Cysteine Sulfenylation Directs IRE-1 to Activate the SKN-1/Nrf2 Antioxidant Response

Abstract

Emerging evidence suggests that many proteins may be regulated through cysteine modification, but the extent and functions of this signaling remain largely unclear. The endoplasmic reticulum (ER) transmembrane protein IRE-1 maintains ER homeostasis by initiating the unfolded protein response (UPR(ER)). Here we show in C. elegans and human cells that IRE-1 has a distinct redox-regulated function in cytoplasmic homeostasis. Reactive oxygen species (ROS) that are generated at the ER or by mitochondria sulfenylate a cysteine within the IRE-1 kinase activation loop. This inhibits the IRE-1-mediated UPR(ER) and initiates the p38/SKN-1(Nrf2) antioxidant response, thereby increasing stress resistance and lifespan. Many AGC-family kinases (AKT, p70S6K, PKC, ROCK1) seem to be regulated similarly. The data reveal that IRE-1 has an ancient function as a cytoplasmic sentinel that activates p38 and SKN-1(Nrf2) and indicate that cysteine modifications induced by ROS signals can direct proteins to adopt unexpected functions and may coordinate many cellular processes.

Figure 1. Distinct IRE-1-dependent responses to ROS and ER stress

(A–B) IRE-1 is required for the C. elegans p38/SKN-1 response to ROS. ire-1 RNAi diminished the p38 (A) and gst-4P::GFP (B) responses to sublethal exposure to AS (30 min) or PQ (30 min), but not Sorbitol (ST, 30 min) or heat (30 min). Phosphorylation of p38 (P-p38) indicates p38 activation (Inoue et al., 2005). (C–E) Acute AS treatment does not activate the UPR^ER. AS (30 min) did not upregulate hsp-4P::GFP (C), with intestinal GFP scoring in (D), or promote IRE-1 and HSP-3/4 dissociation as determined by co-immunoprecipitation (co-IP) (E). (F–I) AS pre-treatment blocks IRE-1 and UPR^ER activation in response to ER stress. Prior AS exposure (30 min) inhibited IRE-1 phosphorylation (F), xbp-1 mRNA splicing, as indicated by quantitiative PCR assay of levels of the indicated mRNA forms (G), XBP-1S protein accumulation (H), and hsp-4P::GFP induction (I–J) by ER stress (TM; 5 hr). AS-treated samples were allowed to recover for 30 min. prior to TM exposure. (K) AS does not inhibit the PERK kinase PEK-1, as indicated by phosphorylation of its substrate eIF2-α. In the right panels PEK-1 expression was blocked by RNAi. (L–O) Pre-exposure to ER stress inhibits the antioxidant response. Exposure to TM (5 hr) prior to AS treatment (30 min.) decreased p38 activation (L), SKN-1 nuclear localization (M) and intestinal gst-4P::GFP expression (N–O). In all figures, mock incubations and zero time points are indicated by “C”. (B, D, J, M, O) GFP quantification with high (H), medium (M) or low (L) scoring. p < 0.0001****; p < 0.001***; p < 0.01**. All immunoblots in the paper are representative of at least two and in most cases three experiments. See also Figure S1.

Figure 2. IRE-1 sulfenylation within the kinase activation loop

(A) Capture of sulfenylated (SOH) proteins with DCP-BIO1. (B) ROS induce sulfenylation of IRE-1. Lysates from animals treated with AS (30 min) in the absence or presence of GSH were incubated with DCP-BIO1, with sulfenylated IRE-1 (IRE-1:SOH) identified by immunoblot. (C–D) sulfenylation of IRE-1 parallels p38 activation. Animals treated with AS for 0–120 min were assessed for IRE-1:SOH (C) and active p38 (D) by immunoblot. (E) ER stress interferes with IRE-1 sulfenylation. Animals pretreated with TM (5 hr) show reduced IRE-1:SOH in response to AS (30 min). (F–G) Conservation of C663 within the IRE-1 kinase domain. In (F), adjacent basic amino acids are labeled in bold. A model of the IRE-1 kinase domain based upon the human Ire1α crystal structure (PDB ID: 3P23) shows C663 protruding into the ATP-binding pocket (G). (H) C663 is the site of IRE-1 sulfenylation. IRE-1:SOH was not detected in IRE-1^C663S animals after AS treatment (30 min). In (H–J), results are representative of two independent transgenic lines. (I) C663 mutation does not interfere with ER stress-induced XBP-1S production. XBP-1S protein levels were determined in ire-1 (ok799) animals and animals expressing either IRE-1^WT or IRE-1^C663S following TM (5 hr) treatment. (J) C663 is required for ROS-induced p38 activation. IRE-1^C663S animals exhibit diminished AS-induced p38 activity comparable to ire-1(ok799) null animals. (K) IRE-1^C663S animals exhibit enhanced oxidative stress sensitivity compared to ire-1(ok799) that was rescued with WT IRE-1. Day-1 adult worms were exposed to 5mM AS (p < 0.01**). (L) IRE-1 sulfenylation is not compatible with IRE-1 phosphorylation. The sulfenylated IRE-1 fraction, which was isolated by Biotin (Dimedone) pull-down, does not contain any detectable phosphorylated IRE-1. Short exposure (S.E); long exposure (L.E). See also Figure S2 and Table S1.

Figure 3. A local ROS signal initiates the IRE-1/p38/SKN-1 response

(A) AS promotes IRE-1 and NSY-1 complex formation. IRE-1 was IP’d from C. elegans that had been treated with vehicle or AS (30 min), then assayed by immunoblot for NSY-1. Non-specific IgG acted as control. (B) trf-1 promotes AS-induced IRE-1 and NSY-1 interaction. Lysates of animals fed control (EV) or trf-1 RNAi and treated with AS (30 min) were assayed by immunoblot for IRE-1/NSY-1 interaction. (C–E) trf-1 is required for the p38/SKN-1 response to AS. Knockdown of trf-1 diminished the p38 (C) and gst-4P::GFP (D) responses to AS (30 min), with GFP scoring shown in (E). (F) Reduced TRF-1 levels sensitize to oxidative stress. Day 1 adults were scored for survival in 5mM AS. (G) Sulfenylation of NSY-1 but not IRE-1 requires trf-1. Knockdown of trf-1 blocked AS-induced (30 min) sulfenylation of NSY-1. (H) AS induces NSY-1 phosphorylation comparably in control (EV) and ire-1(RNAi) animals. (I) AS induces NSY-1 sulfenylation in an ire-1-dependent manner. Animals fed either control or ire-1 RNAi were assessed for NSY-1:SOH after AS (30 min) treatment. (J–K) bli-3 (DUOX) is required for the p38/SKN-1 response to AS. bli-3 RNAi reduced intestinal gst-4P::GFP activation by AS (30 min) (J) with intestinal GFP scoring in (K). (L) AS exposure (30 min) increases BLI-3 activity. NOX-dependent superoxide production was measured by chemiluminescence. (M) AS-induced sulfenylation of IRE-1 was diminished by bli-3 RNAi. (N) bli-3 is required for the p38 response to AS. bli-3 RNAi decreased AS-induced p38 activity. (O) Interaction of IRE-1 and BLI-3 is enhanced by AS, assessed by co-IP. (E, K) GFP scored as high (H), medium (M) or low (L). p < 0.0001****; p < 0.01**; p < 0.05*. See also Figure S3.

Figure 4. ER-associated ROS activate the IRE-1/p38/SKN-1 pathway to increase lifespan

(A) IRE-1 is sulfenylated in response to mitochondrial ROS production. IRE-1 sulfenylation (SOH) that is induced by PQ (30 min) was inhibited by pre-treatment of C. elegans with the mitochondrial ROS scavenger mito-tempo (MT) (1 hr). (B–D) Mitochondrial ROS activate the p38/SKN-1 response. Mito-tempo (MT) (1 hr) reduced the p38 response to PQ (30 min) (B) and diminished PQ-induced gst-4P::GFP expression (C), with GFP scoring shown in (D). GFP was quantified as high (H), medium (M) or low (L). (E–I) ero-1 RNAi induces ROS production (E), IRE-1:SOH (F), and p38 signaling (G) to extend lifespan (H, I). The MAPKK SEK-1 phosphorylates p38 in response to ROS and is required for p38 activity (Inoue et al., 2005). Lifespan extension from ero-1 RNAi depended upon ire-1- and sek-1 (H), and in a presumed null skn-1 mutant ero-1 RNAi shortened lifespan (I). Additional lifespan experiments and statistics are described in Table S2. p < 0.0001****; p < 0.05*. See also Figure S4 and Table S2.

Figure 5. Evolutionary conservation of the IRE-1 antioxidant response function

(A) Human Ire1α senses ROS. Ire1α sulfenylation was assessed in HepG2 cells following AS (30 min) treatment. (B) Ire1α is required for the p38 and Nrf2 response to AS-induced ROS. Levels of Nrf2 accumulation, p38 phosphorylation and activation were assayed in HepG2 cells treated with either control siRNA (si-SCR) or Ire1α siRNA (si-Ire1α) following AS (30 min) treatment. (C) NOX-derived ROS activate the p38/Nrf2 pathway. Levels of Nrf2 and p38 activation were determined in HepG2 cells treated with AS (30 min) in the absence or presence of the pan-NOX inhibitor VAS2870 (NOX-i)(1 hr). (D–E) Ire1α senses ROS and ER stress in a mutually exclusive manner. HepG2 cells were treated with AS (30 min) prior to TM (5 hr) treatment, with Xbp1S protein assayed by immunoblot (D). Levels of active p38 were determined in HepG2 cells following treatment with AS (30 min) alone, TM (5 hr) alone or pretreatment with TM prior to acute AS exposure (E). (F) Nrf2 activation by AS requires p38. HepG2 cells were treated with vehicle control or the p38 inhibitor SB203580 (p38-i) (2 hr) prior to AS (30 min). (G) Ire1α regulates Nrf2 independently of Keap1. HepG2 cells treated with either control siRNA (si-SCR) or Ire1α siRNA (si-Ire1α) were exposed to either the Keap1 inhibitor sulforaphane (SF) (30 min) or AS (30 min). (H) Activation of the p38/Nrf2 response by ER-derived ROS. Treatment of HepG2 cells with Ero1 inhibitor II (2 hr) activated p38 and Nrf2 comparably to AS. See also Figure S5.

Figure 6. Sulfenylation of AGC kinases by NOX-derived ROS

(A) Alignment of the DFG+2 Cys of selected human AGC kinases. Basic amino acids are underlined. (B) The C. elegans Rock1 ortholog (LET-502) and p70S6K (RSKS-1) are sulfenylated in vivo. Lysates from animals treated with AS (30 min) were assessed for levels of Rock1:SOH and RSKS-1:SOH. (C) NOX-derived ROS sulfenylate multiple AGC kinases. HepG2 cells were treated with AS (30 min) in the absence or presence of the NOX inhibitor (NOX-i) VAS2870 (1 hr).

Figure 7. A sulfenylation-mediated switch between IRE-1 functions

(A) Model of the UPR^ER and antioxidant response functions of IRE-1 (see text). (B) NOX-generated ROS sulfenylate multiple AGC kinases, resulting in IRE-1 adopting a distinct function.