Processing of and responding to various signals is an essential cellular function that influences survival, homeostasis, development, and cell death. Extra- or intracellular signals are perceived via specific receptors and transduced in a particular signalling pathway that results in a precise response. Reversible post-translational redox modifications of cysteinyl and methionyl residues have been characterised in countless signal transduction pathways. Due to the low reactivity of most sulfur-containing amino acid side chains with hydrogen peroxide, for instance, and also to ensure specificity, redox signalling requires catalysis, just like phosphorylation signalling requires kinases and phosphatases. While reducing enzymes of both cysteinyl- and methionyl-derivates have been characterised in great detail before, the discovery and characterisation of MICAL proteins evinced the first examples of specific oxidases in signal transduction. This article provides an overview of the functions of MICAL proteins in the redox regulation of cellular functions.
In the past, redox-dependent modifications were supposed to be the result of a global cellular redox state, which is determined by the ratio of pro- and anti-oxidants within the cell. This concept defined ‘oxidative stress’ as excess of the ‘pro-oxidant’ molecules or decreased levels of ‘anti-oxidant’ molecules . Over the past decade, however, redox-biochemistry experienced a dramatic shift in paradigm as redox modifications were more and more shown to be specific, rapid, physiological, and reversible signalling events, see for instance [2–4], and [5–9] for in-depth discussions. Redox mediated signal transduction occurs in different cellular compartments, often at the same time, both by oxidation and reduction of key molecules. Signal transduction depends on the reaction kinetics and thus enzymatic activity [2,5,6,10].
Redox modifications of cysteinyl and methionyl side chains
The sulfur moieties of cysteinyl and methionyl side chains are prime targets for reversible redox modifications (Figure 1) summarised, e.g. in . The oxidation of thiols groups can lead to the formation of inter- or intramolecular disulfides and thus the generation of e.g. hetero- or homo-dimers of proteins (Figure 1, reaction 2). Such intra- or intermolecular disulfides can be reduced by specific thiol-containing reductases such as glutaredoxins (Grx) or thioredoxins (Trx). In addition, cysteinyl residues may also form disulfides with other thiol-containing molecules such as glutathione (GSH), the most abundant cellular thiol component. The oxidation of thiol groups by hydrogen peroxide evinces sulphenic acids (Figure 1, reaction 3). Cysteinyl sulphenic acid derivates are rather unstable and will readily react with thiols yielding disulfides (Figure 1, reaction 4). If these derivates react with further hydrogen peroxide molecules, sulphinic and sulphonic acids may form, reactions that are considered irreversible (Figure 1, reactions 5 and 6). Another form of post-translational redox modification is the reaction of thiol groups and nitric oxide (NO), resulting in S-nitroso-thiols (Figure 1, reaction 1).
Redox modifications of cysteinyl and methionyl amino acid side chains.
The thiol-ether groups of methionine or methionyl residues can be reversibly oxidised by hydrogen peroxide to methionine or methionyl sulfoxide (Figure 1, reactions 7 and 8), i.e. to a mixture of two diastereomers, methionine-S-sulfoxide, and methionine-R-sulfoxide. This reaction is today also recognised as a redox signalling mechanism . Methionine/methionyl-S-sulfoxide is specifically reduced by methionine sulfoxide reductase (Msr) A, methionine/methionyl-R-sulfoxide by MsrB . Human MsrA is expressed in various isoforms from a single gene, giving rise to mitochondrial, cytosolic, and nuclear variants [13–15]. Oxidised Msrs form an intramolecular disulfide that is reduced by Trxs or Grxs [12,16]. Most mammalian genomes possess three MsrB genes. MsrB1 is a seleno-protein. MsrB2 and 3 contain cysteinyl residues only and are catalytically less efficient [17,18].
Signal transduction depends on specificity and reversibility. In fact, redox modifications do not occur randomly at any given cysteinyl or methionyl side chain. Today, we see more and more evidence that many — if not most — reactions outlined above are catalysed by enzymes (see Figure 1). This ensures not only a fine tuned, enzymatically catalysed activation and inactivation of protein functions, but also allows for spatio-temporal control of these events through cell signalling, see for instance . In this model, the specificity of the redox modification does not depend on the accessibility or reactivity/nucleophilicity of the modified residue, but arises primarily from the specificity of the protein–protein interactions between catalyst and target proteins. These interactions may be influenced by conformational changes of the proteins induced by other signalling events as, for instance, seen in phosphorylation cascades or G-protein-coupled receptor signalling pathways.
Hydrogen peroxide production, reactivity, and reduction
Outside the endoplasmic reticulum and peroxisomes, H2O2 is primarily the product of superoxide dismutases (SOD). Mammalian cells contain two SOD isoforms: the cytosolic SOD1 that is Cu/Zn-dependent, and the mitochondrial SOD2 that is Mn-dependent. These enzymes catalyse the alternate reduction and oxidation of superoxide yielding hydrogen peroxide and molecular oxygen [19–21]. Superoxide is produced by numerous enzymes in the cell, for instance: mitochondrial complexes I  and III , NADPH oxidases , cyclooxygenases , xanthine oxidoreductase , and cytochrome p450 enzymes .
In principle, hydrogen peroxide readily reacts with cysteinyl and also methionyl groups (see above). However, the second-order rate constant of hydrogen peroxide with free cysteine is ∼k = 2.9 M−1 s−1, with GSH ∼0.9 M−1 s−1 , and with the active site motif of Trx ∼1 M−1 s−1 . The rate constants of hydrogen peroxide with methionine and methionyl residues in peptides and proteins were determined to be as low as 7 × 10−4–8 × 10−3 M−1 s−1 . Not only are these rate constants really low, these reactions also compete for hydrogen peroxide with some of the most abundant proteins of all cells, i.e. peroxiredoxins (Prxs) and glutathione peroxidases (Gpxs). The rate constant of the reaction of hydrogen peroxide with the peroxidatic cysteinyl residue at the active site of Prxs and the cysteinyl or selenocysteinyl residue in the active sites of Gpxs ranges from 3 × 105 to 107 M−1 s−1 [29,31]. Peroxynitrite may also react with the sulfur moieties of cysteinyl and methionyl residues at considerably higher rate, however, Prxs are also highly reactive with this reactive nitrogen species [32,32,33]. We can thus conclude that by far most hydrogen peroxide (and also peroxynitrite) in vivo will react with dedicated peroxidases, for detailed quantitative calculations see also . The low rate constants of regulatory methionyl or cysteinyl residues with these species compared with the dedicated peroxidases makes a direct reaction highly unlikely. This, in turn, necessitates enzymatic catalysis of the modifications in physiological redox signalling. Most cysteinyl and methionyl residues modified in signalling pathways may only be oxidised in the presence of a catalyst or when tethered to a site of peroxide production. For more comprehensive discussions on this topic, see also [5,10,35–37].
Flavin monooxygenases of the MICAL family
MICALs (molecule(s) interacting with CasL) are a family of multidomain proteins that participate in cellular processes such as axonal growth cone repulsion, membrane trafficking, and actin dynamics [38–41]. MICALs were first described in 2002 as regulators of axon repulsion . The human MICAL family consists of three isoforms (MICAL 1–3) that differ in their domain composition (Figure 2a). All genuine MICAL proteins contain an N-terminal monooxygenase (MO) domain with FAD as prosthetic group. Subsequently, calponin homology (CH) and LIM domains follow. An additional C-terminal Rab-binding domain (RBD) was identified for some MICAL family proteins, e.g. human MICAL1 and MICAL3. The MO domain enables MICAL proteins to function as flavin MO and oxidise substrates using oxygen and NADPH as substrates. The CH domains are important to facilitate the actin binding by the MO domain [43–45]. LIM domains have been shown to play roles in cytoskeletal organisation and mediate protein–protein interactions . The RBD incorporates two distinct binding sites for small GTPases of the Rab family with different affinities and was initially proposed to form a coiled-coil domain, thus sometimes termed CC domain.
Reactivity, kinetics, and mechanism
MICALs are FAD-containing MO structurally similar to aromatic hydroxylases and amine oxidases . In the presence of NADPH, the proteins reduce molecular oxygen to hydrogen peroxide via a peroxyflavin intermediate (kcat = 77 s−1). It was thus proposed that the proteins may produce hydrogen peroxide as signalling molecules . When methionyl residues in actin were identified as MICAL substrates (see below), it was shown that F-actin, but not G-actin, nor free methionine, stimulates NADPH oxidation by increasing kcat . Moreover, an apparent Km for actin of 4.7 μM could be determined, suggesting a direct oxidation of methionyl residues to methionyl sulfoxides [9,48]. The reductive half-reaction of the MICAL2 hydroxylase domain is stimulated by F-actin. In the absence of actin, NADPH reduces the flavin relatively slowly; the presence of actin increases the velocity of this reaction significantly. Hence, MICAL2 has the classic behaviour of class A MO, i.e. slow reduction in the flavin when the substrate to be oxygenated is absent . Under physiological conditions, this may attenuate the production of hydrogen peroxide released from the enzyme. The comparison of the structures before and after reaction with NADPH revealed that the flavin ring can or even must switch between two discrete positions (Figure 2b) . Most notably, this conformational switch is coupled to the opening of a channel to the active site (Figure 2c). In this structure, a water molecule is placed in close proximity to the carbon C4a (arrow in Figure 2c). Thus, this oxygen molecule is in exactly the same position as the expected peroxy group of the peroxyflavin intermediate. Being placed only 0.2 nm from the surface of the protein, it might well be available for an attack by an amino acid side chain. Alternatively, hydrogen peroxide could be released at this position and presented directly to target proteins bound in close proximity to the channel.
In all MICALS, the MO domain is followed by a type-2 CH domain (Figure 2a). A recent study on the structure of a fragment of MICAL1 containing the MO and the CH domains suggest that the MICAL redox activity is controlled by a cooperative mechanism between the domains . For a more detailed discussion on the structure–function relations of MICALs catalytic mechanisms, we refer to Vanoni .
MICAL proteins as signal transducers
Redox control of protein function involves the specific oxidation and reduction in amino acid side chains and, as discussed above, kinetic constrains require catalysis of both oxidation and reduction. Various specific reductases have been described in great detail, most of them belong to the Trx family of proteins . To the best of our knowledge, MICALs are the first examples of signal-induced specific oxidases. Two methionyl residues (M44 and M47) in F-actin are specifically converted to methionyl-R-sulfoxides by MICALs regulating actin filament dynamics [4,52]. The re-reduction back to methionyl residues is specifically catalysed by MsrB selenoproteins, representing an example for a fully catalysed reversible site- and stereo-specific methionyl oxidation [53–55]. So, what signalling pathways lead to the activation of MICALs?
Semaphorins (Sema) are a large family of secreted signalling proteins involved in many cellular processes . Cellular and genetic work using the model organism Drosophila indicates that Semaphorins regulate the activity of MICALs and their interaction with target molecules including their F-actin substrate [39,42], see  for a review. In mammals, a role for MICALs in Semaphorin signalling has also begun to emerge, e.g. [56–59]. For example, Sema3A, which is one of the most widely studied Sema, influences many cellular processes including kidney development and polarisation , as well as the migration of T-lymphocytes . Neuropilin-1 (NP1) and plexin A (PlexA) form a heterodimeric transmembrane protein which acts as a Sema3A receptor. Sema3A binds to its ligand partner NP1 leading to PlexA activation and further signal transduction, summarised in . MICAL1 was shown to directly interact with PlexA1 and A3, MICAL2 with PlexA4 via their C-terminal domains, overview in . As in Drosophila, results indicate that these interactions regulate the activity of MICALs and their interaction with target molecules (Figure 3).
Signalling networks with mammalian MICAL proteins as transducers.
The C-termini of some MICALs (in human 1 and 3) contain a RBD, suggesting their regulation by small GTPase signalling. Redox-active MICAL1 was shown to be inhibited by its C-terminal domain. This inhibition can be released by a direct interaction of MICAL1 with Rab35. This interaction activates the MO domain and/or allows the interaction with actin filaments and thus the regulation of their dynamics  (Figure 3). The interaction of MICAL1 with Rab8 in the active GTP-bound state stabilises the active MICAL1 conformation causing a specific four-fold increase in kcat of the NADPH oxidase reaction . The interaction of MICAL3 with Rab8 was suggested to play an important role in vesicle docking and fusion contributing to the receptor and channel membrane trafficking mechanisms disturbed in human ciliopathies . Moreover, Rab1, involved in vesicle trafficking, was also identified as an interaction partner of MICAL1 . Rab6, Rab8, and MICAL3 were suggested to co-operate in controlling docking and fusion of exocytotic carriers . Interestingly, interactions of Rab proteins (Rab8, 13, and 15) with MICAL-like proteins, that lack the MO domain, have also been characterised [66,67].
MICALs functions regulate actin dynamics in various processes, for overviews see [41,51,54,68,69]. In addition, MICALs might also act on other proteins as oxidase of specific amino acid side chains. Depletion of MICAL1 markedly reduced cell proliferation in breast cancer cells likely by reduction in hydrogen peroxide production, presumably via maintaining cyclin D expression through the hydrogen peroxide-sensitive PI3K/Akt/ERK signalling pathway . MICAL1 and 3 may also be involved in the formation of a regulatory disulfide in the collapsin response mediator protein 2 through the production of hydrogen peroxide . MICAL1 is a binding partner of nuclear Dbf2-related (NDR) kinases. MICAL1 was suggested to compete with mammalian Ste-20-like kinase (MST1) for NDR binding and thereby to antagonise MST1-induced NDR activation that would otherwise lead to pro-apoptotic signalling . MICAL2 was demonstrated to bind to p53 and oxidises it at M40 and M160, thus promoting ubiquitination, degradation, and inactivation of p53 . MICAL2 was also reported to promote breast cancer cell migration by inhibiting endothelial growth factor receptor degradation in a Rac1-dependent manner .
Importance of the field — The identification and characterisation of MICAL proteins as specific oxidases in signal transduction was a major breakthrough in our understanding of reversible redox signalling. MICALs fulfil all requirements for the proposed catalysts in redox signalling.
Summary of the current thinking — The proteins act in specific signalling events and as specific oxidases of methionyl residues and maybe also as generators of hydrogen peroxide and thereby as oxidases of cysteinyl residues [70,75]. One of the major functions is the regulation of actin dynamics.
Comments on future directions — While our knowledge on MICALs is steadily increasing (see also [41,51,54,68,69] for more comprehensive coverage), various questions have to be addressed in the future: Do MICAL proteins oxidise more target proteins? Why do mammalian genomes encode three MICAL isoforms? The results outlined above imply that they might act, at least in part, on different targets and are activated by different mechanisms. What are the molecular mechanisms for MICAL activation? Especially the activation by the semaphorin receptors were not addressed yet. And, what is/are the molecular mechanism(s) of target oxidation? Do MICALs produce hydrogen peroxide in vivo, or do they act on protein targets exclusively?
The authors declare that there are no competing interests associated with the manuscript.
We gratefully acknowledge the financial support by the Deutsche Forschungsgemeinschaft (DFG, German Research Foundation), grants (to CHL) LI 984/3-2 (SPP 1710) and GRK1947-A1.