Phylogenetic analysis of glutathione-dependent redox-relevant plastid proteins from streptophyte algae to land plants
On the one hand, many protein families of redox cascade components have expanded during land plant evolution and underwent functional diversification [9]. On the other hand, several proteins are persistently conserved as only one isoform in a compartment, suggesting a potential essential biological function and tight regulation of gene copy number. In order to identify glutathione-related redox components in plastids with interesting phylogenetic patterns, we first did an explorative phylogenetic analysis taking advantage of the available sequence coverage of streptophyte algae and non-seed plant lineages. We included all known scavenging and damage repair enzyme families with members that were reported to (1) localise to plastids and (2) use GSH and produce GSSG: dehydroascorbate reductase (DHAR), glutathione S-transferases lambda and iota (GSTL, GSTI), atypical methionine sulfoxide reductase B1 (MSRB1), peroxiredoxin IIE (PRXIIE), glutaredoxins (GRXs), as well as glutathione reductase (GR), which is responsible for reduction of GSSG. All used protein sequences and gene accessions are listed in Table S1 (Additional File 1).
Dehydroascorbate reductases
DHARs belong to the GST superfamily and catalyse the reduction of dehydroascorbate to ascorbate using GSH as electron donor. By maintaining a reduced ascorbate pool, DHARs indirectly assist in the detoxification of H2O2 catalysed by ascorbate peroxidases (APXs). The phylogenetic tree of plant DHARs (Additional File 2) shows several independent gene duplications resulting in a number of 1–3 paralogs per species. Targeting predictions (TargetP [48, 49] as well as the presence or absence of N-terminal sequence extensions suggest a high variability in subcellular targeting. Notably, only a single DHAR gene is present in C. braunii and two bryophyte species (M. polymorpha, A. agrestis) (Fig. 1, Additional File 2). Whereas the C. braunii ortholog is putatively cytosolic, both M. polymorpha and A. agrestis orthologs possess an N-terminal extension and are predicted to be targeted to plastids. This indicates that a stromal DHAR occurred early in land plant evolution. DHAR1 from P. patens (Pp3c22_5470V3) is predicted to be-targeted to plastids but proteomic evidence also indicates its presence in mitochondria [50], which may suggest dual-targeting to plastids and mitochondria. All land plant model species except S. moellendorffii have at least one DHAR isoform with N-terminal extension and/or a plastid targeting prediction (Fig. 2). In S. moellendorffii, the situation remains unclear, as at least one gene model is incomplete at the N-terminus.
Numbers indicate the isoforms present in the genome and the numbers in brackets the number of isoforms with predicted plastid targeting (TargetP2.0, [49]). DHAR, dehydroascorbate reductase; GST, glutathione S-transferase, L: Lambda, I: Iota; MSRB1, methionine sulfoxide reductase B1 (1 Cys); PRX, peroxiredoxin; GRX, glutatredoxin; GR, glutathione reductase. Model species names are abbreviated: Cb = Chara braunii; Aa = Anthoceros agrestis; Mp = Marchantia polymorpha; Pp = Physcomitrium patens; Sm = Selaginella moellendorffii; Sc = Salvinia cucullata; Af = Azolla filiculoides; Bd = Brachypodium distachyon; At = Arabidopsis thaliana.
Iota (I)- and Lambda (L)-type glutathione S-transferases
In addition to DHARs, plastid-targeting has also been reported within the lambda-subclass of GSTs (GSTL) [51]. While canonical GSTs use GSH to conjugate electrophilic compounds, the subclasses theta, phi and tau have also been shown to act as GSH-dependent peroxidases that release GSSG [51, 52]. Because a similar reductive activity towards DHA or hydroperoxides has also been reported for plastid-localized PpGSTL1, we investigated isoforms of this subclass [51]. We did not identify any GSTL homologs in C. braunii, A. agrestis, M. polymorpha, S. moellendorffii but several homologs with variable N-terminal extensions and predicted targeting to plastids in the seed plant models (Additional File 3). On the contrary, the closely related subfamily of iota-type GSTs (GSTI) contains homologs of A. agrestis, M. polymorpha and S. moellendorffii, but is absent in the fern and seed plant models (Fig. 1). The function of GSTIs is yet unclear [51] while our analysis indicates that they may also be targeted to plastids where they might fulfil similar functions to their sister clade GSTL (Additional File 3, Fig. 2). We did not identify a C. braunii GSTI, but a C. reinhardtii GSTI has been reported [51], suggesting that the GSTI subfamily is present in chlorophytes.
Atypical methionine sulfoxide reductase B
Methionine sulfoxide reductases (MSR) catalyse the reduction of methionine sulfoxide (MetSO), repairing ROS-induced damage to proteins (Fig. 2). Oxidation of methionine can generate two diastereomeres that are reduced by two different non-related enzyme families, namely MSRA (reducing methionine-S-sulfoxide) and MSRB (reducing methionine-R-sulfoxide) [53]. MSRA and MSRB proteins containing two cysteines are dependent on TRX for regeneration, while atypical MSRBs such as plastid-targeted AtMSRB1 (AT1G53670), depend on GSH/GRX for regeneration [54]. In AtMSRB1, the second cysteine is replaced by a threonine (Thr132). Investigating AtMSRB1 homologs (Additional File 4a, c, d), we find a complete conservation of the isoform containing one cysteine as the well-supported MSRB1(1Cys) clade contains C. braunii, bryophyte, lycophyte and fern isoforms. The conservation of plastid targeting is unclear but likely variable, based on TargetP [48, 49] predictions and the presence and absence of N-terminal extensions (Additional File 4b, Fig. 2). Notably, M. polymorpha does not possess a MSRB2 homolog while the MSRB1 gene was duplicated and the emerging two isoforms show differing N-terminal extensions and targeting predictions (Additional Files 4 and 9).
Peroxiredoxin II E
PRXs are antioxidant enzymes that catalyze the reduction of hydroperoxides into alcohols using a strictly conserved cysteine (i.e. peroxidatic cysteine). Plastids contain canonical 2-Cys PRXs having two active site cysteines, namely peroxidatic and resolving cysteines, and an atypical 2-Cys PRX named PRXIIE (reviewed in [19]). Whereas the regeneration of canonical 2-Cys PRX is controlled by the plastidial TRX system via a dithiol/disulfide exchange, the recycling of oxidised poplar PRXIIE was demonstrated to be more efficiently catalysed by GRXS12 in vitro [30]. GRXS12 can restore the reduced PRXIIE as the peroxidatic cysteine undergoes glutathionylation during the catalytic cycle, highlighting the dependence of this PRXII subfamily on the GSH/GRX system for functional recycling (Fig. 2).
PRXIIE homologs are evolutionary conserved [55] and our analysis confirms a single highly supported clade of PRXIIE homologs from charophytes to flowering plants (Additional File 5). Most organisms investigated possess a single PRXIIE isoform and all complete protein models confirm an N-terminal extension that is largely predicted to confer targeting to plastids (Fig. 1, Fig. 2, Additional File 5). The S. moellendorffii homolog did not allow a targeting analysis as the N-terminus of the protein model was fragmentary. We find gene duplications in B. distachyon with two isoforms and P. patens with three isoforms, all with conserved plastid targeting. Based on the phylogenetic analysis (Additional File 5), plastid GRX-dependent PRXIIE homologs are likely present in all analysed organisms.
Glutaredoxins
GRXs are oxidoreductases belonging to the TRX family many of which are involved in the control of protein redox state. In photosynthetic eukaryotes, the GRX family comprises four classes that are localized in various subcellular compartments and distinguished by their active site signature and domain organization [3]. The nomenclature of the members of these GRX classes is based on the presence of a Cys or a serine at the last position of the active site signature (CXXC/S) with a limited number of exceptions containing a residue differing from Cys or serine [3]. In the model plant A. thaliana, GRXs belonging to class I, II, and IV are represented by a number of 6, 4, and 2 isoforms, respectively, while class III GRXs (also referred to as CC-type GRX or ROXY proteins) have largely expanded and comprises 21 members. Class III GRXs function involves the interaction with bZIP TGA transcription factors, influencing plant development and flowering [56, 57]. Plastids typically contain GRX members belonging to class I and II, namely class I GRXS12 and the close paralog GRXC5 (in A. thaliana) and class II GRXS14 and GRXS16 [58–60].
Class I GRX
GRXC5 and GRXS12 are formed by a single GRX domain with an active site signature (YCPYC and WCSYS, respectively) that slight differs from the typical YC[P/S/G][Y/F]C motif of class I GRXs. Both GRX isoforms are redox-active being involved in the control of protein glutathionylation of plastidial proteins and in the recycling of antioxidant enzymes such as MSRB1 and PRXIIE (see above). The analysis of the phylogenetic tree for the redox-active class I GRXs revealed several evolutionary ancient clades corresponding to isoforms targeted to plastids (C5/S12 clade), the cytosol (C1/C2 clade), and the secretory pathway (C3/C4 clade) (Additional File 6). We found that only in the two angiosperm model species investigated, the second cysteine of the active site was replaced by a serine, giving rise to the GRXS12 isoform. This substitution can increase protein activity because formation of an internal disulfide that would block the active side can be avoided [59, 61]. The relevance of this mutation is emphasized by the fact that GRXS12 homologs did even become predominant in several angiosperms [62]. While GRXC5 gene models of S. moellendorffii and M. polymorpha may be fragmentary and thus not contain the N-terminal targeting sequence, we did not identify a C5/S12 isoform clustering with the highly conserved GRXC5/S12 clade in the fern A. filiculoides (Additional File 6, Fig. 2).
Class II GRX
Class II GRXS14 and S16 contain a CGFS active site motif that is typically conserved in all other class II GRX members. As observed for plastidial class I GRXs, GRXS14 is formed by a single GRX domain whereas GRXS16 has a modular organization possessing an N-terminal domain (GIY–YIG endonuclease fold) fused to one GRX domain [3, 63, 64]. Class II GRX are mainly thought to be involved in the coordination and transfer of iron-sulfur clusters [65, 66] (Fig. 2), but may also become redox-active after loss of iron-sulfur coordination in response to an oxidative signal [67]. Our phylogenetic analysis showed that GRXS14 and GRXS16 homologs are conserved in all investigated model species, with only one duplicated isoform of GRXS14 in P. patens (Additional File 7). As far as complete gene models are available, N-terminal extensions and predicted plastid targeting confirm the very high conservation of a single plastid-targeted isoform in these class II GRX subfamilies (Additional File 7, Fig. 2).
Glutathione reductase
GRs perform the highly efficient NADPH-dependent recovery of GSH from enzymatically or non-enzymatically generated GSSG (Fig. 2), keeping GSSG as low as nanomolar amounts [25]. The resulting highly negative EGSH in the plant cytosol, peroxisomes, plastids and mitochondria is based on the activity of two isoforms exhibiting dual targeting to either cytosol and peroxisomes (e.g. AtGR1, AT3G24170) or plastids and mitochondria (e.g. AtGR2, AT3G54660) [32, 33]. These two isoforms were already shown to be evolutionary conserved, including the dual-targeting of one isoform to plastids and mitochondria in P. patens [68]. We inferred a phylogeny using our set of model species and found the same conservation of two clades, representing the two GR isoforms, suggesting that these isoforms were established before the emergence of land plants (Additional File 8, Fig. 2).
Schematic overview of the plastid GSH-dependent redox network in land plant model species and the streptophyte alga C. braunii. Electrons from photosynthetic electron transport (PET) contribute to ROS generation and at the same time to ROS scavenging, damage repair and redox homeostasis. H2O2 leads to lipid peroxidation (L-O-O-H) as well as oxidation of protein methionine (Met-R-SO), of ascorbic acid (AsA) to dehydroascorbate (DHA) or protein thiol oxidation to the respective sulfenic acid (RS-OH) that can react with GSH to form an S-glutathionylated adduct (RS-SG). Glutaredoxins (GRX) can (de)glutathionylate proteins. The balance between the reduced tripeptide glutathione (GSH) and glutathione disulfide (GSSG) is influenced by GSSG generation via enzymes involved in ROS/RNS scavenging or protein as well as lipid repair, such as dehydroascorbate reductase (DHAR), atypical (1 Cys) methionine sulfoxide reductases B1 (MSRB1), lambda and iota-type (?, function not confirmed in vitro) glutathione S-transferases and type II peroxiredoxins (PRX). Glutathione reductase (GR, NADPH-dependent) safeguards a highly reduced GSH-pool. The presence of at least one plastid-targeted isoform of a protein in a model species (see Fig. 1 and Additional Files 1–8) is represented by a coloured box next to the protein. A coloured box with question mark means the potential presence of an isoform as targeting prediction is unclear, but N-terminal extension indicating a targeting peptide is present (see Additional Files 1–8). Absence of a box can either mean absence of homologs from that species (see Fig. 1), or that all homologs do not have N-terminal extensions or that gene models are fragmentary (see Additional Files 1–8). Chara braunii (Cb), Anthoceros agrestis (Aa), Marchantia polymorpha (Mp), Physcomitrium patens (Pp), Selaginella moellendorffii (Sm), Salvinia cucullata (Sc), Azolla filiculoides (Af), Brachypodium distachyon (Bd) and Arabidopsis thaliana (At).
The S. moellendorffii model for the mitochondria/plastid GR clade was potentially truncated at the N-terminus, not allowing for a targeting prediction. We did not identify an isoform in the mitochondria/plastid GR clade for S. cucullata nor for A. agrestis (Bonn). A BLASTN search revealed a possible locus for S. cucullata on scaffold s0092, however without a gene model present. The presence of a plastidial/mitochondrial isoform in A. agrestis (Oxford) suggests that the respective gene is present but not correctly predicted for A. agrestis (Bonn). Except for B. distachyon, that possesses two plastidial/mitochondrial GR isoforms, each present species has a single isoform in each clade. Notably, one A. agrestis isoform (Sc2ySwM_228.5258.1) did show higher sequence similarity to bacterial than plant GRs. We identified similar isoforms in the other sequenced Anthoceros species and strains, namely A. agrestis (Oxford) and A. punctatus [43] (Additional File 8), suggesting a horizontal gene transfer (HGT) that occurred before the split of these species. Notably, these isoforms possess N-terminal extensions compared to bacterial sequences (Additional File 8).
Evolutionary conservation of putative glutathionylation sites on plastid proteins
Many target thiol switches on cyanobacterial or plastidial proteins were acquired early in evolution, while others were reported to have evolved regulatory cysteines later. For example, the C-terminal extension in the plastidial glyceraldehyde-3-phosphate dehydrogenase isoform B (GAPB) evolved in streptophytes and the N-terminal cysteine pair of the plastidial NADPH-dependent malate dehydrogenase in land plants [9, 18, 69].
The vast majority of these enzymes are regulated by TRX through dithiol/disulfide interchanges that induce conformational changes either negatively or positively modulating protein activity [5, 6]. Besides TRX-dependent regulation, S-glutathionylation has recently emerged as an important regulatory mechanism in plants. It is involved in the recycling of antioxidant enzymes, but it can also protect protein cysteines from irreversible oxidation and modulate protein function/activity [3, 61, 70]. Many glutathionylation target proteins and the exact Cys that undergo S-glutathionylation remain unknown (Fig. 2). S-glutathionylation is not routinely detected in proteomics experiments in which cysteines are usually reduced, removing reversible modifications such as S-glutathionylation, and subsequently treated with Cys-blocking agents as a standard modification for MS/MS. However, several proteomic studies have developed specific protocols to detect S-glutathionylation and identified hundreds of putative target proteins highlighting the role of S-glutathionylation as thiol switching regulatory mechanism in eukaryotic oxygenic phototrophs ([3] and references therein). A BioGSSG-based (biotinylated glutathione disulfide) proteomic study was carried out in the cyanobacterium Synechocystis PCC 6803 and 383 proteins were identified as putative S-glutathionylated targets [71]. This study underpins the hypothesis that S-glutathionylation might have a regulatory role in all oxygenic phototrophs. Due to the importance of this post-translational redox modification, we decided to examine its relevance for plastidial proteins by analysing data from proteomic studies that used different methodologies, i.e. biotinylated GSH (biotinylated GSH ethyl ester (BioGEE) [72], biotinylated GSSG (BioGSSG) [73, 74], anti-GSH antibodies [75] or radiolabelling of the glutathione pool using 35S-cysteine [76]. In order to obtain an exhaustive and complete list of S-glutathionylated proteins, we also considered research studies carried out on purified proteins in vitro [72, 77–81, 81–94] (see Additional File 9: Table S2). Combining all these studies, we compiled a list of 364 proteins known to undergo S-glutathionylation in green eukaryotes (Additional File 9: Table S2, Fig. 3).
We determined the subcellular localisation of glutathionylated proteins based on biological function and prediction tools such as TargetP [48, 49] and SUBA (Subcellular Localisation Database for Arabidopsis thaliana [95]. Among the 364 proteins, 151 proteins are localized to plastids (Additional File 9: Table S2, Fig. 3), corresponding to c. 41% of the known plant glutathionylome. Subsequently, we explored for which plastidial proteins the exact site of S-glutathionylation was determined. One proteomic work in the chlorophyte C. reinhardtii combined the identification of glutathionylated targets with streptavidin enrichment using biotinylated-tagged peptides [74]. This approach allows establishing the exact S-glutathionylation sites if the identified peptide contains only one Cys residue. To further extent the analysis, we also considered studies on recombinant plastidial proteins in which the cysteine residues undergoing S-glutathionylation were identified by in vitro oxidant treatments coupled to mass spectrometry analysis [77–83]. Among 151 plastidial proteins, we found 37 glutathionylation sites with known target Cys within 26 different proteins (Table 1 and Additional File 9: Table S2), representing c. 17% of the known plastidial glutathionylome (Fig. 3).
Table 1
Overview of glutathionylation sites on plastid proteins and evolutionary conservation
|
Protein name
|
Cys
|
Org.
|
Cat.
|
Ref.
|
Redox regulation
|
2-Cys peroxiredoxin
|
172
|
Ps
|
yes
|
Calderón et al. 2017 [93]
|
|
Glutaredoxin S12 (GRXS12)
|
29
|
Pt
|
yes
|
Zaffagnini et al. 2012 [74]
|
|
Thioredoxin f (TRX-f)
|
60
|
At
|
no
|
Michelet et al. 2005 [14]
|
Photosynthesis
|
Ferredoxin 1
|
48
|
Cr
|
no
|
Zaffagnini et al. 2012 [74]
|
|
|
69
|
Cr
|
yes
|
Zaffagnini et al. 2012 [74]
|
|
Fructose-1,6-bisphosphatase
|
109
|
Cr
|
no
|
Zaffagnini et al. 2012 [74]
|
|
Fructose-1,6-bisphosphate aldolase
|
58
|
Cr
|
no
|
Zaffagnini et al. 2012 [74]
|
|
Glyceraldehyde-3-phosphate dehydrogenase, A subunit (GAPA)
|
156
|
At
|
yes
|
Zaffagnini et al. 2007 [94]
|
|
Plastocyanin
|
130
|
Cr
|
yes
|
Zaffagnini et al. 2012 [74]
|
|
Phosphoglycerate kinase
|
159
|
Cr
|
no
|
Zaffagnini et al. 2012 [74]
|
|
|
412
|
Cr
|
no
|
Zaffagnini et al. 2012 [74]
|
|
Phosphoribulokinase
|
47
|
Cr
|
no
|
Zaffagnini et al. 2012 [74]
|
|
|
274
|
Cr
|
no
|
Zaffagnini et al. 2012 [74]
|
|
Photosystem II (PSII) core phosphatase (PBCP)
|
168
|
Os
|
no
|
Liu et al. 2019 [80]
|
|
|
176
|
Os
|
no
|
Liu et al. 2019 [80]
|
|
|
195
|
Os
|
no
|
Liu et al. 2019 [80]
|
|
Ribulose bisphosphate carboxylase large chain
|
172
|
Cr
|
no
|
Zaffagnini et al. 2012 [74]
|
|
|
247
|
Cr
|
no
|
Zaffagnini et al. 2012 [74]
|
|
|
427
|
Cr
|
no
|
Zaffagnini et al. 2012 [74]
|
|
Transketolase
|
84
|
Cr
|
no
|
Zaffagnini et al. 2012 [74]
|
|
Triose phosphate isomerase (chloro TPI)
|
15
|
At
|
no
|
López-Castillo et al. 2016 [81]
|
Carbohydrate metabolism
|
ADP-glucose pyrophosphorylase large subunit
|
112
|
Cr
|
no
|
Zaffagnini et al. 2012 [74]
|
|
Alpha-amylase 3 (AMY3)
|
499
|
At
|
yes
|
Gurrieri et al. 2019 [78]
|
|
|
587
|
At
|
yes
|
Gurrieri et al. 2019 [78]
|
|
Beta-amylase 3 (BAM3)
|
433
|
At
|
no
|
Storm et al. 2018 [79]
|
Biosynthesis
|
Acetohydroxy acid isomeroreductase
|
439
|
Cr
|
no
|
Zaffagnini et al. 2012 [74]
|
|
Full-length thiazole biosynthetic enzyme
|
106
|
Cr
|
no
|
Zaffagnini et al. 2012 [74]
|
|
Isopropylmalate dehydratase, large subunit
|
444
|
Cr
|
no
|
Zaffagnini et al. 2012 [74]
|
|
Magnesium-chelatase subunit chlI
|
184
|
Cr
|
no
|
Zaffagnini et al. 2012 [74]
|
others
|
3′-phosphoadenosine 5′-phosphate phosphatase SAL1
|
119
|
At
|
no
|
Chan et al. 2016 [77]
|
|
|
190
|
At
|
no
|
Chan et al. 2016 [77]
|
|
Chaperonin 60B2
|
249
|
Cr
|
no
|
Zaffagnini et al. 2012 [74]
|
|
|
537
|
Cr
|
no
|
Zaffagnini et al. 2012 [74]
|
|
Heat shock protein 70B (HSP70B)
|
349
|
Cr
|
no
|
Michelet et al. 2008 [76]
|
|
Phosphorylase
|
171
|
Cr
|
no
|
Zaffagnini et al. 2012 [74]
|
|
Protein tyrosine phosphatases (PTP)
|
78
|
At
|
no
|
Dixon et al. 2005 [73]
|
|
|
176
|
At
|
no
|
Dixon et al. 2005 [73]
|
Cys = position of identified cysteine, Org. = organism, Cat. = catalytic cysteine, Ref. = literature reference
|
(a) Overview of all known glutathionylated proteins in green eukaryotes with emphasis on the plastidial glutathionylation target proteins. (b) Overview of all known plastidial glutathionylation target proteins with focus on known glutathionylation sites. (c) Overview of all known plastidial glutathionylation sites with classification in evolutionary conserved and non-/partially conserved cysteine sites and subdivision in catalytic and non-catalytic function (for accessions, annotation and references, see Additional File 9: Table S2).
The number of sites exceeds the number of proteins as several glutathionylated proteins contain multiple S-glutathionylation sites (Table 1 and Additional File 9: Table S2). More precisely, seven proteins contain two glutathionylated Cys and two proteins contain three glutathionylated Cys sites (Table 1 and Additional File 9: Table S2). The proteins with identified S-glutathionylation sites are involved in diverse cellular processes such as photosynthesis, carbohydrate metabolism, biosynthetic pathways, redox regulation, signalling and protein homeostasis (Table 1).
To assess the evolutionary conservation of Cys undergoing glutathionylation, we constructed multiple sequence alignments of the 26 plastidial proteins (Additional File 10) from all analysed plant model species. We found that among the 37 known S-glutathionylation sites seven sites were involved in catalytic activity, of which six sites were fully conserved from green algae to flowering plants (Fig. 3, Additional File 10). Thirty glutathionylation sites were identified on non-catalytic cysteines (Fig. 3) of which eight Cys were nevertheless fully conserved in all investigated model species. The remaining 22 Cys were not or only partially conserved (Additional File 10), exhibiting different patterns of evolutionary gains and losses (Fig. 3 and Additional File 9: Table S2).
Five examples of interesting gain and losses of putative S-glutathionylation sites are illustrated in Fig. 4. In alpha-Amylase 3 (AMY3) and SAL1 (3′-phosphoadenosine 5′-phosphate phosphatase) two S-glutathionylation sites were identified in each of the A. thaliana homologs. However, Cys499 in AMY3 is only present in the investigated angiosperm models and Cys119 in SAL1 is only present in A. thaliana, suggesting the gain of these Cys late in land plant evolution. In contrast, Cys106 of thiamine thiazole synthase (THI1) is present in C. reinhardtii and all homologs of investigated land plant models, except for A. thaliana, suggesting a late loss in land plant evolution. Several of the putative S-glutathionylation sites show variable conservation patterns that suggest several independent gains and losses in land plant lineages, such as Cys190 of SAL1, Cys159 and Cys412 in phosphoglycerate kinase (PGK) and Cys49 in ferredoxin.