In higher plants, biosynthesis of cysteine is catalysed by OAS-TL [O-acetylserine(thiol)lyase], which replaces the activated acetyl group of O-acetylserine with sulfide. The enzyme is present in cytosol, plastids and mitochondria of plant cells. The sole knockout of mitochondrial OAS-TL activity (oastlC) leads to significant reduction of growth in Arabidopsis thaliana. The reason for this phenotype is still enigmatic, since mitochondrial OAS-TL accounts only for approximately 5% of total OAS-TL activity. In the present study we demonstrate that sulfide specifically intoxicates Complex IV activity, but not electron transport through Complexes II and III in isolated mitochondria of oastlC plants. Loss of mitochondrial OAS-TL activity resulted in significant inhibition of dark respiration under certain developmental conditions. The abundance of mitochondrially encoded proteins and Fe–S cluster-containing proteins was not affected in oastlC. Furthermore, oastlC seedlings were insensitive to cyanide, which is detoxified by β-cyano-alanine synthase in mitochondria at the expense of cysteine. These results indicate that in situ biosynthesis of cysteine in mitochondria is not mandatory for translation, Fe–S cluster assembly and cyanide detoxification. Finally, we uncover an OAS-TL-independent detoxification system for sulfide in mitochondria of Arabidopsis that allows oastlC plants to cope with high sulfide levels caused by abiotic stresses.
Biosynthesis of cysteine in higher plants is catalysed by OAS-TL [O-acetylserine(thiol)lyase; EC 220.127.116.11). It transfers sulfide to the activated hydroxy group of OAS (O-acetylserine), whereby cysteine and acetate are formed. The acceptor of sulfide, OAS, is synthesized by SAT (serine acetyltransferase; EC 18.104.22.168) from serine and acetyl-CoA. SAT and OAS-TL are encoded by small gene families in Arabidopsis thaliana and form the CSC (cysteine synthase complex) that has a regulatory function for the rate of cysteine synthesis depending on the ambient concentrations of OAS and sulfide [1,2]. Isoforms of both proteins and thus the CSC are present in the cytosol, plastids and mitochondria in Arabidopsis [3,4], allowing the production of OAS and cysteine in these subcellular compartments. The cellular distribution of SAT and OAS-TL activities has been suggested to reflect the need for cysteine production in every subcellular compartment in which protein synthesis takes place . However, the viability of T-DNA (transferred DNA) insertion mutants lacking functional OAS-TL proteins in the cytosol, plastids and mitochondria (oastlA, oastlB and oastlC) respectively, strongly indicated that cysteine can be exchanged between the cytosol and organelles. Remarkably, only the oastlC mutant showed a significant retardation in growth (25% retardation in comparison with the wild-type), although it contributes to only 5–10% of total OAS-TL activity of the cell .
Since sulfate reduction is localized in plastids, the question is, how do the cytosolic and mitochondrial OAS-TL obtain sulfide for incorporation into cysteine? The mode of sulfide transport through the plastidic membrane system for sufficient sulfide supply in cytosol and mitochondria is still unknown. Mathai et al.  showed that sulfide can penetrate membranes easily without any facilitator protein, which questions an efficient control of sulfide release from plastids by a specific transporter system. Sulfide rather appears to be released from the plastid in accordance with its concentration gradient. Other possible sources of sulfide in mitochondria include degradation of sulfur compounds, turnover of Fe–S clusters, and detoxification of cyanide by CAS (β-cyano-alanine synthase). The high reactivity of sulfide in combination with its potentially uncontrolled distribution within plant cells raises the question of its toxic potential in plants. Indeed, plant growth may be negatively affected by H2S fumigation , although many plant species tolerate quite high levels of sulfide (up to 0.5 p.p.m. H2S) with no or only minor retardation in growth. Fumigation of plants with H2S is always accompanied by a massive accumulation of soluble thiols, such as cysteine and glutathione [7,8], and cysteine synthesis may be involved in the detoxification of excessive sulfide in plants.
In contrast with plants, animals are not able to fix sulfide in the form of cysteine. Nevertheless, sulfide is produced in animals as a by-product of cysteine catabolism . It is well established that sulfide acts as a toxin in animals in three major loci: the cardiovascular system, the central nervous system and energy metabolism. In the latter, sulfide toxicity is based on inhibition of COX (cytochrome c oxidase; Complex IV) in the ETC (electron transport chain) of mitochondria [10,11].
Cysteine produced in mitochondria could be used directly for protein synthesis in this compartment. In addition, cysteine is the substrate for NFS1, the mitochondrial cysteine desulfurase. This enzyme provides persulfide for the synthesis of Fe–S clusters , and also for biotin and lipoate. Fe–S clusters are abundant in Complex I and Complex II (succinate dehydrogenase) of the mitochondrial ETC and aconitase. Another consumer of cysteine in mitochondria is CAS (AtCysC1; EC 22.214.171.124 ). The enzyme takes cysteine as the acceptor for detoxification of cyanide, a further potent inhibitor of COX, and releases sulfide.
The purpose of the present study was to elucidate the relevance of mitochondrial cysteine synthesis for mitochondria. Therefore the impact of cysteine synthesized in mitochondria for Fe–S protein biogenesis, cyanide detoxification and mitochondrial translation were assessed. Furthermore, direct evidence for the toxicity of sulfide on COX and the existence of an OAS-TL-independent sulfide detoxification system is provided using a combination of a reverse genetic approach and biochemical analyses of isolated mitochondria.
Plant genotypes, growth conditions and sample preparation
All experiments were performed using A. thaliana, ecotype Columbia-0 as the wild-type control, and a mitochondrial OAS-TL C mutant (oastlC, At3g59760) in the same background .
Plants for fumigation experiments were germinated and grown on soil under short day conditions according to , but five instead of single plants were transferred to a single pot 2 weeks after germination. The 3-week-old plants were then transferred into fumigation cabinets to allow acclimatization to constant temperature (18°C) during the day and night. Sulfide (1 p.p.m. H2S) was exposed to the plants for 2 weeks according to . The fresh weight of the rosette leaves of each individual plant was determined. For determination of metabolites and extraction of RNA, leaf material of 5-week-old plants (n=25) for each genotype and treatment was pooled and immediately frozen in liquid nitrogen. The resulting samples were ground in liquid nitrogen to a fine powder and stored at −80°C.
Callus from root tissue was obtained by growing wild-type and oastlC plants for 10 days on agar dishes containing 0.5× MS medium (Duchefa) in a vertical position. The roots of the seedlings were cut off and transferred to callus induction plates [1× Gamborg's B5 medium including vitamins, 1% (w/v) glucose, 0.05% Mes, 0.5 μg/ml 2,4-dichlorophenoxyacetic acid, 5 μg/ml kinetin and 0.8% microagar, pH 7.5]. The roots were wounded with a scalpel and the plates were incubated under short day conditions in the dark for callus induction. The calli were transferred every second week to freshly prepared plates. For isolation of mitochondria from seedlings, wild-type and oastlC were grown for 2 weeks in liquid medium containing 1% (w/v) sucrose as described previously . Total respiration rates were determined from rosette leaves of 5-week-old soil grown plants and 2-week-old seedlings grown in liquid medium or on soil.
Cyanide sensitivity was tested using seedlings grown in dishes on AT medium  under short day conditions in a vertical position. Seedlings (1-week-old) were transferred from AT medium to AT medium supplemented with 0, 0.1 and 1 mM KCN and root growth was monitored for 2 weeks.
Isolation of mitochondria from callus and seedlings
Callus material (2-week-old; 0.5 g) was ground in 5 ml of extraction buffer [0.03 M Mops, 0.3 M mannitol, 1 mM EDTA, 0.1% BSA, 0.6% PVP-40 (polyvinylpyrrolidone 40) and 4 mM cysteine, pH 7.5] and mitochondria were purified according to . Mitochondria from 2-week-old seedlings were isolated as described previously . The protein concentration was determined with Roti-Quant™ (Roth) using BSA as a reference.
Enzyme activity assays
COX activity (Complex IV) and intactness of purified mitochondria (corresponding to 0.2 mg of total protein) were determined according to  with a Clark-type oxygen electrode (Hansatech Instruments) in a total volume of 1 ml containing 0.3 M mannitol, 10 mM Tes, 3 mM MgSO4, 10 mM NaCl, 5 mM KH2PO4, 0.1% BSA, 1 mM NADH, 5 mM sodium ascorbate and 0.02% cytochrome c. The intactness of mitochondria varied between 60 and 80%.
Inhibition constants (IC50) of COX for sulfide and cyanide were determined after solubilization of mitochondria (38 μg of mitochondrial protein) with Triton X-100. Concentrations of sulfide and cyanide used in the COX activity assay were between 0 and 0.5 μM (n=3). The data were fitted according to the equation v=Vmax[I]/IC50+[I] (where I is inhibitor) to determine the IC50 value.
Complex II activity, coupled to complex III, was determined photometrically (ϵ550 of cytochrome c: 20 mM−1·cm−1) according to  using 0.1 mg of total mitochondrial protein.
Aconitase activity was measured using 20 μg of total mitochondrial protein in a reaction volume of 1 ml containing 100 mM Tris, 1.5 mM MgCl2, 0.1% Triton X-100, 1 mM NADP and 1 unit of isocitrate dehydrogenase as described previously . The reaction was started by addition of 150 μM cis-aconitate (ϵ340 of NADP: 6.2 mM−1·cm−1).
Sulfide detoxification capacity of mitochondria
Total protein of mitochondria isolated from wild-type and oastlC seedlings grown in liquid medium (108 and 78 μg respectively) was diluted in 0.1 ml of 10 mM Tris, pH 8.0, and incubated with 0.1 mM sulfide for up to 180 min at room temperature (22°C). Samples were taken at 0, 60, 120 and 180 min for determination of the residual sulfide content with the Methylene Blue test . In order to determine the non-enzymatic evaporation rate of sulfide during the incubation period, an aliquot was kept on ice to inhibit enzymatic detoxification of sulfide in the presence and absence of mitochondria. Quantification of sulfide was based on an external calibration curve (0–30 nmol of sulfide).
Dark respiration of intact tissues
Dark respiration, as determined by O2 consumption of rosette leaves from 5-week-old plants and 2-week-old seedlings grown in liquid medium or on soil, was determined in a total volume of 1.5–2 ml containing 15 mM Tes and 0.2 mM CaCl2, pH 7, using a Clark-type oxygen electrode (Hansatech Instruments) in the dark. Cyanide-resistant O2 consumption was determined upon the addition of 2.5 mM potassium cyanide. Activity of COX could be determined as the difference between respiration rate in the absence and presence of cyanide.
Sulfide and sulfite were captured during extraction of leaf samples (20 mg) for 15 min with the fluorescent dye monobromobimane (3 mM), which was dissolved in 0.16 ml of extraction buffer (160 mM Hepes, pH 8.0, and 16 mM EDTA), and quantified after separation of the bimane conjugate by reverse-phase chromatography using an external calibration curve . Cysteine was determined upon labelling with monobromobimane according to  using the same HPLC system. Quantification of ADP and ATP was performed as described previously .
Immunological detection of proteins
Purified mitochondrial protein (15 μg) from wild-type and oastlC were separated by SDS/PAGE and transferred on to nitrocellulose membrane, which was subsequently blocked with 0.5% non-fat dried skimmed milk powder in PBS-T (3.2 mM Na2HPO4, 0.5 mM KH2PO4, 1.3 mM KCl, 135 mM NaCl and 0.05% Tween 20, pH 7.4) or TBS-T (20 mM Tris/HCl, 500 mM NaCl and 0.05% Tween 20, pH 7.5). For immunological detection of Cox2 (COX subunit 2), the primary antibody (AS04053A; Agrisera) was diluted 1:2000 in TBS-T with 5% (w/v) non-fat dried skimmed milk powder and used in combination with an horseradish peroxidase-conjugated secondary antibody and chemiluminescent detection. Anti-OAS-TL C antibody was diluted 1:1000 , anti-aconitase isoform 1 (AS09521, Agrisera) and anti-Nad9 (NADH dehydrogenase subunit 9)  antibodies were diluted 1:10000. Loading of protein and transfer efficiency were tested by staining the gel after transfer with Instant Blue (Expedeon), and the nitrocellulose blot with Ponceau S.
qRT-PCR (quantitative real-time PCR)
Expression of AtAOX1a (At3g22370) and AtSO (At3g01910) was analysed in rosette leaves (three aliquots from a pool of 20–25 plants) and seedlings grown on soil (n=3, 20 seedlings each). Total RNA was extracted using the RNeasy® Kit (Qiagen). Total RNA (4 μg) was reverse-transcribed with the RevertAid™ H Minus First Strand cDNA synthesis kit (Fermentas) and oligo(dT)18 primers. qRT-PCR was performed in a Rotor-Gene Q cycler (Qiagen) using the SensiMix™ SYBR No-ROX Kit (Bioline) with the primers given in Supplementary Table S1 (at http://www.BiochemJ.org/bj/445/bj4450275add.htm).
Sulfide is a potent inhibitor of plant COX
The susceptibility of plants towards H2S varies strongly between species . Since the potential main target of H2S toxicity in plants has not so far been determined, the activity of COX was investigated after purification of mitochondria from seedlings of wild-type plants grown in liquid culture (Figure 1A). For comparison, the known inhibitory effect of cyanide on plant COX was also determined in this experimental setup (Figure 1B). COX activity, as determined by oxygen consumption of 38 μg of mitochondrial protein, was inhibited by sulfide with a 50% inhibition constant (IC50) of 6.9 nM sulfide. The amount of mitochondrial protein was the lower limit for our COX activity assay, but may be not low enough to exclude the fact that the IC50 is dependent on the protein amount (see inset of Figure 1A), indicating that the true IC50 for sulfide could be even lower than 6.9 nM sulfide. In contrast, the IC50 for cyanide was determined as 7 nM cyanide and found to be independent of the amount of mitochondrial protein at low protein concentrations (inset of Figure 1B). Sulfide is therefore a comparable or probably even stronger inhibitor of plant COX than cyanide, which is also reported for the animal COX proteins . The IC50 of plant COX for sulfide is at least 25-fold lower than the IC50 of purified animal COX proteins for sulfide (IC50=200 nM sulfide ). Thus inhibition of COX activity by sulfide is a potential target of sulfide toxicity in plants.
Inhibition of COX activity by sulfide and cyanide
COX is inhibited in mitochondria of oastlC mutants
The oastlC mutant constitutively lacks the ability to produce cysteine in the mitochondria, which could possibly result in the accumulation of the substrates sulfide and OAS. The significant inhibition of COX by sulfide in isolated mitochondria of wild-type plants prompted us to test the COX activity in the oastlC mutant. Mitochondria from in vitro grown seedlings and from callus were isolated and assayed for COX activity (Figures 2A and 2B). COX activity in the mitochondria of oastlC was consistently only approximately half of the activity in wild-type, independent of the tissue used for the preparation of mitochondria. To investigate whether the decrease in COX activity is due to lower Complex IV protein levels, we assayed the levels of Cox2 by Western blot analysis. Although Cox2 levels were decreased in mitochondria from oastlC seedlings in the results shown (Figure 2F), this was not consistently observed, and it was also not observed in oastlC callus mitochondria. We therefore conclude that the 50% lower COX activity is not due to decreased Complex IV levels, but that some degradation of the protein may occur upon inactivation.
Impact of cysteine synthesis on respiration and Fe–S enzymes in mitochondria
Assembly of Fe–S clusters and mitochondrially encoded Nad9 levels are not affected in oastlC mutants
As cysteine is required for the synthesis of mitochondrially encoded proteins and for Fe–S cluster assembly, we investigated the enzymatic activity and abundance of selected proteins. The activity of the Fe–S cluster containing Complexes II and III (Figures 2C and 2D) was not affected in mitochondria of oastlC prepared from callus or seedlings. Similarly, activity (Figure 2E) and abundance (Figure 2F) of mitochondrial aconitase, a [4Fe–4S] cluster-containing enzyme, were found to be unaffected in the oastlC mutant. These results indicate sufficient supply of cysteine for the formation of mitochondrial Fe–S cluster-containing enzymes in oastlC. Immunological detection of Nad9 (ATMG00070) and Cox2 indicated that the level of proteins encoded by the chondriome was not affected because of the lack of mitochondrial cysteine synthesis in oastlC (Figure 2F). These results are in agreement with the unaffected foliar cysteine steady-state level of oastlC compared with wild-type .
oastlC is not sensitive to cyanide
Cyanide and sulfide efficiently inhibit COX activity of isolated mitochondria (Figure 1). Cyanide is assumed to be detoxified in mitochondria of Arabidopsis via CAS (AtCysC1, At3g61440) which transfers cyanide to cysteine . Mitochondrial cysteine levels are assumed to be low in the oastlC mutant. In order to rule out that inhibition of COX is a result of impaired detoxification of cyanide in oastlC, wild-type and oastlC mutants were challenged with cyanide. Growth of wild-type seedlings was not affected at cyanide concentrations of up to 0.1 mM, but displayed a significant decrease of root growth in response to 1 mM cyanide (Figure 3). The oastlC seedlings showed the same sensitivity to cyanide compared with the wild-type at sub-effective and effective concentrations of cyanide. The result strongly indicates that sufficient levels of cysteine were present in mitochondria of oastlC to serve as a substrate for detoxification of cyanide, and that inhibition of COX in oastlC is not a result of higher sensitivity to cyanide.
Effect of cyanide on root growth of wild-type and oastlC plants
Dark respiration is lower in young, but not in adult, oastlC plants
The inhibition of COX activity in isolated mitochondria prompted us to test dark respiration rates of intact seedlings grown hydroponically (Figure 4A) or on soil (Figure 4B). Dark respiration was approximately 2-fold higher in hydroponic seedlings (grown with sucrose in the medium) compared with soil-grown seedlings. Nevertheless, dark respiration was significantly lower in oastlC seedlings in comparison with the wild-type at both growth conditions. This decrease was owing to reduced cyanide-sensitive respiration (COX activity), since cyanide-resistant oxygen consumption was similar in oastlC and wild-type. This result confirms the inhibition of COX activity measured in isolated mitochondria of oastlC. However, dark respiration rates of leaves from 5-week-old oastlC and wild-type plants were identical (Figure 4C). This finding argues for an inhibition of COX by sulfide in young plants, but against a permanent inactivation of COX during the entire life cycle. The significant decrease of COX in seedlings could result in an induction of alternative oxidase (AtAOX1a) expression as a terminal electron acceptor . Neither in seedlings nor in leaves of oastlC was an up-regulation of AtAOX1a transcript level observed (Figure 4D). The marginally decreased AtAOX1a transcription in leaves of oastlC compared with wild-type is not considered to be functionally relevant.
Developmental and nutritional status affects dark respiration in wild-type and oastlC
Mitochondria lacking OAS-TL activity can scavenge sulfide
To investigate the fate of sulfide in oastlC mitochondria, an in vitro assay was developed to monitor sulfide detoxification. Mitochondria were isolated from seedlings and incubated with exogenous Na2S. The amount of sulfide was monitored over the course of 180 min using the Methylene Blue reaction (Figure 5A). Mitochondria cooled to 4°C consumed very little sulfide (Figure 5A), demonstrating that the observed sulfide removal is enzyme catalysed. Evaporation of volatile sulfide (H2S) contributed only insignificantly to the decrease of sulfide in comparison with the enzyme-catalysed removal (Figure 5A, buffer control). Mitochondria isolated from seedlings of oastlC plants were equally able to consume sulfide, which hints at the existence of a so far unknown sulfide detoxification system in the mitochondria of plants (Figure 5B). The sulfide detoxification capability of oastlC mitochondria, prepared from callus material, was even higher than in mitochondria isolated from wild-type callus (Supplementary Figure S1 at http://www.BiochemJ.org/bj/445/bj4450275add.htm), suggesting that the unknown mechanism is induced and important in oastlC, at least under conditions of the in vitro test. Taking into consideration that COX activity in leaves of mature oastlC plants was not inhibited, these results strongly support the existence of an OAS-TL-independent detoxification system for sulfide in mitochondria of Arabidopsis.
Sulfide detoxification capacity of isolated mitochondria from wild-type and oastlC mutants
Fumigation with H2S reveals sufficient sulfide detoxification ability in oastlC
The fumigation of plants with sub-lethal concentrations of H2S for prolonged times is an established tool to investigate processes related to sulfur metabolism [7,8]. To ultimately prove the significance of sulfide detoxification via formation of cysteine in mitochondria, wild-type and oastlC plants were challenged with a high dose of sulfide (1 p.p.m. H2S). When the biomass of the aerial parts of plants were determined after 2 weeks of exposure, the growth was lower in both genotypes by approximately 50% compared with controls without H2S. oastlC mutants were affected to the same degree as the wild-type (Figure 6A). Concentrations of endogenous sulfide in the fumigated plants increased approximately 2–3-fold in both genotypes. The increase and absolute level of sulfide was significantly lower in oastlC compared with wild-type after fumigation with H2S, suggesting that the lack of oastlC had prompted the induction of the OAS-TL-independent sulfide-scavenging activity (Figure 6B). The high dose of sulfide caused a 30-fold accumulation of cysteine in the wild-type, indicating a high flux of sulfide into the cysteine synthesis pathway. The oastlC mutant had only a minor decrease in the overall capacity of cysteine synthesis , nevertheless, cysteine accumulated to only half of the wild-type cysteine level upon sulfide fumigation (Figure 6C). This lowered increase of cysteine could be explained by induction of the OAS-TL-independent mitochondrial sulfide detoxification system in oastlC that competes with plastidic and cytosolic cysteine synthesis for sulfide. Fumigation of both genotypes with sulfide caused a significant increase in the ADP/ATP ratio, which could be a consequence of inhibition of COX by sulfide or a pleiotrophic response to excess sulfide stress. However, the increase of the ADP/ATP ratio after sulfide fumigation was significantly lower in oastlC in comparison with the wild-type (Figure 6D), which is in line with the proposed protecting function of an OAS-TL-independent sulfide detoxification system. The alternative sulfide-scavenging capacity was discovered using isolated mitochondria (Figure 5). However, SO (sulfite oxidase) in peroxisomes could contribute to sulfide detoxification in oastlC at the cellular level, since sulfide can be spontaneously oxidized to sulfite. The latter can then be further oxidized by SO to non-hazardous sulfate. Neither sulfite nor AtSO expression was altered in oastlC seedlings or leaves of 5-week-old plants (Supplementary Figure S2 at http://www.BiochemJ.org/bj/445/bj4450275add.htm), questioning a significant contribution of SO to sulfide detoxification in oastlC.
Effect of H2S exposure on biomass and metabolite levels of wild-type and oastlC mutants
Toxicity of sulfide in mammals is based on reversible binding of H2S to oxidized COX in the ETC of mitochondria [10,11]. The protection from free sulfide involves an enzyme-catalysed oxidation sequence in mitochondria that starts with membrane-bound sulfide:quinone oxidoreductase and produces thiosulfate . This enzyme is not present in plants. However, sulfide is also toxic for higher plants , but must be produced in significant amounts for synthesis of cysteine, which is the precursor for all reduced sulfur-containing metabolites in photoautotrophic organisms. In the present study, we identified plant COX as one target of H2S toxicity in plants. The IC50 of Arabidopsis COX for sulfide is lower than that for COX proteins from animals [10,11,26,29], indicating an efficient inhibition of COX-dependent respiration even when low concentrations of sulfide are present in plant mitochondria. The concentration of sulfide in plant mitochondria is currently not known. Krüger et al.  calculated the plastidic, cytosolic and vacuolar sulfide concentrations to be 125, 55 and 5 μM respectively, after non-aqueous fractionation of Arabidopsis leaf material. The mitochondria were not resolved from the vacuole and the cytosol in this experiment. Nevertheless, the results clearly indicated a gradient of sulfide within the plant cell with the highest concentration in the plastids and significantly lower concentrations in the remaining sub-cellular compartments. This is in agreement with the exclusive reduction of sulfate to sulfide by the activity of sulfite reductase in plastids  and an efficient diffusion through membranes without the necessity of a facilitator protein . In contrast, OAS, the carbon backbone for incorporation of sulfide, is almost exclusively produced in the mitochondria of Arabidopsis leaves [4,32]. Since the availability of OAS is generally considered as rate limiting for cysteine synthesis [22,33], the probable high concentration of OAS in mitochondria [30,32] could be used to remove free sulfide in this compartment. The high excess of OAS-TL over SAT activity in plastids and cytosol further indicates that sulfide should be efficiently fixed in these sub-cellular compartments in order to avoid accumulation of sulfide in the mitochondria [3,4].
A concentration of sulfide in the mitochondria similar to those found in the plastids or cytosol (>10000 nM) would result in a complete inhibition of COX (IC50 ~10 nM). In agreement with this hypothesis, dark respiration was inhibited by approximately 50% in oastlC seedlings due to specific inhibition of COX activity. The oastlC mutant displayed a retarded growth phenotype, although OAS-TL C contributes to only 5% of total OAS-TL activity. In contrast, knockout of the highly abundant cytosolic OAS-TL A and plastidic OAS-TL B does not cause a visible phenotype . One could postulate that the phenotype of oastlC is mainly caused by inactivation of mitochondrial COX. A note of caution must be added to this interpretation. The substrate affinity of OAS-TL C for sulfide (Km=4.7 μM ) would allow efficient cysteine synthesis, but is three orders of magnitude higher than the IC50 of COX for sulfide, which makes an efficient protection of COX against sulfide exclusively by OAS-TL C implausible. However, as a result of its high specific activity (534 μmol·min−1·mg−1), OAS-TL C may contribute to the protection of COX against sulfide by keeping the sulfide concentration in mitochondria in the low micromolar range. This is especially important when sulfide production is high to meet demand for ample cysteine synthesis, such as in fast-growing tissues. Remarkably, COX activity was significantly inhibited in oastlC seedlings, but not affected in 5-week-old oastlC plants. The conservation of sensitivity towards sulfide between COX proteins from animals and Arabidopsis strongly indicates that inhibition of COX by sulfide is a general feature and takes place also in other plant species. In spite of this, Warrilow and Hawkesford  showed that no mitochondrial OAS-TL exists in spinach, which further questions the exclusiveness of sulfide detoxification by mitochondrial OAS-TL in higher plants.
Indeed, we could demonstrate the presence of a sulfide-consuming process in isolated mitochondria of oastlC, which could also contribute to detoxification of sulfide in other plant species. This detoxification mechanism most likely allowed oastlC plants to overcome severe inhibition of COX due to high mitochondrial sulfide levels caused by fumigation. The nature of the newly identified detoxification mechanism for sulfide is not known so far. In a similar manner to plants, animals also face the problem of sulfide intoxication of COX, since sulfide is generated as a by-product of cysteine catabolism . Very recently, the functional knockout of mitochondrial sulfur dioxygenase (ETHE1) in mouse was reported to mimic the phenotype of ethylmalonic encephalopathy, which is an autosomal recessive disease of humans accompanied by inhibited COX activity in muscle and brain caused by high sulfide levels . Human ETHE1 shares 60% identity with the gene products of At1g53580, an uncharacterized protein with a predicted mitochondrial pre-sequence and 27% identity with At3g10850, a protein annotated as hydroxyacylglutathione hydrolase. The protein encoded by At1g53580 is a prime target for the sulfide-consuming mechanism in mitochondria of oastlC, either via the same or a similar mechanism as ETHE1. The existence of an alternative sulfide detoxification system which is able to compensate for OAS-TL-dependent sulfide detoxification in oastlC, and the unaffected COX activity in 5-week-old oastlC plants, makes sulfide toxicity unlikely to be the sole cause for the decrease in growth of oastlC . In animals, sulfide does not only act as a toxic by-product of cysteine catabolism, but also functions as a gasotransmitter that controls blood pressure by activation of ATP-sensitive K+ channels in vascular smooth muscle cells [35–37]. Recent studies in animals indicate that sulfide predominantly acts as a signal by S-sulfhydrating cysteine residues in target proteins . Although no such target proteins are currently known in plants, it is conceivable that sulfide also has a signalling function in higher plants. In contrast with animals, higher plants have a receptor for sulfide. This sensor system is built up by reversible association of SAT and OAS-TL within the CSC. The CSC is stabilized by sulfide and thereby controls the formation of OAS depending on the actual cysteine content [1,33,38]. The absence of sulfide-dependent stabilization of the mitochondrial CSC would result in an inadequate supply of the plant with OAS [4,32] and would therefore contribute to the observed retardation in growth. In agreement with this hypothesis, oastlC plants have lower OAS steady-state levels .
In summary, our results demonstrate that: (i) COX is a target of sulfide toxicity in higher plants; (ii) synthesis of cysteine in mitochondria can add to the capacity for detoxification of sulfide, (iii) plants possess an OAS-TL-independent detoxification system for sulfide in mitochondria; and (iv) lack of mitochondrial cysteine synthesis does not limit cyanide detoxification, Fe–S cluster biosynthesis and protein translation in mitochondria.
Hannah Birke performed the experiments, except for characterization of respiration and Fe–S enzymes (Figure 2; performed by Florian Haas and Janneke Balk), contributed to writing the paper and prepared the Figures. Luit De Kok contributed to sulfide exposure experiments and approved the paper. Janneke Balk supervised parts of the experiments and contributed to development of the concept and writing the paper. Markus Wirtz developed the concept of the study and wrote the paper. Rüdiger Hell supervised the experiments, contributed to the development of the concept and approved the paper prior to submission.
We thank Catherine Ainsworth for her valuable contribution and Chung Pong Lee for carefully reading the paper prior to submission.
H.B. was funded by the Landesgraduiertenförderung Baden-Württemberg for funding within the Hartmut Hoffmann-Berling International Graduate School of Molecular and Cellular Biology and the Schmeil Stiftung Heidelberg. This work was funded by the German Research Society (Deutsche Forschungsgemeinschaft) [grant number He1848/13-1/14-1] and travel was supported by the Excellence Initiative Mobility Program of the University of Heidelberg as well as the A. and R. Bauer Foundation. J.B. is supported by a University Research Fellowship of the Royal Society.