Heat stress induces ferroptosis-like cell death in plants

In eukaryotes, regulated cell death (RCD) is an active mechanism essential for development, immune responses, and many pathophysiological processes (Hakem et al., 1998; Yoshida et al., 1998; Lindsten et al., 2000; Van Hautegem et al., 2015). In animals, cell death was traditionally classified into three categories, based on the morphological features exhibited by cells during the death process, as apoptotic, autophagic, or necrotic (Kroemer et al., 2009). In recent years, this tripartite classification scheme has been challenged by the discovery of new forms of cell death that differ from these canonical categories, such as pyroptosis and necroptosis (Bergsbaken et al., 2009; Christofferson and Yuan, 2010). Most recently, a new form of cell death that was shown to be morphologically, biochemically, and genetically distinct from apoptosis, necrosis, and autophagy was described in cancer cells, in kidney tissue, and in the nervous system. This type of cell death was named ferroptosis, as it is dependent on intracellular iron, which is required for the accumulation of toxic lipid reactive oxygen species (ROS; Yang and Stockwell, 2008; Dixon et al., 2012; Friedmann Angeli et al., 2014; Linkermann et al., 2014; Skouta et al., 2014; Yang et al., 2014). A defining feature of ferroptosis is that it can be prevented by treatment of cells with iron-chelating agents, such as deferoxamine or ciclopirox olamine (CPX), or with lipophilic antioxidants, such as ferrostatin 1 (Fer-1) or vitamin E (Yagoda et al., 2007; Yang and Stockwell, 2008; Dixon et al., 2012; Skouta et al., 2014; Yang et al., 2014). Although the specific role of iron in ferroptosis remains unclear, it is unlikely to be explained by a simple increase in iron-catalyzed ROS production (i.e., Fenton chemistry), as cell death induced by peroxide is in many aspects different from ferroptosis (Dixon et al., 2012).

In plants, RCD mechanisms play critical roles during several developmental processes, such as megasporogenesis, fertilization, embryogenesis, development of tracheary elements, and regulation of leaf shape (Gunawardena et al., 2004; Drews and Koltunow, 2011; Bollhöner et al., 2012; Choi, 2013). Additionally, abiotic stresses, such as salt stress, drought, and nutrient starvation, are able to induce plant cell death (Liu et al., 2009). Plant cells do not undergo apoptosis as defined by classic morphological features observed in animal cells. The rigid plant cell wall prevents the cells from breaking into apoptotic bodies, and even though protoplasts are known to shrink in response to diverse abiotic stresses, they do not further fragment into discrete bodies. Thus, a recent classification of plant programmed cell death (PCD) stated that the use of the term “apoptosis-like” is not justified in plants, although it remains a subject of debate (van Doorn et al., 2011).

Vacuoles may be involved in a specific form of plant cell death. A large vacuolar system occupies most of the plant cell volume (Marty, 1999). These vacuoles have important functions during a plant-specific type of cell death, termed vacuolar cell death, that involves a gradual decrease in the cytoplasm volume and formation of small lytic vacuoles, resembling autophagy (Jones, 2001). Several A. thaliana autophagy genes (ATG) have been implicated in this type of cell death and were first identified based on sequence similarity to yeast autophagy genes (Xie and Klionsky, 2007). Remarkably, mitochondria and other organelles remain intact until the final stages of vacuolar cell death, which involves tonoplast rupture, disassembly of the nuclear envelope, and nuclear segmentation (van Doorn et al., 2011). Vacuolar cell death is associated with several developmental pathways, including aerenchyma formation, leaf remodeling, and xylem differentiation (Drew et al., 2000; Gunawardena, 2008; Courtois-Moreau et al., 2009; Kwon et al., 2010), and has also been linked to plant responses to environmental stress conditions and to leaf senescence (Liu et al., 2009; Kariya et al., 2013).

In this work, we examined whether iron-dependent, oxidative processes similar to those recently described to occur during ferroptosis in animal cells could be relevant to plant cell death during development or after abiotic stress. Remarkably, ferroptosis was implicated in the RCD that follows heat shock (HS) stress, but not in several other cell death events. These results show that a group of morphological and biochemical features that are specific for ferroptosis are present in Arabidopsis roots in response to HS and suggest an underlying similarity between ferroptosis-like plant cell death and animal cell death.

Diverse environmental stresses, such as salt stress, high temperatures, drought, and nutrient starvation, are able to induce cell death in plants (Liu et al., 2009). Stress-induced cell death can be studied by following the response to HS, hydrogen peroxide (H₂O₂), and salt (NaCl) stress in cell suspensions and root hairs (Reape and McCabe, 2008; Blanvillain et al., 2011; Hogg et al., 2011). A 10-min heat treatment at 55°C (55°C HS) triggers RCD in A. thaliana, whereas exposure to higher temperatures promotes necrosis (Reape and McCabe, 2008; Blanvillain et al., 2011). The mechanism involved in the cell death triggered by moderate heat (i.e., 55°C HS) is not clear, but is known to involve the accumulation of ROS (Hogg et al., 2011). We hypothesized that processes related to the iron-dependent, oxidative ferroptotic pathway described in animal cells could be relevant to stress-induced cell death in A. thaliana. To test this hypothesis, diverse lethal treatments were performed in the presence of two small-molecule ferroptosis inhibitors discovered and characterized in animal cells: the lipophilic antioxidant Fer-1 and the membrane-permeable iron chelator CPX (Dixon et al., 2012). CPX has a very high affinity for iron, comparable to that of deferoxamine (Linden et al., 2003). Its higher lipophilicity, specificity, and availability make this iron chelator a very useful tool in cell biology studies (Kuriki et al., 1975).

When 6-d-old seedlings were preincubated for 16 h before HS with 1 µM Fer-1 or 10 µM CPX, the death of root hairs triggered by 55°C HS, as assayed by Sytox green nucleic acid stain, was significantly prevented (Fig. 1 a). In contrast, neither Fer-1 nor CPX prevented cell death triggered by 77ºC H₂O₂ or NaCl treatments (Fig. 1 a), suggesting that ferroptosis inhibitors specifically block cell death triggered by 55°C HS. Necrostatin 1 (Nec-1), a potent inhibitor of a different nonapoptotic cell death pathway in animal cells, RIPK1-mediated, did not prevent cell death triggered by HS at either 55°C or 77°C (Fig. S1 a). This suggests that necroptosis is not involved in HS-induced death in plant cells.

In plants, RCD is calcium dependent in numerous systems and tissues (Ma and Berkowitz, 2007); thus, the effect of calcium chelators was studied in 55°C HS–triggered cell death. Whereas at 6 h after treatment, ∼70% of the root hairs in A. thaliana roots were dead, only ∼10% died when cotreated with the calcium chelator EGTA, a value comparable to that seen in untreated roots, suggesting that influx of calcium from the extracellular space is required for HS-induced, iron-dependent cell death in plants (Fig. 1 b). In addition, dose–response curves were constructed in which we measured the ability of Fer-1 and two structural analogues to prevent HS-induced cell death in roots. Overall, plant cells were more sensitive to Fer-1 and structurally related compounds than human cells: whereas Fer-1 had an EC₅₀ of 60 nM in cancer cells (Dixon et al., 2012), the EC₅₀ in plant cells was 2.5 pM (Fig. 1 c). Similarly, the two Fer-1 analogues tested (SRS9-01 and SRS8-24) showed lower EC₅₀ values, and the tendency observed was equivalent to the one described for these analogues in tumor cells (Dixon et al., 2012). Substitution of the primary aromatic amine, which is essential for antioxidant function (SRS9-01 and SRS8-24) resulted in less protection against cell death, as reported for animal cells (Dixon et al., 2012; Skouta et al., 2014). Incubation of plants with an exogenous source of iron (200 µM FeNa-EDTA) did not potentiate the 55°C HS–induced death in root cells. Likewise, the application of other divalent metal ions (e.g., Cu²⁺ and Mn²⁺) did not have any effect on 55°C HS–induced cell death (Fig. S1 b). One possibility is that the requirement of iron was already fulfilled in planta, and thus cell death was occurring at its maximum rate already. Alternatively, exogenous iron may not reach the intracellular sites or enzymes that promote cell death under these conditions.

Analyses of root hair morphology using differential interference contrast (DIC) microscopy showed that 55°C HS produced cytoplasmic retraction and strong nuclear Sytox green staining (Fig. 2 a). No cytoplasmic retraction was observed in root hairs treated at 77°C, and neither Fer-1 nor CPX was protective at this temperature (Fig. 2 a). Root hairs treated with H₂O₂ or NaCl showed partial cytoplasmic retraction and cell death that was not prevented by Fer-1 or CPX (Fig. 2 a). Using transmission EM (TEM) on root cortical cells, we confirmed that after 6 h, 55°C HS resulted in cytoplasmic retraction, and we observed the presence of numerous lytic vacuoles and shrunken and less abundant mitochondria showing a dark (condensed) matrix. In tumor cells, the only distinctive morphological characteristic of ferroptotic cells was the presence of smaller mitochondria that showed increased membrane density compared with normal cells (Yagoda et al., 2007; Dixon et al., 2012). Plant cells from roots treated at 77°C showed no cytoplasmic retraction, and the presence of mitochondria was rare, as was the presence of lytic vacuoles (Fig. 2 b). To know how mitochondrial morphology was affected after HS, we measured mitochondria at different times after HS using MitoTracker green and confocal microscopy. As shown in Fig. 2 c, mitochondrial planar area increased during the first hours after 55°C HS, consistent with mitochondrial swelling in cells undergoing cell death. After 6 h of 55°C HS, mitochondria were slightly smaller than in control cells. Thus, 55°C HS resulted in mitochondrial morphological changes that mimic those observed in cancer cells undergoing ferroptosis, and in the accumulation of lytic vacuoles that are not observed in cancer cells.

Using Sytox green and confocal microscopy, we investigated how the different cell types that compose the roots were dying over time after 55°C HS treatment (Videos 1, 2, 3, and 4). We observed that root hairs and cells composing the main root, principally those related to the central cylinder, were dying during the first hours after exposure to 55°C. Interestingly, the vast majority of iron in the root was reported to locate within the central cylinder, whereas only very small amounts of iron can be detected in the epidermis and cortex (Kuriki et al., 1975).

In tumor cells, ferroptosis is characterized by the iron-dependent accumulation of lethal cytosolic and lipid ROS. In these cells, the accumulation of cytosolic and lipid ROS is prevented by iron chelation (Dixon et al., 2012). To investigate whether a similar process takes place in Arabidopsis, we measured changes in cytosolic and lipid ROS using the fluorescent probes H₂DCFDA and C11-BODIPY^581/591, respectively, after inducing cell death with a 55°C HS in A. thaliana roots and in cell suspensions. No change in lipid ROS was detected in root cells under these conditions, perhaps because of the limited sensitivity of the C11-BODIPY dye. However, H₂DCFDA fluorescence increased as early as 30 min after treatment and preceded cell death (Fig. 3 a). ROS were detected at high levels until 6 h after HS, when the cells died (Fig. 3 b). ROS accumulation and cell death were both prevented by cotreatment with CPX or Fer-1. When cell death was induced in A. thaliana cell suspensions, both cytosolic and lipid ROS were increased (Fig. S2). These increases were suppressed by cotreatment with CPX or Fer-1 (Fig. S2). To test whether mitochondrial ROS were involved in this type of cell death, mitochondrial superoxide (mitoSOX)­­–sensitive ROS production was measured after HS. No increase in mitoSOX was detected either in Arabidopsis root hairs or in cell suspensions (Figs. S2 and S3).

The mechanism of lipid peroxidation during ferroptosis was recently reported to involve peroxidation of polyunsaturated fatty acids (PUFAs) at the bis-allylic position in human cells. Pretreatment of cancer cells with PUFAs containing the heavy hydrogen isotope deuterium at the site of peroxidation (D-PUFAs) prevented PUFA peroxidation, blocking ferroptosis (Yang et al., 2016a). Likewise, we observed that preincubation with 8 µM D4-linoleate (16 h) protected plants against 55°C HS but not against 77°C HS (Fig. 3 e). Together, these results suggested that plant cells exposed to HS undergo an oxidative, iron-dependent form of cell death similar to ferroptosis observed in animal cells.

To investigate another possible source of ROS that might be involved in 55°C HS–induced cell death, A. thaliana roots and cells were exposed to 55°C HS in the presence of diphenyleneiodonium (DPI). DPI is an inhibitor of NADPH oxidase (NOX) and other flavo-enzymes such as NO synthase and xanthine oxidase (Wind et al., 2010). DPI treatment prevented the increase in ROS after HS (Fig. S2). Thus, as in human tumor cells (Dixon et al., 2012), HS-induced cell death in A. thaliana does not involve mitoSOX production and may involve NOX enzymes and lipid peroxidation.

Reduced glutathione (GSH) and ascorbic acid (AA) are important redox buffers in eukaryotic cells (Pavet et al., 2005; Kranner et al., 2006; Franco and Cidlowski, 2012). AA depletion has been linked to ROS accumulation and cell death associated with plant defense responses during pathogen attack and senescence (Pavet et al., 2005). GSH depletion is considered a central event that activates cell death pathways in both animal and plant cells (Kranner et al., 2006; Franco and Cidlowski, 2012). In mammalian cells, class I ferroptosis-inducing agents are thought to trigger ferroptosis by depleting the cell of GSH (Skouta et al., 2014; Yang et al., 2014). In plants, it is likely that AA levels are affected as well. We therefore measured the levels of GSH and AA in root tissues after 55°C HS. Both total GSH and AA concentrations declined after HS (Fig. 4, a and b). An increase in the fraction of the oxidized AA was also detected after HS (Fig. 4 a). These effects were not reverted by cotreatment with Fer-1 or CPX, demonstrating that GSH and AA depletion is upstream of lipid ROS accumulation, as in animal cells (Dixon et al., 2012; Skouta et al., 2014; Yang et al., 2014). To determine whether GSH depletion was necessary for the ferroptotic-like death observed upon 55°C HS in plant roots, plants were cotreated with GSH, which can be taken up by plant cells (Schneider et al., 1992; Jamai et al., 1996; Zhang et al., 2004). Notably, when GSH was added to the culture medium, HS-induced cell death was prevented (Fig. 4 c). Additionally, we determined that GSH depletion was sufficient to trigger cell death, as Arabidopsis roots treated with l-buthionine-(S,R)-sulfoximine, a specific inhibitor of GSH biosynthesis, showed ∼50% of cell death after 4 d of treatment (Fig. 4 d). Thus, GSH depletion is critical for HS-induced cell death in plants, probably by compromising the ROS-scavenging capacity of the cell. Altogether, these results suggest that plant cells can also undergo an oxidative, iron-dependent type of cell death similar to ferroptosis in animal cells.

Despite the absence of plant orthologues for caspase genes, proteases with caspase-like enzymatic activities are required for plant cell death (Danon et al., 2004; Ge et al., 2016). To assess whether caspase-like activity is involved in cell death triggered by HS in Arabidopsis, we studied the effect of preincubation with the caspase-3 inhibitor Ac-DEVD-CHO (CHO). We observed that preincubation with CHO partially reduced the percentage of cell death triggered by 55°C HS but not 77°C HS (Fig. S4 a). This suggested that a caspase-like activity was involved in the pathway triggered after 55°C exposure. To study another aspect linked to apoptosis-like cell death, we assessed DNA fragmentation by using TUNEL assays (Fig. S4 b). No TUNEL-positive nuclei were detected in roots 6 h after the initiation of 55°C HS. Although we cannot exclude DNA fragmentation as a late outcome for this cell death pathway, these results suggest that it might not take place at least at 6 h after HS, when cell death is already detected in root tissues.

The potential involvement of ferroptosis-like cell death during reproductive development was investigated. A. thaliana inflorescences were treated with Fer-1 or CPX, and pistils composing the inflorescence, ranging in size between 1 and 8 mm, were examined for two key events involving cell death: (1) megaspore cell death that occurs as the final step of megasporogenesis and (2) synergid cell death, the death of the cells that receive the pollen tube during fertilization (Christensen et al., 2002; Sandaklie-Nikolova et al., 2007). After treatment, the cleared ovules were examined for defects in megasporogenesis or fertilization (see Materials and methods for details). However, no abnormalities were detected, suggesting that ferroptosis might not be involved in these developmental cell death processes (Fig. S5, a–c).

During vascular differentiation, tracheary elements undergo PCD, which results in a continuous transport system of empty cells (Bollhöner et al., 2012). However, no evident defects in vascular tissues were detected in A. thaliana plants growing in the presence of 1 µM Fer-1, indicating that vascular differentiation does not likely involve a ferroptotic type of cell death (Fig. S5 d).

As an initial approach to identify markers of ferroptotic plant cell death, we examined the expression pattern of 30 candidate genes in roots from plants exposed to 55°C HS ± Fer-1 (Table S1 and Table S2). As can be observed in Fig. 5 a, the death of root cells triggered by 55°C HS is prevented by Fer-1, as previously shown for root hairs. The genes tested were (a) asparagine synthetase genes ASN1, ASN2, and ASN3, as ASN has been identified as a candidate ferroptosis marker in tumor cells (Dixon et al., 2014); (b) genes encoding voltage-dependent channels located in the mitochondrial membrane, as they have been reported to be essential for erastin-induced ferroptosis in human cancer cells VDAC1, VDAC2, VDAC3, VDAC4, and VDAC5 (Yagoda et al., 2007); (c) cation transport regulator (Chac)-like genes, also reported as ferroptosis markers in cancer cells (CCL1, CCL2, and CCL3; Dixon et al., 2014); (d) BAXI-1, HRD1, SEL1, BiP1, BiP2, BiP3, and bZIP60, encoding for Arabidopsis ER stress markers, as an ER-stress signature has been found in erastin-induced cell death (Dixon et al., 2014); (e) PR2, PR5, ACS2, LSD, and WRKY33 (Zheng et al., 2006), implicated in the hypersensitive response; (f) Arabidopsis autophagy markers ATG7, ATG8, and ATG9 (Kim et al., 2012); (g) the vacuolar cell death–related gene VPE (Hara-Nishimura et al., 2005); (h) metacaspase genes MC1 and MC2 (Coll et al., 2010); (i) kiss of death (KOD), a gene encoding for a 25-aa peptide that induces cell death in A. thaliana roots (Blanvillain et al., 2011); and (j) as a control, RHD6 and WER, implicated in root development (Han et al., 2003; Coll et al., 2010; Hara-Nishimura and Hatsugai, 2011; Bruex et al., 2012; Robert et al., 2012). We analyzed the expression of these 30 genes 2 h after HS treatment and observed down-regulation of most genes, or slight up-regulation that was further enhanced by Fer-1 (e.g., GPT2a, VDAC5, and VPE). One exception was KOD, which was significantly up-regulated in a Fer-1–sensitive manner (17-fold; Fig. 5, b and c). Together with previous results showing that KOD is induced by heat stress and peroxide (Blanvillain et al., 2011), our results suggest that KOD might be acting downstream of GSH depletion and ROS accumulation in the cascade of events that lead to 55°C HS–induced ferroptotic-like cell death and could serve as a molecular marker, albeit nonspecific, for this process.

Our results indicated that 55°C HS triggered ferroptosis-like cell death in plants. We next assessed the potential relevance of ferroptosis in vivo under more physiological conditions. Plants’ inherent ability to tolerate high-temperature stress without prior conditioning is called basal thermotolerance. Basal thermotolerance is crucial, and together with acquired thermotolerance (acquired after acclimation), ultimately determines how well a plant copes with thermal changes in its environment. It has been shown that sudden HS induces the accumulation of ROS dependent on NOX activity (Wahid et al., 2007; Pucciariello et al., 2012). These data and our results with ferroptosis inhibitors suggest that plant cells might be dying in response to moderate HS via a NOX-dependent, ferroptosis-like pathway. Thus, ferroptosis inhibitors might ultimately find utility in the protection of crops during times of extreme temperature fluctuations, which is of increasing concern because of ongoing climate change. Thus, we assessed whether ferroptosis inhibitors could improve the ability of plants to cope with exposure to lethal temperatures at which many crops could be exposed in the environment. A temperature of 43°C was tested for 1 h, as when exposed to these conditions, plants without acclimation treatment collapse and bleach (Fig. 6; Meiri and Breiman, 2009). When plants were cotreated with Fer-1 or CPX, they showed a significantly increased ability to survive at this otherwise lethal temperature (Fig. 6). Although control plants submitted to a 43°C HS bleached in the course of 3 d, around 70% of the plants pretreated with CPX survived the treatment. The fraction of surviving plants was 30% when plants were pretreated with Fer-1. In addition, treatment with DPI also increased the rate of survival to 42% (Fig. 6). This result suggested that NOX activity was implicated in the pathway triggered by this treatment, meaning that oxidative stress might be a significant component of the heat-induced damage induced. Also, this shows that a similar process to the one observed in roots and cell suspensions in response to the 55°C HS is taking place in these plants.

Previous experiments in nonphotosynthetic tissues (roots and cell suspensions grown in the dark) showed that active chloroplasts are not essential for ferroptosis-like cell death. However, the aerial parts of plants exposed to 43°C were observed to die after HS, suggesting that chloroplasts might be involved in the cell death pathway triggered at this temperature in photosynthetic tissues. We tested this possibility by stressing the plants and maintaining them in the dark, before analysis. As shown in Fig. 6, seedlings kept in the dark died at lower rates than the ones exposed to the light, suggesting that active chloroplasts are contributing to cell death in leaves, either as a source of iron or as a source of ROS. This cell death was prevented when plants were preincubated with Fer-1 before 43°C HS. These results, obtained at temperatures that can be easily reached in the field, suggest that HS normally causes ferroptosis-like cell death in plants and that suppressing this death may be of practical value in agricultural applications.

Ferroptosis was described in tumor cells as a cell death mechanism involving iron-dependent ROS accumulation. This process can be triggered by GSH depletion (Yang et al., 2014) and results in cell death that is morphologically distinct from other known forms of cell death such as apoptosis, necrosis, and autophagy (Dixon et al., 2012). The results presented here demonstrate that a similar form of iron-dependent, ferroptosis-like cell death also takes place in plant cells. In particular, our results suggest a physiological role for ferroptosis-like death in plant responses to environmental stress.

Many of the hallmarks that characterize ferroptosis in animal cells are conserved in plants exposed to heat stress. Notably, dying cells showed normal nuclei but shrunken mitochondria, an accumulation of intracellular ROS, and depletion of GSH and AA. Moreover, death was prevented by the canonical ferroptosis inhibitors Fer-1 and CPX. These results suggest that plant cells undergoing ferroptosis cannot overcome the ROS accumulation induced by HS. As the depletion of neither GSH nor AA is reverted back by cotreatment with ferroptosis inhibitors, depletion of GSH and reduced AA may be an early event, probably before ROS accumulation through NOX activity and lipid peroxidation, as observed in tumor cells (Dixon et al., 2012; Yang et al., 2014). The mechanisms through which GSH is depleted in plant cells are not yet well understood. However, because of accumulation of unfolded proteins in the ER during heat stress (Yang et al., 2016b), more GSH might be consumed to reduce incorrectly formed disulfide bonds (Ozgur et al., 2014). This, together with the unfolding and inhibition of GSH biosynthetic enzymes, might explain GSH depletion and ROS accumulation in plant cells exposed to HS. Additionally, the low levels of reduced AA detected after HS might be caused by a failure in AA recycling attributable to the drop in GSH levels (Foyer and Noctor, 2011). In addition to soluble ROS via NOX activity, ferroptosis involves the accumulation of lipid ROS, which are typically formed through peroxidation of PUFA chains of membrane lipids (Skouta et al., 2014). The requirement of PUFA peroxidation for ferroptosis was recently reported in human cells by the use of PUFAs containing the heavy hydrogen isotope deuterium at the site of peroxidation (Yang et al., 2016a), which prevented PUFA oxidation and blocked ferroptosis. A similar result was obtained in Arabidopsis root hairs, where pretreatment with D-PUFAs protect root hairs exposed to a 55°C HS, preventing cell death (Fig. 3 e).

The study of the expression pattern of several genes related to cell death processes in plants or tumor cells showed that a recently described gene, KOD, which encodes a short peptide that regulates plant PCD, seems to be specifically regulated in Arabidopsis ferroptosis-like cell death (Blanvillain et al., 2011). Not only was KOD up-regulated after HS, but it was also induced in a ferroptosis-dependent manner (Fig. 5). This small peptide was shown previously to be induced by heat stress and peroxide (Blanvillain et al., 2011), which indicates that KOD might be acting downstream of GSH and AA depletion in the cascade of events that lead to cell death after HS in Arabidopsis. As Fer-1 is not able to prevent cell death induced directly by peroxide, but KOD has been shown to be induced upon peroxide application (Blanvillain et al., 2011), these results together suggest that the iron-dependent pathway shown herein might merge with the peroxide-induced pathway in which KOD was previously reported to act. The ability of Fer-1 to suppress KOD up-regulation suggests that ROS accumulation, downstream of GSH depletion, is an essential signal regulating KOD expression. Additionally, calcium efflux from the ER has been suggested to act downstream KOD. Thus, KOD may regulate calcium mobilization, which is in agreement with the requirement of calcium that we found for plant ferroptosis-like cell death (Fig. 1 b). Concordantly, a calcium requirement was also shown in a recently described form of oxidative cell death in mouse cells that appears to be related to ferroptosis (oxytosis). In oxytosis, calcium influx is observed downstream of GSH depletion and ROS accumulation (Henke et al., 2013). In summary, KOD is a candidate molecular marker of ferroptosis-like cell death in plants.

Fluctuations in environmental conditions occur naturally during plant development and reproduction. Extreme variations in temperature during hot summers are already known to cause significant agricultural yield losses (Bita and Gerats, 2013). Here we find that inhibitors of ferroptosis prevent the death of plants exposed to otherwise lethal temperatures. Importantly, the temperatures used in our thermotolerance experiments are conditions that can be reached in the field (HanumanthaRao et al., 2016). This finding indicates that the death of plants exposed to these lethal temperatures could involve a ferroptotic-like pathway. This may suggest new ways to protect crops during times of extreme temperature fluctuations, which is crucial as more extreme temperature events are expected with the ongoing climate change. Notably, a connection between HS and ferroptosis was also recently found in cancer cells (Sun et al., 2015). HSPB1 (also called mouse HSP25 or human HSP27), a member of the family of the small HS proteins (HSPs), is strongly induced after treatment with the ferroptosis trigger erastin in cancer cells, and the phosphorylated form of HSPB1 acts as a negative regulator of ferroptosis by inhibiting cellular iron uptake and lipid ROS production (Sun et al., 2015). In Caenorhabditis elegans, HS triggers a regulated form of cell death that was described as a regulated necrosis that involves calcium and the expression of HSPs (Kourtis et al., 2012). Overexpression of HSF-1, a transcription factor that regulates the expression of HSPs, suppresses HS-induced cell death in C. elegans. Notably, members of class A HSFs, including Arabidopsis HSF-1 and -3, play a fundamental role in regulating the HS response in plants. Furthermore, it was recently reported that overexpression of a constitutively active form of A-HSF1 caused the induction of many HSP genes and improved the thermotolerance of plants (Ohama et al., 2016), suggesting that similar pathways might be triggered in response to HS in plant and animal systems.

In conclusion, an iron-dependent, oxidative cell death process with biochemical and morphological similarities to ferroptosis, as described in mammalian cells, has a physiological role in plants, regulating plant cell death in response to heat stress. Although additional factors involved in this ferroptosis-like pathway remain to be identified in plants, many characteristics are conserved between plants and animal cells. These results not only suggest that ferroptosis might be a conserved form of cell death, but also raise interesting evolutionary questions. Ferroptosis might be an ancient form of cell death or the result of convergent evolution in plants. Either way, this iron-dependent oxidative form of cell death represents a common solution across kingdoms, whose evolutionary basis can be the subject of future studies.

Cell suspension cultures of A. thaliana were grown essentially as previously described (May and Leaver, 1993). In brief, cultures were grown in 50 ml of liquid Murashige and Skoog medium (basal salts 4.3 g/l) containing 0.5 mg/l 1-naphthylacetic acid (NAA), 0.05 mg/l kinetin, and 3% (wt/vol) sucrose, pH 5.8, with agitation on an orbital shaker (100 rpm) under dark conditions in a controlled environment room at 23°C. Cultures were subcultured every 7 d into fresh medium.

A. thaliana (Col-O) seeds were sterilized in 50% ETOH for 1 min and then 10 min in a bleach-SDS solution (50% bleach and 1% SDS), followed by extensive washing with sterile distilled water.

After sterilization, all seeds were plated in a single line on ATS growing medium to allow the roots to grow down the surface and stratified at 4°C for 48 h, before being placed vertically at 23°C, 16 h light, 8 h dark (Estelle and Somerville, 1987).

10 ml of a 7-d-old cell culture pretreated according each experiment was placed in sterile 50-ml flasks and treated with the indicated temperature in a water bath with shaking (85 oscillations/min). After heat treatment, flasks were returned to the controlled environment room, with shaking, until scoring 24 h later.

6-d-old seedlings were pretreated with the indicated solutions in a 1.5-ml tube in a controlled environment room. Heat treatments were performed in sterile distilled water, placing pretreated 6-d-old seedlings in a water bath at the indicated temperature without shaking for 10 min. After HS, the seedlings were returned to a controlled environment room at a constant temperature of 23°C in the light for the indicated times, after which cell death was scored.

Fer-1, CPX, DPI, Nec-1, CHO, and D-PUFAs (Retrotope) were applied to Arabidopsis roots 16 h before the HS treatment. For H₂O₂ treatments, seedlings were pretreated with or without Fer-1 in a 1.5-ml tube in a controlled environment room for 16 h. Seedlings were then treated with 150 mM H₂O₂ for 6 h. For NaCl treatments, seedlings were pretreated with or without Fer-1 in a 1.5-ml tube in a controlled environment room for 2 h, and then treated with 150 mM NaCl for 16 h. To study the optimal concentration to determine the effect of NaCl, three different concentrations of NaCl (50, 80, and 150 mM) were tested, and cell morphology was observed by DIC microscopy, to avoid a concentration that triggers a necrotic cell death response. Cell death after the three treatments was observed at similar rates, and for the three concentrations tested, cell death was not prevented with Fer-1 or with CPX. In addition, root hairs showed only partial retraction of cytoplasm, which appeared highly granulated for all three concentrations tested (a representative photo is shown in Fig. 2). The morphology observed was not necrotic. For the case of H₂O₂ treatments, ROS in root cells was assayed before treatments to ensure that plants were not undergoing oxidative stress. To adjust the experiment, different concentrations of H₂O₂ were tested (1, 25, 30, and 150 mM) with and without the addition of ferroptosis inhibitors. Whereas 1 mM was not enough to trigger cell death in root hairs, 15 mM H₂O₂ induced 40% of cell death, 25 mM and 30 mM induced ∼55%, and 150 mM induced cell death ∼65%. In all cases, the morphology of the root hairs was not necrotic; a representative picture is shown in Fig. 2.

For calcium requirement studies, 6-d-old seedlings were preincubated with 3 mM CaCl₂ for 16 h, 1 mM EGTA for 2 h, or EGTA for 2 h and then with 3 mM CaCl₂ for 16 h before inducing cell death by treating roots at 55°C for 10 min as explained earlier. 200 µM FeNa-EDTA, 5 µM CuSO₄, 3 mM Cl₂Mn, and 3 mM Cl₂Mg were applied for 16 h before inducing cell death by treating roots at 55°C for 10 min as explained earlier. For all treatments, root hairs were stained with Sytox green, and dead and living root hairs were quantified.

Cells/roots were stained with Sytox green, a nucleic acid stain that easily penetrates cells with compromised plasma membranes. Root hairs/cells were stained in a 1-µg/ml solution of Sytox green for 10 min. After incubation, root hairs were washed with 20 mM Hepes buffer, pH 7.2 (buffer A), and immediately quantified using confocal microscopy (Eclipse C1 Plus Confocal microscope using EZ-C1 3.80 imaging software and Ti-Control; Nikon). All images were taken at RT using buffer A as the medium with a CFI Super Fluor 40× oil lens (NA 1.3; Nikon). Roots were imaged using 15% laser power and a small pinhole. Images were organized using Photoshop CC (Adobe Systems). Cell death in cell suspensions was quantified by using a fluorometer (Fluoroscan II).

H₂DCFDA was prepared as a 10 mM stock solution in DMSO and frozen at −20°C. MitoSOX red was prepared fresh at a 5 mM concentrated solution in DMSO for each use. C11-BODIPY^581/591 was prepared at 2 mM in DMSO. H₂DCFDA, mitoSOX red, and C11-BODIPY^581/591 were purchased from Molecular Probes. Stock solutions were diluted 1:1,000 in buffer A in each experiment.

Cell cultures and 6-d-old plants were incubated for 10 min at RT with the working solution containing the dyes, in 12- or 24-well plates. After incubation, cells were washed with buffer A and immediately scored using a Fluoroscan. Roots were washed with buffer A and immediately quantified using confocal microscopy (Eclipse C1 Plus Confocal microscope using EZ-C1 3.80 imaging software and Ti-Control). All images were taken at RT using buffer A as the medium with a CFI Super Fluor 40× oil lens (NA 1.3). Roots were imaged using 15% laser power and a small pinhole. Images were organized using Photoshop CC.

Seedlings grown vertically for 6 d on ATS plates and subjected to the indicated treatments were fixed using 2.5% (vol/vol) glutaraldehyde in PBS buffer overnight at 4°C. In parallel, a fraction of the seedlings were not fixed and were stained with Sytox green to score for cell death in cortical cells. Cells from roots submitted to heat stress showed Sytox green staining in cortical cells as shown in Videos 1, 2, 3, and 4. After washing three times in PBS buffer, the specimens were postfixed with 1% (vol/vol) osmium tetroxide in the medium buffer for 1 h and washed twice in distilled water. Samples were dehydrated with 50%, 70%, 95%, and 100% ethanol and infiltrated and embedded in Spurr’s resin. Roots were sectioned from the tip upward using a glass knife on an EM UC7 ultramicrotome (Leica Biosystems). Roots were continuously oriented transversely to the root axis to allow recording of the distance from the root tip by counting sections. Ultrathin (80-nm) sections were made using a Diatome diamond knife on an EM UC7, collected onto copper grids, and treated with 5% uranyl acetate in water for 60 min followed by Sato’s lead staining for 5 min. Sections were examined in a transmission electron microscope (1230; Jeol). Digital images were captured using a Gatan MSC 600CW.

We examined buds and pistils composing the inflorescence ranging in size between 1 and 8 mm, looking for two key events involving cell death: megaspore cell death and synergid cell death. To test the effect of Fer-1 and CPX on reproductive development, Arabidopsis inflorescences were dipped for three consecutive days in a work solution containing either 10 µM CPX or 1 µM Fer-1 (in 0.1% DMSO and 0.01% Silwet L-77) or in a mock solution (0.1% DMSO and 0.01% Silwet L-77). For each treatment, five plants were treated, and a minimum of three inflorescences per plant were used for studies. Analysis of the treated inflorescences started on day 4. To analyze the buds in which we were expecting to find only a functional megaspore inside the ovules as a result of megaspore death, 1–1.25-mm pistils were selected. Embryo sacs were evaluated by DIC microscopy. To analyze whether synergid cell death was compromised, which would lead to fertilization problems, the ratio of fertilized to unfertilized embryo sacs was analyzed in self-pollinated flowers after treatment. To perform this experiment, we selected those flowers exposed to the treatments that showed signs of pollination (pollen on the stigma) at day 3. At day 4, the pistils were dissected, and fertilization was assessed by DIC microscopy.

Total RNA from leaves and roots was extracted using TRIzol reagent (Invitrogen) according to the manufacturer’s recommendations, and samples were treated with RQ1 RNase-free DNase (Promega) to remove DNA contamination. For cDNA synthesis, 1 µg of total RNA was reverse transcribed by IMPROM II (Thermo Fisher Scientific) using random primers (Biodynamics). Real-time PCRs were performed in a Step One real-time PCR system (Applied Biosystems) using SYBR green PCR master mix (Applied Biosystems) as described by the manufacturer. The cycling program was 40 cycles at 95°C for 10 min and 1 min at 60°C; a melting curve analysis was performed at the end of the PCR. Primer pairs were tested for specificity and for amplification efficiency with a standard cDNA dilution curve. The different biological samples were subsequently normalized against expression of UBQ5, encoding a ubiquitin gene. Quantitative expression analysis was conducted for at least three independent experiments including at least three independent biological replicates per sample in each experiment.

6-d-old seedlings growing at 22°C were exposed to 43°C for 1 h and then allowed to recover at 22°C for 5 d. Those seedlings that were still green and continued to produce new leaves were scored as survivors. Experiments were performed after pretreating the seedlings on a plate with 1:1,000 DMSO (control plates) or with 1 µM Fer-1 or 10 µM CPX for 16 h. Control plates exhibited a survival rate <10%.

Roots were harvested (∼400 mg) and powdered with liquid N₂, and AA and GSH were extracted and determined as previously reported (Griffith, 1980; Bartoli et al., 2006, respectively). Concentrations of both AA and GSH obtained in control seedlings root were as previously described for plants grew under similar conditions (Ramírez et al., 2013).

6-d-old seedlings were submitted to heat treatments at the indicated temperature for 10 min. After HS, the seedlings were returned to a controlled environment room at a constant temperature of 23°C in the light for 6 h. Then, TUNEL assays were performed with the Click-iT TUNEL Alexa Fluor imaging kit (Thermo Fisher Scientific) according to the manufacturer’s instructions.

Roots of 6-d-old seedlings were stained with MitoTracker green FM (200 nM; Molecular Probes) for 45 min. Mitochondrial images were obtained using confocal microscopy (Eclipse C1 Plus Confocal microscope using EZ-C1 3.80 imaging software and Ti-Control) at different times (0, 1, 3, and 6 h after HS). Automated quantification of mitochondrial area was performed using the Mito-Morphology macro for ImageJ software (version 1.39; National Institutes of Health) as described previously (Dagda et al., 2009). In brief, the green channel of cells stained with MitoTracker green FM was converted to grayscale and inverted (to show mitochondria-specific fluorescence as black pixels), and the threshold was set to optimally resolve individual mitochondria. The macro traces mitochondrial outlines using the command “analyze particles.”

Sequence similarity searches were performed on the National Center for Biotechnology Information website using protein–protein BLAST programs against the A. thaliana database.

Fig. S1 shows the effect of the necroptosis inhibitor Nec-1 and divalent transition metal ions on cell death induced by a 55°C HS treatment. Fig. S2 shows that 55°C HS in Arabidopsis cell suspensions triggers the accumulation of ROS, which can be inhibited by Fer-1, CPX, and DPI. Fig. S3 shows that mitoSOX is not involved in the cell death triggered by a 55°C treatment in Arabidopsis root hairs. Fig. S4 shows that caspase-like activity is required for cell death after 55°C HS, but DNA fragmentation is not detected. Fig. S5 shows the effect of ferroptosis inhibitors on reproductive and vascular development in A. thaliana. Table S1 shows the identity of the transcriptional pharmacodynamic ferroptosis markers reported in Dixon et al. (2014) and putative orthologues found in A. thaliana. Table S2 lists the primer pairs used for quantitative PCR. Videos 1, 2, 3, and 4 show Sytox green staining of the different cell types that compose the root after a 55°C HS treatment.

We thank Mikhail S. Shchepinov for providing D-PUFAs, Monica Kotler for providing Nec-1, and Maria Ana Contin for providing Ac-DEVD-CHO. We thank Daniela Villamonte for technical assistance with confocal microscopy.

This research was funded by grants to G.C. Pagnussat from Agencia Nacional de Promoción Científica y Técnica Argentina (PICT 1809) and from the Howard Hughes Medical Institute (International Early Career Scientist award 55007430); grants to B.R. Stockwell from the National Institutes of Health (5R01CA097061 and R35CA209896); R00 Award to S.J. Dixon from the National Cancer Institute (4R00CA166517-03); and a grant to D.F. Fiol from Agencia Nacional de Promoción Científica y Técnica Argentina (PICT 1524). A.M. Distéfano and S. D’Ippólito are postdoctoral fellows of Consejo Nacional de Investigaciones Científicas y Técnicas (CONICET); J.P. Córdoba, A.M. Bellido, and D. Soto are graduate fellows of CONICET. M.V. Martin, C.G. Bartoli, E.J. Zabaleta, D.F. Fiol, and G.C. Pagnussat are CONICET researchers.

The authors declare no competing financial interests.

Author contributions: Conceptualization, G.C. Pagnussat, S.J. Dixon, B.R. Stockwell, D.F. Fiol, E.J. Zabaleta, M.V. Martin, and A.M. Distéfano; methodology, G.C. Pagnussat, S.J. Dixon, B.R. Stockwell, D.F. Fiol, M.V. Martin, and A.M. Distéfano; investigation; M.V. Martin, A.M. Distéfano, G.C. Pagnussat, J.P. Córdoba, J.A. Roldán, A.M. Bellido, S. D’Ippólito, S.L. Colman, D. Soto, and C.G. Bartoli; formal analysis, M.V. Martin, A.M. Distéfano, and D.F. Fiol; supervision, G.C. Pagnussat, S.J. Dixon, B.R. Stockwell, and D.F. Fiol; and writing of the original draft, G.C. Pagnussat, S.J. Dixon, B.R. Stockwell, D.F. Fiol, M.V. Martin, and A.M. Distéfano.