Hydrogen sulfide modulates eukaryotic translation initiation factor 2α (eIF2α) phosphorylation status in the integrated stress-response pathway

Hydrogen sulfide (H2S) regulates various physiological processes, including neuronal activity, vascular tone, inflammation, and energy metabolism. Moreover, H2S elicits cytoprotective effects against stressors in various cellular models of injury. However, the mechanism of the signaling pathways mediating the cytoprotective functions of H2S is not well understood. We previously uncovered a heme-dependent metabolic switch for transient induction of H2S production in the trans-sulfuration pathway. Here, we demonstrate that increased endogenous H2S production or its exogenous administration modulates major components of the integrated stress response promoting a metabolic state primed for stress response. We show that H2S transiently increases phosphorylation of eukaryotic translation initiation factor 2 (eIF2α) resulting in inhibition of general protein synthesis. The H2S-induced increase in eIF2α phosphorylation was mediated at least in part by inhibition of protein phosphatase-1 (PP1c) via persulfidation at Cys-127. Overexpression of a PP1c cysteine mutant (C127S-PP1c) abrogated the H2S effect on eIF2α phosphorylation. Our data support a model in which H2S exerts its cytoprotective effect on ISR signaling by inducing a transient adaptive reprogramming of global mRNA translation. Although a transient increase in endogenous H2S production provides cytoprotection, its chronic increase such as in cystathionine β-synthase deficiency may pose a problem.

Cells respond to endogenously produced and external stressors that perturb cellular homeostasis by activating stress-response pathways to adapt to stressful conditions and to minimize damage to cellular components. Recent findings have uncovered a significant regulatory role for H 2 S signaling on this front and fueled a growing pharmacological interest in H 2 S for treatment of cardiovascular diseases and inflammation where stress-induced cell injury contributes significantly to disease progression (1,2). H 2 S is a signaling mole-cule produced endogenously from sulfur-containing amino acids, cysteine and homocysteine, by the actions of two enzymes in the trans-sulfuration pathway, cystathionine ␤-synthase (CBS) 2 and cystathionine ␥-lyase (CSE) (3), and from 3-mercaptopyruvate catalyzed by mercaptopyruvate sulfur transferase (4,5). Metabolic removal of H 2 S involves its oxidation in mitochondria to thiosulfate and sulfate (6), which are excreted in urine (7). H 2 S is a weak acid and ionizes to HS Ϫ and H ϩ with a pK a of 6.9 (8) resulting in an estimated 80% of it being in the ionized form at physiological pH. Although HS Ϫ is likely confined to cells in which it is produced, H 2 S gas can freely diffuse across membranes (9) to initiate paracrine signaling, i.e. at sites remote from its production site. H 2 S-based signaling is mediated by formation of persulfides at reactive cysteine residues on target proteins to change activity (10). Persulfides can form via reaction of HS Ϫ with oxidized cysteine on proteins such as cysteine sulfenic acid or by the reaction of cysteine thiolates on oxidized sulfide species such as HSSH and polysulfides (11).
The key biochemical step for ISR is the induction of eIF2␣ phosphorylation, an evolutionarily conserved cytoprotective response, which has broad cellular consequences including translational and transcriptional reprogramming (41,42). Phosphorylation of eIF2␣ at Ser-51 is catalyzed by four kinases, GCN2, PERK, HRI, and PKR, each responding to different stresses (42). Phosphorylation of eIF2␣ blocks global mRNA translation, whereas translation of select cytoprotective proteins, including ATF4, continues via regulatory uORFs in their 5Ј-UTRs (42). ATF4 activates expression of stress-response genes and promotes proteostasis via a feedback loop that involves induction of GADD34, a regulatory subunit of protein phosphatase-1 (PP1c), which dephosphorylates eIF2␣-P. Low basal level of eIF2␣-P in unstressed cells is maintained by the action of PP1c in complex with the constitutively expressed regulatory subunit CReP (43)(44)(45).
In this study, we tested the hypothesis that H 2 S regulates the ISR signaling pathway. Herein, we describe the cellular response to H 2 S and show that exogenous H 2 S or induction of its endogenous synthesis leads to increased eIF2␣ phosphorylation. H 2 S leads to increased eIF2␣-P levels by inhibiting PP1c phosphatase via persulfidation, which in turn leads to transient suppression of global translation and activation of ATF4 expression.

H 2 S induces phosphorylation of eIF2␣
To test whether H 2 S modulates eIF2␣ phosphorylation, we treated mouse embryonic fibroblast (MEF) cells and HeLa cells with 100 M NaHS for 2 h. NaHS treatment resulted in an ϳ2.5-fold increase in eIF2␣ phosphorylation in both cell types, whereas the total eIF2␣ levels did not change (Fig. 1, a and b). The increase in eIF2␣ phosphorylation was evident as early as 1 h after H 2 S exposure, and it decayed to baseline levels after 8 -12 h (Fig. 1, c and d). A 25 M concentration of NaHS was sufficient to increase eIF2␣-P levels, and no further increase was seen at concentrations up to 200 M (Fig. 1, e and f). In comparison, the ER stress-inducing agent, tunicamycin (Tn) resulted in a more robust increase in eIF2␣ phosphorylation (Fig. 1a). However, repeated exposure to H 2 S (100 M NaHS added every 4 h) resulted in a gradual increase in eIF2␣ phosphorylation with no change in eIF2␣ level (Fig, 1g).
Next, we tested whether induction of endogenous H 2 S production elicits similar effects on eIF2␣ phosphorylation levels. We have recently described a regulatory switch whereby inhibition of CBS by CO increases H 2 S production by CSE (46). We exploited this regulatory strategy by overexpressing heme oxygenase-2 (HO-2), a CO-producing enzyme, in HEK293 cells. Transient overexpression of HO-2 increased endogenous H 2 S levels, as detected in live cells using the fluorescent H 2 S probe, 7-azido-4-methylcoumarin that was sensitive to propargylglycine, an inhibitor of CSE (Fig. 2a), and resulted in a 4-fold increase in basal eIF2␣-P level compared with cells transfected with an empty plasmid (p ϭ 0.02) (Fig. 2b). We obtained similar results by overexpressing nitric-oxide synthase (NOS), a source of NO, which also inhibits CBS activity (Fig. 2d) (47). Although the increase in eIF2␣-P levels in response to a single dose of exogenous H 2 S was transient, induction of endogenous H 2 S production by HO-2 overexpression resulted in a persistent increase in eIF2␣-P (compare Fig. 2, b and c, with Fig. 1, a and  b). This result reveals that both induction of endogenous H 2 S synthesis and exogenous H 2 S addition are associated with increased eIF2␣-P levels. To further test whether the increase in eIF2␣-P level by overexpression of CO-or NO-producing enzymes is mediated by CBS inhibition, we analyzed eIF2␣-P levels in liver tissue homogenates prepared from CBS Ϫ/Ϫ mice, as described previously (48). An ϳ2-fold higher level of eIF2␣-P levels was consistently seen in CBS Ϫ/Ϫ liver compared with wild-type control (Fig. 2, e and f). Cells were washed twice with cold PBS on ice and lysed in RIPA buffer supplemented with complete protease inhibitor mixture, and 50 g of total protein was loaded per lane. b, signal intensity was quantified for eIF2␣-P and eIF2␣ from replicate Western blotting experiments and expressed as eIF2␣-P/eIF2␣ ratio. c, time-dependent effect of H 2 S on eIF2␣-P levels after a single dose of 100 M NaHS treatment. Cells grown in the absence of NaHS were processed for each time point as controls. d, quantification of eIF2␣-P signal from replicate Western blotting experiments. e, changes in eIF2␣-P levels in response to increasing concentrations of H 2 S. Cells were treated with the indicated concentrations of NaHS for 1 h prior to sample preparation. f, quantification of eIF2␣-P signal from replicate Western blotting experiments. g, increase in eIF2␣-P level in response to repeated exposure to NaHS. Cells were supplemented with H 2 S every 4 h and were harvested at 12 h after the first H 2 S addition. Error bars represent Ϯ S.D., n ϭ 3. Signal for eIF2␣, which represent total eIF2␣, serves as an equal loading control. Error bars represent Ϯ S.D., n ϭ 3.

Inhibition of global protein synthesis by H 2 S
We tested whether H 2 S-induced eIF2␣ phosphorylation leads to inhibition of protein translation. For this, we monitored incorporation of [ 35 S]methionine into the protein pool in MEF cells (Ϯ100 M NaHS treatment) and in HEK293 cells transiently overexpressing HO-2 in the absence of exogenous H 2 S. We observed a significant decrease in translation in H 2 Streated cells (p ϭ 0.007) and in HO-2 overexpressing HEK293 cells (p ϭ 0.005) compared with controls ( Fig. 3, a and b). Because exogenous H 2 S increased eIF2␣-P levels transiently (Fig. 1d), we determined whether the kinetics of translational suppression was correlated with this behavior. For this, we monitored the time-dependent changes in protein translation in MEF cells exposed to a single dose of H 2 S (100 M NaHS). H 2 S-induced inhibition of protein synthesis was exerted over 4 h and returned to baseline levels over 8 -12 h mirroring the pattern of H 2 S-induced increase in eIF2␣-P levels (Figs. 3c and 1c). This result is consistent with the involvement of H 2 S-induced eIF2␣ phosphorylation in translational suppression.
Next, we tested whether H 2 S induced ATF4 expression, which is associated with increased eIF2␣-P levels and inhibition of global translation. ATF4 was increased in MEF cells after a single dose of H 2 S treatment (Fig. 4a), and in cells stably overexpressing HO-2, compared with control cells (Fig. 4b) confirming induction of ATF4 expression by H 2 S.

Inactivation of protein phosphatase-1 by persulfidation
The transient increase in eIF2␣ phosphorylation levels in response to H 2 S treatment can result from activation of one of the four upstream kinases and/or by inhibition of the phosphatase, PP1c. We hypothesized that the increase in eIF2␣-P levels by H 2 S results from inhibition of the basal activity of PP1c for the following reason. H 2 S-induced increase in eIF2␣-P levels was lower compared with the effect of ER stress-inducing m. b, Western blot analysis for eIF2␣-P in HEK293 cells overexpressing HO-2 and in control cells transfected with an empty plasmid in the presence and absence of the ER stress-inducing agent, thapsigargin. c, signals for eIF2␣-P and eIF2␣ were quantified from replicate Western blotting experiments. d, Western blot analysis for eIF2␣-P in HEK293 cells overexpressing (OE) inducible nitric-oxide synthase (iNOS). e, Western blot analysis for eIF2␣-P in liver tissue homogenates from wild-type and CBS Ϫ/Ϫ mice. A total 50 g of protein was loaded in each line. Extracts from HeLa cells Ϯ thapsigargin, 0.4 M, were used as control. f, signals for eIF2␣-P and eIF2␣ were quantified from three replicate experiments and expressed as eIF2␣-P/eIF2␣ ratio. agents ( Fig. 1a) and was independent of H 2 S concentration between 25 and 200 M NaHS (Fig. 1e). Additional increase in eIF2␣-P levels required either repeated exposure to H 2 S (Fig.  1g) or sustained H 2 S overproduction (Fig. 2, b and c). These results suggested to us that the increase in eIF2␣-P levels upon H 2 S exposure is limited by its rate of basal phosphorylation.
To test our hypothesis, we expressed and purified recombinant human PP1c to ϳ95% purity and analyzed the effect of H 2 S on dephosphorylation of eIF2␣-P in extracts prepared from cells exposed to ER stress. Addition of PP1c to extracts reduced eIF2␣-P (p ϭ 0.007), whereas NaHS-treated PP1c had no effect (Fig. 5, a and b). PP1c contains 13 cysteines, including several reactive ones, Cys-127, Cys-273, and Cys-291 (49). We hypothesized that the observed decrease in PP1c activity in the presence of H 2 S was due to persulfidation, which was characterized by mass spectroscopic analysis. A single cysteine, corresponding to Cys-127, was identified as being persulfidated (Fig.  6). To test whether Cys-127 mediates the effect of H 2 S on PP1c activity, we substituted Cys-127 with serine. Whereas PP1c-C127S efficiently dephosphorylated eIF2␣ in cell extracts (p ϭ 0.01), it was unresponsive to H 2 S treatment (Fig. 5, c and d) consistent with the importance of Cys-127 in mediating the H 2 S effect. To further validate these results, we overexpressed wild-type and C127S-PP1c in HEK293 cells. Overexpression of wild-type and mutant PP1c significantly reduced eIf2␣-P levels (Fig. 5, e and f) as expected (43,44). Treatment of these cells with 100 M H 2 S increased the eIF2␣-P level in cells overexpressing wild-type PP1c but not in cells overexpressing C127S-PP1c (Fig. 5e) confirming that Cys-127 is required to mediate H 2 S inhibition of PP1c activity. Interestingly, H 2 S had no effect on dephosphorylation when p-nitrophenyl phosphate or a phosphopeptide corresponding to residues 45-56 of eIF2␣ (ILLSEL(pS)RRRIR), was used as substrates (data not shown). This difference in H 2 S effects on PP1c presumably results from differences in its interaction between the phosphopeptide and full-length protein substrates as also demonstrated for eIF2␣ phosphorylation, which requires at least 80 amino acids from the N terminus (50).

Involvement of PERK kinase in H 2 S-induced phosphorylation of eIF2␣
H 2 S reportedly inhibits PTP1B phosphatase during ER stress (51). PTP1B dephosphorylates the PERK kinase, an ER stress sensor that autophosphorylates and induces the PERK branch of the ISR during ER stress by phosphorylating eIF2␣. We tested whether the H 2 S-mediated increase in eIF2␣-P levels was contributed by increased phosphorylation of PERK due to inhibition of PTP1B by H 2 S. For this, we analyzed the effect of H 2 S on eIF2␣-P levels in Perk Ϫ/Ϫ and Perk ϩ/ϩ MEF cells. H 2 S treatment for 1 h resulted in an ϳ2-fold increase in eIF2␣-P levels in both wild-type (p ϭ 0.02) and Perk Ϫ/Ϫ MEF cells (p ϭ 0.009) (Fig. 7, a and b). These results were mirrored by changes in translation rates measured in the presence and absence of H 2 S in Perk ϩ/ϩ and Perk Ϫ/Ϫ MEF cells (Fig. 7c). These results indicate that H 2 Sinduced increase in eIF2␣-P level is independent of PERK activation.

Effect of HO-2 overexpression on stress resistance
We determined the viability of three cell lines exposed to H 2 S for 2 h prior to induction of ER stress with thapsigargin. H 2 S protected cells from ER stress-induced cell death in all cell lines (Fig. 8a) consistent with previous reports (34 -37). H 2 S treatment alone in the absence of stress had no effect on cell viability (Fig. 8b). We tested the effect of sustained high H 2 S levels in cells stably overexpressing HO-2. Although these cells are more resistant to transient stress induced by thapsigargin (Fig. 8a), they are slow growing and die off after 4 -6 passages.
The transient pre-emptive induction of eIF2␣ phosphorylation is known to be cytoprotective for further stress (52). Therefore, we examined eIF2␣-P levels in response to ER stress induced with Tg in cells with prior H 2 S exposure. The increase in eIF2␣-P levels upon Tg treatment was significantly lower in H 2 S-pretreated cells compared with Tg alone treatment (Fig. 8,  c and d) suggesting that H 2 S increases cellular threshold to stress. This is consistent with the report that the increase in eIF2␣-P levels induced by a mild transient stress in neuroblastoma cells prevents further increase in eIF2␣-P levels in response to a subsequent acute stress (53). Our results suggest that the cytoprotective function of H 2 S is in part mediated by inhibition of PP1c and transient modulation of the ISR components, eIF2␣-P phosphorylation and global protein synthesis (52).

Discussion
In this study, we have characterized an H 2 S signaling pathway using a cellular model system where endogenous H 2 S production was induced by overexpressing HO-2. Our results indicate that H 2 S is a physiological modulator of eIF2␣ phosphorylation status and that it exerts its effect via a mechanism involving persulfidation and concomitant inhibition of PP1c. Phosphorylation of eIF2␣ is typically induced under stress conditions by activation of upstream kinases to guard against dysregulation of cellular homeostasis. However, the existence of signaling pathways to transiently induce eIF2␣ phosphorylation in the absence of overt stress is largely unexplored. Herein, we show that H 2 S-induced inhibition of PP1c provides an alternative route to modulate eIF2␣-P levels independent of upstream kinases. We propose that although a transient increase in H 2 S production induces an acute response, which is consistent with its cytoprotective effects, continuous exposure results in a persistent increase in eIF2␣-P levels but is tolerated in cells due to the presence of an efficient H 2 S oxidation pathway present in mitochondria. Consistent with this model, disruption of the first sulfide oxidation pathway enzyme in Caenorhabditis elegans leads to death upon H 2 S exposure resulting from both ER and mitochondrial stress (54).
A cytoprotective effect for H 2 S-induced inhibition of PP1c leading to a transient increase in basal eIF2␣-P levels is consistent with other reports. For instance, inhibition of PP1c activity by knocking down CReP activates the ISR and is cytoprotective against stressors, including oxidative and ER stress (43). Similarly, inhibition of PP1c interaction with the regulatory subunits by salubrinal (55) or by mutagenesis (44) and by GADD34 knock-out (56) increases eIF2␣-P levels, inhibits translation, Figure 5. H 2 S induced inhibition of PP1c activity. a, H 2 S inhibits eIF2␣ dephosphorylation by PP1c. PP1c activity to dephosphorylate eIF2␣ was determined in reactions containing extracts (50 -100 g of total protein) from Tg-treated HEK293 cells with or without NaHS pretreatment. Reactions were separated on a 10% SDS gel, and eIF2␣-P and eIF2␣ levels were monitored by Western blot analysis. b, quantified eIF2␣-P and eIF2␣ signals presented as the ratio of eIF2␣-P to eIF2␣. c, loss of the H 2 S effect on the activity of PP1c-C127S. d, signals for eIF2␣-P and eIF2␣ in Western blots from replica assays were quantified. e, H 2 S effect on eIF2␣-P levels in cells overexpressing WT-PP1c and C127S-PP1c compared with control cells transfected with an empty plasmid. Thirty to 48 h after transfection, cells were treated with 200 M H 2 S and continued to grow for 2 h prior to sample preparation. f, quantified eIF2␣-P signals from samples prepared from HEK293 cells expressing wild-type or C127S mutant PP1c in the presence or absence of NaHS. EP donates empty plasmid. TR denotes transfection. Error bars represent S.D. Asterisks represent p Ͻ 0.05.

Regulation of integrated stress-response pathway by H 2 S
and in some systems has a protective role against stress (43,55,56). Our results might be relevant for understanding the protective effect of H 2 S on protein metabolism in response to hypoxia-induced stress as discussed below (57).
PP1c is rich in cysteine residues and contains several reactive ones. Cys-273 is critical for activity, and microcystin, an inhibitor of PP1c, functions by binding covalently to the sulfur in Cys-273 (49,58). In the crystal structure of PP1c, Cys-127 is oxidized to sulfinic or sulfonic acid, and Cys-291 forms a mixed disulfide with mercaptoethanol (49). Our results validate that Cys-127 in PP1c is a target of persulfidation, which was initially picked up in a persulfide proteomic analysis under ER stress conditions that is accompanied by increased H 2 S synthesis (39). We predict that PP1c activity is sensitive to cysteine modifications, and its inhibition by persulfidation results in increased eIF2␣-P levels and to modulation of global translation.
PP1c associates with a variety of regulatory subunits that dictate target specificity (59). Although the dependence of the H 2 S effect on the identity of the regulatory domains remains to be demonstrated, our model is consistent with the reported effect of H 2 S on increasing phosphorylation levels of AMP-activated protein kinase (60), which is dephosphorylated by PP1c in complex with the regulatory subunit R6 (61). AMP-activated protein kinase functions as an energy sensor, and its phosphorylation under hypoxia activates the AMPK/TSC2/Rheb/mTOR signaling pathway, which inhibits mTOR activity. These changes lead to suppression of the initiation and elongation phases of translation (62,63). Although it is not understood how H 2 S treatment modulates this signaling network during hypoxia, we speculate that H 2 S-induced inhibition of PPIc activity might increase AMPK phosphorylation for inhibition of eIF2B activity.
In summary, we have shown that transient exposure of cells to H 2 S leads to increased eIF2␣ phosphorylation by PP1c persulfidation at Cys-127, which leads to its inhibition. This study reveals a previously unknown mode of regulation for the eIF2␣-P level that may underlie the cytoprotective effects of H 2 S. The ISR/ATF4 program mediates metabolic reprogramming of cells exposed to ER stress via H 2 S-mediated protein persulfidation (39). The current findings suggest that H 2 S might also contribute to the outcome of ISR in part by modulating translational recovery required for transcriptional reprogramming and adaptation (64,65). Translational recovery depends on the phosphorylation status of eIF2␣ and is critical in most chronic stress conditions as uncontrolled translational recovery decreases survival of stressed cells (39,66,67). Our data suggest that inhibition of PP1c by H 2 S can potentially dampen translational recovery and be important in delaying the onset of diseases involving chronic stress.

Cell culture
ATF ϩ/ϩ and ATF4 Ϫ/Ϫ MEFs were obtained from Dr. Ronald Wek (Indiana University School of Medicine) and were described previously (41,68). HEK293, RWPE (prostate cells), and HeLa cells were grown in DMEM. LNCaP cells (prostate cancer cells) were grown in RPMI 1640 media. Perk ϩ/ϩ MEF, and Perk Ϫ/Ϫ MEF cells were grown in DMEM. ATF4 Ϫ/Ϫ MEF cells were grown in DMEM supplemented with 1% non-essential amino acid solution (Life Technologies, Inc.) and 55 M mercaptoethanol (Life Technologies, Inc.). All media were supplemented with 10% FBS and penicillin/streptomycin (100 units/ml and 100 g/ml, respectively). Transient transfections were performed using the Xtreme Gene HP transfection reagent (Roche Applied Science) according to the manufacturer's protocol. HeLa cells stably expressing HO-2 were obtained by selection with geneticin, 0.6 g/ml. ER stress was induced with thapsigargin or tunicamycin in the medium at a final concentration from 0.5 to 1 M and 5 g/ml, respectively, as specified in the figure legends.

Metabolic labeling and determination of protein translation rate
Cells were grown in either 6-cm plates or in 12-well plates to a confluency of 80%. Then they were labeled with 15 or 30 Ci/ml [ 35 S]methionine (PerkinElmer Life Sciences) and con-tinued to grow for 1-2 h depending on the experiment. Fresh medium was added to plates 1 h before radiolabel addition. At the end of the labeling time, cells were washed twice with PBS and scraped off the plates in RIPA buffer. After two freeze/thaw cycles, 10% (v/v) of trichloroacetic acid (TCA) was added to precipitate proteins. Aliquots from extracts were used to measure protein concentration. The protein precipitate was washed twice with 5% TCA and dissolved in 200 l of 1 M NaOH, and the radioactivity was counted in a liquid scintillation counter. Radioactivity was normalized to protein concentration measured using Bradford reagent (Bio-Rad) with bovine serum albumin as the standard. For radioactive gels, extracts were denatured in SDS dye loading buffer and boiled for 5 min before electrophoresis. Equal amounts of protein were loaded in each well. Radioactive gels were dried in a gel dryer attached to a vacuum pump, placed on a phosphor storage screen cassette for 24 h, and imaged on a STORM 860 phosphorimager. Autoradiograms were quantified using the software ImageJ.

Western blot analysis
Cells were washed three times with PBS on ice and scraped in 50 mM Tris, pH 7.4, containing 0.1% Triton X-100, complete protease inhibitor mixture (Sigma) or in RIPA buffer supplemented with protease inhibitor mixture. Extracts were incubated on ice for 30 min before centrifuging at 13,000 ϫ g for 10 min at 4°C. Aliquots of supernatants were added to SDS denaturing buffer and boiled for 5-10 min. Equal amounts of proteins were separated on 10 -12% SDS-polyacrylamide gel and then transferred to a PVDF membrane. Membranes were blocked with 5% non-fat dry milk in TTBS (Tris-buffered saline containing 0.1% Tween 20) for 1 h at room temperature with shaking and washed four times with TTBS before overnight incubation with the primary antibody at 4°C. Membranes were washed for 4 -5 times, 20 min each, with TTBS and then incubated with the secondary antibody conjugated to horseradish peroxidase for 1 h at room temperature. Signal intensities for specific proteins were estimated using ImageJ (National Institutes of Health) software. Antibody dilutions were as follows: anti-eIF2␣, 1:1000; anti-eIF2␣-P, 1:1000; anti-HO-2, 1:1000; anti-PP1c, 1:1000; and anti-ATF4, 1:500.

Western-based eIF2␣ dephosphorylation assay
Cells were treated with thapsigargin for Ͼ24 h to induce eIF2␣ phosphorylation. Dephosphorylation of eIF2␣ was performed in a 50-l reaction mixture containing 50 mM Tris, pH 7.4, 100 mM KCl, 4 mM MgCl 2 , and 50 -100 g of protein from cell extracts prepared as described above from Tg-treated cells but without protease inhibitors. Reactions were started by adding 5-10 g of purified wild-type or mutant PP1c with or without H 2 S pretreatment (250 M final) and incubated at room temperature for 10 min. At the end of the incubation time, SDS denaturing buffer was added to the reaction mixtures and boiled for 5 min at 95°C. Proteins were separated on a 10% SDS gel and transferred to PVDF membrane for Western blot analysis as described above to detect eIF2␣ and eIF2␣-P levels. Signal intensities for eIF2␣ were quantified from three independent experiments.

Regulation of integrated stress-response pathway by H 2 S
Fluorescence microscopy H 2 S was visualized in HEK293 cells transfected with mammalian expression construct for HO-2 or with an empty plasmid using 7-azido-4-methylcoumarin as described previously (69). Briefly, 30 -36 h post-transfection, 7-azido-4-methylcoumarin was added to the culture medium to a final concentration of 50 M, and incubation was continued for 30 min. Then, the cells were washed three times with PBS and visualized using an IX70 inverted microscope, connected with Photometrics Cool-SNAP HQ2 camera, using 720-nm laser excitation. Metamorph software was used to acquire and analyze images. Propargylglycine, an inhibitor of the H 2 S-producing enzyme, CSE, was added 6 h prior to imaging at a 2 mM final concentration.

Cloning, expression, and purification of PP1c
The mammalian expression vector containing PP1c, pEZ-M01, was purchased from GeneCopoeia (Rockville, MD). The PP1c-coding region was PCR-amplified using the forward (5Ј-GGAAGGAGTTCGACATATGGCGGATTTAGATA-AACTCAACATCG) and reverse (5Ј-GCGGCCGCACTC-GAGCTATTTCTTTGCTTGCTTTGTGATCATAC) primers containing the XhoI and NdeI restriction sites, which were used for subcloning into the pET28b bacterial expression vector to generate the expression construct pET28b-PP1c. E. coli BL21 (DE3) cells transformed with the pET28b-PP1c construct were grown overnight at 37°C in 100 ml of LB media containing kanamycin (50 g/ml) and used to inoculate 6 liters of LB media. Cells were grown at 25°C. Expression of PP1c was induced with isopropyl ␤-D-thiogalactopyranoside (25 mg/li-ter) when the A 600 reached 0.5. Cells were harvested after 16 h, and fresh or frozen cell pellets were suspended in 200 ml of lysis buffer containing 20 mM sodium phosphate, pH 7.4, 150 mg of lysozyme, 10 mM MgCl 2 , 500 mM NaCl, 20 mg of DNase, and 10% glycerol. The cell suspension was stirred at 4°C for 30 min and then sonicated at a power setting of 7 for 10 min in 30-s intervals separated by 1 min of cooling. The sonicate was centrifuged at 17,000 ϫ g for 30 min to obtain the soluble fraction. The N-terminal His-tagged PP1c was affinity-purified using a nickel-nitrilotriacetic acid column in 20 mM sodium phosphate buffer, pH 7.4, 500 mM NaCl, 10% glycerol. Further purification was achieved by gel filtration chromatography using a prepacked Hiload 16/60 Superdex 200 column (GE Healthcare) using the same buffer.

Mass spectrometry
Recombinant PP1␥ (20 g) was treated with NaHS (1 mM) at room temperature for 30 min and then incubated with the thiol-blocking buffer containing 20 mM NEM for 30 min. NEMalkylated PP1␥ was subjected to a non-reducing SDS gel and stained with Coomassie Blue. The PP1␥ protein band was excised and digested with trypsin and chymotrypsin. The peptide samples were analyzed by capillary column LC-tandem MS and the CID spectra searched against the human reference sequence database using the program Mascot and more specifically against the protein sequences with the program Sequest. All analyses utilized the standard LC gradient from 2 to 70% acetonitrile in 110 min. Cells were grown to a 60 -70% confluency, and NaHS was added to a final concentration of 200 M. Cell viability was determined after 20 -24 h using trypan blue to stain dead cells, which were counted using a microscope (with HeLa and RWPE cells) or analyzed by flow cytometry using propidium iodide to stain dead cells (for MEF cells). c, Western blot analysis for eIF2␣-P levels in cells with ER stress induced by thapsigargin, 0.4 M, with and without NaHS pretreatment. d, quantification of signals from three independent experiments. Error bars represent S.D., n ϭ 3-4.

Cell viability analysis
Because H 2 S can impact metabolic rate and energy production, we used the trypan blue assay for staining and counting live and dead cells under the microscope using a hemocytometer. Flow cytometry was used for MEF to determine the effect of H 2 S on cell viability using propidium iodide (Calbiochem) to stain dead cells. Cells were harvested by trypsinization and washed twice with PBS. To the cell suspensions, 1% BSA (w/v) and propidium iodide (ϳ1 M final) were added for flow cytometry or, alternatively, mixed with trypan blue for counting under the microscope. The data are presented as percent dead or live cells.

Statistical analysis
The statistical significance of observed differences was evaluated using paired t test.
Author contributions-V. Y. performed the majority of the experiments and analyzed the data. X.-H. G. contributed his expertise in preparing samples for detection of PP1c persulfidation. B. W. performed the mass spectroscopy analysis with LC-MS/MS. M. H. contributed to designing the study related to ISR, data analysis, and manuscript preparation. R. B. contributed to data analysis and helped in manuscript preparation. O. K. conceived and designed the study, analyzed and interpreted the data, performed experiments, and prepared the manuscript. All authors approved the final version of the manuscript.