Report

Regeneration of Peroxiredoxins by p53-Regulated Sestrins, Homologs of Bacterial AhpD

See allHide authors and affiliations

Science  23 Apr 2004:
Vol. 304, Issue 5670, pp. 596-600
DOI: 10.1126/science.1095569

Abstract

Acting as a signal, hydrogen peroxide circumvents antioxidant defense by overoxidizing peroxiredoxins (Prxs), the enzymes that metabolize peroxides. We show that sestrins, a family of proteins whose expression is modulated by p53, are required for regeneration of Prxs containing Cys-SO2H, thus reestablishing the antioxidant firewall. Sestrins contain a predicted redox-active domain homologous to AhpD, the enzyme catalyzing the reduction of a bacterial Prx, AhpC. Purified Hi95 (sestrin 2) protein supports adenosine triphosphate–dependent reduction of overoxidized PrxI in vitro, indicating that unlike AhpD, which is a disulfide reductase, sestrins are cysteine sulfinyl reductases. As modulators of peroxide signaling and antioxidant defense, sestrins constitute potential therapeutic targets.

Reactive oxygen and nitrogen species (ROS and RNS, respectively) are by-products of metabolism that are destroyed by antioxidant systems. However, the same compounds function as important signal-transducing messengers. Hydrogen peroxide bursts are elicited by ligand-receptor interaction in major pathways, which include mitogen-activated protein kinase and nuclear factor κB pathways (1), affecting cell proliferation and cell death. Antioxidant mechanisms must be thus tightly regulated to enable signaling but prevent oxidative damage.

Peroxiredoxins (Prxs), a family of thiol-containing peroxidases conserved from bacteria to mammals (2), are major reductants of endogenously produced peroxides in eukaryotes (3). In addition, Prxs catalyze decomposition of RNS (47). How do hydrogen peroxide signals evade this antioxidant defense? The answer seems to lie in the structure of eukaryotic 2-Cys Prxs (8) which, unlike their prokaryotic ancestors, have acquired increased sensitivity to inactivation through overoxidation of the catalytic center (9, 10). During background scavenging, the peroxidatic cysteine is oxidized to Cys-SOH, which forms a disulfide bridge with the resolving cysteine located in the other subunit of the Prx dimer. This disulfide bond is then reduced by thioredoxin (1113). However, because the formation of the resolving disulfide bond is slow, high concentrations of ROS cause further oxidation of the peroxidatic cysteine to Cys-SO2H, yielding an inactive form of Prx that cannot be reduced by thioredoxin (8). Oxidative inactivation of Prxs was apparently advantageous during the evolution of eukaryotes, which use hydrogen peroxide as a signaling molecule (8). This mechanism allows hydrogen peroxide to halt peroxidase while passing its message to redox-sensitive components of signaling pathways (8, 14). However, after the peroxide signal is conveyed, the impaired antioxidant firewall needs to be promptly reestablished to prevent oxidative stress. Indeed, gradual recovery of Prxs has been detected after initial oxidative inactivation (13, 15, 16), and an enzyme catalyzing adenosine triphosphate (ATP)–dependent regeneration of Cys-SO2H, sulphiredoxin (Srx), has been detected in yeast (17). However, a similar activity remains to be identified in mammals and other animals in which the rate of regeneration of Prxs is critical for modulation of peroxide signaling and thus for cell-specific responses to oxidative stress.

We have recently shown that overexpression of the p53-regulated gene Hi95 (18), a member of the sestrin gene family (19), protects cells against apoptosis induced by ischemia or H2O2, suggesting that Hi95 is involved in defense against ROS (18). To gain insight into the functions of sestrins [Hi95, PA26 (20), and Sesn3], we undertook a detailed sequence analysis of these proteins (21). An iterative database search performed with the use of the PSI-BLAST program (22) revealed statistically significant sequence similarity (random expectation value, E ∼ 0.001) between a conserved region in the N-terminal part of the sestrins and a family of prokaryotic proteins, which includes the Mycobacterium tuberculosis AhpD, an enzyme that catalyzes reduction of the peroxiredoxin AhpC (23, 24). The presence of a domain with a fold common to the AhpD family proteins and the sestrins was confirmed by threading analysis using the 3D-PSSM program (25). The homology spans at least the five N-terminal α-helices of the conserved region of the sestrins and the C-terminal α-helical domain of AhpD (23, 24). The structural and evolutionary autonomy of the conserved domain is supported by the fact that certain members of the AhpD family, such as caroboxymuconolactone decarboxylases, consist of this domain alone (Fig. 1). However, only the proximal cysteine of the essential cysteine dyad of AhpD (26) is conserved in sestrins.

Fig. 1.

Multiple alignment of sestrins and selected AhpD-family proteins (31). The alignment was constructed with the use of the MACAW program (32). The secondary structure assignments are from the AhpD structure (Protein Data Bank code 1KNC); H indicates α-helix [numbered as in (23)], and T indicates a hydrogen-bonded turn. Amino acid residues conserved (identical or similar) in aligned sequences are shown by shading; darker shading indicates identical or closely similar residues. The two cysteines of the AhpD family proteins that form a transient disulfide bridge are shown by reverse shading, and the cysteine that is conserved between the sestrins and the AhpD family is denoted by an asterisk. The positions of the aligned regions in the respective protein sequences are indicated in parentheses. CAEEL indicates Caenorhabditis elegans; DROME, Drosophila melanogaster; MYCTU, M. tuberculosis; DEIRA, Deinococcus radiodurans; METAC, Methanosarcina acetivorans; NEIME, Neisseria meningitides; SESN, sestrin; and CMLD, carboxymuconolactone decarboxylase. The GenBank accession number for Hi95_HUMAN is NP_113647; PA26_HUMAN, NP_055269; SESN_DROME, Q9W1K5; SESN_CAEEL, NP_490664; AhpD_MYCTU, NP_216945; DR1765_DEIRA, NP_295488; CMLD_METAC, NP_618609; and NMA1203_NEIME, NP_283969.

AhpD is a component of alkylhydroperoxide reductase, which in M. tuberculosis takes part in the defense against oxidative and nitrosative components of the host immune response (24). AhpD is responsible for regeneration of Prx AhpC, which is oxidized during reduction of peroxides and RNS (4, 24, 26). The homology of sestrins and AhpD and the indirect link of Hi95 to the regulation of ROS homeostasis (18) suggested that, like AhpD, sestrins might be redox enzymes catalyzing regeneration of Prxs. In AhpD, both cysteines of the Cis-Ser-His-Cys motif (Fig. 1) are essential for redox activity, with the distal cysteine directly involved in the nucleophilic attack on the substrate and the proximal cysteine catalyzing the resolution of the reaction intermediate via formation of an intramolecular disulfide bond in AhpD.

To determine whether Hi95 and PA26 affect hydrogen peroxide amounts in cell culture, we inhibited expression of Hi95 or PA26 more than 50-fold with the use of small interfering RNAs (siRNAs) (Fig. 2A). The level of intracellular ROS determined by dichlorodihydrofluorescein (DCF) fluorescence 48 hours after introduction of sestrin siRNAs was significantly increased compared to control cells, regardless of pretreatment with 500 μM H2O2 (Fig. 2B and fig. S3, A and B). Knocking down expression of Hi95 also compromised the ability of cells to detoxify RNS generated after treatment with NO donors (fig. S3C). By contrast, overexpression of Hi95 and PA26 cDNAs lead to a notable reduction in ROS (Fig. 2, C to E, and figs. S3D and S4).

Fig. 2.

Expression of Hi95 and PA26 affects intracellular ROS amounts. (A) Northern (top three rows) and Western (bottom row) analyses of untreated or H2O2-treated (500 μM, 6 hours) RKO cells. Inhibition of endogenous Hi95 and PA26 was achieved by infection with lentivirus vectors expressing the appropriate siRNAs 48 hours before H2O2 treatment. (B) Basal and H2O2-induced (500 μM, 6 hours) ROS amounts in RKO cells expressing siRNAs to Hi95 or to PA26, measured by fluorescence-activated cell sorting (FACS) after DCF treatment. FL1-H, fluorescence level. (C) Expression of Hi95 and PA26 proteins in MCF7 cells overexpressing appropriate cDNAs from lentiviral constructs as determined by Western analysis with antibodies against Flag M2. (D) Basal and H2O2-induced (500 μM, 2 hours) ROS amounts in MCF7 cells overexpressing Hi95 or PA26, measured by FACS after DCF treatment. The less prominent relative antioxidant effect of PA26 overexpression in H2O2-treated cells can be explained by the more robust induction of endogenous PA26 mRNA in response to H2O2 (fig. S5), which precludes an additional effect of the PA26 construct. (E) Basal and H2O2-induced ROS amounts in RKO and RKO/siHi95 cells expressing Ser125 mutant of Hi95 or Ser130 mutant of PA26. The siRNA to Hi95 only slightly decreased the amount of exogenous Hi95 protein. The ROS amounts were measured by FACS after DCF treatment and expressed as the mean ± SEM intensity of cell fluorescence.

Inhibition of Hi95 or PA26 negatively affected the growth rate and viability of the cells (Fig. 3 and fig. S6), especially after introduction of siRNA to PA26, which resulted in deterioration of the cultures within a week (Fig. 3A and fig. S6A). Inhibition of Hi95 in human embryonic fibroblasts (HEFs) resulted in slower proliferation, expression of senescence-associated β-galactosidase, and a dramatic increase in p53 (fig. S6B). In RKO cells, inhibition of Hi95 elicited notable sensitization to H2O2-induced apoptosis, which contrasts with the increase in resistance after overexpression of Hi95 or PA26 cDNAs (Fig. 3C).

Fig. 3.

Inhibition of Hi95 and PA26 affects cell viability and proliferation. (A) Human embryonic fibroblasts (HEFs) 7 days after infection with empty lentiviral vector (pLV) or vectors expressing siRNAs to HPV18 E6, Hi95, or PA26 (phase-contrast microscopy). (B) Growth curves of HEFs after introduction of siRNAs to Hi95 or PA26. (C) Effect of Hi95 and PA26 expression on survival of RKO cells after H2O2 treatment (1 mM, 10 hours) as revealed by Annexin-V (Roche, Indianapolis, IN) staining.

We next tested the role of the cysteine that is conserved between sestrins and AhpD in the apparent redox function of sestrins. Replacement of the conserved Cys125 of Hi95 or Cys130 of PA26 with serine resulted in complete loss of the sestrins' ability to decrease intracellular ROS (Fig. 2E) and to protect against cell death after H2O2 treatment (fig. S6C). Moreover, overexpression of the mutant proteins resulted in an increase in ROS and a decrease in viability after H2O2 treatment, possibly because of their dominant-negative effect (Fig. 2E and fig. S6C). To test the possibility that another cysteine of sestrins could be involved in the formation of the transient disulfide bond, we mutated each of the two cysteines of Hi95 that are conserved in all three sestrins (Cys399 and Cys430) as well as the nonconserved Cys314. The latter substitution did not affect the activity of Hi95 in H2O2-treated or untreated cells, whereas replacement of either Cys399 or Cys430 reduced the antioxidant potential of Hi95 in H2O2-treated cells in which the endogenous Hi95 was inhibited by siRNA (fig. S4A). One possibility is that two or more cysteines of sestrins catalyze the resolution of intermediates during catalysis. However, given the mild effect of Cys399 and Cys430 mutations, it seems more likely that the conserved Cys125 is the only redox-active cysteine in Hi95, whereas substitution of Cys399 or Cys430 affected the enzymatic function indirectly by changing the protein conformation.

Given the homology with AhpD, it seemed likely that sestrins function similarly in the Prx pathway. Alternatively, however, the ability of sestrins to down-regulate peroxides could be because of their participation in other antioxidant systems. In eukaryotes, glutathione peroxidase controls peroxide amounts in addition to and independently of the Prxs. Unlike Prxs, which function in conjunction with the thioredoxin reductase (TrxR) system (2), glutathione peroxidase is reduced by glutathione (27). Treatment of RKO cells with buthioninesulfoximine (BSO), which depletes intracellular glutathione (28), or with the TrxR inhibitor 2,4-dinitro-1-chlorobenzene (DNCB) (29) resulted in an increase in intracellular ROS. However, in Hi95-deficient cells, only treatment with BSO produced an additional increase in ROS (Fig. 4A), suggesting that Hi95 is functionally coupled to TrxR rather than to glutathione.

Fig. 4.

Sestrins are involved in Prx regeneration. (A) The effect of BSO (300 μM, 6 hours) or DNCB (20 μM, 6 hours) on ROS in control and Hi95-deficient RKO cells, estimated by FACS after DCF staining. (B) Colocalization graphs and Pearson's correlation (Rp) were obtained with the use of VayTek software (VayTec, Incorporated, Fairfield, IA) by analyzing corresponding confocal microphotographs (fig. S8A). (C) Coimmunoprecipitation of Hi95Flag or PA26Flag proteins with PrxI. The samples were precipitated with antibodies against either Flag or PrxI, and the corresponding Western blots were developed with reciprocal sets of antibodies. (D) Kinetics of PrxI regeneration after H2O2 treatment of RKO-PrxIFlag cells expressing different amounts of Hi95 or PA26. The cells were incubated with 35S methionine for 12 hours, treated with H2O2 (500 μM), and incubated for different periods in medium containing unlabeled methionine. Cell extracts were precipitated with antibodies against Flag, followed by separation of the proteins on 2D gels. (E) Cell-free assay demonstrating the enzymatic activity of Hi95 toward PrxI. Hi95Flag and Hi95Flag-125 were immunopurified. The purity of the obtained proteins was confirmed by Coomassie staining of acrylamide gels; no extra bands were detected (fig. S8A). Reduction of overoxidized PrxI by recombinant Hi95 was performed in the reaction mixture containing 1 μg of overoxidized PrxI (LabFrontiers, Seoul, South Korea), 4 μg or 0.4 μg Hi95Flag (or 4 μg Hi95Flag-125), 50 mM tris-HCl (pH = 7.0), 150 mM KCl, 1 mM DTT, 1 mM ATP, and 1 mM MgCl2, incubated at 30°C for 1 hour. The proteins were resolved on 2D electrophoresis and visualized with antibodies against PrxI (LabFrontiers).

To investigate whether the sestrins directly regulate Prxs, we examined whether Hi95 and PA26 interact with Prxs in the cell. Confocal microscopy after immunofluorescent staining revealed strict colocalization of PrxI with exogenously expressed Hi95 and PA26 in MCF7 cells, whereas there was no colocalization of Hi95 with mitochondrial PrxIII (Fig. 4B and fig. S7A). In immunoprecipitation experiments, antibodies to both PA26 and Hi95 precipitated PrxI equally efficiently in H2O2-treated and untreated cells (Fig. 4C and fig. S7B).

To evaluate the effect of sestrins on the redox state of Prxs, we measured the proportion of the native and Cys-SO2H forms of Prx on two-dimensional (2D) gels (9, 10). We introduced N-terminal Flag-tagged PrxI into RKO cells, labeled the cells with S35 methionine, and analyzed redox changes of the presynthesized protein after immunoprecipitation with antibodies against Flag. Two-dimensional gel separation showed that <5% of labeled PrxI was in the overoxidized, inactive form. Ten minutes after treatment with 500 μM of H2O2, >80% of PrxI migrated as the Cys-SO2H form, with nearly 40% recovering to the reduced form within the next 2 hours. Neither overexpression of Hi95 or PA26 cDNA nor inhibition of Hi95 with siRNA affected the initial extent of PrxI overoxidation when analyzed 10 min after treatment with H2O2. However, Hi95 and PA26 had a substantial effect on the rate of regeneration of the reduced form. Two hours after H2O2 treatment, there was almost no reduction of overoxidized PrxI in cells deficient in Hi95, in contrast to the nearly complete recovery of PrxI in the cells overexpressing either Hi95 or PA26 (Fig. 4D). The difference was less significant when the cells were allowed to recover for 6 hours after H2O2 treatment, which correlated with the induction of endogenous PA26 and Hi95 (fig. S5). A similar effect of sestrins was observed for PrxII (fig. S8A).

To test whether sestrins are directly involved in regeneration of Prxs, we assayed the activity of immunopurified Hi95Flag protein in vitro with the use of H2O2-pretreated PrxI. Incubation with Hi95Flag protein in the presence of ATP, MgCl2, and dithiothreitol (DTT) resulted in notable reduction of PrxI, whereas purified Ser125-Hi95Flag mutant protein was inactive; the reduction required ATP and MgCl2 (Fig. 4E). Thus, the activity of Hi95 is sufficient to catalyze reduction of PrxI in vitro, indicating that Hi95 is an essential, although perhaps not the only, component of cysteine sulfinyl reductase. In yeast that does not have sestrins, regeneration of Prx is catalyzed by Srx (17). We overexpressed recombinant human Srx ortholog in RKO and MCF7 cells and observed notable decrease in ROS after H2O2 treatment (30). However, in the in vitro assay, purified recombinant human Srx protein did not reduce PrxI and did not affect the reaction catalyzed by Hi95 when added in an equimolar amount (fig. S8C). Hence, the human Srx either functions in a separate pathway of ROS regulation or plays an auxiliary role in Prxs regeneration mediated by sestrins.

Our results indicate that, whereas Hi95 and PA26 do not protect Prxs from initial overoxidation after a peroxide burst, they substantially increase the rate of recovery of the overoxidized Prxs. Thus, sestrins appear to catalyze the reduction of Prxs in a manner that is functionally analogous to but mechanistically distinct from the role of AhpD in the AhpD/AhpC complex. Being a robust peroxidase, AhpC does not form overoxidized species, and, accordingly, AhpD acts as a disulfide reductase. In Prxs, the disulfides formed during background scavenging are reduced by thioredoxin. However, to reactivate the overoxidized form produced after peroxide bursts, cysteine sulfinyl reductase activity is required. Therefore, concomitantly with evolution of two-stage oxidation of Prxs in eukaryotes, which allows peroxide signaling, the redox activity of AhpD/sestrins apparently underwent a qualitative change to meet the new requirements.

Inactivation of Prxs by overoxidation during physiological signaling opens a gateway for peroxide-dependent signals (8, 14), allowing the peroxides to circumvent antioxidant control mechanisms. However, this gateway then needs to be closed to prevent development of oxidative stress. Our findings indicate that sestrins play a major role in reestablishing the antioxidant firewall. Modulation of sestrin expression by p53 might be part of an adaptive response mechanism that protects the cell against oxidative stress. In addition, substantial variations in the expression of PA26 and Hi95 in different tissues (18, 20) might be responsible for tissue-specific differences in peroxide-dependent signaling and for variability in the effects of elevated ROS and RNS. Lastly, modulation of sestrins' activity may have promising therapeutic implications for ROS-related pathologies.

Supporting Online Material

www.sciencemag.org/cgi/content/full/304/5670/596/DC1

Materials and Methods

Figs. S1 to S8

References and Notes

References and Notes

View Abstract

Navigate This Article