Do Fungi Undergo Apoptosis-Like Programmed Cell Death?

This question of whether fungi undergo apoptosis-like programmed cell death can be separated into two questions. One question is about applying the term “apoptosis” to fungi, and the other is a more challenging question of whether fungi have evolved mechanisms that inflict self-injury.

manner). If the dying cell makes no contribution to its own death, this death can be considered equivalent to murder-the cell was simply hammered by forces beyond its control (Fig. 1A). An analogous example from mammals might be perforin, which is released from cytotoxic lymphocytes and forms pores in target cell membranes (43). (The NCCD unfortunately refers to death by assault as "accidental" cell death [ACD] [39].) A possible variant of death by assault is the complete absence of any death mechanisms or processes; in such a case, death is defined simply as the absence of life (life simply stops, like a car that runs out of gas). This way of thinking was once prominent but has faded with the accumulation of knowledge about death (Fig. 1A). However, even if there is compelling evidence that a dying cell is capable of contributing to its own death, this is still not sufficient to qualify as "programmed cell death." There is one more hurdle. Theoretically, a microbial toxin could trigger two conceptually distinct types of events in the target cell destined to die; both events require contributions by the dying cell, but only one represents programmed cell death. The distinction between them depends on the answer to the following question: did a gene product or other component originating from the dying cell ЉpurposelyЉ contribute to its own death? If yes, the death fits the long-held definition of "programmed cell death" as currently interpreted (39).

PROGRAMMED VERSUS GENE-DEPENDENT CELL (OR CENTRIFUGE) DEATH
To illustrate that final distinction between programmed and nonprogrammed cell death, take the case of the laboratory centrifuge. Let the centrifuge represent the dying cell, and the centrifuge rotor represent the death-promoting gene product encoded by the dying cell. If the rotor fractures while spinning and wrecks the centrifuge beyond Cell death that occurs without any contributions from the cell that dies is analogous to death by assault or murder (i.e., death is not dependent on a gene or other component originating in the cell destined to die), or, potentially, a cell could expire (like a car running out gas), though enthusiasm for the latter idea has waned in recent years. (B and C) A gene product or other component of the dying cell can contribute to its demise (gene-dependent, cell-autonomous death) in two conceptually distinct processes: accidental/extemporaneous cell death occurring via mechanisms not selected by evolution (B) and programmed cell death occurring via mechanisms selected during evolution (C). Cell death in all three categories (A to C) can be induced by conditions/factors (hammer, toxin, death receptor ligand, growth factor withdrawal, irradiation, sunlight, etc.) external to the cell that dies, while only cell death in categories B and C can also be triggered by events inside the cell, for example, events analogous to those involving the centrifuge rotor (B) or inherent errors in replication (C). These definitions differ somewhat from the NCCD definitions (39).
Perspective ® repair, this would be analogous to a cell's protein actively contributing to the cell's own death. In this case, the centrifuge/cell-killing function of the rotor/protein (e.g., an unleashed protease or nuclease) was not a feature of the rotor/protein design. Cell death analogous to this centrifuge example can be classified as cell-autonomous and gene-dependent death (i.e., a gene or other component originating from the dying cell helps promote self-elimination). Here I refer to this type of cell death as accidental or extemporaneous (metaphorically, "unintended cell death") ( Fig. 1B). However, if the centrifuge rotor had been designed by its engineers to destroy the centrifuge when its time was up, this would be considered programmed death (metaphorically, "deliberate cell death") ( Fig. 1C). Thus, cell-autonomous mechanisms that evolved for cell death, or that can be reasonably anticipated as such, are generally considered to be programmed death mechanisms. Note, however, that the rotor/protein has another "day job" function in the centrifuge/healthy cell.
Importantly, both types of gene-dependent death (programmed and nonprogrammed) are potentially druggable; thus, the question of their evolutionary origins seems less important. Furthermore, distinguishing between accidental/unintended death and deliberate/programmed death may not be feasible experimentally, even in animals. Therefore, a more readily testable definition would be useful. We ask whether the death of a cell is dependent on its own gene products or other components (e.g., proteases, reactive oxygen species [ROS], oxidized lipids, unfolded proteins, etc.), regardless of whether such pro-death functions arose through a selection process during evolution (implying a purposeful death) and regardless of whether they wholly or partly cause cell death. For simplicity, we refer to this as "gene-dependent cell death," without any expectations or requirements that the process would be programmed or regulated ( Fig. 1B and 1C). Death-promoting factors can be either direct executors of death (e.g., mammalian caspase-3) or indirect contributors to death (e.g., death signaling pathways). However, applying the term "gene-dependent death" to fungi still requires further considerations.

CAN WE EXTRAPOLATE FROM ANIMAL KNOCKOUT DATA TO FUNGAL KNOCKOUT DATA?
The evidence suggesting that yeast undergo gene-dependent cell death (PCD, RCD) relies heavily on genetic approaches and the analysis of mutants that are resistant or sensitive to death stimuli. The same powerful genetic approach was used in the landmark C. elegans studies that defined the apoptosis pathway (19), but there is a critical difference. In sharp contrast to yeast, the worm cell death is physiologically relevant by definition-it occurs during normal development. Analogous physiological model systems are limited in availability for the laboratory workhorse Saccharomyces cerevisiae, though there are reports of efforts in this direction (population dynamics, failed mating, quorum sensing, virus infection, sporulation, aging, colony differentiation, and many others) (44)(45)(46)(47)(48)(49)(50)(51).
This limitation raises a looming question. Is it reasonable to conclude that a gene has pro-death activity if deletion of that gene confers death resistance to the cell, or could there be unrelated explanations? This is a troublesome question. In fact, researchers in the larger (non-cell death) yeast genetics field are adamant about this issue in my personal experience, and they have an excellent point worthy of consideration. They argue that the improved survival is likely explained by reasons other than the loss of a death-promoting gene. Consider the death-resistant strains of S. cerevisiae lacking either YCA1/MCA1 (metacaspase related to mammalian caspases) or DNM1/DRP-1 (conserved dynamin-like mitochondrial fission factor implicated as an accessory to mammalian cell death) (52)(53)(54). In both cases, enzymatic activity is required for the death function. However, if these knockout strains also have heightened defenses or stress responses simply as a consequence of their gene deficiencies, this could potentially account for their ability to survive stress better than the wild type. The same applies to metazoans.
This criticism is difficult to overcome, and yet it has its own caveats. Testing all known adaptive stress responses cannot resolve the criticism; there could always be another untested compensatory mechanism. Conversely, even if adaptive stress responses are elevated, and even if these adaptations partly contribute to death resistance, it remains possible that the gene in question had evolved in part to inflict self-harm. Thus, the possibility of the existence of a cell death mechanism cannot be dismissed because stress responses are heightened, just as a cell death mechanism cannot be inferred from the absence of detectable stress responses. Adaptive responses and pro-death functions are not mutually exclusive possibilities and can even be expected to co-occur. For example, careful analyses revealed that depletion of mammalian apoptosis regulators alter antioxidant defenses at the steady state (55). Other potential confounders for studying fungal cell death include the secondary mutations affecting cell death that frequently arise in gene knockout strains (56) and the day job functions of most, if not all, pro-death factors that could also significantly impact susceptibility to stress (57)(58)(59). Our studies of animal caspases and other pro-death factors in the nervous system revealed alternative and nonapoptotic roles (60)(61)(62). Antiapoptotic proteins also have day jobs. How the effects of antiapoptotic mammalian BCL-xL on mitochondria and cellular energetics in healthy cells (57,58,63) are related to its antiapoptotic role (BH3/BAX-inhibition) is not yet known. How are we to address all these issues for fungi?

WHAT IS ACCEPTABLE EVIDENCE OF PROGRAMMED FUNGAL CELL DEATH?
Cell death mechanisms could potentially have evolved in single-cell species to enable them to respond to inevitable pathogens (47) or to live in the differentiated communities of a simple colony harboring layers with distinct transcriptional profiles and death-susceptible subpopulations that likely benefit the community (50,51,64). However, except for a few salient examples supporting the existence of fungal cell death pathways (below), the evidence is limited thus far. Based on the history of the metazoan PCD field, the idea of the existence of PCD mechanisms in fungi will remain controversial and lack general acceptance until molecularly defined fungal death pathways are compellingly demonstrated. This includes identification of the direct effectors of fungal cell death and the biochemical mechanisms involved. Until then, claims that fungi undergo gene-dependent PCD/RCD may be overstated without further clarification. Admittedly, the bar is higher now than it was for mammalian PCD in the mid-20th century.

REASONS TO KEEP OR ABANDON THE TERM "APOPTOSIS" FOR FUNGI
The application of the term ЉapoptosisЉ to fungi is difficult to defend by any definition and has been decisively rejected (2,(65)(66)(67), and yet studies on "fungal apoptosis" continue to populate the literature (40,68,69). This may be an extension of the early metazoan and nonmetazoan literature that was published when ЉapoptosisЉ was the only relevant word available and served as a blanket term to describe any type of regulated cell death. However, this practice has declined in recent years to separate apoptosis from several other molecularly defined nonapoptotic death pathways in metazoans (notably necroptosis, ferroptosis, and pyroptosis).
Thus, apoptosis is now more narrowly defined, demanding consideration of authors' intentions before citing their earlier works as evidence of "apoptosis" as currently defined. The current working definition of mammalian "apoptosis" is caspase-3mediated cell death (39), and yet fungi do not encode true caspases. Therefore, ЉapoptosisЉ applied to fungi presumably has a different definition. Although a defensible consensus definition has not emerged, several criteria have been reported (40). Related fungal metacaspases appear not to behave like mammalian caspases or to be regulated like caspases (70)(71)(72). Thus, the use of promiscuous mammalian caspase reporters (e.g., FITC-VAD-FMK) as evidence of "caspase-like" activities in fungi currently lacks rigorous justification (73), and newer better reagents may be helpful (74). However, these facts do not in any way rule out a role for fungal metacaspases or other proteases in cell death, and the possibility remains open.
Is the term "yeast apoptosis" justifiable in other ways? Some justify the use of the term ЉapoptosisЉ (or Љapoptosis-likeЉ) by restricting its use to morphological and biophysical features of dying fungal cells without further implications. Indeed, the word ЉapoptosisЉ was put forth in a landmark paper in 1972 (75) without any direct evidence for programmed cell death, supported only by the distinct morphological features of occasional cells in normal human tissues and in rat liver at delayed times after ischemic injury, though the claims have not been free of controversy (76). However, unlike fungi, the apoptotic features of mammalian cells are now recognized as the handiwork of caspase-3, including chromatin condensation, membrane blebbing, and phosphatidylserine exposure on the cell surface to facilitate engulfment into phagocytic cells (where some cells may finish dying) (77)(78)(79)(80). Accordingly, the field has advanced. Importantly, apoptosis-like features of mammalian cells not caused by caspases are no longer classified as a type of apoptosis in the latest NCCD guidelines (39). New biochemical or molecular validations may eventually justify the use of the term "apoptosis-like" for fungal death, though it is not presently clear if apoptosis-like fungal death is more like apoptosis than like any other known death mechanism. Perhaps the best justification of the use of the term Љfungal apoptosisЉ is according to the original definition of apoptosis-for use not as a morphological term but to conceptualize the idea of a deliberate cell-autonomous death mechanism. However, an explicit explanation of this otherwise retired definition will be required to avoid the inevitable misunderstandings that can compromise a field's credibility.

CONSERVED DEATH PROGRAMS IN FUNGI AND HUMANS?
The new focus is on mammalian cell death by programmed necrosis resulting in plasma membrane rupture. One potentially conserved necrotic cell death mechanism is represented by the proposed N-terminal pore-forming domain of mammalian MLKL, mediator of necroptosis (81). The HET or HeLo-like (HELL) domains of filamentous fungi that mediate cell death upon fusion of two incompatible cells (heterokaryon incompatibility) (82,83) have an MLKL-like structure prediction (e.g., Phyre2). A different form of death occurs in conidia of the agriculturally important plant pathogen Magnaporthe oryzae (rice blast) during germination (84). Deletion of any of 16 conserved autophagy genes causes conidia to remain alive, blocking both appressorium formation and pathogenicity (85), raising the possibility of autophagy-dependent cell death (ADCD) (86,87). However, more studies are needed to rule out indirect effects of autophagy (not true ADCD), for example, by degrading an inhibitor of the primary death effector (39). One study suggested that fungi may also undergo iron-dependent ferroptosis, which is thought to result from loss of membrane integrity as a consequence of lipid peroxidation (88), conceivably shared across many species.
Fungi may also encode inhibitors of programmed cell death. Conserved BIR (baculovirus inhibitor of apoptosis repeat) domains that potently suppress cell death were first identified in insect virus genomes and later in humans and Drosophila (89). Interestingly, BIR-containing proteins of the plant-pathogenic fungus Botrytis cinerea and the human-pathogenic fungus Aspergillus fumigatus were found to suppress fungal cell death and increase virulence (68,90). Although the mechanisms are not known, the suggestion that BIR1 suppresses caspase-dependent apoptosis-like fungal cell death is debatable (73), and with rare exceptions, BIR-containing proteins are not direct caspase inhibitors (72,91). Continuing to stretch mammalian cell death nomenclature to accommodate fungi requires changing ingrained assumptions, and the field may be better served by new nomenclature to convey new discoveries.