Transgenic system for conditional induction and rescue of chronic myocardial hibernation provides insights into genomic programs of hibernation
- Dalit May*,
- Dan Gilon†,
- Valentin Djonov‡,
- Ahuva Itin*,
- Alon Lazarus*,
- Oren Gordon*,
- Christian Rosenberger§, and
- Eli Keshet*,¶
- Departments of *Molecular Biology and
- †Cardiology, The Hebrew University–Hadassah University Hospital, Jerusalem 91120, Israel;
- ‡Institute of Anatomy, University of Berne, CH-3000 Berne 9, Switzerland; and
- §Department of Nephrology, Charite University Clinic, 13353 Berlin, Germany
-
Edited by Napoleone Ferrara, Genentech, Inc., South San Francisco, CA, and approved November 9, 2007 (received for review September 7, 2007)
Abstract
A key energy-saving adaptation to chronic hypoxia that enables cardiomyocytes to withstand severe ischemic insults is hibernation, i.e., a reversible arrest of contractile function. Whereas hibernating cardiomyocytes represent the critical reserve of dysfunctional cells that can be potentially rescued, a lack of a suitable animal model has hampered insights on this medically important condition. We developed a transgenic mouse system for conditional induction of long-term hibernation and a system to rescue hibernating cardiomyocytes at will. Via myocardium-specific induction (and, in turn, deinduction) of a VEGF-sequestering soluble receptor, we show that VEGF is indispensable for adjusting the coronary vasculature to match increased oxygen consumption and exploit this finding to generate a hypoperfused heart. Importantly, ensuing ischemia is tunable to a level at which large cohorts of cardiomyocytes are driven to enter a hibernation mode, without cardiac cell death. Relieving the VEGF blockade even months later resulted in rapid revascularization and full recovery of contractile function. Furthermore, we show that left ventricular remodeling associated with hibernation is also fully reversible. The unique opportunity to uncouple hibernation from other ischemic heart phenotypes (e.g., infarction) was used to determine the genetic program of hibernation; uncovering hypoxia-inducible factor target genes associated with metabolic adjustments and induced expression of several cardioprotective genes. Autophagy, specifically self-digestion of mitochondria, was identified as a key prosurvival mechanism in hibernating cardiomyocytes. This system may lend itself for examining the potential utility of treatments to rescue dysfunctional cardiomyocytes and reverse maladaptive remodeling.
Ischemic heart disease, a leading cause of death worldwide, represents a heterogeneous group of pathological conditions underlined by insufficient perfusion of the heart muscle. Different types of vessel injury may result in various patterns of cardiac ischemia and, correspondingly, trigger a host of cardiac responses. The latter may range from asymptomatic ischemia to transient stunning, chronic myocardial hibernation (MH), and myocyte death and infarction. MH, the condition of contractile dysfunction in patients with chronic coronary artery disease, is characterized by a reduced or complete arrest of myocyte contractility that, together with other energy-saving adjustments, spare the myocardium from ischemic insults that otherwise would have been lethal. Importantly, these dysfunctional cardiomyocytes can fully resume normal function upon restoration of an adequate blood supply, sometimes years after the onset of ischemia. Hence, hibernating cardiomyocytes have been recognized as the critical reserve of dysfunctional cells that can be potentially rescued (reviewed in ref. 1).
Despite the clinical importance of MH, the underlying cellular and molecular mechanisms are poorly understood. This is primarily attributable to the lack of an experimental system for studying MH without the confounding factors of cell death, inflammation, fibrosis, and scarring, unavoidable consequences in current experimental models. We, therefore, devised a conditional transgenic system for generating a state of ventricular hypoperfusion and myocardial ischemia that can be tuned to a level where a large fraction of cardiomyocytes are driven to enter the hibernation mode yet without a detectable cell death. This VEGF-based conditional system is superior to current animal models relying on experimental vessel occlusion (reviewed in refs. 1 and 2) because of its unique ability to produce long-term, chronic MH and, furthermore, its ability to conditionally reverse hibernation via induced revascularization anytime after its onset.
Cardiac remodeling, manifested as changes in size, shape, and function of the heart after cardiac injury represent an adaptive process that may, however, progress to a maladaptive one. Thus, patients in whom remodeling initially aids in coping with pressure or volume overload may subsequently experience progressive worsening of cardiac function, advocating that reversing remodeling in the latter circumstances should be highly beneficial (3). Currently, it is not known under which circumstances left ventricular (LV) remodeling is still reversible. As described below, extensive LV remodeling also occurred in our experimental MH model, thus providing a unique opportunity to assess the reversibility of remodeling in settings in which significant cardiomyocyte cell death has not yet occurred.
VEGF is a hypoxia-inducible angiogenic factor assumed responsible for maintaining oxygen homeostasis in the heart. However, it is not known whether this VEGF function is backed by other angiogenic factors (4). Here, we provide evidence that VEGF is indispensable for matching microvascular density (MVD) according to an increased demand for oxygen.
Adaptive responses to hypoxia are coordinated by the transcription factors hypoxia-inducible factor (HIF)-1 and HIF-2 and mediated by a large number of their respective target genes. Superimposed on adjustments common to all cells experiencing hypoxia (primarily common metabolic adjustments) there are also tissue-specific regulations, and, moreover, different gene cohorts may be induced under acute vs. chronic hypoxia. The hibernating myocardium represents a prime example of a successful cellular adaptation to chronic hypoxia, as reflected in withstanding ischemia for months without loosing viability. Two critical prosurvival processes in MH are cessation of contractility, a cellular process consuming most oxygen and evading apoptosis. The underlying mechanism of either process is poorly understood. The latter is enigmatic considering that hypoxia is known to promote HIF-1-dependent apoptosis (5).
To account for the concerted induction of many genes acting together to sustain chronic ischemia in MH, it has been proposed that a “hibernation program” is likely to operate (2). Here, we took advantage of the unique opportunity to filter out entirely all other myocardial phenotypes to perform a high-throughput analysis toward elucidating the MH “program,” in general, and its “stop work” and “stay alive” signals, in particular.
Results
A Transgenic System for Conditional and Reversible Suppression of VEGF Signaling in the Heart.
To examine the role of VEGF in maintaining oxygen homeostasis in the heart, we devised a transgenic system to preclude cardiac VEGF signaling in a conditional and reversible manner, based on the induction of a secreted soluble VEGF decoy receptor 1 (sVEGF-R1) that is known to inhibit expression of VEGF-A, VEGF-B, and placenta growth factor (PlGF) (reviewed in ref. 6). Briefly, transgenic mice expressing a tetracycline-regulated transactivator protein (tTA) from a myosin heavy chain (MHC)α promoter exclusively in the heart were mated with transgenic mice harboring a transgene encoding a chimeric protein regulated by tetracycline that includes the VEGF-binding domain of VEGF-R1 (Fig. 1 A). Double-transgenic mice were selected for VEGF modulations, whereas littermates that inherited only one of the two transgenes served as controls. The onset of sVEGF-R1 activation in double-transgenic animals and the duration of its expression were tightly controlled by including or omitting tetracycline from the drinking water (“off” and “on” modes, respectively), also allowing restoration of VEGF function at will (Fig. 1 B). For further characterization of the system, see Materials and Methods.
Bitransgenic system for conditional and reversible suppression of VEGF signaling in the heart. (A) Schematic representation of the transgenic lines used in the bitransgenic inducible system. (B) Northern blot analysis of cardiac RNA with a probe specific for transgenic (human) sVEGF-R1. The term “off” indicates a control littermate; “on” indicates that sVEGF-R1 expression was induced for 1 month; “on>off” indicates that sVEGF-R1 was induced for 1 month, switched off by adding tetracycline, and analyzed 7 days later. Note the nonleaky expression, efficient induction, and complete reversibility of the genetic switch system.
VEGF Is Indispensable for Ischemia-Induced Heart Angiogenesis and Matching MVD to Increased Oxygen Demand.
To generate various degrees of cardiac ischemia, a VEGF-blocking strategy was used, with the reasoning that a sustained VEGF blockade would create a progressive perfusion deficit. This was based on the hypothesis that VEGF mediates vascular adjustments solely to meet changes in oxygen demand and supply, including compensating for a natural vascular turnover. To rule out a backup angiogenic mechanism, the sVEGF-R1 was activated at different times, both in neonatal and adult mice, and for various durations. Transgene expression was suppressed throughout embryonic development with tetracycline. Nonleaky expression (Fig. 1 B) ruled out developmental heart defects that could have impacted the phenotypes observed after birth. As anticipated, a significant microvascular deficit developed rapidly when VEGF was inhibited during the early postnatal period (starting at postnatal days 0–2), the outcome of inability to ramify the vasculature to match the dramatic increase in heart mass taking place at this time. The normally high MVD, a reflection of the exceptionally high metabolic demands of cardiomyocytes, was markedly reduced (by 2.8-fold) in the face of a VEGF blockade (Fig. 2). Inability to mount a rectifying angiogenic response persisted during the many weeks when sVEGF-R1 was present. This was despite the development of a significant myocardial ischemia and a positive-feedback response of up-regulated production of endogenous VEGF-A (see Fig. 5). Expression of the two other proteins sequestered by sVEGF-R1, namely VEGF-B and PlGF, which are not hypoxia-inducible remained unchanged [supporting information (SI) Fig. 7]. Remarkably, after relieving the VEGF blockade, angiogenesis was reactivated and normal MVD was regained within a few days (Fig. 2), also restoring the native 3D architecture of the vascular network (SI Fig. 8). These findings provide formal in vivo evidence that VEGF is indispensable for adjusting the coronary vasculature to increased oxygen consumption, as well as providing a framework for experimental induction of heart hypoperfusion.
Generating undervascularized heart and restoring normal MVD through VEGF manipulations. (A) Visualization of vessels in heart sections by using endothelial cell-specific lectin staining (brown). (B) Relative quantification of microvessels, expressed as the number of lectin-positive capillaries per high-power field (HPF). Off, 88.4 (n = 11 mice; 33 HPF counted); on, 31.6 (n = 7 mice; 16 HPF counted); on>off, 88.5 (n = 6 mice; 22 HPF counted).
Myocardial Ischemia Can Be Tuned to Levels Conducive for Chronic Hibernation Without Cell Death.
The nature and severity of the ischemic insult are likely to be a major factor in the cellular outcome, i.e., whether cardiomyocytes will enter a hibernation mode or die. Therefore, we sought experimental conditions that would drive a large fraction of cells into hibernation without inflicting cell death and, at the same time, would avoid forcing too many cardiomyocytes to stop contraction, thereby over-compromising heart function to an extent that would jeopardize the animal. The microvascular deficit and level of ischemia could be controlled by the timing of activating the sVEGF-R1, when early postnatal onset led to a greater microvascular deficit than late switching. Interestingly, a marked reduction in MVD was observed even when sVEGF-R1 was activated in adult mice. In this case, however, it took a longer time (up to 6–8 wk) for a significant microvascular deficit to develop (data not shown), likely reflecting the slow turnover rate of coronary vessels that could not be replaced in the face of a VEGF blockade. As shown in Fig. 3, the degree of myocardial dysfunction, measured by echocardiographic monitoring of the shortening fraction (Fig. 3 A), was linearly related to the reduction in MVD (Fig. 3 B). This allowed us to select experimental settings that fulfill the above set parameters. Importantly, under these conditions, cardiomyocytes could be maintained in a hibernating mode for weeks (Fig. 3 D) without cell death whatsoever, as determined by both morphological inspection of heart sections, by a failure to detect TUNEL-positive cells or caspase 3-positive cells in the heart, and by a failure to detect elevated levels of cardiac troponin T in the serum of hibernating mice (data not shown). There was also no evidence of inflammation, fibrosis, or scarring.
Induction of chronic, yet reversible, cardiomyocytes dysfunction. (A) M-mode echocardiography showing reduced ventricular contraction after activation of the sVEGF-R1. (B) An inverse linear relationship between the MVD (expressed as lectin-positive capillaries per HPF) and contractility [expressed as shortening fraction (SF)]. The data shown were compiled from measurements taken of control mice (diamonds), during hibernation (squares), and after switching off and revascularization (triangles). R 2 = 0.772, and, when including only the on and on>off groups, R 2 = 0.955. (C) M-mode echocardiography showing a complete functional recovery after terminating the VEGF blockade. Both images are from the same mouse before (on) and 1 wk after (on>off) silencing sVEGF-R1 expression. (D) Mice maintained up to 6 wk at the on mode (solid squares) show chronic cardiomyocyte dysfunction. Switching off sVEGF-R1 5 wk from the onset of hibernation (the group marked by empty triangles and the broken line; n = 5) resulted in functional restoration, reaching the normal SF value measured for the control group (solid circles).
The Hibernating Myocardium Can Be Rescued by Revascularization.
To determine whether the left ventricle can resume normal contractile function upon restoring adequate blood supply, thus qualifying cardiomyocytes as truly hibernating cells (1), the VEGF blockade was terminated at different times after its onset. As already shown in Fig. 2, switching off the sVEGF-R1 led to rapid revascularization and regain of normal MVD. Revascularization was accompanied by a complete recovery of contractile function. (See Fig. 3 C for an illustrative example of a single same mouse analyzed before and after silencing sVEGF-R1, and see Fig. 3 D for compiled data.) Of note, cardiomyocyte function could be rescued fully even after 3 months of continual MH, which is the longest period that we examined (data not shown). These findings indicate that adjustments to ischemia in hibernating cells succeed in maintaining cardiomyocytes at a fairly stable state of homeostasis, such that their functional deterioration is still reparable. These findings also underscore the potential utility of revascularization therapy in cases in which there is a large reserve of hibernating cardiomyocytes.
Ventricular Remodeling Associated with Hibernation Is Fully Reversible.
A compromised cardiac output and the resultant pressure and volume overload usually lead to extensive remodeling of the ventricle. Ischemic LV dysfunction in our experimental system also led to a marked remodeling, as evidenced by the appearance of its hallmarks, namely a change in shape and a significant increase in heart size (SI Fig. 9A), a progressive increase in LV volume (SI Fig. 9B), thinning of the ventricular wall (SI Fig. 9C), and marked collagen deposition (SI Fig. 9D). Capillaries in the ventricular wall were also remodeled, as manifested by significant thickening of the vascular basement membranes (SI Fig. 9E). As expected, processes accompanying myocardial infarction (MI)-induced remodeling such as infarct expansion, inflammation, and scar formation were undetectable, and myocyte hypertrophy, another manifestation of MI-associated remodeling (7), also was not observed.
To determine whether MH-associated remodeling is reversible, we rectified LV dysfunction via VEGF-mediated revascularization and reexamined remodeling parameters. Remarkably, remodeling was fully reversed upon rescue of MH, as evidenced by the return to normal values of all these parameters (SI Fig. 9). These findings indicate that, without a significant tissue damage incurred by infarction, the process is fully reversible.
Hibernating Cells Are Interspersed Among Functional Cardiomyocytes in Hypoxic Subendocardial Microenvironments.
We wanted to visualize hibernating cardiomyocytes and determine their spatial distribution within the ventricle wall. As shown in Fig. 4 A, histochemical staining of semithin sections with toluidine blue clearly distinguished a subpopulation of weakly stained cells, presumed to be hibernating cardiomyocytes, from a population of densely stained cells identical to those found in control heart. Inspecting these cells at EM resolution revealed that reduced staining was attributable to mitochondrial changes, including a decrease in electron density of the mitochondrial matrix and substantially fewer and irregularly distributed cristae, resulting in increased electron lucid zones (Fig. 4 B and C). Notably, these mitochondrial phenotypes resemble fully those detected in human patients with coronary insufficiency (8). The two subpopulations were found to reside side-by-side, often as intermingled clusters (Fig. 4 A) with a clear tendency of the hibernating microenvironments to reside in the subendocardium (data not shown). In this respect, our experimental system resembled the situation in human patients in which the subendocardium is more affected by ischemic insults than other myocardial regions. The ability to visually discern hibernating and functional cardiomyocytes in heart sections was exploited to establish that hibernating microenvironments were coincidental with the region of hypoxia. Immunohistologic staining of serial sections with a hypoxia marker (Hypoxiprobe; Fig. 4 D), specific HIF-1 antibodies (data not shown), or HIF-2 antibodies (Fig. 4 E) indicated that hibernating cardiomyocytes indeed resided in hypoxic regions of the ventricle wall. In situ hybridization with HIF target genes further corroborated this spatial relationship (see below).
Coincidental distribution of hibernating zones and hypoxic microenvironments in the subendocardium. (A) Toluidine blue staining of a semithin section through the ventricular wall. Note a clear demarcation of two cell populations arranged in clusters. Only the darkly stained population was detected in controls (data not shown). (B) A high-magnification EM image of a border zone showing abnormal mitochondria only in the hibernating cell. (C) EM images of control (off) and MH (on). Note that only the latter contains cells with apparently damaged mitochondria, residing side by side with cells having normal, densely compacted mitochondria. (D) Immunohistochemical staining for hypoxia in the control (off) and hibernating (on) subendocardium by using Hypoxyprobe-1. Note the localization of hypoxic cells (brown) near the endocardium. (E) Immunohistochemical staining for HIF-2 in the control (off) and hibernating (on) subendocardium. v, ventricle.
High-Throughput Analysis of the Hibernation Genetic Program.
The lack of an appropriate animal model to induce chronic MH and filter out irrelevant changes has precluded high-throughput analysis of a chronic hibernation program that, therefore, has been limited to analysis of genes in short-term ischemia (9). To elucidate the hibernation transcriptome, RNAs extracted from control and hibernating hearts were analyzed by using oligonucleotide arrays (Affymetrix). Only a relatively small number of genes showed significantly altered expression: 94 genes displaying up-regulated and 108 genes showing down-regulated expression (using a cutoff of a change >2-fold relative to control). The unique opportunity to reverse MH was used to determine whether expression of induced or repressed genes returned to normal values upon restoration of cardiomyocyte function. With few exceptions, reversal of hibernation (verified by echocardiography) indeed led to a return to normal expression levels (see Fig. 5 for representative examples), reinforcing the notion that even after weeks of MH, cardiomyocytes have not yet reached a no-return point with respect of their ability to readjust gene expression.
Attention was naturally drawn to genes likely to play a role in beneficial adjustments as grouped in the following categories.
Metabolic adjustments.
Up-regulated expression of the glucose transporter-1 (Fig. 5) and key glycolytic enzymes, including triphosphate isomerase, pyruvate kinase, and GAPDH (data not shown), was evident in the chronically hibernating hearts. These findings indicate that enhanced glucose uptake and glycolysis are integral components of the metabolic homeostasis under conditions of chronic ischemia. A previously proposed mechanism for keeping HIF levels in check under chronic hypoxia is the elevation of activities promoting its degradation, particularly prolyl-hydroxylase-3 (PHD3) (10). Interestingly, PHD3 was markedly induced during chronic MH and deinduced after restoration of oxygen homeostasis (Fig. 5).
Cessation of contractile function.
Stopping contraction is the most dramatic energy-saving adjustment in hibernation. Expression levels of many genes involved in different aspects of contractility were altered in the hibernating myocardium, including down-regulation of motor proteins (myosin 1B and tropomodulin 4), potassium pumps, and channels (kcnip2, kcnj3, kcnj8, and atp1a2) and up-regulation of kcne4, an inhibitory subunit of potassium currents (11). It remains to be determined which of these genes, if any, is causally related to cessation of contractility.
Cardioprotection.
Activation of the neurohormonal axis is considered to be an important facet of cardioprotection. Brain natriuretic peptide (BNP), atrial natriuretic peptide (ANP), and adrenomedullin (ADM) are thought to be cardioprotective by virtue of promoting vasodilatation, natriuresis, and diuresis, acting together to reduce cardiac load (reviewed in ref. 12). Importantly, BNP, ANP, and ADM were all strongly induced (by 11.7-, 9.7-, and 7.3-fold, respectively) in MH (Fig. 5). Up-regulation of ANP and BNP were associated with different heart pathologies, and BNP is used as a diagnostic marker for heart failure (13). We, therefore, wanted to determine whether these peptides are specifically induced in hibernating cardiomyocytes. To this end, in situ hybridization using ANP- and BNP-specific riboprobes was carried out. Results showed that expression of ANP and BNP was mostly confined to cardiomyocytes residing in the subendocardium of the left ventricle, coinciding with regions of cellular hypoxia and hibernation (SI Fig. 10). This finding suggests that ANP and BNP are secreted by hibernating cells, rather than uniformly expressed throughout the myocardium (e.g., as a result of secondary remodeling). Reversing MH was accompanied by down-regulation of BNP and ADM to normal levels, but, interestingly, high levels of ANP persisted, at least during the initial reversal stage (Fig. 5), presumably reflecting additional regulations superimposed on its hypoxic regulation.
Altered gene expression during hibernation and its reversal. Real-time PCR analysis of representative HIF target genes (Left) and neurohormones (Right). Expression levels are normalized relative to those determined in control hearts (designated as 1). Open columns represent control hearts (off), filled columns represent MH hearts (on), and filled/open columns represent hibernating hearts that were subsequently rescued (on>off). Glut-1, glucose transporter-1.
Autophagy As a Prosurvival Pathway in Hibernation.
To provide a mechanistic explanation for how cells survive during sustained hypoxia, we searched the transcriptome of hibernating hearts for altered expression of proapoptotic and antiapoptotic genes. The most significant change detected was induction of the proapoptotic gene Bcl2-interacting protein (Bnip)3, a member of the Bcl-2 family of mitochondrial cell-death-regulating factors (Fig. 6). Prompted by findings that Bnip3 is a key regulator of autophagy (a self-digestion process responsible for the turnover of unnecessary or dysfunctional organelles) in the heart (14), we searched for evidence of autophagy in chronic MH.
Autophagy in hibernating cardiomyocytes. Examples of cardiomyocytes undergoing autophagy, as evidenced by the presence of double-membrane autophagosomes. Note the discernable zones of autophagy within a single cell (lower part of the cell shown in A) and an example of progressive stages of mitochondrial degradation (black arrow in B).
The most reliable visual marker of autophagy is the typical double-membrane-bound vesicles, termed autophagosomes, containing the sequestered damaged organelles destined for lysosomal degradation. As shown in Fig. 6, a high-resolution transmission EM of the chronically hibernating myocardium indeed showed numerous double-membrane-bound vesicles, a reflection of ongoing autophagy. Autophagosomes were detected mostly in zones where hibernating cells are abundant. Interestingly, evidence for active autophagy was often detected in a cell that also contained zones of unaffected, apparently healthy mitochondria (Fig. 6). It should be stressed that ongoing autophagy could be detected at any time during the many weeks that cardiomyocytes were maintained in the hibernating mode, suggesting that continual autophagy is an essential prosurvival mechanism in MH. Evidence of autophagy could not be detected, however, after the termination of the VEGF blockade and regain of normal vascular density (data not shown).
Discussion
Our conditional system of VEGF blockade has several advantages over the previously used approach of cardiac VEGF deletion (15). First, it allows exclusion of affects attributed to developmental defects. Second, whereas it is difficult to obtain a complete deletion in all cells of the organ by using cre recombinase, complete inhibition is readily achieved through titration with inducible sVEGF-R1. Third, it allows restoration of VEGF function and examination of the potential for phenotypic reversion. Fourth, exercising different timings and durations of VEGF sequestration provides better control over the magnitude of functional loss. In the context of this study, the latter advantage has proven instrumental for precise tuning of the perfusion deficit to obtain a chronic sublethal ischemia.
Here, we used a prolonged VEGF blockade to generate a state of heart underperfusion, building on the assumption that compensatory angiogenesis to match increased demand for oxygen will not take place in the face of VEGF inhibition. Our success in generating a sustained microvascular deficit solely by VEGF inhibition retrospectively proves this assumption and, moreover, indicates that no other endogenous angiogenic activity replaces VEGF in this capacity. It should be noted, however, that the sVEGF-R1 described here sequesters not only VEGF-A but also VEGF-B and PlGF; therefore, observed phenotypes also could result from blockade of the latter factors, particularly blockade of VEGF-B, which has been shown to be abundantly expressed in the heart and the functions of which are not fully understood (16, 17).
The “genetic switch” system developed in this study is unique among animal models of ischemic heart diseases in at least three different aspects. First, in comparison with models based on coronary vessel occlusion, the opportunity to tightly control the level of hypoperfusion provided by our system enables driving large cohorts of cardiomyocytes into long-term hibernation while completely avoiding hypoxia-induced death. Second this is an animal model in which MH can be fully rescued and, moreover, in a noninvasive fashion and at any desired schedule (even weeks after the onset of MH). Third, the system provides direct experimental access to the clinically important process of reversing LV remodeling.
It has been suggested that MH represents an incomplete adaptation to a reduced myocardial oxygen supply and that, consequently, there is progressive diminution of the chance for complete structural and functional recovery after restoration of blood flow (18, 19). Here, we show that a fairly stable state of hibernation can be instituted and that the potential for full recovery maintained for a long period (provided, of course, that the perfusion deficit does not exceed a critical level). Thus, in principle, MH might represent successful institution of a new homeostatic state enforced by chronic ischemia.
Although it is generally assumed that not all HIF target genes are necessarily coinduced and that different conditions of oxygen and nutrient deprivation may induce different subsets of target genes, the particular constellations distinguishing, for example, adaptations to acute vs. chronic hypoxia have not been defined, let alone in vivo. In this regard, our ability to generate chronic ischemic conditions in the native heart environment provides a useful platform to address this question.
There is no evidence for the existence of a master gene (a presumed HIF target gene) coordinating the individual responses of cardiomyocytes to hypoxia. Rather, it is currently best described as parallel adjustments in multiple pathways acting together to restore homeostasis under chronic hypoxia. It should be pointed out that in the tissue retrieved for expression analysis, hibernating cells comprise necessarily only a fraction of cardiomyocytes, estimated by morphology not to exceed 30–40%, and, therefore, the reported changes in expression levels should be increased by this factor. Also, key regulatory steps might be at the posttranscriptional and posttranslational levels. To obtain a more holistic view of how metabolic homeostasis is attained, for example, elucidation of the hibernation proteome and metabolome may be beneficial. Preliminary results indicate that concomitantly with increased anaerobic glucose metabolism, MH is also characterized by decreased lipid metabolism (M. Mayr, D.M., and E.K., unpublished data).
Mechanistic insights into cessation of contractile function are of obvious importance. Although any of the changes in expression of genes encoding motor proteins and K channels reported here could potentially play a role in this process, it is equally possible that the critical changes are in channel activity. In this regard, recent studies show that three different glycolytic enzymes (triphosphate isomerase, pyruvate kinase, and GAPDH) can physically associate with the complex of ATP-regulated K channels and directly control its activity (20, 21). Interestingly, all three enzymes were found to be induced in our hibernating cardiomyocytes, prompting the speculation of a functional link between up-regulated glycolysis and activity of critical channels.
Extensive self-digestion of mitochondria occurs throughout hibernation. Whereas many studies have shown that autophagy is a prodeath process, others have shown that autophagy is a prosurvival process (reviewed in ref. 22). Recent studies have suggested that autophagy is associated with both cell death and survival taking place in myocardial cells in physiological and pathological conditions (23, 24). Clearly, in our MH system, autophagy can only play a prosurvival role, because cardiomyocytes remain fully viable. Self-digestion of mitochondria for the purpose of recycling much needed amino acids may not be deleterious in MH because maintaining a large number of these organelles becomes obsolete under conditions of limited oxygen and reduced contractility.
Considering that the population size of hibernating cells is a major determinant predicting the success of contemplated therapies, noninvasive means to determine abundance of hibernating cells are of great importance. Serum biomarkers of MH should be advantageous over the costly imaging techniques that are currently used. Our approach toward this goal is to first identify induced genes encoding secreted proteins and then examine whether they are specifically elaborated by hibernating cells by using in situ mRNA hybridization. As an illustration, serum BNP is currently used as markers of left ventricle overload (12, 13). However, our analysis shows that BNP and also ANP are not uniformly expressed throughout the myocardium but instead colocalize with hibernating zones. These findings provide evidence for up-regulated ANP and BNP expression in MH and may also suggest that their circulating levels are indicative of the extent of hibernation.
Recent studies have shown that disruption of coordinated tissue growth and angiogenesis in the heart contributes to the progression from adaptive cardiac hypertrophy to heart failure (25, 26). It is generally accepted that progressive remodeling beyond a certain degree (e.g., after MI) should be considered deleterious, hence advocating reversing or attenuating remodeling as a therapeutic goal (3, 7). However, it is reasonable to assume that the ability to reverse remodeling will progressively decrease with increasing tissue damage. In this regard, our finding that during chronic hibernation the myocardium retains full capacity for reversing remodeling argues for the potential benefit of revascularization therapy in MH patients (27). Identifying key genes mediating LV remodeling may suggest new targets for manipulating the process.
In conclusion, the genetic animal model that we developed provides valuable mechanistic insights into the highly important phenomenon of MH. It may serve as a platform for evaluating emerging pharmaceuticals and therapies intended for rescuing dysfunctional, yet viable cardiomyocytes or for reversing maladaptive remodeling.
Materials and Methods
Transgenic Mice and Conditional Modulations of VEGF Signaling.
A bitransgenic system for organ-specific, tetracycline-regulated transgene expression was used. Heart-specific induction was achieved by using a transgenic driver line in which tTA expression is driven by a myosin heavy chain (MHC)α heart-specific promoter (28). The tet–sVEGF-R1 transgenic line encodes a tetracycline-inducible protein composed of an IgG1–Fc tail fused to the extracellular domain of VEGF-R1 (corresponding to amino acid residues 1–631 of human VEGF-R1 containing the ligand-binding domain but lacking the transmembrane and cytoplasmic domains). The induction of sVEGF-R1 in double-transgenic mice by tetracycline withdrawal and the termination of sVEGF-R1 by tetracycline addition (0.5 mg/ml tetracycline and 3% sucrose in the drinking water) were carried out as described (29).
Gene Expression Profiling.
Four different litters, each containing two control mice and two mice in which sVEGF-R1 expression in the postnatal heart was induced 4 wk earlier, were analyzed. A successful induction of hibernation in these mice was verified by echocardiographic examination before they were killed. Hearts were divided into two parts. One part was quick-frozen and used for DNA microarray analysis by using Affymetrix Mouse Genome Array 430A (≈22,000 genes), and data were analyzed by using Mas 5 array analysis software. RNA extracted from the other part was used for real-time PCR quantification of selected genes. For further details, see SI Text.
Immunohistochemistry.
Paraffin sections (5 μm) were used, and antigen was retrieved by microwaving at 92°C in a citrate buffer (pH 6; Zymed Laboratories) for 20 min.
Apoptosis was determined by using both a TUNEL assay and immunostaining for activated caspase-3 (Cell Signaling). Endothelial cells were visualized by using Bandeiraea simplicifolia isolectin B4 staining. Hypoxia in the sections was determined by using Hypoxyprobe-1 (Chemicon International) according to the instructions of the manufacturer. Anti-HIF-1 antibodies were from Novus (catalog no. NB 100-449; diluted 1:200) and rabbit anti-mouse HIF-2 (PM9, gift from Patrick Maxwell, Wellcome Trust Centre for Human Genetics, Oxford, U.K.).
Transmission EM.
For EM studies, anesthetized mice were perfusion-fixed by using a solution of cold 2.5% glutaraldehyde in PBS, postfixed in osmium tetroxide, block-stained by using uranyl acetate, dehydrated through ascending concentrations of ethanol, and embedded in epoxy resin (30). Ultrathin sections were obtained at 90 nm, counterstained with lead citrate, and viewed on an EM-300 microscope (Philips).
Echocardiography was performed as described in SI Text.
Acknowledgments
We thank Dr. Eli Pikarsky for helpful discussions. This work was supported by the Israel Science Foundation (E.K.). E.K. is the incumbent of the Woll Brothers and Sisters Chair of Cardiovascular Diseases at the Hebrew University.
Footnotes
- ¶To whom correspondence should be addressed. E-mail: keshet{at}cc.huji.ac.il
-
Author contributions: D.M., D.G., V.D., A.I., A.L., O.G., and C.R. performed research; and E.K. wrote the paper.
-
This article is a PNAS Direct Submission.
-
The authors declare no conflict of interest.
-
This article contains supporting information online at www.pnas.org/cgi/content/full/0707778105/DC1.
- © 2007 by The National Academy of Sciences of the USA