Molecular Mechanisms That Differentiate Apoptosis from Programmed Necrosis

The Role of Programmed Cell Loss in Heart Failure and Diabetes

Shortly after birth (i.e., over the ensuing several days in mice or months in humans), cardiac myocytes lose the ability to undergo cell proliferation; further heart growth is almost entirely through hypertrophic cell enlargement. In the adult heart, normal hypertrophic growth is constrained by physical limits to oxygen diffusion that, when cell cross-sectional area exceeds the ability of oxygen to fully diffuse into the cell, produces ischemic cellular cores (Vatner 1988). Thus, myocardial homeostasis is not achieved as in most other tissues by maintaining a balanced rate of cardiomyocyte loss and renewal, but instead by minimizing the loss of cardiomyocytes that cannot be replaced. Accordingly, there seems to be no “physiological” rate for programmed cardiac myocyte death in normal hearts. Indeed, “normal” cardiomyocyte TUNEL staining is observed only in ∼1/10,000 normal human cardiomyocytes. By comparison, cardiac myocyte apoptosis is 10 or more times that normal value in failing, ischemic, and pressure overloaded hearts, and programmed loss of cardiomyocytes is believed to play a causal role in progression from functionally compensated hypertrophy to decompensated or failing cardiomyopathy (Whelan, Kaplinskiy, and Kitsis 2010). Genetic reprogramming leading to cardiomyocte death is a molecular hallmark of reactive pathological hypertrophy that is induced in response to myocardial injury or hemodynamic overload (Dorn 2005, 2009).

In contrast to terminally differentiated and nonproliferative adult cardiac myocytes, insulin producing β-cells of pancreatic islets are constantly remodeling according to the physiological requirements. β-cells are stimulated to proliferate (Kulkarni et al. 2004) or undergo programmed elimination (Bonner-Weir 2000) as needed to maintain physiological insulin homeostasis according to dietary intake and nutritional status. Diabetes occurs when endogenous insulin secretion by β-cells is insufficient to meet metabolic demands, either because of an absolute insulin deficiency in type 1 diabetes (Notkins and Lernmark 2001) or under conditions where insulin levels are normal, but are still functionally inadequate for a given body mass because of obesity and/or insulin resistance (type 2 diabetes; Pimenta et al. 1995; Rhodes 2005). Thus, both type I and type II diabetes are caused by (absolute or relative) deficiency in insulin-producing β-cells, (Mathis, Vence, and Benoist 2001). While β-cell loss in diabetes had been mechanistically linked to increased β-cell apoptosis (Johnson et al. 2003; Thomas and Biden 2009; Eizirik, Colli, and Ortis 2009), inhibiting apoptosis in β-cells has been incompletely effective as a therapeutic approach (Choi et al. 2009). This suggests that an additional or parallel mechanism leading to β-cell loss is also provoked by diabetes (Cnop et al. 2005). Recent studies have identified this mechanism as programmed necrosis.

A Brief Review of the Molecular Pathways for Apoptosis and Programmed Necrosis

Apoptosis signaling occurs via cascade activation of caspase proteases. There are two different proximal pathways leading to caspase activation: (1) the extrinsic pathway initiated via activated cytokine “death” receptors that process initiator caspase 8; and (2) the intrinsic pathway initiated after cytosolic release of mitochondrial derived cytochrome c, in which caspase 9 is activated. Both pathways terminate with activation of the executioner caspase, caspase 3. As the focus of this review is on different mitochondrial pathways leading to programmed cell death, here I review details of the intrinsic pathway only.

The sine qua non of mitochodrial pathway apoptosis is activation of Bcl-2 family of mitochondrial “death” proteins. Bcl-2 family proteins are broadly classified as pro-apoptotic or anti-apoptotic. The pro-apoptotic group is further subdivided according to structural/functional characteristics as either regulatory BH3 domain-only factors or pore-forming multidomain factors (Danial and Korsmeyer 2004). As discussed below, dynamic regulation of Bcl-2 factors can contribute to heart disease and diabetes by shifting the balance away from cell survival and toward cell apoptosis.

Cellular stress initiates an apoptotic response when transcriptional upregulation and/or posttranslational processing activates pro-apoptotic BH3-only factors. BH3-only proteins interact with the multidomain Bcl-2 proteins, Bax or Bak (Wei et al. 2001), to cause mitochondrial outer membrane permeabilization that releases mitochondrial apoptogens into the cytosol (Kluck et al. 1997). Mitochondrial outer membrane permeabilization by Bax and Bak is suppressed through the actions of antiapoptotic Bcl-2 proteins such as Bcl-2 and Bcl-x_L, either indirectly through their binding and sequestration of proapoptotic BH3-only factors or directly by their binding and inhibition of pore-forming Bax and Bak (Kharbanda et al. 1997). Bax/Bak-mediated destabilization of mitochondrial outer membranes leads to cytosolic relocalization of mitochondrial cytochrome c. This is the initial step in forming the macromolecular apoptosome, consisting of cytochrome c complexed with dATP, the adaptor protein Apaf-1 (apoptotic protease activating factor-1), and caspase 9. Apoptosome formation facilitates Apaf-1 oligomerization that is required for recruitment and activation of procaspase-9, which undergoes autoproteolysis to initiate cascade activation of downstream caspases.

Because unregulated apoptosis would be catastrophic, apoptosis signaling is tightly constrained at multiple levels. As noted, mitochondrial outer membrane pore formation is limited by antiapoptotic Bcl-2 protein binding to Bax and/or Bak. Inhibitor of apoptosis (IAP) family proteins bind to and prevent substrate access to downstream caspases-3 and 7 (Riedl et al. 2001), while concomitantly ubiquitinating and targeting these caspases for proteasomal degradation (Yang et al. 2000). Thus, mitochondrial cytochrome c release may not be sufficient to cause cell apoptosis unless these and other factors that chronically restrain apoptosis are simultaneously inhibited or suppressed, shifting the overall cell homeostatic balance toward cell elimination.

Signaling for programmed necrosis is less complex than that for apoptosis, in the same way that a car wreck is mechanistically less complex than an orchestral performance of Hayden’s Farewell Symphony. The inciting event in programmed necrosis (i.e., the first automobile collision of the chain reaction) is calcium-mediated opening of the MPTP, a nonselective pore in the mitochondrial inner membrane (Haworth and Hunter 1979). Creation of an ion-permeant pore connecting mitochondrial matrix with cytosol leads to dissolution of matrix protons essential for ATP production and dissipation of the normal mitochondrial inner membrane electrochemical potential (▵ψ_m) that sustains oxidative phosphorylation (Krasnikov et al. 2005). The immediate metabolic effect of calcium-mediated MPTP opening is reversal of ATP synthesis, that is, the mitochondrion becomes a net consumer (rather than a producer) of ATP. Loss of cellular ATP then causes general cell dysfunction and necrotic death. An indirect consequence of MPTP opening is osmotic swelling of the mitochondrial matrix, which, because the inner mitochondrial membrane is redundantly folded into numerous cristae, expands outward and ruptures the mitochondrial outer membrane. Outer membrane rupture permits apoptogenic cytochrome c that is normally sequestered in the compartment between the mitochondrial outer and inner membranes to leak into the cytosol and activate caspases through the intrinsic pathway as described above. For this reason, programmed necrosis is frequently associated with positive biochemical (activated caspase 3) and histological (TUNEL staining) markers of apoptosis. Nevertheless, cell death after MPTP opening does not occur by apoptosis because processing and packaging of cellular DNA, organelles, and soluble proteins into apoptotic bodies require energy in the form of ATP that is critically deficient after mitochondrial depolarization. Instead, cell death occurs via metabolic shutdown.

Research is still underway identifying all of the molecular components of the MPTP (Halestrap and Pasdois 2009; Javadov, Karmazyn, and Escobales 2009). Roles in MPTP function have been established for the voltage-dependent anion channel (VDAC), the adenine nucleotide translocator (ANT), and cyclophilin D (CyP-D; Szabo, De Pinto, and Zoratti 1993; Beutner et al. 1996; Marzo et al. 1998; Baines et al. 2007; Kokoszka et al. 2004). Genetic ablation of CyP-D has been especially useful in uncovering roles for MPTP opening and resulting programmed cell necrosis in murine Alzheimer’s disease, muscular dystrophy, diabetes mellitus, and heart failure (Du et al. 2008; Millay et al. 2008; Fujimoto, Chen, et al. 2010; Chen et al. 2010).

Nix, Cardiomyocyte Apoptosis, and Heart Failure

The genetic program that is induced in adult cardiac hypertrophy and heart failure includes multiple genes normally expressed during embryonic heart development (Dorn, Robbins, and Sugden 2003). Among these are genes encoding cell growth and death factors. Over a decade ago, our laboratory took advantage of a genetic model of cardiac hypertrophy that transitions to heart failure, the Gq transgenic mouse (D'Angelo et al. 1997; Adams et al. 1998), to define specific cardiac death genes that are expressed in hypertrophy and heart failure. Cardiomyocyte-autonomous activation of Gq signaling by overexpressing the alpha subunit of the heterotrimeric Gq signaling protein, Gαq, activated pathways normally stimulated by angiotensin II, α1-adrenergic, or endothelin receptors in the heart. The phenotypic consequence was recapitulation of many molecular, histological, morphological, and functional characteristics of pressure overload hypertrophy (D'Angelo et al. 1997). We also observed that the Gq hypertrophy phenotype progressed over time to classic dilated cardiomyopathy associated with greatly increased cardiomyocyte apoptosis (Adams et al. 1998; Sakata et al. 1998). The molecular underpinnings of cardiomyocyte apoptosis were defined using transcriptional profiling by RNA expression array. Surprisingly, the mRNA signature of non-failing cardiac hypertrophy included markedly increased levels of multiple apoptosis genes (Aronow et al. 2001; Yussman et al. 2002). The most thoroughly studied and important transcriptionally induced cardiac pro-apoptosis gene encodes Nix, one of the BH3-only pro-apoptotic Bcl-2 proteins. Nix is upregulated in a protein kinase C–dependent manner (Galvez et al. 2006) and stimulates mitochondrial outer membrane permeabilization by Bax and Bak. The discovery that Nix expression is intrinsic and apparently inseparable from the genetic reprogramming that occurs in reactive hypertrophy suggested a mechanism linking altered cardiac gene expression and programmed cardiomyocyte loss to the characteristic progression of functionally compensated hypertrophy to dilated cardiomyopathy.

To interrogate the pathophysiological consequences of increased cardiomyocyte Nix, we employed cardiomyocyte-specific forced Nix gene expression using murine transgenesis. Under conditions where cardiac Nix expression was increased shortly after birth the transgenic pups died of fulminant dilated cardiomyopathy on the 7th to 10th day of life. Massive increases in TUNEL positive cardiomyocytes (15–20% of all cardiac myocytes) implicated apoptosis as the mechanism for cardiomyocyte dropout and myocardial insufficiency (Yussman et al. 2002). A follow-up study in which forced cardiomyocyte Nix expression was conditionally induced in adult mouse hearts revealed synergism between genetic reprogramming and physiological stress for inducing cardiomyocyte apoptosis in adult hearts, suggesting how cardiomyocyte apoptosis can be activated after injury or hemodynamic overload in normally apoptosis-resistant adult hearts (Syed et al. 2004).

We further examined the role of Nix as the essential mediator of programmed cardiomyocyte death by creating cardiomyocyte-specific Nix gene knockout mice. Cardiac Nix knockout mice were subjected to chronic pressure overload resembling severe human hypertension by surgical transverse aortic banding. Pressure overloaded Nix-deficient hearts hypertrophied to a normal extent, but showed a ∼50% reduction in the “normal” amount of cardiomyocyte TUNEL positivity seen in this model. Additionally, cardiac Nix knockout mice showed substantial reductions in cardiomyocyte replacement fibrosis (a late marker of apoptotic cardiomyocyte dropout) and almost none of the left ventricular chamber dilatation, wall thinning, and decrease in contractile systolic function that is typical of late decompensation after pressure overload hypertrophy (Diwan et al. 2008). We therefore concluded that Nix is one of the major stress-inducible factors responsible for apoptotic cardiomyocyte loss, ventricular remodeling, and functional decompensation of pressure overload hypertrophy. The related BH3-only protein Bnip3, which is structurally and functionally similar to Nix (but encoded by a different gene that is regulated by tissue hypoxia), plays an analogous role mediating programmed cardiomyocyte loss and adverse ventricular remodeling after myocardial infarction (Diwan, Krenz, et al. 2007).

Nix-stimulated Programmed Necrosis Identifies Parallel Cardiomyocyte Death Signaling Pathways

As reviewed above, all pro-apoptotic Bcl-2 family members (including Nix and Bnip3) promote mitochondrial outer membrane permeabilization that activates endogenous pathways leading to caspase-dependent apoptosis. Accordingly, Nix primarily localizes to mitochondria (Yussman et al. 2002). However, we unexpectedly detected a fraction of both endogenous and transfected Nix protein that localized at the endoplasmic reticulum (ER) or its analogous organelle in cardiac myocytes, the sarcoplasmic reticulum (SR; Diwan et al. 2009). We further observed that dynamic regulation of Nix affected SR calcium homeostasis: cells/hearts with greater Nix expression had a ∼20% increase in SR calcium content, whereas cells/hearts lacking Nix had a ∼20% lower SR calcium content. These results led to the hypothesis that Nix-mediated regulation of SR calcium was increasing cardiomyocyte sensitivity to programmed cell necrosis through calcium-dependent MPTP opening. Consistent with this paradigm, genetic manipulations that normalized the characteristically depressed SR calcium levels of Nix knockout mice largely abolished the cardioprotection typically conferred by Nix gene ablation (Diwan et al. 2009). We further observed that Nix mutants engineered to localize specifically and exclusively to either cell/cardiomyocyte mitochondria or to ER/SR activated only one of the two different programmed cell death pathways, apoptosis and programmed necrosis, respectively (Chen et al. 2010). Transfecting cultured fibroblasts with mitochondrial–targeted Nix induced pure apoptosis defined as caspase activation without dissipation of the mitochondrial inner membrane electrochemical gradient, ▵ψ_m. By comparison, forced expression of ER-targeted Nix caused MPTPs to open, revealed by dissipation of mitochondrial ▵ψ_m. Pharmacological (cyclosporin A) or genetic (CyP-D/Ppif ablation) inhibition of MPTP opening largely prevented programmed cell death induced by ER-targeted Nix without affecting apoptotic death induced by mitochondrial targeted Nix (Chen et al. 2010).

Concomitant activation of parallel but independent apoptotic and necrotic cell death pathways by translationally upregulated Nix in hypertrophied hearts raised the question as to which pathway of cardiomyocyte elimination is more important for the pathogenesis of heart failure. To answer, we engineered transgenic mice in which we could force in vivo cardiomyocyte-specific expression of either mitochondrial–targeted or SR-targeted Nix, and examined the consequences on cardiac remodeling, function, and development of overt heart failure (Chen et al. 2010). Because apoptosis and programmed necrosis can both increase TUNEL positivity and biochemical caspase activity (see above), we assayed cardiac myocyte necrosis using histological staining for complement 9 protein, a marker of local inflammation. Anti-complement 9 labelling or mitochondrial swelling, matrix degeneration, and outer membrane disruption (visualized by transmission electron microscopy) were observed only in cardiomyocytes isolated from SR-targeted Nix expressing mice. This result established an in vivo link between SR localization of Nix and programmed necrosis. We further determined that MPTP opening is an essential part of the mechanism of cardiomyocyte death induced by SR-targeted Nix by genetically ablating ppif (encoding CyP-D) in mice expressing the mitochondrial  or ER/SR-targeted Nix mutant. Abrogation of the MPTP by ppif ablation prevented complement 9 staining, normalized mitochondrial ultrastructure, and prevented the dilated cardiomyopathy typically induced by cardiac expression of SR-targeted Nix. In contrast, ppif ablation only slightly improved apoptotic cardiomyopathy induced by mitochondrial targeted Nix. To our knowledge, these are the first findings to prove that programmed necrosis mediated by a SR/ER localized Bcl-2 factor can significantly impact end-organ function in vivo.

β-Cell Apoptosis and Programmed Necrosis in Diabetes

As introduced earlier, programmed elimination of β-cells from pancreatic islets and resulting insulin insufficiency are implicated in the pathogenesis of diabetes. We have studied apoptosis and programmed necrosis in a genetic model of diabetes, the Pdx1 haploinsufficient (+/−) mouse, that mimics the human conditions of heritable maturity onset diabetes of the young type 4 and type 2 diabetes (Stoffers et al. 1997; Stoffers, Stanojevic, and Habener 1998; Macfarlane et al. 1999). Whereas homozygous germ-line Pdx1 gene deletion produces lethal pancreatic agenesis (Jonsson et al. 1994), heterozygous germ-line or β-cell-specific Pdx1 null mutations induce a diabetic phenotype (Brissova et al. 2002; Gannon et al. 2008) that has been mechanistically attributed to both decreased β-cell proliferation (Sharma et al. 1999) and increased β-cell apoptosis and/or autophagy (Johnson et al. 2003, 2006; Fujimoto et al. 2009). (The consensus opinion seems to be that increased apoptotic β-cell death, and not decreased proliferation, is the most important mechanism; Thomas and Biden 2009.) Because the commonly used research markers do not faithfully distinguish between apoptotic and necrotic cell death, we addressed the question of “By what mechanism do β-cells die?” using pharmacological or genetic inhibition of the essential MPTP regulator, CyP-D. Our first studies were performed in vitro using cultured mouse insulinoma-derived β-cells (MIN6 cells). In these cells, Pdx1 expression can be suppressed using specific shRNA, thereby mimicking Pdx1 insufficiency caused by loss-of-function mutations in the human disease (Fujimoto, Chen, et al. 2010). Pdx1 shRNA increased MIN6 cell death and was associated with loss of MIN6 mitochondrial inner membrane potential, that is, with opening of MPTP. Treatment of MIN6 cells with cyclosporin A, a pharmacological inhibitor of CyP-D, prevented MPTP opening and protected Pdx1-deficient cells from programmed death.

The above results suggested a role for programmed necrosis in β-cell demise. However, one cannot evaluate diabetes in a petri dish. To translate the cell biology findings to an experimental disease model, we designed parallel studies that used analogous approaches in genetically manipulated mice. Pdx1+/− mice develop diabetes (measured as abnormal glucose tolerance) when fed a high fat diet: Pancreatic islets of Pdx1+/− mice are abnormally small with decreased insulin staining, indicating loss of pancreatic β-cells. The rate of Pdx1+/− β-cell TUNEL positivity is abnormally high, supporting conclusions that apoptosis is increased by Pdx1 insufficiency. We also observed increased complement 9 staining of Pdx1+/− islets, revealing local inflammation that would be generated by programmed necrosis, but not from apoptosis.

We developed Pdx1+/− diabetic mice with either functional MPTP or nonfunctional MPTP by cross-breeding to ppif null (i.e., CyP-D knockout) mice. Genetic ablation of CyP-D (and the resulting loss of functional MPTPs) restored Pdx1+/− pancreatic islet size and insulin staining while decreasing β-cell TUNEL labeling and islet complement 9 staining, demonstrating at least a partial role for programmed necrosis in β-cell loss. Finally, we found that CyP-D ablation almost fully normalized the characteristic decrease in circulating insulin and increase in blood glucose (at baseline and in response to an acute glucose challenge) seen in the Pdx1 haploinsufficient mouse model of diabetes. These results reveal an important, perhaps even dominant, role for MPTP-mediated cell necrosis in β-cell death caused by Pdx1 deficiency.

Unexpectedly, the pathways leading to programmed β-cell death in diabetes caused by Pdx1 deficiency were found to be directed by Nix. Unbiased transcriptional profiling of Pdx1-shRNA treated cultured MIN6 cells identified Nix as one of the upregulated mRNAs, and forced Nix expression was sufficient to induce MIN6 cell death. We also detected increased Nix mRNA in pancreatic islets isolated from Pdx1+/− mice, suggesting a role for Nix in the in vivo diabetic phenotype (Fujimoto, Ford, et al. 2010). We examined this by deleting Nix in Pdx1+/− mice, that is, we crossed the two knockout models to achieve concomitant hemizygous (+/−) Pdx1 deletion and homozygous (−/−) Nix ablation (Diwan, Koesters, et al. 2007) in the same animal. In this manner, we prevented upregulation of Nix in Pdx1+/− mice by interrupting its gene. Nix ablation eliminated the high pancreatic islet β-cell TUNEL labeling characteristic of Pdx1+/− mice and normalized islet size and insulin staining. Insulin secretion was normalized (i.e., increased) and reactive hyperglycemia limited by Nix ablation, thus “rescuing” or “curing” the diabetes phenotype.