Messenger Molecules and Cell Death: Title and subTitle BreakTherapeutic Implications

Nobel Laureate Julius (“Julie”) Axelrod often commented, “Nature is very efficient. If she finds a good molecule, she uses it over and over again.” He was likely thinking of molecules such as histamine, which, besides its role in allergic reactions, possesses distinct roles as a mediator of gastric acid secretion and as a neurotransmitter in the brain. Sometimes, diverse roles of messenger molecules reflect coordinated functions. Thus, angiotensin II stimulates sodium retention through adrenal formation of aldosterone, enhances blood vessel contraction, and, in the brain, augments drinking behavior. More instances are appearing that neurotransmitter-related systems play major roles in biology beyond neurotransmission. A similar diversity of function is evident for nonneural messenger systems. In many cases, these actions influence cell death and suggest novel approaches to therapy in diverse conditions such as stroke, myocardial infarction, and neurodegenerative disease.

In this account, drawn in large part from work conducted in our own laboratory, we review 3 novel signaling pathways that reflect novel ways in which cells are protected from potentially lethal insults and also review new cascades involved in programmed cell death, also known as apoptosis.

Just as there are any number of means by which a person might meet his or her demise, individual cells can die by a variety of processes. Two of the best-described types of cell death are known as necrosis and apoptosis.^{1 - 3} In necrosis, the process perhaps best known to clinicians, cells are exposed to a pathophysiologic stress—eg, myocardial ischemia, hypoglycemia, infection, or inflammation—leading to a florid swelling and rupture of the cell. As the mitochondria—the principal source of energy in the cell—disintegrate, cellular energy levels are depleted, leading to disintegration of various other organelles. To use a legal metaphor, necrosis is murder of the cell, which is an unwilling participant.

By contrast, apoptosis is cellular suicide—ie, the cell is an active participant in its own death. For instance, death receptors (eg, Fas) on the cell surface activate a cascade of executioner enzymes known as caspases, leading to an irreversible commitment to cellular death.

Morphologically, necrosis differs from apoptosis. Whereas necrosis is characterized by cellular rupture, in apoptosis the cell membrane becomes permeable to the outside environment yet remains relatively contiguous; cellular chromatin condenses and is cleaved into small fragments as the nucleus disintegrates. Although energy depletion can result in necrosis, energy is required for apoptosis. Whereas necrosis also generally produces an inflammatory response, apoptosis does not.

Although necrosis is typically pathologic, apoptosis may be either physiologic or pathologic. For instance, excess numbers of neurons in the developing brain are eliminated by normal apoptotic death. If this process is disrupted, brain dysfunction and malformation may occur.^{4 - 7} Normal apoptosis is also required to avoid webbing between the digits of the extremities and to remove autoreactive or improperly developed T and B cells of the immune system.^{7 - 8} Inefficient or dysregulated apoptosis may lead to cancer, and necrosis and apoptosis both occur in neurodegenerative conditions such as amyotrophic lateral sclerosis, as well as Alzheimer, Huntington, and Parkinson diseases.^{2 - 3}

Cytoprotection by the Bilirubin Pathway. Neurotransmitters come in chemical classes. The biogenic amines—acetylcholine, norepinephrine, serotonin, and others—were the first neurotransmitters to be identified, followed by amino acids such as γ-aminobutyric acid, glutamate, and glycine. Peptides such as the opioid peptides, the enkephalins and endorphins, and substance P were subsequently identified. Gaseous transmitters are the most recent class, with nitric oxide functioning both in the brain and peripheral autonomic nervous system in addition to its roles as a determinant of blood vessel relaxation and regulator of macrophage function.

More recently, carbon monoxide has been shown to be a neurotransmitter formed by heme oxygenase (HO), which exists in 2 forms, HO-1 and HO-2. First discovered was HO-1, which is concentrated in the spleen, where it degrades heme released from aging red blood cells. Heme oxygenase breaks open the heme ring, releasing iron and carbon monoxide and giving rise to the green pigment biliverdin, which is rapidly reduced by biliverdin reductase (BVR) to the yellow pigment bilirubin (Figure 1). In the liver, conjugation enzymes attach sugar residues to bilirubin to make it more water soluble, allowing its passage into bile. Aside from its neurotransmitter properties, carbon monoxide has also been shown to activate a variety of intracellular signaling pathways.⁹ Its antiproliferative properties at low doses may offer promise in treating atherosclerotic disease.^{10 - 11}

HO-1 is an inducible enzyme whose synthesis is rapidly and profoundly augmented by myriad cell stressors, including most inflammatory stimuli as well as heat shock. Hence, it is also known as heat shock protein 32. Heme itself is one of the most powerful inducers of HO-1.

Evidence for a role of the HO system in cellular protection came from studies of HO-2 gene knockout mice,^{12 - 14} the enzyme form predominantly found in the brain and testes. These mice are much more sensitive to stroke damage than are normal mice. Moreover, brain cultures of these knockout mice display far greater degrees of cell death in response to oxidant stressors such as hydrogen peroxide than do control cultures.¹⁵ Loss of formation of biliverdin and bilirubin explains the augmented neurotoxicity.

Remarkably, bilirubin acting as an antioxidant combats the oxidative stress elicited by 10 000-fold higher concentrations of hydrogen peroxide, which at first seems impossible.¹⁵ The resolution to this dilemma came with the identification of a novel BVR cycle¹⁶ that regenerates bilirubin (Figure 1). Thus, bilirubin acting as an antioxidant can itself be oxidized to its precursor, biliverdin. Tissue levels of biliverdin do not accumulate, because biliverdin is rapidly reduced to bilirubin through the action of biliverdin reductase, an enzyme abundantly expressed in many tissues. Therefore, as soon as a molecule of bilirubin is oxidized to biliverdin, it is immediately recycled to bilirubin.

Assuming that bilirubin is a physiologic cytoprotectant, this regenerative cycle makes good sense. Very high levels of bilirubin can have toxic effects such as kernicterus, the deposition of bilirubin in the basal ganglia of the brain. Recycling of bilirubin by the BVR cycle allows the cell to keep bilirubin levels low enough to be safe but still effective as an antioxidant. Evidence that this BVR cycle is a physiologic cytoprotectant comes from experiments in which BVR was depleted from tissues by RNA interference technology. In these tissues, reactive oxygen species accumulated to very high levels, and the cells became much more sensitive to cell stressors.¹⁶

Reactive oxygen species are generally regarded as a final common pathway for cell death and are elicited by all manner of tissue insults, such as ischemia associated with myocardial infarction or stroke, inflammatory stimuli such as endotoxin, and many anticancer drugs. Reactive oxygen species are also generated in the mitochondria as part of the multiple metabolic processes of normal physiology. In susceptible cells, low levels of oxidants can lead to apoptosis, but high levels can cause necrotic death.

To protect cells from physiologic and pathologic oxidative stress, cells require endogenous, physiologic cytoprotectants. The tripeptide glutathione (γ-glutamyl-cysteinyl-glycine), which occurs in all tissues in high millimolar levels, has generally been accepted as the major cytoprotectant for most tissues. Depleting tissue glutathione levels by 95% through treatment with buthionine sulfoximine, which inhibits the glutathione-synthesizing enzyme γ-glutamylcysteine synthase, increases levels of reactive oxygen species and augments cell death but to a lesser extent than does depletion of bilirubin.¹⁶

It appears likely that both bilirubin and glutathione are physiologic cytoprotectants. By detoxifying oxidants, they can block both necrotic and apoptotic types of cell death. They also serve complementary functions. Thus, the lipophilic bilirubin may protect cells against damage to cell membranes caused by lipid peroxidation, while the hydrophilic glutathione may protect water-soluble proteins. Evidence for this comes from experiments in which depletion of BVR leads to a relatively selective augmentation of lipid peroxidation, while depletion of glutathione selectively augments oxidative damage to water-soluble proteins.¹⁷

Bilirubin and glutathione are clearly not the only endogenous cytoprotectant systems. Enzymes such as superoxide dismutase and catalase, which act in concert, convert superoxide to water. Also, the vitamins ascorbate (vitamin C) and α-tocopherol (vitamin E) exert antioxidant actions, as does uric acid.¹⁸

Mildly Elevated Bilirubin Levels: Clinical Implications. There are clinical ramifications of bilirubin's physiologic cytoprotectant role.^{19 - 20} Several studies suggest that mildly to moderately elevated levels of serum bilirubin offer therapeutic advantage in diseases associated with oxidative stress. Multiple investigators have linked elevated serum bilirubin levels with diminished risk of coronary artery disease. In studies using data from the Framingham cohort, higher bilirubin levels were associated with a lower risk of myocardial infarction and other cardiovascular events.^{21 - 22} In one case-control study, patients with a familial history of coronary artery disease demonstrated lower serum bilirubin levels than did those without such a history.²³ The protective effect of bilirubin in this study was comparable with that afforded by high-density lipoprotein cholesterol.

Gilbert syndrome is a common genetic disorder presenting as mild to moderate augmentation of unconjugated bilirubin caused by impairment of bilirubin conjugation.²⁴ In one study, the prevalence of ischemic heart disease in individuals with Gilbert syndrome was 2%, or one sixth that of the control population, and elevated bilirubin level was more prominent as a protective index than was high-density lipoprotein cholesterol.²⁵ A meta-analysis of 11 studies revealed a diminished risk of atherosclerosis in individuals with elevated bilirubin levels.²⁶ In a large patient population screened for carotid stenosis, elevated bilirubin levels were associated with a 32% reduction in the risk of carotid plaques.²⁷

While very high levels of serum bilirubin in infants are unquestionably toxic, some investigators have reported a decreased incidence of the retinopathy of prematurity in infants with elevated serum bilirubin levels,^{28 - 29} though others have not observed such a relationship.^{30 - 35} Neonatal Gunn rats, whose elevated levels of serum unconjugated bilirubin result from a genetic defect causing a lack of glucuronidation, resist oxidative damage when they are exposed to high oxygen levels.³⁶

These findings may have direct therapeutic implications. In rodents, induction of HO-1 by treatment with porphyrin derivatives protects against ischemic insults.³⁷ Intravenous administration of bilirubin diminishes drug-elicited pulmonary fibrosis³⁸ and intestinal ischemic-reperfusion injury³⁹ in animal models. Moreover, rinsing liver grafts with bilirubin leads to diminished postgraft injury.⁴⁰ Although bilirubin may become toxic when its concentrations increase to levels 15 to 50 times normal, this is actually a much safer toxicity profile than that of sodium, potassium, calcium, or cortisol, in which morbidity results from doubling or tripling of serum levels. Although most studies measure levels of serum bilirubin, it may actually be tissue bilirubin—which is more difficult to assess—that mediates beneficial effects.

A single case of human HO-1 deficiency has been reported, occurring in an individual who inherited a defective allele from each parent.^{41 - 43} The patient displayed hemolytic anemia, asplenia, cellular sensitivity to oxidative damage, damaged vascular endothelia, renal tubular injury, and coagulopathy before succumbing to intracranial hemorrhage at age 6 years. The human HO-1 gene has at least 3 different genetic polymorphisms associated with a reduced ability to increase its expression following oxidative stress.⁴⁴ These polymorphisms have been associated with emphysema in smokers, renal allograft deterioration, idiopathic miscarriage, coronary artery disease, and restenosis following angioplasty or stenting.^{44 - 45}

Genetic studies have suggested that uridine diphosphate glycosyltransferase 1 contributes to mildly elevated bilirubin levels, which in turn protect against coronary artery disease.⁴⁶ This enzyme conjugates bilirubin with sugar residues so that it may be excreted in bile. Inhibitors of bilirubin conjugation and excretion represent a means by which bilirubin levels could be mildly increased for beneficial effects.

Glyceraldehyde-3-Phosphate Dehydrogenase and a Novel Cell Death Cascade. The concept of multiple, diverse functions for proteins and small molecules has recently been extended to common “housekeeping” glycolytic enzymes. For instance, in iron-depleted cells, cytosolic aconitase loses its catalytic activity and is transformed into the protein called iron regulatory protein, which can bind specific RNA hairpin loop structures called iron-responsive elements. Activation of IRPs augments expression of the transferrin receptor, thus facilitating iron entry into cells, but decreases expression of the iron storage protein ferritin.^{47 - 48} Glyceraldehyde-3-phosphate dehydrogenase (GAPDH) also displays nonglycolytic functions. During apoptotic cell death, GAPDH translocates to the nucleus.^{49 - 50} Antisense oligonucleotides that deplete GAPDH prevent this nuclear translocation and reduce cell death.^{49 ,51 - 52} These events appear to reflect a novel signaling pathway involving nitric oxide as an initial trigger to cell death (Figure 2).⁵²

Nitric oxide is derived from the amino acid arginine following catalytic activity of nitric oxide synthase (NOS), which abstracts nitric oxide from the guanidino group of arginine, giving rise to citrulline as a by-product (Figure 2). There are 3 distinct forms of NOS derived from 3 different genes.⁵³

Neuronal NOS is the predominant form in the brain.^{54 - 56} It is not inducible but instead is rapidly activated by calcium via the accessory protein calmodulin.⁵⁷ In this way, generation of nitric oxide can take place following neuronal depolarization, which is associated with rapid calcium influx.

Endothelial NOS occurs predominantly in blood vessels and, like neuronal NOS, is a constitutive enzyme that is stimulated by calcium-calmodulin, affording rapid, transient activation.⁵⁸ More prolonged activation of endothelial NOS occurs in response to shear stress associated with turbulent blood flow, which, via the phosphoinositide 3 kinase cascade, leads to phosphorylation and long-term activation of endothelial NOS by the enzyme Akt (protein kinase B).^{59 - 60}

Inducible NOS (iNOS) exists at very low tissue levels under basal conditions, but massive new synthesis of iNOS occurs in response to almost any form of cell stress, especially inflammatory stimuli.^{53 ,61} In this way, the behavior of iNOS closely resembles that of HO-1.⁶²

Nitric oxide signals via 2 principal systems. In one mode, nitric oxide activates guanylyl cyclase to form guanosine 3′,5′-cyclic monophosphate, which accounts for its ability to relax smooth muscle and dilate blood vessels. A more recently appreciated signaling mode for nitric oxide is called S-nitrosylation, whereby nitric oxide covalently modifies cysteine in target proteins by forming a S-nitrosothiol group.^{63 - 64} This process is reversible spontaneously or by selective denitrosylation enzymes.

Targets for S-nitrosylation include a wide range of proteins, such as the sodium pump, sodium-potassium ATPase, the microtubule protein alpha-tubulin, the N-methyl-D-aspartate subtype of receptor for actions of the excitatory neurotransmitter glutamate, and GAPDH.^{52 ,65} Evidence for physiologic S-nitrosylation comes from experiments showing that these proteins are no longer S-nitrosylated in the brains of mice with targeted deletion of nNOS.⁶⁴ Recent studies indicate that a wide range of apoptotic cell stressors acting on many different cell types consistently induce iNOS and lead to S-nitrosylation of GAPDH, triggering its nuclear translocation.⁵²

Proteins such as GAPDH, which move from the cytoplasm to the nucleus of cells, typically contain a “nuclear localization signal” that enables the importin system of the nuclear membrane to grant them entry. However, GAPDH lacks a nuclear localization signal, which raises the question of how it gets into the nucleus. Recent studies show that S-nitrosylation of GAPDH confers on it the ability to bind to a protein called Siah1, which possesses a nuclear localization signal.

Siah1 is an E3-ubiquitin-ligase, a member of a class of enzymes that mediate the ubiquitin-associated degradation of proteins in the proteasome, and is well known to elicit cell death by causing the degradation of selected nuclear proteins.^{66 - 68} When S-nitrosylated GAPDH binds to Siah1, the 2 proteins enter the nucleus. Normally Siah1 turns over very rapidly, but after binding to GAPDH its turnover is slowed down, enabling it to more efficiently degrade its nuclear substrates, thus leading to cell death.⁵² Evidence for the GAPDH-Siah1 apoptotic cascade includes experiments in which selected mutations that prevent GAPDH from binding to Siah1 prevent nuclear translocation of GAPDH.⁵² Similarly, inhibition of iNOS and depletion of GAPDH or Siah1 from cells by RNA interference techniques prevent cell death.⁵²

The iNOS-GAPDH-Siah1 cell death cascade provides a unique pathway extending from the external surface membrane of cells to the nucleus. Diverse cell stressors all have in common the ability to induce iNOS, leading to the formation of nitric oxide, which S-nitrosylates GAPDH. S-nitrosylated GAPDH binds to Siah1, which escorts it to the nucleus. In the nucleus, GAPDH stabilizes the rapidly turning over Siah1, enabling it to use its E3-ubiquitin-ligase activity to degrade nuclear target proteins and bring about apoptotic cell death.

iNOS-GAPDH-Siah1: Clinical Implications. Though only recently identified, the iNOS-GAPDH-Siah1 pathway already has demonstrated clinical ramifications. Huntington disease is caused by a dominant genetic abnormality leading to the formation of a mutant form of the protein huntingtin.^{69 - 70} Huntington disease is associated with an expansion of repeats of the nucleotide triplet CAG, which codes for glutamine so that mutant huntingtin (mHtt) possesses 30 to 200 glutamine repeats.

Abundant evidence indicates that neuronal damage in Huntington disease derives from the translocation to the nucleus of the N-terminal fragment of mHtt.^{71 - 78} We recently showed that this translocation is mediated via the binding of mHtt to GAPDH.⁷⁹ Molecular manipulations that prevent this binding in neuronal cell lines prevent nuclear translocation of mHtt and prevent neurotoxicity. A genetic association study recently found an association between GAPDH and late-onset Alzheimer disease, suggesting that GAPDH variants may predispose to dementia.⁸⁰

These findings have clear therapeutic implications. For example, drugs that block the binding of GAPDH to Siah1 would be anticipated to prevent cell death. Interestingly, such drugs already appear to exist. The monoamine oxidase inhibitor deprenyl has been used in the treatment of Parkinson disease for many years.^{81 - 82} It had been assumed that deprenyl acts by simply increasing brain levels of dopamine, hence correcting the dopamine deficiency of the disease in a fashion analogous to that exerted by levodopa. However, in animal studies deprenyl has been found to be neuroprotective, protecting against the loss of dopamine neurons elicited via various dopamine neuronal toxins.⁸³

Human studies have also suggested that deprenyl has neuroprotective effects, though some controversy exists as to whether the benefit is symptomatic or neuroprotective.^{84 - 85} To aid in distinguishing the inhibitory effect of monoamine oxidase from the neuroprotective effects of deprenyl, very potent derivatives have been synthesized that do not inhibit monoamine oxidase. One of these, TCH346 (CGP3466), exerts neuroprotective effects even at concentrations as low as 1 nanomolar.^{86 - 92} TCH346 binds to a single brain protein, identified as GAPDH,^{86 ,93} and deprenyl and TCH346 both prevent the binding of GAPDH to Siah1.⁹⁴ The therapeutic applicability of drugs such as deprenyl and TCH346 may extend well beyond Parkinson disease and may be relevant to many nonneurologic as well as neurologic conditions in which cell death is a mediator of clinical pathologic processes.

Cytochrome c and Calcium Release in Apoptotic Cell Death. Inositol-1,4,5-trisphosphate (IP3) is one of the 2 principal second messengers of biology, the other being cyclic adenosine monophosphate (AMP). When neurotransmitters or hormones act on cell-surface membrane receptors, they activate phospholipase C, which cleaves lipid membranes to generate IP3, which in turn then acts on specific receptors located on the endoplasmic reticulum, leading to the release of intracellular calcium.⁹⁵ In addition to its role in physiological intracellular signaling, calcium release elicited by IP3 can also lead to apoptotic cell death. Release of intracellular calcium causing apoptosis can arise from a variety of sources in addition to IP3 receptors, as damage to cell membranes and opening of cell membrane calcium channels may also be involved.⁹⁶

In some types of cell death, mitochondria become overloaded with calcium, activating a mitochondrial opening referred to as the permeability transition pore. This opening is associated with permeabilization of the mitochondria, leading to the release of a variety of proteins that activate cell death.^{97 - 98} Most prominent among these is cytochrome c, a small, heme-containing protein of the electron transport chain. Cytochrome c binds to a protein called apoptotic protease activating factor 1 (APAF-1), which recruits one of the caspase enzymes, caspase-9, to initiate a sequence of cleavage and activation of other caspases that ultimately results in cell death.⁹⁹

Recent studies indicate an intimate link between cytochrome c and IP3-elicited release of calcium, reflecting a novel cell death pathway^{100 - 101} that might be approached therapeutically (Figure 3). Thus, when cytochrome c is released from mitochondria in the initial stages of apoptosis, it passes to the closely adjacent membrane of the endoplasmic reticulum and physiologically binds IP3 receptors there. Extremely low concentrations of cytochrome c block a feedback system in which released calcium normally inhibits the ability of IP3 to provoke further calcium release. Blocking this feedback system leads to unrestrained release of calcium. This calcium enters mitochondria to provoke further release of cytochrome c. This feed-forward system spreads throughout the cell, leading to cell-wide coordinated release of cytochrome c.¹⁰⁰

Cytochrome c–IP3 Receptor: Clinical Implications. The cytochrome c–IP3 binding system may provide novel therapeutic strategies. It has been possible to map the exact amino acids in the IP3 receptor responsible for binding to cytochrome c. A cell-permeable peptide derived from this sequence prevents the binding of cytochrome c to IP3 receptors and prevents cell death in various cell lines.¹⁰¹ This peptide could represent a novel therapeutic approach to preventing cell death, but clinicians may also readily screen for small-molecule drugs to block the binding of cytochrome c to the IP3 receptor. Such drugs might be beneficial in preventing cell death in conditions such as stroke, in which excessive cellular calcium is associated with cell death. The IP3 receptor was also recently shown to interact with the antiapoptotic protein, Bcl-2, underlining its role in regulating cell death.¹⁰²

Mechanisms and Potential Drug Targets. In some conditions, eg, neurodegenerative conditions such as Alzheimer, Huntington, and Parkinson diseases, the relative roles of necrosis and apoptosis are controversial. Apoptosis is currently better understood, and several molecular pathways have been characterized, consistent with a need for redundancy in such a critical aspect of cellular disposition.

Less molecular characterization has been achieved for necrotic death. One necrotic pathway involves activation of poly(adenosine diphosphate [ADP]-ribose) polymerase (PARP),^{103 - 104} a nuclear enzyme that adds poly(ADP-ribose) groups both to itself and to nuclear proteins such as histones. Nicotinamide adenine dinucleotide (NAD) is the donor of the ADP-ribose groups for this process. The massive activation of PARP that occurs during necrosis leads to a depletion of NAD and, secondarily, to a depletion of adenosine triphosphate (ATP) so that cells die of energy deficit.^{105 - 106} Targeted deletion of PARP-1, the major nuclear form of this enzyme, greatly reduces stroke infarct size,¹⁰⁷ pancreatic islet cell damage in models of diabetes,¹⁰⁸ and damage from myocardial infarction.¹⁰⁹

Necrotic Death and PARP: Clinical Implications. There are more than 2 dozen PARP inhibitor drugs, several of which have shown efficacy in animal models for treating stroke, Parkinson disease, myocardial infarction, ischemic colitis, and other conditions.¹⁰⁴ Two of these compounds, INO-1001 and AG140699, are currently being tested in human phase 1 or phase 2 trials for a number of applications, including treatment of myocardial infarction and prevention of complications associated with surgical repair of abdominal aortic aneurisms. Although there are likely multiple undiscovered molecular mechanisms by which cells may die of necrosis, PARP inhibitor drugs are attractive candidates for attenuating cell injury in stroke, myocardial infarction, and other disease states.

Therapeutic agents acting via cell death and cell protective pathways are likely to become increasingly prominent in the next few decades. Prevention of cell death will be of importance in stroke, myocardial infarction, and neurodegenerative conditions, while highly selective augmentation of cell death will be sought in anticancer drugs.^{110 - 112} Apoptosis is a normal physiologic event in organs such as the developing thymus, brain, and bone marrow. Antiapoptotic drugs might disrupt this normal cell death, and their use may have to be avoided during critical periods of development or limited to acute phases of disease pathogenesis. Neurodegenerative diseases might require chronic treatment with antiapoptotic drugs, necessitating careful monitoring of adverse effects. Drugs that promote apoptosis may have the adverse effect of killing healthy cells, requiring exquisite dosing and follow-up as is done with cancer chemotherapeutic agents that are similarly toxic to healthy cells.

The cytoprotective actions of bilirubin derive from its antioxidant properties and would presumably be relevant to both necrotic and apoptotic instances of cell death. Drugs that prevent cytochrome c–IP3 receptor binding and the iNOS-GAPDH-Siah1 cascades would presumably be relevant only for apoptotic cell death, the only form for which these 2 pathways have been implicated.