Phosphotyrosine redox-regulation and the Src model
The involvement of redox regulation of PTPs in phosphotyrosine (p-Tyr) signal transduction prompts the question of whether the PTP counterpoint, the protein tyrosine kinases (PTKs), are similarly regulated by ROS. PTKs are the other half of the p-Tyr signalling dynamic, described as the ‘writer’ to the PTPs ‘eraser’ in the p-Tyr toolkit model proposed by Lim and Pawson . Such an interdependent system of signal transduction requires precise and co-ordinated regulation at every step, so that the powerful p-Tyr signal featuring so prominently in almost every cellular function does not become derailed. With a precedent set by the dual roles of phosphorylation as both a kinase-activating and phosphatase-inactivating mechanism, is there sufficient evidence to support the hypothesis that oxidation fulfils a similar function in the p-Tyr signalling machine?
The indirect redox regulation of PTKs through the oxidative inactivation of inhibitory PTPs, has been widely documented  and, although not the focus of the present review, represents an important mechanism of redox-mediated p-Tyr signalling. The present review aims to provide evidence to complement rather than confront this model, with the objective of broadening the current redox-signalling paradigm to include the direct redox regulation of PTKs as a crucial component.
Early evidence for the direct modification of kinases by ROS centred on the in vitro oxidation of Src tyrosine kinase by nitric oxide (NO)-releasing agents, resulting in enhanced catalytic activity [53, 54]. Subsequent studies by Senga et al.  identified a number of cysteine residues where Src is necessary for the enzyme's stability and transformative ability. Giannoni et al.  focused on Cys245 and Cys487, which were demonstrated to undergo intramolecular disulfide bridge linkage upon exposure to ROS, leading to a level of Src activation not observed in the redox-insensitive Cys-Ala mutants. Such work was crucial for formulating the concept of phases of Src activation via phosphorylation and de-phosphorylation steps, with a final super-activation state being achieved through oxidation (Fig. 4). This concept and the supporting literature that has emerged subsequently, demonstrating direct redox regulation of Src kinase in a variety of systems, was recently and comprehensively reviewed by Giannoni et al. .
Figure 4. Models of Src kinase redox regulation. Contrasting models of the outcome of Src oxidation as proposed by Giannoni et al.  (left) and Kemble et al.  (right). The model of redox-activation of Src involves phases of Src activation, from the generally accepted inactive conformation as a result of inhibitory phosphorylation at C-terminal tyrosine Y527 to the first phase of activation via relief of this inhibitory phosphorylation and autophosphorylation at Y416 of the catalytic domain. The final stage of activation requires oxidation of Cys245 (C245) of the SH2 domain and Cys487 (C487) of the catalytic domain, resulting in intramolecular disulfide bond formation and an increase in kinase activity. Kemble et al.  support the conflicting theory of redox inactivation of Src via oxidation of Cys277 (C277) on different Src molecules resulting in intermolecular disulfide bond formation, producing an inactive dimer.
Download figure to PowerPoint
Even in the 2 years since this review, further studies have emerged attesting to the direct redox regulation of Src in different cell systems and with diverse signalling outcomes. Most recently, Xi et al.  showed that, under conditions of hyperglycaemia, insulin-like growth factor-1 signalling leads to enhanced Src activation through direct oxidation by Nox4 in vascular smooth muscle cells. Lee et al.  similarly demonstrated that vascular c-Src forms a signalling axis with vascular endothelial growth factor-derived ROS through the direct oxidation of cysteine residues on both c-Src and vascular endothelial growth factor receptor (VEGFR), resulting in activation of the c-Src-phosphoinositide 3-kinase-Akt pathway downstream of VEGFR but not the phospholipase Cγ1-extracellular signal-regulated kinase (Erk) 1/2 pathway, which is also downstream of the receptor. The redox-specific activation of the VEGFR appears to lead exclusively to redox activation of c-Src, allowing the cell to sense the redox status of the endothelium and respond with specific pathway activation. Further experiments will determine whether a disulfide bond between VEGFR and c-Src contributes to this activation, as well as which cysteine residues are affected, and whether endocytosis of the two kinases (perhaps with a Nox protein and Src effector PI3K) facilitates this relationship.
The vulnerability of cells to de-regulated kinase activity and the subsequent risks of tumorigenesis have been widely documented, with an emphasis on the oncogenic activation of PTKs such as Src through constitutive phosphorylation or overexpression. However, the ability of redox-regulated Src to become activated through oxidation represents another potential avenue for pro-tumourigenic activity, especially within the oxidizing environment of many cancer cells that display higher levels of ROS than their noncancer counterparts [89-92].
One of hallmark tumourigenic characteristics described by Hanahan and Weiberg  is the focus of recent study by Zhu et al. , where a role for redox-regulation of Src in conferring anoikis-resistance to tumours is demonstrated. The role for oxidized Src in the mediation of integrin signalling, as outlined by Zhu et al. , echoes the findings of Giannoni et al. . In the situation described by Zhu et al. , cell attachment to the extracellular matrix via integrins is mimicked by angiopoietin-like 4 protein engagement of the integrins of cells when in suspension, stimulating ROS production, as well as activation of Src and subsequent pro-survival signalling normally associated with cell adhesion . Thus, the dependence of tumour cells on extracellular matrix or cell–cell contact is eliminated.
An intriguing aspect is the data presented in favour of the O2•− mediated oxidation of Src, whereas previous studies have strongly implicated NO [53, 54] and H2O2 [95-97] as being the relevant oxidizing agents with regard to Src. However, Zhu et al.  show that oxidative-activation of Src in tumour cells co-incides with a high ratio of O2•−:H2O2 caused by Nox activation downstream of integrin stimulation, and facilitated and sustained by a reduction in Cu/ZnSOD expression . Further investigations of the effects of O2•−-specific scavengers on the proportion of oxidized-Src, as well as an examination of the subcellular proximity of Src to the relevant O2•−-producing Nox, will help to clarify categorically whether O2•− is directly oxidizing Src in this situation.
Although these studies support a model in which Src is activated in response to ROS, data consistent with an inhibitory role for ROS in Src activity have also been provided by Kemble et al. . By contrast to the majority of published studies, yet echoing the earlier results of Cunnick et al. , Kemble et al.  propose that, upon oxidation, intermolecular disulfide bridges form between the Cys277 residues in the glycine loops of two different Src molecules, resulting in the formation of inactive Src dimers (Fig. 4). This scenario is also proposed to be applicable to FGFR1 because of its corresponding cysteine residue, which interacts with the γ-phosphate of ATP.
The contrast and apparent disparity between studies can be explained in part by the different conditions used by the different groups. Kemble et al.  highlight two advantages of a cell-free assay over whole-cell techniques when examining the redox-regulation of proteins. First, in a cell-free system, the absence of any Src-regulatory proteins, which may be the legitimate targets of oxidants, ensures that any effects on Src observed can be attributed to the direct oxidation of the kinase itself and not to indirect effects via oxidation of mediators. Second, great care must be taken during the processes of lysis and immunoprecipitation (i.e. as used to isolate Src from cells) to preserve the intracellular redox state of the kinase, by minimizing both exposure to environmental O2 and the release of cellular oxidants during lysis. However, cell-free assays alone cannot faithfully replicate the physiological scenario of localized ROS production in proximity to relevant protein targets, nor do they address the issue of oxidant concentration carefully tempered by the presence of cellular antioxidants and peroxiredoxins. In addition to this, it is likely that the intracellular redox regulation of Src is sufficiently complex to involve both the redox-activation and inactivation of Src depending on the ROS concentration, as well as the sensitivity and accessibility of individual cysteine residues. As argued by Giannoni et al. , oxidation of Cys277 might well be an artefact of the cell-free assay, and may not occur intracellularly as a result of the native conformation of the enzyme.
It is interesting at this point to note that Tang et al.  also observed redox-inactivation of Src kinase, although within a cellular context, through the addition of H2O2 to primary human umbilical vein endothelial cells, human adult aorta cells and E6 fibroblasts. However Tang et al.  reported that the kinase activity was not affected in a cell-free assay, again clearly demonstrating a disconnect between the two systems. By contrast to Kemble et al. , Tang et al.  apparently support whole cell-based methods, yet arrive at the same conclusion with respect to the inhibitory effect on Src.
To address the conflict between their results and the earlier studies supporting redox-activation of Src, Tang et al.  propose an elegant compromise by outlining a scenario in which both the localization of the kinase and the dose of H2O2 used affect the outcome in terms of positive or negative regulation. The immunofluorescence data of Tang et al.  appear to demonstrate that Src kinases residing in focal adhesion complexes or at the plasma membrane become inactivated after the addition of H2O2, whereas those localized to the cytoplasm and compartments such as endosomes become activated.
This provides convincing evidence for the spatially defined redox-regulation of Src, with the differential subcellular localization of the kinase determining the consequences of its oxidation. The scenario is promising in terms of unifying the apparently disparate hypotheses of Src redox-inactivation and redox-activation because it provides a possible explanation for the observations of both phenomena.
Tang et al.  also comment on the importance of the dose of H2O2 used, describing how very high concentrations (in the mm range), which cannot be claimed to be physiologically relevant, may have pleiotrophic effects on the intracellular signalling network, and produce highly artificial results . This would obscure observation of the innate redox regulation of the kinase.
The addition of exogenous H2O2 is a device that is often used to investigate intracellular redox-signalling. However, even at physiologically appropriate concentrations, this method cannot accurately reproduce the temporally- and spatially-regulated intracellular production of H2O2 for signalling purposes, which occurs in discrete subcellular domains. As an alternative, the induction or overexpression of ROS-producing enzymes such as Nox, allows the concentration, proximity, duration and nature of the species produced to more closely reflect that of the physiological situation, and comprises one method of introducing a much-needed cellular context into the debate on the redox regulation of Src.
Src family kinases
Irrespective of its consequence for Src kinase activity, the susceptibility of Src to redox regulation is a characteristic that is assumed to be common to other members of the Src family kinases (SFK). The first kinase hypothesized to be redox regulated was the lymphocyte-specific SFK p56lck (Lck) when, in 1993, four groups independently published data demonstrating increased Lck activity upon oxidant exposure [101-104]. Although this was interpreted as signifying that the kinase is active when oxidized , at the time, it was not clear whether Lck was directly oxidized, although it was emphasized by Schieven et al.  that the kinase activity observed was not simply a result of phosphatase inhibition. Nakamura et al.  provided the most detail with regard to oxidation-induced modifications to Lck by examining the tyrosine phosphorylation profile of T-cells after exposure to oxidative stress, in the form of the thiol-oxidizing agent diamide, and finding that Lck was phosphorylated at its autophosphorylation site Tyr394 after exposure to the oxidant. They also detected increased activity of Lck presumably as a consequence of this phosphorylation event, whereas other similar kinases such as p59fyn did not become activated.
Nakamura et al.  subsequently demonstrated that oxidant exposure also prompted the association of Lck with PI3K, although in a binding conformation, which was not observed in the absence of oxidizing agents . This is perhaps the most convincing detail so far in support of direct oxidation of Lck because it implies a conformational change to the enzyme that cannot be attributed to established phosphorylation-induced modifications. There are a number of cysteine residues in Lck that have been shown to be crucial to the enzymes stability and function , including Cys475, which is analogous to a conserved oxidation-sensitive residue in c-Ret , as discussed further below. However, it remains to be established definitively whether Lck is directly redox-activated.
More recently, the SFK Lyn was demonstrated by Yoo et al.  to be a ‘redox sensor’ in zebrafish neutrophils. In this elegant model of wound healing, neutrophilic Lyn undergoes oxidation at Cys466 upon exposure to extracellular H2O2 produced at the site of the wound, resulting in the autophosphorylation and activation of the kinase and selective stimulation of its downstream pathways. The redox activation of Lyn appears to be the dominant factor driving the directional migration of the neutrophils towards the wounded area. Although the paracrine signalling properties of H2O2 in wound healing have been well described , Yoo et al.  were the first to demonstrate a molecular connection between the redox signal and the neutrophil recruitment response.