Dissecting the interaction of photosynthetic electron transfer with mitochondrial signalling and hypoxic response in the Arabidopsis rcd1 mutant.

The Arabidopsis mutant rcd1 is tolerant to methyl viologen (MV). MV enhances the Mehler reaction, i.e. electron transfer from Photosystem I (PSI) to O2, generating reactive oxygen species (ROS) in the chloroplast. To study the MV tolerance of rcd1, we first addressed chloroplast thiol redox enzymes potentially implicated in ROS scavenging. NADPH-thioredoxin oxidoreductase type C (NTRC) was more reduced in rcd1. NTRC contributed to the photosynthetic and metabolic phenotypes of rcd1, but did not determine its MV tolerance. We next tested rcd1 for alterations in the Mehler reaction. In rcd1, but not in the wild type, the PSI-to-MV electron transfer was abolished by hypoxic atmosphere. A characteristic feature of rcd1 is constitutive expression of mitochondrial dysfunction stimulon (MDS) genes that affect mitochondrial respiration. Similarly to rcd1, in other MDS-overexpressing plants hypoxia also inhibited the PSI-to-MV electron transfer. One possible explanation is that the MDS gene products may affect the Mehler reaction by altering the availability of O2. In green tissues, this putative effect is masked by photosynthetic O2 evolution. However, O2 evolution was rapidly suppressed in MV-treated plants. Transcriptomic meta-analysis indicated that MDS gene expression is linked to hypoxic response not only under MV, but also in standard growth conditions. This article is part of the theme issue 'Retrograde signalling from endosymbiotic organelles'.

functions. However, the studies in plants with enhanced MDS gene expression, including the rcd1 mutant [3,[5][6][7], the ANAC013 overexpressor [1] and the ANAC017 overexpressor [2], have also revealed alterations in their chloroplasts. The characteristic feature of the MDS-overexpressing lines is their tolerance to methyl viologen (MV, also known as paraquat) [1][2][3]. MV catalyses electron transfer from Photosystem I (PSI) to molecular oxygen (O 2 ), referred to as the Mehler reaction [8,9]. MV promotes oxidation of PSI, modifies redox state of the photosynthetic electron transfer (PET) chain and allows the Mehler reaction to outcompete other electron fluxes downstream from PSI, including cyclic electron transfer (CET) [3,[10][11][12]. The Mehler reaction is the main source of reactive oxygen species (ROS) in illuminated chloroplasts. Accordingly, in MV-treated plants, gradual light-dependent increase in ROS production rate ultimately leads to destabilization of Photosystem II (PSII) and to cell death [3,13,14].
How the MDS gene products provide resistance to MV is unknown. One possible component of the resistance is altered redox state of chloroplast thiol enzymes. The rcd1 mutant was isolated in several genetic screens related to altered tolerance to ROS [15][16][17]. In addition, it has been identified in a genetic screen for chloroplast redox imbalance [5,6]. Recently, we discovered that the pool of the abundant chloroplast 2-Cys peroxiredoxin (2-CP) is more reduced in rcd1 than in the wild type [3]. This could contribute to MV tolerance, as the ability of 2-CP to scavenge ROS is well established [18][19][20][21]. Another possible basis for MV tolerance of rcd1 is alterations of electron transfer downstream of PSI. These alterations have been associated with the activity of mitochondrial alternative oxidases (AOXs) [3]. AOXs are encoded by MDS genes. These mitochondrial enzymes with ubiquinol:oxygen oxidoreductase activity provide an extrachloroplastic electron sink for PET [22][23][24][25]. Pharmacological or genetic inhibition of AOX activity suppresses photosynthesis [24,26], modifies PET and decreases the tolerance of plants to MV [3], but the mechanisms remain unknown.
Here, we studied the response of wild type, rcd1 and other MDS-overexpressing plants to MV to gain insight into the mechanisms whereby MDS gene products affect the chloroplast. Our results suggest that in the plants with enhanced MDS expression, the electron transfer through MV was inhibited by hypoxic environment. One of the possible explanations for this is that the interaction between the organelles may be linked to alterations in cellular oxygen availability.

(b) Chemical and hypoxic treatments
Leaf discs were placed on Milli-Q water with 0.05% Tween-20 (Sigma-Aldrich) +/− MV. Unless specified otherwise, 1 µM MV was used. Final concentration of antimycin A (AA) was 2.5 µM [3]. Dark pre-treatment with MV and AA was overnight. To generate hypoxic atmosphere, nitrogen gas was flushed inside a custom-built chamber containing plant material. Imaging was performed through the glass cover. Alternatively, plant material was placed into the AnaeroGen Compact anaerobic gas generator bag (Oxoid). AnaeroGen decreases oxygen concentration below 0.5% producing 9-13% CO 2 [32]. CO 2 accumulation was prevented by LoFloSorb non-caustic containing carbon dioxide absorbent (Intersurgical). O 2 was controlled with resazurin anaerobic indicator (Oxoid).
(c) Thiol-specific labelling of protein extracts Thiol-specific labelling of protein extracts was done and interpreted as described in [3].
(d) Feeding with 14 C glucose and analysis of metabolic fluxes 14 C glucose labelling, fractionation and analysis of metabolic fluxes were performed as described in [3]. Arabidopsis leaf discs were incubated for 150 min in light with 5 ml of 10 mM MES-KOH ( pH 6.5) containing 1.85 MBq mmol −1 [U-14 C] glucose (Hartmann Analytic) in a final concentration of 2 mM. Leaf discs of the dark experiment were incubated similarly, but under the green light. Samples were washed with distilled water, harvested and kept at −80°C for further analysis. The evolved 14 CO 2 was collected in 0.5 ml of 10% (w/v) KOH. Samples were extracted, fractioned and metabolic fluxes were analysed according to [3]. Material from frozen leaf discs was extracted in a two-step ethanolic extraction of 80% (v/v) and 50% (v/v). Supernatants were combined, dried and resuspended in 1 ml of water [33,34]. The soluble fractions were separated into neutral, anionic and basic fractions by ion-exchange chromatography as described in [34]. About 2.5 ml of the neutral fraction were freeze-dried and resuspended in 100 µl of water for further enzymatic digestions as described in [35]. Phosphate esters of the soluble fractions were measured as in [3] and starch of the insoluble fractions was measured as described in [33]. Calculation of the fluxes was performed according to the assumptions described by Geigenberger et al. [36,37].
(e) Pulse-amplitude-modulation chlorophyll fluorescence imaging Measurements of chlorophyll fluorescence kinetics and long-term PSII inhibition were performed using Walz Imaging pulseamplitude-modulation (PAM) fluorometer as described in [3].
(f ) Measurement of proton motive force by electrochromic shift Measurement of proton motive force was done as described in [38].
(g) Measurement of gas exchange by membrane inlet mass spectrometry Fourteen-mm leaf discs were floated overnight in Milli-Q water with Tween-20 ±1 µM MV at 20°C. Following the overnight incubation, 12.5 mm discs were cut from the centre of the pretreated discs in very dim light and loaded into a sealed membrane inlet mass spectrometry (MIMS) cuvette calibrated to 22°C. The cuvette was purged using air scrubbed of 12 CO 2 with carbosorb before 13  . Data processing was based on concepts and methods described by [39].

(h) Direct fast imaging of OJIP chlorophyll fluorescence kinetics
The imaging of OJIP (Fo, Fj, Fi, Fp) fluorescence kinetics was performed using FluorCam FC800F from Photon Systems Instruments, Czech Republic (www.psi.cz). The instrument is described in [40]. It contains an ultra-fast sensitive CMOS camera, TOMI 3, developed by Photon Systems Instruments, which performs image acquisition with maximum frame rate of 20 µs. FluorCam software was used to control the instrument and to analyse the data. The OJIP imaging protocol involved the triple measurement of the background signal followed by three 20 µs flashes of saturating light to measure Fo and then a 1 s saturating light pulse (3 500 µmol m −2 s −1 ) to follow OJIP kinetics. Both the background and the Fo values were averaged for further calculations. The frame period was set at 300 µs, and the integration time was 35-50 µs. The excitation light was generated by a pair of blue LED panels (470 nm) and filtered by dichroic filters that block light at 490-800 nm to avoid crosstalk with the detection. To record chlorophyll fluorescence signal, the camera was equipped with 700-to 750-nm band filters.

(i) Transcriptomic meta-analyses
Gene expression data were acquired from ArrayExpress E-MTAB-662 (rcd1) [31], E-GEOD36011 (3 h antimycin A) [41] and E-GEOD-9719 (2 h hypoxia) [42,43]. Genes that both showed at least a 2-fold change and had a statistical significance of p < 0.05 were considered as differentially expressed and were categorized as up-or down-regulated based on the direction of the change. The overlap of multiple gene lists was analysed using Venn diagrams. Pairwise Fischer's exact test was performed on the gene lists.

Results
(a) NADPH-thioredoxin oxidoreductase type C is more reduced in rcd1, but this does not explain methyl viologen tolerance of the mutant Previous study suggested that the tolerance the Arabidopsis mutant rcd1 to MV could be explained by more reduced thiol redox state of the chloroplast thiol enzymes, in particular 2-CPs, and/or by alterations in the electron transfer downstream of PSI [3]. We first addressed thiol redox enzymes. The main in vivo reductant of chloroplast 2-CPs is NADPH-thioredoxin oxidoreductase type C (NTRC) encoded by a single gene in Arabidopsis [44][45][46]. We performed thiol bond-specific labelling of NTRC and found that it was more reduced in vivo in rcd1 than in the wild type (Col-0), both in Thiol bond-specific labelling was performed as described in [3]. In brief, leaf protein extracts were treated with N-ethylmaleimide to block free thiol groups, then with DTT to reduce in vivo thiol bridges, and then with 5-kDa methoxypolyethylene glycol maleimide to label all the newly opened thiol groups. The samples were separated in SDS-PAGE and immunoblotted with the αNTRC antibody. DTT(-) control contains mainly unlabelled form. The unlabelled form (0)  To address the role of NTRC in the phenotypes of rcd1, we generated an rcd1 ntrc double mutant. Analysis of metabolic fluxes revealed that the metabolism of glucose, significantly elevated in rcd1 in light, was suppressed to wild-type levels in rcd1 ntrc (figure 1b; electronic supplementary material, table S1 for the full dataset). This hinted that alterations in the energy metabolism observed in rcd1 were partially mediated by NTRC.
In addition to reducing 2-CPs, NTRC controls many chloroplast processes, including the activity of ATP synthase [47,48] and of thylakoid NADH dehydrogenase (NDH) complex that mediates CET from PSI to the plastoquinone pool [38]. Hence, we next evaluated NDH activity in rcd1 by assessing the redox state of the plastoquinone pool in darkness. Reduced plastoquinone pool activates the chloroplast state transition kinase STN7 that phosphorylates the lightharvesting antenna complex II (LHCII). Therefore, LHCII phosphorylation may serve as an indirect measure of the plastoquinone redox state. Immunoblotting of total protein extracts from dark-acclimated seedlings with anti-phospho-threonine antibody revealed increased phosphorylation of LHCII in rcd1 that was suppressed in rcd1 ntrc (figure 1c). This suggested that NDH activity was increased in rcd1, which is likely mediated by NTRC. Thus, the alterations of chloroplast electron flows in rcd1 were partially NTRC dependent.
We next measured tolerance of rcd1 ntrc to MV-induced PSII inhibition. It was only slightly lower than in rcd1 (figure 1d). It is noteworthy that, the single ntrc mutant was more sensitive, while the NTRC overexpressor line [29] was more tolerant to MV than Col-0 (figure 1d). Thus, the activity of NTRC contributed to, but was not the main reason for the MV resistance of rcd1. An alternative explanation for MV tolerance of rcd1 suggests that it arises from changed electron transfer downstream of PSI [3]. To explore this possibility, we next addressed the effects of MV on PET in planta.

(b) Methyl viologen induces non-photochemical quenching and suppresses photosynthetic oxygen evolution
To understand how MV treatment affects the performance of photosynthetic apparatus in planta, we first measured the kinetics of chlorophyll fluorescence using the imaging PAM. The wild-type leaves were pre-treated with the catalytic amounts of MV and then exposed to low intensity light. Importantly, the measurements were performed within the first minutes of illumination, which minimizes ROS-induced irreversible damage to PSII. MV led to pronounced decrease in maximal fluorescence under light (Fm') as compared to the MVuntreated control (figure 2a). In contrast with Col-0, MV did not lower Fm' in the npq4 mutant that is deficient in non-photochemical quenching (NPQ; figure 2b,c). This suggested that the quenching effect of MV on chlorophyll fluorescence was owing to accelerated generation of NPQ. We next aimed to explore the reasons for the MVstimulated NPQ. NPQ is activated by light-dependent acidification of thylakoid lumen. To find out whether MV promoted lumen acidification, we assessed proton motive force (pmf) by measuring electrochromic shift. As expected, MV strongly inhibited thylakoid proton conductivity (gH+), resulting in elevated pmf and rapid decrease in proton flux from stroma into the lumen (vH+; figure 2d-f). Proton conductivity largely depends on the activity of thylakoid ATP synthase [49]. Taken together, these results suggested that the illumination of MV-treated plants triggered fast inhibition of thylakoid ATP synthase, which caused the rapid acidification of thylakoid lumen and increase in NPQ.
NPQ competes with charge separation in PSII, thus suppressing photosynthetic O 2 evolution. To test whether treatment with MV indeed inhibited O 2 evolution in planta, we addressed the effect of MV using MIMS. This technique n.s. allows simultaneous real-time monitoring of multiple compounds produced and absorbed through leaf gas exchange [39]. As expected, treatment with MV markedly inhibited O 2 evolution already during the first minutes of illumination (figure 2g). Importantly, the concomitant MIMS measurement of CO 2 evolution suggested that MV did not affect respiration (details in electronic supplementary material, figure S1). Thus, treatment with MV largely eliminated photosynthetic oxygen evolution but not respiration.
(c) The effect of methyl viologen on photosynthetic electron transfer is reversibly suppressed by light in rcd1 We next assayed the above-described reactions to MV in the rcd1 mutant. In MV-treated rcd1, the dynamics of oxygen evolution and chlorophyll fluorescence was identical to Col-0 during the first minutes of illumination (electronic supplementary material, figures S1, S2A, B). However, longer light treatment led to gradual recovery of Fm' in rcd1, but not in Col-0 nor in rcd1 ntrc ( figure 3). To find out whether the changes in Fm' were related to release of NPQ, we generated an rcd1 npq4 double mutant. In rcd1 npq4, the recovery of Fm' was suppressed (electronic supplementary material, figure S2C). These experiments indicated that exposure to light gradually released MV-induced NPQ in rcd1, and this process was NTRC dependent.
To test whether in rcd1 light promoted export of MV from its chloroplast site of action, we interrupted illumination with 20 min dark periods. After each dark treatment, NPQ restored and then gradually relaxed again (electronic supplementary material, figure S2D), making the possibility of MV export unlikely. Collectively, these observations suggested that electron flow through MV was reversibly inhibited by light in rcd1, but the mechanism remained unclear.
The results hinted that in illuminated rcd1 chloroplasts, MV displays different physiological activities from in Col-0 and this is not owing to removal of MV from its site of action. MV tolerance of rcd1 has previously been linked to altered mitochondrial respiration [3]. Both mitochondrial respiration and the electron flow through MV depend on molecular oxygen. MV inhibited photosynthetic O 2 evolution, but not respiration (figure 2g; electronic supplementary material, figure S1). This raised the question whether MV treatment of rcd1 could affect the availability of O 2 inside the leaf, prompting us to study the effects of MV in hypoxic environment.

(d) The methyl viologen-dependent response in rcd1 is eliminated in hypoxic environment
We thus exposed leaf discs to hypoxic atmosphere by flushing nitrogen gas for 20 min in darkness and monitored PET during 5 min of low intensity light as in figure 2a-c. In all the tested lines, hypoxia led to increased fluorescence, as compared to the aerobic controls (electronic supplementary material, figure S3A). This was anticipated, since O 2 acts as an electron sink for several chloroplast processes including the Mehler reaction and activity of the chloroplast terminal oxidase PTOX. In Col-0, MV markedly diminished the hypoxia-related rise in chlorophyll fluorescence (figure 4a; electronic supplementary material, figure S3A). This was likely owing to the fact that MV catalysed the Mehler reaction, thus compensating for oxygen deficiency. Importantly, the same effect of MV was observed in the ptox mutant, indicating that it was not associated with the PTOX activity. Similarly, MV lowered chlorophyll fluorescence in hypoxia-treated npq4 and stn7, suggesting that the shift was not owing to NPQ or chloroplast state transitions. In striking contrast to all of the above plant lines, in rcd1, MV did not lower chlorophyll fluorescence under hypoxic conditions (figure 4a; electronic supplementary material, figure S3A). This implies that in rcd1, but not in other lines, hypoxic environment compromised the electron flow through MV.  This effect has previously been ascribed to the oxidative action of MV on the electron-acceptor side of PSI [10,11,53]. It has been proposed that MV releases 'a traffic jam of electrons caused by a transient block at the acceptor side of PSI' [10], thereby lowering fluorescence. In hypoxic conditions, the effect of MV on the Fi-Fm phase was still detected in Col-0; however, it disappeared in rcd1 (figure 4b). The Fi-Fm phase can be expressed by the parameter wRE1o = 1 -Fi/Fm, called the quantum yield of the electron flux to PSI electron acceptors [52]. Accordingly, in aerobic conditions MV affected wRE1o both in Col-0 and in rcd1, while in hypoxic conditions, the effect of MV was only observed in Col-0, but absent from rcd1 (figure 4c). Importantly, in an analogous dark hypoxic assay stn7 performed similarly to Col-0, while rcd1 ntrc was indistinguishable from rcd1 (electronic supplementary material, figure  S3B, C). This indicated that the differential response to MV was most probably not owing to dark anaerobic-induced state transitions [50] nor to NTRC-dependent CET. The disappearance of MV activity in rcd1 as early as after 40 ms of illumination strongly indicated that hypoxia counteracted electron flow through MV directly at the electron-acceptor side of PSI, and not through the changes outside of the chloroplast. Fast change in OJIP kinetics induced by nitrogen gas flush made it difficult to observe the transition of PET from the aerobic to the hypoxic state and to address the stability of the difference between the genotypes. Thus, we generated hypoxia with an alternative approach, using the AnaeroGen anaerobic gas generator, and imaged OJIP kinetics once in 30 min over 12 h of darkness. The raw OJIP kinetics are presented in electronic supplementary material figure S3D and the calculated wRE1o parameter in figure 4d. The treatment resulted in similar changes of OJIP as those induced by nitrogen gas (figure 4b,c). In MV-treated rcd1, hypoxic atmosphere restored wRE1o to the levels observed in MV-untreated controls, while this did not happen in MV-treated Col-0 (figure 4d). The difference between the genotypes remained stable during several hours of incubation, suggesting that it was not owing to possible alterations in gas exchange rates or different starting O 2 reserves in Col-0 and rcd1. Collectively these experiments implied that hypoxic environment rapidly blocked the PSI-MV electron transfer in rcd1, but not in Col-0.

(f ) Increased expression of mitochondrial dysfunction stimulon genes is linked to hypoxic response
The results indicated that the PSI-MV electron transfer was suppressed in rcd1 during illumination or under hypoxic environment. The rcd1 mutant is characterized by increased expression of the MDS genes, which results in altered mitochondrial respiration [3]. We thus assessed whether the altered response to MV under hypoxia observed in rcd1 exists in other MDS-inducing lines or treatments. The mitochondrial   figure S4). Measurement of OJIP kinetics in plants overexpressing ANAC013 [1,2] revealed the suppression of PSI-MV electron transfer by hypoxia similar to that in rcd1 (figure 5b). These observations demonstrated that oxygen availability affected the MV response not only in rcd1, but also under other perturbations activating MDS gene expression. Thus, the studied effects are not a peculiarity of rcd1, but rather a common feature associated with MDS overexpression.
Mitochondrial AOXs were proposed to be implicated in MV tolerance [3]. To test the contribution of the individual AOX isoforms, we performed analogous experiments in aox1a, aox1c and aox1d knockout mutants and in the AOX1a overexpressor line [30]. All of them showed wild-type response to hypoxia (electronic supplementary material, figure S5A). Similarly, the rcd1 aox1a double mutant [31] was indistinguishable from rcd1 (electronic supplementary material, figure S5B). Thus, the studied effect on chloroplast electron transfer could not be attributed to a single AOX isoform.
Taken together, the results demonstrated altered response to the hypoxia of the MDS-overexpressing plants. Interestingly, transcriptomic changes occurring in plants exposed to hypoxia are similar to the changes triggered by AA [54]. Thus, the two types of stress are also linked in physiological conditions unrelated to MV. To find out whether similar changes in gene expression occurred in rcd1, we compared publicly available transcriptomic datasets obtained in the rcd1 mutant and in Col-0 treated with AA or hypoxia. The lists of genes with changed expression under these three perturbations showed a statistically significant overlap (figure 5c,d). The 19 genes that were activated in all the perturbations included the hypoxia-responsive universal stress protein 1 (HRU1), the stress-responsive transcription factor ZAT10, transcription factor WRKY25 and the MDS genes AOX1a and SOT12. These results implied that transcriptional reprogramming induced by hypoxia bears similarity to the changes in gene expression triggered by MDS signalling and to those observed in rcd1 under standard growth conditions (figure 5c,d; electronic supplementary material, table S3).

Discussion
Expression of the MDS genes is related to mitochondrial retrograde signalling, but it also affects chloroplast functions. One prominent feature of MDS-overexpressing plants is their tolerance to MV [1][2][3]. The Arabidopsis mutant rcd1 that among several distinct phenotypes [55] overexpresses MDS genes and is MV tolerant [3,31] provides an experimental tool to dissect the interaction between the organelles. MV tolerance of rcd1 was suggested to be owing to altered redox state of chloroplast thiol enzymes, and/or modified electron transfer pathways downstream of PSI [3]. Knockout of the master royalsocietypublishing.org/journal/rstb Phil. Trans. R. Soc. B 375: 20190413 regulatory chloroplast thiol enzyme NTRC in the rcd1 background largely suppressed rcd1 phenotypes related to PET (figures 1 and 3). However, the long-term tolerance of rcd1 ntrc to MV-induced PSII inhibition was only slightly lower than in rcd1 (figure 1d). These observations imply that the more reduced states of thiol enzymes are not the primary reason for MV tolerance of rcd1. Several recent studies have suggested that ROS generated by the Mehler reaction are the main electron sink for chloroplast 2-CP and its reductants, including NTRC [19][20][21]. Thus, rcd1-specific alternations of the Mehler reaction may underlie both its MV tolerance and more reduced state of 2-CP [3] and NTRC ( figure 1). Consequently, the changes in the activities of NTRC and its targets affect PET, including operation of the NDH complex (figure 1) and possibly of the malate shuttle [3]. Thus, we suggest that the following chain of events could be responsible for the control of chloroplast processes by MDS gene products. The signals initiated in the mitochondria induce expression of MDS genes. This affects electron transfer at the electron-acceptor side of PSI suppressing the Mehler reaction and ROS production in the chloroplast. Lower ROS production leads to more reduced chloroplast thiol enzymes, which alters photosynthesis.
The question of how MDS gene products influence the chloroplast Mehler reaction remained unanswered. In search for mediators linking the two organelles, we discovered that hypoxic treatment in darkness caused rapid inactivation of PSI-MV electron transfer in rcd1, but not in Col-0 (figure 4). One possible explanation for this is that in rcd1 unknown readjustments of PET disrupt the access of MV to PSI. Such readjustments would likely be post-translational, as the change occurred during only 20 min of hypoxic treatment. The changes were not associated with dark hypoxia-induced state transitions [50] because the stn7 mutant, deficient in state transitions, performed in a similar way to Col-0 that had functional state transitions [50] (electronic supplementary material, figure S3B). The effect was also not owing to the NTRC-dependent CET, as rcd1 ntrc performed similarly to rcd1 (electronic supplementary material, figure S3C). Further research into chloroplast electron flows in rcd1 and, more generally, in MDS-overexpressing plants, is required to address the implication of other possible CET pathways in MV tolerance.
As another possibility, the rcd1-specific response could rely on some other pre-existing adaptations. Several comprehensive studies in rcd1 did not reveal chloroplast defects that could unambiguously explain its MV tolerance [3,6,7]. By contrast, defects have been found in the rcd1 mitochondria. The mutant has increased respiration metabolic flux (electronic supplementary material, table S1) and higher AOX capacity [3]. AOX activity is implicated in the control of PET [3,[22][23][24][25] and in tolerance to MV [3]. Thus, AOXs and/or other mitochondrial components could be involved in the rcd1-specific reaction to the hypoxic treatment in darkness.
AOX respiration consumes reducing power and molecular oxygen [22]. Hence, one possible mechanism connecting AOX activity to the Mehler reaction is electron transfer between the organelles. Photorespiration [26] and malate shuttle [3] have been proposed to mediate this pathway. Importantly, however, the hypoxic treatments described in figure 4b-d were performed in darkness, ruling out the involvement of photorespiration, NADPH-MDH-dependent malate shuttle and NADP-GAPDH-dependent glyceraldehyde 3-phosphate shuttle, as these pathways are light dependent [56].
Alternatively, AOXs and/or other MDS components could affect the response to MV through decreasing cellular availability of O 2 . AOXs have previously been proposed to act as oxygen sink in plant mitochondria [57,58] and nonphotosynthetic tissues [59,60]. In photosynthesizing plants, this effect is certainly masked by O 2 evolution. However, the results described in figure 4b-d were obtained in dark-acclimated plants using a 1 s flash of saturating light. Hence, photosynthetic O 2 evolution was inactive. In darkness, increased respiratory activity possibly makes rcd1 plants more dependent on the uptake of atmospheric O 2 through stomata. Therefore, disrupting external O 2 supply could lead to higher O 2 deficit in rcd1 compared to Col-0. This could explain the suppressed PSI-MV electron transfer in rcd1 ( figure 4).
It remains unclear how the effects observed in rcd1 in darkness under hypoxic conditions are related to the mutant's long-term MV tolerance (e.g. figure 1d), or to specific lightdependent alterations of PET (figure 3), occurring in rcd1 under aerobic atmosphere. It is worth pointing out that according to our results the treatment with MV in light rapidly suppressed photosynthetic O 2 evolution in both rcd1 and Col-0, while respiration remained unaffected (figure 2 and electronic supplementary material, figure S1). Besides, MV has been shown to promote stomatal closure [61], which could limit the supply of atmospheric O 2 . This hints that the treatment of plants with MV can be associated with O 2 deficiency in photosynthesizing tissues. However, the direct assessments of in vivo O 2 tissue concentrations are needed to test this assumption. Transcriptomic meta-analysis revealed similarities of gene expression changes induced by AA, in the rcd1 mutant and by hypoxia ( figure 5; electronic supplementary material, table S3). This suggests that the expression of MDS genes is linked to hypoxic response not only in artificial physiological situations such as MV treatment, but also in standard growth conditions. Whether this similarity is related to altered cellular O 2 availability is yet to be determined. Our observations open up new experimental possibilities to explore the mechanisms of interactions between the mitochondria and the chloroplasts.
Data accessibility. Data are presented as electronic supplementary material.