Mitochondrial and Chloroplast Stress Responses Are Modulated in Distinct Touch and Chemical Inhibition Phases

Mitochondria and chloroplasts adapt to environmental changes via complex transcriptional regulation, the transcription factor ANAC017 regulating chemical inhibition responses and WRKY transcription factors modulating touch responses in Arabidopsis.

Abstract

Previous studies have identified a range of transcription factors that modulate retrograde regulation of mitochondrial and chloroplast functions in Arabidopsis (Arabidopsis thaliana). However, the relative importance of these regulators and whether they act downstream of separate or overlapping signaling cascades is still unclear. Here, we demonstrate that multiple stress-related signaling pathways, with distinct kinetic signatures, converge on overlapping gene sets involved in energy organelle function. The transcription factor ANAC017 is almost solely responsible for transcript induction of marker genes around 3 to 6 h after chemical inhibition of organelle function and is a key regulator of mitochondrial and specific types of chloroplast retrograde signaling. However, an independent and highly transient gene expression phase, initiated within 10 to 30 min after treatment, also targets energy organelle functions, and is related to touch and wounding responses. Metabolite analysis demonstrates that this early response is concurrent with rapid changes in tricarboxylic acid cycle intermediates and large changes in transcript abundance of genes encoding mitochondrial dicarboxylate carrier proteins. It was further demonstrated that transcription factors AtWRKY15 and AtWRKY40 have repressive regulatory roles in this touch-responsive gene expression. Together, our results show that several regulatory systems can independently affect energy organelle function in response to stress, providing different means to exert operational control.

As plants are sessile organisms, they must respond in an energy-efficient and effective way to cope with continuous change in the environment. Chloroplasts provide the energy equivalents to reduce CO₂ to carbohydrates via photosynthesis. Plant mitochondria are crucial to support energy metabolism, photosynthetic and photorespiratory functions, survival during dark periods, and meristem activity (Ishizaki et al., 2006; Van Aken et al., 2007; Howell et al., 2009; Araújo et al., 2014). As mitochondria and chloroplasts are tightly involved in cellular metabolism, they are in a prime position to sense cellular dysfunction and are therefore thought to be one of the initial sites of stress recognition and can contribute to programmed cell death execution (Vanlerberghe, 2013). A significant body of evidence supports a crucial role for mitochondria and chloroplasts in perceiving and responding to biotic and abiotic stress conditions (Skirycz et al., 2010; Leister et al., 2011; Nomura et al., 2012; Zhang et al., 2014; Stael et al., 2015). This is further supported by several studies demonstrating that the transcriptomic response to mitochondrial or chloroplast dysfunction by chemical or genetic disruption appears to be part of broad biotic and abiotic stress responses (Schwarzländer et al., 2012; Van Aken and Whelan, 2012).

For mitochondria and chloroplasts to be involved in this cellular signaling network, communication from nucleus to organelle (anterograde) and from organelle to nucleus (retrograde) is essential (Ng et al., 2014). In terms of anterograde signals, it is known that nuclear genes encoding mitochondrial proteins respond to the diurnal cycle (Giraud et al., 2010), sugar content (Comelli et al., 2012), stress conditions (Van Aken et al., 2009), and germination cues (Law et al., 2012). On the other hand, retrograde signaling has been studied extensively by using chemical inhibition of organellar function (e.g. antimycin A, methyl viologen, and norflurazon). Several studies showed that a subset of nuclear genes consistently responds to inhibition of mitochondrial function (Van Aken et al., 2007; Schwarzländer et al., 2012; De Clercq et al., 2013). This group of genes was termed the mitochondrial dysfunction stimulon (De Clercq et al., 2013) and contains several genes that play important roles in stress tolerance in Arabidopsis (Arabidopsis thaliana). For example, the glycosyltransferase UGT74E2 targets auxins and upon overexpression triggers profuse shoot branching and drought tolerance (Tognetti et al., 2010). Also, the classical mitochondrial stress marker gene alternative oxidase AOX1a was found to be important during osmotic stress (Dojcinovic et al., 2005; Dahal et al., 2014), and recently it was shown that the mitochondrial outer membrane ATPase AtOM66 (AtBCS1) is a regulator of cell death that controls pathogen resistance, senescence, and drought tolerance (Zhang et al., 2014). Therefore, it is evident that energy organelle signaling is an important part of general stress responses, and understanding the way it is controlled could provide a means to enhance stress tolerance and, thus, productivity in crop species.

Over the past few years, regulatory proteins that are involved in retrograde responses in plants have been identified. Using a reverse genetic approach, a number of transcription factors were identified, such as ABI4 (Koussevitzky et al., 2007; Giraud et al., 2009), AtWRKY15, and AtWRKY40 (Shang et al., 2010; Vanderauwera et al., 2012; Van Aken et al., 2013). In a forward genetic screen using the AOX1a promoter fused to luciferase, several genes that influenced plant retrograde signaling were identified. A cyclin-dependent kinase, CDKE;1, was found as a positive regulator of both mitochondrial retrograde signaling and chloroplast signaling (Ng et al., 2013a; Blanco et al., 2014). Additionally, a group of transmembrane NAC-type transcription factors was identified and shown to positively regulate mitochondrial retrograde signaling (De Clercq et al., 2013; Ng et al., 2013b) by binding a mitochondrial dysfunction DNA motif. Surprisingly, subcellular localization studies showed that some of these NAC transcription factors are anchored in the endoplasmic reticulum (ER) by their transmembrane domain. This suggests that the NAC transcription factors are released from the ER upon stress perception and subsequently translocate to the nucleus. Furthermore, application of rhomboid protease inhibitors was able to partially suppress retrograde induction, indicating NAC transcription factors are released by proteolytic cleavage (Ng et al., 2013b). Inhibitors targeting ER and mitochondrial Ca²⁺ channels were demonstrated to enhance or reduce stress responsiveness of MDR genes (Vanderauwera et al., 2012), further pointing toward a role for the ER. It was also found that an antagonistic relationship occurs between retrograde and auxin signaling. Loss of transporters involved in auxin distribution such as PIN1 and BIG results in an enhanced activation of mitochondrial retrograde response upon stimulation, while exogenous application of auxin suppresses retrograde signaling (Ivanova et al., 2014; Kerchev et al., 2014).

The relative importance of these regulators and whether they act downstream of separate or overlapping signaling cascades feeding into expression of common genes is still unclear. Here, we demonstrate that multiple stress-related signaling pathways, with distinct time signatures, converge on overlapping gene sets involved in chloroplast and mitochondrial function. When plants are spray-treated with mitochondrial or chloroplast inhibitors, one expression phase related to inhibitory activity of the active compound peaks around 3 to 6 h after treatment and is largely controlled by ANAC017. Furthermore, a rapid phase of gene expression within 10 min of treatment also affects stress signaling target genes and correlates with rapid changes in metabolic substrate concentrations. We further demonstrate that this rapid phase is related to “touch” signaling and is modulated by WRKY transcription factors.

RESULTS

NAC Transcription Factors Involved in Mitochondrial Retrograde Signaling Show Divergent Expression Patterns

Two previous studies identified five membrane-anchored NAC transcription factors as transcriptional regulators of plant mitochondrial retrograde signaling (De Clercq et al., 2013; Ng et al., 2013b), so the first aim was to clarify their relative contributions to retrograde signaling. To do this, we assessed transcript abundance of NACs in 2-week-old seedlings treated with monofluoroacetate (MFA; a tricarboxylic acid [TCA] cycle inhibitor) for 6 h (Fig. 1). ANAC017 transcripts were most highly abundant in both untreated and stressed conditions, but showed no significant stress induction. Its closest homolog, ANAC016, was almost 1,300 times less abundant and was not further characterized in this study. ANAC013 was highly inducible, but even after induction transcript levels were still less abundant than ANAC017 in untreated conditions. Similar induction patterns were observed for ANAC053 and ANAC078: Both were significantly induced by MFA, but did not reach the high transcript abundance of ANAC017.

Figure 1.

Expression of NAC transmembrane transcription factors. A, mRNA abundance of NAC transcription factors was quantified using qRT-PCR in 2-week-old in vitro-grown pooled Arabidopsis seedlings, before treatment and 6 h after treatment with 50 µM MFA (n = 3; mean ± se). Expression levels were normalized to ANAC016 expression in untreated conditions (set to 1). Asterisk indicates significant induction of mRNA levels after MFA treatment compared to untreated control (P < 0.05). B, Two-week old in vitro-grown seedlings were spray treated with Flg22, antimycin A, MFA, or mock solution (water + Tween). Pooled plants were collected, and mRNA levels were measured by qRT-PCR (± se) and normalized to untreated samples (0 h). *P < 0.05 versus 0 h (n = 3). C, Promoter-GUS analysis of NAC expression patterns. The 1.5-kb promoter regions of selected NAC transcription factors fused to a GUS reporter gene were stably transformed into Arabidopsis Col-0 plants. Three-week-old in vitro-grown plants were stained for GUS activity before and after 6 h of treatment with 50 µM antimycin A.

Next, detailed time-course experiments were performed with a wider range of stress-inducing factors: bacterial elicitor Flg22 and UV and chemical inhibitors operating on complex III (antimycin A) and the TCA cycle (MFA) were applied or sprayed on 2-week-old seedlings, along with a mock spray treatment control. Samples were collected at a range of time points over 24 h, and mRNA abundance of the transcription factors and their previously identified target genes AOX1a and UGT74E2 were measured (Fig. 1B). In accordance with the previous MFA experiment, ANAC016, ANAC017, and ANAC078 displayed no or very mild stress response. ANAC013, ANAC053, and the target genes AOX1a and UGT74E2 showed a clear induction after 3 h of treatment. Transcript levels steadily decreased to almost basal levels 12 and 24 h after treatment, except for the MFA treatment, which provoked longer term induction. UV treatment caused the most gradual rates of induction over time and in some cases expression increased at 1 h (Supplemental Fig. S1).

To assess promoter activities, 3-week-old GUS-based reporter lines were sprayed with antimycin A (Fig. 1C). In untreated conditions the promoter of ANAC017 showed by far the strongest activity throughout the entire root and shoot of the plants, but in accordance with its transcript levels no obvious changes after stress treatment were observed (a magnified image of ANAC017:GUS activity in root tips is shown in Supplemental Fig. S1B). The promoter of ANAC013 showed low but discernible expression in young leaves and root tips. After stress treatment, a clear induction in the leaves was observed, particularly in the youngest leaves, in line with a previous report (De Clercq et al., 2013). ANAC053 showed clear promoter activity throughout the root, but expression in leaf tissues was only observed after stress treatment. ANAC078 showed the weakest expression with some GUS staining in root tips, but no response to the stress treatment. Overall, these experiments show that ANAC017 is the most strongly expressed but was not induced by stress treatments.

ANAC017 Is a Key Regulator of Responses to Chemical Mitochondrial Inhibition

To determine the effect of antimycin A treatment on the respiratory capacity of plants, dark respiration measurements were performed and showed no gross depletion of oxygen consumption after 4 h of antimycin A treatment compared to untreated controls (Supplemental Fig. S2B; P > 0.50, n = 4). This indicates that spray treatment with 50 µm antimycin A, while known to lead to cytochrome pathway inhibition and induction of AOX (Vanlerberghe and McIntosh, 1992; Ng et al., 2013b), is not a lethal chemical treatment, does not stop plant respiratory metabolism, and thus represents a reasonable model for assessing signaling responses following a partial mitochondrial dysfunction.

Next, to assess the individual contributions of previously identified transcription factors in plant retrograde signaling (belonging to NAC and WRKY families, and ABI4), we collected a set of unpublished and previously published loss- and gain-of-function mutants (see “Materials and Methods”). This set of 15 lines including the wild type (Col-0) was subjected to antimycin A treatment. Samples were collected at 0, 0.5, 1, 3, 6, and 12 h after treatment in three independent pools of approximately 20 2-week-old seedlings per genotype to account for individual variations. Transcript abundance of target genes was measured using qRT-PCR (Fig. 2). In untreated conditions, overexpression lines of ANAC013, ANAC017, ANAC053, and ANAC078 already showed significantly elevated expression of AOX1a (P < 0.05) compared to Col-0 (Fig. 2, AOX1a, left panel, time 0). ANAC013 itself was significantly (P < 0.05) induced in ANAC017, ANAC078, and ANAC013 overexpression lines (Fig. 2; Supplemental Fig. S2, ANAC013 panels). ANAC053 was significantly induced in ANAC017 and ANAC053 overexpression lines (P < 0.05; (Fig. 2, ANAC053 panels; Supplemental Fig. S2). Overall, these data suggest that all four NAC transcription factors upon overexpression have a basic capability of inducing mitochondrial retrograde signaling marker genes and some of its coregulators.

Figure 2.

ANAC017 is a key regulator of antimycin A-induced mitochondrial retrograde signaling. Two-week-old in vitro-grown seedlings were spray treated with antimycin A. Plants were collected in pools, and relative mRNA levels were measured by qRT-PCR and normalized to Col-0 untreated samples (0 h). For clarity, ANAC transcript levels in their respective overexpression and loss-of-function lines were not displayed. Colored asterisks indicate statistically significant differential expression in a specific genotype versus Col-0 at the marked time point (P < 0.05; n = 3); white asterisks indicate statistically significant differential expression in Col-0 at the marked time point versus Col-0 0 h (P < 0.05; n = 3).

By far the most drastic effect was observed in anac017 loss-of-function plants, where at the 3 h time point no significant induction of AOX1a (1.71-fold, P = 0.47) was observed compared to Col-0 0 h, as opposed to 10.4-fold induction in Col-0 (P < 0.01; Fig. 2). anac017 loss-of-function lines also showed no induction of ANAC013 and ANAC053 at the 3 h time point, as opposed to 7.3-fold (ANAC013) and 2.9-fold (ANAC053) in Col-0 (P < 0.05). This indicates that ANAC013 and ANAC053 are target genes downstream of ANAC017 in the signaling cascade, further supported by the observation that ANAC013 and ANAC053 are constitutively induced in ANAC017 overexpression plants (Fig. 2). At 6 h in anac017, only a mild induction of AOX1a and ANAC013 was observed, which was significantly lower than in Col-0 (P < 0.05). None of the loss- or gain-of-function lines for other NAC, WRKY, or ABI4 transcription factors showed similarly strong reductions in antimycin A-induced target gene expression. In line with previous reports of ABI4 being a repressor of AOX1a expression, an increased expression of AOX1a was observed in abi4-1 at the 0 h time point (P < 0.05; Giraud et al., 2009).

To confirm the robust nature of the upstream signaling role of ANAC017 in antimycin A-induced gene expression, a repeat experiment was performed on plants grown in soil and Col-0 was also compared to two independent anac017 alleles (Supplemental Fig. S2C). In agreement, both anac017 lines showed 4- to 6.5-fold reductions in AOX1a transcript induction at 3 h (P < 0.01, n = 3). At 6 h, AOX1a transcript levels were already falling in the Col-0 plants, but showed a delayed increase in the anac017 lines that was well below the AOX1a levels reached in Col-0. Overall, these results indicate that ANAC017 is a key regulator of the transcriptional stress response to mitochondrial inhibition in seedlings grown in both Murashige and Skoog (MS) media and soil, and it is also regulating induction of ANAC013 and ANAC053.

ANAC017 Affects Plant Resistance to Energy Organelle Inhibition

To consolidate that the role of ANAC017 extends beyond transcriptional responses, we performed a comprehensive phenotype analysis of all 15 NAC, WRKY, and ABI4 lines for resistance to constitutive energy organelle function inhibition (Fig. 3). To do this, the different plant lines were incubated on vertically positioned plates supplemented with antimycin A (complex III inhibitor), rotenone (complex I inhibitor), and methyl viologen (MV; a superoxide generator that leads to damage in chloroplasts in lighted conditions). Plants were grown for 7 d and primary root length was measured (Fig. 3). On control plates, a number of lines showed mild but significant growth reductions of between 13 and 30% compared to Col-0 plants. Antimycin A treatment resulted in significant growth reduction in all lines; anac017 plants were particularly highly sensitive. In contrast, ANAC017- and ANAC013-overexpressing lines were significantly more tolerant to high concentrations of antimycin A. Rotenone had a milder effect on root length than antimycin A, but again plants overexpressing ANAC017 showed significantly higher resistance than Col-0, while anac017 plants were most susceptible. MV caused the most severe growth reductions (Fig. 3). Again, plants overexpressing ANAC017 and ANAC013 showed a significant resistance to MV compared to Col-0, while the anac017 and anac013 loss-of-function lines showed increased susceptibility. On the other hand, AtWRKY15, 40, 63, and ABI4 have no discernible effects specific to any of the treatments. From these results, it is apparent that ANAC017 is crucial for optimal plant growth under conditions of organellar inhibition and to a lesser extent also ANAC013 as reported previously (De Clercq et al., 2013).

Figure 3.

ANAC017 is a determinant of organelle dysfunction tolerance. Seeds of different genotypes were grown vertically on MS media plates supplemented with antimycin A, rotenone, or MV. Primary root growth was measured after 7 d in the growth chamber (±se). Significant differences with Col-0 within the same treatment are indicated (n > 7, *P < 0.05, **P < 0.01, and ***P < 0.001).

ANAC017 Is Also a Regulator of Chloroplast Retrograde Signaling

Root growth characteristics showed a clear role for ANAC017 in resistance to antimycin A and rotenone. Interestingly, MV caused similar phenotypes in the gain- and loss-of-function lines of ANAC017 (Fig. 3). MV is generally considered as a photosynthesis inhibitor that accepts electrons directly from thylakoid photosystem I (Fd subunit) and transfers them to molecular oxygen generating superoxide. In light conditions, MV is regarded as a chloroplast-specific ROS generator (Tsang et al., 1991). We sprayed 2-week-old Col-0 and anac017 seedlings with MV and assessed the effects on gene expression (Fig. 4). To confirm that MV did not cause a significant mitochondrial defect, the expression of mitochondrial dysfunction marker genes AOX1a and AtOM66 was measured. The mock spray treatment caused mild inductions of both genes; however, no additional inductions were caused by the MV treatment (Fig. 4). It was also evident that ANAC017 is required for the mock induction at around 3 h. Based on an analysis of public microarray data, multiple marker genes were selected that respond more to MV treatment than mock (Kilian et al., 2007; Yoshida and Noguchi, 2009). qRT-PCR analysis revealed very high induction by MV of HSP17.6A (∼80-fold) and HSP17.6CII (∼250-fold) in Col-0, whereas no or minor inductions were caused in the mock treatment (Fig. 4). Other genes such as the TINY AP2 transcription factor, sHSP23.5, AtWRKY30, and AtWRKY33 also showed severalfold higher induction in MV treatment compared to mock in Col-0. In contrast, the MV-inducible genes were up to 10-fold less induced in anac017 plants compared to Col-0 (Fig. 4). These experiments indicate that ANAC017 is also involved in the transcriptional response that is triggered by chloroplast-generated superoxide signaling in response to MV.

Figure 4.

ANAC017 is a regulator of chloroplast stress response induced by MV. Two-week-old Col-0 and anac017 plants were spray treated with mock solution or 50 µM MV. Pools of plants (n = 3) were collected in triplicate, and mRNA levels were measured by qRT-PCR (±se). Statistically significant changes in Col-0 caused by treatments are indicated (#P < 0.05; ##P < 0.01). Statistically significant changes of anac017 versus Col-0 within the same treatment are indicated (*P < 0.05; **P < 0.01).

The further assess the role of ANAC017 in chloroplast retrograde signaling, Col-0 and anac017 plants were exposed to high light conditions (750 µmol s⁻¹ m⁻²). Samples were collected in biological triplicate and marker genes were tested using qRT-PCR (Supplemental Fig. S3). HSP17.6 A and HSP17.6 CII were highly induced (>1,000-fold) by 1 h of high light in Col-0, while AtAPX2 was induced ∼150-fold. To determine if the transcripts responded to temperature changes caused by the high light treatment or the light itself, an additional control experiment was performed where plants were incubated under normal light conditions but at a temperature matching that observed in the plates during high light treatment (28°C). From this it was clear that exposure to high light, rather than elevated temperature, was the main contributor to the gene expression increases observed. Mitochondrial stress marker gene AOX1a showed mild but significant high light induction (3-fold). Overall, anac017 plants showed no significant differences in high light-induced gene expression compared to Col-0 for any of the marker genes tested, indicating that ANAC017 plays no significant role in high light-induced signaling from the chloroplasts.

To clarify if ANAC017 plays a role in the previously described GUN (genomes uncoupled) pathway of chloroplast retrograde signaling (Susek et al., 1993), Col-0 and anac017 plants were treated with the carotenoid biosynthesis inhibitor norflurazon, and the expression of GUN marker genes LHCB1.1 and LHCB2.4 was monitored (Supplemental Fig. S3B). As expected, transcript levels of LHCB2.4 and LHCB1.1 were strongly repressed within 3 h of norflurazon application, but no genotype-specific differences between Col-0 and anac017 plants were found.

Stress Treatment by Spray Application Induces a Two-Phase Response in AtOM66 and WRKY Transcription Factors

The above experiments showed a clear role for NAC transcription factors, especially ANAC017, in retrograde responses to inhibitors around 3 h after treatment. Previous studies also identified AtOM66 (encoding an outer mitochondrial membrane AAA ATPase) as a target for mitochondrial and chloroplast retrograde signaling but suggested it may be regulated differently to mitochondrial dysfunction stimulon genes such as AOX1a and UGT74E2 (Ho et al., 2008; Van Aken and Whelan, 2012). As AtOM66 was previously shown to be regulated by WRKY transcription factors via direct binding of the AtOM66 promoter (Van Aken et al., 2013), stress-related (antimycin A, MFA, Flg22, and UV) WRKY expression patterns were examined by qRT-PCR as described in Figure 1B (Fig. 5; Supplemental Fig. S1). AtWRKY63 showed an expression pattern very similar to AOX1a and ANAC013 with stress-inducible expression peaking around 3 to 6 h after treatment for Flg22 and antimycin A and 12 h for MFA. Surprisingly, a more complex expression pattern was observed for AtWRKY40, AtWRKY15, and AtOM66, with a very high level of induction by the 30 min time point (>380-fold for AtWRKY40, >100-fold for AtOM66, and >10-fold for AtWRKY15), which rapidly declined by 1 h after treatment. Following this initial transient peak, these genes showed a second phase of induction similar in timing to AOX1a (Figs. 1B and 5). The mock treatment triggered a similar induction of the first peak, but showed much weaker induction than the inhibitors around 3 to 6 h. This indicates that the spray treatment itself appears to be the main cause for the first peak, while the stress-inducing active compounds are mostly responsible for the second phase for both NAC and WRKY transcription factors. To determine the effect of Tween 20 on the early response, a repeat experiment using water without Tween 20 for the spray treatment was performed, and this showed similar highly significant inductions within 25 min (e.g. AtWRKY40 was induced 226-fold, P < 0.001; Supplemental Fig, S4). In conclusion, AtOM66 and its regulators AtWRK40 and AtWRKY15 display a two-phase expression pattern, with a rapid peak at 30 min caused by the physical spray treatment and a second peak around 3 h or later caused by the sprayed compound.

Figure 5.

Spray treatment causes a two-phase stress response. A, Two-week-old in vitro-grown seedlings were spray treated with Flg22, antimycin A, MFA, or mock solution. Pooled plants were collected, and mRNA levels were measured by qRT-PCR and normalized to untreated samples (0 h). *P < 0.05 versus 0 h (n = 3). B, Two-week-old Col-0 plants were sprayed with water + Tween (mock), and pools of plants were collected in triplicate. mRNA levels were measured using qRT-PCR (mean ± se) and normalized to time point 0 h. C, Two-week-old in vitro or soil-grown seedlings were treated as indicated. Pools of plants were collected (n = 3), and mRNA levels were measured by qRT-PCR and normalized to untreated samples (0 h). Statistically significant changes compared to untreated samples at time 0 h are indicated (*P < 0.05; **P < 0.01).

The Rapid Response Peak Is Related to Touch and Wounding Responses

To investigate this 30 min peak in more detail, a mock treatment time course was performed and samples were collected every 10 min during the first hour, then at 3, 6, and 24 h in triplicate pools (Fig. 5B). Significant induction of AtOM66 (5.1-fold; P < 0.02) and AtWRKY40 (67-fold; P < 0.005) was already observable by 10 min, with the highest expression level reached in the 20 to 30 min time points. AOX1a showed no significant induction at these early time points. Both AtOM66 and AOX1a showed a mild but significant induction at the 3 h time point, possibly linked to the increased humidity and wetness caused by the spray treatment.

The rapid and highly transient nature of this early response could be caused by a number of different factors, including the mechanical stimulation of the spray treatment (similar to touch responses; Braam and Davis, 1990) or the changes in humidity and air composition by opening the petri dish plates. To investigate this, the effect of opening plates and mechanical stimulation by touching with blunt tweezers were each compared with the water spray treatment and a negative control. A similar control experiment was also performed on soil-grown plants. Samples of pooled plants were collected in triplicate after 25 min and mRNA expression levels of AtOM66 and AtWRKY40 were measured by qRT-PCR (Fig. 5C). The touch and water spray treatments caused rapid and strong inductions both in soil and in vitro, while just transiently opening the lid that led to minor vibrations resulted in minor inductions. The water spray was the most uniform manipulation of the leaf surfaces of the plants and gave the highest fold changes overall.

A number of previous studies have looked at touch and early wounding responses using genome-wide expression studies. In the Affymetrix ATH1 data set E-GEOD-48676, 5-week-old Arabidopsis leaves were wounded and collected 10 min after treatment. Although raw data files are publicly available on ArrayExpress, we could not find a published analysis. Therefore, we analyzed the data in detail. In total, 1127 probe sets of 22810 were significantly altered after wounding (>2-fold change, posterior probability of differential expression [PPDE] > 0.95; Supplemental Table S1). A total of 745 probe sets were up-regulated and 382 were down-regulated. In agreement with our observations in Figure 5, A and B, AtWRKY40 (131-fold), AtOM66 (9.7-fold), and AtWRKY15 (2.1-fold) were significantly induced within 10 min after wounding, with AtWRKY40 being the second most induced gene across the genome. In contrast, AOX1a and UGT74E2 were not induced significantly, which confirms the specificity of the response and similarity with our experiments. A study looking at changes in gene expression within 5 min of wounding also noted a significant induction of AtWRKY40 using both Agilent microarrays and qRT-PCR (Walley et al., 2007).

In another study, 4-week-old plants were touched by gently folding back and forth the leaves, not damaging them (Lee et al., 2005). Samples were collected 30 min after treatment. The authors reported that expression of 589 genes was significantly touch-induced, while 171 were repressed. AtWRKY40 (43-fold), AtOM66 (10.7-fold), and AtWRKY15 (4.3-fold) were also found to be significantly touch-induced, while AOX1a and UGT74E2 were not. In conclusion, the rapid first peak observed in Figure 5 appears to reflect touch and wounding responses, which affects approximately 3 to 5% of the Arabidopsis transcriptome. This includes induction of several genes encoding mitochondrially targeted proteins, including AtOM66, as well as mitochondrial dicarboxylate substrate carriers (DIC2, 48-fold; DIC1, 21-fold; PPDE > 0.99), which are thought to provide the primary sources of reductant for mitochondrial function (Palmieri et al., 2008).

Touch Responses Directly Affect Substrate Levels and the Expression of Genes for Transporters for Energy Metabolism

Our initial interest in the early touch response was the strong induction of AtOM66, which encodes a mitochondrial protein. Touch responses caused by physical manipulation with water jets result in almost immediate mitochondrial and cytosolic Ca²⁺ spikes (Logan and Knight, 2003; Loro et al., 2012). This suggests mechanical manipulations may directly affect mitochondrial function. The touch- and wounding-related microarray expression data suggest that mitochondrial dicarboxylate carriers (DICs) are among the most highly induced genes (Lee et al., 2005). Therefore, we tested if transcripts for the three Arabidopsis DIC genes are responding to the mock spray treatment presented in Figure 5B. DIC2 was induced 91-fold, DIC1 was induced 17-fold, and DIC3 was induced 25-fold (Fig. 6). We also tested a chloroplast-located Glc-6-phosphate transporter, GPT2, reported to be crucial for metabolic readjustment during dynamic acclimation (Athanasiou et al., 2010). GPT2 was 13-fold induced by touch treatment. Together, these findings suggest that transcriptional changes during touch responses are directed to modulate import/export of metabolic substrates from energy organelles.

Figure 6.

Touch responses are coupled to rapid changes in energy metabolism. Heat map of metabolite and transcript levels at indicated intervals (in minutes) after mock spray treatment, after hierarchical clustering (Pearson correlation and average linkage). Values represent fold changes normalized to 0 min time point. Asterisks indicate differential abundance during at least one of the time points (P < 0.05). Transcript levels are indicated by red arrows to differentiate from metabolites (n = 3).

To confirm if touch responses indeed affect mitochondrial metabolism, metabolite levels were analyzed in the same samples used for the detailed mock spray time course used in Figure 5B (Fig. 6). In line with the strong touch-induced expression of DIC transcripts, several of their major substrates, including succinate and citrate, decreased in abundance approximately 2-fold within 10 min (P < 0.05). These metabolite levels started recovering after 30 min and returned to normal levels by 3 to 24 h. A number of amino acids, such as l-Ala and l-Thr, followed a similar trend, while other components such as glycerol showed increases toward the later time points. These results indicate that touch responses can affect mitochondria both through substrate availability and transcriptional signaling as part of a wider cellular response.

AtWRKY15 and AtWRKY40 Are Involved in Regulation of Touch Responses

To assess if the previously identified regulators could affect touch-related expression, AtOM66 expression was measured in the time course experiment with antimycin A treatment (Figs. 2 and 7; n = 3). At the 3 h time point, anac017 plants again showed a complete loss of AtOM66 induction, whereas none of the tested WRKY lines negatively affected AtOM66 induction. Similar reduced AtOM66 levels were observed during a treatment time course using two independent alleles of anac017 mutants grown in soil (Supplemental Fig. S2C). However, loss of ANAC017 function had no effect on the touch-related peak at 30 min. This indicated that two separated signaling cascades were involved in these transcriptional responses. None of the transgenic lines tested above had a completely repressed touch-response peak, but our screen suggested that plants overexpressing AtWRKY15 and AtWRKY40 had an attenuated response (Fig. 7A). Due to the highly transient nature of the touch response resulting in rapid changes in AtOM66 expression within minutes (see Fig. 5), the interexperiment variability caused these differences between AtWRKY15 OE or AtWRKY40 OE and Col-0 at the 30 min time point not to pass a standard unpaired t test. However, within each of three experiments the AtOM66 expression values in AtWRKY15 OE and AtWRKY40 OE were consistently lower than in Col-0 and were found to show significantly reduced induction compared to Col-0 by a paired t test (P < 0.05).

Figure 7.

AtWRKY15 and AtWRKY40 modulate touch-responsive gene expression. A, Two-week-old in vitro-grown seedlings were spray treated with antimycin A. Pooled plants were collected, and mRNA levels were measured by qRT-PCR (n = 3) and normalized to Col-0 untreated samples (0 h). B, Two-week-old seedlings of different genotypes were spray-treated and pools of plants were collected in triplicate before treatment and at the indicated times after treatment. mRNA levels were calculated by qRT-PCR (±se) and normalized to Col-0 untreated samples (0 h). Statistical significance is indicated by asterisks: *P < 0.05, **P ≤ 0.01, and ***P < 0.001.

To identify other regulators of the rapid touch responses, we carried out transcriptomic analysis focusing on the first 60 min following spray treatment. A range of mutants with previously described roles potentially relating to very early touch and wounding responses were included (see “Discussion” for more details): touch-related plasma membrane Ca²⁺ channels mca1 mca2, extracellular ATP receptor atdorn1-3, OXDORN1 (AtDORN1 overexpressor), and coi1 (coronatine-insensitive jasmonic acid [JA] receptor). Two-week-old in vitro-grown seedlings were mock spray treated and triplicate pools of plants were collected at 0, 30, and 60 min. mRNA abundance of touch-response marker genes was measured by qRT-PCR analysis (Fig. 7B; Supplemental Fig. S4). Overexpression lines of AtWRKY15 and AtWRKY40 (independent allele) were also included and again displayed a significant reduction of AtOM66 expression at 30 min after treatment (∼35% reduction; P < 0.05). In contrast to our expectations, a significantly higher induction was found in atdorn1-3 plants (∼50% increase; P = 0.01), while the OXDORN1 line did not show any significant differences (Fig. 7B). Homozygosity of the T-DNA insertion in atdorn1-3 was confirmed by PCR, and AtDORN1 transcript levels were confirmed to be overexpressed in OXDORN1 plants (Supplemental Fig. S4). AtWRKY40 induction at 30 min was 50% reduced in AtWRKY15 OE plants (P < 0.01) and 42% more induced in atdorn1-3 plants (Fig. 7B). Also, AtWRKY15 induction was significantly repressed in AtWRKY40 OE plants (∼55% reduction; P < 0.05). The mca1 mca2 and coi1 plants showed no significant differences compared to the wild type (Fig. 7B). A mutant in gibberellin oxidase ga2ox7 that was recently reported to be involved in touch signaling (Joao Pimenta Lange and Lange, 2015) was also analyzed, but no significant effect in transcriptional touch responses were observed compared to Col-0 at 25 min (Supplemental Fig. S4). In conclusion, our results show that AtWRKY40 and AtWRKY15 are negative regulators of touch-induced transcriptional responses.

DISCUSSION

Of several previously identified transcription factors involved in mitochondrial retrograde signaling in response to organelle inhibition, ANAC017 appears to be the main positive transcriptional regulator known to date. ANAC017 is most strongly expressed and appears to be epistatic over ANAC013 and ANAC053, as loss of ANAC017 abolishes their stress induction. This is in line with the direct binding of the ANAC013 promoter by ANAC017 as reported previously (De Clercq et al., 2013), and the ANAC053 promoter contains a consensus ANAC017 mitochondrial dysfunction DNA motif binding site. At 3 h after treatment, the downstream target genes show an almost complete lack of induction in anac017 lines compared to wild-type plants. Some weak induction at 6 h is observed, which may be due to the partial redundancy with the other NAC family members, mainly ANAC013, ANAC053, and/or ANAC078. The later increase in signal may occur through a positive feedback loop by ANAC013 as it binds and stimulates its own promoter (De Clercq et al., 2013), which may take some hours to reach a noticeable threshold in the absence of ANAC017. As previously published, ABI4 appeared to have a significant repressive effect mainly on AOX1a in untreated conditions (Giraud et al., 2009). The role of WRKY transcription factors in response to mitochondrial inhibition is less clearly defined during the 3 to 6 h gene expression peak. This may be due at least in part to the large redundancy within the WRKY gene family, as previous studies have shown that more than 10 different WRKY proteins can bind the promoters of AOX1a and AtOM66 (Van Aken et al., 2013). For instance, AtWRKY40 is known to form a functional complex with related proteins AtWRKY18 and AtWRKY60 (Xu et al., 2006). Analysis of root growth also showed that ANAC017, and to a lesser extent ANAC013, provided the plants with enhanced resistance to inhibitors of organelle function. Altered expression of a number of transcription factors resulted in reduced root growth (Fig. 3), which likely reflected a perturbation of basal gene expression resulting in mild growth penalties to the plants.

Our previous analysis suggested that the role of ANAC017 in stress signaling is mostly confined to regulation of gene expression after mitochondrial function was inhibited at the level of Complex III (antimycin A), Complex IV (myxothiazol), and/or the TCA cycle (MFA; Ng et al., 2013b). The experiments shown here suggest that the role of ANAC017 may stretch further than just mitochondrial dysfunction, as shown by the very clear inhibition of stress signaling in response to chloroplast superoxide generator MV (Tsang et al., 1991). MV induces a transfer of electrons from photosystem I to oxygen in illuminated conditions, resulting in superoxide generation and oxidative stress that harms chloroplast function. The lack of mitochondrial stress marker gene induction (e.g. AOX1a and AtOM66) following treatment with MV compared to the mock treatment further supports that MV triggered signaling in the light is largely independent of mitochondria. Nevertheless, ANAC017 does not seem to mediate the signals for every type of stress that induces ANAC017 target genes. Stress response triggers such as UV, salicylic acid, cold, abscisic acid, and Flg22 can also induce gene expression, e.g. AOX1a expression (Ng et al., 2013b; this study). However, loss of ANAC017 did not significantly reduce the induction of AOX1a to many of these triggers (Ng et al., 2013b). From our results ANAC017 is also not involved in every type of chloroplast retrograde response. High light stress causes a rapid induction within 1 h of many of the same target genes as MV (e.g. HSP17.6A, HSP17.6 CII, TINY TF, and dual-targeted sHSP23.5); however, loss of ANAC017 function did not significantly reduce these high light responses, nor did anac017 plants have a GUN phenotype in our assays. Evidently, these genes are steered by separate pathways in the case of MV or high light, but have similar gene expression outcomes, providing an elegant example of convergent signaling. Previous reports also suggested a large overlap between target genes of ANAC017 and singlet oxygen-induced genes in the flu mutant (Meskauskiene et al., 2001; Ng et al., 2013b). Our results indicate that both ANAC017 and flu-mediated signaling may converge on multiple common target genes, but the signaling cascades that lead there appear to be distinct, shedding further light on the apparent cross talk between different ROS stress signaling pathways (Laloi et al., 2007; Lee et al., 2007). Indeed, several of these genes contain consensus ANAC017 binding sites [CA(A/C)G] in their promoters, such as HSP17.6A and sHSP23.5.

Our results also raise the question of whether the ANAC017 pathway requires energy organelle dysfunction as part of the signaling event. Perhaps ANAC017 on the ER membrane receives inputs from many different upstream signaling pathways, of which organellar dysfunction, ER stress, or unfolded protein responses may be some examples (Senft and Ronai, 2015). To a large extent, the cell can distinguish the source of the stress despite the fact that ANAC017 is mediating multiple signals and overlapping target genes. This is evident from the observation that treatment with antimycin A or MV causes very different levels of induction of ANAC017 target genes: e.g. AOX1a and AtOM66 are 4- to 10-fold more induced by antimycin A than by MV, while HSP17.6A and HSP17.6 CII are 8- to 16-fold more induced by MV than antimycin A (Supplemental Fig. S5; Fig. 4). It appears that additional (currently unknown) factors may codetermine the specificity of ANAC017.

Unexpectedly, gene expression of mitochondrial and chloroplast proteins also responded to a much more immediate transient signaling cascade. This transient expression was initiated within 10 min of inhibitor application and appeared uncoupled from the organellar inhibition caused later on by the active compound. Again, overlapping genes with organelle inhibitor-induced expression were affected (e.g. AtOM66), but distinct pathways lead to activation, implying convergent signaling. This rapid response is very similar to mechanical manipulation signaling, such as touch and wounding responses, which induce Ca²⁺ spikes in mitochondria within seconds (Logan and Knight, 2003). Our control experiments showed that while water treatment induced the largest signaling responses, other mechanical processes also yielded significant responses (Fig. 5). Overall the effects observed here appeared to be proportional to the degree of mechanical stimulation. Touch-responsive gene expression was first described in the literature several decades ago (Braam and Davis, 1990), but only limited information on the genes that directly regulate these very rapid gene expression responses has been published. A number of studies suggested that plasma membrane-located Ca²⁺ transporters MCA1 and MCA2 are involved in touch responses and loss-of-function mca1 mca2 mutants showed impaired mechanosensing (Nakagawa et al., 2007). An extracellular ATP (eATP) receptor AtDORN1, also on the plasma membrane, rapidly relays the presence of eATP (which is known to be released upon wounding) and triggers transcriptional responses (Choi et al., 2014), including induction of AtWRKY15, AtWRKY40, and AtOM66. Phytohormones are also thought to be involved in touch responsiveness, and plants show altered morphology and pest resistance when touched on a regular basis, a process termed thigmomorphogenesis (Chehab et al., 2012). JA and gibberellic acid (GA) are thought to be main contributors to thigmomorphogenesis, but it is unclear whether JA or GA signaling is also involved in the very early transcriptional responses (Chehab et al., 2012; Joao Pimenta Lange and Lange, 2015).

Here, we found that AtWRKY40 and AtWRKY15 may play repressive roles in the early touch response. Their transcripts are strongly induced by the touch treatment itself, suggesting that they may contribute to the rapid decline of mRNA expression between 30 and 60 min after treatment. We selected 10 of the most strongly induced touch-responsive genes, as identified by Lee et al. (2005) (Supplemental Fig. S6), for a more detailed promoter analysis, including AtOM66, DIC1, DIC2, AtWRKY40, and AtWRKY15. In line with a direct role for WRKY transcription factors in touch responses, all 10 genes contained W-boxes (TTGACY) in their 1-kb upstream promoter region (P < 0.004) using the Athena tool (O’Connor et al., 2005). Accordingly, the promoter of AtOM66 was previously demonstrated to be directly bound by AtWRKY15 and AtWRKY40 (Van Aken et al., 2013). The presence of WRKY binding sites in stress-responsive genes encoding energy organelle proteins has been described previously (Van Aken et al., 2009; Leister et al., 2011; Van Aken and Whelan, 2012). However, not all these genes exhibit touch-responsive expression patterns (e.g. AOX1a), suggesting that WRKY transcription factors play complex roles in multiple signaling cascades.

Although several of the genes induced by touch responses play important roles in biotic and abiotic stress responses, the touch response also targets energy organelle function and substrates for metabolism (Xu et al., 2006; Shang et al., 2010; Sheard et al., 2010; Zhang et al., 2014). All three DICs in Arabidopsis are highly induced by touch responses. DICs are the main transporters that allow substrates such as succinate, malate, oxaloacetate, and 2-oxoglutarate to pass the inner mitochondrial membrane, in exchange for anions including phosphate, sulfate, and arsenate (Palmieri et al., 2008). DICs can also function in exporting TCA intermediates as substrates for anaplerotic metabolism. For instance, DICs transport 2-oxoglutarate, which is important for anaplerotic pathways and nitrogen cycling, as well as malate for photorespiration and reductant exchange with chloroplasts. A metabolite analysis revealed that DIC substrates such as succinate and citrate are significantly reduced within 10 min of spray treatment, indicating that such rapid transcriptional responses occur to address a physiological change. Citrate is also known to trigger specific transcriptional response, but the target genes appear not to overlap with our findings, and the timeframe of 8 h used in these experiments is different (Finkemeier et al., 2013).

Rapid (<1 h) transient changes in organic acid contents in Arabidopsis seedlings in response to CO₂ levels have also been reported, further showing that such dynamic responses in the metabolome are possible (Dutta et al., 2009). Recent metabolic studies have suggested a role for dicarboxylates in rapid respiratory responses to Pi starvation, and DICs are indeed induced within 1 h of transfer of plants to Pi depleted medium (Lin et al., 2011; Alexova et al., 2015). Rapid transcriptional responses within 10 min have also been found in response to high-light treatment and involve chloroplast-located triose phosphate/phosphate translocator (TPT; Vogel et al., 2014). Another gene related to energy metabolism and transport that is highly touch-responsive is the plastid Glc-6-phosphate translocator GPT2. GPT2 is involved in regulating transitions in metabolic states, with plants lacking GPT2 being unable to undergo dynamic photosynthetic acclimation under naturally viable conditions (Athanasiou et al., 2010). This further underlines the importance of short-term responses in maintaining plant energy organelle function.

Taken together, these findings suggest that rapid touch and wounding responses could prepare energy metabolism for a change in substrate availability and potentially redirect nutrients and substrates from primary catabolic processes to secondary metabolism or biosynthesis. Touch-induced transcriptional responses are closely related to dark-shift responses (Lee et al., 2004), which also target AtWRKY40, DIC2, and AtOM66. Such a shift in light input could have dramatic effects on energy metabolism that need to be addressed quickly, perhaps by shifting to anaplerotic mitochondrial metabolism and usage of starch reserves, to allow sustained energy supply for the cell (Athanasiou et al., 2010). Plants thus seem to use overlapping signaling systems to respond to metabolic changes and biotic stress. Recent work showed that plants respond rapidly to even just the leaf vibrations caused by insect herbivores (Appel and Cocroft, 2014). Such a mechanical signal can travel much faster through the plant than, for example, phytohormones. From this perspective, touch, wounding, and dark-shift responses may all precede similar challenges, thus initiating overlapping defense responses that would have advantages for the plant.

CONCLUSION

In conclusion, we found that multiple independent signaling pathways operate following a wide range of treatments that involve physical manipulations, changes in light intensity, and inhibitor/elicitor treatments (Fig. 8). The first gene expression peak within 10 min after physical manipulation or dark shifts is clearly related to pathogen defense signaling and appears to be regulated at least in part by WRKY transcription factors. Our study shows that this first expression phase may also facilitate rapid readjustments in energy metabolism and energy organelle functions, as plants anticipate or respond to changes in substrate or nutrient availability very soon after a manipulation. A second signaling system occurs around 3 to 6 h after chemical chloroplast or mitochondrial inhibition and is regulated to a large extent by ANAC017. This peak affects, e.g. AOX1a, heat shock proteins, chaperones etc., and appears enriched in gene products primarily associated with repairing/maintaining the energy organelle machinery. This response may thereby prevent excessive ROS production during functional inhibition, rather than optimizing substrate availability and usage. A third response triggered by excessive light, which causes damage to the photosystems, induces overlapping genes with the second pathway. However, this high light pathway is ANAC017-independent and may at least partly be coordinated by the flu/executer and/or TPT pathways (Meskauskiene et al., 2001; Lee et al., 2007; Vogel et al., 2014). Our results extend the idea that retrograde signals affect organelle biogenesis (biogenic control) or organelle performance (operational control; Pogson et al., 2008), postulating that operational control involves separate signaling pathways for opportunity and risk (metabolic substrate availability and pathogen defense regulated, e.g. by WRKYs) and for damage repair (organelle inhibitors, high light, and ETC disruption regulated, e.g. by NACs).

Figure 8.

A working model for retrograde, high light-, and touch-responsive operational control in plants. A, High light treatment leads to chlorophyll overexcitation and reactive oxygen species (ROS) production. Signals are transduced to the nucleus possibly via flu/executer or TPT signaling, leading to retrograde transcriptional responses. B, Mechanical stress (e.g. by touch or spray application of an exogenous compound) triggers a transcriptional response within minutes. Studies suggest signaling molecules such as Ca²⁺, eATP, ROS, JA, and GA are involved. This first touch response is likely to be of key importance for stress recognition, successful pathogen resistance, morphological alterations, and rapid readjustments of energy metabolism. A second, independent signaling network caused by the active inhibitor compound (e.g. affecting mitochondria and/or chloroplasts such as antimycin A [AA] and MV) then launches a more general stress response, possibly dependent on ROS, via activation of NAC transcription factors present on the ER. NAC target genes are at least partially overlapping with high light-induced genes, but different signaling pathways are involved.

MATERIALS AND METHODS

Materials and Growth Conditions

Arabidopsis (Arabidopsis thaliana) Col-0 was used in all experiments. Seeds were sown on soil mix or MS media with 2% Suc and stratified for 2 to 3 d at 4°C, then grown under long-day conditions (16 h light/8 h dark) at 22°C and 100 µmol m⁻² s⁻¹. Previously published transgenic lines were obtained from Ng et al. (2013b) (anac017: rao2.1; anac017-2: SALK_022174), De Clercq et al. (2013) (anac013 amiRNA5.2, ANAC013 OE6, ANAC053 OE, and ANAC078 OE), Van Aken et al. (2013) (AtWRKY15 OE1, atwrky40, AtWRKY40 OE2, atwrky63, and AtWRKY63 OE1), Giraud et al. (2009) (abi4-1), Choi et al. (2014) (atdorn1.3 and OXDORN1), and Yamanaka et al. (2010) (mca1 mca2). coi1-1 was kindly provided by Dr. Louise Thatcher (CSIRO, Western Australia). We are grateful to Dr. Gary Stacey (University of Missouri), Dr. Roberto Solano (Centro Nacional de Biotecnología, Spain), Dr. Kemal Kazan (CSIRO Agriculture Flagship, Australia), and Dr. Hidetoshi Iida (Tokyo Gakugei University, Japan) for promptly sharing transgenic materials. Anac053 (SALK_009578), anac078 (SALK_025098), and ga2ox7 (SALK_055721C) were obtained from NASC. ANAC017 OE plants were generated by cloning the full-length ANAC017 coding sequence into pB7WG2 (Karimi et al., 2002) using standard Gateway (Life Technologies) cloning strategies. ANAC017, ANAC078, and ANAC053 expression levels were tested by qRT-PCR (Supplemental Fig. S2). The 1.5-kb promoter regions of ANAC013, ANAC017, ANAC053, and ANAC078 were amplified and cloned into pBGWFS7 (Karimi et al., 2002) via Gateway recombination and transformed by floral dipping as described previously (Clough and Bent, 1998; Zhang et al., 2012). At least two independent lines per construct were analyzed with comparable results. All primers used in this study are listed in Supplemental Table S2.

Stress Treatments of Plants

Seeds were sown on petri dishes containing MS media + 2% Suc, stratified for 2 to 3 d in the cold room, and then incubated in long-day growth conditions for 2 weeks. Pools of plants were then collected before or after treatment and immediately placed in liquid nitrogen for storage and further processing. For transcript analysis, plants were sprayed with mock (water + Tween), 50 µM antimycin A, 50 µM monofluoroacetate, 1 µM Flg22, 50 µM MV, or 20 µM norflurazon. For UV treatment, plants received 5 kJ/cm² UV radiation (254 nm) using a CX-2000 UV cross-linker (UVP Ltd.). In vitro stress assays were performed as previously described using 50 µM antimycin A, 100 µM rotenone, or 20 µM MV (De Clercq et al., 2013). Statistical analysis was performed using Student’s t test. Additional control experiments for touch-related treatments were as follows: (1) unopened/untreated negative control, (2) plates opened for the same duration as spray treatment but otherwise untreated, (3) plates opened and gently mechanically stimulated by touching with blunt tweezers, and (4) spray treatment with water. A similar control experiment was also performed on soil-grown plants that were (1) untreated, (2) gently mechanically stimulated with blunt tweezers, and (3) spray treated with water. For high light treatment, in vitro-grown 2-week-old plants were exposed to 750 µmol m⁻² s⁻¹ for 1 h and compared to untreated control. Furthermore, to allow differentiation between high light treatment and associated increase in temperature, a heat control set was used that was exposed to normal light (100 µmol m⁻² s⁻¹) but incubated at 28°C (the temperature that was measured in the plates during the high light treatment). For root growth assays, the different plant lines were incubated on vertically positioned plates supplemented with 50 µM antimycin A, 10 µM rotenone, and 50 µM MV. Plants stratified in the cold room for 3 d and were kept in long-day conditions for 7 d. Primary root length was measured using ImageJ. Statistical analysis was performed using Student’s t test.

Promoter-GUS Analysis

The 1.5-kb promoter regions were cloned using standard Gateway (Life Technologies) procedures into pBGWFS7 (Karimi et al., 2002). Three-week-old in vitro-grown plants were stained for GUS activity before and 6 h after spray treatment with 50 µM antimycin A as described previously (Van Aken et al., 2007).

Quantitative RT-PCR and Microarray Analysis

RNA isolation, cDNA generation, and qRT-PCR were performed as described (Van Aken et al., 2013) using Spectrum RNA Plant extraction kits (Sigma-Aldrich), iScript cDNA synthesis kit (Bio-Rad), and a Roche LC480 light cycler using SYBRgreen detection assays. All primers for qRT-PCR are shown in Supplemental Table S2. Relative expression values were normalized with untreated Col-0 samples set as 1. Statistical analyses were performed using Student’s t test throughout the manuscript, except where indicated. The E-GEOD-48676 microarray data set was obtained from ArrayExpress and analyzed using Cyber-T as previously described (Zhang et al., 2014). Changes were considered significant at false discovery rate correction level of PPDE (<P) > 0.95 and 2-fold change.

Metabolite Analysis

Approximately 20 to 30 mg of frozen tissue from pooled 2-week-old plants grown in vitro were analyzed in biological triplicate per time point after mock spray treatment using gas chromatography-mass spectrometry. Full description of metabolite analysis is available in Supplemental Methods. Data were imported and statistically analyzed in Metabolome Express (Carroll et al., 2010).

Supplemental Data

The following supplemental materials are available.

• Supplemental Figure 1. UV-induced expression of retrograde-related genes.

• Supplemental Figure 2. Confirmation of ANAC expression levels in transgenic lines, leaf respiration, and antimycin A response.

• Supplemental Figure 3. ANAC017 is not a regulator of high light- and norflurazon-responsive gene expression.

• Supplemental Figure 4. Touch-responsive gene expression.

• Supplemental Figure 5. Expression patterns of dicarboxylate carriers and HSP17.6 during antimycin A treatment.

• Supplemental Figure 6. Promoter analysis of touch-responsive genes.

• Supplemental Table 1. Wounding microarray analysis.

• Supplemental Table 2. RT-PCR primers.

• Supplemental Methods.

Glossary

MFA

monofluoroacetate

TCA

tricarboxylic acid

MV

methyl viologen

JA

jasmonic acid

eATP

extracellular ATP

GA

gibberellic acid

TPT

triose phosphate/phosphate translocator