The antagonism between thiol oxidation and reduction enables efficient control of protein function and is used as central mechanism in cellular regulation. The best-studied mechanism is the dithiol-disulfide transition in the Calvin Benson Cycle in photosynthesis, including mixed disulfide formation by glutathionylation. The adjustment of the proper thiol redox state is a fundamental property of all cellular compartments. The glutathione redox potential of the cytosol, stroma, matrix and nucleoplasm usually ranges between −300 and −320 mV. Thiol reduction proceeds by short electron transfer cascades consisting of redox input elements and redox transmitters such as thioredoxins. Thiol oxidation ultimately is linked to reactive oxygen species (ROS) and reactive nitrogen species (RNS). Enhanced ROS production under stress shifts the redox network to more positive redox potentials. ROS do not react randomly but primarily with few specific redox sensors in the cell. The most commonly encountered reaction within the redox regulatory network however is the disulfide swapping. The thiol oxidation dynamics also involves transnitrosylation. This review compiles present knowledge on this network and its central role in sensing environmental cues with focus on chloroplast metabolism.
The reactivity of cysteine thiols
Structural stabilization of proteins by disulfides and detoxification of reactive oxygen species (ROS) by thiol peroxidases evolved early during evolution. Nowadays, cysteinyl residues in proteins function in catalytic mechanisms, structural stabilization and regulation of proteins. The unique properties of cysteinyl sulfur are linked to its reactivity, its ability to adopt multiple oxidation states and to form thiolate anions (Figure 1). As consequence of this peculiar reactivity, Cys is under-represented in plant proteins with a molar share of ∼1.6% among the 20 proteinogenic amino acids . The targeted positioning of Cys in polypeptides implies strict selection based on functionality and discrimination against random Cys residues .
Schematics of orbital occupation in sulfur and thiol reactions.
Sulfur has two half-occupied p-orbitals corresponding to two free valencies. The thiol, disulfide, persulfide and polysulfide, S-nitrosyl and sulfenic acid derivatives all use the ordinary valencies of the Cys sulfur, one for C-bonding in the peptide or glutathione, and the other one for a single bond to hydrogen, sulfur, nitrogen or oxygen. Hypervalency bonding is realized if the Cys-sulfur is oxidized further to sulfinic or sulfonic acid (Figure 1). Other reactive sulfur species (RSS) such as thiosulfinate (RS(O)SR’) and thiosulfonate (RS(O)2SR’) which occur in biological systems also rely on more than two valencies .
The protonated thiol group –SH is rather unreactive, and only after deprotonation the thiolate anion –S− with its high electron density and nucleophilicity is prone to multiple chemical reactions such as oxidation by H2O2 . The pK-value of each thiol determines its ability to adopt the thiolate form in the cellular pH-environment of the bulk phase. Formation of the thiolate form also depends on the protein microenvironment which affects the electrophilicity of the oxidant or stabilizes the transition state of the reaction . Another important parameter of thiol chemistry is linked to the midpoint redox potential which defines the reduction potential relative to the 2H++2e− → H2 reaction set to −0.414 V at pH = 7. The physiological redox potentials can be calculated using the Nernst equation taking into consideration the concentrations found in the cellular compartments .
All cells contain a thiol redox regulatory network, which relies on one or more of the three most commonly encountered thiol modifications, the formation of intra- or intermolecular disulfide bonds, S-glutathionylation and S-nitrosylation . The thiol redox regulatory network as suggested in 2008 by Dietz  consists of redox input elements for feeding electrons from metabolism into the network, e.g. NADPH-dependent thioredoxin reductases (NTR) and ferredoxin (FDX), redox transmitters like thioredoxins (TRX), redox target proteins, redox sensors and final electron acceptors. The most prominent final electron acceptors are ROS and reactive nitrogen species (RNS). Jacob (2011)  termed the system thiolstat. This regulatory network exists in all cells and sensitively monitors the metabolic state with high spatial and temporal resolution. It adjusts the redox state of target activities, which essentially comprise all cellular processes ranging from metabolism, gene transcription, translation, cell division to development.
Adjusting the redox state of the network in subcellular compartments
The eukaryotic cell concept developed by Schnepf  distinguishes plasmatic compartments and extraplasmatic compartments within the cell. Crossing a membrane always means switching from one compartment type to another one, e.g. if moving from the cytosol as plasmatic compartment across a membrane to any neighboring compartment like endomembrane systems, vacuole or outer chloroplast envelope, the conditions change to those of an extraplasmatic compartment. Plasmatic compartments have a rather neutral pH, a very low free Ca2+ activity, contain nucleic acids and protein synthesis, and also display a highly negative glutathione redox potential close to −310 mV . Peroxisomes might be exceptional extraplasmatic compartments due to their highly reduced glutathione content .
Plasmatic compartments comprise the cytosol, the mitochondrial matrix, the stroma and the nucleoplasm. Their negative thiol redox potential is buffered by the highly reduced glutathione pool in conjunction with glutathione reductases (GR) and the protein thiols. Live cell imaging with redox sensitive fluorescent proteins fused to glutaredoxin (roGFP-GRX) introduced into the cell and targeted to different cell compartments allows researchers to estimate the glutathione redox potential of the cytosol, mitochondrial matrix, the chloroplast stroma and the endoplasmatic reticulum (ER), the latter in HeLa cells [12–14]. Thiol oxidation for structural stabilization of proteins in the lumenal spaces of the endomembrane system, belonging to the extraplasmatic compartments, the intermembrane space of mitochondria and the thylakoid lumen depends on protein disulfide isomerase (PDI), thiol oxidases and an oxidant .
H2O2- and .NO-dependent oxidation of Cys thiols
Thiol oxidation ultimately is linked to reactive molecular species like H2O2 and .NO (Figure 2). However, only certain protein thiols are amenable to efficient direct oxidation by H2O2 and .NO. Molecular dioxygen (O2) contributes to relatively slow thiol oxidation. The protonated form of thiols is poorly reactive. The reaction of H2O2 with the thiolate was recently suggested to proceed via a slightly charged peroxide oxygen which turns even more negative in the product via an electrophilic attack on the negative sulfur atom . The authors employed hybrid quantum-classical (QM-MM) molecular dynamics simulations to describe the reaction (Figure 2A). Water molecules as solvent play an important role in the reaction by modulating the solvation structure and the electron distribution. The sulfenic acid then reacts with another thiol to form an intra-, intermolecular or mixed disulfide.
Reaction mechanism of sulfur with H2O2 and .NO.
One of the most reactive and best studied Cys thiols is the peroxidatic thiol in peroxiredoxins (PRX). Based on thermodynamic considerations and reaction kinetics, thiol peroxidases were proposed to function as exclusive ROS sensors in unstressed cells, e.g. in redox regulation of transcription factors . The chloroplast contains four PRXs (2-CysPRXA, 2-CysPRXB, PRXQ and PRXIIE) and two glutathione peroxidase-like proteins (GPX1 and GPX7) which function as thiol peroxidases .
.NO is a moderately reactive radical, which even diffuses through cell membranes. Time-dependent and spatial release of .NO determines its reaction with biomolecules . .NO-dependent oxidation of thiols may occur after oxidation of .NO to NO2, N2O3 or ONOO.− which then react with thiols (Figure 2D). ONOO.− is generated if .NO reacts with O2.− and is considered as most reactive. .NO also reacts with thiyl radicals (RS.). Transition metals, e.g. in heme groups, catalyze the reaction of .NO with thiolates. The relative contribution of these different pathways of S-nitrosothiol formation in the cellular context is unknown so far. A global assessment of S-nitrosylated proteins by a method called biotin switch with subsequent mass spectrometric identification revealed many chloroplast proteins as target for this PTM [18,19].
Similar to disulfide shuffling, trans-nitrosylation occurs by transfer of .NO from nitrosothiols as donor to suitable thiols as acceptor. Trans-nitrosylation is the most prominent reaction within the S-nitrosothiol network and proceeds via a nucleophilic attack of a thiolate on the donor S-nitrosothiol. The nitroso-group is exchanged intra- and intermolecularly between proteins, S-nitrosoglutathione and other S-nitroso compound like S-nitrosocysteine .
Chloroplast triosephosphate isomerase (TPI) of Chlamydomonas reinhardtii is S-nitrosylated by S-nitrosoglutathione but not by .NO released from a pharmacological .NO-donor . Cys219 in CrTPI is located in a suitable environment of an acid-base motif suggested to flank reactive target Cys , easing thiolate formation and thus, transnitrosylation. Since lower plant TPI has dual functions in Calvin Benson Cycle (CBC) and glycolysis, the type of regulation of CrTPI in algae may differ from that of TPI in land plants . Denitrosylation of S-nitrosylated proteins involves glutathione, TRX, nitrosoglutathione reductase or non-enzymatic reactions . The stability of an S-nitroso-group in a particular protein molecule and the dynamics of the S-nitrosylation pattern in a specific subcellular compartment needs attention and should be explored both within a pool of homogeneous proteins and in a complex mix of diverse proteins found in a (sub-)cellular scenario such as the stroma.
Persulfidation as thiol oxidation
D-cysteine desulfhydrases (DCD) release H2S from D-Cys. H2S in turn can react with Cys-thiols most likely after previous sulfenylation of the thiol to form Cys-S-SH . This reaction is named persulfidation and participates in hormonal signaling. Chen et al.  reported DCD-dependent H2S production in guard cells. H2S-dependent persulfidation takes place at two specific Cys residues (Cys131, 137) of the Open Stomata 1 (OST1) protein kinase (SnRK2.6) which in turn activates ABA signaling and induces ABA-dependent stomatal closure. The chloroplast is a main source of cellular H2S by sulfite reduction. Thus, it may be expected that a significant portion of the persulfidated polypeptides in Arabidopsis thaliana are associated with the chloroplast (∼22%). This group comprises many polypeptides of the photosynthetic electron transport chain (PET) and the CBC , albeit the physiological role of persulfidation of these polypeptides remains to be shown .
Glutathionylation as thiol oxidation mechanism
The tripeptide glutathione (γ-Glu-Cys-Gly) represents the most abundant redox-active low molecular mass component in eukaryotic cells with cytosolic concentrations of 1–10 mM . The redox couple of oxidized (GSSG) and reduced glutathione (GSH) has a midpoint redox potential of -240 mV at pH 7 and the pK of its Cys residue is in the range of 8.6 to 9 and thus, close to the pK of free Cys of 8.5 . Glutathione is found in the cytosol, peroxisomes, the apoplast, vacuoles, nuclei, mitochondria and chloroplasts. The redox potential of glutathione is adjusted with respect to the required redox conditions in the compartments. GRs reduces GSSG at the expense of NADPH + H+ and ensures a reducing environment of ∼−310 mV in the cytosol, corresponding to only micromolar concentrations of oxidized glutathione and a GSH/GSSG ratio of more than 1000 : 1 [12,28,29]. In a converse manner, oxidative conditions and a GSH/GSSG ratio of up to 1 : 1 are required in the ER to ease appropriate folding by protein disulfide isomerase (PDI) . In general, the ratio reflects the redox state of the cell and GSSG can accumulate due to ROS-scavenging so that the GSSG-amount becomes significant for spontaneous reaction with protein thiols . Its high abundance and impact on cellular responses led to comparisons with hormones such as salicylic acid, abscisic acid and ethylene .
Stress conditions result in ROS signaling by increased release of H2O2, which shifts the GSH/GSSG ratio towards a more oxidizing state. This favors the reaction of reduced protein thiols with GSSG leading to S-glutathionylation (Figure 3). But to facilitate this reaction a drastic change of the GSH/GSSG ratio would be necessary. It is interesting to note that the ascorbate-dependent water-water cycle, which uses ascorbate and ascorbate peroxidase to detoxify H2O2 and generates mono- and dehydroascorbate, links the ascorbate to the glutathione pool. As shown for the cytosolic water-water-cycle, DHAR plays a major role in oxidizing GSH [33,34]. DHARs mediate the H2O2-dependent effect on the GSH/GSSG ratio in the cat2 (At4g35090) background where the major isoform of peroxisomal catalase is missing. The connection between the ascorbate and glutathione is not very tight, but chloroplasts use this mechanism as well.
A spontaneous glutathionylation occurs more likely among the sulfenic acid of an oxidized cysteine residue and reduced glutathione [30,35]. Prerequisite of glutathionylation is a redox-sensitive cysteinyl residue with a low pK in the range 3–7, which can be achieved by basic amino acids near the cysteinyl residue [30,36]. This condition also favors spontaneous oxidation of the cysteinyl residue and thus, irreversible hyper-oxidation to sulfinic or sulfonic acid, so that reversible glutathionylation competes with this process and prevents irreversible inactivation in a protective manner [37,38]. Last but not least, nitrosoglutathione (GSNO) can cause glutathionylation of cysteinyl residues, too, as reported for the cytosolic isocitrate dehydrogenase from A. thaliana . Enzymatic glutathionylation has been described for glutaredoxin C2 (GRX C2) in A. thaliana, which glutathionylated the BRI1-associated receptor like kinase (BAK1) and is suggested to thereby control the activity of BAK1 [40,41].
De-glutathionylation of proteins occurs either spontaneously by GSH, if a highly reducing GSH/GSSG ratio is restored, or enzymatically by GRXs and cysteine-containing glutathione-S-transferases [38,40]. In A. thaliana, 31 GRXs are present, of which type I GRXs are capable to de-glutathionylate as shown for the cytosolic GRXC1 and C2 and the plastidic GRXS12 and C5 [39,42,43]. The catalytic cycle of GRXs involves a GRX-SSG intermediate, which is attacked by reduced GSH in a monothiol mechanism so that the activity of the GRX is restored and GSSG is formed [30,42]. Finally, GSSG is reduced by glutathione reductases (GR). Two GRs are known for A. thaliana, which are organelle-specific: GR1 locates to the cytosol, peroxisomes and nucleus, and GR2 is dual-targeted to chloroplasts and mitochondria, although it is essential in plastids only [44,45].
Glutathione functions in cellular redox-homeostasis, detoxification of heavy metals and xenobiotics, and glutathionylation allowing for protecting protein thiols against irreversible oxidation. More recently, glutathionylation has been suggested as signaling mechanism [30,36,38]. Since most of the data were collected in vitro, the importance of the data remains unclear, for instance glutathionylation of alcohol dehydrogenase from A. thaliana had no impact on its activity .
Furthermore, the proteomic data on glutathionylation is scarce for autotrophic cells and organisms; 349 proteins were found glutathionylated in the cyanobacterium Synechocystis sp. PCC6803, among them mostly proteins of central pathways like the CBC, sugar and fatty acid metabolism and nitrogen assimilation . In Chlamydomonas reinhardtii 225 proteins were identified, including ten out of eleven enzymes of the CBC . A set of 79 putative proteins was detected subsequent to labeling in A. thaliana, but only nine were identified in vivo .
Regulating the CBC by glutathionylation appears to be a major mechanism in autotrophic cells, and independent data on TPI from C. reinhardtii and glyceraldehyde-3-phosphate dehydrogenase (GAPDH), phosphoribulokinase (PRK) from A. thaliana confirm the findings [20,50]. In addition, oxidation sensitivity of GAPDH and PRK is influenced by the light-dependent regulator chloroplast protein 12 (CP12), which protects the enzymes from oxidation, while CP12 itself might be subject of glutathionylation [47,50].
With respect to the daily variation in glutathione abundance and oxidant-dependent ratio of GSH/GSSG , it is intriguing that among the plastidic TRXs only TRX-f is targeted by glutathionylation. TRX-f plays a major role in light acclimation of the CBC and carbon metabolism. Glutathionylation of an additional conserved, but non-catalytic cysteinyl residue within TRX-f results in reduced ability to activate its targets . Taken together, glutathionylation acts directly on the enzymes of the CBC, which might represent a protective action under stress as shown for PRK of C. reinhardtii . Glutathionylation also modifies enzyme activity indirectly via TRX-f, which might represent a signaling mechanism.
Besides enzymes of the CBC, plastidic starch catabolism is subject of protective glutathionylation as reported for the plastidic α-amylase 3 of A. thaliana, which is prone to irreversible oxidation due to the low pK (5.7) of its catalytic Cys499, which also favors glutathionylation. Interestingly, both GRX and TRX were required to restore full activity of the formerly glutathionylated enzyme . This might be due to the TRX-mediated regulatory disulfide bridge between Cys499 and Cys587 .
The chloroplast possesses a peculiar NTR with fused TRX domain (NTRC) and GR, but does lacks a backup system like NTRA and NTRB or a GSSG-transporter delivering GSSG to cytosolic GR1 as additional flood gate . It is therefore hypothesized that glutathionylation of highly abundant proteins helps to cope with severe stress conditions and to restore the initial GSH/GSSG-ratio.
Thiol-disulfide exchange reaction
All redox regulatory networks rely on cascades of thiol-disulfide exchange reactions. In fact, such reaction that a disulfide oxidizes two thiols is likely the most common oxidation step. Modeling of the thiol-disulfide exchange reaction indicates the occurrence of a transient state where (i) the thiolate of a protein attacks a disulfide, (ii) three sulfur atoms align and (iii) another thiolate leaves the tripartite intermediate (Figure 4) . This mechanistic scenario does not explain the preferential occurrence and direction of disulfide shuffling. A hybrid Monte Carlo and Molecular Dynamics approach revealed that proximity and accessibility were the main criteria for the initiation of thiol-disulfide isomerization . The model predicted the region-selectivity of the isomerization reaction in a mutated immunoglobulin domain.
Disulfide shuffling between protein thiols.
Protein disulfide isomerase-dependent oxidation of thiols in the secretory pathway as prototypic mechanism
Disulfide formation in the ER may serve as a prototypic thiol-disulfide shuffling cascade. Protein disulfide isomerases (PDI, EC 184.108.40.206) are long known to assist in proper folding of proteins synthesized in the secretory pathway which starts in the ER. Two PDIs coexist in the microsomal fraction from rat liver, a soluble and a membrane-bound version . PDI activity was then found in wheat embryos  and in all eukaryotic cells. Ten years later, Bulleid and Freedman (1988)  demonstrated that dog microsomes deficient of PDI cannot form properly folded secreted protein by lack of disulfide formation. PDIs catalyze dithiol-disulfide exchange reactions and are composed of an arrangement of two catalytic (a, a′) and two non-catalytic (b, b′)-domains in the arrangement a-b-b′-a′ (Figure 5). They rely on the TRX fold. In order to function as a dithiol oxidase of target proteins in the ER, PDIs must get oxidized .
The electron flow and disulfide shuffling in the protein disulfide isomerase (PDI) and endoplasmic reticulum oxidase (ERO1).
One of the involved pathways employs the flavoprotein ERO1α which has 15 Cys residues in humans and 18 in Arabidopsis (Figure 5). Some of these Cys have regulatory functions, while others serve in catalysis. Disulfide formation between Cys99–Cys104 and Cys94–Cys131 inhibits the enzymatic activity of ERO1α. The PDI oxidase function of ERO1α relies on the disulfide formed between Cys394–Cys397 which attacks the Cys-X-X-Cys motif in the a-domain of the PDI. The disulfide of a-domain then accepts the electrons from the a′-domain which oxidizes the secreted target proteins .
Overall the ER manages a cascade of electron transfer reactions from the target protein, e.g. the secreted storage protein or immunoglobulin, to the a‘-domain of PDI, the a-domain of PDI, the C-terminal Cys394–Cys397 of ERO1α to O2 which is converted to H2O2. This was demonstrated with a Clark electrode monitoring O2 consumption in a glutathione oxidation test (10 mM GSH) reconstituted in vitro with recombinant ERO1α and PDI of either wildtype (a and a′ present), PDI-a′ (with functional a′ only) and PDI-a (with only functional a-domain), where WT PDI depicted the highest activity and PDI-a lacked any activity . Sequence and structure similarities between human and plant PDIs and EROs suggest conservation of the mechanism  (Figure 5). Thus, this thiol oxidase cascade uses thiol disulfide exchange reactions to transfer electrons from reduced sulfur to O2 and may be considered as exemplary mechanism how thiol oxidation efficiently functions in context of structural stabilization. The PDI/ERO1/O2-pathway may also serve as blueprint how thiol oxidation functions in context of regulation.
Regulatory oxidation of thiols in the chloroplast
Photosynthetic metabolism relies on thiol-based redox regulation. Initially discovered as mechanism to align the activity of the PET with CO2 fixation and, thereby, ATP and NADPH consumption in the CBC, thiol redox regulation is now known to control diverse metabolic activities in the chloroplast, but also in other plastids and in any plasmatic cell compartment [63,64]. The A. thaliana genome codes for 20 TRX and TRX-like proteins targeted to plastids [65,66]. This high number demonstrates the outstanding importance of thiol redox regulation in plastid metabolism. For a long time research focused on reductive activation of CBC enzymes which is most efficiently achieved by TRX-f and TRX-m . Two reductive pathways feed electrons either from ferredoxin (FDX) via FDX-dependent TRX reductase (FTR) or from NADPH via NADPH-dependent TRX reductase C (NTRC) to target proteins . Most chloroplast proteins are active in their reduced dithiol state with few established exceptions, e.g. reduction inactivates NADPH-dependent glucose-6-phosphate dehydrogenase in order to suppress futile cycles of ATP hydrolysis in the light .
Efficient regulation by dithiol-disulfide transitions requires mechanisms not only for reduction, but also for oxidation. The regulatory adjustment of the thiol redox state enables fine-tuning of target protein activity . As pointed out above, there are several possibilities of oxidation of thiols in target proteins, namely oxidation by molecular dioxygen O2, by ROS, in particular H2O2, or by dithiol-disulfide exchange. Maintenance of active CBC enzymes, e.g. in in vitro tests requires protective reductants like dithiothreitol, and TRXs oxidize in the presence of atmospheric O2. The second-order rate constant for 4 RSH + O2 → 2 RSSR + H2O in alkaline solution is four to five orders of magnitude smaller than with H2O2 . A resting concentration of H2O2 in the chloroplast of 30 nM as derived from a simulation as described below  would be 10 000-fold lower than the O2 concentration at 20°C in equilibrium with the 21% atmospheric O2 (284 µM at 20°C). The relative contribution of H2O2 and O2 to cellular thiol oxidation awaits elucidation and likely differs between root cells with less O2 in the medium and photosynthesizing leaves.
Chloroplasts generate O2.− in the PET. After dismutation of O2.− by superoxide dismutase, H2O2 functions in the redox regulatory network as final redox acceptor (Figure 6). Driever and Baker  suggested that H2O2 released by PET does not act as major alternative electron sinks under stress but serves regulatory functions.
Modeling and simulation of a redox subnetwork of the chloroplast.
H2O2 is the final redox acceptor in the redox regulatory network  and preferentially reacts with thiol peroxidases . The network consists of redox input elements with shared function as electron donors in metabolism and in the redox regulatory network (Figure 6). Redox transmitters mediate the electron transfer to redox regulated target proteins. Thiol peroxidases such as 2-Cys peroxiredoxins (2-CysPRX) most efficiently react with peroxides, turn oxidized and get re-reduced by withdrawing electrons from redox transmitters. Oxidized transmitters can oxidize target proteins, thus transmitters like TRXs serve in both directions, as reductants and oxidants. This mechanism in part employs the same TRXs like TRX-f1 for FBPase and TRX-m1 for NADPH-dependent malate dehydrogenase . In addition, atypical TRXs like ACHT1–4  and TRX-like 1–3  participate in oxidizing target proteins. Structure analysis together with site-directed mutagenesis and molecular dynamics simulation revealed amino acid residues e.g. Cys-126 and Thr-158 in TRX-f1, which determine the specificity for target reduction and other amino acid residues for target oxidation in case of ACHT .
The experimental work leading to this knowledge relied on in vitro work with recombinant proteins and ex vivo studies with mutants lacking certain TRX isoforms. However, not a single approach provides access to quantitative network understanding. Kinetic modeling of the redox network allows simulating redox changes of network elements, e.g. dependent on H2O2 concentration  (Figure 6). The simulated network included the regulatory TRX-dependent branch reducing FBPase or 2-CysPRX and the metabolic branch using ferredoxin (FDX)-dependent NADP-reductase to generate NADPH + H+, which either drives metabolism or reduces NADPH-dependent TRX reductase C (NTRC). NTRC is also an efficient electron donor to 2-CysPRX.
Several parameters inaccessible in experiments could tentatively be determined by network simulation . Relating the redox state of 2-CysPRX as determined in leaf extracts  to the results of the network simulation allowed estimating the resting H2O2 concentration in the chloroplast with 30 nM. According to this simulation, electron consumption in metabolism exceeds electron flow to FBPase for reductive activation ∼7000-fold. NTRC was 5.32-times more efficient than TRX-f1 to reduce 2-CysPRX. Accumulation of 30 μM H2O2 as assumed under severe stress increased the electron flow to the thiol network relative to metabolism ∼28-fold to a ratio of 1:251 .
Among the different PRXs in the plastids, 2-CysPRX was most efficient as TRX oxidase. PRXQ was less efficient but still able to inactivate target enzymes via TRX oxidation, while PRXIIE has a different function in the redox network . This is in line with the peculiar observation that 2-CysPRX is oxidized by 50%–65%, PRXQ by 50%, while PRXIIE is highly reduced if extracted from leaves . PRXQ associates with the thylakoids and may play a role in redox regulation of PET components . The specificity of PRXs as TRX oxidases within the chloroplast should receive more attention. The study should expand to other thiol peroxidases, in particular to the GPX-like proteins, since they likely play a major role as redox sensors in the network as well.
The role of proteins as thiol buffer
The total concentration of protein thiols in plasmatic compartments exceeds the thiol concentration of low molecular mass thiol compounds such as glutathione and cysteine more than 10-fold. This can be estimated for the chloroplast stroma, where RubisCO concentration is ∼0.5 mM. Each RubisCO molecule consists of eight large subunits with nine Cys and eight small subunits with four Cys residues . This means that the RubisCO Cys thiol pool is equivalent to 52 mM compared with a stromal glutathione pool of ∼3 mM . RubisCO only represents about half of the chloroplast protein.
However, not all Cys are surface-accessible. In wheat leaves protein thiols exceeded the glutathione thiols by a factor of 22, and this value dropped to twelve in drought stressed plants . Apparently, a large reservoir of protein thiols participates in adjusting the thiol redox state of a particular compartment. Such a role as thiol buffer is ascribed to serum albumin, which constitutes the major thiol pool in blood serum . Muthuramalingam et al.  determined the Cys thiol pool in stromal proteins with 58 µmol/g protein after denaturing and reducing the protein. Stroma protein has a high number of total Cys thiols, many of which are accessible. Disulfide-dithiol exchange reactions and nitrosothiol hopping by transnitrosylation may facilitate dynamic propagation of thiol oxidation throughout the accessible pool of thiols.
Outlook, where to go?
The mechanism of thiol-disulfide shuffling and trans-nitrosylation adds a dimension of ‘volatility' to thiol-based redox regulation. Studies and experimental approaches usually present a focused snapshot, but there is a need for exploring fluxes and distribution of such S-related PTMs in more detail. How do S-nitrosyl groups, S-glutathionyl groups and disulfide bonds propagate through the thiol landscape of an organelle in a temporal and spatial manner?
What happens with the thiol proteome in a kinetic experiment if a certain amount of thiol oxidant like GSSG, GSNO or H2O2 is applied as individual stimulus or in combination? Proteomics approaches usually focus on single types of PTMs. These independent studies identified identical target proteins as subject of different S-PTMs, e.g. the peroxidatic Cys of PRXIIE can be subjected to disulfide bond formation, S-nitrosylation, S-glutathionylation, sulfenylation, sulfinylation and sulfonylation. Knowledge on the state of the different S-PTMs determined in parallel will improve our understanding of the S-PTMs network in its entirety.
Despite the methodological progress in studying the redox state of proteins and identifying involved targets, there remain gaps in assessing certain decisive biochemical features of the network in vivo. This drawback in particular concerns the electron fluxes through the network and the concentration of reactive species. Mathematical modeling provides computational access to such parameters. Gerken et al.  introduced two partial and a combined solutions to a small chloroplast redox regulatory network. This type of computational approach should be expanded to include more elements of the network and to improve our understanding.
The multiplication and functional diversification of elements within the chloroplast redox regulatory network, e.g. with 20 TRX and TRX-like redox transmitters, pose challenging research questions. Molecular structures and complementary surface areas differ between oxidized and reduced targets. We need to explore the interacting interfaces between the transmitters and reduced and oxidized targets and determine the kinetic constants of redox reactions. The recent work by Yokochi et al.  provides excellent experimental guidance toward this direction.
Improved biochemical and proteomic knowledge will provide the basis for designing novel in vivo test systems. Knock out mutants may be complemented with tailored variants of redox elements, which have altered interaction abilities and increased reduction or oxidation efficiencies. We have to test our hypotheses not only in loss- and gain-of-function mutants but in a highly refined manner. In the next step, attention should be given to ecological and phylogenetic variation of the network. Which modifications to the network are needed if a plant inhabits the understory of a dark forest with short sun flecks and, therefore, must manage a rapid response system to efficiently harvest photons during the short periods of light in comparison with sun leaves? Full understanding of the system is expected to provide strategies for improved performance of crop plants in stressful growth conditions.
The authors declare that there are no competing interests associated with the manuscript.
The own cited work was supported by the German Science Foundation (DFG, Di346/17).
L.V. and K.J.D. discussed and wrote the paper.
BRI1-accociated receptor like kinase
Calvin Benson Cycle
chloroplast protein 12
ER oxidase 1α
FDX-dependent TRX reductase
glutathione peroxidase like
NADPH-dependent TRX reductase
protein disulfide isomerase
photosynthetic electron transport
post translational modification
reactive nitrogen species
reduction-oxidation sensitive green fluorescent protein
reactive oxygen species