Ethylene signalling enhances mitochondrial stress tolerance independently of the transcription factor ANAC017 in Arabidopsis

Abstract

Mitochondria are central to plant metabolism, yet the diversity of mechanisms plants use to cope with mitochondrial stress and its implications in cellular signalling are not fully understood. In this study, we analysed Arabidopsis noxy (nonresponding to oxylipins) mutants affected in 9-HOT (9(S)-hydroxy-10,12,15-octadecatrienoic acid) signalling, mitochondrial function, and ethylene (ET) signalling to dissect plant responses to a range of mitochondrial stresses, including inhibitors of all electron transport chain complexes and mitochondrial translation. All noxy mutants showed resistance to antimycin A (AA), implicating Complex III and 9-HOT signalling in mitochondrial stress adaptation. Notably, noxy22/eto1-14, an ET overproducer mutant, displayed resistance to all tested inhibitors independently of the canonical mitochondrial retrograde pathway mediated by ANAC017. We found similar results in eto1-5 and eto1-13 alleles, thus sustaining a role for ET in mitochondrial protection. Histochemical and RNA-sequencing analysis revealed that AA induced ANAC017-regulated genes early and independently of ET signalling, whereas EIN2 contributed to later induction of AA-associated immune responses. EIN2 was required for full activation of AA-induced resistance against the biotrophic pathogen Hyaloperonospora arabidopsidis, but not against the necrotroph Plectosphaerella cucumerina. Collectively, our findings point to a complex network that co-ordinates distinct but overlapping responses to mitochondrial dysfunction and integrates them into broader stress pathways.

Ethylene signalling promotes mitochondrial stress tolerance independently of ANAC017. In response to antimycin A, it modulates Arabidopsis immunity and co-operates with 9(S)-hydroxy-10,12,15-octadecatrienoic acid (9-HOT) oxylipin and ANAC017 pathways to restore mitochondrial homeostasis.

Introduction

Mitochondria are essential organelles that generate ATP through oxidative phosphorylation and integrate metabolic and signalling functions critical to plant growth and stress responses. Because the mitochondrial proteome is encoded in two separated genomes, plants rely on tightly co-ordinated anterograde (nucleus-to-organelle) and retrograde (organelle-to-nucleus) signalling pathways to maintain mitochondrial function under changing physiological and environmental conditions (Møller et al., 2021; Selinski et al., 2024). Mitochondrial mutants and chemical inhibitors that disrupt mitochondrial electron transport or protein synthesis—such as antimycin A (AA), oligomycin, or doxycycline—have been widely used to examine mitochondrial dysfunction and trigger retrograde responses (Rhoads et al., 2006; Ho et al., 2008; Van Aken and Whelan, 2012; Wang and Auwerx, 2017; Kacprzak et al., 2020). In particular, AA inhibits Complex III of the electron transport chain, leading to reactive oxygen species (ROS) production and transcriptional reprogramming of nuclear genes. These model systems have helped to identify signalling components that reprogram gene expression, promote mitochondrial repair, and enhance cellular stress tolerance.

A key mitochondrial retrograde signalling mechanism involves the NAC transcription factor ANAC017, which is cleaved from the ER membrane upon mitochondrial perturbation and translocated to the nucleus to regulate target genes (Ng et al., 2013). ANAC017 controls the expression of a broad network of genes involved in alternative respiratory pathways, detoxification, mitochondrial protein import, and redox balance (De Clercq et al., 2013; Van Aken et al., 2016; Meng et al., 2019). Comparative transcriptomic studies using inhibitors of different mitochondrial activities have established a core set of ‘mitochondrial dysfunction stimulon’ (MDS) genes that serve as markers of retrograde signalling activation (De Clercq et al., 2013). While ANAC017 is a major regulator, evidence also points to a complex regulatory network that includes additional transcription factors and interacts with other organelle signalling pathways (De Clercq et al., 2013; Ng et al., 2013; Wang et al., 2016).

Despite extensive characterization of ANAC017-mediated transcriptional regulation, the existence of ANAC017-independent regulation mechanisms, as well as the identity of the primary retrograde signals, are still a matter of debate. It is widely assumed that ROS, such as hydrogen peroxide and superoxide, are implicated in this early signalling and can originate from mitochondrial electron leakage during stress (Møller, 2001; Dourmap et al., 2020; Khan et al., 2024). Other mitochondrial-derived signals include changes in ATP levels, pH, calcium flux, nitric oxide, oxidized peptides, and membrane potential (Møller and Sweetlove, 2010; Crawford et al., 2018; Wagner et al., 2018; Welchen et al., 2021; Gupta et al., 2022). Hormones such as ethylene (ET) and auxins also modulate mitochondrial stress responses (Wang and Auwerx, 2017; Kacprzak et al., 2020; He et al., 2022). For instance, doxycycline-induced mitochondrial stress is accompanied by a ROS burst that precedes ethylene biosynthesis, suggesting that ROS act upstream of hormone-mediated signalling (Wang and Auwerx, 2017). On the other hand, 9-lipoxygenase (9-LOX)-derived oxylipins, including 9-HOT and its ketone derivative 9-KOT, have been shown to induce mitochondrial stress and activate protective responses, whereas mutants defective in 9-HOT signalling (noxy mutants) bear constitutive mitochondrial alterations that lead to AA resistance (Vellosillo et al., 2013; Izquierdo et al., 2021).

Continuing with our previous identification of noxy (nonresponding to oxylipins) Arabidopsis mutants, which exhibit altered sensitivity to 9-HOT-induced mitochondrial stress (Izquierdo et al., 2021), we investigated their response to a broader range of compounds inhibiting distinct mitochondrial activities. Our results highlight ethylene as a modulator of stress perception downstream of mitochondrial dysfunction and point to an integrated signalling network involving ANAC017, 9-HOT, and ET in shaping plant stress and immune responses.

Materials and methods

Plants and growth conditions

Arabidopsis thaliana wild-type (wt) and mutants noxy3/lon1-3, noxy23, noxy54/ahg2-3, noxy22/eto1-14 (9-HOT insensitive), eto1-5 and eto1-13 (SALK_113656) (ethylene overproducers), ein2-5 (ethylene signalling deficient), and anaca017-1 (AA response deficient, SALK_022174) are derived from the Col-0 ecotype, whereas NahG plants expressing an salicylate hydroxylase that degrades salicylic acid are derived from the Wassilewskija Ws ecotype. For in vitro analyses, sterilized seeds were vernalized (3 days, 4 °C) and allowed to grow on plates containing Murashige & Skoog (MS) medium pH 6.0 with 1.5% (w/v) sucrose and 1.5 or 0.7% (w/v) agar (Bacto Agar; Becton-Dickinson) for vertical or horizontal plates, respectively; growth conditions were 14 h light, 10 h dark, 22 °C, and 250 μmol s⁻¹m⁻² fluorescent illumination. For in planta analyses, seeds were sown on soil, vernalized (3 days, 4 °C), and grown in chambers (22 °C, 70% relative humidity, with 14 h light, 10 h dark photoperiod at 250 μmol s⁻¹m⁻² fluorescent illumination.

Phenotypic analyses after chemical treatment

For root growth phenotypic analysis, seeds were germinated on vertical MS plates and transferred 4 days later to plates supplemented with the specific compounds at the indicated concentrations. In pretreatment assays, plants were germinated on plates containing 9-HOT and transferred as described. The root length was measured 3 days after transfer using the software Fiji (https://fiji.sc/). Significant differences were detected with one-way ANOVA and the Tukey post-hoc test. To measure the aerial part of wt and ein2-5 seedlings, seeds were germinated on horizontal plates containing the specific products and phenotype was examined 7 days after germination; plates were photographed and inter-cotyledon distance measures taken with Fiji. Unless otherwise stated, root phenotypes were analysed in ∼15 seedlings and aerial phenotypes in ∼40 seedlings. Ethylene response was assessed by evaluating the triple response assay (root length, hypocotyl length and presence of apical hook) in seedlings germinated 4 days in the dark with or without 1 µM ACC. For analysis of in planta AA-induced resistance, 50 µM AA with 0.02% Tween-20 was sprayed onto 2.5-week-old plants 3 days before pathogen inoculation. At least three independent experiments were done in each case. Mitochondrial inhibitors were purchased from Sigma Aldrich with at least 95% purity: rotenone (R8875), sodium malonate (63409), antimycin A (A8674), KCN (31252), oligomycin A (75351), doxycycline (D9891). Chemicals were prepared according to the manufacturer’s specifications. The 9-HOT used in this study was produced as described in Prost et al. (2005).

RNA isolation and qPCR analysis

For in vitro analyses, seeds were germinated on horizontal plates and covered at 12 days post-germination with liquid MS containing AA or EtOH as a control. AA was added to a final concentration of 20 µM assuming a homogeneous distribution on the plate. For each genotype/condition, total RNA was isolated from three independently grown plates according to Logemann et al. (1987) and contaminant DNA was removed with the TURBO DNA-free kit (Thermo Fisher). For qPCR, 1 μg RNA per sample was reverse-transcribed with M-MuLV Reverse Transcriptase (New England Biolabs) and amplification was done with HOT FIREPol® EvaGreen® qPCR Mix Plus (Solis Biodyne) in a 7500 Real-Time PCR system (Applied Biosystems). Accession numbers for the genes analysed are AOX1a (AT3G22370), ANAC013 (AT1G32870), and NDB3 (AT4G21490). Expression was normalized combining 18S rRNA and SAND gene expression (AT2G28390), described in Czechowski et al. (2005), as references. Primers for these analyses and lengths of amplification products are described in Supplementary Table S1.

Protein extraction and western blot

For protein extraction, seedlings were grown in vertical planes and treated with AA (20 μM) alone or in combination with the ET pathway inhibitors AVG (2-aminoethoxyvinyl glycine, 1 μM), pyrazinamide (20 μM), or AgNO₃ (1 μM). Seedlings were collected at the indicated times, frozen in liquid nitrogen and stored at −80 °C. Protein extracts were prepared by grinding seedlings to a fine powder in liquid nitrogen and extracted in 50 mM Tris–HCl pH 7.5, 150 mM NaCl, 0.1% (w/v) NP-40, 1 mM phenylmethylsulfonyl fluoride (PMSF), and 1× complete protease inhibitor cocktail (Roche). Supernatants were clarified by centrifugation (3000 g, 10 min, 4 °C). Twenty-five microgrammes of protein per sample were separated by SDS-PAGE in 12.5% acrylamide gels, transferred to nitrocellulose membranes (GE Health Care) by electroblotting, stained with red Ponceau S (0.1% w/v) to control protein loading, washed, blocked with 5% defatted powder milk, and incubated with antibody specific to plant AOX1/2 (Agrisera) diluted 1:1000 in Tris-buffered saline (TBS) buffer with 1% (w/v) defatted powder milk. Blots were then incubated with a peroxidase-labelled anti-rabbit antibody (Agrisera) and developed with ECL Select^TM (Cytiva). Blots were imaged in a ChemiDoc^TM imaging system (BioRad).

Histochemical analysis

For histological analysis, seeds were germinated on MS plates and transferred 4 days post-germination to plates supplemented with rotenone (50 µM), malonate (3.5 mM), AA (10 µM), KCN (300 µM), oligomycin (5 µM), doxycycline (5 µM), or ACC (1-aminocyclopropane-1-carboxylic-acid, 5 µM) and examined 3 days later. For callose detection, seedlings were stained with 0.1 mg ml⁻¹ aniline blue (SiroBiosupplies) for 30 min in the dark. Samples were washed, mounted in 50% (v/v) glycerol on glass microscope slides and examined with a Leica DMR fluorescence microscope as carried out in Vellosillo et al. (2007). For lignin detection, seedlings were stained with 2% (w/v) phloroglucinol-HCl (Sigma Aldrich) as described by Schrick et al. (2004). Stained seedlings were examined and photographed under a Leica M165 FC stereomicroscope. For each treatment, over 15 plants were examined in three independent experiments.

RNA-sequencing and bioinformatic analysis

Total RNA was extracted as described above from three independently grown plates for each genotype and treatment. RNA was quantified using a Nanodrop ND-1000 UV-Vis spectrophotometer (Nanodrop Technology Inc.) and quality was assessed using an Agilent 2100 bioanalyzer (Agilent Technologies). RNA samples (250 ng) were used for library preparation and sequencing at Novogene Co., Ltd. (Beijing, China). Briefly, mRNA was isolated with poly-T oligo-attached magnetic beads. After fragmentation, the first strand cDNA was synthesized using random hexamer primers followed by second strand cDNA synthesis. After end repair, A-tailing and adapter ligation, size selection, amplification, and purification, the libraries were checked with a Qubit fluorometer (Thermo Fisher) and quantified byreal-time PCR. Band size distribution was evaluated with Bioanalyzer (Agilent). Quantified libraries were pooled and paired-end sequenced on the Illumina platform NovaSeq X Plus (PE150). Raw reads were filtered to remove adapters and low-quality reads (i.e. reads with over 10% non-called bases and over 50% bases with Qscore <5). In all samples, over 95% of reads had an error rate of less than 0.001. Filtered reads were mapped to the Arabidopsis thaliana genome (TAIR10 release) using hisat2 (Kim et al., 2019). Alignment files in SAM format were converted to BAM format, sorted by genomic co-ordinates and indexed with SAMtools (Li et al., 2009). Alignment results were visually inspected with the IGV browser (Robinson et al., 2011). Transcripts were quantified with featureCounts (Liao et al., 2014) (parameters: -s 0 -p –countReadPairs -F SAF -g gene -a Arabidopsis_thaliana.TAIR10.57.saf). For each comparison, differentially expressed genes were detected with the DESeq function of the DESeq2 Bioconductor package (Love et al., 2014) using default options. Gene names and descriptions were downloaded using BioMart (Kinsella et al., 2011) and attached as new informative columns to DESeq2 results. Plotting and filtering of differential expression results were performed with the FIESTA viewer (Oliveros, 2022). For each genotype and time, differential expression was assessed between AA-treated samples and MS controls. Gene Ontology (GO) enrichment analysis of genes was carried out with PANTHER (Thomas et al., 2022). Co-expression networks with selected genes were built with ‘WGCNA’ (Langfelder and Horvath, 2008) and built-in packages in R. Expression data was clustered with the ‘hclust’ function of ‘cluster’ package (Maechler, 2022), using the average silhouette width to determine the optimum number of clusters. The resulting network was visualized with Cytoscape (Shannon et al., 2003).

In vivo analyses of pathogen growth in antimycin A-pretreated plants

For examination of pathogen growth, inoculations were performed in plants 3 days after pretreatment with 50 µM AA, or the same volume of ethanol used as control, in 0.02% (v/v) Tween-20. Hyaloperonospora arabidopsidis (Hpa) was maintained in a running culture on Ws NahG seedlings, replicated weekly. For long-term storage, infected plants with sporulating Hpa were kept at −80 °C. Spore suspensions (10⁵ conidiospores ml⁻¹) from freshly infected Ws NahG plants were spray-inoculated onto 2.5-week-old plants as described in Sánchez et al. (2016). Plants were collected in 96% ethanol 5 days after inoculation, and stained with trypan blue to visualize fungal structures. Over 100 Hpa-infected plants per treatment, distributed in 10 pots, were stained and visualized with a stereo microscope and classified into one of the four different Hpa colonization classes (I: no visible colonization; II: oomycete growth without sporulation; III: presence of sporangiophores and asexual sporulation; IV: sexual sporulation). Differences in infection severity among treatments were assessed by multiple chi-square tests with false discovery rate (FDR) correction. Additionally, a colonization index was calculated for each pot of plants, as the weighted mean of infection classes, by the formula (n1+2n2+3n3+4n4)/(n1+n2+n3+n4), where n1, n2, n3, and n4 are the number of leaves in each colonization class. Significance of differences between treatments were assessed by one-way ANOVA and the Tukey post-hoc test. Plectosphaerella cucumerina strain BMM (Pc) was cultivated on half-strength Potato Dextrose Agar (BD Difco, BD-213400) in the dark. Spores from agar plates were collected by scraping, resuspended in water and adjusted to 10⁶ spores ml⁻¹. Over 15 plants per treatment were inoculated by pipetting 6 μl droplets onto each of three fully expanded leaves per plant. Infection was evaluated 12 days after inoculation by measuring the lesion area in each infected leaf with the software Fiji. Significant differences between treatments were assessed by one-way ANOVA and the Tukey post-hoc test.

Results

9-HOT, noxy mutations, and ethylene activate protection against mitochondrial stress

Previous studies demonstrated that treatment with 9-LOX derived oxylipins such as 9-HOT or 9-KOT induces a mild mitochondrial stress that protects Arabidopsis seedlings against AA, while their simultaneous application enhances AA-induced damage (Izquierdo et al., 2021). Further analyses using noxy (nonresponding to oxylipins) mutants revealed that alterations in mitochondrial proteins can trigger mitochondrial retrograde signalling enhancing tolerance to AA (Izquierdo et al., 2021).

To further explore the role of oxylipins in mitochondrial stress responses and signalling, we evaluated the effect of 9-HOT pretreatment on the response of wt plants to inhibitors of mitochondrial respiratory complexes, i.e. rotenone (complex I), malonate (complex II), AA (complex III), KCN (complex IV), oligomycin (complex V), and also to doxycycline, which inhibits mitochondrial protein synthesis. As observed with AA (Izquierdo et al., 2021), transference from MS to media supplemented with each of the inhibitors examined significantly inhibited root growth. However, when seedlings were germinated on medium containing 9-HOT and then transferred to inhibitor-containing media, plants were more tolerant to AA, but not to any of the remaining compounds (Fig. 1).

Fig. 1.

Root growth of 4-day-old plants grown in MS medium with or without 9-HOT (25 μM), and then transferred for 3 additional days to MS containing rotenone (50 µm), malonate (3.5 mM), antimycin A (10 µM), KCN (300 µM), oligomycin (5 µM), or doxycycline (5 µM). (A) Representative pictures; white marks indicate the root length at the moment of transfer. (B) Root length (mm) after transfer. (C) Relative root growth measurements. Data are means±SE from three independent experiments. Significance of differences was assessed by Student’s t-test (***P<0.001, n=15).

As noxy mutants are insensitive to 9-HOT and AA (Izquierdo et al., 2021), we analysed their root growth in the presence of the additional inhibitors tested here. We selected three noxy mutants affected in mitochondrial proteins—noxy3/lon1-3, noxy23, and noxy54/ahg2-3—previously characterized in Izquierdo et al. (2021). In addition, we included the ET overproducer mutant noxy22/eto1-14 (López et al., 2011) due to the previously reported association of ET with mitochondrial stress signalling (Wang and Auwerx, 2017). Our results showed that noxy mutants exhibited different degrees of resistance to the mitochondrial damage (Fig. 2). Firstly, all were resistant to both rotenone (50 µM) and AA (10 µM) when compared with wt plants. Additionally, (i) noxy3/lon1-3, noxy23, and noxy22/eto1-14 were resistant to KCN (300 µM) and doxycycline (5 µM), whereas noxy54/ahg2-4 was not; (ii) noxy3/lon1-3, noxy54/ahg2-4, and noxy22/eto1-14, but not noxy23 were resistant to oligomycin (5 µM); and (iii) in response to malonate (3.5 mM), noxy3/lon1-3, noxy23, and noxy54/ahg2-3 mutants showed root growth inhibition similar to wt plants, while noxy22/eto1-14 exhibited resistance.

Fig. 2.

Effect of mitochondrial inhibitors on root growth of noxy mutants. Four-day-old vertically grown seedlings were transferred to MS (control) or MS containing rotenone (50 µM), malonate (3.5 mM), antimycin A (AA) (10 µM), KCN (300 µM), oligomycin (5 µM), or doxycycline (5 µM). (A) Relative growth with respect to MS and (B) Root length 3 days after transference of wild-type (wt), noxy22 (noxy22/et1-14), noxy3 (noxy3/lon1-3), noxy23, and noxy54 (noxy54/ahg2-4) plants. Data are means±SE from three independent experiments. Different letters above the bars indicate significant differences (P<0.05) as detected with one-way ANOVA and the Tukey post-hoc test.

Notably, noxy22/eto1-14 plants showed a broad tolerance to mitochondrial inhibition, which prompted us to examine the root growth response of two additional eto1 mutant alleles under these conditions: eto1-5, harbouring a missense mutation in the first TPR domain (Wang et al., 2004), and eto1-13 (SALK_113656), a knockout T-DNA insertion line. As a result, both mutants displayed resistance levels comparable to noxy22/eto1-14, indicating that constitutive ET confers protection against mitochondrial stress (Supplementary Fig. S1). To confirm this conclusion, we analysed the response of ein2-5, a mutant with defective ET signal transduction leading to ET insensitivity (Alonso et al., 1999). Surprisingly, root growth inhibition in ein2-5 was not significantly different to that in wt plants (Supplementary Fig. S2). However, when shoot growth was evaluated, ein2-5 plants were more sensitive than wt plants to all inhibitors of mitochondrial activity, with the exception of doxycycline (Fig. 3). To ensure that the observed responses were due to ET signalling, phenotypes of ET overproduction (in eto1 mutants) and ET insensitivity (in ein2-5) were checked by the triple response assay and ACC treatment (Supplementary Fig. S3).

Fig. 3.

Effect of mitochondrial inhibitors on leaf growth of wild-type (wt) and ein2-5 plants. wt and ein2-5 seedlings were grown for 7 days in MS (control) or MS containing rotenone (25 µM), malonate (2.5 mM), antimycin A (AA) (2.5 µM), KCN (300 µM), oligomycin (1 µM), or doxycycline (1 µM). (A) Representative pictures of seedlings. (B) Inter-cotyledon distance. Data are means±SE from three independent experiments. Significance of differences was assessed by Student’s t-test (***P<0.001, **P<0.001, n=40).

Antimycin A-induced mitochondrial damage enhances callose and lignin production

We had shown that 9-HOT induces mild mitochondrial stress and callose deposition (Vellosillo et al., 2007, 2013), and that mitochondrial noxy mutants were affected in lignin accumulation in response to cell wall damage (Vellosillo et al., 2013; Marcos et al., 2015), altogether suggesting a connection between mitochondrial dysfunction and cell wall reinforcement. Based on these findings, we investigated whether the mitochondrial damaging products used in this study could also induce callose or lignin accumulation.

In wt plants, AA treatment led to callose deposition in root tips and vascular tissue; however, no substantial callose accumulation was observed in response to the remaining inhibitors (Fig. 4A). Furthermore, to investigate whether AA-induced cell wall modification is influenced by ET or ANAC017-dependent signalling, we analysed callose accumulation in eto1-13, ein2-5, and the knockout mutant anaca017-1 (Ng et al., 2013). AA induced a similar increase in callose deposition in root tips across all genotypes tested, indicating that this response is independent or occurs upstream of both ET and ANAC017 signalling (Fig. 4B). Phloroglucinol staining showed increased lignin accumulation in the aerial parts of AA-treated seedlings, but not in those treated with rotenone, malonate, KCN, oligomycin, or doxycycline (Fig. 5A). Moreover, lignin accumulation in response to AA was evident in wt, eto1-13, and anaca017-1 seedlings, but not in ein2-5 mutants (Fig. 5B). These results suggest that AA-induced lignin production depends on ET signalling but is independent of ANAC017 activity.

Fig. 4.

Callose formation in response to mitochondrial inhibitors. (A) Fluorescence visualization of callose by aniline blue staining in wild-type (wt) roots 3 days after passing to MS (control) or MS containing rotenone (50 µM), malonate (3.5 mM), antimycin A (AA) (10 µM), KCN (300 µM), oligomycin (5 µM), doxycycline (5 µM), or ACC (2 µM). (B) Callose formation in the roots tips of wt, ein2-5, eto1-13 or anac017-1 seedlings 3 days after passing to MS containing AA (10 µM). In each case, three independent experiments were performed with similar results.

Fig. 5.

Lignin formation in response to mitochondrial inhibitors. (A) Transmitted light visualization of lignin by phloroglucinol staining of wild-type (wt) seedlings, 3 days after passing to MS (control) or MS containing rotenone (50 µM), malonate (3.5 mM), antimycin A (AA) (10 µM), KCN (300 µM), oligomycin (5 µM), doxycycline (5 µM), or ACC (2 µM). (B) Lignin formation in wt, ein2-5, eto1-13 or anac017-1 seedlings 3 days after passing to MS (control) or MS containing AA (10 µM). In each case, three independent experiments were performed with similar results.

Induction of canonical antimycin A-responsive genes is not affected by ET signalling

The mitochondrial retrograde signalling pathway triggered by AA has been extensively characterized, with the transcription factor ANAC017 acting as a major regulator of gene expression to overcome mitochondrial stress. To assess the influence of ET signalling on AA-induced retrograde responses we examined the expression of well-established AA marker genes in ET pathway mutants (noxy22/eto1-14, eto1-5, eto1-13, and ein2-5) alongside the anac017-1 knockout line as a control.

We analysed transcript levels of AOX1A (Alternative Oxidase1A), ANAC013 (a NAC domain transcription factor), and NDB3 (NAD(P)H dehydrogenase B3) as well as AOX protein amount, in plants treated with MS or AA. No significant differences in basal transcripts levels were observed among wt, eto1, or ein2-5 plants under control conditions (Fig. 6A). Three hours after AA treatment, we detected a similar increase of AOX1A, NDB3, and ANAC013 transcript levels in wt, eto1, and ein2-5 plants. As expected, this transcriptional activation was significantly reduced in anaca017-1. At the protein level, untreated wt, eto1, ein2-5 and anac017-1 plants contained similar amounts of AOX. Upon AA treatment, AOX protein levels increased similarly in wt, eto1, and ein2-5 plants at both 8 and 24 h. In contrast, anac017-1 mutants displayed only a modest increase in the amount of AOX protein (Fig. 6B).

Fig. 6.

Induction of the mitochondrial retrograde response in wild-type (wt), noxy22/et1-14 (noxy22), eto1-5, eto1-13, ein2-5, and anac017-1 plants exposed to antimycin A (AA). (A) Seedlings were grown in horizontal plates and covered 12 days post-germination with liquid MS medium (control) or liquid MS containing AA (20 µM final concentration). AOX1A, ANAC013, and NDB3 expression was determined by quantitative real timePCR in RNA samples extracted 3 h after treatment. Data are means±SE from three independent experiments and relative to MS-treated wt plants. Significance of differences within each genotype was assessed by Student’s t-test (***P<0.001, **P<0.01, *P<0.05). (B) AOX protein levels in MS or AA-treated seedlings at 8 h or 24 h after treatment. (C) AOX protein levels in wt seedlings 3 days after transfer to MS medium or MS containing AA (20 μM) alone or in combination with ethylene inhibitors AVG (1 μM), Pyr (Pyrazinamide 20 μM) or Ag⁺ (AgNO₃ 1 μM). Immunoblots were repeated three times with similar results using an anti-AOX antibody. Ponceau Red staining (0.1%) was used as control for protein loading.

Additionally, chemical inhibition of ET biosynthesis using AVG (aminoethoxyvinylglycine, an ACC synthase inhibitor) or PZA (pyrazinamide, an ACC oxidase inhibitor), or ET perception with AgNO₃ (Binder and Schaller, 2017; Sun et al., 2017), did not affect AOX levels in AA-treated wt plants (Fig. 6C). Together, these results indicate that the canonical mitochondrial retrograde response to AA functions independently of ET signalling and that the AA resistance observed in ET overproducing mutants is likely not mediated via ANAC017.

Transcriptional changes in wild-type and ein2-5 plants treated with antimycin A

To investigate the contribution of ET signalling in the plant response to AA, we performed RNA sequencing (RNA-seq) on wt and ein2-5 seedlings treated with AA for 3 or 24 h, alongside untreated and MS-treated controls (Supplementary Table S2).

Gene expression responses to AA differed between wt and ein2-5 plants. At 3 h post-treatment, 536 genes were up-regulated in wt seedlings (log₂ fold change ≥1, FDR <0.05) relative to MS controls. Of these, approximately 70% were also induced in ein2-5, indicating that their regulation is independent of the ET pathway. The remaining 30% were not induced in ein2-5 plants, defining a subset of ET-dependent AA-responsive genes (Fig. 7). At 24 h, ET signalling seemed to have a higher weight on the transcriptional response to AA; of the 339 genes up-regulated in wt plants, only 159 (47%) were also induced in ein2-5 (Fig. 8).

Fig. 7.

ET influence on GO term enrichment of genes induced by antimycin A (AA) at 3 h. The tool DeepVenn ( Hulsen, 2022, Preprint) was used to draw an area-proportional Venn diagram of genes induced by AA at 3 h in wild-type (wt) and ein2-5 plants. The resulting gene sets, wt-exclusive (left), common (centre), and ein2-5-exclusive (right), were analysed for GO term enrichment of biological processes with PANTHER (Thomas et al., 2022) and visualized as bubble plots with the tool SRPlot (Tang et al., 2023). Bubble position, size, and colour represent fold enrichment, number of genes, and statistical significance, respectively. GO categories were grouped for clarity.

Fig. 8.

ET influence on GO term enrichment of genes induced by antimycin A (AA) at 24 h. The tool DeepVenn (Hulsen, 2022, Preprint) was used to draw an area-proportional Venn diagram of genes induced by AA at 24 h in wild-type (wt) and ein2-5 plants. The resulting gene sets, wt-exclusive (left), common (centre) and ein2-5-exclusive (right), were analysed for GO term enrichment of biological processes with PANTHER (Thomas et al., 2022) and visualized as bubble plots with the tool SRPlot (Tang et al., 2023). Bubble position, size and colour represent fold enrichment, number of genes and statistical significance, respectively. GO categories were grouped for clarity.

GO term enrichment analysis revealed distinct functional profiles for ET-independent and ET-dependent AA-responsive genes. Among the ET-independent genes, we found significant enrichment for GO terms related to mitochondrial processes (Fig. 9). Notably, 22 and 20 of the genes that were up-regulated at 3 and 24 h, respectively, belong to the MDS group, a set of 24 genes previously defined as core components of mitochondrial and oxidative stress responses regulated by ANAC017 (De Clercq et al., 2013). These MDS genes displayed similar temporal expression patterns in wt and ein2-5 plants, with maximal induction occurring at 3 h after AA exposure.

Fig. 9.

Expression profiles of MDS genes (A) and lignin synthesis genes (B) in response to antimycin A. In each case, relative expression values are represented as line plots and heatmaps for wt and ein2-5 plants. In the line plots, each grey line corresponds to a gene, and the average expression is represented as a red line.

Further GO analysis of ET-independent AA up-regulated genes, also revealed enrichment in terms related to abiotic and biotic stress, ROS, and hypoxia responses at both time points. Additionally, terms associated with root development and hormone regulation were enriched at 3 h, while genes involved in iron homeostasis were over-represented at 24 h.

The set of genes up-regulated by AA in an ET-dependent manner was enriched in terms associated with plant defence responses and hypoxia at both 3 and 24 h. At 3 h, additional enrichment in terms related to senescence and abscisic acid (ABA) signalling was observed. Interestingly, the group of AA-up-regulated genes that were also induced in ein2-5 (i.e. negatively regulated by ET) was similarly enriched in defence-related terms.

Given our earlier observation that AA-induced lignin accumulation depends on ET signalling, we searched for AA-regulated genes involved in lignin formation. We identified AA-inducible genes encoding key enzymes of the monolignol biosynthesis pathway, such as Caffeoyl-CoA 3-O Methyl Transferase and Hydroxycinnamate Glucosyl Transferase as well as multiple Berberine Bridge Enzyme-like genes implicated in monolignol metabolism and lignin formation (Liu et al., 2011; Daniel et al., 2015). Many of these genes showed reduced expression in ein2-5 mutants compared to wt plants, supporting the role of ET signalling in mediating AA-induced lignin biosynthesis.

Regarding down-regulated genes, 308 and 350 transcripts were repressed by AA in wt plants at 3 and 24 h, respectively. Of these, 64% (3 h) and 42% (24 h) were simultaneously down-regulated in ein2-5, indicating their ET-independent repression. The remaining 36% (at 3 h) and 58% (at 24 h) were not significantly down-regulated in ein2-5, suggesting ET-dependent regulation (Supplementary Figs S4–S5). ET-independent down-regulated genes were enriched for GO terms related to root development, cell-wall dynamics, and hypoxia. In contrast, ET-dependent down-regulated genes were associated with developmental growth and cell–cell junction organization.

Co-expression network of antimycin A-induced defence genes

Previous studies demonstrated that AA can enhance plant resistance to pathogen infection (Sánchez et al., 2021). Consistent with this, our RNA-seq results showed increased expression of immune-related genes following AA treatment. To explore the contribution of AA to plant defence and assess the role of ET signalling in this response, we examined the expression patterns of genes assigned to the GO term ‘defense response’ (GO:0006952) and up-regulated by AA in at least one of the experimental conditions tested. Hierarchical clustering of gene expression profiles grouped these defence-related genes into four clusters based on their temporal response to AA (Fig. 10A). Cluster 1, the largest group, consisted of genes with an early AA induction profile. Cluster 2, included genes with delayed induction, showing peak expression at 24 h. Clusters 3 and 4 were smaller, each containing only three genes. Cluster 3 was characterized by genes with high basal expression in ein2-5 and early induction by AA in wt plants. Cluster 4 genes were primarily induced at 24 h in ein2-5, suggesting negative regulation by AA.

Fig. 10.

Expression of defence genes induced by antimycin A (AA) is modulated by ethylene signalling. (A) Clustering of defence genes induced by AA by their expression profiles. Expression profiles were grouped into four clusters by maximizing average cluster width, and then hierarchical clustering heatmaps were drawn. (B) Co-expression network of defence genes induced by AA. The network was built and then grouped into the four clusters shown in (A) for easier interpretation. Nodes (genes) were coloured by their ET-dependent fold change, estimated as the average induction by AA in wild-type (wt) versus ein2-5 plants. Transcription factors are highlighted in bold.

To further dissect the functional relationship and ET dependency among these genes, we constructed a co-expression network based on the AA-induced immune-related genes (Fig. 10B). We tested multiple network topologies and selected a scale-free network with a correlation threshold of 0.91, which resulted in two main subnetworks and several unconnected nodes. The network topology largely reflected the clustering, with the major subnetwork corresponding to Cluster 1 and the secondary subnetwork matching Cluster 2. Genes from Cluster 3 and 4 were weakly connected or unconnected.

To visualize ET dependency, we calculated the average difference in AA-induced expression between wt and ein2-5 plants for each gene and used this value to generate a heatmap overlay on the co-expression network. Cluster 1 genes, mostly early-induced, formed a highly interconnected module. This cluster was enriched in transmembrane receptor proteins with Ser/Thr kinase activity, including six lectin receptor kinases (LecRKs). Notably, LecRK-1.7 and LecRK-4.2 have been previously linked to pathogen resistance, and LecRK-6.3 is known to negatively regulate ABA signalling. Additionally, three cell wall-associated receptor kinases (WAKL4, WAKL5, and WAKL6) were identified, suggesting that AA treatment might enhance surface-level immune perception.

Cluster 1 also contained 12 transcription factors involved in diverse defence and stress-related processes, such as glucosinolate biosynthesis (MYB122, MYB51), salicylic acid signalling (WRKY51), camalexin biosynthesis (WRKY33), ABA regulation (ABR1), salt stress (WRKY8, AtC3H29), and pathogenesis-related responses (EFR018, ERF2, ERF011). The salicylic acid pathway regulator NPR2 was also present in this cluster. No consistent pattern of ET dependency was observed across Cluster 1; genes ranged from being positively to negatively regulated by EIN2.

To identify network motifs specifically regulated by ET within Cluster 1, we filtered for genes with an ET induction factor (wt—ein2-5) >0.5. This highlighted two interconnected motifs. The first contained MYB122, ABR1, the kinase ZRK4, and senescence-associated protein SAG13. The second included WRKY51, secreted peptides PIP2, PIP3, and IDA, cytochrome CYP81F2, and membrane proteins NHL10, HSR4, and CRK20. Filtering for an ET induction factor of <0.5 identified seven genes with low or no connectivity, including the negative regulator of JA responses CYP94B3, the ET response factor ERF2, and CAF1-9, a CCR4-associated factor 1 homologue.

Genes in Cluster 2, generally induced at 24 h post-AA treatment, showed a stronger dependence on ET signalling. This group was enriched in genes encoding oxygenating enzymes, including Lipoxygenase 1 (LOX1) and Dioxygenase 1 (DOX1), both involved in oxylipin production during pathogen attack. Additionally, three cytochrome P450 enzymes were identified: two (AT2G30750 and AT4G31970) associated with indole cyanohydrin biosynthesis, and one (AT4G19230) involved in ABA catabolism. Other notable enzymes included DLO1 (involved in salicylic acid inactivation) and FMO1, a key regulator of systemic acquired resistance.

Clusters 3 and 4 each contained only three genes. Cluster 3 transcripts were induced early by AA in wt plants, but had high basal expression in ein2-5 plants. These included LecRK-3.1 and the cysteine-rich receptor-like protein kinase CRK32. Cluster 4 genes were primarily induced at 24 h in ein2-5, suggesting negative regulation by ET. These included the defensin-like gene PDF2.1, an ATP-binding protein (AT1G57650), and a bifunctional inhibitor/lipid transfer protein (AT2G37870).

ET signalling modulates antimycin A-induced defence against obligate biotrophs

We show that AA treatment induces the early expression (3 h) of defence-related genes in wt and ein2-5 plants, but this activation is more sustained in wt plants at later stages of the response (24 h). To assess if this difference has implications in AA-induced plant immunity, we analysed the effect of AA pretreatment on wt and ein2-5 resistance against two different pathogens: the obligate biotrophic oomycete Hyaloperonospora arabidopsidis (Hpa) and the necrotrophic fungus Plectosphaerella cucumerina (Pcc).

Plants were sprayed with either AA or a control solution and inoculated with the pathogens 3 days later. In the case of Hpa infection, conidiophore development was visible 5 days after inoculation, predominantly in control-pretreated plants. Infection severity was classified on a scale from class I (little or no infection) to class IV (severe colonization), and the distribution of infection classes was represented as stacked bar plots for each treatment and genotype combination.

AA pretreatment markedly increased the proportion of class I leaves in both the wt and ein2-5 plants compared to the control pretreatment. However, the protective effect of AA seemed more pronounced in wt plants, where class I leaves increased from 25% in control plants to 75% after AA pretreatment compared to an increase from 25% to 45% in ein2-5 mutants. Multiple chi-square tests confirmed statistically significant differences in infection rates between control- and AA-treated plants, as well as between AA-treated wt and ein2-5 plants. These results demonstrate that: (i) AA induces resistance to Hpa, and (ii) the extent of AA-induced defence priming varied between the two genotypes.

To quantify this difference, we calculated a colonization index for each pot of plants by computing the weighted mean of infection classes. No significant differences were found between control-pretreated wt and ein2-5 plants, confirming similar basal susceptibility. AA pretreatment significantly reduced the colonization index in both genotypes, confirming AA-induced resistance to Hpa. However, the colonization index of AA-pretreated plants was significantly higher in ein2-5 compared to wt plants, supporting a partial dependency of AA-induced resistance on EIN2 signalling (Fig. 11A).

Fig. 11.

Induction of resistance to pathogens by antimycin A (AA) pretreatment in wild-type (wt) and ein2-5 plants. (A) Response to Hyaloperonospora arabidopsidis (Hpa) in AA- or control-pretreated plants; left panel: distribution of infected leaves in colonization classes assessed by trypan blue staining (I: no visible colonization; II: oomycete growth without sporulation; III: oomycete growth with sporangiophores and asexual sporulation, IV: sexual sporulation); right panel: colonization index. (B) Response to Plectosphaerella cucumerina (Pcc) in control or AA-pretreated plants: lesion area measured 12 days after inoculation. In both pathogens, pretreatments were carried out 3 days before inoculation. Data are means±SE from three independent experiments. Different letters represent significant differences (P<0.05) detected by FDR-corrected multiple chi-square tests (A, left) or one-way ANOVA followed by Tukey post-hoc test (A, right, and B).

Resistance to Pcc was evaluated by measuring lesion area 12 days after fungal inoculation (Fig. 11B). AA pretreatment significantly reduced lesion size in both genotypes, indicating enhanced resistance. Unlike with Hpa, no significant differences were observed between wt and ein2-5 plants in AA-induced resistance to Pcc, suggesting that ET signalling does not contribute to AA-mediated resistance against this pathogen.

Discussion

Mitochondria are essential organelles for maintaining cellular homeostasis. Disruption of mitochondrial function can lead to ROS accumulation and oxidative damage, as well as energy deficits and metabolic imbalances. Environmental stresses such as hypoxia, heat, cold, and pathogen attack are known to cause mitochondrial dysfunction (Dahal and Vanlerberghe, 2017; Kerbler et al., 2019; Meng et al., 2020; Barreto et al., 2022; Wang et al., 2022; Réthoré et al., 2024), which has shaped the evolution of mechanisms to restore mitochondrial homeostasis, placing mitochondria as a component of the cellular stress sensing machinery.

In order to expand our knowledge of mitochondrial stress responses, we examined the growth of previously characterized noxy mutants (López et al., 2011; Izquierdo et al., 2021), under various forms of mitochondrial stress, including inhibitors of all electron transport chain (ETC) complexes and mitochondrial translation. These mutants were noxy3/lon1-3, affected in the mitochondrial protease LON1; noxy23, mutated in a mitochondrial protein with a ribosomal S5 DRBM domain; noxy54/ahg2-3, which contains a truncated allele of the mitochondrial poly(A)-specific ribonuclease AHG2; and noxy22/eto1-14, an ET overproducer mutant affected in ETO1, a negative regulator of ET synthesis (Chae et al., 2003; Wang et al., 2004). A common response among all mutants was resistance to AA, consistent with prior observations for noxy3/lon1-3, noxy23, and noxy54/ahg2-3, now extended to noxy22/eto1-14. In line with this, 9-HOT pretreatment improved the adaptation of wt plants to AA, but not to the other inhibitors, confirming that the mitochondrial activity of this oxylipin is mediated by Complex III (Izquierdo et al., 2021). Mutations in noxy3/lon1-3, noxy23, and noxy54/ahg2-3 mimic mitochondrial damage, leading to reduced root length and increased levels of AOX which explains resistance to AA (Izquierdo et al., 2021). However, AA-resistant noxy22/eto1-14 shows wild-type levels of AOX, suggesting that ethylene protects mitochondria from AA damage in an AOX-independent way. The tolerance of the studied noxy mutants to the rest of inhibitors was variable, indicating that effective responses against mitochondrial stress vary depending on the specific processes affected.

In Arabidopsis, mitochondrial stress signalling has largely focused on the ANAC017-dependent retrograde pathway, using AA as a model for mitochondrial stress (Ng et al., 2013). AA inhibits ETC Complex III, blocking electron transport and inducing ROS production towards the cytosolic side of the mitochondrial membrane (Bleier and Dröse, 2013). This oxidative signal activates ANAC017 which, in turn, induces alternative respiration genes such as AOX to restore electron flow, stabilize ubiquinone (UQ) redox status, and mitigate oxidative stress (Ng et al., 2013; Khan et al., 2024). This mechanism is effective when inhibition takes place downstream of UQ or causes oxidative damage but may not be suited for other types of mitochondrial stress. For instance, malonate inhibits succinate dehydrogenase (upstream of UQ) without producing ROS, resembling a scenario of low TCA cycle activity (Quinlan et al., 2012; Jardim-Messeder et al., 2015). KCN inhibits cytochrome c oxidase and could produce ROS indirectly (Umbach et al., 2005; Vergara et al., 2012). Oligomycin, on the other hand, blocks ATP synthesis, resembling a low ADP/ATP ratio and inducing strong ROS production (Liu and Schubert, 2009). Interestingly, we have seen that anac017-1 plants have increased susceptibility to AA and oligomycin, but show wt responses to rotenone, malonate, KCN, and doxycycline (Supplementary Fig. S6), suggesting that ANAC017-mediated responses are only effective against a set of mitochondrial stresses, particularly where oxidative damage is higher. Therefore, additional mechanisms may operate to counterbalance other alterations of mitochondrial function.

ET signalling may control one of these mechanisms. Mutants affected in the negative regulator of ET biosynthesis ETO1 (noxy22/eto1-14, eto1-5, and eto1-13) showed enhanced resistance to all the tested inhibitors, whereas blocking ET signalling in ein2-5 plants caused a more pronounced growth arrest in most of the cases, when shoot growth was analysed 12 days post-germination. Interestingly, Kacprzak et al. (2020) found that blocking ET signalling led to longer hypocotyls in AA-treated plants 4 days after germination. This apparent contradiction suggests that blocking early ethylene-mediated responses results in poor adaptation to mitochondrial stress in the long term, and that ET contributes primarily to long-lasting acclimation.

Pre-adaptation of ET overproducing eto1 mutants would thus explain their broad tolerance. This idea is also consistent with recent reports of constitutive ET phenotypes in mitochondrial deficiency mutants (Merendino et al., 2020; Song et al., 2024). However, the mechanism by which ET influences mitochondria remains obscure. One possibility is that ET increases mitochondrial translation and the amount of ETC complexes (Wang and Auwerx, 2017; Yu et al., 2019), but further studies are required to clarify this matter.

An important question is whether or not ET-mediated tolerance is associated with the activation of the mitochondrial retrograde pathway. Gene expression analysis by qPCR (eto1 alleles and ein2-5) and RNA-seq (ein2-5) indicate that the canonical ANAC017-dependent pathway is largely independent of ET signalling. RNA-seq data showed that almost all the genes of the previously identified MDS group (De Clercq et al., 2013) are induced similarly in wt and ein2-5 plants, and significant enrichment of GO terms associated with mitochondria was only detected in the gene set induced by AA independently of EIN2. Furthermore, genes indirectly associated, but possibly contributing to responses to mitochondrial stress were also induced in wt and ein2-5 plants (GO terms: ‘response to hydrogen peroxide’, ‘response to ROS’ and ‘glutathione metabolic process’). At 24 h, both genotypes induced genes involved in protein import into the mitochondrial matrix, indicating the activation of the mitochondrial retrograde pathway. Several GO terms related to iron homeostasis were also induced at this time, possibly associated with the recycling and de novo assembly of iron-sulfur centres in ETC complexes and controlling cellular damage (Naranjo-Arcos et al., 2017; Distéfano et al., 2021). These findings agree with those of Kacprzak et al. (2020) who had already indicated that ET signalling promoted resistance to mitochondrial stress independently of ANAC017, even though these authors reported an increase in AA-dependent MDS gene expression when ET signalling was activated. Differences in growth and treatment conditions between both studies could account for this discrepancy.

ET-independent gene induction by AA was not limited to restoration of mitochondrial function, but also included abiotic and biotic stress responses, pointing to a signalling role of mitochondria in plant adaptation to stress. Thus, genes involved in response to hypoxia, salt, and pathogen attack were induced 3 h after treatment in an ET-independent manner, whereas anthocyanin biosynthesis and response to abiotic stimulus were induced at 24 h. Hormonal and developmental categories, including auxin signalling, senescence, and root development, were also activated in both genotypes at this time.

Other AA-induced stress responses were, however, modulated by ET signalling. At 3 h after treatment, subsets of hypoxia response genes were positively or negatively regulated by EIN2, and a similar scenario was found for several GO categories associated with response to pathogens. ABA response genes were also up-regulated in an ET-dependent manner. ET influence was even clearer at 24 h. At this time point, hypoxia and defence-related GO terms were exclusively enriched in wt plants. This supports prior evidence of ET influence on hypoxia response, and the role of mitochondria as an oxygen sensor in this process (Licausi et al., 2013; Gibbs et al., 2015; Hartman et al., 2019). Other authors had demonstrated the induction of defence and stress-related genes after AA treatment (Zhu et al., 2023). Here, we show that ET signalling is required for the full induction of this response. Early response to AA increased the expression of a considerable number of transcription factors and kinases (some located on the cell surface) involved in defence responses, suggesting that mitochondrial stress prepares the cell for pathogen perception.

EIN2 affected the induction of co-regulated groups of defence genes, suggesting that activation of ET signalling by mitochondrial stress controls specific physiological responses during pathogenesis. For instance, EIN2 promoted the induction of the glucosinolate-controlling transcription factor MYB122 and downstream indole cyanohydrin synthesis genes at 3 and 24 h, respectively. As another example, several oxygenating enzymes required for defence responses (LOX1, DOX1, DLO1, and FMO1) were simultaneously induced at 24 h in an EIN2-dependent way. The LOX1 and DOX1 enzymes initiate the synthesis of oxylipins that regulate mitochondrial stress and protect cells against oxidative damage and pathogen-induced cell death (Ponce de León et al., 2002; Hamberg et al., 2003; Vicente et al., 2012; Izquierdo et al., 2021); their induction thus suggests a role in signalling mitochondrial damage. ET signalling also influenced the induction of lignin biosynthesis genes in response to AA, consistent with higher lignin accumulation in wt versus ein2-5 plants. Many of the induced lignin-related genes encode berberine bridge enzyme-like proteins, implicated in lignin biosynthesis through the oxidation of monolignols (Daniel et al., 2015), and also in the modulation of immune responses through the oxidation of oligogalacturonides, which are DAMPs released from the cell wall after pathogen invasion (Salvati et al., 2025). Lignin deposition is a known defence mechanism against physical damage and infection (Caño-Delgado et al., 2003; Lee et al., 2019), pointing to a role of mitochondria in cellular protection, and in line with previous studies showing enhanced resistance to cell wall damage in mitochondrial mutants (Vellosillo et al., 2013; Hu et al., 2016).

Infection with the biotroph oomycete H. arabidopsidis demonstrated that EIN2 is required for full induction of defences by AA. Accordingly, the AA induction of key regulators of SA responses, such as the transcription factor WRKY51 and the SA-inactivating enzyme DLO1, was influenced by EIN2. By contrast, in the case of the necrotroph P. cucumerina, AA was also able to induce resistance but this was independent of EIN2, indicating that ET influence on defence induction by mitochondrial stress depends on pathogen-specific factors such as lifestyle. Previously, we had reported that AA-induced resistance against Hpa was associated with transcriptional priming of defence responses at the epigenetic level (Sánchez et al., 2021). Further studies will be needed to demonstrate if both processes affecting mitochondrial stress influence on defence responses, i.e. ET signalling and epigenetics, are connected.

Finally, our results uncover distinct roles for ET and ANAC017 signalling in mitochondrial stress adaptation. While ANAC017 governs early responses to oxidative stress, ET signalling appears to support long-term adaptation and defence integration. Oxylipins derived from 9-LOX also contribute to mitochondrial signalling, particularly under Complex III-associated stress. Indeed, approximately 22% of genes up-regulated 2 h after treatment with 9-KOT (Vicente et al., 2012) were also induced by AA at 3 h in our study. Given their roles in environmental stress responses, understanding the crosstalk among ET, ANAC017, and 9-HOT signalling will be critical for decoding how plants detect and respond to mitochondrial dysfunction, and the corresponding implications in plant adaptation to adverse conditions.

Supplementary data

The following supplementary data are available at JXB online.

Fig. S1. Effect of mitochondrial inhibitors on the root growth of eto1 mutants.

Fig. S2. Effect of mitochondrial inhibitors on the root growth of wild-type and ein2-5 plants.

Fig. S3. Ethylene responses in eto1 and ein2-5 mutants.

Fig. S4. ET influence on GO term enrichment of genes repressed by AA at 3 h.

Fig. S5. ET influence on GO term enrichment of genes repressed by AA at 24 h.

Fig. S6. Effect of mitochondrial inhibitors on the root growth of anac017-1 compared to wild-type plants.

Table S1. Primers used in this study.

Table S2. Differentially expressed genes in wild-type and ein2-5 plants after AA treatment.

Author contributions

YI and CC designed the research; YI, BL, and TC performed the research; YI and CC analysed the data; YI and CC wrote the manuscript. All authors approved the final version of the manuscript.

Funding

This work was supported by grant PID2021-1225240B-1000 (Spanish Ministry of Science and Innovation) to CC. The postdoctoral contract to BL was supported by grant PID2020-118607RB-I00 (Spanish Ministry of Science and Innovation) to Pablo Pulido (CNB-CSIC, Madrid, Spain).

Data availability

RNA-seq data generated in this study has been deposited in the Gene Expression Ombibus database under the accession number GSE299881 (https://www.ncbi.nlm.nih.gov/geo/query/acc.cgi?acc=GSE299881). The original contributions presented in this study are included in the article and its supplementary material. Further enquires can be directed to the corresponding authors.