The Role of Peroxiredoxins in the Transduction of H 2 O 2 Signals

Introduction

Since the discovery that hydrogen peroxide (H₂O₂) produced in response to activation of various receptors in mammalian cells propagates receptor signaling by oxidizing the active-site thiols of protein tyrosine phosphatases (PTPs) (96, 97, 113), the reversible oxidation of cysteine residues in proteins has emerged as an important facet of intracellular signaling pathways (13, 22, 31, 40, 146, 158, 181, 188).

H₂O₂ oxidizes the cysteine thiol (–SH) to sulfenic acid (–SOH), which then either forms an intramolecular or intermolecular disulfide bond, a mixed disulfide with glutathione (glutathionylation), or sulfenylamide (–S–NR–CO–) or is further oxidized to sulfinic (–SO₂H) and sulfonic (–SO₃H) acids (Fig. 1). Oxidation to sulfenic acid, disulfide, or sulfenylamide can be reversed by various cellular thiols, whereas that to sulfonic acid is irreversible. Oxidation to sulfinic acid appears to be irreversible for most proteins, with one exception being the active-site cysteine of the family of antioxidant enzymes known as peroxiredoxins (Prxs). Changes in cysteine oxidation state can affect the susceptibility of a protein to other posttranslational modifications, influence protein turnover or enzymatic activity, regulate protein–protein interactions, and alter subcellular protein trafficking.

OPEN IN VIEWER

FIG. 1.

Scheme for oxidation of proteinaceous cysteine thiols by H₂O₂. The oxidation of cysteine thiol (–SH) groups by H₂O₂ results in the reversible formation of cysteine sulfenic acid (–SOH). The sulfenic intermediate can then form an intramolecular disulfide bond (S–S) if it is in close proximity to another cysteine, a sulfenylamide bond with a neighboring amide nitrogen, or an intermolecular disulfide with a thiol of another protein or glutathione (GSH). The formation of sulfenic acid, disulfide, or sulfenylamide can be reversed by various thiol-containing electron donors. The sulfenic intermediate can also be further oxidized to sulfinic (–SO₂H) and sulfonic (–SO₃H) states, from which cysteine thiol cannot be regenerated. H₂O₂, hydrogen peroxide.

Many proteinaceous cysteines have now been identified as primary sites of reaction with H₂O₂ (see section on Protein Whose Function Is Regulated via Reversible Cysteine Oxidation). However, the mechanism by which H₂O₂ transmits its signal by oxidizing the specific cysteine residues of its target proteins in the presence of vigilant antioxidant enzymes is not well understood. H₂O₂ molecules are eliminated by several such enzymes, including catalase, glutathione peroxidase (GPx), and Prx. Whereas catalase is present mostly in peroxisomes, GPx and Prx exist in multiple isoforms that are found in a variety of subcellular compartments. The primary sites of reaction with H₂O₂ in catalase, GPxs, and Prxs are heme, selenocysteine, and cysteine, respectively.

Kinetic and structural analyses have revealed that all Prxs possess an active-site pocket that gives rise to a high-affinity (submicromolar) peroxide binding site that is lacking in catalase and GPxs (123, 135, 147). In addition, this pocket of Prxs is arranged so as to stabilize the transition state intermediates that aid breaking of the O-O bond of H₂O₂, thus eliciting rapid oxidation of the catalytic cysteine by H₂O₂, with the second-order rate constant for the oxidation of Cys–SH to Cys–SOH being several orders of magnitude greater than that for the oxidation of cysteine in well-characterized H₂O₂ target proteins (135, 188). Given that Prxs are abundant proteins and are present in various compartments of the cell, cysteine residues of most target proteins are at a competitive disadvantage for reaction with H₂O₂.

It has recently become increasingly apparent that Prxs are more than simple peroxide-eliminating enzymes and that they actively participate in H₂O₂ signaling by assisting H₂O₂ effector molecules to overcome their kinetic disadvantage with regard to cysteine oxidation. Although the role of Prxs in H₂O₂ signaling remains largely uncharacterized, a full understanding of this role will require identification both of the relevant sources of H₂O₂ and of the effector proteins. In this review, we have compiled various known sources of H₂O₂ production as well as classified proteins known to be modulated by H₂O₂ through cysteine oxidation according to their biochemical and cellular functions.

We then describe how some of these H₂O₂ target proteins are able to overcome their kinetic disadvantage with regard to cysteine oxidation by H₂O₂. One way that this can be achieved is through temporary inactivation of Prx so as to protect H₂O₂ from destruction. Prx I is reversibly inactivated either through phosphorylation by Src family kinases at the plasma membrane in response to growth factor receptor or immune receptor activation (190) or through phosphorylation by cyclin B-dependent kinase activity at the centrosome during mitosis (27, 101). Such localized inactivation of Prx allows H₂O₂ to accumulate in specific regions of the cell without global redox disturbance.

Another way in which oxidation of H₂O₂ target proteins can be achieved is through oxidation of Prx I or Prx II by H₂O₂ and subsequent transfer of their oxidation state to the target protein. Such target proteins include apoptosis signal regulating kinase 1 (ASK1), the transcription factor STAT3 (signal transducer and activator of transcription 3) (72, 126, 168), the redox-sensitive chaperone DJ-1 (48), and APE1/Ref-1 (apurinic/apyrimidinic endonuclease 1/redox effector factor-1), a reductive activator of many transcription factors (125).

Sources of Intracellular H₂O₂

The complete reduction of one oxygen molecule (O₂) to water requires four electrons, with incomplete reduction by one electron producing the superoxide anion (O₂ ^−•), which is spontaneously or enzymatically (by superoxide dismutase) converted to H₂O₂. Incomplete reduction of molecular oxygen by two electrons generates H₂O₂ directly. Enzymes that generate H₂O₂ directly or indirectly in mammalian cells include NADPH oxidase (Nox) (5, 92), the mitochondrial electron transport chain (ETC) complexes (5, 152, 164), cytochrome P450 (CYP) (5, 55, 60), arachidonic acid-metabolizing enzymes such as lipoxygenase and cyclooxygenase (32), xanthine oxidase (78, 128), and monoamine oxidase (MAO) (46, 109) (Fig. 2). Among these enzymes, Nox is the only one whose primary function is to generate superoxide or H₂O₂, with the others producing these oxidants as a by-product of their main catalytic function.

OPEN IN VIEWER

FIG. 2.

Intracellular sources of H₂O₂ and types of effector molecules with H₂O₂-sensitive cysteine residues. Proteins with cysteine residues susceptible to oxidation by H₂O₂ to one of the states shown in Figure 1 are grouped on the basis of their biochemical and cellular functions. Not all proteins with known H₂O₂-sensitive cysteines are included here.

Members of the Nox family of enzymes exist as membrane-bound complexes that face the extracellular space when located at the plasma membrane. Nox generates superoxide by transferring electrons from intracellular NADPH across the membrane to molecular oxygen. The Nox family consists of seven members—Nox1 to Nox5 as well as two dual oxidases designated Duox1 and Duox2—that are differentially expressed and regulated in various tissues and which have different subcellular localizations (20). Nox1 is present in lipid rafts at the plasma membrane as well as in endosomes; Nox2 in phagosomes, endosomes, and at the leading edge of lamellipodia; Nox3 at the plasma membrane; Nox4 in focal adhesions, the nucleus, and endoplasmic reticulum (ER); and Nox5, Duox1, and Duox2 at the plasma membrane.

The signaling targets of H₂O₂ molecules produced by Nox enzymes are located in close proximity to the site of H₂O₂ generation (65, 129, 182). Early studies revealed that growth factors such as platelet-derived growth factor (PDGF) and epidermal growth factor (EGF) induce the production of H₂O₂ and that this H₂O₂ is required for the mitogenic action of these proteins (4, 173). Nox enzymes were subsequently identified as the source of H₂O₂ for such proliferation signaling: Nox1 for PDGF- or EGF-induced mitogenesis in nonphagocytic cells (114, 171) and for angiotensin II-induced proliferation of vascular smooth muscle cells (93), and Nox4 for proliferation of pulmonary artery smooth muscle cells induced by transforming growth factor-β (170).

The first proteins identified as targets of H₂O₂ produced by Nox were members of the PTP family. H₂O₂ generated in response to EGF, PDGF, or insulin signaling was thus found to inactivate PTPs by oxidizing the active-site cysteine of these enzymes and thereby to enhance growth factor-induced protein tyrosine phosphorylation (96, 106, 113). Many other signaling proteins, including the transcription factor AP1 (activator protein 1) and Src protein kinase, were later shown to be regulated via cysteine oxidation [reviewed in Refs. (31, 71, 148)].

Mitochondria generate ATP by transferring electrons derived from the tricarboxylic acid cycle to the ETC, which ultimately delivers them to molecular oxygen. The ETC comprises four protein complexes (complexes I–IV) that are located at the mitochondrial inner membrane. Electron flow through these complexes requires a diverse set of cofactors that include quinones, flavin groups, heme moieties, and iron/sulfur centers. Electrons can escape from the flavin groups or iron/sulfur centers of complexes I, II, and III in the ETC and can then be captured by O₂ to generate superoxide, which then undergoes dismutation to yield H₂O₂ (152, 164). Many other mitochondrial proteins, including those that participate in the tricarboxylic acid cycle, contain flavin groups or iron/sulfur centers. As with the ETC complexes, electrons can leak from the cofactors of these proteins and react monovalently with O₂ to form superoxide (152, 164).

The production of H₂O₂ by mitochondria was long viewed as a wasteful side reaction of ATP production. However, substantial evidence now indicates that such mitochondrially generated H₂O₂ serves an important signaling function in the regulation of diverse biological processes such as cell proliferation, differentiation, autophagy, immunity, the response to hypoxia, and, most recently, circadian rhythm (37, 59, 80, 81, 134, 152, 164). Some of the H₂O₂ produced in mitochondria is released into the cytosol, where it serves as a key regulator of various signaling pathways. However, mitochondria are equipped with several H₂O₂-eliminating enzymes, with Prx III being an especially abundant and efficient such enzyme in mitochondria of most cell types. As the production of H₂O₂ increases in mitochondria, Prx III undergoes reversible inactivation through hyperoxidation of its catalytic Cys–SH to Cys–SO₂H, likely resulting in the release of mitochondrial H₂O₂ into the cytosol [reviewed in Ref. (151)].

Mitochondria are highly dynamic organelles that frequently undergo fission, fusion, and translocation along the cytoskeleton. The overall morphology of mitochondria is therefore highly variable, ranging from numerous fragmented individual organelles to a single large interconnected tubular network depending on environmental conditions and cell type. Recent studies have revealed that the mitochondria of HEK293 cells are arranged in elongated tubules that undergo fragmentation during mitosis (74, 75). Given that cell division represents a major commitment to energy expenditure, mitochondrial H₂O₂ likely plays a major role in coordinating the cell cycle with metabolism.

The CYP family comprises many enzymes that catalyze the monooxygenation of hydrophobic organic molecules. CYPs bind and activate oxygen via a heme group in their active site, and they use one atom of an oxygen molecule to hydroxylate their substrate and the other to produce H₂O. The two electrons required for the conversion of an oxygen atom to H₂O are provided by NADPH. During the catalytic cycle, a substantial portion of the heme-bound (activated) oxygen is released from the catalytic intermediate and forms superoxide or H₂O₂ (55, 60, 199). CYPs are abundant in the ER of liver cells as well as in mitochondria of steroidogenic cells. The production of H₂O₂ by CYPs is generally considered to be a side reaction with accompanying adverse effects. However, H₂O₂ produced by a CYP in mitochondria of the mouse adrenal gland was recently shown to contribute to a negative feedback mechanism that is critical for the circadian rhythm of corticosterone production (80).

Arachidonic acid metabolism by lipoxygenase and cyclooxygenase enzymes also plays a role in the generation of reactive oxygen species (ROS) in various cell types. Arachidonic acid is released from glycerophospholipids in the nuclear envelope and plasma membrane through the activity of cytosolic phospholipase A₂, and it can then be oxidized to a variety of bioactive eicosanoids by lipoxygenase and cyclooxygenase enzymes. Oxidation of arachidonic acid by these two groups of iron-containing dioxygenases is accompanied by the side production of ROS, including H₂O₂ (32). Certain arachidonic acid metabolites have also been found to increase H₂O₂ levels by activating Nox enzymes (32).

Xanthine oxidase is a large (270 kDa) protein that contains two molybdopterin and two flavin groups as well as eight iron atoms and which catalyzes the oxidation of hypoxanthine to xanthine as well as the further oxidation of xanthine to uric acid (78, 128). Substrate-derived electrons are first transfered to the molybdenum cofactor and then, via several intermediates, to the flavin center, where they can reduce O₂ either univalently to yield the superoxide anion or divalently to yield H₂O₂ (78, 128).

MAO enzymes are flavoproteins and catalyze the oxidative deamination of neurotransmitters (such as dopamine, serotonin, and norepinephrine) and food-derived biogenic amines (such as tyramine and phenethylamine). Substrate-derived electrons reduce the flavin groups, which then react with O₂ stochiometrically to form H₂O₂ (46). Two forms of MAO are found in mammals, with MAOA being located at the cytosolic face and MAOB at the intermembrane face of the mitochondrial outer membrane (46). Although a signaling role has been demonstrated for CYP-derived H₂O₂ in steroidogenic cells, H₂O₂ produced by CYP, lipoxygenase, cyclooxygenase, xanthine oxidase, and MAO enzymes has been studied mainly in connection with pathological conditions such as hepatotoxicity, vascular inflammation, and neurodegeneration.

Proteins Whose Function Is Regulated via Reversible Cysteine Oxidation

Many proteins are susceptible to modification by cysteine oxidation, with their subcellular localization and sensitivity to H₂O₂-mediated oxidation varying widely. The regulation of cellular processes by H₂O₂ thus depends on the concentration of the oxidant, the duration of exposure of the target protein, and cell type. Taking cell cycle progression as an example, low levels of H₂O₂ generated by Nox enzymes promote the transition from G₀ to G₁ phase by increasing the expression of cyclin D, whereas prolonged exposure to higher levels of H₂O₂ induces adaptive cellular responses by activating the tumor suppressor protein p53, which triggers either growth arrest at the G₁-S or G₂-M transitions of the cell cycle or apoptosis depending on the extent of such exposure [reviewed in Refs. (22, 31)].

The conversion of H₂O₂ to hydroxyl radicals and consequent induction of DNA damage result in recruitment of the protein kinase ATM (ataxia telangiectasia mutated) to the damage sites, where it phosphorylates and activates p53. H₂O₂ can also activate ATM directly in the absence of DNA damage by inducing the formation of a disulfide-linked dimer of the kinase (57). Low levels of H₂O₂ produced by Nox4 and Duox2 in response to stimulation of normal human fibroblasts with PDGF were also found to promote cell cycle entry not by increasing cyclin D expression but by downregulating the p53-dependent checkpoint (156). This attenuation of p53 function is likely achieved through a reduction in p53 concentration or through oxidation of cysteine residues of p53 and consequent inhibition of its DNA binding activity (156, 172).

Given that many regulatory proteins participate in multiple signaling pathways, it has also become clear that the outcome of activation of a specific regulator depends on contextual factors such as cell type and the activation state of related signaling pathways. For example, activation of the three types of mitogen-activated protein kinase (MAPK)—extracellular signal-regulated kinase (ERK), c-Jun NH₂-terminal kinase (JNK), and p38—all of which are sensitive to H₂O₂, has been implicated in cellular outcomes as diverse as proliferation, survival, differentiation, and death (94). Activation of ERK and p38 thus contributes not only to promotion of the G₀-G₁ transition via the AP1–cyclin D pathway but also to growth arrest at the G₁-S or G₂-M transitions through phosphorylation of p53 and consequent inhibition of p53 degradation (66, 165, 178).

In the following sections, we will group proteins whose function is regulated via H₂O₂-mediated cysteine oxidation on the basis of their biochemical and cellular functions (Fig. 2).

Intracellular Messenger Function of H₂O₂ and Its Regulation by Prx

The Prx family of peroxidases reduces H₂O₂ or alkyl peroxide (ROOH) with a conserved cysteine residue (the peroxidatic Cys or C_P–SH) serving as the site of oxidation by peroxides (147, 149). In addition to the C_P–SH located in the NH₂-terminal region of the molecule, most Prx enzymes contain an additional conserved cysteine (the resolving Cys or C_R–SH) in the COOH-terminal region. On the basis of the location or absence of the C_R residue, Prxs are classified into 2-Cys, atypical 2-Cys, and 1-Cys subfamilies. Mammalian cells express six isoforms of Prx: four 2-Cys Prxs (Prx I to IV), one atypical 2-Cys Prx (Prx V), and one 1-Cys Prx (Prx VI).

These isoforms vary in subcellular localization, with Prx I, II, and VI being localized mainly in the cytosol; Prx III being restricted to mitochondria; Prx IV being found predominantly in the ER; and Prx V being present in the cytosol, mitochondria, and peroxisomes. Peroxiredoxins are abundant proteins, with Prx I and Prx II together constituting a total of 0.2%–1% of soluble protein in cultured mammalian cells (25).

During the catalytic cycle of Prxs, the C_P–SH is first oxidized to sulfenic acid (C_P–SOH) by H₂O₂, which promotes a local conformational change that allows the reaction of C_P–SOH with C_R–SH to form a disulfide that is subsequently reduced by an appropriate electron donor to complete the cycle (136, 150). The C_P–SOH of 2-Cys Prxs is occasionally further oxidized to sulfinic acid (C_P–SO₂H) before the disulfide can be formed, resulting in inactivation of peroxidase function (197). This hyperoxidation is reversed by the ATP-dependent enzyme sulfiredoxin, which restores peroxidase activity (14, 189). The reversible hyperoxidation confers a chaperone function on 2-Cys Prxs in cells subject to severe oxidative stress (69). It has also been shown that the hyperoxidation allows accumulation of H₂O₂ and so H₂O₂-mediated signaling (80).

Kinetic analysis revealed that peroxides bind to Prx with submicromolar affinity (110, 132, 180). A specific binding site for peroxide was confirmed by determination of the crystal structure of Prx with bound peroxide or peroxide-mimicking molecules (58, 123, 136). Structural analysis also yielded a model for the transition state of the peroxidase reaction. In this model, all Prxs have an active-site structure with a nearly universal sequence, PXXXTXXC (where X is any amino acid), as well as a conserved arginine residue that is distant in sequence but located nearby in the three-dimensional fold. The transition state is characterized by an extensive hydrogen bond network formed by the C_P thiolate anion (C_P–S⁻), peroxide, the threonine and proline residues located within the universal PXXXTXXC motif, and the conserved arginine residue. The importance of the conserved arginine was also revealed by mutational study (121).

This network of hydrogen bonds not only provides a binding site for peroxide but also properly aligns the peroxide for attack by the C_P sulfur atom as well as lowers the activation energy for both breaking of the O–O bond of peroxide and formation of the S–O bond of C_P–SOH by stabilizing the transition state intermediate. In accord with this model, the second-order rate constants for the oxidation of C_P–SH to C_P–SOH measured in several laboratories are very high, lying in the range of 1 × 10⁶–1 × 10⁸ M ⁻¹ s⁻¹ (38, 110, 131, 132, 135, 139, 180).

H₂O₂ is distinct from other messenger molecules in that it acts not by binding to effectors but by oxidizing cysteine residues of effector proteins. The mechanism by which H₂O₂ selects its target proteins and oxidizes specific cysteine residues thereof is not well understood, however. It has been generally accepted that such selective oxidation is attributable to the fact that thiol oxidation is affected by pK _a (where K _a is the acid dissociation constant). The pK_a of a typical proteinaceous cysteine thiol is ∼8.4, but characteristics of the microenvironment centered on the cysteine residue, such as proximity to a positively charged basic residue, can lower this value and thereby increase the susceptibility of the thiol group to deprotonation (oxidation) at physiological pH.

The active-site cysteines of members of the PTP family thus have low pK_a values (4.6–5.5) (105, 169), and the rates of Cys–SH oxidation to Cys–SOH by H₂O₂ for PTPs (PTP1B, Cdc25, leukocyte antigen-related, and vaccinia H1 related) have been kinetically measured as 9–160 M ⁻¹ s⁻¹, values that are 10–180 times that (0.87 M ⁻¹ s⁻¹) for oxidation of the glutathione thiol (pK_a of 8.4) (42, 169, 188). Although the effect of a low pK _a on thiol reactivity is clear, the target thiol of PTPs still reacts with H₂O₂ at a rate that is several orders of magnitude slower than that for the catalytic thiol (C_P–SH) of Prxs. Given that Prxs are abundant proteins that are present in various compartments of the cell, cysteine residues of H₂O₂ target proteins such as PTPs are at a competitive disadvantage for reaction with H₂O₂.

Recent studies have revealed that this disadvantage can be overcome by transient inactivation of neighboring Prx molecules so as to allow the target cysteines to react with H₂O₂ (see below). Furthermore, in some instances, redox-regulated proteins are not directly oxidized by H₂O₂, their oxidation instead being mediated by Prxs. In such a scenario, Prx is first oxidized by H₂O₂ and then transfers its oxidation state to the redox-regulated protein, thus serving as both a sensor and transducer of H₂O₂ signaling (also see below).

Localized Regulation of H₂O₂ Concentration at the Plasma Membrane

Light was shed on the mechanism by which H₂O₂ produced in response to receptor engagement is able to propagate its signal by the finding that Prx I localized in the vicinity of the receptor can be reversibly inactivated. Prx I becomes phosphorylated on Tyr¹⁹⁴ by an Src family kinase in several cell types stimulated via various receptors, including the PDGFR, EGFR, TCR, and the B cell receptor (190).

In vitro analysis revealed that such tyrosine phosphorylation of Prx I impairs its peroxidase function. Importantly, tyrosine-phosphorylated Prx I, which accounts for about 0.2%–0.4% of total Prx I in PDGF-stimulated NIH3T3 cells, was found to be present exclusively in the low-density, detergent-insoluble fraction of these cells. This fraction is derived mainly from lipid rafts, which are membrane microdomains enriched in glycosphingolipid-cholesterol and are similar in their lipid composition and resistance to detergent solubilization to other biochemically well-defined membrane structures known as caveolae.

Lipid rafts and caveolae compartmentalize cellular processes by serving as organizing centers for the assembly of signaling molecules. Growth factor and immune receptors, Nox enzymes, and Src family kinases are all membrane-associated proteins that are highly localized to lipid rafts (3, 85, 133). Nox1 appears to be the major source of H₂O₂ in cells stimulated with PDGF or EGF (190), with this Nox1-derived H₂O₂ being critical for the propagation of growth factor signaling, including protein tyrosine phosphorylation, Akt and MAPK activation, and target gene transcription (4, 90, 146, 173).

These various findings indicate that localized H₂O₂ production by Nox1 might not be sufficient to support the messenger function of this molecule in growth factor-stimulated cells, with inactivation of Prx I by phosphorylation on Tyr¹⁹⁴ being necessary to protect the pool of H₂O₂ signaling molecules from destruction by this abundant enzyme so as to allow the oxidation of target molecules such as PTPs, Src family kinases, and PTEN (Fig. 3). PTPs such as SHP-1 and SHP-2 associate with tyrosine-phosphorylated (activated) receptors via their SH2 domains. PTEN is present mostly in the cytosol, but it transiently associates with lipid rafts to antagonize PI3K signaling (87).

OPEN IN VIEWER

FIG. 3.

Scheme depicting the role of tyrosine phosphorylation of Prx I in H₂O₂ accumulation around lipid rafts as well as that of the accumulated H₂O₂ in intracellular signaling. Ligation of a growth factor receptor induces the production of H₂O₂ through activation of Nox located near the receptor. The activated receptor also induces the tyrosine phosphorylation of lipid raft-associated Prx I through activation of an Src family PTK, resulting in inactivation of Prx I, which would otherwise destroy the newly generated H₂O₂. Inactivation of Prx I thus allows the accumulation of H₂O₂ in the submembrane compartment, which in turn promotes further phosphorylation and inactivation of Prx I through activation of Src family kinases. The accumulated H₂O₂ also inactivates PTPs in the vicinity of the activated receptor by oxidizing their active-site cysteine thiol (–SH) to various oxidation states (–S_OX) via formation of an intramolecular disulfide, glutathionylation, or formation of sulfenamide. Such inhibition of PTPs by H₂O₂ is critical for growth factor signaling because activation of receptor PTKs alone is not sufficient to achieve the levels of protein tyrosine phosphorylation necessary for such signaling, with the concurrent inhibition of PTPs by H₂O₂ preventing the futile cycle of phosphorylation and dephosphorylation. For a similar reason, the receptor-induced activation of PI3K is not sufficient for the accumulation of PI(3,4,5)P₃ (PIP₃) needed for Akt signaling, with H₂O₂-dependent inactivation of PTEN through oxidation of its active-site cysteine thiol to disulfide also being necessary. The inactivation of PTPs further promotes Prx I phosphorylation, whereas PI(3,4,5)P₃ enhances growth factor-induced Nox activation, thus completing two more positive feedback loops for the elevation of H₂O₂ levels in addition to that involving Src phosphorylation. Nox, NADPH oxidase; PI3K, phosphatidylinositol 3-kinase; Prx, peroxiredoxin; PTEN, phosphatase and tensin homolog; PTK, protein tyrosine kinase; PTP, protein tyrosine phosphatase.

The accumulation of H₂O₂ around lipid rafts would be expected to promote further phosphorylation and inactivation of Prx I through both activation of Src family kinases and inactivation of PTPs (Fig. 3). These two positive feedback loops likely sustain the H₂O₂ levels necessary for the oxidation of target cysteines. The oxidative inhibition of PTPs by H₂O₂ is critical for growth factor signaling because activation of receptor PTKs alone is not sufficient to achieve the levels of protein tyrosine phosphorylation necessary for such signaling, with the concurrent inhibition of PTPs by H₂O₂ preventing the futile cycle of phosphorylation and dephosphorylation.

For a similar reason, oxidative inactivation of PTEN by H₂O₂ is necessary for the accumulation of PI(3,4,5)P₃ that is generated by PI3K and needed for Akt signaling (90). PI(3,4,5)P₃ was also found to be required for PDGF-induced Nox activation (6), thus revealing the operation of another positive feedback loop (Fig. 3). A similar scenario likely applies to H₂O₂ signaling in cells stimulated via the B cell receptor or TCR.

Whereas both Prx I and Prx II are present in the low-density, detergent-insoluble membrane fraction, only Prx I is phosphorylated by PTKs. Given that the accumulation of a small amount of H₂O₂ at the plasma membrane can trigger rapid signal amplification through the action of multiple positive feedback loops (Fig. 3), this presence of Prx II in addition to Prx I might serve as a safeguard to prevent inappropriate activation of signaling cascades by basal production of H₂O₂ at the cell membrane. Consistent with this notion, cells derived from Prx II-deficient mice show enhanced tyrosine phosphorylation of PDGFR when stimulated with PDGF (34).

A similar protective role for Prx II at specific microdomains was also demonstrated in human aortic vascular endothelial cells stimulated with VEGF (76). A fraction of Prx II, but not of Prx I, was thus detected in caveolae of these cells. Furthermore, depletion of Prx II rendered VEGFR2 insensitive to VEGF stimulation as a result of the formation of an intramolecular disulfide between Cys¹¹⁹⁹ and Cys¹²⁰⁶ in its COOH-terminal tail, suggesting that Prx II protects the PTK activity of VEGFR2 from oxidative inactivation (76). Whereas VEGFRs are located in both caveolae and other regions of the plasma membrane, the protective function of Prx II was found to be restricted to caveolae, with the oxidized form of VEGFR2 being detected only in this fraction of the endothelial cells after Prx II knockdown. Deficiency of Prx II was also shown to impair VEGFR2 activation and angiogenesis in mice (76).

Functional differences between Prx I and Prx II are also apparent in the phenotypes of the corresponding knockout mice. Mice lacking Prx I thus manifest an increased susceptibility to cancer, whereas those deficient in Prx II do not (127), suggesting that Prx I may function as a tumor suppressor. Prx I directly binds to PTEN and thereby protects its lipid phosphatase function from inactivation by removing H₂O₂, whereas Prx II does not bind to PTEN (23).

Whether PTEN located at lipid rafts is also protected by Prx I binding and whether such PTEN-bound Prx I becomes inactivated through phosphorylation are questions that remain unanswered, however. Nevertheless, tumor suppression by Prx I is likely due, at least in part, to its ability to remove H₂O₂ from the region surrounding growth factor receptors and its ability to protect PTEN by binding to the lipid phosphatase. As described in the following section, the tumor suppressor function of Prx I is also likely related to its role as a regulator of H₂O₂ concentration around the centrosome.

Localized Regulation of H₂O₂ Concentration Around the Centrosome via Cdk1-Dependent Phosphorylation of Prx I

Another example of spatially confined inactivation of Prx is provided by Prx I phosphorylation at Thr⁹⁰ in mitotic cells. Mammalian 2-Cys Prx enzymes (Prx I–IV) contain a consensus sequence, (S/T)PX(K/R), for phosphorylation by Cdks. Among 2-Cys Prxs, Prx I is phosphorylated most efficiently by the purified Cdk1–cyclin B complex, with the susceptibility to such phosphorylation then decreasing in the order Prx II, Prx III, and Prx IV, and such phosphorylation inactivates peroxidase function (27).

Threonine phosphorylation of Prx I occurs during mitosis, being virtually undetectable in interphase. The amount of threonine-phosphorylated Prx I in mitotic HeLa cells was estimated to be ∼0.4% of total Prx I, suggesting that phosphorylation of Prx I at Thr⁹⁰ is likely a localized event, as in the case of its phosphorylation at Tyr¹⁹⁴. Indeed, Prx I was found to be present at the centrosome, and such centrosome-associated Prx I, but not cytosolic Prx I, was enriched in the Thr⁹⁰-phosphorylated form of the protein. Furthermore, threonine-phosphorylated Prx I was detected at the centrosome of HeLa cells during the early stages of mitosis (prophase, prometaphase, and metaphase) but not during interphase or late mitotic stages (anaphase, telophase, and cytokinesis) (101).

In addition to being the major microtubule-organizing center of mammalian cells, the centrosome acts as a centralizing platform that coordinates the complex sequence of events that lead to the activation of Cdk1 and the consequent entry of cells into mitosis (18). Many proteins that function in the mitotic entry network—such as cyclin B, the phosphatases Cdc25B and Cdc25C, and the kinases Polo-like kinase 1 (Plk1) and Aurora A—are specifically recruited to the centrosome in G₂ phase of the cell cycle, resulting in the generation of high local concentrations of these mitotic activators (18).

Activation of the Cdk1–cyclin B complex occurs first at the centrosome during prophase, and its amplification through multiple feedback loops involving cyclin B, Plk1, and Aurora A also occurs at this organelle (103). Exit of cells from mitosis also requires that these regulators (cyclin B, Plk1, and Aurora A) be degraded in a timely manner. Such degradation is mediated by the 26S proteasome and is triggered by ubiquitylation of the regulators by the multisubunit ubiquitin ligase APC/C (anaphase-promoting complex/cyclosome), which requires CDC20 homolog 1 (Cdh1) as an activator protein (138, 140, 192).

Cdh1 is prevented from interacting efficiently with APC/C as a result of its phosphorylation by Cdks during S and G₂ phases, but it is able to activate APC/C after its dephosphorylation by Cdc14B (138, 192). A fraction of each of APC/C, Cdh1, and Cdc14B is present at the centrosome (140, 191), as are Cdk1, cyclin B, Plk1, and Aurora A. Activation of APC/C also occurs at the centrosome (18, 103). At the onset of mitosis, cyclin B, Plk1, and Aurora A need to be protected from APC/C–Cdh1-dependent ubiquitylation and degradation to ensure the robust and irreversible activation of Cdk1–cyclin B. Such protection appears to be achieved through suppression of the Cdc14B-mediated dephosphorylation of Cdk1-phosphorylated Cdh1.

In the absence of such suppression of centrosomal Cdc14B activity, further activation of Cdk1 would not be expected to occur because of the premature degradation of cyclin B, Plk1, and Aurora A. The catalytic Cys³¹⁴ residue of Cdc14B readily forms an intramolecular disulfide with Cys²²⁸ in cells exposed to H₂O₂, with Cdc14B appearing to be more sensitive to such oxidation than is PTEN.

The intracellular concentration of ROS—as measured with the chloromethyl derivative of 2′,7′-dichlorodihydrofluorescein diacetate, an irreversible fluorescent probe sensitive to various oxidants—has been found to change during cell cycle progression (62). Analysis with a genetically encoded fluorescent probe (HyPer) that reacts reversibly and specifically with H₂O₂ (10) revealed that the intracellular level of H₂O₂ increased steadily from G₁ to S phase and was maximal at G₂-M (101). Whereas HyPer fluorescence reflects global changes in H₂O₂ concentration during cell cycle progression, however, the production of H₂O₂ and its elimination occur via highly localized and regulated processes. The measured global changes in H₂O₂ levels may thus not be representative of local changes.

The accumulation of H₂O₂ in S and G₂-M phases might be due to increased energy metabolism, which is accompanied by mitochondrial H₂O₂ production. H₂O₂ derived from Nox4 has also been shown to be required for G₂-M progression (195). Indeed, the Nox inhibitor diphenyleneiodonium attenuates mitotic entry in HeLa cells (101). In addition, inhibition of cytosolic phospholipase A₂ activity—which produces arachidonic acid for H₂O₂-generating lipoxygenases—as well as inhibition of lipoxygenase activity each delayed mitotic entry, suggesting that G₂-M progression is affected by H₂O₂ produced from multiple sources (101).

Mitochondrially produced H₂O₂ might also be directed to the centrosome, as appears to be the case in T cells. Ligation of the TCR thus induces H₂O₂ production that is required for the activation of various signaling molecules, including ERK (43). This H₂O₂ might be produced in T cells by different sources, including Nox2 and mitochondria (68, 162). Mitochondria accumulate in a region close to the immunological synapse formed between T cells and antigen-presenting cells and trigger localized Ca²⁺ entry that is critical for T cell activation and proliferation (88, 142, 160).

TCR activation also induces translocation of the centrosome to the vicinity of the synapse (1). Changes in mitochondrial morphology and positioning might therefore determine the site of H₂O₂ release from these organelles and thus the proteins targeted for oxidation by the released H₂O₂. The immediate targets of H₂O₂ in the region close to the synapse include two Src family kinases (Lck and Fyn) that are associated with the TCR.

Taken together, these many observations indicate that, during interphase, the centrosome is shielded from the effects of H₂O₂ by Prx I that is physically associated with the organelle (Fig. 4A). During early mitosis, however, centrosomal Prx I is inactivated as a consequence of phosphorylation by Cdk1, resulting in exposure of the centrosome to the high tide of H₂O₂ and consequent inactivation of Cdc14B (Fig. 4A). This inactivation of Cdc14B is critical to prevent the premature dephosphorylation of Cdh1 and the consequent untimely ubiquitylation of cyclin B, Plk1, and Aurora A by APC/C–Cdh1 (Fig. 4B). Consistent with this scenario, downregulation of the pericentrosomal H₂O₂ level through forced expression of a centrosome-targeted form of catalase delayed mitotic entry (101).

OPEN IN VIEWER

FIG. 4.

Scheme for the control of pericentrosomal H₂O₂ levels through Cdk1-mediated phosphorylation of Prx I and for regulation of the stability of the mitotic activators cyclin B, Aurora A, and Plk1. (A) Control of H₂O₂ concentration around the centrosome through reversible phosphorylation of Prx I. Whereas the intracellular concentration of H₂O₂ increases as the cell cycle progresses to G₂ phase, the centrosome is shielded from the high tide of H₂O₂ by the peroxidase activity of Prx I associated with the organelle. At the onset of mitosis, centrosome-bound Prx I is inactivated through threonine phosphorylation by Cdk1–cyclin B, resulting in exposure of the centrosome to H₂O₂ and consequent inactivation of centrosome-bound Cdc14B through oxidation of its catalytic cysteine thiol to disulfide. Subsequent dephosphorylation of phosphorylated Prx I (Prx I-T-P) by PP1 and PP2A and consequent reactivation of its peroxidase function during late mitosis again shield the centrosome from H₂O₂. (B) Control of the stability of the mitotic activators cyclin B (Cyc B), Aurora A (AurA), and Plk1 by H₂O₂ around the centrosome. The sequential events of Prx I phosphorylation, exposure of the centrosome to H₂O₂, and inactivation of Cdc14B depicted in (A) prevent the dephosphorylation of Cdh1 phosphorylated by Cdk1–cyclin B (Cdh1–P). Given that the multisubunit ubiquitin ligase APC/C is activated specifically by the nonphosphorylated form of Cdh1, ubiquitylation and subsequent 26S proteasome-mediated degradation of centrosome-bound mitotic activators are prevented while Prx I is inactive. APC/C, anaphase-promoting complex/cyclosome; Cdh1, CDC20 homolog 1; Cdk, cyclin-dependent kinase; Plk1, Polo-like kinase 1; PP1, protein phosphatase 1; PP2A, protein phosphatase 2A.

Later during mitosis, Prx I is dephosphorylated by protein phosphatase 1 (PP1) and PP2A, which are also concentrated at the centrosome (192), and the centrosome is consequently again shielded from H₂O₂, allowing Cdc14B activation, Cdh1 dephosphorylation, and activation of APC/C–Cdh1 (Fig. 4).

The phosphatases Cdc25B and Cdc25C, which function during the entry of cells into mitosis, are not sensitive to oxidation by H₂O₂ under the same conditions as those supportive of the oxidation of PTEN and Cdc14B. The accumulation of H₂O₂ at the centrosome during the early stages of mitosis as the result of Prx I inactivation therefore does not lead to the inactivation of centrosome-associated Cdc25B and Cdc25C, with the role of these phosphatases in the feedback loops underpinning activation of Cdk1–cyclin B thus being maintained. The oxidative inactivation of Cdc25C has been suggested to occur at higher levels of H₂O₂ as a means to induce cell cycle arrest (159).

The regulation of APC/C–Cdh1 by endogenous ROS (H₂O₂) was first demonstrated by the observation that antioxidant treatment of synchronized cells induced cell cycle arrest in late G₁ phase (62). Although cyclin A expression is induced at the transcriptional level in early G₁ phase, cyclin A protein does not accumulate until late during the G₁-S transition as a result of its ubiquitylation by APC/C–Cdh1. The activity of cyclin A-associated Cdk2 (or Cdk1) is required to initiate DNA synthesis, but endogenous H₂O₂ is required to inactivate APC/C–Cdh1 activity and thereby to allow cyclin A accumulation at late G₁ phase (62).

The mechanism of APC/C–Cdh1 inactivation by H₂O₂ at late G₁ phase is not known, but our knowledge of the control of pericentrosomal H₂O₂ levels (101) suggests that it is also achieved through a combination of Cdh1 phosphorylation by Cdks and Cdc14B inactivation by H₂O₂. Given that Prx I associated with the centrosome remains active (not phosphorylated) during all of interphase, however, the centrosome is not likely the site of Cdc14B inactivation by H₂O₂ at this time.

In quiescent cells, APC/C–Cdh1 must remain active to ensure that certain positive regulators of S and M phases do not accumulate prematurely. Given that mitosis persists in the absence of Cdk1 activity when phosphatase activity is suppressed (167) and that H₂O₂ can inactivate phosphatases and APC/C–Cdh1 at the centrosome, the presence of Prx I at the centrosome might serve as a safeguard to prevent unscheduled mitotic entry as a result of an accidental increase in H₂O₂ abundance such as might occur in response to ultraviolet radiation or in association with inflammation. Consistent with this notion, Cdc14B, Cdh1, and Prx I have all been suggested to function as tumor suppressors (51, 127, 191).

Sensor and Transducer Function of Prx

In addition to being associated with subcellular compartments such as lipid rafts and the centrosome and its regulation of the susceptibility of proteins present in these compartments to oxidation by H₂O₂, Prx is able to catalyze the oxidation of H₂O₂ effector molecules. In this process, the catalytic cysteine sulfhydryl group (C_P–SH) of Prx is first oxidized to sulfenic acid or disulfide intermediates, either of which can then transfer its oxidation state to an effector protein to transduce the H₂O₂ signal (Fig. 5). Binding of the effector protein to Prx can theoretically occur at any stage of and persist throughout the Prx catalytic cycle. It is also possible that Prx and the effector protein are recruited for transient disulfide formation by a scaffold protein.

OPEN IN VIEWER

FIG. 5.

Model for the sensor and transducer function of Prx in H₂O₂ signaling. The catalytic cysteine thiol of Prx is rapidly and selectively oxidized by H₂O₂. The oxidized Prx, in either the sulfenic or disulfide state, then forms an intermolecular disulfide-linked intermediate with a bound effector protein. Resolution of the disulfide by reaction with another Cys–SH of the effector results in regeneration of reduced Prx and oxidation of the effector protein, the latter process leading to a change in effector function.

Peroxidase-catalyzed protein thiol oxidation by H₂O₂ was first demonstrated in yeast, in which the AP1-like transcription factor Yap1 was shown to be oxidized indirectly through the action of a thiol-dependent peroxidase (41, 185).

It was later shown that H₂O₂ produced by ER membrane-associated oxidoreductin 1 (Ero1) in mammalian ER can be used to oxidize Prx IV, the ER-specific isoform of Prx, and that the oxidized state of Prx IV can then be transferred to the thiols of protein disulfide isomerase (PDI). The resulting formation of an intramolecular disulfide in PDI then allows PDI to promote protein folding in the ER by transfer of the disulfide to a target protein (177, 201). The physical interaction of Prx IV with PDI was revealed by determination of the crystal structure of oxidized Prx IV bound to ERp44 (an ER-specific PDI) (196). It was thus suggested that Prx IV functions as a sensor of H₂O₂ in PDI-mediated protein folding.

ASK1 is activated subsequent to oligomerization mediated by the formation of intermolecular disulfide bonds in cells exposed to H₂O₂ (118). This disulfide-dependent oligomerization is catalyzed by Prx I, which forms a transient disulfide-linked intermediate with ASK1 in H₂O₂-treated cells (72). Knockdown of Prx I thus resulted in inhibition of both ASK1 oxidation and phosphorylation of the downstream kinase p38, suggesting that Prx I can function as a sensor of H₂O₂ and transduce its signal to ASK1. This ASK1-activating function was found to be specific to Prx I, with Prx II exhibiting a negative effect on ASK1 oligomerization, probably because it lowers cytosolic H₂O₂ levels.

Prx I was also shown to play a role in the oxidation of APE1/Ref-1, a multifunctional protein that contributes to both DNA repair and transcriptional regulation, with Prx I-catalyzed APE1/Ref-1 oxidation having been proposed as a mechanism to prevent NF-κB activation and upregulation of interleukin-8 expression (125).

On the contrary, Prx II is thought to play a sensor and transducer role in the formation of disulfide-linked oligomers of STAT3 or DJ-1 (48, 168). Dimerization and tetramerization of STAT3 were initially observed in cells exposed to H₂O₂ (100), and this oligomerization was shown to be mediated by the oxidation of several cysteine residues located in the DNA binding and COOH-terminal transactivation domains of STAT3 and to result in diminished STAT3 binding to serum-inducible elements in cells treated with H₂O₂ or interleukin-6 (100). It was then shown that Prx II catalyzes both the dimerization and tetramerization of STAT3 in a reaction involving the formation of disulfide-linked Prx II–STAT3 intermediates.

Prx I did not form such disulfide-linked conjugates with STAT3. Two cysteine residues (Cys⁷¹² and Cys⁷¹⁸) in the COOH-terminal transactivation domain and two cysteines (Cys⁴¹⁸ and Cys⁴²⁶) in the DNA binding domain of STAT3 were found to be critical for dimerization and tetramerization, respectively.

Importantly, Prx II-catalyzed STAT3 oligomerization was observed in cells stimulated with interleukin-6 or oncostatin M (168). In contrast to the case of Prx IV and PDI, direct evidence for the physical association of Prx II and STAT3 has not been obtained to date. An observed dominant negative effect of a catalytically inactive form of Prx II (in which both C_P and C_R are substituted by serine) on STAT3 oligomerization, however, suggests that such physical association occurs before the formation of disulfide-linked conjugates, providing specificity for the relay of oxidative equivalents from H₂O₂ to Prx II and then to STAT3 (168). Prx II also interacts with DJ-1 and promotes its disulfide-linked homodimerization in cardiomyocytes exposed to H₂O₂ (48).

The catalytic cysteine (C_P) of Prx enzymes is highly sensitive to oxidation by H₂O₂, whereas direct oxidation of the cysteines of H₂O₂ target proteins is generally very slow. It is also likely in the interest of cell survival that the concentration of the potentially toxic oxidant H₂O₂ not remain elevated for the time required for direct oxidation of kinetically inefficient targets. The Prx-catalyzed oxidation of target proteins is thus likely a biologically selected mechanism of H₂O₂ signaling.

The few examples of such regulation identified to date likely reflect difficulties in capturing the disulfide-linked Prx-target intermediate, which is formed transiently and in specific compartments. It is important to note, however, that Prx isoforms are known to interact with many other proteins (11, 150). Moreover, Prx I and Prx III each transiently form a number of disulfide-linked higher molecular weight intermediates in H₂O₂-treated cells, suggesting that cysteine residues of multiple proteins are targets for Prx-catalyzed oxidation (72).

Perspective

As our understanding of signaling networks has become more advanced, the importance of strict temporal and spatial regulation of signaling events has become evident. This is especially so for events related to signaling by the precarious molecule H₂O₂. The outcome of oxidative modification of an H₂O₂ effector molecule depends on contextual factors such as cell type as well as the subcellular domain and stage of the cell cycle in which the oxidation occurs. Oxidative modification and signaling events observed in cells exposed to exogenous H₂O₂, even at physiologically relevant concentrations, can thus be very different from those that occur in response to a specific physiological signal.

Several challenges remain to achieving a full understanding of H₂O₂ signaling. First, it is important that changes in H₂O₂ levels be monitored at discrete locations during cell cycle progression. Although sensitive probes for such monitoring are not yet available, substantial progress has recently been made toward this goal by taking advantage of the high sensitivity of Prx enzymes to H₂O₂ and their capacity to transduce the H₂O₂ oxidation state to target proteins. Genetically encoded fluorescent probes for H₂O₂ that are based on the fusion of a Prx isoform to green or yellow fluorescent protein and in which Prx catalyzes the formation of an intramolecular disulfide within the probe and thereby changes its fluorescence properties have been designed (115, 183).

Second, given that Prxs are localized to various subcellular compartments and regulate local H₂O₂ levels, unraveling of Prx regulation at specific locations—as exemplified by Prx I regulation via tyrosine phosphorylation at lipid rafts and via threonine phosphorylation at the centrosome—will be required. Prx enzymes also undergo additional posttranslational modifications such as lysine acetylation (163) as well as cysteine glutathionylation (26) and nitrosylation (145), although the in vivo relevance of these additional modifications remains to be established.

Third, given that Prx-catalyzed oxidation appears to be an economic means by which to convey the H₂O₂ message with spatiotemporal precision and target specificity, cysteine residues of many signaling proteins are likely to be oxidized indirectly via the disulfide transferase activity of Prx. Identification of such target proteins will be difficult, however, because only a small fraction of the total pools of a given Prx and target protein is likely to form a disulfide-linked intermediate and such an intermediate is likely to be transient.

Fourth, tools to selectively detect the oxidized form of H₂O₂ target proteins are needed. Again, given that Prx-catalyzed oxidation is likely restricted to a small subpopulation of target molecules localized in a specific subcellular compartment, the development of such tools for various target proteins will be a long-term goal. However, the successful development of phage antibodies that specifically recognize the conformation of oxidized PTP1B but not that of the reduced form (61) may provide a guide for the accomplishment of this goal.