PARP-1 (Poly-ADP Ribose Polymerase) dependent cell death is critically important in a number of experimental disease models including diabetes, inflammation, arthritis, myocardial infarction, Parkinson’s disease, and cerebral ischemia. The mechanism of PARP-1-dependent cell death has been proposed to be due to energy depletion; however, data supporting this hypothesis is not conclusive. Our paper identifies an alternate cell death program involving apoptosis-inducing factor (AIF). We and others have been observing cell death that utilizes some, but not all, of the death machinery described in models of classic apoptosis. In some cases, investigators observed what appeared to be caspase-independent cell death. Some argued that the relevant caspase had just not yet been discovered, while others questioned whether there might be other specialized forms of cell death beyond the dichotomy of necrosis or apoptosis. In the nervous system this is quite important. There have been many disparate observations made in excitotoxicity and experimental stroke models. For many years neuronal death due to glutamate excitotoxicity had been classified as necrotic. However, investigators have observed the appearance of biochemical markers associated with apoptosis—including Annexin-V positive cell membranes, cytochrome c release, caspase activation, nuclear condensation, and DNA fragmentation. Although there are indices of apoptosis, caspase inhibition is not overly effective in blocking neuronal toxicity. The presence of apoptotic markers but the inability to block cell death with inhibitors of apoptosis has confounded neuroscientists. Since PARP-1-dependent cell death is critical in the nervous system, linking PARP-1-mediated toxicity to AIF, a molecule that can initiate the appearance of these biochemical markers in a caspase-independent manner, resolves the apparent conflict. These disparate observations can now be explained by the action of AIF mediating these events in a caspase-independent manner.

Discovering that AIF mediates neurotoxicity in the nervous system provides investigators with another link in the cell death chain to study. Scientists can now ask if the cell death they observe is caspase independent or dependent and determine if, or when, AIF plays a role. In terms of studying PARP-1-dependent cell death it is critically important as it provides a mechanism to test hypotheses about how PARP-1 activation kills cells. The current hypothesis is that PARP-1 activation depletes cells of critical energy pools and this leads to cell death. We can now directly test whether loss of energy is sufficient to trigger AIF translocation and cell death or if other factors are at play?

Many different diseases across organ systems resulting from loss of blood flow, traumatic injury, or inflammatory processes involve activation of the enzyme, poly(ADP-ribose) polymerase-1. Drugs that block poly(ADP-ribose) polymerase-1 activity or genetic deletion of poly(ADP-ribose) polymerase-1 confer profound protection in experimental models of diseases ranging from diabetes and arthritis to heart attack and stroke. How cell death occurs after activation of poly(ADP-ribose) polymerase-1 was not known. We have identified apoptosis-inducing factor (AIF) as the next step in poly(ADP-ribose) polymerase-1-mediated cell death. AIF may be the final executioner of cell death. AIF moves from the mitochondria to the nucleus. After AIF enters the nucleus, DNA is fractured and the nucleus shrinks to a quarter of its normal size. At this point many scientists consider the nucleus to be dead. Now that we know AIF is an important molecule in this cell death pathway we can begin to look for ways to block its release from mitochondria or its entry into the nucleus. We may be able to develop new drugs for the treatment of diseases that involve poly(ADP-ribose) polymerase-1 or AIF.

During a journal club presentation of the Nature paper from Guido Kroemer’s research team describing the identification of AIF, I was struck by the similarity between the morphology of cell death his group observed and the morphology that we had observed following glutamate excitotoxicity. Glutamate excitotoxicity is thought to mediate, in large part, neuronal injury from a wide variety of neurodegenerative diseases, stroke, and trauma. We are interested in how glutamate excitotoxicity kills neurons and previously we had identified nitric oxide/peroxynitrite activation of poly(ADP-ribose) polymerase as a key cell death pathway. To better understand the death process we examined the morphology of the neurons as they died following lethal exposure to a glutamate agonist by electron microscopy and high-resolution fluorescent microscopy. While the final morphology of the neurons resembled necrosis, the morphology of the neurons during the death process did not conform to the published descriptions of either necrosis or apoptosis. The nucleus rapidly shrank and within one hour was one quarter in size. This phenomenon was not observed in neurons treated with poly(ADP-ribose) polymerase inhibitors or in poly(ADP-ribose) polymerase knockout cells. We also had observed caspase activation, but were not able to rescue the neurons with caspase inhibitors. Dr. Kroemer’s paper raised the possibility that AIF could be the "missing link" we were looking for between PARP activation and neuronal death. We generated specific antibodies to AIF and when we activated PARP-1-dependent death in cultured fibroblasts or neurons we saw AIF translocate into the nucleus. In PARP knockout cultures there was no AIF translocation. We characterized the translocation of AIF from the mitochondria to the nucleus and ordered the death pathway in relationship to other known biochemical markers for the death process including phosphatidyl serine exposure, loss of mitochondrial membrane potential, cytochrome c release from mitochondria, activation of caspase-3, and nuclear shrinkage. The final piece of data that indicates AIF is a mediator of PARP-1-triggered cell death and not just a consequence of PARP-1 activation was the blockade of cell death by a neutralizing antibody to AIF.

Since the publication of our findings, other investigators have reported caspase-dependent release of AIF in models of apoptosis. The biochemical pathways that lead to cell death in these models are very different and distinct from the biochemical pathways that lead to cell death in excitotoxicity and PARP-1-dependent cell death. In PARP-1-dependent cell death caspases are activated after the final commitment point of the cell to die. Consistent with this notion is the observation that inhibition of caspases does not promote cell survival. AIF plays a key role in PARP-1-dependent cell death, and blocking its translocation is protective. Caspase-dependent apoptosis and PARP-1-dependent cell death highlight the diversity in cell death signaling and demonstrate the importance of defining the biochemical cell death cascade for each death stimulus and cell type. There is no single pathway that is sufficient to define all forms of cell death.