ANAC044 orchestrates mitochondrial stress signaling to trigger iron-induced stem cell death in root meristems

Significance

Iron toxicity is a frequent environmental stress induced by flooded soil conditions that can cripple root growth and harm crop production. S-nitrosoglutathione reductase (GSNOR) has emerged as a crucial gene in safeguarding root growth from iron toxicity in various plant species, yet the underlying mechanism remains unclear. Our study exposes the mitochondrial retrograde signaling pathway that involves the transcription factor ANAC044, which constitutes the primary mechanism downstream of GSNOR. Pharmacological and genetic abolishment of the identified retrograde signal pathway could largely eliminate iron-induced stem cell death in the root meristem, thus not only exposing a highly relevant retrograde signaling pathway but also offering a promising target for engineering plant resilience to iron toxicity.

Abstract

While iron (Fe) is essential for life and plays important roles for almost all growth related processes, it can trigger cell death in both animals and plants. However, the underlying mechanisms for Fe-induced cell death in plants remain largely unknown. S-nitrosoglutathione reductase (GSNOR) has previously been reported to regulate nitric oxide homeostasis to prevent Fe-induced cell death within root meristems. Here, we found that in the absence of GSNOR, exposure to high Fe treatment results in DNA damage–dependent cell death specifically in vascular stem cells in root meristems within 48 h. Through a series of time-course transcriptomic analyses, we unveil that in the absence of GSNOR, mitochondrial dysfunction emerges as the most prominent response to high Fe treatment. Consistently, the application of mitochondrial respiratory inhibitors leads to stem cell death in root meristems, and pharmacological blockage of the voltage-dependent anion channel that is responsible for the release of mitochondrial-derived molecules into the cytosol or genetic changes that abolish the ANAC017- and ANAC013-mediated mitochondrial retrograde signaling effectively eliminate Fe-induced stem cell death in gsnor root meristems. We further identify the nuclear transcription factor ANAC044 as a mediator of this mitochondrial retrograde signaling. Disruption of ANAC044 completely abolishes the GSNOR-dependent, Fe-induced stem cell death in root meristems, while ectopic expression of ANAC044 causes severe root stem cell death. Collectively, our findings reveal a mechanism responsible for initiating Fe-induced stem cell death in the root meristem, which is the ANAC044-mediated GSNOR-regulated mitochondrial stress signaling pathway.

Multicellular organisms depend on stem cells and their differentiated progeny for their growth and development. In plants, stem cells in the root meristem are crucial for establishment and growth of root systems, which frequently encounter challenges posed by various environmental stresses such as drought, flooding, chilling, alkaline, and heavy metal exposure. Consequently, these environmental stresses can induce DNA damage, triggering a programmed cell death response specifically in stem cells within the root meristems (1, 2). A plant-specific NAC-type transcription factor SUPPRESSOR OF GAMMA RESPONSE 1 (SOG1) is a functional orthologue of p53 in animals that mediates DNA damage responses such as cell cycle arrest, DNA repair, and cell death (3). SOG1 regulates the expression of almost all the genes in response to DNA damage (4). Among direct targets of SOG1, two SOG1-like transcription factors ANAC044 and ANAC085 play a crucial role in causing G2 arrest and cell death in response to DNA damage in root meristems (4, 5). Distinct molecular regulatory modules of these NAC-type transcription factors are activated by various environmental stresses, including radiomimetic compounds, gamma (γ)-irradiation, heat, and osmotic stresses (5, 6). It remains unknown which mechanisms cause DNA damage and activate specific molecular regulatory modules to initiate stem cell death in the root meristem upon environmental stresses.

Plant mitochondria not only provide the primary energy to cells by producing ATP through oxidative phosphorylation but also act as key sensors to monitor cellular changes in response to harmful environmental conditions. As such, they can trigger retrograde signaling to adjust the nuclear transcriptional response to balance their function and plant growth (7, 8). Plant mitochondria harbor a standard oxidative phosphorylation machinery composed of four electron transport chain (ETC) complexes (I–IV), ATP SYNTHASE (Complex V), and two intermediary substrates ubiquinone and cytochrome c (9). Disruption of mitochondrial function by ETC complex inhibitors leads to the activation of mitochondrial retrograde signaling. The endoplasmic reticulum (ER)-anchored transcription factor ANAC017-ANAC013 module has been identified as the master regulator of this mitochondrial retrograde signaling in Arabidopsis (10–12). Interestingly, overexpression of Arabidopsis ANAC013 can constitutively cause cell death in the root meristems (13). It also has been reported that mitochondrial dysfunction and reactive oxygen species (ROS) accumulation are linked to the cell death of Arabidopsis protoplasts suffering from aluminum toxicity (14, 15). In animals, ROS derived from mitochondria can act as important signaling molecules to cause cell death (16).

Iron (Fe) is an essential micronutrient for cell and organismal function, but it also can damage cell growth when presented in excess. In animals, Fe has been found to be a trigger or mediator of cell death signaling and linked to many diseases (17, 18). In higher plants, Fe toxicity not only represents a common stress in flooded soil but also is a critical intrinsic factor in inhibiting root growth and crop production. Fe toxicity of rice can potentially occur in around 18% of global paddy fields and cause 10 to 100% yield loss depending on the different rice varieties (19–21). The toxicity of cellular Fe accumulation can also suppress plant growth and limit the generation of Fe-fortified plant varieties that are required for eradicating the hidden hunger of the growing population (22, 23). Moreover, it has been reported that the toxicity of cellular Fe accumulation contributes to the root growth inhibition caused by other stresses including phosphate deficiency (24) and ammonium toxicity (25). However, it is yet not clear whether like in animals, Fe could be a trigger or mediator of stress-responsive programmed cell death in plants.

S-nitrosoglutathione reductase (GSNOR) is an enzyme that modulates a multitude of abiotic and biotic stress responses including cell death in higher plants through its conserved enzyme function in the S-nitrosoglutathione metabolism and nitric oxide homeostasis (26). GSNOR was previously identified to determine plant root growth in response to high Fe in different plant species and found that loss of function of GSNOR leads to cell death in root meristems upon high Fe stress (27). This property of cell death showed a high similarity to DNA damage–induced stem cell death in root meristems (1, 5). In this study, we show that in the absence of GSNOR, high Fe indeed induces DNA-damage-dependent stem cell death in root meristems. We demonstrate that mitochondria dysfunction signaling emerges as the early primary response of GSNOR-regulated Fe-induced oxidative damage and activates the nuclear transcription factor ANAC044 to initiate Fe-induced stem cell death in root meristems.

Results

High Fe Induces DNA Damage–Dependent Stem Cell Death in Root Meristems of GSNOR Null Mutant.

Our previous work had identified GSNOR to be involved in preventing high Fe-induced cell death in the root meristem (27). We therefore set out to investigate the spatiotemporal characteristics of Fe-induced cell death in the Arabidopsis root apical meristem. As shown in Fig. 1A and SI Appendix, Fig. S1A, cell death occurred not at all or only very sporadically in the root meristems of the wild type (Col-0), gsnor mutant (hot5-2), and the complemented mutant line (GSNOR-GFP) one day after transfer to 350 µM Fe(III)-EDTA (high Fe). At day 2 and day 3 after transfer to high Fe, we observed frequent cell death specifically in the vascular stem cell population (stele initial cells) of root meristems in the hot5-2 mutant, while this did not occur in Col-0 and the complemented line (Fig. 1 A and B and SI Appendix, Fig. S1 A and B). To exclude that this might be an effect of the chelating agent EDTA, we also treated the hot5-2 mutant with 350 µM EDTA-Na₂ and 500 µM ferric citrate. EDTA did not significantly cause cell death in the hot5-2 root meristem while ferric citrate significantly elevated both frequency and area of cell death (SI Appendix, Fig. S1 C–H). These results demonstrated that the cell death in the hot5-2 root meristem is caused by high Fe treatment.

Fig. 1.

High Fe induces DNA damage–dependent stem cell death in root meristems in the absence of GSNOR. (A) Propidium iodide (PI) staining showing vascular stem cell death (stele initial cells are indicated by the white arrow) in the hot5-2 root meristem upon high Fe treatment. Four-day-old seedlings of Col-0, the gsnor mutant (hot5-2), and the complemented line (GSNOR-GFP) grown in the normal medium (5 μM Fe) were transferred to high Fe (350 μM Fe) for additional 0, 1, 2, and 3 d. PI is limited to cell walls in live cells but penetrates into dead cells. “n” indicates the number of seedlings used for cell death staining. (Scale bar, 20 μm.) (B) Frequency of stem cell death in the root meristem occurred in each genotype under the given condition shown in (A). (C) Knockout of SOG1 alleviated the frequency of Fe-induced stem cell death in hot5-2 root meristems. Four-day-old seedlings were transferred to 350 μM Fe for 2 d. In (B) and (C), the number of seedlings with and without stem cell death is shown on the red and gray bars, respectively. Different letters indicate significant differences between groups by pairwise Fisher’s exact test (P < 0.05). The representative image of cell death in root meristems is shown in (A), when the frequency of cell death in the given genotype was significantly higher than that of Col-0 in control based on Fisher’s exact test.

Environmental stresses such as irradiation can cause DNA damage, specifically leading to the death of stele stem cells within the root apical meristem (1, 5). Considering that the cell death caused by high Fe treatment in hot5-2 root meristem shares similarities with this phenotype (Fig. 1A and SI Appendix, Fig. S1A), we therefore asked whether GSNOR is implicated in the pathway associated with DNA damage–induced stem cell death. To test this, we first assessed the effect of high Fe treatment on DNA damage in the root apical meristem by using TUNEL staining (28). The result indicated much severe DNA damage in the root meristem of the hot5-2 mutant compared to Col-0 after treatment with 350 µM Fe for 2 d (SI Appendix, Fig. S2 A and B); however, the staining was only observed on the root surface rather than stem cells due to the inherent limitation of this technique, such as permeability and sensitivity of dye (29). To further test whether this Fe-induced stem cell death depends on the DNA damage response pathway, we pursued a genetic approach by generating a hot5-2sog1-101 double mutant, as knockout of SOG1 can abolish the DNA damage–induced stem cell death in the root apical meristems (5). Consistent with our hypothesis, knockout of SOG1 greatly alleviated both frequency and area of Fe-induced stem cell death in root apical meristems of the hot5-2 mutant plants (Fig. 1C and SI Appendix, Fig. S2 C and D). Taken together, these results revealed that high Fe treatment induces the DNA damage–dependent stem cell death in the hot5-2 root meristem.

To test whether GSNOR’s function in regulating nitric oxide was responsible for Fe-induced stem cell death, we utilized the REPRESSOR OF GSNOR1 (ROG1/CAT3, a transnitrosylase) null mutation, which rescues hot5-2 mutant shoot phenotypes through regulating nitric oxide signaling (30). Consistent with the role of nitric oxide regulation, in contrast to the hot5-2 single mutant, the double mutant of hot5-2rog1-10 showed comparable stem cell death in the root meristem induced by high Fe treatment to that of Col-0 (SI Appendix, Fig. S3). This result indicated that GSNOR-regulated nitric oxide signaling is critical to cause Fe-induced stem cell death in the root meristem.

As high Fe-induced damage is linked to ROS accumulation (17), we assessed the accumulation of two major ROS molecules, hydrogen peroxide (H₂O₂) and superoxide anion (O₂^·−) by 3,3′-diaminobenzidine (DAB) staining and nitroblue tetrazolium (NBT) staining in the root meristem upon high Fe treatment, respectively. Quantification of DAB staining indicated that the accumulation of H₂O₂ in the hot5-2 root meristems was lower than that of Col-0 at all time points of high-Fe treatment (SI Appendix, Fig. S4 A and B). Unlike the stable level of H₂O₂ accumulation observed in Col-0, a significant increment of H₂O₂ was detected in hot5-2 root meristems at 6 h (Dunnett’s multiple comparisons test, P = 0.03) after high Fe treatment, followed by a sharp decrease starting from 24 h posttreatment (SI Appendix, Fig. S4 A and B). However, neither knockout of the major catalase CATALASE 2 (CAT2), a peroxisomal catalase functioning as a negative regulator of H₂O₂ accumulation, nor the supply of potassium iodide (KI), an H₂O₂ scavenger, could clearly change the frequency of Fe-induced stem cell death in the hot5-2 mutant root meristems (SI Appendix, Figs. S3 and S4 C–E). Interestingly, the area of cell death was statistically reduced by the knockout of CAT2 in the hot5-2 mutant (SI Appendix, Fig. S3). Taken together, these results suggest that the observed H₂O₂ alteration in the hot5-2 root meristem may be not the major inducer of stem cell death; however, sufficient accumulation of H₂O₂ may prevent the expansion of stem cell death. Quantitative analysis of NBT staining demonstrated that O₂^·− accumulation in root meristems of hot5-2 plants was higher than that of Col-0 at the early time points of high Fe treatment, but significantly decreased at 48 h posttreatment (SI Appendix, Fig. S4 F and G). Importantly, the application of N, N'-dimethylthiourea (DMTU), a scavenger of O₂^·−, significantly reduced both frequency and area of stem cell death in the hot5-2 root meristems (SI Appendix, Fig. S4 H–J). Taken together, these results indicate that a defect of GSNOR strongly alters the pattern of H₂O₂ and O₂^·− accumulation in the root meristem in response to high Fe and that O₂^·− production is likely associated with Fe-induced stem cell death in the hot5-2 root meristem.

To test whether the GSNOR-regulated Fe-induced stem cell death is like ferroptosis that is driven by Fe-dependent lipid peroxidation (31), we monitored lipid peroxidation in the root meristem by measuring the content of malondialdehyde (MDA), a lipid peroxidation marker (32), with the staining of Schiff’s reagents (SI Appendix, Fig. S5A). No difference in MDA accumulation indicated that lipid peroxidation was not increased in hot5-2 root meristems upon high Fe treatment (SI Appendix, Fig. S5 A and B). In parallel, application of ciclopirox olamine (CPX) and ferrostatin-1 (Fer-1), two selective ferroptosis inhibitors that inhibiting lipid peroxidation (31), did not alleviate stem cell death induced by high Fe in the root meristem of the hot5-2 mutant in terms of either frequency or area (SI Appendix, Fig. S5 C–E). Thus, the GSNOR-regulated Fe-induced stem cell death in Arabidopsis root meristems is not ferroptosis-like cell death.

Lack of GSNOR Activates Mitochondrial Dysfunction Signaling upon High Fe Treatment.

To investigate the early molecular signatures of GSNOR-regulated root responses to high Fe treatment, we performed a time-course transcriptome analysis of Col-0 and the hot5-2 mutant roots in response to high Fe treatment at different time points (0 h, 3 h, 6 h, and 24 h). Using these data, we performed a principal component analysis (PCA) of all detected genes in these samples. While the biological replicates clustered together, there were significant differences between genotypes as well as treatment duration (SI Appendix, Fig. S6). The result indicated that our time-course transcriptome analysis could distinguish the effects of GSNOR and high Fe treatment on root growth.

We then wanted to identify genes and biological processes that depend on GSNOR and high Fe treatment. By utilizing DEseq2 with the time-course model (likelihood ratio test, adjusted P-value < 0.001), we identified a total of 970 differential expressed genes (DEGs) that were affected by both GSNOR mutation and over the time-course of high Fe treatment (Fig. 2A and Dataset S1). We then classified the 970 DEGs into 10 clusters using a divisive hierarchical clustering approach (Fig. 2B and SI Appendix, Fig. S7) and performed overrepresentation analysis (ORA)-based gene ontology (GO) enrichment for the genes within each cluster. Both unique as well as shared GO terms were found to be enriched in the clusters (SI Appendix, Fig. S8). One of the most interesting clusters was cluster 4 in which genes in Col-0 were induced at early stages (3 h and 6 h) followed by a recovery at 24 h posttreatment, which was not present in the hot5-2 mutant. In this cluster, the second most enriched GO term was “response to heat” (Fig. 2C and SI Appendix, Fig. S8), which matches the role of GSNOR in heat tolerance that was initially discovered in Arabidopsis mutant hot5-2 through an EMS-mutagenized forward genetic screening (33), even though the relationship between Fe level and heat tolerance has not yet been reported. One group of GO terms that was enriched in cluster 4 caught our attention. Among the top 10 enriched GO terms in cluster 4, several are related to mitochondria and encompassed “mitochondria organization,” “protein targeting to mitochondrion,” “NADH oxidation,” and “mitochondrial transport” (Fig. 2C and SI Appendix, Fig. S8). Mitochondria are the primary source of ROS production and their function has been shown to be regulated by GSNOR in animals (34, 35). To test high Fe and GSNOR-dependent gene enrichment in an orthogonal manner, we conducted a Gene Set Enrichment Analysis (GSEA) using all genes. Among the most enriched GO terms, GO terms “DNA repair” and “mitochondrial gene expression” were identified to have a significantly activated tendency (NES > 1, adjusted P-value < 0.05) in the hot5-2 mutant throughout the entire high Fe treatment (Fig. 2 E and F, and SI Appendix, Figs. S9–S11), while genes involved in direct organization, substance transportation, and membrane protein targeting of mitochondria were also significantly induced after 24 h of high Fe treatment in the hot5-2 mutant (Fig. 2 E and F and SI Appendix, Figs. S8–S11). We then wanted to test whether marker genes of mitochondrial dysfunction stimulus (MDS) (10) were differentially regulated in the expression dataset. All MDS genes that are expressed in the root were more strongly induced by high Fe treatment in the hot5-2 mutant than in Col-0 (Fig. 2D). Overall, these results indicated that high Fe is likely to disrupt mitochondrial function and trigger mitochondrial dysfunction signaling in the hot5-2 roots. As the “mitochondrial envelope” represents the most enriched cellular component GO term among the genes induced in the hot5-2 mutant at 24 h Fe posttreatment (Fig. 2F), we therefore wanted to test whether the functionality of mitochondria is affected in the hot5-2 mutant by determining the mitochondria membrane potential (MMP) with Tetramethyl rhodamine methyl ester (TMRM) staining (36). The TMRM fluorescence was comparable between Col-0 and the hot5-2 mutant under normal Fe supply, while it was significantly higher in the hot5-2 mutant compared to Col-0 under high Fe treatment (Fig. 3 A and B). This result suggests a hyperpolarized MMP, which would lead to an alteration of mitochondrial functionality in the root of the hot5-2 mutant in response to high Fe treatment.

Fig. 2.

High Fe treatment activates mitochondrial dysfunction signaling in the GSNOR null mutant. (A) Scheme of the experimental design for time-course transcriptomic analysis to identify GSNOR-regulated and Fe-responsive genes in Arabidopsis roots. Consequently, 970 differentially expressed genes (DEGs) were identified using the DEseq2 time-series analysis module at a threshold of P.adj < 0.001. (B) Clustering of the 970 DEGs based on the transcriptional expression patterns. Normalized expression data (read counts/sample specific size factor) were scaled to Z-score values (across genotypes as well as time points) for all of the 970 DEGs and plotted. (C) Network map of the top 30 overrepresented biological process Gene Ontology (GO) terms enriched in cluster 4 genes shown in (B). Circle size and color represent the number of genes belonging to a certain GO term and the significance of enrichment, respectively (P < 0.05 after Benjamini–Hochberg adjustment). GO terms with gene overlaps higher than 20% are connected with gray lines, and the thickness of the lines reflects the degree of gene overlap. (D) Expression profiles of mitochondrial dysfunction stimulus (MDS) genes in Col-0 and hot5-2 roots. Read count normalization and conversion were consistent with those described in (B). (E and F) Standard gene set enrichment analysis (GSEA) showing mitochondria-related biological process (E) and cellular component (F) GO terms enriched in the hot5-2 mutant upon high Fe treatment. Break lines in the Upper panel represent the running enrichment scores of the GO terms; the position of vertical bars in the Middle panel shows the fold change rank of a certain input gene belonging to a specific GO term; while the gray lines in the Lower panel showing the log2 fold change value [hot5-2/Col-0] of the genes in a decreasing order. The normalized enrichment scores (NES) as well as the adjusted P values of the enriched GO terms are also indicated in the Upper panel.

Fig. 3.

Disruption of mitochondrial function or mitochondrial retrograde signaling modulates stem cell death in the root meristems. (A) Representative images of tetramethyl rhodamine methyl ester (TMRM) staining for mitochondrial membrane potential assay. Four-day-old seedlings were treated with 5 μM and 350 μM Fe for 48 h. (Scale bars, 20 μm.) (B) Quantification of TMRM fluorescence intensity in epidermal cells of root elongation zone in (A). Data are presented as mean ± SD alongside individual values. “n” indicates the number of epidermal cells from at least six independent seedlings for each genotype. Different letters indicate the significant differences by one-way ANOVA followed by Tukey’s HSD test (P < 0.05). (C) Representative images showing the expression of pAOX1a:GFP reporter activated by high Fe in the root tips of hot5-2. Four-day-old seedlings grown in 5 μM Fe were exposed to 350 μM Fe for 0, 6, 24, and 48 h, respectively. (Scale bar, 20 μm.) (D) Quantification of fluorescence intensity of pAOX1a:GFP shown in (C). Data are presented as mean ± SD alongside individual values. “n” indicates the number of seedlings used for the assay. Different letters indicate the significant differences by one-way ANOVA followed by Tukey’s HSD test (P < 0.05). (E) Scheme showing mitochondrial respiration electron transport chain (ETC) complexes and their inhibitors. (F) Frequency of stem cell death in root meristems of Col-0 and hot5-2 seedlings under different ETC inhibitors’ treatment. Five-day-old seedlings were transferred to half MS medium supplied with different ETC inhibitors and the corresponding control as indicated in the figure for 2 d. (G) DIDS treatment significantly reduced the frequency of stem cell death in hot5-2 root meristems. Four-day-old seedlings were transferred to the medium containing 350 μM Fe with or without 100 μM DIDS for 48 h. (H) Overexpression of ANAC013 causes stem cell death in the root meristem. Four-day-old root tips of Col-0, the ANAC013 overexpression lines (anac13-1D, pUBQ10:ANAC013), and the ANAC017 overexpression line (35S:ANAC017) were exposed to 350 μM Fe for 48 h. (I) Simultaneously disruption of ANAC013 and ANAC017 reduces the frequency of stem cell death in hot5-2 root meristems upon high Fe treatment. Four-day-old seedlings were exposed to 350 μM Fe for 48 h. For (F–I), the number of seedlings with and without stem cell death is shown on the red and gray bars, respectively. Different letters indicated significant differences between groups by pairwise Fisher’s exact test (P < 0.05).

Next, we asked whether mitochondrial dysfunction responses can be observed in the hot5-2 root meristem when treated with high Fe. To test this hypothesis, we generated a GFP-fusion reporter using the promoter of a well-known MDS marker gene, ALTERNATIVE OXIDASE 1a (AOX1a) (10, 11). We then tested its response to high Fe treatment in both Col-0 and the hot5-2 mutant background. The fluorescence of pAOX1a:GFP in the root meristem of the hot5-2 mutant was significantly higher than that of Col-0 at early time points of high Fe treatment (Fig. 3 C and D). The difference between the hot5-2 mutant and Col-0 amounted to 45-fold and 137-fold when they grew in the medium with 350 µM Fe at 24 h and 48 h after treatment onset (Fig. 3 C and D). Overall, these results supported the hypothesis that high Fe disrupts mitochondrial function and triggers mitochondrial dysfunction signaling in root meristems of hot5-2 mutant plants.

Mitochondrial Dysfunction Causes DNA Damage and Stem Cell Death in Root Meristems.

We next set out to test the causality between DNA damage and mitochondrial dysfunction during GSNOR-regulated Fe-induced stem cell death in the root meristem. We first asked whether mitochondria dysfunction is a consequence of DNA damage. We confirmed that 10 µg/ml zeocin, a known DNA damage–inducing compound, indeed causes stem cell death in Col-0 root meristems (SI Appendix, Fig. S12 A–C). This was even more pronounced in the hot5-2 mutant background that displayed a larger area of cell death (SI Appendix, Fig. S12 A and C). However, the same treatment with 10 µg/ml zeocin did not activate the expression of pAOX1a: GFP, a marker of mitochondrial dysfunction response (SI Appendix, Fig. S12 D and E). This result therefore indicates that mitochondrial dysfunction is not the consequence of DNA damage in the root meristem of the hot5-2 mutant upon high Fe treatment.

Next, we asked whether mitochondria dysfunction causes stem cell death in the root meristem. To test this hypothesis, we first examined the effect of different mitochondrial respiration electron transport chain (ETC) complex inhibitors on stem cell death of root meristems in both Col-0 and the hot5-2 mutant backgrounds (Fig. 3 E and F). We found that all mitochondrial ETC inhibitors were able to increase the frequency of stem cell death in root meristems in the hot5-2 mutant (Fig. 3F and SI Appendix, Fig. S13 A and B). Interestingly, rotenone, NaN₃, and oligomycin, inhibitors of mitochondrial ETC complex I, IV, and V, caused a significantly increased frequency of stem cell death in root meristems compared to other mitochondrial ETC inhibitors, while oligomycin can even significantly induce stem cell death in the root meristem in Col-0 background (Fig. 3F and SI Appendix, Fig. S13 A and B). These results revealed that mitochondrial dysfunction can cause stem cell death in the root meristem independent of Fe toxicity and GSNOR function and suggested that functionality of mitochondria exhibits heightened sensitivity to ETC inhibitors in the hot5-2 mutant compared to that of Col-0. Accordingly, we hypothesized that mitochondrial dysfunction could be a consequence of Fe toxicity, thus leading to stem cell death in the hot5-2 root meristem. To further test whether mitochondrial dysfunction causes DNA damage in the root meristem, we examined the effect of NaN₃ and oligomycin on DNA damage in the root meristem using TUNEL staining. DNA damage in the root meristem was significantly increased by NaN₃ and oligomycin treatment, but no statistical difference was observed between Col-0 and the hot5-2 mutant (SI Appendix, Fig. S13 C and D). Overall, these results demonstrated that mitochondrial dysfunction can cause DNA damage and stem cell death in the root meristem.

Given that mitochondrial dysfunction can cause stem cell death in the root meristem, we assumed that mitochondrial dysfunction signaling is required to induce stem cell death in the hot5-2 root meristem in response to high Fe. To test this hypothesis, we applied 4,4′diisothiocyanatostilbene-2,2′-disulfonate (DIDS), an inhibitor of the voltage-dependent anion channel (VDAC) located on the mitochondrial outer membrane that mediates the transport of mitochondrial molecules into the cytosol (37). Application of DIDS significantly reduced the frequency and area of cell death in the hot5-2 root meristems to the wild-type level (Fig. 3G and SI Appendix, Fig S13 E and F), supporting that mitochondrial dysfunction signaling plays a vital role in Fe-induced stem cell death.

The ANAC017-ANAC013 Module Contributes to Fe-Induced Stem Cell Death in the hot5-2 Root Meristem.

Next, we wanted to explore the molecular mechanisms underlying high Fe-induced mitochondrial dysfunction signaling in the hot5-2 root meristem. We first wanted to look at known regulators of mitochondrial retrograde signaling and chose the transcription factor ANAC017-ANAC013 module as it acts as a master regulator of mitochondrial retrograde signaling in Arabidopsis leaves (10–12). Consistent with a previous finding (13), we found that overexpression of ANAC013 causes serious cell death in the root meristem under the normal condition (Fig. 3H and SI Appendix, Fig. S14A), and anac13-1D had a larger area of cell death in the root meristem in response to high Fe (SI Appendix, Fig. S14B). However, we did not detect a comparable level of frequency of cell death in the root meristem of the ANAC017 overexpression line compared to ANAC013 overexpressing plants (Fig. 3H and SI Appendix, Fig. S14A). Nevertheless, the area of cell death was significantly increased in the ANAC017 overexpression line under the high Fe condition compared to Col-0 but still significantly lower compared to ANAC013 overexpressing plants (SI Appendix, Fig. S14B). Importantly, we further found that the transcriptional level of ANAC013 was increased upon high Fe treatment in hot5-2 roots (SI Appendix, Fig. S14C). While the transcript level of ANAC017 was much higher than that of ANAC013 but not elevated upon high Fe treatment (SI Appendix, Fig. S14C), the level of GFP-ANAC017 fusion protein (pANAC017:GFP-ANAC017) was increased, especially in the nucleus, by high Fe treatment in hot5-2 root meristem cells (SI Appendix, Fig. S15). Therefore, we wondered whether the canonical mitochondrial retrograde signaling regulatory component ANAC017-ANAC013 module contributes to the GSNOR-regulated and Fe-induced stem cell death in the hot5-2 root meristem. We generated hot5-2 mutant combinations harboring anac13-1 CRISPR/Cas9 knockouts and/or the anac17-1 mutation from a T-DNA line (Fig. 3I and SI Appendix, Fig. S16 A and B). Both frequency and area of stem cell death in the root meristem of hot5-2anac13-1 and hot5-2anac17-1 double mutants were similar or slightly lower than that of the hot5-2 mutant, while they were greatly reduced in hot5-2anac13-1anac17-1 triple mutants (Fig. 3I and SI Appendix, Fig. S16 C and D). However, it should be noted that the frequency of stem cell death in the root meristem of hot5-2anac13-1anac17-1 triple mutants was still significantly higher than that of Col-0 (Fig. 3I and SI Appendix, Fig. S16C). Taken together, these results suggest that ANAC017 and ANAC013 contribute to GSNOR-regulated Fe-induced stem cell death in the root meristem, while additional transcription factor(s) involved in this process likely exist.

The Nuclear Transcription Factor ANAC044 Mediates the Signaling of Mitochondrial Dysfunction.

We next wanted to identify other transcription factors that mediate the mitochondrial retrograde signaling triggered in GSNOR knockout plants upon high Fe treatment. Notably, ANAC044, another NAC transcription factor, stood out as one of the most prominent genes (adjusted P-value, 8.83E-21) within the 970 DEGs and displayed a strong correlation (correlation coefficient, 0.985) with ANAC013 (Dataset S1). Moreover, according to the ATTED-II database, approximately 40% of the top 100 coexpression genes of ANAC013 and ANAC044 are overlapping (Fig. 4A). Importantly, 21 of the 24 MDS genes (10) are coexpressed either with ANAC013 or ANAC044, and 15 of those are coexpressed with both of these ANAC genes (Fig. 4A). GO enrichment analysis of the top 100 ANAC044 coexpressed genes confirmed that, in addition to genes consistent with the reported biological processes such as DNA repair and cell cycle, ANAC044 is also coexpressed with the genes involved in the organization of mitochondria as well as the mitochondria-nucleus signaling pathway (Fig. 4B). In order to determine the role of ANAC044 in response to high Fe treatment, we performed a single gene GSEA of ANAC044 with its Spearman correlation coefficients of all the 20,619 expressed genes in our 24 time-course transcriptomic samples. Results demonstrated that mitochondrial-related GO terms are the most abundant among the top 30 enriched GO terms whose genes show positively correlated expression patterns to ANAC044 (Fig. 4C and SI Appendix, Figs. S17 and S18). ANAC044 and its closest relative ANAC085 were previously reported to be involved in cell death induced by DNA damage in the root meristem (5), but ANAC085 was not significantly induced in the hot5-2 roots upon high Fe treatment (SI Appendix, Fig. S14C). This suggested that ANAC044 might have a role in the transduction of mitochondrial retrograde signaling triggered by GSNOR knockout and high Fe treatment.

Fig. 4.

Mitochondrial dysfunction signaling activates the nuclear transcription factor ANAC044 in root meristems. (A) MDS genes are enriched in the top 100 coexpression genes of ANAC013 and ANAC044. The top 100 coexpression genes of ANAC013 and ANAC044 are collected from an online database (https://atted.jp), and the 24 MDS marker genes refer to a previous report (11). (B) Network map of the top 30 overrepresented biological process GO terms enriched in the top 100 ANAC044 coexpression genes. Circle size and color represent the number of genes belonging to a certain GO term and the significance of enrichment, respectively (P < 0.05 after BH correction). GO terms with gene overlap higher than 20% were connected with gray lines, and the thickness of the lines reflects the degree of gene overlap. (C) Single gene GSEA identified mitochondria-related biological processes GO terms are enriched in ANAC044 positively correlated genes using the time-course RNA-seq data obtained in this study. (D) The expression of pANAC044:GUS in Col-0 and hot5-2 roots in response to high Fe. Four-day-old seedlings were incubated at 350 μM Fe for 0, 6, 24, and 48 h. (E) ETC inhibitors NaN₃ and oligomycin induce the expression of pANAC044:GUS in the root meristem. Five-day-old seedlings were incubated at the mock conditions and ETC inhibitors for 48 h. (F) High Fe induces the expression and accumulation of GFP-ANAC044 fusion protein in the nucleus of the hot5-2 root meristem. Four-day-old seedlings were transferred to 5 μM Fe and 350 μM Fe for 48 h. (G) Knockout of ANAC017 significantly reduces the expression of ANAC044 in the GSNOR null mutant. Five-day-old seedlings were incubated at 5 μM Fe and 350 μM Fe for 24 h. Relative changes in gene expression were determined by qRT-PCR and compared to normal Col-0 samples. Data are presented as mean ± SD alongside individual values, n = 3 biological replicates. Different letters indicate the significant differences between groups with one-way ANOVA followed by Tukey’s HSD test (P < 0.05). (Scale bars in (D–F) are 20 μm.)

In agreement with this hypothesis, the pANAC044:GUS reporter was only induced in hot5-2 root meristems upon high Fe and not in the corresponding Col-0 controls (Fig. 4D and SI Appendix, Fig. S19A). We then tested the effect of inducing mitochondrial dysfunction on the expression of pANAC044:GUS by using different mitochondrial ETC inhibitors. The expression of pANAC044:GUS in the root meristem was slightly induced by mitochondrial ETC inhibitors rotenone, malonate, and Antimycin A, and strongly induced by NaN₃ and oligomycin (Fig. 4E and SI Appendix, Fig. S19B). To explore the role of the ANAC044 protein, we generated a GFP-ANAC044 fusion protein (pANAC044:GFP-ANAC044). We tested the functionality of this fusion protein by complementing the anac44-1 mutant with it and found that it could recover the sensitivity of cell death to zeocin in the root meristem (SI Appendix, Fig. S19 C–E). Consistent with the transcriptional GUS reporter data, the fluorescence of GFP-ANAC044 was not detected in Col-0 background, while it was clearly induced in the nucleus of hot5-2 root meristem cells when treated with high Fe for 2d (Fig. 4F and SI Appendix, Fig. S19F). As it was reported that the expression of ANAC044 is directly regulated by ANAC017 (38), we tested whether knockout of ANAC017 would reduce the induction of ANAC044 in the hot5-2 root meristem cells upon treatment with high Fe. Indeed, the expression of ANAC044 in the roots of hot5-2 mutant under high Fe condition was significantly reduced by 50% (P < 0.05) when present in a knockout of ANAC017 (Fig. 4G). This result suggested the activation of ANAC044 expression is partially dependent on the ANAC017-mediated canonical mitochondrial retrograde signaling pathway and that mitochondrial dysfunction signaling can activate the expression of DNA damage–responsive nuclear transcription factor ANAC044.

We next tested whether ANAC044 is required to activate the expression of MDS genes. To confirm this hypothesis, we examined the effect of ANAC044 knockout on the expression of MDS genes that are induced in the roots of hot5-2 mutant treated with high Fe by using qRT-PCR. The expression level of 75% of MDS genes that we tested was significantly reduced in the hot5-2anac44-1 double mutant upon high Fe treatment for 1d compared to the hot5-2 single mutant. These genes include UGT74E2, ERF71, CRF6, AT2G41730, AT2G21640, and ANAC013 (Fig. 5A). This suggested that the nuclear transcription factor ANAC044 is required to activate the expression of most MDS genes induced by GSNOR knockout and high Fe treatment. While transcription levels of ANAC017 were comparable in Col-0 and hot5-2 mutant under both control and high Fe conditions, it was significantly reduced in the hot5-2anac44-1 double mutant under the high Fe condition compared to Col-0 and the hot5-2 single mutant (Fig. 5A). Taken together, our results identified that the transcription factor ANAC044 controls mitochondrial retrograde signaling in response to mitochondrial dysfunction.

Fig. 5.

ANAC044 activates the GSNOR-regulated Fe-induced MDS signaling and stem cell death in root meristems. (A) ANAC044 is required to activate the transcriptional expression of MDS genes as well as ANAC017 in the roots of the hot5-2 mutant. Five-day-old seedlings were transferred to 5 μM and 350 μM Fe for 24 h. Transcript abundance was determined by RT-qPCR and compared to normal Col-0 samples. UBC21 was used as the internal control. Data are presented as mean ± SD alongside individual values, n = 3 biological replicates. Different letters indicate significant differences between groups with one-way ANOVA followed by Tukey’s HSD test (P < 0.05). (B) Knockout of ANAC044 abolishes the Fe-induced stem cell death in the root meristem of the hot5-2 mutant. Four-day-old seedlings grown under 5 μM Fe were transferred to 5 μM and 350 μM Fe for 2 d. The number of seedlings with and without observable stem cell death is shown on the red and gray bars, respectively. Different letters indicate significant differences between groups by pairwise Fisher’s exact test (P < 0.05). (C) Schematic model showing the ANAC044-mediated GSNOR-regulated mitochondrial signaling to trigger Fe-induced stem cell death in the root meristem.

ANAC044 Is Epistatic to ANAC013 and ANAC017 in GSNOR-Regulated Fe-Induced Stem Cell Death.

We next set out to test the function of ANAC044 for the Fe-induced DNA damage–dependent stem cell death in hot5-2 root meristems. Surprisingly, both of the ANAC044 knockout lines, anac44-1 and anac44-2, fully abolished stem cell death induced by high Fe treatment in the hot5-2 mutant root meristems (Fig. 5B and SI Appendix, Fig. S20 A and B). Moreover, the ANAC044 knockout also alleviated stem cell death induced by different mitochondrial ETC inhibitors in root meristems (Fig. 3F and SI Appendix, Figs. S13 A and B and S21). However, the ANAC044 knockout did not abolish DNA damage induced by high Fe in the root meristem of hot5-2 mutants (SI Appendix, Fig. S20 C and D). Collectively, these results indicated that ANAC044 is essential for mitochondrial dysfunction-triggered stem cell death in root meristems but acts downstream of DNA damage.

To examine whether ectopic expression of ANAC044 leads to stem cell death in the root meristem, we introduced the pUBQ10:ANAC044, the pANAC044:GFP-ANAC044, as well as the empty vector into the Col-0 background. None of the 54 independent T1 lines harboring the empty vector showed stem cell death in the root meristem, while 8 of 84 pUBQ10:ANAC044 T1 lines and 15 of 136 pANAC044:GFP-ANAC044 T1 lines exhibited severe stem cell death in the root meristem (SI Appendix, Fig. S22). Fisher’s exact test showed a significant difference in stem cell death in the root meristem between the empty vector and pUBQ10:ANAC044 or pANAC044:GFP-ANAC044 (SI Appendix, Fig. S22B). Moreover, we observed that the nuclear GFP fluorescence was weakly expressed in the pANAC044:GFP-ANAC044 T1 lines with stem cell death, while it was almost not detected in the root meristems of these T1 lines without stem cell death (SI Appendix, Fig. S22 D–H). This result indicates that the induction of ANAC044 protein expression tends to result in stem cell death in the root meristem, despite variations in ANAC044 protein sensitivity among individual seedlings. Taken together, these results revealed that the ectopic expression of ANAC044 driven by either the UBQ10 promoter or the native promoter in Col-0 background could be sufficient to trigger stem cell death in the root meristem.

To test the genetic relationship between ANAC044 and the ANAC013-ANAC017 module in GSNOR-regulated and Fe-induced stem cell death in root meristems, we generated hot5-2anac17-1anac44-1 and hot5-2anac13-1anac44-1 triple mutants and tested their responses to high Fe treatment. The ratio of stem cell death in root meristems of hot5-2anac17-1anac44-1 and hot5-2anac13-1anac44-1 triple mutants was similar to that of hot5-2anac44-1 double mutants (Fig. 5B and SI Appendix, Fig. S23), which was much lower than that of hot5-2anac17-1 and hot5-2anac13-1 double mutants (Fig. 3I). These results indicate that ANAC044 is epistatic to ANAC013 and ANAC017 in the context of GSNOR-regulated Fe-induced stem cell death in root meristems.

Discussion

Cellular Fe is emerging as a significant trigger for inducing cell death or damage in both animals and plants. Previously, we discovered that root stem cells, critical for root growth and development, are particularly susceptible to Fe toxicity when GSNOR is disrupted in Arabidopsis (27). While mitochondria play a crucial role in regulating cell death in animals, their impact in plant cell death has remained unclear. In this study, we demonstrate that Fe toxicity can compromise mitochondrial function, leading to a mitochondrial stress response in the root of the GSNOR defective mutant. The dysfunctional mitochondria release a retrograde signal through the mitochondrial outer membrane channel VDAC, activating the expression of the nuclear transcriptional factor ANAC044, thereby triggering DNA damage–dependent stem cell death in root meristems (Fig. 5C). This Fe-induced stem cell death in the GSNOR defective meristems exhibits important differences compared to ferroptotic cell death as we did not find evidence for Fe-dependent lipid peroxidation, a hallmark of ferroptotic cell death. Overall, the nuclear transcription factor ANAC044-mediated mitochondria dysfunction signaling pathway is responsible for initiating DNA damage– dependent stem cell death in roots.

Our study revealed a function of GSNOR in maintaining mitochondrial function. As a major regulator in protein S-nitrosylation that controls multiple physiological processes through posttranslational modification in plants, GSNOR has been extensively investigated at the genome-wide level for its regulatory roles by different proteomics approaches (39–41). However, mitochondria-related biological processes have not been found in these studies. Our findings suggest that in addition to the proteomics research, the time-course transcriptomic approach can provide clues from another dimension to extend our current knowledge of the role of GSNOR. Therefore, this study also provides an important transcriptomic data resource for the research communities for both GSNOR function and Fe biology studies. Nevertheless, the mechanism by which GSNOR protects mitochondrial function from the damage of Fe toxicity remains elusive in this study.

It is interesting that the pattern of ROS accumulation was greatly altered in the GSNOR defective root meristem in response to high Fe. Compared to the control condition, ROS accumulation was slightly increased in the gsnor root meristem upon high Fe treatment at early stages, but it was seriously impaired in gsnor root meristem after high Fe treatment for 48 h (SI Appendix, Fig. S4 A, B, F, and G). The reduction of ROS accumulation might be a consequence of seriously impaired mitochondrial function because it also occurred in the root meristem when subjected to mitochondrial inhibitor treatments for longer than 24 h (SI Appendix, Fig. S24). Considering that stem cell death occurred in the gsnor root meristem until 48 h after high Fe treatment (Fig. 1A), these results do not support the notion that ROS originating from mitochondria act as toxic substances to trigger DNA damage–dependent stem cell death in the stressed root meristem. Instead, the short-term ROS burst might further impair mitochondrial function and initiate mitochondrial stress signaling or directly act as a mitochondrial signaling molecule to trigger the downstream response in the root meristem upon high Fe treatment. In animals, mitochondrial ROS can be released into the cytosol and nucleus, where they serve as crucial signaling molecules, triggering DNA damage and cell death (42, 43). The second possibility is partially supported by our finding that blocking mitochondrial VDAC with DIDS fully abolishes high Fe-induced cell death in the gsnor root meristem (Fig. 3G and SI Appendix, Fig. S13 E and F), considering the reported role of VDAC in releasing ROS from mitochondria to the cytosol to transmit the retrograde signal in mammalian cells (44). However, given that VDAC functions as a multisubstrate transporter, it cannot be excluded that other molecules are transported and play a role. For instance, VDAC is also required to transport cytochrome c from mitochondria into the cytosol (45). Cytochrome c is able to activate the protease enzyme caspases, which sets off a cascade of events that ultimately result in DNA fragmentation and cell death (46). Interestingly, it has been reported that cytochrome c requires nitrosylation posttranslation modification (47). Hence, it is plausible that GSNOR-mediated nitric oxide signaling could modulate the release of cytochrome c and DNA damage in cases of mitochondrial dysfunction. Therefore, it is worth clarifying which molecule derived from mitochondria is transported into the cytosol to trigger a cascade of events that ultimately results in DNA damage–dependent stem cell death in the root meristem in the future.

Given the contrast between the specific induction of stem cell death and nonspecific induction of ANAC044 expression at both mRNA and protein levels in the root meristem upon high Fe treatment, nuclear ANAC044 should be considered as an intermediary rather than the ultimate executor of the mitochondrial retrograde signal, thereby specifically triggering DNA damage–dependent cell death in the root meristem. Recently, two Aux/IAA family members, IAA5 and IAA29, were identified as target genes of SOG1. Both genes were specifically induced in vascular stem cells and their daughters in root meristems, resulting in stem cell death in Arabidopsis roots in response to DNA damage (48). Future studies are necessary to investigate direct target genes of ANAC044 and whether IAA5 and IAA29 are targets of ANAC044.

Materials and Methods

Details of plant materials and chemicals used in this study are provided in SI Appendix, Materials and Methods. Vector construction, transgenic line and CRISPR/CAS9 line establishment, cell death staining, GUS staining, TUNNEL assay, ROS measurement, MMP measurement, microscopy observation, image quantification, RNA isolation, quantitative real-time PCR (qRT-PCR), transcriptome library preparation, and sequencing were performed according to the protocols described in SI Appendix, Materials and Methods. RNA-seq read mapping and raw read counting were conducted using STAR (v 2.6.1) and HTSeq (v 0.6.0), respectively; further downstream analysis was all achieved in R (v 4.3.3) as described in SI Appendix, Materials and Methods. Statistical analysis of the rest of the experimental results is conducted in R (v 4.3.3), Data Processing System (v 20.00), or Microsoft Excel 2019 as described in SI Appendix, Materials and Methods.

Supplementary Material

Appendix 01 (PDF)

Dataset S01 (CSV)

Data, Materials, and Software Availability

Raw RNA-seq datasets data have been deposited in BioProject database (PRJNA1042792) (49). All study data are included in the article and/or supporting information.