TriPer, an optical probe tuned to the endoplasmic reticulum tracks H2O2 consumption by glutathione

The fate of H2O2 in the endoplasmic reticulum (ER) has been inferred indirectly from the activity of ER localized thiol oxidases and peroxiredoxins, in vitro, and the consequences of their genetic manipulation, in vivo. Here we report on the development of TriPer, a vital optical probe sensitive to changes in the concentration of H2O2 in the thiol-oxidizing environment of the ER. Consistent with the hypothesized contribution of oxidative protein folding to H2O2 production, ER-localized TriPer detected an increase in the luminal H2O2 signal upon induction of pro-insulin (a disulfide bonded protein of pancreatic β-cells), which was attenuated by the ectopic expression of catalase in the ER lumen. Interfering with glutathione production in the cytosol by buthionine sulfoximine (BSO) or enhancing its localized destruction by expression of the glutathione-degrading enzyme ChaC1 in lumen of the ER, enhanced further the luminal H2O2 signal and eroded β-cell viability. Tracking ER H2O2 in live cells points to an unanticipated role for glutathione in H2O2 turnover. Significance statement The presence of millimolar glutathione in the lumen of the endoplasmic reticulum has been difficult to understand purely in terms of modulation of protein-based disulphide bond formation in secreted proteins. Over the years hints have suggested that glutathione might have a role in reducing the heavy burden of hydrogen peroxide (H2O2) produced by the luminal enzymatic machinery for disulphide bond formation. However, limitations in existing in vivo H2O2 probes have rendered them all but useless in the thiol-oxidizing ER, precluding experimental follow-up of glutathione’s role ER H2O2 metabolism. Here we report on the development and mechanistic characterization of an optical probe, TriPer that circumvents the limitations of previous sensors by retaining specific responsiveness to H2O2 in thiol-oxidizing environments. Application of this tool to the ER of an insulin-producing pancreatic b-cells model system revealed that ER glutathione antagonizes locally-produced H2O2 resulting from the oxidative folding of pro-insulin. This study presents an interdisciplinary effort intersecting cell biology and chemistry: An original redox chemistry concept leading to development of a biological tool, broadly applicable for in vivo studies of H2O2 metabolism in the ER. More broadly, the concept developed here sets a precedent for applying a tri-cysteine relay system to discrimination between various oxidative reactants, in complex redox milieux.


Significance statement
The presence of millimolar glutathione in the lumen of the endoplasmic reticulum has been difficult to understand purely in terms of modulation of protein-based disulphide bond formation in secreted proteins. Over the years hints have suggested that glutathione might have a role in reducing the heavy burden of hydrogen peroxide (H2O2) produced by the luminal enzymatic machinery for disulphide bond formation. However, limitations in existing in vivo H2O2 probes have rendered them all but useless in the thiol-oxidizing ER, precluding experimental follow-up of glutathione's role ER H2O2 metabolism.
Here we report on the development and mechanistic characterization of an optical probe, TriPer that circumvents the limitations of previous sensors by retaining specific responsiveness to H2O2 in thiol-oxidizing environments. Application of this tool to the ER of an insulin-producing pancreatic b-cells model system revealed that ER glutathione antagonizes locally-produced H2O2 resulting from the oxidative folding of pro-insulin.
This study presents an interdisciplinary effort intersecting cell biology and chemistry: An original redox chemistry concept leading to development of a biological tool, broadly applicable for in vivo studies of H2O2 metabolism in the ER. More broadly, the concept developed here sets a precedent for applying a tri-cysteine relay system to discrimination between various oxidative reactants, in complex redox milieux.

Abstract:
The fate of H 2 O 2 in the endoplasmic reticulum (ER) has been inferred indirectly from the activity of ER localized thiol oxidases and peroxiredoxins, in vitro, and the consequences of their genetic manipulation, in vivo. Here we report on the development of

Introduction:
The thiol redox environment of cells is compartmentalized, with disulfide bond formation confined to the lumen of the endoplasmic reticulum (ER) and mitochondrial inter-membrane space in eukaryotes and the peri-plasmic space in bacteria and largely excluded from the reducing cytosol (Sevier and Kaiser, 2008). Together, the tripeptide glutathione and its proteinaceous counterpart, thioredoxin, contribute to a chemical environment that maintains most cytosolic thiols in their reduced state. The enzymatic machinery for glutathione synthesis, turnover and reduction is localized to the cytosol, as is the thioredoxin/thioredoxin reductase couple (Toledano et al., 2013). However, unlike the thioredoxin/thioredoxin reductase system that is largely isolated from the endoplasmic reticulum, several lines of evidence suggest equilibration of glutathione pools between the cytosol and ER.
Isolated microsomes contain millimolar concentrations of glutathione (Bass et al., 2004); an estimate buttressed by kinetic measurements (Montero et al., 2013). Yeast genetics reveals that the kinetic defect in ER disulfide bond formation wrought by lack of an important luminal thiol oxidase, ERO1, can be ameliorated by attenuated glutathione synthesis in the cytosol (Cuozzo and Kaiser, 1999), whereas deregulated import of glutathione across the plasma membrane into the cytosol compromises oxidative protein folding in the yeast ER (Kumar et al., 2011). Import of reduced glutathione into the isolated rat liver microsomal fraction has been observed (Banhegyi et al., 1999) and in a functional counterpart to these experiments, excessive reduced glutathione on the cytosolic side of the plant cell ER membrane compromised disulfide formation (Lombardi et al., 2012). In mammalian cells experimental mis-localisation of the reduced glutathione-degrading enzyme ChaC1 to the ER depleted total cellular pools of glutathione (Tsunoda et al., TriPer_glut_H2O2_2.4.2f 2014), arguing for transport of glutathione from its site of synthesis in the cytosol to the ER. Despite firm evidence for the existence of a pool of reduced glutathione in the ER, its functional role has remained obscure, as depleting ER glutathione in cultured fibroblasts affected neither disulfide bond formation nor their reductive re-shuffling (Tsunoda et al., 2014).
The ER is an important source of hydrogen peroxide production. This is partially explained by the activity of ERO1, which shuttles electrons from reduced thiols to molecular oxygen, converting the latter to hydrogen peroxide (Gross et al., 2006).
Alternative ERO1-independent mechanisms for luminal hydrogen peroxide production also exist , yet the fate of this locally generated hydrogen peroxide is not entirely clear. Some is utilized for disulfide bond formation, a process that relies on the ER-localized peroxiredoxin 4 (PRDX4) (Tavender and Bulleid, 2010;Zito et al., 2010) and possibly other enzymes that function as peroxiredoxins (Nguyen et al., 2011;Ramming and Appenzeller-Herzog, 2013). However, under conditions of hydrogen peroxide hyperproduction (experimentally induced by a deregulated mutation in ERO1), the peroxiredoxins that exploit the pool of reduced protein thiols in the ER lumen as electron donors, are unable to cope with the excess of hydrogen peroxide and cells expressing the hyperactive ERO1 are rendered hypersensitive to concomitant depletion of reduced glutathione (Hansen et al., 2012). Besides, ERO1 overexpression leads in increase of cell glutathione content (Molteni et al., 2004). These findings suggest a role for reduced glutathione in buffering excessive ER hydrogen peroxide production. Unfortunately, limitations in methods for measuring changes in the content of ER luminal hydrogen peroxide have frustrated efforts to pursue this hypothesis. Here we describe the development of an optical method to track changes in hydrogen peroxide levels in the ER lumen. Its application to the study of cells in which the levels of hydrogen peroxide and glutathione were selectively manipulated in the ER and cytosol revealed an important role for glutathione in buffering the consequences of excessive ER hydrogen peroxide production. This process appears especially important to insulin-producing β-cells that are encumbered by a heavy burden of ER hydrogen peroxide production and a deficiency of the peroxide degrading calatase. TriPer_glut_H2O2_2.4.2f

Results:
Glutathione depletion exposes the hypersensitivity of pancreatic β-cells to hydrogen peroxide Insulin producing pancreatic β-cells are relatively deficient in the hydrogen peroxide-degrading enzymes catalase and GPx1 (Lenzen et al., 1996;Tiedge et al., 1997) and thus deemed a sensitized experimental system to pursue the hypothesized role of glutathione in ER hydrogen peroxide metabolism. Compared with fibroblasts, insulinproducing RINm5F cells (a model for pancreatic β-cells) were noted to be hypersensitive to inhibition of glutathione biosynthesis by buthionine sulphoximine (BSO, Figure 1A & Figure S1). Cytosolic catalase expression reversed this hypersensitivity to BSO ( Figure 1B & C).
Induction of pro-insulin biosynthesis via a tetracycline inducible promoter ( Figure   1D), which burdens the ER with disulfide bond formation and promotes the associated production of hydrogen peroxide, contributed to the injurious effects of BSO. But these were partially reversed by the presence of ER-localized catalase ( Figure 1E). The protective effect of ER localized catalase is likely to reflect the enzymatic degradation of locally-produced hydrogen peroxide, as hydrogen peroxide is slow to equilibrate between the cytosol and ER . Together these findings hint at a role for glutathione in buffering the consequences of excessive production of hydrogen peroxide in the ER of pancreatic β-cells.

A probe adapted to detect H 2 O 2 in thiol-oxidizing environments
To further explore the role of glutathione in the metabolism of ER hydrogen peroxide we sought to measure the effects of manipulating glutathione availability on the TriPer_glut_H2O2_2.4.2f 8 changing levels of ER hydrogen peroxide. Exemplified by HyPer (Belousov et al., 2006), genetically encoded optical probes responsive to changing levels of hydrogen peroxide have been developed and via targeted localization, applied to the cytosol, peroxisome and mitochondrial matrix (Belousov et al., 2006;Gehrmann and Elsner, 2011;Gutscher et al., 2009;Malinouski et al., 2011). Unfortunately, in the thiol-oxidizing environment of the ER, the optically-sensitive disulfide in HyPer (that reports on the balance between hydrogen peroxide and contravening cellular reductive processes) instead forms via oxidized members of the protein disulfide isomerase family (PDIs), depleting the pool of reduced HyPer that can sense hydrogen peroxide Mehmeti et al., 2012).
To circumvent this limitation we sought to develop a probe that would retain responsiveness to hydrogen peroxide in the presence of high concentration of oxidized PDI. HyPer consists of a circularly permuted yellow fluorescent protein (YFP) grafted with the hydrogen peroxide-sensing portion of the bacterial transcription factor OxyR (Belousov et al., 2006;Choi et al., 2001). It possesses two reactive cysteines: a peroxidatic cysteine (OxyR C199) that reacts with H 2 O 2 to form a sulfenic acid and a resolving cysteine (OxyR C208) that attacks the sulfenic acid to form the optically distinct disulfide. We speculated that introduction of a third cysteine, vicinal to the resolving C208, might permit a rearrangement of the disulfide bonding pattern that could preserve a fraction of the peroxidatic cysteine in its reduced form and thereby preserve a measure of H 2 O 2 responsiveness, even in the thiol-oxidizing environment of the ER.
Replacement of OxyR alanine 187 (located ~6 Å from the resolving cysteine 208 in PDB1I69) with cysteine gave rise to a tri-cysteine probe, TriPer, that retained responsiveness to H 2 O 2 in vitro but with an optical readout that was profoundly different from that of HyPer: Whilst reduced HyPer exhibits a monotonic H 2 O 2 and time-dependent TriPer_glut_H2O2_2.4.2f increase in its excitability at 488 nm compared to 405 nm (R 488/405 , Figure 2A), in response to H 2 O 2 , the R 488/405 of reduced TriPer increased transiently before settling into a new steady state ( Figure 2B). TriPer's optical response to H 2 O 2 was dependent on the peroxidatic cysteine (C199), as its replacement by serine eliminated all responsiveness ( Figure 2C). R266 supports the peroxidatic properties of OxyR's C199, likely by deprotonation of the reactive thiol (Choi et al., 2001). The R266A mutation similarly abolished H 2 O 2 responsiveness of HyPer and TriPer indicating a shared catalytic mechanism for OxyR and the two derivative probes ( Figure S2A).
The optical response to H 2 O 2 of TriPer correlated in a dose and time-dependent manner with formation of high molecular weight disulfide bonded species, detectable on non-reducing SDS-PAGE ( Figure 2D & Figure S3A). These species were not observed in H 2 O 2 -exposed HyPer and their presence in TriPer, depended on both the peroxidatic C199 and on R266 ( Figure 2E, 2F and S2B). Furthermore, H 2 O 2 promoted such mixed disulfides in probe variants missing the resolving C208 or both C208 and the TriPer-specific C187 ( Figure 2G). The high molecular weight TriPer species induced by H 2 O 2 , migrate anomalously on standard SDS-PAGE, however on neutral pH gradient SDS-PAGE their size is consistent with that of a dimer ( Figure S2C).
The observations above indicate that in the absence of C208, H 2 O 2 induced C199 sulfenic intermediates are resolved in trans and suggest that formation of the divergent C208-C187 pair, unique to TriPer, favors this alternative route. To test this prediction we traced the R 488/405 of TriPer's under conditions mimicking oxidizing environment of the ER.
TriPer's time-dependent biphasic optical response (R 488/405 ) to H 2 O 2 contrasted with the hyperbolic profile of its response Diamide or PDI mediated oxidation ( Figure 3A). The latter is by far the most abundant ER thiol-oxidizing enzyme. PDI catalyzed HyPer TriPer_glut_H2O2_2.4.2f oxidation likewise had a hyperbolic profile but with a noticeable higher R 488/405 plateau ( Figure 3A). However, whereas TriPer retained responsiveness to H 2 O 2 , even from its PDI-oxidized plateau, PDI-oxidized HyPer lost all sensitivity to H 2 O 2 ( Figure 3B). Unlike H 2 O 2 driven formation of the optically active C199-C208 disulfide in HyPer enjoys a considerable kinetic advantage over its reduction by DTT . This was reflected here in the high R 488/405 of the residual plateau of HyPer co-exposed to H 2 O 2 and DTT ( Figure 3C). Thus, HyPer and TriPer traces converge at a high ratio point in the presence of H 2 O 2 and DTT ( Figure 3C & S3C); convergence that requires both C199 and C208 ( Figure S2D). In these conditions DTT releases TriPer's C208 from the divergent disulfide, allowing it to resolve C199-sulfenic in cis, thus confirming C199-C208 as the only optically distinct (high R 488/405 ) disulfide. It is worth noting that the convergence of TriPer and HyPer traces in these conditions confirms that in both C199-C208 in the sole high ratio states; consistent with the lack of optical response in all monomeric/dimeric configuration ( Figure S2). Thus, TriPer's biphasic response to H 2 O 2 , which is preserved in the face of PDI-driven oxidation (a mimic of conditions in the ER), emerges from the competing H 2 O 2 -driven formation of a trans-disulfide, imparting a low R 488/405 ( Figure 3D).

TriPer detects H 2 O 2 in the oxidizing ER environment
To test if the promising features of TriPer observed in vitro enable H 2 O 2 sensing in the ER, we tagged TriPer with a signal peptide and confirmed its ER localization in transfected cells ( Figure. 4A). Unlike ER-localized HyPer, whose optical properties remained unchanged in cells exposed to H 2 O 2 , ER-localized TriPer responded with a H 2 O 2 concentration-dependent decline in the R 488/405 ( Figure 4B & 4C).

TriPer_glut_H2O2_2.4.2f
The H 2 O 2 -mediated changes in the optical properties of ER-localized TriPer were readily reversed by washout or by introducing catalase into the culture media, which rapidly eliminated the H 2 O 2 ( Figure 4D & 4E). Both the slow rate of diffusion of H 2 O 2 into the ER  and the inherent delay imposed by the two-step process entailed in TriPer's responsiveness to H 2 O 2 (Figure 3), contribute to the sluggish temporal profile of the changes observed in TriPer's optical properties in cells exposed to H2O2.
Further evidence that TriPer was indeed responding to changing H 2 O 2 content of the ER was provided by the attenuated and delayed response to exogenous H 2 O 2 observed in cells expressing an ER-localized catalase ( Figure 4F).
The response of TriPer to H 2 O 2 could be tracked not only by following the changes in its excitation properties (as revealed in the R 488/405 ) but also by monitoring the fluorophore's fluorescence lifetime, using Fluorescent Lifetime Imaging Microscopy (FLIM) (as previously observed for other disulfide-based optical probes (Avezov et al., 2013;Bilan et al., 2013)).

Exposure of cells expressing ER TriPer to H 2 O 2 resulted in highly reproducible
increase in the fluorophore's fluorescence lifetime (with a dynamic range > 8 X SD, Figure   5A). HyPer's fluorescence lifetime was also responsive to H 2 O 2 , but only in the reducing environment of the cytoplasm ( Figure 5B); the lifetime of ER localized HyPer remained unchanged in cells exposed to H 2 O 2 ( Figure 5C). These findings are consistent with nearly complete oxidation of the C199-C208 disulfide under basal conditions in ER-localized HyPer and highlight the residual H 2 O 2 -responsiveness of ER-localized TriPer ( Figure   5D) Mehmeti et al., 2012).
Both ratiometry and FLIM trace alterations in the fluorophore resulting from C199-C208 formation. However, FLIM has important advantages over ratiometric measurements TriPer_glut_H2O2_2.4.2f of changes in probe excitation, especially when applied to cell imaging: It is a photophysical property of the probe that is relatively independent of the ascertainment platform and indifferent to photobleaching. Therefore, though ratiometric imaging is practical for short-term tracking of single cells, FLIM is preferable when populations of cells exposed to divergent conditions are compared. Under basal conditions ER-localized TriPer's lifetime indicated that it is found in a redox state where C199-C208 pair is nearly half-oxidized ( Figure 5D), resembling that of PDI exposed TriPer in vitro ( Figure 3A) and validating the use of FLIM to trace TriPer's response to H 2 O 2 in vivo.

Glutathione depletion leads to H 2 O 2 elevation in the ER of pancreatic cells
Exploiting the responsiveness of ER-localized TriPer to H 2 O 2 we set out to measure the effect of glutathione depletion on the ER H 2 O 2 signal as reflected in It is noteworthy that the increase in ER H 2 O 2 signal in BSO-treated cells was observed well before the increase in the cytosolic H 2 O 2 signal (Fig. 6C) and also preceded death of the glutathione depleted cells (Fig. S4B).

TriPer_glut_H2O2_2.4.2f
The ability of ER catalase to attenuate the optical response of ER-localized TriPer to BSO or pro-insulin induction argues for an increase in ER H 2 O 2 as the underlying event triggering the optical response. Two further findings support this conclusion: 1) The disulfide state of the ER tuned redox reporter, ERroGFPiE (Avezov et al., 2013;Birk et al., 2013;van Lith et al., 2011) remained unaffected by BSO. This argues against the possibility that the observed TriPer response is a consequence of a more reducing ER thiol redox poise induced by glutathione depletion ( Figure 6D). 2) TriPer's responsiveness to BSO and pro-insulin induction was strictly dependent on R266, a residue that does not engage in thiol redox directly, but is required for the peroxidatic activity of TriPer C199 ( Fig   6E & S4C). In addition, the above H 2 O 2 specificity controls of TriPer response exclude other possible artificial affects on the probe's fluorophore, such as pH changes.
To further explore the links between glutathione depletion and accumulation of ER H 2 O 2 , we sought to measure the effects of selective depletion of the ER pool of glutathione on the ER H 2 O 2 signal. ChaC1 is a mammalian enzyme that cleaves reduced glutathione into 5-oxoproline and cysteinyl-glycine (Kumar et al., 2012). We have adapted this normally cytosolic enzyme to function in the ER lumen and thereby deplete the ER pool of glutathione (Tsunoda et al., 2014). Enforced expression of ER localized ChaC1 in RINm5F cells led to an increase in fluorescence lifetime of ER TriPer, which was attenuated by concomitant expression of ER-localized catalase (Fig. 6F). Cysteinyl-glycine, the product of ChaC1 has a free thiol, but its ability to balance ER H 2 O 2 , may be affected by other factors such a clearance or protonation status. Given the relative selectivity of ER localized ChaC1 in depleting the luminal pool of glutathione (which equilibrates relatively slowly with the cytosolic pool (Kumar et al., 2011;Tsunoda et al., 2014)), these observations further support a role for ER-localized glutathione in the elimination of

Analysis of the potential for uncatalyzed quenching of H 2 O 2 by the ER pool of glutathione
Two molecules of reduced glutathione can reduce a single molecule of H 2 O 2 yielding a glutathione disulfide and two molecules of water, equation (1): However there is no evidence that the ER is endowed with enzymes capable of catalyzing this thermodynamically-favored reaction. For while the ER possesses two glutathione peroxidases, GPx7 and GPx8, both lack key structural determinates for interacting with reduced glutathione and function instead as peroxiredoxins, ferrying electrons from reduced PDI to H 2 O 2 (Nguyen et al., 2011). Therefore, we re-visited the feasibility of a role for the uncatalyzed reaction in H 2 O 2 homeostasis in the ER. Considering a scenario whereby oxidative folding of pro-insulin (precursor of the major secretory product of β-cells) proceeds by the enzymatic transfer of two electrons from di-thiols to O 2 , (as in ERO1/PDI catalysis, generating one molecule of H 2 O 2 per TriPer_glut_H2O2_2.4.2f disulfide formed (Gross et al., 2006)) and given, three disulfides in pro-insulin, a maximal production rate of 6*10 -4 femtomoles pro-insulin/min/cell (Schuit et al., 1991) and an ER volume of 280 femtoliters (Dean, 1973), the resultant maximal generation rate of H 2 O 2 has the potential to elevate its concentration by 0.098 µM per second. Given the rate constant of 29 M -1 s -1 for the bi-molecular reaction of H 2 O 2 with reduced glutathione and an estimated ER concentration of 15 mM GSH (Montero et al., 2013), at this production rate the concentration of H 2 O 2 would stabilize at 0.23 µM, based solely on uncatalyzed reduction by GSH. Parallel processes that consume H 2 O 2 to generate disulfide bonds would tend to push this concentration even lower (Nguyen et al., 2011;Ramming et al., 2014;Tavender and Bulleid, 2010;Zito et al., 2010), nonetheless this calculation indicates that GSH can play an important role in the uncatalyzed elimination of H 2 O 2 from the ER. TriPer_glut_H2O2_2.4.2f

Discussion:
The sensitivity of β-cells to glutathione depletion, the accentuation of this toxic effect by pro-insulin synthesis and the ability of ER catalase to counteract these challenges all hinted at a role for glutathione in coping with the burden of H 2 O 2 produced in the ER. However, without means to track H 2 O 2 in the ER of living cells, this would have remained an untested idea. TriPer has revealed that ER hydrogen peroxide levels increase with increased production of disulfide bonds in secreted proteins, providing direct evidence that the oxidative machinery of the ER does indeed produce H 2 O 2 as a (bi)product. Less anticipated has been the increased H 2 O 2 in the ER of glutathionedepleted cells. The contribution of glutathione to a reducing milieu in the nucleus and cytoplasm is well established, but it has not been easy to rationalize its presence in the oxidizing ER; especially as glutathione appears dispensable for the reductive step of disulfide isomerization in oxidative protein folding (Tsunoda et al., 2014). Consistently, the ER thiol redox poise resisted glutathione depletion. TriPer has thus pointed to a role for glutathione in buffering ER H 2 O 2 production, providing a plausible benefit from the presence of a glutathione pool in the ER lumen.
TriPer's ability to sense H 2 O 2 in a thiol-oxidizing environment relies on the presence of an additional cysteine residue (A187C) near the resolving C208 in the OxyR segment. The presence of this additional cysteine attenuates the optical responsiveness of the probe to oxidation by PDI by creating a diversionary disulfide involving C208.
Importantly, this diversionary disulfide, which forms at the expense of the (optically active) C199-C208 disulfide preserves a fraction of the peroxidatic C199 in its reduced form.
Thus, even in the thiol-oxidizing environment of the ER a fraction of C199 thiolate is free to form a reversible H 2 O 2 driven sulfenic intermediate that is resolved by disulfide formation TriPer_glut_H2O2_2.4.2f in trans. Furthermore, by preserving a fraction of C199 in its reduced form, the diversionary C187-C208 disulfide also maintains a pool of C199 to resolve the C199 sulfenic to form the trans-disulfide bond. Thus, the H 2 O 2 induced formation of the trans disulfide is a feature unique to TriPer, and it too is a consequence of the diversionary disulfide, which eliminates the strongly competing reaction of the resolving C208 in cis with the sulfenic acid at C199. The aforementioned theoretical arguments for TriPer's direct responsiveness to H 2 O 2 are further supported by empirical observations: the peroxidatic potential (unusual for cysteine thiols) of the probe's C199 is enabled through a finally balance charge distribution in its vicinity, of which R266 is a crucial determinant (Choi et al., 2001). Eliminating this charge yielded a probe variant with its intact cysteine system, but unresponsive to H 2 O 2 . Further, the ability of ER catalase to reverse the changes in TriPer's disposition argues that these are initiated by changes in H 2 O 2 concentration.
Both Hyper and TriPer react with elements of the prevailing ER thiol redox buffering system (exemplified by their equilibration, in vitro, with PDI). In the case of HyPer this reactivity is ruinous, but even in the case of ER TriPer, which retains a modicum of sensitivity to H 2 O 2 , the elements of the complex kinetic regime that drive its redox state are TriPer_glut_H2O2_2.4.2f not understood in quantitative terms. Thus, it is impossible to fully deconvolute the potential impact on TriPer of changes in the ER thiol redox milieu from changes in H 2 O 2 concentration wrought by a given physiological perturbation -TriPer is sensitive to both.
However, it is noteworthy that oxidation of TriPer by H 2 O 2 , leading to a mixed disulfide state, shifts the optical readout towards lower R 488/405 and shorter fluorescent lifetimes.
Whilst such shifts are also consistent with a surge in thiol reductive activity, it seems unlikely that exposure of cells to H 2 O 2 results in a more thiol-reducing ER. Similar considerations apply to the state of the ER in cells depleted of glutathione, as it is hard to imagine how this would lead to a more thiol-reducing ER. Indeed the H 2 O 2 -insensitive redox probe roGFPiE, which is know to equilibrate with ER localized PDI, was unaffected by glutathione depletion. Thus, while we cannot formally exclude that glutathione depletion also affects TriPer's redox status independently of changes in H 2 O 2 concentration, the bulk of the evidence favors a role for TriPer in tracking the latter and in reporting on an increase in ER H 2 O 2 in glutathione-depleted cells.
TriPer has been instrumental in flagging glutathione's role in buffering ER luminal H 2 O 2 . This raises the question as to whether the thermodynamically favored reduction of H 2 O 2 by GSH is accelerated by ER-localized enzymes or proceeds by uncatalyzed mass action. The cytoplasm and mitochondria possess peroxide consuming enzymes that are fueled by reduced glutathione (Lillig et al., 2008). However the ER lacks known counterparts. Such enzymes may be discovered in the future, as well as possible pathways of GSH mediated PDRX4 modulation. But meanwhile it is notable that the kinetic properties of the uncatalyzed reduction of H 2 O 2 by GSH are consistent with its potential in keeping H 2 O 2 concentration at bay.  ; which provides an explanation for the observed delay between the increase in ER and cytosolic H 2 O 2 when glutathione synthesis was inhibited. Sequestration of H 2 O 2 in the lumen of the ER protects the genome from this potentially harmful metabolite and enables the higher concentrations needed for the uncatalyzed reaction to progress at a reasonable pace.
The implementation of a probe that detects H 2 O 2 in thiol oxidizing environments has revealed a remarkably simple mechanism to defend the cytosol and nucleus from a (bi)product of oxidative protein folding in the ER. This mechanism is especially important in secretory pancreatic β cells that are poorly equipped with catalase/peroxidase.

Materials and Methods:
Plasmid construction: Table S1 lists the plasmids used, their lab names, description, published reference and a notation of their appearance in the figures. The readouts of each set were normalized its the maximum value (untreated sample).

Quantification of catalase enzymatic activity
Catalase enzyme activity was quantified as described earlier (Tiedge et al., 1998).
Briefly, whole-cell extracts were homogenized in PBS through sonication on ice with a Braun-Sonic 125 sonifier (Braun, Melsungen, Germany). Subsequently the homogenates were centrifuged at 10,000 x g and 4 °C for 10 min. The protein content of the supernatant was assessed by the BC Assay (Thermo Fisher Scientific, Rockford, IL, USA). For quantification of the catalase enzyme activity 5 µg of the total protein lysate were added to 50 mmol/L potassium phosphate buffer (pH 7.8) containing 20 mmol/L H 2 O 2 . The specific catalase activity was measured by ultraviolet spectroscopy, monitoring the decomposition of H 2 O 2 at 240 nm and calculated as described in (Tiedge et al., 1998).
Fluorescence ratiometric intensity images (512 x 512 points, 16 bit) of live cells were acquired. A diode 405 nm and Argon 488 nm lasers (6 and 4% output respectively) were used for excitation of the ratiometric probes in the multitrack mode with an HFT 488/405 beam splitter, the signal was detected with 506-568 nm band pass filters, the detector gain was arbitrary adjusted to yield an intensity ratio of the two channels to allow a stable baseline and detection of its redox related alterations.
FLIM experiments were performed on a modified version of a previously-described laser scanning multiparametric imaging system (Frank et al., 2007) After filtering out autofluorescence (by excluding pixels with a fluorescence lifetime that out of range of the probes) mean fluorescence lifetime of single cells was established.
Each data point is constituted by the average and SD of measurements from at least 20

Protein purification and kinetic assays in vitro
For in-vitro assays, human PDI (PDIA1 18-508), HyPer and TriPer were expressed in the E. coli BL21 (DE3) strain, purified with Ni-NTA affinity chromatography and analyzed by fluorescence excitation ratiometry as previously described . Briefly, HyPer, TriPer and their mutants were assayed in vitro in Tris-HCl buffer, pH 7.4, 150mM NaCl after being reduced with 50mM DTT for one hour followed by gel filtration to remove DTT.
To establish the reactivity of H 2 O 2 with GSH different amounts of GSH (0-3 mM, pH adjusted to 7.1; Sigma, Gillingham, Dorset, UK) were mixed with 10 µM of H 2 O 2 for a fixed time period, and then exposed to recombinant HyPer (2 µM, reduced by 40 mM DTT and gel filtered to remove DTT). The relationship between the rate of HyPer oxidation and [GSH], described by equation (2), was used to extract the remaining [H 2 O 2 ] after its exposure to various [GSH] for a given period, using equation (3).
Where the rates of HyPer oxidation (s -1 ) in a given [H 2 O 2 ] are denoted by R g (GSHaffected) and R i (in the absence of GSH), S 1 is the experimental coefficient (s -1 mM -1 ). The latter two are the y-intercept and the slope of the curve in Figure 6D, accordingly. corresponding reaction rate according to equation (4), developed based on equation (3) for the special case of H 2 O 2 -GSH reaction time (t) and [H2O2] (resulting curve shown in Fig. 6D inset, S 2 is the slope). The bi-molecular rate constant (k) is given by equation (5).
The concentration of H 2 O 2 at the equilibrium where the rate of its supply equals the rate of its reaction with GSH was calculated according to the equation (6): Where V [H 2 O 2 ] is the assumed rate of H 2 O 2 generation and K is the bi-molecular rate constant for the GSH/H 2 O 2 reactivity. TriPer_glut_H2O2_2.4.2f Lortz, S., Lenzen, S., and Mehmeti, I. (2015b). N-glycosylation-negative catalase: a useful tool for exploring the role of hydrogen peroxide in the endoplasmic reticulum. Free radical biology & medicine 80, 77-83.
Mehmeti, I., Lortz, S., and Lenzen, S. (2012). The H2O2-sensitive HyPer protein targeted to the endoplasmic reticulum as a mirror of the oxidizing thiol-disulfide milieu.   Shown are representatives of n ≥ 3.
(C) Ratiometric traces (as in "a") of HyPer or TriPer exposed to H 2 O 2 (4 µM) followed by DTT (2.6 mM). The excitation spectra of the phases of the reaction (1-4) are analyzed in Figure S3C.
TriPer_glut_H2O2_2.4.2f (E) A ratiometric trace of TriPer ER expressed in RINm5F cells, exposed to H 2 O 2 (0.2 mM) followed by bovine catalase for the indicated duration.
(F) A ratiometric trace of TriPer ER expressed alone or alongside ER catalase (CAT ER ) in RINm5F cells, exposed to increasing concentrations of H 2 O 2 (0 -0.2 mM). Shown are representatives of n ≥ 3.
TriPer_glut_H2O2_2.4.2f ng/ml) and the cells were exposed to 0.3 mM BSO (18 hours) (B) As in "a" but TriPer ER -expressing cells were exposed to 0.15 mM BSO (18 hours.  HyPers oxidation rate and the length of the preceding H 2 O 2 -GSH pre-incubation. (B) As in (A) but following HyPer introduction into pre-mixed solutions of H 2 O 2 (10 µM) and reduced glutathione (GSH) at the indicated concentrations (0 -3 mM).