Abstract
Reactive oxygen species (ROS) play a key role in the regulation of many biological processes in plants. Nonetheless, they are considered highly reactive and toxic to cells. Owing to their toxicity, as well as their important role in signaling, the level of ROS in cells needs to be tightly regulated. The ROS gene network, encoding a highly redundant arsenal of ROS scavenging mechanisms and an array of enzymes involved in ROS production, regulates ROS metabolism and signaling in plants. In this article, we review the role of the ROS gene network in plants and examine how it evolved. We identify key components of the ROS gene network in organisms that likely originated as early as 4.1–3.5 billion years ago, prior to the great oxidation event that resulted from the rise of cyanobacteria on Earth. This estimate concurs with recent evidence for the appearance of oxygenic photosynthetic organisms on Earth, suggesting that low and/or localized levels of photosynthetically produced oxygen necessitated the emergence of ROS scavenging mechanisms to protect life. Life forms have therefore evolved in the presence of ROS on Earth for at least 3.8–3.6 billion years, highlighting the intimate relationship that exists today between many physiological and developmental processes and ROS.
Introduction
Reactive oxygen species (ROS; e.g. O2·−, H2O2, OH·, 1O2) represent partially reduced or excited forms of atmospheric oxygen that can react with different cellular components and cause their oxidation (Fridovich, 1978; Halliwell and Gutteridge, 2015). ROS are continuously formed in different compartments of the cell as an unavoidable byproduct of aerobic metabolism, but are also formed as a result of activation of specific ROS-producing enzymes that mediate signal transduction processes (e.g. Foyer and Noctor, 2013; König et al., 2012; Mittler, 2002). In plants, the major ROS-producing compartments are the chloroplast, which mainly produces O2·−, H2O2, and 1O2 as a byproduct of photosynthesis; the mitochondrion, which mainly produces O2·− and H2O2 as a byproduct of respiration; the peroxisome, which mainly produces H2O2 as a byproduct of photorespiration; and the apoplast, which mainly produces O2·− (and subsequently H2O2) as key signaling molecules generated by NADPH oxidases termed respiratory burst oxidase homolog (RBOH), peroxidases, and superoxide dismutases (SODs; Mignolet-Spruyt et al., 2016; Mittler et al., 2011; Vaahtera et al., 2014). Although ROS were originally thought to be toxic byproducts of aerobic metabolism that must be removed to prevent the oxidative destruction of the cell, more recent studies revealed that ROS are used by most organisms as key signal transduction molecules (Mittler et al., 2004). Even more recently, it was discovered that a basal level of ROS is actually required to support life (reviewed in Mittler, 2016). The quantity of ROS in cells therefore needs to be maintained above a low cytostatic level, but below a high cytotoxic level (Schieber and Chandel, 2014). This allows ROS to function safely as signal transduction molecules that mediate many different processes in cells, including the regulation of metabolic pathways, physiological processes such as stomatal responses, activation of acclimation responses to abiotic stress, activation of defense responses against different pathogens and pests, activation of different developmental programs, and coordination of systemic plant responses to different environmental stimuli (e.g. Frederickson Matika and Loake, 2014; Mittler and Blumwald, 2015; Noctor et al., 2014; Scheler et al., 2013; Schmidt and Schippers, 2015; Song et al., 2013; Wendehenne et al., 2014). Because ROS can directly affect the oxidation state of proteins, for example, by oxidizing cysteine or methionine residues, they can alter the function of different proteins and affect the activity of various protein phosphorylation relays, as well as transcription factors and other regulatory proteins (e.g. Akter et al., 2015; König et al., 2012; Reczek and Chandel, 2015; Truong and Carroll, 2013; Waszczak et al., 2015). This process is generally termed ‘redox biology’ and serves as an interface between ROS and the different biological processes they control. Because ROS production by, for example, RBOHs is directly linked to changes in cellular calcium levels and/or protein phosphorylation events (Gilroy et al., 2014; Suzuki et al., 2011), ROS production can be directly linked to the function of many different receptors, hormones, and signal transduction and developmental events that alter calcium levels and/or protein phosphorylation. Activation of ROS production, in turn, affects cellular calcium levels and/or protein phosphorylation events, placing ROS as upstream as well as downstream signaling molecules. Recent studies have linked ROS signaling with different plant hormones, such as abscisic acid, salicylic acid, and nitric oxide, suggesting that ROS can also act as a bridge between different plant hormones, redox biology, and changes in cellular calcium levels and/or protein phosphorylation events. This view of ROS as indispensable for plant signaling and metabolism should come as no surprise, considering that life on Earth most likely evolved in the presence of ROS (e.g. Boyd et al., 2014; Miller, 2011; Mittler, 2016; Wood, 2003). Controlling the overall levels of ROS in cells and allowing ROS to exert their beneficial functions is mediated by constantly balancing the rates of ROS production and scavenging in the different cellular compartments (Mittler et al., 2004). This process yields a particular ROS signature in cells that changes based on the developmental state of the cell and the environmental conditions it encounters (Choudhury et al., 2016). The particular ROS signature of the cell, reflecting the compiled ROS levels in all the different cellular compartments, will in turn activate different cellular responses via redox reactions with different proteins and adjust metabolism, development, and acclimation/defense pathways. Key to the control of ROS levels in cells, the generation of spatial and temporal ROS signatures, and the activation of different cellular responses is the ROS gene network (Mittler et al., 2004). In this review, we focus on the distribution and evolution of the ROS gene network, with an emphasis on superoxide dismutase (SOD), ascorbate peroxidase (APX), and NADPH oxidase (NOX)/RBOH proteins.
The ROS gene network
The ROS gene network of plants includes all the genes that encode ROS-detoxifying and ROS-producing proteins in the cell. In Arabidopsis, the basic ROS gene network is composed of over 150 genes (Mittler et al., 2004). These include genes encoding iron (Fe)-, manganese (Mn)-, and copper–zinc (CuZn)-SODs, which scavenge O2·−; APXs, catalases (CATs), glutathione peroxidases (GPXs), peroxiredoxins (PRXR), and other peroxidases, which scavenge H2O2; enzymes such as monodehydroascorbate reductase, dehydroascorbate reductase, and glutathione reductase, which participate in reducing/recycling the antioxidants ascorbic acid and glutathione, used in the removal of H2O2 and O2·−; and thioredoxins and glutaredoxins, which participate in different redox reactions (Fig. 1; Mittler et al., 2004). In addition, the ROS gene network of plants includes genes that encode RBOHs (10 genes in Arabidopsis) and other ROS-producing enzymes such as xanthine, glycolate, and oxalate oxidases. The various proteins encoded by the ROS gene network are localized to the different cellular compartments of the cell and mediate ROS production and scavenging. Many of the enzymes encoded by the ROS network function in defined pathways for ROS removal, such as the SOD–Asada–Foyer–Halliwell pathway (also termed the SOD–ascorbate–glutathione pathway) found in chloroplasts, mitochondria, cytosol, and peroxisomes (Fig. 1). In addition to the genes encoding the proteins described above, the ROS gene network includes genes that belong to several biosynthetic pathways that mediate the production and degradation of different low molecular weight antioxidants, such as ascorbic acid, glutathione, α-tocopherol, carotenoids, flavonols, and other phenolic compounds. Together, the high subcellular levels of antioxidants, the presence of multiple ROS detoxification enzymes and proteins in all compartments of the cell, and the high redundancy in gene and enzymatic function within the network provide plant cells with a high level of protection against ROS toxicity. Additional enzymes and proteins that play a key role in the network are those that control the level of labile iron in cells. Because iron (Fe2+) can react with H2O2 (or dismutated O2·−) to form the highly reactive hydroxyl radical (OH·; the Fenton reaction; Halliwell and Gutteridge, 2015), the level of labile iron in cells must be kept under tight control. This is especially critical under conditions of enhanced O2·− production because O2·− can react with Fe-S proteins and cause the release of labile iron from them. Thus, proteins such as ferritin that store iron, different iron uptake systems, and different Fe-S biogenesis and repair systems play a key role in protecting the cell from ROS damage (Cabantchik, 2015; Gammella et al., 2016; López-Millán et al., 2016). Of course, many more regulatory proteins and transcription factors involved in ROS metabolism and signaling need to be included in the ROS gene network (Gadjev et al., 2006; Mittler et al., 2004; Willems et al., 2016). It is plausible that the number of genes that comprise the extended ROS gene network could exceed 250–300 once all regulatory, biosynthetic, degradation, uptake, and repair genes have been added to the basic ROS gene network (Mittler et al., 2004). In our current discussion of the evolution of the ROS gene network we will, however, focus on only a limited number of key genes (Fig. 2).
Major superoxide- and H2O2-scavenging pathways of plants. Superoxide dismutase and the major H2O2-scavenging and antioxidant-recycling pathways of plants are shown, with the number of genes encoding each enzyme in Arabidopsis indicated. The SOD–Asada–Foyer–Halliwell (SOD–ascorbate–glutathione) pathway is indicated in bold. Adapted and modified from Mittler et al. (2004).
Occurrence of ROS-producing and ROS-scavenging proteins in different species. A common tree of species was obtained from NCBI Taxonomy (http://www.ncbi.nlm.nih.gov/Taxonomy/CommonTree/wwwcmt.cgi). The tree was populated with one representative fully sequenced genome on each of its terminal branches, indicative of the different lineages (obtained from NCBI Genome; https://www.ncbi.nlm.nih.gov/genome/browse/; Supplementary Table S2). Each genome represented on the species tree was individually subjected to a PSI-BLAST search (Altschul et al., 1997) following translation to all six reading frames with PSI-BLAST threshold of 5 and Expect value of 10, with the following as queries: Arabidopsis thaliana proteins (RBOHD, respiratory burst oxidase homolog D, AT5G47910.1; FSD1, Fe-SOD, AT4G25100.3; MSD1, Mn-SOD, AT3G10920.1; CSD1, CuZn-SOD, AT1G08830.1; CAT1, catalase, AT1G20630.1; APX1, ascorbate peroxidase 1, AT1G07890.1; GPX, glutathione peroxidase, AT2G48150.1; and PRXR, peroxiredoxin, AT1G48130.1; Supplementary Table S3), human proteins (NOX, NADPH oxidase, NP_008983.2 and DUOX, AAI14939.1), and the Saccharomyces protein CCP, cytochrome C peroxidase, CAA44288.1. The presence or absence of the respective genes in the genomes is indicated on the right for each organism (red, present; green, absent). Additional criteria for the detection of homologs were 60% or more of the query coverage and at least 25% amino acid identity. Protein homologs that could not be unambiguously classified as FSD or MSD or APX or CCP (i.e. its similarity to each of these proteins was below 50%) are marked with an asterisk (representing potentially hybrid FSD/MSD or APX/CCP proteins). Organisms that perform aerobic photosynthesis are highlighted with a green background.
Occurrence of ROS-producing and ROS-scavenging proteins in organisms from different lineages
To obtain an evolutionary perspective of ROS-metabolizing enzymes, we determined the presence or absence of key ROS-metabolizing proteins in organisms from different lineages as they appear on a common tree of species obtained from NCBI Taxonomy (Fig. 2). We divided our analysis into genes encoding ROS-producing [RBOH, dual NADPH oxidase (DUOX), and NOXs] and ROS-scavenging [SODs, CAT, APX, cytochrome C peroxidase (CCP), GPX, and PRXR] enzymes, and indicated whether these genes are present or absent in the completely sequenced genomes of organisms representing different lineages on the species tree (Fig. 2). The presence of an RBOH or a NOX-type enzyme-coding gene was previously thought to be associated with increased organism complexity and multicellularity, reflecting the key role of ROS signaling in the development and maintenance of multicellular organisms (Lalucque and Silar, 2003; Lambeth, 2004; Sumimoto, 2008). Indeed, with the exception of brown algae, all other multicellular organisms included in our analysis contained at least one copy of an RBOH or NOX-type protein-coding gene. It is currently unknown how brown algae can generate multicellular structures without an NOX/RBOH homolog, and this question should be addressed in future studies. Owing to its early evolutionary origin (Miller, 2011), Fe-SOD could be found in archaea, bacteria, algae, and plant chloroplasts, but it is absent in organisms such as animals or fungi. Fe-SOD was most likely lost in these latter organisms during the course of evolution, and its function replaced by Mn- and CuZn-SOD. Aside from SODs, the other highly conserved ROS-removal enzymes are PRXRs and CATs; all organisms included in our analysis contained at least one copy of PRXR- and/or CAT-encoding genes. Interestingly, APX, which was traditionally considered to be a plant enzyme, was found in several other organisms, including hydra, sea shell (Lingula), sponge, fungi, and brown and red algae (Fig. 2). In addition, CCP, which was originally identified in fungi, was also found in some plant species and in green, red, and brown algae. In contrast to plant, algae and bacterial genomes, mammalian, reptile, avian, fish, amphibian, and insect genomes contained a very similar set of genes encoding ROS-producing (NOX/DUOX) and ROS-scavenging (MnSOD, CuZn-SOD, CAT, GPX and PRXR) proteins, suggesting that this set of proteins is sufficient for all different functions and developmental stages of these organisms. The lineages of SOD, APX/CCP, and RBOH/NOX are discussed in more detail below.
Early evolution of the ROS gene network
The analysis presented in Fig. 2 highlights SOD (in particular Fe-SOD), CAT, and PRXR as the most ancient among the different ROS gene network proteins. This finding prompted us to examine how ancient these proteins are. In addition, we wanted to find out whether they evolved before or during the great oxidation event (Anbar, 2008; Boyd et al., 2014) that resulted from the rise of oxygen-evolving photosynthetic cyanobacteria on Earth about 2.5–2.2 billion years ago. To address these questions we conducted phylogenetic analyses of these proteins in different unicellular organisms and identified the putative most ancient unicellular origins of Fe-SOD (Candidatus Nanosalinarum), PRXR (Nanoarchaeota archaeon), CAT (Methanolacinia paynteri and/or Methanoplanus limicola), and Class I catalase-peroxidase (CAT/PER; Thermovibrio aminoficans) (Fig. 3; Supplementary Figs S1–S4 at JXB online). These data were then placed in context with the evolution of different life forms on Earth, the great oxidation event, and the different geological periods (Fig. 3; Battistuzzi et al., 2004). As shown in Fig. 3, SOD, CAT, and PRXR can be found in organisms that appeared on Earth about 4.1–3.5 billion years ago. This timing is in accordance with recent estimates for the earliest appearance of oxygen-evolving photosynthetic organisms on Earth (about 3.8 billion years ago; Blankenship, 2010; Fischer et al., 2016; Nelson and Junge, 2015; Rosing and Frei, 2004; Shaw, 2015; Shih et al., 2016; Ward et al., 2015), and at least 1 billion years prior to the great oxidation event. The timescale presented in Fig. 3 highlights the important role ROS-scavenging mechanisms might have played in the evolution of life on Earth. Thus, low but localized levels of photosynthetically produced oxygen (which did not significantly accumulate in the atmosphere 3.8 billion years ago; Anbar, 2008; Boyd et al., 2014; Fischer et al., 2016; Hohmann-Marriott and Blankenship, 2011; Holland, 2009; Nelson and Junge, 2015; Rosing and Frei, 2004; Ward et al., 2015) necessitated the evolution of ROS-scavenging mechanisms to protect life. Moreover, without the prior evolution of ROS-scavenging mechanisms by different organisms it is highly likely that most life forms on Earth would have perished during the great oxidation event. The evolution of life on Earth therefore occurred in the presence of ROS for at least 3.8–3.6 billion years, highlighting the intimate relationships that exist today between many physiological and developmental processes and ROS. Moreover, many milestones in the evolution of life on Earth, such as the appearance of eukaryotic organisms and the endosymbiotic events that resulted in the formation of the chloroplast and mitochondrion, might therefore have occurred in the presence of ROS.
Early evolution of the ROS gene network. Top: Geological timescale of Earth, adapted from the International Commission on Stratigraphy (http://www.stratigraphy.org/). Middle: Evolution of ROS-producing and ROS-scavenging proteins in response to increasing oxygen concentration in the biosphere. The appearance of microorganisms that contain early versions of several ROS gene network proteins was placed in the context of concentrations of iron and oxygen in the primordial oceans (adapted from Anbar, 2008). To obtain the estimated time of evolution of superoxide dismutase, catalase, peroxiredoxin, ferric reductase/oxidoreductase, and catalase/peroxidase, the origin of each organism (taxonomic class/order) that contained the most ancestral form of these proteins in our phylogenetic analysis (Supplementary Figs S1–S5) was traced back using the TimeTree of Life (http://www.timetree.org). Bottom: An estimated timeline of major biological/metabolic events on Earth (adapted from Battistuzzi et al., 2004; Rosing and Frei, 2004). The estimated earliest origin of life on Earth is around 4.5 billion years ago, while the first major occurrence of atmospheric oxygen, represented as the ‘great oxidation event’, happened approximately 2.3–2.6 billion years ago (Anbar, 2008; Boyd et al., 2014). ‘Colonization of land’ refers to the appearance of microorganisms such as the Actinobacteria, Cyanobacteria, and Deinococcus, capable of growing on dry land (Battistuzzi et al., 2004). Horizontal lines represent credibility intervals, white boxes indicate minimum and maximum time constraints on the origin of the metabolic event, and colored boxes indicate the duration or presence of the respective metabolic event as shown by Battistuzzi et al. (2004).
In an attempt to determine whether NOX-like proteins also appeared early in evolution, we identified bacterial homologs of the most conserved form of NOX (represented in our analysis by Dictyostelium discoideum; Supplementary Fig. S5) and mapped the appearance of organisms containing these proteins (oxidoreductase and ferric-reductase; e.g. Cylindrospermum stagnale; Supplementary Fig. S6) on to the chart shown in Fig. 3. These putative ancestral forms of NOX appeared to have evolved about 4.1–3.5 billion years ago, succeeding the evolution of SOD, CAT, and PRXR, and coinciding with the evolution of CAT/PER. It thus appears that the earliest form of ROS protection was mediated by Fe-SOD coupled with CAT and/or PRXR, and that NOX-like and CAT/PER proteins evolved later. Based on the origin of organisms as illustrated in Fig. 3, it seems likely that the origins of ROS-related proteins preceded the great oxidation event, and was most likely key to the survival of organisms during this planet-altering event (Fig. 3). Although we identified ancestral forms of ROS-metabolizing genes in genomes of organisms that appeared on Earth billions of years ago (Fig. 3), care should be exercised in interpreting these findings because it is not known at what exact point these genes appeared in those organisms’ genomes. In addition, the function of many of the proteins thought to be found in ancestral genomes is unknown and could be predicted only on the basis of sequence homology.
Superoxide dismutase
SODs (EC 1.15.1.1; Fridovich, 1975; Fig. 4); catalyze the dismutation of O2·− into H2O2 (2 O2·− + 2 H+ −> O2 + H2O2). The term SOD actually describes three different enzymes with different structures: a Fe-, Mn- and/or Fe/Mn-SOD, a Ni-SOD, and a CuZn-SOD (Fig. 4A). The most ancient of the three types is Fe-SOD, which is thought to have evolved even before the differentiation of eubacteria from archaea (Miller, 2011). Fe-SOD and Mn-SOD are very similar in structure, and some forms of Fe-SOD can actually function with a Mn ion replacing the Fe (Fig. 4A; Edward et al., 1998; Joshi and Dennis, 1993; Miller, 2011; Muñoz et al., 2005; Perry et al., 2010). It is thought that, due to the high availability of Fe in the primordial oceans, Fe-SOD evolved first; once the availability of Fe declined, Mn replaced Fe and the enzyme evolved to function with Mn (Anbar, 2008; Bafana et al., 2011; Huaiyang and Wangyi, 1996; Miller, 2011). In plants, Fe-SOD is found in the chloroplast and Mn-SOD in the mitochondrion, a remnant of the bacterial origins of these organelles (Fig. 2). Mn-SOD is also found in the mitochondria of most multicellular organisms (Fig. 2). In addition to Fe- and Mn-SOD, plants contain CuZn-SOD in both the cytosol and chloroplast (Fink and Scandalios, 2002). CuZn-SOD is found in many different bacteria and in highly evolved multicellular organisms (Fig. 2; Bannister et al., 1991; Bloomfield, 2003). The third type of SOD, Ni-SOD, is mainly found in a selected group of Gram-negative bacteria and cyanobacteria, and is hypothesized to have independently emerged much later in evolution, after the differentiation of eukaryotes and a substantial diversification among bacteria (Wuerges et al., 2004). The fact that SODs have independently evolved on at least three separate occasions (Fe/Mn-SOD, CuZn-SOD, and Ni-SOD) highlights the important role that SOD enzymatic activity plays in different biological systems (Miller, 2011). The finding of Fe-, Mn- and CuZn-SODs in mosses, spruce, and higher plants (Fig. 2) further reinforces the importance of SODs in protecting photosynthetic multicellular organisms. A phylogenetic tree for SODs constructed for the organisms shown in Fig. 2 highlights the ancestral link between Mn- and Fe-SODs (Fig. 4B). In addition, it demonstrates that chloroplastic CuZn-SOD is distinct from cytosolic and peroxisomal CuZn-SODs, suggesting that it could have originated earlier (Fig. 4B).
Superoxide dismutase. (A) Three-dimensional structures of (i) superimposed Fe-SOD (1isb.pdb) and Mn-SOD (1vew.pdb) of Escherichia coli demonstrate a high degree of homology (Green, Fe-SOD; Red, Mn-SOD; Purple sphere, Mn ligand; Green sphere, Fe ligand; Red sphere, OH attached to Mn); (ii) Spinach CuZn-SOD (1srd.pdb); (iii) Streptomyces Ni-SOD (1q0d.pdb). The Protein Data Bank (PDB) structures of all the proteins were downloaded from RCSB.org (http://www.rcsb.org/pdb/home/home.do). The structures were superimposed based on protein topology, i.e. on the orientation of strands and helices, using WinCOOT 0.8.3 (Emsley et al., 2010). The images in (i), (ii), and (iii) were generated using Discovery Studio 4.1 (Discovery Studio, 2011). (B) Representative phylogenetic tree of plant SOD genes. For simplification purposes, the different clades were collapsed based on the type of SOD. Plant SOD protein sequences in the NCBI Protein database (http://www.ncbi.nlm.nih.gov/protein) were identified using the PSI-BLAST algorithm with PSI threshold value of 5 and Expect threshold of 10; Arabidopsis FSD1, Fe-SOD, AT4G25100.3; MSD1, Mn-SOD, AT3G10920.1; and CSD1, CuZn-SOD, AT1G08830.1 were used as query sequences. The sequences were then aligned using MUSCLE (MUltiple Sequence Comparison by Log-Expectation; Edgar, 2004). The multiple sequence alignment was treated with GBlocks and all columns with more than 95% gaps were eliminated. A 100 bootstrap maximum-likelihood phylogenetic tree was then generated using the PhyML program (Guindon et al., 2010). To estimate the optimal model of substitution, ProtTest 3.0 was used for the aligned proteins (Darriba et al., 2011). The tree was finally edited with FigTree 1.4.0 software (Rambaut, 2012). A full version of the tree with complete protein annotations and bootstrap confidence values is provided as Supplementary Fig. S7.
Ascorbate peroxidase
APX (EC 1.11.1.11; Asada, 2000) was initially isolated from various plant tissues and green algae, and was considered to be a plant-specific protein. APX is an ascorbic acid-specific class I peroxidase that catalyzes the reaction 2 L-ascorbate + H2O2 + 2 H+ −> 2 monodehydroascorbate + 2 H2O. It functions mainly as part of the Asada–Foyer–Halliwell pathway (Fig. 1) or the water–water cycle, and is required for cellular protection against ROS (Asada, 2000; Noctor and Foyer, 1998). In recent years a key signaling role for several types of APXs has been identified, especially in responses to abiotic stress and systemic signaling (Davletova et al., 2005; Suzuki et al., 2013). The amino acid sequence of APX is highly homologous to that of fungal CCP, and recent studies have identified hybrid APX/CCP proteins (Fig. 5; Zámocký et al., 2010). Despite this similarity, APX is unable to use cytochrome c as a substrate, and CCP is unable to use ascorbic acid as its substrate (Lad et al., 2002; Macdonald et al., 2006; Meharenna et al., 2008; Poulos, 2010). Following the crystallization of APX, its similarity to CCP was also noted at the structural level (Fig. 5B; reviewed in Raven, 2000; Sharp et al., 2003). As indicated above, the occurrence of APX is not limited to plants and green algae, and homologs of this protein can be found in hydra, sea shell, sponge, fungi, and brown and red algae (Fig. 2). An APX/CCP phylogenetic tree constructed for the organisms shown in Fig. 2 reveals that hydra, sea shell and sponge APXs are distinct from all other APXs and could represent an ancestral form of this protein (Fig. 5C). Chloroplastic, cytosolic, and peroxisomal APXs from higher plants form a distinct clade, while fungal CCP, fungal and diatom APX, and hybrid APX/CCP from green, red and brown algae and diatoms form a separate clade. Interestingly, APX 6 and APX 4 of Arabidopsis, moss, diatom, and green, brown, and red algae appear to be the most ancient plant-type APXs, and could represent a common ancestor for APXs and CCPs. APX 4/6 are also distinct from chloroplastic, cytosolic, and peroxisomal APXs (Fig. 5C). Because APXs require L-ascorbic acid to detoxify H2O2, the occurrence of L-ascorbic acid should be determined in the different organisms that appear to contain different forms of APX (Fig. 2; Gest et al., 2012) in order to confirm the biological function of APX. The distribution of APXs among different lineages suggests that this protein evolved early during evolution and that it was likely lost in certain organisms, probably replaced by GPX, which uses the more common glutathione in place of ascorbate for its reaction (Fig. 1). The loss of APX might have been tied to the loss of ability to synthesize or regenerate L-ascorbic acid by different organisms (e.g. humans), although rats maintained the pathway for ascorbate biosynthesis but lack APX.
Ascorbate peroxidase. (A) Classification of peroxidases. Adapted and modified from Zámocký et al. (2010). (B) Three-dimensional structures of (i) Arabidopsis APX (1apx.pdb), (ii) yeast CCP (2cyp.pdb), and (iii) superimposed APX and CCP demonstrate a high degree of homology between APX and CCP (Green, Arabidopsis APX; Red, yeast CCP). The PDB structures of both proteins were downloaded from RCSB.org (http://www.rcsb.org/pdb/home/home.do). The structures were superimposed as described for Fig. 4A. (C) Representative phylogenetic tree of APX and CCP proteins. For simplification purposes, the different clades were collapsed based on the type of APX/CCP. Homologs of APX and CCP in the NCBI Protein database were identified using the PSI-BLAST algorithm with PSI threshold value of 5 and Expect threshold of 10; Arabidopsis APX1 (AT1G07890.1) and Saccharomyces CCP (CAA44288.1) were used as query sequences (Altschul et al., 1997). The sequences were then aligned and a maximum-likelihood phylogenetic tree generated and edited as described for Fig. 4B. A full version of the tree with complete protein annotations and bootstrap confidence values is provided as Supplementary Fig. S8.
Respiratory burst oxidase homologs
RBOHs are membrane-bound calcium- and phosphorylation-dependent NADPH oxidases (EC 1.6.3.1) that catalyze the formation of superoxide using reducing power provided by NADPH (Lambeth, 2004; Sumimoto, 2008). They belong to a large gene family also known as NOX, which includes mammalian, insect, reptile, avian, fish, and amphibian members. The NOX family can be roughly divided into NOX, DUOX, and RBOH proteins (Fig. 2), all of which play a key signaling role via the regulated production of ROS (Lambeth, 2004; Sumimoto, 2008). The core structure of the NOX/RBOH family proteins includes a transmembrane domain that harbors two heme groups and a long NADP-binding cytoplasmic C-terminal (Fig. 6A). The N-terminal of NOX/RBOH proteins can contain a series of EF-hand calcium binding domains, as well as multiple phosphorylation sites and, in the case of DUOX, an entire peroxidase domain (Sumimoto, 2008). In mammalian cells, additional subunits join the core protein structure to generate a highly regulated protein complex. These subunits include p22, p40, p47, and p67 (Fig. 6A; Sumimoto, 2008). In addition, a regulatory Rac GTPase is associated with all types of NOX/RBOH (Fig. 6A; Wong et al., 2007). The basic evolutionary path of the NOX/RBOH family is shown in Fig. 6B. The ancestral NOX protein that most likely evolved in unicellular organisms is shown to have generated at least six types of NOX/RBOH proteins (Han, 1998; Hata et al., 1997; Sumimoto, 2008; Takemoto et al., 2007). These include fungal NOX protein, which could be associated with a p67 homolog, mammalian NOX4, which is associated with p40 and p67, and mammalian NOX1–3, which are associated with p22 and p47 (Han, 1998; Hata et al., 1997; Sumimoto, 2008; Takemoto et al., 2007). In addition, the ancient NOX protein acquired an EF-hand domain to generate mammalian NOX5, which is associated with p40 and p67 (Hata et al., 1997; Sumimoto, 2008). NOX5 with its EF-hand domain is also very similar to plant RBOH proteins, but RBOHs are not known to associate with any of the type p22, p40, p47, and p67 homologs, and bind only the Rac GTPase at their N terminal (Hata et al., 1997; Oda et al., 2010; Sumimoto, 2008). Finally, following the acquisition of a peroxidase transmembrane domain by the NOX5-type NOX ancestor, it evolved to become the DUOX protein (Fig. 6B; Sumimoto, 2008). A phylogenetic analysis of all NOX/RBOH proteins included in Fig. 2 revealed that the Dictyostelium discoideum NOX (Lardy et al., 2005) is closest in sequence to the likely ancestor of NOX/RBOH proteins (Supplementary Fig. S5). Because this organism transitions between a free-living unicellular amoeba and a multicellular structure-forming organism (Lardy et al., 2005), the finding of an ancestral NOX homolog in this organism emphasizes the key role that NOX/RBOH might have played in the evolution of multicellular organisms (Lalucque and Silar, 2003). A phylogenetic tree of selected plant and algae RBOHs (Fig. 2) is shown in Fig. 6C. According to this analysis, green algae RBOH is the closest to the ancient plant RBOH, followed by spruce RBOH. In addition, Arabidopsis RBOHH and RBOHJ form a distinct clade together with moss RBOHs that could also represent an ancient form of this protein. Two additional distinct clades are also evident, one containing Arabidopsis RBOHI, F, and E, and the other Arabidopsis RBOHA, C, B, G, and D. At least three distinct groups of plant RBOHs can therefore be found in plants, highlighting the variability in their expression and function. In Arabidopsis, for example, RBOHD is involved in mediating systemic signaling (Miller et al., 2009), RBOHD and RBOHF coordinate leaf stomatal (Kwak et al., 2003) and pathogen (Torres et al., 2001) responses, and RBOHC regulates the development and growth of root hairs (Foreman et al., 2003). Although the importance of RBOH proteins is thought to be paramount to plant development and response to the environment (Suzuki et al., 2011), the possible redundancy in RBOH gene function and the large number of genes encoding RBOH proteins in plants (e.g. 10 in Arabidopsis) complicate the study of this important gene family.
RBOH and NOX proteins. (A) Domain and subunit architecture of RBOH/NOX proteins. The basic domain common to all RBOH/NOX proteins is the major transmembrane domain, which includes two heme groups (Fe) and a long NADP-binding cytoplasmic C-terminal (1). The NOX5 and RBOH isoforms have an N-terminal extension that contains two to four EF-hand Ca2+-binding sites (2). The DUOX isoform contains the EF-hand domain but has an N-terminal transmembrane domain as well as an extracellular domain with peroxidase homology (3). NOX1–3 isoforms form a complex with the p22 and p47 proteins (4), and NOX4 forms a complex with the p67 and p40 proteins (5). All NOX/RBOH proteins are associated with Rac GTPase. (B) Classification of NOXs adapted and modified from Bedard et al. (2007). (C) Phylogenetic tree of RBOH proteins. Homologs of RBOH and NOX proteins in the NCBI Protein database were identified using the PSI-BLAST algorithm with PSI threshold value of 5 and Expect threshold of 10; Arabidopsis RBOHD (AT5G47910.1) was used as the query sequence. The sequences were then aligned and a maximum-likelihood phylogenetic tree was generated and edited as described in Fig. 4B. A full version of the tree with complete protein annotations and bootstrap confidence values is provided as Supplementary Fig. S9.
Conclusions
The origin of the ROS gene network is rooted deep in the primordial history of life on our planet, with enzymes such as SOD, CAT, and PRXR likely originating as early as 4.1–3.6 billion years ago (Fig. 3). Because the majority of life as we know it on Earth evolved in the presence of ROS, the evolution of ROS-metabolizing enzymes was most likely directed by the needs of the different evolving organisms, and each step of increasing complexity and altered metabolism had to find a suitable solution to maintain ROS within the redox biology range before it could be established (Mittler, 2016). Once a solid network that could maintain ROS at the proper redox levels was established, ROS were recruited to function as signal transduction messengers (Mittler et al., 2011). Because ROS can be produced in cells as a result of the effects of stress on metabolism, it is plausible that ROS initially functioned as signal transduction molecules that alerted the cell to the occurrence of stress in its environment (Mittler et al., 2011). Because different cells produce different levels of ROS depending on their metabolic and genetic make-up, a later evolutionary stage could have involved the use of ROS to sense neighboring or invading cells (Mittler et al., 2011), initiating one of the currently known roles of ROS in signaling cell-to-cell communication (Miller et al., 2009). Because of their early evolution and involvement in almost all biological systems on Earth, ROS could be considered to be one of the earliest signal transduction molecules to have emerged. In this context, it should be noted that, due to their varying molecular properties, which include features such as degree and mode of reactivity, diffusion rate, and ability to cross membranes, different types of ROS could convey or carry different signals in cells (Foyer and Noctor, 2013; König et al., 2012; Mignolet-Spruyt et al., 2016; Vaahtera et al., 2014). Given their highly variable properties, ubiquity in cells, and ancient origins, new signaling roles for ROS are likely to be discovered in future studies. Moreover, future research may also identify new and yet unknown proteins that scavenge or produce ROS (e.g. Luhua et al., 2008, 2013). These will of course be added to the ROS gene network and enhance our understanding of ROS biology. Although very old in origin, the future of ROS research is highly exciting!
Supplementary data
Figure S1. Phylogenetic tree of archaea and bacteria Fe-SOD.
Figure S2. Phylogenetic tree of archaea and bacteria peroxiredoxin.
Figure S3. Phylogenetic tree of archaea and bacteria catalase.
Figure S4. Phylogenetic tree of archaea and bacteria peroxidase.
Figure S5. Phylogenetic tree of NOX-DUOX-RBOH.
Figure S6. Phylogenetic tree of archaea and bacteria oxidoreductase-ferric reductase.
Figure S7. Phylogenetic tree of plant SOD.
Figure S8. Phylogenetic tree of APX and CCP.
Figure S9. Phylogenetic tree of RBOH.
Table S1. The ROS gene network.
Table S2. List of fully sequenced genomes used for the phylogenetic analysis.
Table S3. List of ROS proteins with accession numbers used to generate phylogenetic trees.
Author contribution
MI and SSG performed the meta-analysis of publicly available data. MI, SSG, RKA, ARD, and RM wrote the manuscript.
Conflict of interest
The authors declare no conflict of interest.
Abbreviations:
- APX
ascorbate peroxidase
- CAT
catalase
- CAT/PER
catalase/peroxidase
- CCP
cytochrome c peroxidase
- DUOX
dual NADPH oxidase
- GPX
glutathione peroxidase;
- NOX
NADPH oxidase
- PRXR
peroxiredoxin
- RBOH
respiratory burst oxidase homolog
- ROS
reactive oxygen species
- SOD
superoxide dismutase.
Acknowledgments
This work was supported by funding from the National Science Foundation (IOS- 1353886, IOS-0639964, IOS-0743954, IOS-1557787, MCB-1613462), and the University of North Texas, College of Arts and Sciences. The funders had no role in the design, data collection, analysis, decision to publish, or preparation of the manuscript.
References
Author notes
* These authors contributed equally to this work.
† Correspondence: [email protected]
Editor: Donald Ort, University of Illinois
Comments