A novel mitochondrial regulon for ferroptosis during fungal pathogenesis

ABSTRACT

Ferroptosis remains an underexamined iron- and lipid peroxides-driven cell death modality despite its importance to several human and plant diseases and to immunity thereof. Here, we utilized chemical cell biology, molecular genetics and biochemical analyses to gain insights into how the fungal pathogen Magnaporthe oryzae undergoes ferroptosis strictly in the spore cells to successfully transit to infectious development. We reveal a complex functional interdependency and crosstalk between intrinsic ferroptosis and autophagy-mediated mitochondrial degradation. Mechanistically, the requirement of mitophagy for ferroptotic cell death was attributed to its ability to maintain a pool of metabolically active mitochondria. Pharmacological disruption of the electron transport chain or membrane potential led to complete inhibition of ferroptosis, thus simulating the loss of mitophagy phenotypes. Conversely, increased mitochondrial membrane potential in a mitophagy-defective mutant alleviated the ferroptosis defects therein. Graded inhibition of mitochondrial coenzyme Q biosynthesis with or without ferroptosis inhibitor liproxstatin-1 distinguished its antioxidant function in such regulated cell death. Membrane potential-dependent regulation of ATP synthesis and iron homeostasis, as well as dynamics of tricarboxylic acid cycle enzyme AcoA (aconitase A) in the presence or absence of mitophagy, mitochondrial poisoning or iron chelation further linked mitochondrial metabolism to ferroptosis. Last, we present an important bioenergetics- and redox-based mitochondrial regulon essential for intrinsic ferroptosis and its precise role in fungal pathogenesis leading up to the establishment of the devastating rice blast disease.

Abbreviation: 4-CBA: 4 chlorobenzoic acid; AcoA: aconitase A; Atg24: autophagy related 24; CoQ: coenzyme Q; CPX: ciclopirox olamine; ETC: electron transport chain; GSH: glutathione; Gpx4: glutathione peroxidase 4; HPI: hours post inoculation; MMP: mitochondrial membrane potential; MitoQ: Mitoquinone; ROS: reactive oxygen species; TCA: tricarboxylic acid

Introduction

Ferroptosis is an evolutionarily conserved cell demise modality caused by iron-dependent peroxidation of membrane lipids that contain polyunsaturated fatty acid tail(s) [1–4]. Such nonapoptotic cell death was first reported in and/or well-studied in cancer cells, or engineered mammalian cell lines, and can be induced in several types of tumor cells through depletion of cellular glutathione (GSH) or inhibition of GSH-dependent GPX4 (glutathione peroxidase 4) that directly converts the lipid hydroperoxides into nontoxic lipid alcohols [5,6]. A small molecule, erastin, causes GSH depletion and ferroptosis induction by inhibiting cellular uptake of cystine, which upon conversion to cysteine in vivo is used to synthesize GSH [6]. In tumor cells that are insensitive to GPX4 inhibition, a second ferroptosis resistance factor, i.e. AIFM2/FSP1 (AIF family member 2/ferroptosis suppressor protein 1) [7,8] is post-translationally modified and recruited to the plasma membrane where it reduces coenzyme Q (CoQ)) [7,8] and vitamin K [9] so as to stop phospholipid peroxidation via their radical-trapping antioxidant activities. STARD7 was subsequently identified as the CoQ transporter that shuttles between mitochondria and cytosol [10], and thus unmasks the source of CoQ located near the plasma membrane.

Ever since the term ferroptosis was coined, punctate/smaller mitochondria with increased membrane density were observed using transmission electron microscopy as the lone morphological marker [3], but such morphological changes and the fate of abnormal mitochondria remained unresolved. Ferroptosis can be induced by erastin or the GPX4 inhibitor RSL3 in 143B osteosarcoma cells devoid of mitochondria DNA [3], thus questioning the importance of respiration-based reactive oxygen species (ROS) as the primary source of lipid peroxidation. Given the significant involvement of mitochondria in cellular iron and lipid metabolism [11,12], it became important to investigate the mitochondria-ferroptosis connection from these two perspectives. Conversely, tricarboxylic acid (TCA) cycle and electron transport chain (ETC) have been found to enable ferroptosis in mouse embryonic fibroblasts but only when it is specifically induced by cysteine starvation and not by GPX4 inhibition [13]. It remains unclear as to why context-dependent requirement of intact mitochondria for ferroptosis is evident in different cell types. One possibility could be that tumor cells, and probably also other cultured cell lines, modify cell metabolism in different ways for survival. For example, 143B osteosarcoma cells without mitochondrial DNA can rely on glycolysis for survival as far as the F₁-ATPase capable of hydrolyzing ATP and the reversible ADP/ATP translocase named adenine nucleotide translocator are still functional [14]. Since membrane potential is necessary for import of most precursor proteins into the mitochondria [15] to enable proper metabolism therein, such ρ⁰ tumor cells also maintain sufficient mitochondrial membrane potential (MMP) to sustain cell growth [14]. Thus, a clear and comprehensive understanding of the role of mitochondria in ferroptosis demands efforts in first understanding the regulation of intrinsic ferroptosis in healthy cells.

Rice plants, for example, undergo highly compartmentalized ferroptosis [16], in epidermal cells invaded by the incompatible isolates of the blast fungus. Such strong immune response blocks and then kills the invading fungus within these dead cells. Interestingly, ferroptosis also occurs in the spore/conidium of the rice blast fungus M. oryzae that triggers such host immune response, and represents a crucial determinant of the essential pathogenic development and thus the infection ability leading to the destructive blast disease [2,17]. Particularly, such intrinsic ferroptosis initiates sequentially in the terminal conidial cell followed by the middle and proximal conidial cells [17], although the three interconnected cells in the conidium share the cytoplasm (Figure 1A). Apart from the gradient iron accumulation prior to death in the three cells [17], little is known about the regulators controlling such precise ferroptosis relay in the healthy interconnected cells.

Figure 1.

Regulated mitochondrial fission and degradation precede ferroptosis in M. oryzae conidial cells. (A) programmed ferroptosis occurs sequentially in the 3 connected conidial cells during the pathogenic development. Cells marked 1,2,3 in (A) and (C) are terminal, middle, and proximal conidial cells that initiate ferroptosis first, second, and last, respectively, and time depicted as hours post inoculation (hpi). M. oryzae from 4 h to 24 h is shown in a way that from left to right are the three-celled conidium, germ tube, and appressorium. Red asterisk marks the dead cell, which is also indicated by the disappearance of the nucleus (hH1-GFP, color inverted) in the projection image. DIC images are single plane; scale bar: 5 μm for both (A) and (B). The same experiment has been repeated 3 times with consistent results. (B) mitochondrial fusion, fission, and degradation along with sequential ferroptosis events. Bright field images are single plane and overlapped in the 3D images of mitochondria displayed by a mitochondrial targeting sequence fused with GFP (MTS-GFP). Red asterisks mark collapsed conidial cells. Data shown is representative of 4 replicates of the experiment. (C) sequential mitochondrial degradation in the conidium correlates with ferroptosis therein. Mitochondrial degradation finished first in the terminal cell (6:45 hpi), and then in the middle cell (8:45 hpi) and lastly in the proximal cell (12:00 hpi), precisely occurring before the sequential ferroptosis cycle. Blue rectangles highlight the cells that just completed mitochondrial degradation. V: vacuole. Images in (B) and (C) are from the same time lapse experiment.

Iron that drives ferroptosis in mammalian cells is found to be released by autophagic degradation of ferritin [18,19], which is an iron storage protein complex. However, ferritin is not conserved in Magnaporthe albeit ferroptosis therein being autophagy dependent [17,20]. Thus, the question of whether other types of selective autophagy, such as mitophagy that targets dysfunctional or damaged mitochondria, bridge the gap between ferroptosis and autophagy in fungi remains to be addressed. Here, using time lapse live-cell imaging, we report that mitochondrial fission and then vacuolar degradation occur prior to ferroptosis initiation in the conidium of M. oryzae during pathogenic development. Mitophagy, which mediates selective mitochondrial degradation, is shown to be essential for ferroptosis and plays a crucial role in maintaining metabolically active mitochondria. Systematic investigation of the contributions of mitochondrial metabolism to ferroptosis including CoQ and ATP synthesis, iron homeostasis and fatty acid catabolism provide a comprehensive view of a novel mitochondrial regulon essential for intrinsic ferroptosis and its role in fungal pathogenesis leading to the devastating blast disease in rice. Such knowledge highlights the conserved ferroptosis as a novel target for crop protection, and in turn will help better understand the mitochondria-ferroptosis nexus in human diseases too.

Results

Mitochondrial fission and degradation precede intrinsic ferroptosis-based cell death in M. oryzae

Pathogenesis in M. oryzae is a spatio-temporally controlled developmental process featuring the formation and maturation of the infection structure called the appressorium (Figure 1A). As a critical and programmed step of such development, the 3 connected cells in the asexual spore or conidium undergo precisely controlled death individually and sequentially (Figure 1A), and such regulated cellular demise was unequivocally shown to be ferroptosis in M. oryzae [17]. Interestingly, a similar sequential behavior prior to conidial cell death was observed for mitochondrial degradation, which commenced and completed first in the terminal cell prior to its collapse and death, and then the same two processes repeated subsequently in the middle and proximal cells around 2 and 5 h later, respectively (Figure 1B,C; Figure S1A). In each of the dying conidial cells, mitochondria showed a trend changing from tubular filaments to punctate structures, thus suggesting the involvement of the fission process prior to selective degradation in the vacuole (Figure S1A; Movie S1; https://zenodo.org/records/7943903). The occurrence of such regulated conidial death specifically correlated with and followed precise mitochondrial degradation thus underscoring the importance of this organellar homeostasis in ferroptosis early in the infection cycle.

Mitochondrial degradation via mitophagy is necessary for ferroptosis in M. oryzae

Mitophagy, which targets dysfunctional, damaged or excess mitochondria for vacuolar degradation using the autophagy machinery, is one of the pathways responsible for organellar turnover and homeostasis [21]. Weak free GFP signal in the vacuole (Figure 1C) implied that mitophagy could be an important pathway responsible for the observed mitochondrial degradation. Therefore, we first examined the link between mitophagy and ferroptosis by pharmacological inhibition of mitochondrial fission. Such treatment with mitochondrial division inhibitor 1 (Mdivi-1) led to stably fused mitochondria that persist and were unable to be cleared via mitophagy due to the larger size and/or tubular constraints (Figure 2A), and thus indirectly suppressed mitophagy. Indeed, disrupting mitochondrial fission led to cell death inhibition and consequently increased the conidial viability in a dose-dependent manner (Figure 2A). To confirm the requirement of mitophagy for such conidial cell death, a mitophagy-deficient mutant, atg24Δ [21] was included in the examination. An extensively fused, stable and dense network of mitochondria was evident in atg24Δ conidial cells but rarely seen in the wild-type Magnaporthe conidia wherein the mitophagy was fully functional (Figure 2B). Not surprisingly, cell death was found significantly suppressed in atg24Δ conidia as compared to the wild-type M. oryzae (Figure 2C). We previously found that such cell death, which is essential for pathogenesis M. oryzae, is executed through ferroptosis driven by iron and lipid peroxides [17]. Consistent with these findings, cell death in wild-type conidia was promoted by iron supplementation and suppressed upon iron chelation via ciclopirox olamine (CPX) (Figure 2C). Such iron-dependent cell death was greatly suppressed in atg24Δ conidia as compared to the wild type and could be further reduced via CPX treatment (Figure 2C). On the other hand, lipid peroxides which drive ferroptosis were evident in the dying wild-type conidial cell (cell no. 1) but failed to accumulate along the outer membranes in the atg24Δ conidia (Figure 2D). Together, these data confirmed a lack of ferroptosis in the atg24Δ conidia as compared to the wild-type counterpart. Consequently, atg24Δ mutant displayed a significant defect in infection (Figure 2E) and was incapable of causing blast disease in rice plants (Figure 2F) as compared to wild-type M. oryzae. Taken together, we conclude that mitophagy is necessary for developmental ferroptosis to proceed properly during the pathogenesis cycle in M. oryzae.

Figure 2.

Mitochondrial degradation via mitophagy is required for ferroptotic cell death. (A) chemical disruption of mitophagy through mitochondrial division inhibitor-1 (Mdivi-1) suppresses conidial cell death in a dose-dependent manner. Effectiveness of Mdivi-1 (60 μM) on mitochondrial fission was verified via MTS-GFP (projection) at 7 hpi. Conidial cell viability (green) or death (red) was quantified at 24 hpi and displayed as mean ± SD (3 technical repeats, n = 100 each per dose). DMSO (0.1%): solvent control. (B) mitochondria fuse instead of undergoing degradation when mitophagy is defective. Mitochondria (MTS-GFP shown as 3D images) in the wild type (WT) or mitophagy mutant atg24Δ were observed at 4 or 7 hpi. V for vacuole. (C) conidial cell death is iron dependent and fails to occur in atg24Δ conidia. Conidial cell death/viability of wild type (WT) or atg24Δ in the presence or absence of ferric ion (FeCl₃) or the iron chelator CPX was quantified at 24 hpi and is depicted as mean ± SD, which is derived from 3 technical replicates, each containing 100 conidia per treatment per strain. ** means p < 0.01 and n.S. refers to no significant difference as compared with indicated controls. (D-F) mitophagy defective atg24Δ conidia fail to accumulate lipid peroxides (oxidized lipids) to a level that can cause ferroptosis and are defective in ferroptosis-dependent blast disease infection in rice. (D) oxidized (green) and non-oxidized (red) variants of lipids in wild-type (WT) or atg24Δ conidial cells were observed at 7 to 8 hpi via C11-BODIPY 581/591 staining and ratiometric epifluorescence confocal microscopy and shown as single plane images. (E) the ability to form infection hyphae and spread within rice cells is greatly compromised for atg24Δ appressoria as compared with the wild-type counterpart. Mean ± SD (3 technical replicates, 100 appressoria each per strain per time point) is shown for the pathogenicity quantification measured at 28 or 48 hpi, with representative images on the right. A marks appressorium, and WT stands for wild type. (F) rice blast lesions photographed at 7 days post inoculation. Data shown from (A) to (F) are representative of 3 independent replicates of the experiment. Scale bar: 5 μm for (A), (B), and (D).

Mitophagy maintains a pool of metabolically active mitochondria for ferroptosis

To address the exact function of mitophagy during such developmental cell death, we compared mitochondrial metabolism in wild type and atg24Δ using the fluorescent dye TMRE whose mitochondrial localization depends on membrane potential. Interestingly, although a robust filamentous mitochondrial network persisted in the atg24Δ conidia, such mitochondria showed a massive reduction in the membrane potential as compared to those in the wild-type M. oryzae (Figure 3A; Figure S2). Such dramatic MMP reduction indicated that mitophagy is required for maintaining the overall metabolic activity. Indeed, time lapse imaging showed that all filamentous and punctate mitochondria in the mitophagy-competent wild-type strain, are metabolically active and stain positive with TMRE (Figure S1B). Consistent with this observation, localization of the mitochondrial matrix protein Atp1 (the α subunit of F₁ fraction of ATP synthase) further confirmed that mitophagy helps keep and maintain a pool of metabolically active mitochondria prior to the initiation of ferroptotic cell death (Figure S1C). Together, these data verified the role of mitophagy in maintaining active mitochondria and pointed out the necessity to understand the relationship between mitochondrial metabolism and ferroptosis.

Figure 3.

Mitophagy maintains a pool of metabolically active mitochondria that are important for ferroptosis. (A) mitochondrial membrane potential as detected by TMRE reduces dramatically in the atg24Δ conidia. MTS-GFP outlines mitochondria at 6 hpi. Cyan rectangle marks the conidium for which the TMRE staining is presented as an enlarged image. Conidia of wild type (WT) or atg24Δ were examined in three independent experiments with at least 30 conidia in total for each strain. One representative set of images and corresponding percentages of conidia showing typical TMRE patterns was presented. V, marks the vacuole. Images shown are all projections. (B) schematic model summarizing the mode of action for MitoQ, FCCP, and AcoA. CoQ, ubiquinone/coenzyme Q. ROS, reactive oxygen species. The graphic was created using BioRender. (C and D) positive correlation between mitochondrial metabolism and ferroptosis. (C) mitochondria in wild type or atg24Δ (ACOA-GFP atg24Δ #36) are displayed by AcoA-GFP, and TMRE reflects the mitochondrial membrane potential. Images for both panels are projections, whereas the inserted DIC images are single plane. Images and percentage of conidia showing a TMRE pattern as presented are representatives of 2 independent replications showing consistent results. Total around 30 conidia were examined per treatment per strain. (D) conidial cell viability or death was quantified at 24 hpi and displayed as mean ± SD from 3 technical replicates with n = 100 conidia for each treatment per strain per replicate. ** and !! means p < 0.01 as compared with WT and atg24Δ, respectively. Single plane DIC images showing the morphologies of wild type or atg24Δ with or without MitoQ or FCCP at 24 hpi. Wild type appressorium as well as one typical abnormal appressorium are enlarged and shown as insets. Data in (D) are representative of 2 independent replicates showing consistent results. Total 600 conidia were quantified per treatment per strain. Scale bar: 5 μm for (A), (C) and (D).

Therefore, the mitochondrial electron transport chain was disrupted by replacing ubiquinone/CoQ with its analog mitoquinone; or the membrane potential directly reduced through the protonophore FCCP (Figure 3B). A substantial decrease in ferroptosis along with a dramatic decline in mitochondrial membrane potential was evident upon such disruptive treatments (Figure 3C,D; Figure S3). Such outcomes were reminiscent of and akin to the phenotypic defects associated with the loss of mitophagy in M. oryzae (Figure 3C,D; Figure S3). We hypothesized that the sudden loss of mitochondrial membrane potential in the mitophagy mutant atg24Δ was likely due to oxidative stress or damage therein. Such mitochondrial impairment or oxidative damage, if remains salvageable, can be alleviated via the CoQ/MitoQ function as a potent antioxidant. Indeed, treating atg24Δ with MitoQ significantly increased the membrane potential and alleviated the ferroptosis defect as judged by the increase in the percentage of atg24Δ conidia that showed wild-type-like features i.e. conidial cells dead but appressoria viable and mature (Figure 3C,D; Figure S3). As a negative control, FCCP had no obvious effect on mitochondrial membrane potential in the atg24Δ strain and consequently failed to suppress the ferroptosis defects therein (Figure 3C,D; Figure S3). Instead, it triggered an abnormal and unprecedented spread of cell death from conidia to the appressoria (Figure 3D; Figure S3). The relationship between mitochondrial metabolism and ferroptosis was further verified through analysis of the aconitase A (hereafter AcoA), which catalyzes the conversion of citrate to isocitrate via cis-aconitate in the TCA cycle (Figure 3B). A native in-locus GFP-tagged strain of the mitochondrial AcoA was generated (Figure S4A-C) and confirmed to be wild-type like in terms of its ability to infect rice plants (Figure S4D). Loss of ATG24, significantly affected the mitochondrial localization of AcoA-GFP (Figure S4E-G), suggestive of overall defects in MMP-associated import of freshly synthesized AcoA into the mitochondria upon loss of mitophagy. In line with this finding, AcoA-GFP marked mitochondria became fragmented upon treatment with exogenous MitoQ or FCCP (Figure 3C), further supporting that disruption of mitochondrial oxidative metabolism most likely, albeit indirectly, also affects the TCA cycle, which is consistent with the decrease in membrane potential observed with such drug treatments (Figure 3C). Taken together, we conclude that mitochondrial membrane potential-dependent metabolism is required for proper and timely induction of ferroptosis in rice blast.

Cellular CoQ, ATP and iron link mitochondrial metabolism to ferroptosis

Next, we asked whether mitochondrial metabolism and ferroptosis are functionally linked too. Since CoQ is synthesized in mitochondria and exported to the plasma membrane (Figure 4A), where it acts as an antioxidant to inhibit ferroptosis [7,8,10], the previously identified opposing role of mitochondrial CoQ as an electron carrier that enables ferroptosis, was further validated using the analog idebenone (Figure 4A). As a positive control, the ETC complex III inhibitor antimycin A (Figure 4A) was included for comparison. As expected, idebenone and antimycin A suppressed conidial cell death to a similar level (Figure 4B), thus confirming that the mitochondrial pool of CoQ facilitates ferroptosis. In contrast, imposing CoQ deficiency with low doses of 4-CBA (Figure 4A), likely via reducing the pool of CoQ near the plasma membrane, promoted conidial cell death, and such death promotion could be reversed by the ferroptosis inhibitor liproxstatin-1 (Figure 4C,D). Notably, 4-CBA treatment did not affect mitochondrial fission or turnover via mitophagy (Figure S5A). To further confirm the role of CoQ in suppressing ferroptosis without affecting mitophagy, gene expression of COQ2, the target of 4-CBA, was knocked down using a tetracycline/doxycycline-repressible system (Figure S5B), and the resultant Tet-Off COQ2 strain showed an increase in conidial cell death or ferroptosis in a doxycycline dose-dependent manner (Figure S5C) during pathogenic development. The mycelial growth of the Tet-Off COQ2 strain was unaffected by doxycycline treatment (Figure S5D). Taken together, these findings helped demonstrate and infer a complex link between ferroptosis and mitochondrial oxidative metabolism anchored on CoQ synthesis, with CoQ exerting opposing effects on ferroptosis depending on its intracellular levels and/or distribution.

Figure 4.

Coenzyme Q and ATP modulate ferroptosis. (A) Model summarizing the relationship between mitochondrial metabolism and fungal ferroptosis. (B) Disruption of electron transport chain leads to ferroptosis suppression. Conidial cell viability/death was quantified at 24 hpi and is displayed as mean ± SD from 3 technical replicates with n = 100 conidia for each treatment per replicate. (C) Inhibition of coenzyme Q biosynthesis via exogenous 4-CBA promotes ferroptosis at lower doses. Conidial cell viability (green) or cell death (red) was quantified at 14.5 hpi and is displayed as mean ± SD derived from 3 technical replicates (n = 100 conidia each) of the experiment for each dose. (D) Ferroptosis induction caused by 4-CBA (40 μM) is reversed by liproxstation-1 (lip-1). Conidial cell viability/death presented as mean ± SD (3 technical replicates, n = 100 conidia for each treatment per replicate) was quantified at 14.5 hpi. (E) Effect of glucose on conidial viability in the wild-type (WT) or atg24Δ strain of M. oryzae. death (red) or viability (blue) of the conidium was quantified at 24 hpi and presented as mean ± SD derived from 3 technical replicates, each containing 100 conidia per treatment per strain. (F) Promotion of conidial cell death in atg24Δ conidia caused by glucose is abolished by lip-1. Conidial cell death (red) or viability (green) in atg24Δ was quantified at 24 hpi and presented as mean ± SD (3 technical replicates, n = 100 conidia for each treatment per replicate). For quantification in (B) to (F), ** (p < 0.01) indicates significant differences detected as compared to the solvent (DMSO or water) or the indicated control, and data presented have been confirmed through 2 or 3 biological repeats of the experiments.

Next, mitochondrial bioenergetics or ATP synthesis was tested for such a functional link by promoting glycolysis via exogenous glucose in the low MMP atg24Δ or by directly providing exogenous ATP in trans. Interestingly, such exogenous ATP slightly but significantly increased the conidial cell death in the atg24Δ regardless of the dosage of glucose or ATP (Figure 4E,F; Figure S5E), which further underscored the importance of ATP synthesis and mitochondrial metabolism in regulating ferroptosis.

We further investigated whether bioavailability of iron serves as a functional connection between mitochondria and ferroptosis, because the atg24Δ conidia consistently exhibited signs of iron deficiency based on staining with calcein-AM (Figure 5A), a fluorescent probe or sensor whose signal is quenched upon binding to iron. Such iron deficiency was also evident through AcoA, which requires iron as a cofactor and is sensitive to lack thereof. Iron chelation through CPX caused mild mitochondrial fusion in AcoA-GFP – labeled mitochondria in the wild type but caused strong mitochondrial fragmentation in the atg24Δ mutant (Figure 5B), likely due to the significantly decreased levels of iron sensed by AcoA therein compared to the wild type. Interestingly, short term iron chelation with CPX caused an increase in mitochondrial membrane potential in wild-type M. oryzae (Figure 5B; Figure S6), thus suggesting a role for active mitochondria in mediating iron homeostasis. Consistent with these results, wild-type conidia showed a higher overall iron release capacity (Figure 5A) and a greater abundance of functional mitochondria during early developmental stages (4–7 h; (Figure 1B; Figure S1C)) compared to the later 7–10 h window. In contrast, a similar increase membrane potential in response to iron chelation was also observed in atg24Δ, albeit to a much lesser extent (Figure 5B; Figure S6), indicating that the low MMP mitochondria therein are less capable of sensing and/or regulating iron homeostasis/release, which may account for the iron deficiency observed in the atg24Δ (Figure 5) and the consequent defect in iron-dependent ferroptosis (Figure 2C). Intriguingly, the iron deficiency in the atg24Δ, could not be alleviated by direct provision of FeCl₃ since the atg24Δ strain showed a similar level of ferroptosis defects with or without such iron supplementation (Figure 2C), thus reflecting a unique iron starvation phenotype associated with such low MMP mitochondria in atg24Δ. We infer a mitochondrial membrane potential-dependent regulation of cellular iron homeostasis and/or bioavailability, which together with CoQ and ATP biosynthesis, functionally links mitochondrial metabolism to ferroptosis in the rice blast fungus.

Figure 5.

A role of active mitochondria in mediating cellular iron homeostasis. (A) iron deficiency in the atg24Δ as shown by the FRET sensor calcein-AM signal which reduces upon binding to iron. calcein-AM signal within the conidium marked by the rectangle is presented. Pink arrows mark the dying cell at the indicated time points. Intensity of calcein-AM fluorescence was quantified along the white line starting from cell number 1 to 3 as shown in the insets. X axis of the intensity chart measures the length of the white line. Both calcein-AM and bright field (BF) images are projections. WT: wild type. (B) change of mitochondrial membrane potential and dynamics of AcoA in response to iron starvation in the presence and absence of mitophagy. Mitochondrial membrane potential in wild-type conidia reflected by TMRE increases when iron is chelated by CPX, and such change becomes very small when mitophagy is lost. CPX treatment was carried out for 4–6 h post inoculation of the wild-type conidia, and for 6–8 hpi of the atg24Δ conidia. ACOA-GFP #3 and ACOA-GFP atg24Δ #36 are the two strains used to show normal or disrupted mitophagy, respectively. Time scale (hpi) indicated on the left refers to the specific stages of pathogenic development post conidial inoculation. Calcein-AM staining (A) was repeated six times, whereas mitochondrial response to iron chelation (B) was repeated twice. Overall percentage of conidia showing the depicted phenotype/effect per strain per time point or treatment is provided (A) and (B). Scale bar: 5 μm for all the images.

Mitochondrial β-oxidation is not essential for ferroptosis

Catabolism of polyunsaturated fatty acid (PUFA) in the mitochondria was also examined in M. oryzae using the mitochondrial β-oxidation mutant ech1Δ [22], which shows defects in fatty acid oxidative catabolism [22]. Surprisingly, such mutant was capable of undergoing ferroptosis just like the wild-type M. oryzae (Figure 6A). Furthermore, no change in mitochondrial membrane potential or cellular iron availability was observed in the ech1Δ mutant as compared to the wild type (Figure 6B,C), which in turn supports the membrane potential-dependent roles of mitochondria in cellular iron control and ferroptosis. In conclusion, results here demonstrate that mitophagy helps maintain a pool of metabolically active mitochondria that are important for ferroptosis and unveil an important mitochondrial bioenergetics and redox regulatory network for precise ferroptotic cell death in the rice blast pathosystem.

Figure 6.

The mitochondrial β-oxidation mutant ech1Δ is not defective in ferroptosis. (A) conidial cell viability or death in wild type (WT) or ech1Δ was quantified at the indicated time points and presented as mean ± SD (3 technical replicates, n = 100 conidia for each time point per strain per replicate). n.S. means no significant difference was detected as compared to the WT at the corresponding time points. Experiment has been repeated twice. (B) mitochondrial metabolism reflected by mitochondrial membrane potential/TMRE (projection, only 3 conidial cells are shown) is dramatically reduced in atg24Δ but remains normal in ech1Δ, as compared to the wild-type M. oryzae (WT). Insets are full view projections merging bright field and TMRE images. (C) cellular iron availability shown by calcein-AM at 7 hpi indicates that ech1Δ, does not suffer from iron deficiency as compared to the wild-type (WT) Magnaporthe. Cyan rectangle marks the conidium for which the calcein-AM therein are shown as enlarged images. Both the bright field (BF) and calcein-AM images are projections. Representative percentages of conidia showing typical calcein-AM or TMRE patterns are provided for (B) and (C). Scale bars: 5 μm.

Discussion

The process of mitophagy likely distinguishes between functional and dysfunctional mitochondria based on low levels of mitochondrial membrane potential and clears such damaged organelles via vacuolar degradation in a precise spatio-temporal manner in M. oryzae. This is important for ensuring and enabling a pool of metabolically active mitochondria in the viable conidial cells during the early stages of infection-related development. Such active mitochondria are, in turn, necessary for the programmed ferroptosis during pathogenesis, with the membrane potential-dependent modulation of cellular iron homeostasis and/or bioavailability and/or synthesis of CoQ and ATP synthesis as promising links connecting these two important processes.

Increase in mitochondrial membrane potential has also been observed in mouse embryonic fibroblasts undergoing ferroptosis and is attributed to cysteine deprivation therein [13]. It is worth noting that cysteine is also required for the synthesis of iron-sulfur clusters in mitochondria in a membrane potential dependent manner [11], thus linking the mitochondrial responses to iron shortage (via iron chelation) observed in Magnaporthe (Figure 5B). Proper function of the TCA cycle and the electron transport chain, likewise, depends on iron as a cofactor, which intricately links such iron starvation responses and explains the dependence of ferroptosis on iron from a new perspective given the requirement of mitochondrial metabolism and membrane potential in mounting such a specific regulated cell death modality. The role of metabolically active mitochondria in modulating cellular iron homeostasis also explains why both the mitophagy mutant atg24Δ (Figure 5) and the nonselective autophagy mutant atg8Δ [17] exhibit iron deficiency since Atg24 recruits the core autophagy machinery including Atg8 for clearing dysfunctional or damaged mitochondria. Interestingly, a recent study reported that both autophagy flux and ferroptotic cell death in the rice blast fungus were compromised when synthesis of lipids, phosphatidylethanolamine in particular, was disrupted [23], thus revealing the autophagy-ferroptosis connection from a fresh perspective.

Unlike mitochondria in mitoQ- or FCCP-treated wild type, extensively fused mitochondria were also observed in the atg24Δ mutant, thus suggesting that mitochondrial fusion is likely a salvage option to titrate and reduce the negative effects exerted by damaged or dysfunctional mitochondria with reduced membrane potential that are likely substrates for mitophagy. Supporting such a hypothesis, mitochondrial fusion was recently shown to restore the mitochondrial defects caused by mitochondrial DNA mutations and oxidative stress [24]. Interestingly, ferroptosis was also found to be suppressed by such drug-induced mitochondrial fusion [24], most likely by disrupting the timely clearance of dysfunctional mitochondria through mitophagy.

The finding that mitoQ can alleviate atg24Δ defects is interesting and suggests that atg24Δ suffers similarly from oxidative stresses. Thus, the inability of atg24Δ to execute ferroptosis properly implies that mitochondrial ROS is unlikely to be a pro-ferroptosis factor in M. oryzae. In addition to the defects in ferroptosis, atg24Δ also showed a dramatic reduction in conidiation [21], and some of its conidia contained only two instead of three cells (Figure S3). It remains to be addressed whether a disruption of cellular ROS homeostasis accounts for these additional phenotypes or defects upon loss of mitophagy in rice blast.

Overall, our data highlight the mitochondrial membrane potential-enabled metabolism as an important regulator of developmental ferroptotic cell death in the rice blast fungus and provide a potential strategy for intervention of the devastating blast disease in cereal crops.

Materials and methods

Fungal strains and growth conditions

Magnaporthe oryzae strains were cultivated on prune agar/PA (per liter: 40 mL prune juice [Del Monte], 5 g D-lactose monohydrate [Sigma-Aldrich, L254], 1 g yeast extract [Oxoid Ltd, LP0021B], 10 g sucrose [Sigma-Aldrich, S9378] and 20 g agar [Becton, Dickinson and Company 214,010], pH 6.0) medium for vegetative growth and conidiation as described [25]. Blast isolate B157 obtained from the Directorate of Rice Research (Hyderabad, India) was used as the wild-type strain of choice. Epifluorescence-tagged strains including histone H1 (hH1)-GFP [17], MTS-GFP [22], Atp1-GFP [22], as well as deletion mutants atg24Δ with or without MTS-GFP [21], and ech1Δ [22] have been described in our previous publications.

Generation of ACOA-GFP, ACOA-GFP atg24Δ, and Tet-Off COQ2 strains

The plasmid construct for generating ACOA-GFP in the locus tagged strain (Figure S4A) was made using the Clon Express MultiS One Step Cloning Kit (Vazyme, C113). Briefly, 4 overlapping fragments covering a part of ACOA (MGG_03521) exon3 (1132 bp just before the stop codon), a short sequence encoding a 4 amino acid linker, eGFP encoding sequence without start codon, basta resistance cassette in the reverse direction, and a part of ACOA 3’UTR (1258 bp including the ACOA stop codon) were PCR amplified using primers listed in Table S1 (primer 1 to 8) and purified. Meanwhile, the empty vector pFGL815 (Addgene 52322; Naweed Naqvi’s lab) was linearized using XmaI and XbaI, and subsequently used to assemble the 4 PCR fragments using the MultiS One Step Cloning Kit (Vazyme, C113). Same technology, empty vector, and kit were used for generating the ATG24 deletion construct as indicated in Figure S4E or Tet-Off COQ2 construct as demonstrated in Figure S5B using primers listed in Table S1. The plasmid template for the Tet-Off system was pFGL1252 (Addgene 118993; Naweed Naqvi’s lab). The 3 resultant plasmids were sequence verified and then used to transform Agrobacterium tumefaciens AGL1 strain (Sundaresan Lab, Institute of Molecular Agrobiology, Singapore) individually through electroporation. Wild-type B157 was the parent strain for generating the ACOA-GFP or Tet-Off COQ2 strains, while ACOA-GFP #3 was selected for subsequent ATG24 deletion. Agrobacterium T-DNA mediated transformants of the rice-blast fungus M. oryzae were selected on basal medium/BM (1% glucose [Sigma-Aldrich, G7021], 0.67% yeast nitrogen base without amino acids [Sigma-Aldrich, Y1251], 0.1% yeast extract, NH₄NO₃ [Sigma-Aldrich, A3795] pH 6.0 with Na₂HPO₄ [Sigma-Aldrich 567547]) with basta/glufosinate-ammonium (Sigma-Aldrich 45520) or sulfonylurea/chlorimuron ethyl (Sigma-Aldrich 32874) or complete medium/CM (1% sucrose, 0.6% yeast extract, 0.6% casein hydrolyzate [Sigma-Aldrich, A2427]) with hygromycin B (A.G. Scientific, H-1012-PBS) [25]. Positive transformants were further verified through epifluorescence imaging and/or PCR using primers indicated in Figure S4 or Figure S5B, which are also listed in Table S1.

Staining protocols and confocal microscopy

Sterile water droplets (20 μl) containing freshly harvested conidia at a concentration of 1.5 × 10⁵ conidia/ml were inoculated on hydrophobic cover glass (Matsunami, C218181) for normal imaging, or a 30 μl droplet at the concentration of 1 × 10⁵ conidia/ml were used for inoculating in glass-bottom culture dishes (MatTek Corporation, P35G–0–14-C) for time lapse imaging. Roughly 20–30 min before imaging, conidia were stained with 10 µM C11-BODIPY^581/591 (Thermo Fisher, D3861) to detect lipid peroxidation, or with 250 nM TMRE (Sigma-Aldrich 87,917) to assess mitochondrial membrane potential, or 1 µM calcein-AM (Invitrogen, C3099) to test cellular iron availability.

Laser scanning confocal microscopy was performed using the Leica TCS SP8 X inverted microscope system (Leica Microsystems) under the control of Leica Application Suite X software package (release version 3.5.7.23225). Experiments with C11-BODIPY 581/591 were done using Matsunami micro slide glass (Matsunami, S7213) and an HCX Plan Apochromat lambda blue 63×/1.20 water immersion objective. An argon laser (excitation, 488 nm; emission, 500–535 nm) was used for the oxidized form whereas the white light laser (excitation, 561 nm; emission, 573–613 nm) was used for non-oxidized variant. HC Plan Apochromat CS2 100× or 63×/1.4 oil immersion objectives and white-light laser were used for GFP (excitation, 488 nm; emission, 500–550 nm), calcein-AM (excitation, 494 nm; emission, 510–550 nm), and TMRE (excitation, 540 nm; emission, 580-610 nm). All the lasers associated with Leica TCS SP8 were controlled by the AOTF (Acousto-Optical-Tunable-filter), and fluorescence images were captured using the Leica Hybrid Detector as Z stacks of 10 to 25 sections (0.5 µm-spaced). Intensity of fluorescent signal was directly quantified through SP8 whereas time lapse images were further processed using the IMARIS software (Bitplane AG, v.9.6.0).

Pharmacological treatment and cell viability or cell death measurements

Conidia were inoculated on cover glass as above described. 4-CBA (Sigma-Aldrich 135585) or glucose (Duchefa Biochemie, G0802) were added at 0 h post inoculation (hpi), ATP (Sigma-Aldrich, A6419) was added at 2.5 hpi, while Mdivi-1 (Sigma-Aldrich, M0199), 4 µM MitoQ (Cayman chemical 29317), 2 µM FCCP (Sigma-Aldrich, C2920), 32 µM idebenone (Cayman chemical 15475), 9.4 µM antimycin A (Sigma-Aldrich, A8674), 54 µM liproxstatin-1(Sigma-Aldrich, SML1414), 5 µM CPX (Sigma-Aldrich, C0415), or 5 µM FeCl₃ (Sigma-Aldrich, F2877) was added at 4 hpi to maximally restrict the chemical effect on ferroptosis or conidial cell death and minimize the off-target effects on other processes. For the same reasons, MitoQ or FCCP was added to atg24Δ conidia at 7 hpi. Conidial cell viability or cell death was quantified using 1% trypan blue (Sigma-Aldrich, T6146) at the indicated time points. Conidia capable of appressorium formation and possessing 1 to 3 viable conidial cells were considered “viable,” while those having a viable appressorium but all the 3 conidial cells inviable were regarded as “dead.” This “dead” category does not include the rare or abnormal case wherein both the conidium and appressorium were inviable.

Trypan blue stained samples were imaged using the Zeiss Colour Widefield Upright microscope equipped with Plan Apochromat 100×/1.4 oil immersion objectives and the OCULAR software (Version 2.0).

Statistical analysis

Statistical analysis was achieved via Student’s t-test.

Rice blast infection assays

The youngest leaf or leaf sheath of susceptible CO39 rice seedlings at 4 to 5 leaf stage was used for testing rice infection by wild type, atg24Δ or ACOA-GFP #3. Rice infection by wild type or atg24Δ conidia was imaged at the indicated time points using the Zeiss Colour Widefield Upright microscope equipped with the OCULAR software. Blast disease lesions were then examined at 7 days after inoculation as previously described [17].

Funding Statement

This work was supported by intramural funding from the Temasek Life Sciences Laboratory, Singapore.

Disclosure statement

No potential conflict of interest was reported by the author(s).

Supplementary Information

Supplemental data for this article can be accessed online at https://doi.org/10.1080/15548627.2025.2546944