Classification of cell death

Recommendation to Authors, Reviewers and Editors of Scientific Journals

Authors still make frequent use of expressions like ‘percentage apoptosis’ without mentioning the method actually employed to assess ongoing cell death. In a totally inappropriate fashion, these terms are also employed to describe the results of cell-free assays based on purified cellular components. Such a vocabulary is confusing and imprecise and should definitively be abandoned. From 2009, Cell Death and Differentiation will actively enforce a policy in which terms like ‘percent apoptosis’, ‘percent necrosis’, ‘percent cell death’ and ‘percent cell survival’ must be replaced with more descriptive expressions including ‘percent cells with condensed chromatin’, ‘percent cells with DNA fragmentation’, ‘percent cells with a low mitochondrial transmembrane potential (ΔΨ_m)’, as well as ‘percent propidium iodide (PI) positive’, ‘percent cleaved caspase-3 positive’, ‘percent terminal deoxynucleotidyl transferase-mediated dUTP nick end labeling (TUNEL) positive’ and ‘percent clone forming’ cells. Similarly, the term ‘percent autophagic cells’ should be avoided and replaced with a language that precisely describes what has been measured (such as ‘number of GFP-LC3 puncta/cell’ or ‘percent vacuolated cells’). This applies to the description of experimental results, be it in the text or in the abstract, as well as to the labeling of figures and figure legends. Moreover, NCCD encourages researchers to quantify cell death and/or other catabolic events (such as autophagy) with more than one assay, whenever possible, thereby reducing the probability of artifacts. The NCCD urges all life science journals and, more specifically, all journals in the areas of cell biology, cancer research and pharmacology to pursue a similar policy.

When is a Cell ‘Dead’?

Dying cells are engaged in a process that is reversible until a first irreversible phase or ‘point-of-no-return’ is trespassed (Table 1). It has been proposed that this step could be represented by massive caspase activation,⁶ loss of ΔΨ_m,⁷ complete permeabilization of the mitochondrial outer membrane⁸ or exposure of phosphatidylserine (PS) residues that emit ‘eat me’ signals for normal neighboring cells. However, there are dozens of examples in which caspases are activated in the context of non-lethal processes and differentiation pathways.^9,10 The ΔΨ_m can be dissipated by protonophores without progression to immediate cell death.¹¹ PS exposure can be reversible, for instance in neutrophilic granulocytes.¹² Thus, the concept of a restriction point for cell death, as it was described by Pardee¹³ for the cell cycle, has yet to be specifically defined.

In the absence of a clearly defined biochemical event that can be considered as the point-of-no-return, the NCCD proposes that a cell should be considered dead when any one of the following molecular or morphological criteria is met: (1) the cell has lost the integrity of its plasma membrane, as defined by the incorporation of vital dyes (e.g., PI) in vitro; (2) the cell, including its nucleus, has undergone complete fragmentation into discrete bodies (which are frequently referred to as ‘apoptotic bodies’); and/or (3) its corpse (or its fragments) has been engulfed by an adjacent cell in vivo. Thus, bona fide ‘dead cells’ would be different from ‘dying cells’ that have not yet concluded their demise (which can occur through a variety of biochemically distinct pathways, see below). In particular, cells that are arrested in the cell cycle (as it occurs during senescence) should be considered as alive, and the expression ‘replicative cell death’ (which alludes to the loss of clonogenic potential), as it is frequently used by radiobiologists, should be abandoned.

Considerations on ‘Autophagy’ and ‘Autophagic Cell Death’

Macroautophagy is characterized by the sequestration of cytoplasmic material within autophagosomes for bulk degradation by lysosomes. Autophagosomes, by definition, are two-membraned and contain degenerating cytoplasmic organelles or cytosol,^35,36 which allows them to be distinguished by transmission electron microscopy from other types of vesicles such as endosomes, lysosomes or apoptotic blebs.³ The fusion between autophagosomes and lysosomes generates autolysosomes, in which both the autophagosome inner membrane and its luminal content are degraded by acidic lysosomal hydrolases. This catabolic process marks the completion of the autophagic pathway. When the fusion of autophagosomes with lysosomes is blocked, the former accumulate in spite of autophagy inhibition.^36,37 Hence, a massive increase in the number of autophagosomes is by no means a demonstration that the autophagic pathway is induced, and functional tests are required to investigate autophagy. A very comprehensive description of the assays for monitoring autophagy in higher eukaryotes and a set of guidelines for their interpretation has been recently provided by Klionsky et al.³⁸ One technique commonly employed to detect autophagy relies on the redistribution of GFP-LC3 fusion proteins into vesicular structures (which can be autophagosomes or autolysosomes).^3,39 However, the exclusive use of GFP-LC3 as a marker of autophagy is not sufficient to diagnose an enhanced autophagic catabolism.³⁸

‘Autophagic cell death’ is morphologically defined (especially by transmission electron microscopy) as a type of cell death that occurs in the absence of chromatin condensation but accompanied by massive autophagic vacuolization of the cytoplasm. In contrast to apoptotic cells (whose clearance is ensured by engulfment and lysosomal degradation), cells that die with an autophagic morphology have little or no association with phagocytes.^40,41 Although the expression ‘autophagic cell death’ is a linguistic invitation to believe that cell death is executed by autophagy, the term simply describes cell death with autophagy.^15,16,42 Thus far, involuting Drosophila melanogaster salivary glands provide the only in vivo evidence that the knockdown/knockout of genes required for autophagy truly reduces cell death.⁴³ This may be due to the limited number of studies that have investigated autophagic cell death in vivo, although there are no doubts that autophagy promotes cell survival, in multiple physiological and experimental settings.⁴⁴ Significantly, some reports indicate that cells presenting features of ‘autophagic cell death’ can still recover upon withdrawal of the death-inducing stimulus.⁴⁵ In most cases described to date in which autophagy is suppressed by genetic knockout/knockdown of essential autophagy (atg) genes, cell death is not inhibited but rather occurs at an accelerated pace,¹⁵ pointing to the prominent role of autophagy as a pro-survival pathway. This said, it should be noted that most of these studies have been performed on immortalized cell lines in vitro and that autophagic cell death rarely affects individual cells in vivo.^19,41 Nevertheless, in specific cases, autophagy may participate in the destruction of cells, as a result of a protracted atrophy of the cytoplasm, beyond a not yet clearly defined point-of-noreturn.^43,46 Thus, direct induction of autophagy by overexpression of the Atg1 kinase is sufficient to kill fat and salivary gland cells in Drosophila. Interestingly, although Atg1-driven autophagic cell death entails caspase-dependent mechanisms in fat cells,⁴⁶ the same does not hold true in salivary gland cells (which cannot be rescued from Atg1-induced death by p35 expression).⁴³

Definition of ‘Necrosis’

‘Necrotic cell death’ or ‘necrosis’ is morphologically characterized by a gain in cell volume (oncosis), swelling of organelles, plasma membrane rupture and subsequent loss of intracellular contents. For a long time, necrosis has been considered merely as an accidental uncontrolled form of cell death, but evidence is accumulating that the execution of necrotic cell death may be finely regulated by a set of signal transduction pathways and catabolic mechanisms.^4,47 For instance, death domain receptors (e.g., TNFR1, Fas/CD95 and TRAIL-R) and Toll-like receptors (e.g., TLR3 and TLR4) have been shown to elicit necrosis, in particular in the presence of caspase inhibitors. TNFR1-, Fas/CD95-, TRAILR- and TLR3-mediated cell death seemingly depends on the kinase RIP1,⁴⁸ as this has been demonstrated by its knockout/knockdown and chemical inhibition with necrostatin-1.^49-51 Although there is no generalized consensus on the use of this expression, some authors have proposed the term ‘necroptosis’ to indicate regulated (as opposed to accidental) necrosis. At a biochemical level, necroptosis may be defined as a type of cell death that can be avoided by inhibiting RIP1 (either through genetic or pharmacological methods),^49,50 which may represent a convenient means to discriminate between programmed and fortuitous forms of necrosis.

Several mediators, organelles and cellular processes have been implicated in necrotic cell death, but it is still unclear how they interrelate with each other. The causative elements of necrosis are unclear, as well as its bystander effects. These phenomena include mitochondrial alterations (e.g., uncoupling, production of reactive oxygen species, i.e., ROS, nitroxidative stress by nitric oxide or similar compounds⁵² and mitochondrial membrane permeabilization, i.e., MMP, often controlled by cyclophilin D), lysosomal changes (ROS production by Fenton reactions, lysosomal membrane permeabilization), nuclear changes (hyperactivation of PARP-1 and concomitant hydrolysis of NAD⁺), lipid degradation (following the activation of phospholipases, lipoxygenases and sphingomyelinases), increases in the cytosolic concentration of calcium (Ca²⁺) that result in mitochondrial overload and activation of non-caspase proteases (e.g., calpains and cathepsins).^4,53 In several (but not all) instances of necrotic cell death, a crucial role for the serine/threonine kinase RIP1 has been demonstrated.⁵⁴ Thus far, however, there is no consensus on the biochemical changes that may be used to unequivocally identify necrosis. In the absence of a common biochemical denominator, necrotic cell death is still largely identified in negative terms by the absence of apoptotic or autophagic markers, in particular when the cells undergo early plasma membrane permeabilization (as compared with its delayed occurrence, which is associated with late-stage apoptosis). For these reasons, caution should be used in classifying particular cell death routines as necrotic.

Definition of ‘Cornification’

Cornification is a very specific form of PCD that occurs in the epidermis, morphologically and biochemically distinct from apoptosis. It leads to the formation of corneocytes, that is dead keratinocytes containing an amalgam of specific proteins (e.g., keratin, loricrin, SPR and involucrin) and lipids (e.g., fatty acids and ceramides), which are necessary for the function of the cornified skin layer (mechanical resistance, elasticity, water repellence and structural stability). Cornification is less often referred to as ‘keratinization’ or ‘cornified envelope formation’,^17,55 and it is generally considered as a terminal differentiation program similar to those leading to other anucleated tissues (such as the lens epithelium and mature red blood cells).^56,57 This is mainly due to the fact that these processes display the (often limited) activation of the molecular machinery for cell death, in particular of caspases.^9,10,57,58 In contrast with corneocytes, however, both mature red blood and lens epithelial cells retain the ability to undergo stress-induced death,^59,60 and hence only cornification should be regarded as a bona fide cell death program.

At the molecular level, cornification follows a specific mechanism of epithelial differentiation during which cells express all enzymes and substrates required for building up the epidermal barrier that allows for isolating the body from the external environment. This is obtained by the crosslinking enzymes (e.g., transglutaminase types 1, 3 and 5) acting on several substrates (e.g., loricrin, SPR, involucrin and SP100),¹⁸ as well as through the synthesis of specific lipids that are released into the extracellular space (where they are covalently attached to cornified envelope proteins), and proteases, which are required for impermeability and desquamation, respectively.

Tentative Definitions of Atypical Cell Death Modalities

Postface

As it stands, three distinct routes of cellular catabolism can be defined according to morphological criteria, namely apoptosis (which is a form of cell death), autophagy (which causes the destruction of a part of the cytoplasm, but mostly avoids cell death) and necrosis (which is another form of cell death). Although frequently employed in the past, the use of Roman numerals (i.e., type I, type II and type III cell death, respectively) to indicate these catabolic processes should be abandoned. Moreover, several critiques can be formulated against the clear-cut distinction of different cell types in the triad of apoptosis, autophagic cell death and necrosis.

First, although this vocabulary was originally introduced based on observations of developing animals,^41,90 it has rapidly been adopted to describe the results of in vitro studies performed on immortalized cell lines, which reflect very poorly the physiology of cell death in vivo. In tissues, indeed, dying cells are usually engulfed well before signs of advanced apoptosis or necrosis become detectable. Thus, it may be acceptable - if the irreversibility of these phenomena is demonstrated - to assess caspase activation and/or DNA fragmentation to diagnose apoptotic cell death in vivo.

Second, there are numerous examples in which cell death displays mixed features, for instance with signs of both apoptosis and necrosis, a fact that lead to the introduction of terms like ‘necroapoptosis’ and ‘aponecrosis’ (whose use is discouraged by the NCCD to avoid further confusion).⁵³ Similarly, in the involuting D. melanogaster salivary gland, autophagic vacuolization is synchronized with signs of apoptosis,⁹¹ and results from genetic studies indicate that caspases and autophagy act in an additive manner to ensure cell death in this setting.⁴³ Altogether, these data argue against a clear-cut and absolute distinction between different forms of cell death based on morphological criteria.

Third (and most important), it would be a desideratum to replace morphological aspects with biochemical/functional criteria (Table 3) to classify cell death modalities. Unfortunately, there is no clear equivalence between morphology and biochemistry, suggesting that the ancient morphological terms are doomed to disappear and to be replaced by truly biochemical definitions. In this context, ‘loss-of-function’ and ‘gain-of function’ genetic approaches (e.g., RNA interference, knockout models and plasmid-driven overexpression systems) represent invaluable tools to characterize cell death modes with more precision, but only if such interventions truly reduce/augment the rate of death, instead of changing its morphological appearance (as it is often the case). Present cell death classifications are reminiscent of the categorization of tumors that has been elaborated by pathologists over the last one and a half centuries. As old morphological categorizations of tumors are being more and more supported (and will presumably be replaced) by molecular diagnostics (which allows for a more sophisticated stratification of cancer subtypes based on molecular criteria), the current catalog of cell death types is destined to lose its value as compared with biochemical/functional tests. In the end, such efforts of classification are only justified when they have a prognostic and/or predictive impact, allowing the matching of each individual cancer with the appropriate therapy. Similarly, a cell death nomenclature will be considered useful only if it predicts the possibilities to pharmacologically/genetically modulate (induce or inhibit) cell death and/or if it predicts the consequences of cell death in vivo, with regard to inflammation and recognition by the immune system.