Reactive oxygen species (ROS) are damaging, highly reactive, reduced forms of oxygen, produced in a wide range of physiological processes, including as a by-product of aerobic respiration in the mitochondria, and following exposure of cells to environmental agents such as UV light and drugs/xenobiotics (Gutteridge and Halliwell, 1999). Over recent years, rather than simply acting as toxic agents, it has been established that less-reactive ROS, particularly H₂O₂, also play positive roles as signalling molecules. Key to ROS-signalling functions is the ability to selectively regulate the activity of specific proteins, for example, by oxidising deprotonated cysteine residues, such as are found in the catalytic sites of many enzymes. This selective reactivity is one of the features making H₂O₂ well-suited as a signalling molecule (Winterbourn, 2008). Indeed, H₂O₂ signals have been shown to promote diverse biological responses, from stomatal opening in plants to cell migration in wound healing [For reviews see (Veal and Day, 2011)] and (Holmstrom and Finkel, 2014). However, the most fundamental and universal response to increases in ROS is to limit organismal damage by increasing the levels of detoxification and repair enzymes or initiating apoptosis. Many of these protective responses are mediated by MAPK pathways (Fig. 1). As we discuss below, redox-active cysteines have been shown to play an important part in the activation of these pathways.

Cells have evolved a variety of enzymes which remove ROS before they are able to cause oxidative damage (Bokov et al., 2004; Gutteridge and Halliwell, 1999). Redox-active cysteines are involved in the activity of many of these enzymes. For example, peroxiredoxins (Prx) are a family of extremely abundant, thioredoxin peroxidase enzymes. Prx are themselves highly reactive with H₂O₂ using the reversible oxidation of cysteine residues to reduce peroxides, including H₂O₂. During the catalytic mechanism of typical 2-Cys Prx, the peroxidatic cysteine becomes oxidised to a sulphenic acid, which then reacts with the resolving cysteine of a neighbouring Prx molecule, forming an intermolecular disulphide bond. The resulting Prx disulphides are reduced by thioredoxin using electrons from NADPH (Fig. 2A) [For a review see (Wood et al., 2003)]. Thioredoxin family proteins are oxidoreductases with broad substrate specificity, acting as cofactors in the activity of many other enzymes and participating in the reduction of many transcription factors and signalling molecules (Holmgren and Lu, 2010). However, the abundance of Prx and the high affinity of thioredoxin for Prx disulphides can render Prx disulphides the prevalent thioredoxin substrate when H₂O₂ levels increase (Day et al., 2012). Hence, the activities of Prx and thioredoxin in H₂O₂ signalling are inextricably linked. Here we will discuss the different roles that have been identified for Prx and thioredoxin in the regulation of mitogen activated protein kinase (MAPK) pathways. In some cases, Prx and thioredoxin have been shown to act as barriers to signal transduction, whereas in others Prx and thioredoxin are required for effective signalling. We will discuss the underlying mechanisms that give rise to these apparent contradictions.

In eukaryotes, conserved MAPK pathways mediate responses to various stimuli. In unicellular eukaryotes these MAPK pathways promote changes in gene expression that are vital for adaptation and growth under changing environmental conditions (Toone and Jones, 1998). MAPK signalling pathways are also important mediators of growth factors and cytokine responses. For example, three MAPKs, JNK, p38 and ERK, all become activated following exposure to the cytokine, tumour necrosis factor-α (TNF-α). MAPK pathways consists of a series of protein kinases, which become sequentially activated by phosphorylation (Fig. 1) [For a review see (Marshall, 1994)]. Although some of the protein targets mediating the effects of MAPK remain unclear, MAPK have been found to phosphorylate a variety of target proteins, including transcription factors [For a review see (Sabio and Davis, 2014)]. Work to understand the activation of MAPK has largely focussed on how the phosphorylation of upstream kinases is triggered. However, there is some evidence that the regulation of dephosphorylation reactions, mediated by protein tyrosine phosphatases, doesn’t just allow feedback control of kinase activity, but can also be involved in sensing stress stimuli (Nguyen and Shiozaki, 1999).

Although diverse stimuli activate MAPK pathways, in many cases, secondary generation of ROS/redox changes are responsible for increases in MAPK activity. Consequently, amongst the most studied mechanisms are those by which H₂O₂ stimulates MAPK activation. The MAPKKK ASK1(MAPKKK5) is important in the activation of both p38 and JNK MAPK in response to a variety of stimuli, including H₂O₂ which may also act as a second messenger downstream of other stimuli, such as TNF-α (Ichijo et al., 1997, Tobiume et al., 2001). Following identification in a two hybrid screen for ASK1 interactors, the thioredoxin, Trx1, was found to bind to the N-terminus of ASK1, directly inhibiting its activation under reducing conditions (Saitoh et al., 1998). In this study, treatment of cells with TNF-α, or for 20min with 1mM H₂O₂, was proposed to activate ASK1 by causing Trx1 to become oxidised, promoting its dissociation from ASK1 (Saitoh et al., 1998). ASK1 then oligomerises allowing autophosphorylation of a threonine residue in ASK1’s activation loop that increases its kinase activity (Gotoh and Cooper, 1998; Tobiume et al., 2001). Hence this leads to the increased phosphorylation of p38 and JNK MAPK and apoptosis (Ichijo et al., 1997). More recent studies have suggested that ASK1 oligomerisation also involves the H₂O₂-induced formation of intermolecular disulphide bonds between cysteines in ASK1 which help stabilise the ASK1-ASK1 interactions required for autophosphorylation (Nadeau et al., 2007). This suggests that the ASK1-inhibitory role of Trx1 may be to reduce these oxidised, oligomeric forms of ASK1, preventing autophosphorylation (Nadeau et al., 2009). ASK1 is critically involved in signal transduction to p38 and JNK MAPK in response to a variety of stimuli. In most cases the oxidation state of thioredoxin has not been examined. However, these studies suggest that many activating stimuli may act by causing increased oxidation of thioredoxin. In which case, this would render ASK1 an effective sensor of thioredoxin oxidation. Intriguingly, thioredoxin is actually required for the H₂O₂-induced activation of the p38/JNK-related Hog1 MAPK in the fungal pathogen Candida albicans, raising the possibility that, in some cases, oxidised thioredoxin might even play a positive role in MAPK activation (da Silva Dantas et al., 2010).

However, in addition to inhibiting the activity of MAPKKK, thioredoxins are also implicated in feedback mechanisms to reactivate protein tyrosine phosphatases (PTPs) which de-activate MAPK. The catalytic cysteine of PTP becomes oxidised by H₂O₂ generated in response to a variety of stimuli. In many cases, irreversible oxidation is prevented by the formation of a disulphide bond with another cysteine which may be reduced by thioredoxin Trx1 (Kwon et al., 2004; Schwertassek et al., 2014). Thioredoxin oxidation would therefore also be predicted to inhibit the reduction of oxidised, inactive PTPs, increasing the extent or duration of MAPK activation.

Prx have been shown to have multiple roles in H₂O₂-signalling. Prx have been proposed to act as barriers to H₂O₂ signal transduction and indeed there is plenty of evidence to suggest that Prx inhibit the activation of ROS-activated signalling pathways, including MAPK (Cao et al., 2009; Choi et al., 2005; Kang et al., 2004; Kil et al., 2012; Woo et al., 2010; Yang et al., 2007). Given that antioxidants, such as N-acetyl cysteine (NAC) and catalase, inhibit MAPK activation, this has led to the conclusion that, in such cases, Prx inhibit activation by removing peroxides (Hashimoto et al., 2001; Yang et al., 2007) (Fig. 2A). Indeed, post-translational mechanisms to inhibit the thioredoxin peroxidase activity of Prx have been shown to be important for signal transduction (Kil et al., 2012; Woo et al., 2010).

Conversely, Prx have also been found to be important for H₂O₂ signal transduction. For instance, a genetic screen, using a LacZ reporter under the control of an H₂O₂-activated promoter, unexpectedly identified that the peroxiredoxin TSA1 was required for H₂O₂-induced gene expression in a commonly used laboratory strain of the budding yeast Saccharomyces cerevisiae (Ross et al., 2000) reflecting TSA1’s involvement in the H₂O₂-induced activation of the AP1-like transcription factor Yap1 (Okazaki et al., 2005). Subsequent studies have established that the single 2-Cys Prx, Tpx1, is required for H₂O₂ signal transduction in the unrelated fission yeast Schizosaccharomyces pombe, suggesting a conserved function for Prx in promoting H₂O₂ signal transduction (Veal et al., 2004). Indeed, the thioredoxin peroxidase activity of Tpx1 is required for the H₂O₂-induced activation of the Pap1 transcription factor (Bozonet et al., 2005; Vivancos et al., 2005), that mediates the transcriptional response to low levels of H₂O₂. Moreover, Tpx1 is also required for the activation of the p38/JNK-related MAPK Sty1 (Spc1) that phosphorylates the Atf1 transcription factor, promoting the Atf1-dependent expression of an overlapping set of stress-protective genes (Veal et al., 2004). Intriguingly, in contrast to Pap1, the activation of Sty1 does not require the thioredoxin peroxidase activity of Tpx1 (Veal et al., 2004). Overexpression of Tpx1 also stimulates the increased phosphorylation of Sty1. Indeed, overexpression of Tpx1 restores inducible phosphorylation to cells expressing a constitutively partially active form of the MAPKK Wis1, indicating that Tpx1 acts downstream of the MAPKK to promote Sty1 activation (Veal et al., 2004) [For a review see (Veal et al., 2014)].

Although it was possible that positive ROS-signalling roles of Prx were confined to unicellular eukaryotes, there is increasing evidence that Prx promote ROS-signalling in animals too. For instance, Prx2 was recently found to act as an H₂O₂ receptor, transmitting oxidative signals to the redox-regulated transcription factor STAT3 (Sobotta et al., 2015) (Fig. 2). Prx have also been shown to promote ROS-induced activation of MAPK in animals. For instance, Prdx1 is required for oxidised Low Density Lipoprotein (oxLDL)-induced activation of p38 in macrophage-derived foam cells (Conway and Kinter, 2006), H₂O₂-induced p38 activation in mammalian cells (Jarvis et al., 2012), and for arsenite-induced activation of the PMK-1 MAPK in the nematode worm Caenorhabditis elegans (Olahova et al., 2008). Prdx1 has also been shown to promote p38 activity in pancreatic duct adenocarcinoma cells (Taniuchi et al., 2015). Although the mechanism/s by which Prx promote p38/JNK MAPK activation are not well-established there are a number of possibilities which are discussed below and illustrated in Fig. 2:

[1] As direct redox-transducers: Prdx1 has been found to form intermolecular disulphide bonds with ASK1 that were proposed to initiate the oxidation of ASK1 oligomers (Jarvis et al., 2012). This suggests that in this case Prdx1 is able to act as an H₂O₂ receptor transducing the signal to drive the oxidative activation of ASK1. This is reminiscent of the role of Prx2 in promoting the activation of the STAT3 transcription factor (Fig. 2) (Sobotta et al., 2015). STAT3 is activated by transient formation of disulphide-linked STAT3 oligomers which are subsequently reduced by thioredoxin. The detection of Prx2-STAT3 disulphides suggests that Prx2 participates directly in initiating STAT3 oxidation. Although the abundance and H₂O₂-reactivity of Prx has led to suggestions that such mechanisms must be prevalent in the H₂O₂-induced oxidation of target signalling proteins (Winterbourn, 2008), it remains to be determined whether such peroxidase-based redox relays are widely conserved H₂O₂-sensing mechanisms, or limited to specific situations. Intriguingly, in S. pombe Tpx1 also forms mixed disulphides with the Sty1 MAPK (Veal et al., 2004). Indeed, further cysteines in Sty1 have been identified which form an intracellular disulphide bond which is important for transcriptional responses to H₂O₂ (Day and Veal, 2010). However, it remains to be determined how Tpx1-Sty1 disulphide formation leads to increased phosphorylation of Sty1.

Peroxiredoxins and thioredoxin have multiple redox-signalling activities. Here we have attempted to rationalise the basis for both positive and negative effects of Prx on the activation of MAPK pathways. Some outstanding questions remain, for instance: Is the only role of Prx in promoting ASK1 activity due to a direct ‘redox-transducing’ role or does Prx also promote ASK1 activation by inhibiting thioredoxin from reducing ASK1? Do Prx participate directly in the oxidative inactivation of PTPs either as redox transducers or by promoting the oxidation of thioredoxin? Why, given the high degree of homology between Prdx1 and Prdx2, do they have apparently different effects on MAPK activation? In addition to maintaining MKP-5 activity, do hyperoxidised Prx have other positive signalling/chaperone activities that promote MAPK activation?

Although inhibiting Prx appears to have different effects on MAPK activation depending on the cell or stimuli, in some cases the use of different stimuli, or time-points at which MAPK activation is assessed, make comparisons between studies difficult. In any case, the consequence of inhibiting Prx in any given cell is likely to be determined by the levels of intracellular ROS and intrinsic features of the cell/compartment, such as the capacity to regenerate reduced Prx, using sulfiredoxin, or thioredoxin, using thioredoxin reductase and NADPH (Fig. 2A). For example, it was recently shown that the regulation of mitochondrial sulfiredoxin levels plays an important role in regulating mitochondrial Prx activity and consequently, cytosolic H₂O₂ levels and p38 activity (Kil et al., 2012; 2015). As several studies have illustrated, relative levels and localisation of Prx, thioredoxin and signalling pathway components will also play an important part (Taniuchi et al., 2015; Woo et al., 2010). Accordingly modelling approaches that take into account all these things, developed using quantitative data obtained in well-characterised systems, such as yeast, are likely to be extremely important for identifying key features regulating MAPK activity and predicting the response of these signalling pathways to changing redox conditions.