synthesis compared with the spent O2 is lower when glucose vs. fatty acids is consumed. This mechanism is used by the heart in the case of severe ischaemia and, based on our results demonstrated here, is transcriptionally controlled at least in part by HIF-1. On the molar basis over the long term, however, much more ATP is produced from fatty acid oxidation than from anaerobic glucose utilization.33 In uncompensated hypertrophy and in other forms of heart failure as a consequence of high energy demand, a sole increase in glucose uptake and utilization is therefore not sufficient to compensate for the overall energy demand.34 Thus, the HIF-1α overexpressing heart seems to be well prepared to face an acute ischaemic insult via expression of anaerobic metabolism-related genes and adapting calcium handling. In line with our results, a previous study has shown that constitutive HIF-1α overexpression protects the heart from diabetic-induced glucose metabolism impairment, partly by up-regulation of Glut-1.35 The metabolic preconditioning in HIF-1α transgenic mice, however, seems not to be sufficient for the adaptation towards increased mechanical load, because energy supply may be a limiting factor in this hypertrophy-inducing and energy-wasting setting.

In summary, the cardiac stabilization of HIF-1α results in tissue protection in the case of myocardial infarction.14 HIF-mediated pathways involved may include changes in vascular tone, energy metabolism, and calcium handling (Figure 6). According to our results, long-term activation of HIF-1 and its associated adaptive pathways is detrimental. Although short-term activation of HIF-1 is beneficial and may be used in therapeutic strategies, prolonged activation of HIF-1 over time drives the development of cardiomyopathy.

Figure 6

Scheme indicating possible mechanisms involved in protective and maladaptive effects of cardiac-specific HIF-1α up-regulation. For details see Discussion.

Funding

This work was supported by grants from the Else Kröner-Fresenius-Stiftung (P25/10//A12/10 to A.Z.); and the Deutsche Forschungsgemeinschaft (Ka1269/11-1 to D.M.K.).

Acknowledgements

We would like to acknowledge the technical support of Annette Hillemann, Anika Hunold, Felicia Steuer, and Thomas Sowa. Mareike Lehnhoff, Maren König, and Daniel Bukala were of valuable help in PET measurements. The radiopharmacy group, Gerald Reischl, Denis Lamparter, and Walter Ehrlichmann, of the Department of Preclinical Imaging and Radiopharmacy at the University Hospital of Tübingen, is acknowledged for radiotracer production. We also thank Samantha Whitman (University of Arizona, Tucson, AZ, USA) for her help preparing this manuscript.

Conflict of interest: none declared.

Footnotes

  • These authors contributed equally to this work.

References

View Abstract