Functional Analysis of GSTK1 in Peroxisomal Redox Homeostasis in HEK-293 Cells

Peroxisomes serve as important centers for cellular redox metabolism and communication. However, fundamental gaps remain in our understanding of how the peroxisomal redox equilibrium is maintained. In particular, very little is known about the function of the nonenzymatic antioxidant glutathione in the peroxisome interior and how the glutathione antioxidant system balances with peroxisomal protein thiols. So far, only one human peroxisomal glutathione-consuming enzyme has been identified: glutathione S-transferase 1 kappa (GSTK1). To study the role of this enzyme in peroxisomal glutathione regulation and function, a GSTK1-deficient HEK-293 cell line was generated and fluorescent redox sensors were used to monitor the intraperoxisomal GSSG/GSH and NAD+/NADH redox couples and NADPH levels. We provide evidence that ablation of GSTK1 does not change the basal intraperoxisomal redox state but significantly extends the recovery period of the peroxisomal glutathione redox sensor po-roGFP2 upon treatment of the cells with thiol-specific oxidants. Given that this delay (i) can be rescued by reintroduction of GSTK1, but not its S16A active site mutant, and (ii) is not observed with a glutaredoxin-tagged version of po-roGFP2, our findings demonstrate that GSTK1 contains GSH-dependent disulfide bond oxidoreductase activity.


Introduction
Complex living systems can only survive within a very narrow range of physicochemical conditions [1]. However, they are constantly exposed to a myriad of stimuli that disrupt this equilibrium [2]. To not endanger life, organisms have developed and rely on intricate signaling networks to return to the homeostatic state [1]. An inefficiency of these networks can lead to illness and ultimately death, thereby highlighting their importance for proper cell functioning [2]. From a cellular perspective, a relevant example of a healthy physiological state is redox homeostasis, which refers to the dynamic equilibrium between molecules that regulate the electrophilic and nucleophilic tone [3].
An important molecule for redox homeostasis is the tripeptide γ-glutamyl-cysteinylglycine, commonly known as glutathione, whose thiol group supplies reducing power for numerous cellular redox reactions [4]. This metabolite is part of an intricate thioldependent antioxidant defense mechanism, known as the glutathione system, which depends on NADPH and a group of enzymes that regulate the amount of reduced and This study focuses on the role of GSTK1 in peroxisomal glutathione homeostasis. Given that (i) GSTK1 makes use of GSH as an obligate cofactor or cosubstrate [24,42], (ii) GSH and NAD(P)H levels often correlate [43][44][45], (iii) GSH is a crucial substrate for many enzymes involved in H 2 O 2 detoxification [7], and (iv) depending on the cell type, its physiological state, and the microenvironment, peroxisomes can be a significant source or sink of H 2 O 2 [12]. We examined how GSTK1 inactivation in human embryonic kidney 293 (HEK-293) cells impacted these redox metabolites in peroxisomes and the cytosol under both basal and/or oxidative stress conditions. Since the loss of GSTK1 impairs the intraperoxisomal redox recovery of roGFP2 after an oxidative insult, and this trait can be reversed by expressing po-GRX1-roGFP2, our findings demonstrate that the peroxisomeassociated pool of GSTK1 exhibits glutathione-disulfide oxidoreductase activity in cellulo.
Subcellular fractionations of HEK-293 cells were performed as described [54]. Based on this study, the gradient fractions with a density ranging between 1.20 and 1.08 g/mL (fractions 12 to 20) were processed for immunoblotting.

Generation of Gene Knockout Cells by CRISPR-Cas9 Genome Editing
The CRISPR-Cas9 genome editing technology (as described by [55]) was used to introduce a functional disruption of the GSTK1 gene in DD-DAO Flp-In T-REx 293 cells [46]. To this end, the cells were transfected with the plasmid, encoding the guide RNA and clonal cells with homozygous (clone 2, cl 2; c.133insA) or compound heterozygous (clone 1, cl 1; c.133insA/c.133delA) out-of-frame variants were selected for downstream analysis (NCBI reference sequence: NM_015917.3).
Proteins samples were precipitated by treating DPBS-resuspended cell suspensions with 6% (w/v) trichloroacetic acid and 0.0125% (w/v) deoxycholate, followed by acetonewashing. Thereafter, the protein samples were processed for reducing or nonreducing SDS-PAGE and immunoblotting as described elsewhere [46]. Immunoblot signal intensities were quantified using ImageJ software [59].

Statistical Analysis
Statistical analysis was performed using GraphPad Prism (version 9.0.0 for Windows 64-bit, GraphPad Software, San Diego, CA, USA). The statistical tests used are specified in the figure legends. A p-value lower than 0.05 was considered statistically significant. Considering this is an exploratory study, no Bonferroni corrections were applied.

GSTK1 Displays Dual Peroxisomal and Mitochondrial Localization in HEK-293 Cells
Given the contradictory findings on the subcellular location of GSTK1 [17,[22][23][24][25][26], we first confirmed that at least a portion of GSTK1 cofractionates with peroxisomes in HEK-293 cells. As initial experiments indicated that our anti-GSTK1 antibody was not suitable for immunofluorescence microscopy, we subjected a post-nuclear supernatant from HEK-293 cells to Nycodenz density gradient centrifugation. Immunoblot analysis of the relevant organelle-containing fractions [54] demonstrated that GSTK1 clearly coenriches with the peroxisomal marker CAT and the mitochondrial protein TOMM22 (Figure 1), which is consistent with (i) predictions that the protein contains peroxisomal and mitochondrial targeting information within its mature protein sequence [25] and (ii) its previously described dual localization in mitochondria and peroxisomes in human hepatoblastoma cells [17]. In addition, despite observing a similar distribution pattern between GSTK1 and TOMM22 in the low-density fractions 17 to 20, our findings do not exclude the potential presence of GSTK1 within the ER.

Generation and Validation of the GSTK1 HEK-293 Cell Lines
To study the potential role of GSTK1 in peroxisomal glutathione redox metabolism, we selectively disrupted the corresponding gene in HEK-293 cells (genetic background: DD-DAO Flp-In-T-REx 293 [46]) by using the CRISPR-Cas9 technology [55].
A heterozygous (c.133insA/c.133delA; cl 1) and a homozygous (c.133insA; cl 2) knockout clone were selected and their correctness was validated at the protein level by immunoblot analysis (Figures 2 and S1). If considered meaningful, both clones were incorporated into the analyses to reduce the chance of drawing inaccurate conclusions solely based on a single clonal population of cells. Note that GSTK1 inactivation does not significantly impact the expression levels of any of the other examined peroxisome-related (e.g., the β-oxidation enzymes ACAA1 and HSD17B4, the antioxidant enzymes CAT and PRDX5, and the peroxins PEX5 and PEX14) (Figures 2 and S1), mitochondrial (e.g., COX IV, ND6, and TOMM22) (Figures 3A and S1), and ER (e.g., CALR) (Figures 3A and S1) proteins. As GSTK1 is located both in peroxisomes and mitochondria, we also carried out high-resolution respirometry studies to assess mitochondrial function. Again, no significant differences could be observed between the control and GSTK1 cells ( Figure 3B).

Figure 2.
The absence of GSTK1 does not impact the expression levels of peroxisome-related proteins. Total cell lysates from control (CT), ∆GSTK1 (clone 1, cl1), or ∆GSTK1 (clone 2, cl2) cells, all containing equal amounts of protein, were processed for immunoblot analysis with antisera directed against the indicated proteins. Representative immunoblots are shown (immunoblot quantifications of three biological replicates are provided in Figure S1). The migration points of relevant

Generation and Validation of the ∆GSTK1 HEK-293 Cell Lines
To study the potential role of GSTK1 in peroxisomal glutathione redox metabolism, we selectively disrupted the corresponding gene in HEK-293 cells (genetic background: DD-DAO Flp-In-T-REx 293 [46]) by using the CRISPR-Cas9 technology [55].
A heterozygous (c.133insA/c.133delA; cl 1) and a homozygous (c.133insA; cl 2) knockout clone were selected and their correctness was validated at the protein level by immunoblot analysis (Figures 2 and S1). If considered meaningful, both clones were incorporated into the analyses to reduce the chance of drawing inaccurate conclusions solely based on a single clonal population of cells. Note that GSTK1 inactivation does not significantly impact the expression levels of any of the other examined peroxisome-related (e.g., the β-oxidation enzymes ACAA1 and HSD17B4, the antioxidant enzymes CAT and PRDX5, and the peroxins PEX5 and PEX14) (Figures 2 and S1), mitochondrial (e.g., COX IV, ND6, and TOMM22) (Figures 3A and S1), and ER (e.g., CALR) (Figures 3A and S1) proteins. As GSTK1 is located both in peroxisomes and mitochondria, we also carried out high-resolution respirometry studies to assess mitochondrial function. Again, no significant differences could be observed between the control and ∆GSTK1 cells ( Figure 3B).

Generation and Validation of the GSTK1 HEK-293 Cell Lines
To study the potential role of GSTK1 in peroxisomal glutathione redox metabolism, we selectively disrupted the corresponding gene in HEK-293 cells (genetic background: DD-DAO Flp-In-T-REx 293 [46]) by using the CRISPR-Cas9 technology [55].
A heterozygous (c.133insA/c.133delA; cl 1) and a homozygous (c.133insA; cl 2) knockout clone were selected and their correctness was validated at the protein level by immunoblot analysis (Figures 2 and S1). If considered meaningful, both clones were incorporated into the analyses to reduce the chance of drawing inaccurate conclusions solely based on a single clonal population of cells. Note that GSTK1 inactivation does not significantly impact the expression levels of any of the other examined peroxisome-related (e.g., the β-oxidation enzymes ACAA1 and HSD17B4, the antioxidant enzymes CAT and PRDX5, and the peroxins PEX5 and PEX14) (Figures 2 and S1), mitochondrial (e.g., COX IV, ND6, and TOMM22) (Figures 3A and S1), and ER (e.g., CALR) (Figures 3A and S1) proteins. As GSTK1 is located both in peroxisomes and mitochondria, we also carried out high-resolution respirometry studies to assess mitochondrial function. Again, no significant differences could be observed between the control and GSTK1 cells ( Figure 3B).  Figure S1). The migration points of relevant   (clone 1, cl1), or ∆GSTK1 (clone 2, cl2) cells, all containing equal amounts of protein, were processed for immunoblot analysis with antisera directed against the indicated proteins. Representative immunoblots are shown (immunoblot quantifications of three biological replicates are provided in Figure S1). The migration points of relevant molecular mass markers (expressed in kDa) are shown on the left. (B) The graph depicts the different oxygen consumption rate (OCR)-related parameters normalized to cell number. Each dot or triangle corresponds to an individual data point. The means are represented by horizontal lines. The data for each ΔGSTK1 cell line were statistically compared with those of the CT cell line using the ordinary two-way ANOVA test with Tukey's multiple comparisons test, but no significant differences were found.

GSTK1 Inactivation Does Not Affect the Basal Peroxisomal and Cytosolic Redox States
Given that (i) the overarching aim of this study was to examine the role of GSTK1 in peroxisomal glutathione redox homeostasis, (ii) organellar glutathione pools rely entirely on cytosolic glutathione import [63], and (iii) changes in (local) glutathione metabolism may have a direct impact on the levels of other redox metabolites [7,[43][44][45], we first evaluated if GSTK1 inactivation affected the overall peroxisomal and cytosolic GSSG/GSH, NAD + /NADH, NADPH, and H2O2 levels under basal conditions. To monitor possible changes in these metabolites, we used compartment-specific forms of the ratiometric fluorescent redox sensors roGFP2 [64], SoNar [65], iNAP1 [65], and roGFP2-Orp1 [64], respectively. The typical subcellular distribution patterns for each of these reporter proteins are depicted in Figure S2. All four reporters can exist in different conformational states, each of which corresponds to a particular fluorescent excitation spectrum. In the case of SoNar, the excitation maximum is at 485 nm or around 420 nm, depending on whether it binds NAD + or NADH [51]. Likewise, the excitation maximum of iNAP1 varies depending on whether it binds NADPH or not (±500 nm or ±420 nm, respectively) [52]. As a result, increased NAD + /NADH or NADPH levels lead to higher F480/F400 ratios for SoNar or  Figure S1). The migration points of relevant molecular mass markers (expressed in kDa) are shown on the left. (B) The graph depicts the different oxygen consumption rate (OCR)-related parameters normalized to cell number. Each dot or triangle corresponds to an individual data point. The means are represented by horizontal lines. The data for each ∆GSTK1 cell line were statistically compared with those of the CT cell line using the ordinary two-way ANOVA test with Tukey's multiple comparisons test, but no significant differences were found.

GSTK1 Inactivation Does Not Affect the Basal Peroxisomal and Cytosolic Redox States
Given that (i) the overarching aim of this study was to examine the role of GSTK1 in peroxisomal glutathione redox homeostasis, (ii) organellar glutathione pools rely entirely on cytosolic glutathione import [63], and (iii) changes in (local) glutathione metabolism may have a direct impact on the levels of other redox metabolites [7,[43][44][45], we first evaluated if GSTK1 inactivation affected the overall peroxisomal and cytosolic GSSG/GSH, NAD + /NADH, NADPH, and H 2 O 2 levels under basal conditions. To monitor possible changes in these metabolites, we used compartment-specific forms of the ratiometric fluorescent redox sensors roGFP2 [64], SoNar [65], iNAP1 [65], and roGFP2-Orp1 [64], respectively. The typical subcellular distribution patterns for each of these reporter proteins are depicted in Figure S2. All four reporters can exist in different conformational states, each of which corresponds to a particular fluorescent excitation spectrum. In the case of SoNar, the excitation maximum is at 485 nm or around 420 nm, depending on whether it binds NAD + or NADH [51]. Likewise, the excitation maximum of iNAP1 varies depending on whether it binds NADPH or not (±500 nm or ±420 nm, respectively) [52]. As a result, increased NAD + /NADH or NADPH levels lead to higher F480/F400 ratios for SoNar or iNAP1, respectively. Since iNAP1 and SoNar are pH sensitive, it is critical to conduct parallel measurements with sensors that react similarly to pH but not to NAD(H) or NADPH (e.g., iNAPc, cpYFP, or mKeima). In the case of roGFP2 and roGFP2-Orp1, oxidation results in the formation of an intramolecular disulfide bridge, which shifts the excitation maximum from 488 nm to 405 nm [66]. Hence, the greater the observed F400/F480 ratio, the higher the GSSG/GSH and H 2 O 2 levels, respectively. No significant alterations in any of the examined redox metabolites could be found between control and ∆GSTK1 cells, neither in peroxisomes nor in the cytosol (Figure 4). Potential explanations for this lack of a discernible redox phenotype in ∆GSTK1 cells may include (i) a nonredox role of GSTK1, (ii) a functional redundancy between GSTK1 and other (unidentified) proteins, or (iii) GSTK1's nonessential role in maintaining redox homeostasis under the conditions studied.

of 18
a discernible redox phenotype in ΔGSTK1 cells may include (i) a non-redox role of GSTK1, (ii) a functional redundancy between GSTK1 and other (unidentified) proteins, or (iii) GSTK1's nonessential role in maintaining redox homeostasis under the conditions studied. Two to three days later, the F400/F480 (for roGFP2 and roGFP2-Orp1) and pH-corrected F480/F400 response ratios (for SoNar and iNAP1) were measured, normalized to the average value of the control cells, and shown as violin plots. The data were derived from 30 individual images (represented as dots) from 3 independent experiments, each comprising 10 fields of view. The horizontal solid and dashed lines denote the median and the first and third quartiles, respectively. The data for each ΔGSTK1 cell line were statistically compared with those of the CT cell line using the Kruskal-Wallis test with Dunn's multiple comparisons test, but no significant differences were found.

GSTK1 Aids in the Recovery of Peroxisomal RoGFP2 after Oxidative Insult Withdrawal
To investigate whether GSTK1 regulates antioxidant defenses in response to oxidative stress, we subjected po-roGFP2-expressing control and ΔGSTK1 cells to mild and severe transient oxidative insults. First, we selectively generated H2O2 inside peroxisomes through the expression and D-Ala-mediated activation of DD-DAO [46]. Unfortunately, despite the fact that considerable amounts of po-H2O2 were generated ( Figure S3A), this treatment was unable to sufficiently oxidize po-roGFP2 to reliably quantify potential dif- po-roGFP2-Orp1, (G) po-SoNar or po-mKeima, or (H) po-iNAP1 or po-iNAPc and cultured in rMEMα. Two to three days later, the F400/F480 (for roGFP2 and roGFP2-Orp1) and pH-corrected F480/F400 response ratios (for SoNar and iNAP1) were measured, normalized to the average value of the control cells, and shown as violin plots. The data were derived from 30 individual images (represented as dots) from 3 independent experiments, each comprising 10 fields of view. The horizontal solid and dashed lines denote the median and the first and third quartiles, respectively. The data for each ∆GSTK1 cell line were statistically compared with those of the CT cell line using the Kruskal-Wallis test with Dunn's multiple comparisons test, but no significant differences were found.

GSTK1 Aids in the Recovery of Peroxisomal RoGFP2 after Oxidative Insult Withdrawal
To investigate whether GSTK1 regulates antioxidant defenses in response to oxidative stress, we subjected po-roGFP2-expressing control and ∆GSTK1 cells to mild and severe transient oxidative insults. First, we selectively generated H 2 O 2 inside peroxisomes through the expression and D-Ala-mediated activation of DD-DAO [46]. Unfortunately, despite the fact that considerable amounts of po-H 2 O 2 were generated ( Figure S3A), this treatment was unable to sufficiently oxidize po-roGFP2 to reliably quantify potential differences in recovery of the reduced fraction of po-roGFP2 ( Figure S3B). Within the context of these experiments, it is important to note that (i) under basal conditions, po-roGFP2-Orp1 is almost entirely oxidized in HEK-293 cells, rendering this reporter unsuitable for monitoring changes in po-H 2 O 2 production [46] and (ii) po-H 2 O 2 can rapidly permeate across the peroxisomal membrane, thereby allowing us to monitor peroxisomal H 2 O 2 production with c-roGFP2-Orp1 [54]. Second, we exposed the cells to exogenous H 2 O 2 , but once again, we failed to observe a significant rise in the oxidation state of po-roGFP2 ( Figure S3C), a finding in line with our previous observations that peroxisomes very well resist oxidative stress generated outside the organelle [15].
Given the apparent difficulty in producing insults that po-roGFP2 can detect, we transiently exposed the cells to AT-4 and diamide, two oxidants whose mode of action involves a direct reaction with thiols [67,68] and that have previously been shown to completely oxidize roGFP2 in minutes [69]. Upon addition of AT-4, po-roGFP2 was rapidly oxidized in both control and ∆GSTK1 cells ( Figure 5A). Intriguingly, after washout of the oxidant, analysis of the sensor's redox state at different time intervals revealed that po-roGFP2 recovered from AT-4-induced oxidation in the ∆GSTK1 cells at a considerably slower rate than in the control cells ( Figure 5A). Importantly, reintroducing GSTK1 into the ∆GSTK1 cells restored this deficit ( Figure 5B,D), thereby demonstrating that the observed phenotype is indeed caused by a defect in GSTK1 function.

of 18
completely oxidize roGFP2 in minutes [69]. Upon addition of AT-4, po-roGFP2 was rapidly oxidized in both control and ΔGSTK1 cells ( Figure 5A). Intriguingly, after washout of the oxidant, analysis of the sensor's redox state at different time intervals revealed that po-roGFP2 recovered from AT-4-induced oxidation in the ΔGSTK1 cells at a considerably slower rate than in the control cells ( Figure 5A). Importantly, reintroducing GSTK1 into the ΔGSTK1 cells restored this deficit ( Figure 5B,D), thereby demonstrating that the observed phenotype is indeed caused by a defect in GSTK1 function.  Unexpectedly, diamide administration had no discernible effect on the baseline oxidation of po-roGFP2, even not at a concentration of 5 mM ( Figure S3D). Given this surprising finding, we also investigated the effects of this oxidant on c-roGFP2 in both control and ∆GSTK1 cells. As previously reported [69], the oxidant rapidly oxidized c-roGFP2 ( Figure 6B). Interestingly, no differences in recovery rates could be observed between the control and ∆GSTK1 cells ( Figure 6B). This finding, which was also confirmed in cells transiently challenged with AT-4 ( Figure 5C), demonstrates that the reduced recovery rate of roGFP2 in GSTK1 cells is a peroxisome-specific phenomenon. To potentially increase the intraperoxisomal diamide concentration, we coincubated the cells for 10 min with this oxidant in the presence of 0.001% (w/v) Triton X-100, a nonionic detergent that can permeabilize cell membranes. Under those conditions, po-roGFP2 responded similarly as observed in AT-4-treated control and ∆GSTK1 cells ( Figure 6A). Combined, these findings provide evidence that GSTK1 can catalyze thiol-disulfide exchanges between roGFP2 and the glutathione redox pair in peroxisomes. Given that, under the conditions employed, the response of po-roGFP2 required concentrations in the millimolar range ( Figures 6A and S3E); whereas c-roGFP2 already responded to 50 µM of AT-4 or diamide ( Figures 5C and 6B), the peroxisomal membrane apparently constitutes a permeability barrier for diamide and AT-4.

of 18
the response of po-roGFP2 required concentrations in the millimolar range ( Figures 6A  and S3E); whereas c-roGFP2 already responded to 50 μM of AT-4 or diamide ( Figures 5C  and 6B), the peroxisomal membrane apparently constitutes a permeability barrier for diamide and AT-4. Figure 6. GSTK1 aids in the recovery of po-but not c-roGFP2 after diamide withdrawal. CT or ∆GSTK1 (cl 1 or 2) HEK-293 cells were transfected with a plasmid encoding (A) po-roGFP2 or (B) c-roGFP2 and cultured in rMEMα. After two to three days, the basal oxidation states (b) of po-and c-roGFP2 were measured and the cells were incubated (A) for 10 min with 2 mM diamide and 0.001% (w/v) Triton X-100 or (B) for 5 min with 50 μM diamide. After the removal of the oxidant, the F400/F480 response ratios of the sensors were monitored over time. The response ratios were normalized to the average value of the corresponding basal condition. The data points and vertical bars represent the mean and standard deviation of 30 to 60 individual measurements from 3 to 6 independent experiments, respectively. For each time point, the data obtained for the ∆GSTK1 conditions were statistically compared with those of the CT cells using the ordinary two-way ANOVA test with Tukey's multiple comparisons test (*, p < 0.05; ***, p < 0.001; ****, p < 0.0001).

Po-GRX1-roGFP2′s Recovery Rates Are Comparable in AT-4-Insulted Control and ΔGSTK1 Cells
RoGFP-based probes have been demonstrated to specifically equilibrate with the glutathione redox potential through the action of (endogenous) GRXs, which catalyze the thiol-disulfide exchange between the GSH-GSSG and roGFP2 redox pairs [70]. Given that (i) until now, no peroxisomal GRXs have been identified [18], and (ii) roGFP2-derived probes do respond to a reduction in the peroxisomal redox state [15], a phenomenon impaired in ΔGSTK1 cells (Figures 5 and 6), we examined the recovery behavior of po-GRX1-roGFP2 [47] in AT4-insulted ΔGSTK1 cells. These studies showed that the po-roGFP2 re- Figure 6. GSTK1 aids in the recovery of po-but not c-roGFP2 after diamide withdrawal. CT or ∆GSTK1 (cl 1 or 2) HEK-293 cells were transfected with a plasmid encoding (A) po-roGFP2 or (B) c-roGFP2 and cultured in rMEMα. After two to three days, the basal oxidation states (b) of poand c-roGFP2 were measured and the cells were incubated (A) for 10 min with 2 mM diamide and 0.001% (w/v) Triton X-100 or (B) for 5 min with 50 µM diamide. After the removal of the oxidant, the F400/F480 response ratios of the sensors were monitored over time. The response ratios were normalized to the average value of the corresponding basal condition. The data points and vertical bars represent the mean and standard deviation of 30 to 60 individual measurements from 3 to 6 independent experiments, respectively. For each time point, the data obtained for the ∆GSTK1 conditions were statistically compared with those of the CT cells using the ordinary two-way ANOVA test with Tukey's multiple comparisons test (*, p < 0.05; ***, p < 0.001; ****, p < 0.0001).

Po-GRX1-roGFP2 s Recovery Rates Are Comparable in AT-4-Insulted Control and ∆GSTK1 Cells
RoGFP-based probes have been demonstrated to specifically equilibrate with the glutathione redox potential through the action of (endogenous) GRXs, which catalyze the thiol-disulfide exchange between the GSH-GSSG and roGFP2 redox pairs [70]. Given that (i) until now, no peroxisomal GRXs have been identified [18], and (ii) roGFP2-derived probes do respond to a reduction in the peroxisomal redox state [15], a phenomenon impaired in ∆GSTK1 cells (Figures 5 and 6), we examined the recovery behavior of po-GRX1-roGFP2 [47] in AT4-insulted ∆GSTK1 cells. These studies showed that the po-roGFP2 recovery deficits observed in ∆GSTK1 cells can be recovered by a protein with glutathione-disulfide oxidoreductase activity (Figure 7), indicating a potential functional overlap between the activities of both enzymes. roGFP2 [47] in AT4-insulted ΔGSTK1 cells. These studies showed covery deficits observed in ΔGSTK1 cells can be recovered by a pr disulfide oxidoreductase activity (Figure 7), indicating a potential tween the activities of both enzymes. Figure 7. Fusion of GRX1 to the N-terminus of po-roGFP2 rescues the re glutathione redox sensor in AT-4-insulted ΔGSTK1 cells. CT or ∆GSTK1 were transfected with a plasmid encoding po-GRX1-roGFP2 and cultured three days, the basal oxidation states (b) of po-GRX1-roGFP2 were mea incubated for 10 min with 2 mM AT-4. After the removal of the oxidant, Figure 7. Fusion of GRX1 to the N-terminus of po-roGFP2 rescues the recovery phenotype of the glutathione redox sensor in AT-4-insulted ∆GSTK1 cells. CT or ∆GSTK1 (cl 1 or 2) HEK-293 cells were transfected with a plasmid encoding po-GRX1-roGFP2 and cultured in rMEMα. After two to three days, the basal oxidation states (b) of po-GRX1-roGFP2 were measured, and the cells were incubated for 10 min with 2 mM AT-4. After the removal of the oxidant, the response ratios of the sensors were monitored over time and normalized to the average value of the corresponding basal condition. The data points and vertical bars represent the mean and standard deviation of 30 individual measurements from 3 independent experiments, respectively. For each time point, the ∆GSTK1 data were statistically compared with those of the CT cells using the ordinary two-way ANOVA test with Tukey's multiple comparisons test, but no significant differences were observed.

The GSTK1-Mediated Recovery of Po-RoGFP2 Depends on Its Active-Site Serine Residue
Previous studies have identified serine-16 (S16) as a critical residue for GSTK1 activity [19,31,33]. To evaluate the importance of this amino acid residue in the thioldisulfide exchange between roGFP2 and the glutathione redox pair in peroxisomes, we complemented ∆GSTK1 cells with plasmids encoding wild-type (WT) or S16 to alanine (S16A)-mutated versions of the protein (C-terminally tagged or not with mCherry and a strong PTS1) and monitored the recovery rate of po-roGFP2 after brief exposure to AT-4. Given that the recovery rates of po-roGFP2 were significantly better in cells complemented with GSTK1 WT or po-GSTK1 WT -mCherry as compared to GSTK1 S16A or po-GSTK1 S16A -mCherry ( Figure 8A,C), the recovery phenotype is clearly regulated by the GSTK1 s enzymatic activity. Note that mutating the active site serine to alanine had no effect on the expression levels ( Figure 8B) and peroxisomal localization ( Figure 8D) of the GSTK1 proteins examined.

Oxidative Insults Do Not Induce Intra-or Intermolecular Protein Disulfide Bonds in GSTK1
Given that (i) human GSTK1 has two cysteine residues (C27 and C176), and (ii) the latter cysteine has been reported to be a glutathionylation target in the mouse orthologue [71], we also investigated if GSTK1 can form intra-or intermolecular disulfide bonds in cells exposed to po-H 2 O 2 or AT-4. However, neither internal nor external oxidative insults caused noticeable disulfide bond-induced conformational changes, as evaluated by a redox electrophoretic mobility shift assay (Figure 9). Note that SQSTM1, a protein reported to undergo disulfide-linked oligomerization in response to oxidative insults [48,72], was included as a positive control. mutated versions of the protein (C-terminally tagged or not with mCherry and a strong PTS1) and monitored the recovery rate of po-roGFP2 after brief exposure to AT-4. Given that the recovery rates of po-roGFP2 were significantly better in cells complemented with GSTK1WT or po-GSTK1WT-mCherry as compared to GSTK1S16A or po-GSTK1S16A-mCherry ( Figure 8A,C), the recovery phenotype is clearly regulated by the GSTK1′s enzymatic activity. Note that mutating the active site serine to alanine had no effect on the expression levels ( Figure 8B) and peroxisomal localization ( Figure 8D) of the GSTK1 proteins examined.

Oxidative Insults Do Not Induce Intra-or Intermolecular Protein Disulfide Bonds in GSTK1
Given that (i) human GSTK1 has two cysteine residues (C27 and C176), and (ii) the latter cysteine has been reported to be a glutathionylation target in the mouse orthologue [71], we also investigated if GSTK1 can form intra-or intermolecular disulfide bonds in cells exposed to po-H2O2 or AT-4. However, neither internal nor external oxidative insults caused noticeable disulfide bond-induced conformational changes, as evaluated by a redox electrophoretic mobility shift assay (Figure 9). Note that SQSTM1, a protein reported to undergo disulfide-linked oligomerization in response to oxidative insults [48,72], was included as a positive control. Figure 9. Po-H2O2 and AT-4 do not induce protein disulfide bond formation in GSTK1. CT or ΔGSTK1 (cl 1) HEK-293 cells were (A,B) cultured for three days in rMEMα containing 1 μg/mL doxycycline and 500 nM Shield1 (to express and stabilize DD-DAO, respectively), chased for one day in the same medium lacking doxycycline and Shield1 (to degrade the residual cytosolic pool of DD-DAO that has not yet been imported into peroxisomes), and incubated for 1 h in DPBS containing 10 mM 3-AT and 10 mM L-Ala or D-Ala, or (C,D) incubated for 10 min in rMEMα supplemented or not (b, basal state) with 500 μM AT-4 and chased in the same medium without AT-4 for 0, 1, and 3 h. Next, the cells were treated with 10 mM NEM (to block free thiol groups), and total cell lysates were processed for SDS-PAGE under nonreducing conditions and immunoblot analysis with antibodies specific for GSTK1 and SQSTM1. The migration points of relevant molecular mass markers (expressed in kDa) are shown on the left. The arrows and arrowheads mark nonoxidatively and oxidatively modified immunoreactive protein bands, respectively. The asterisks mark bands of unknown nature.

Discussion
Proteins inside the peroxisome lumen must be constantly safeguarded against the potentially damaging effects of H2O2 [73], and one potential protective mechanism may include the glutathione redox system [7]. However, although previous studies with genetically encoded redox sensors have demonstrated that the peroxisomal glutathione pool is maintained in a reduced state [15,74], virtually nothing is known about the role and metabolism of this low molecular weight antioxidant inside peroxisomes. In this study, we set out to examine the function of the peroxisomal pool of GSTK1, the only glutathione metabolism-related enzyme discovered in the mammalian peroxisomal proteome to date [17,18]. (cl 1) HEK-293 cells were (A,B) cultured for three days in rMEMα containing 1 µg/mL doxycycline and 500 nM Shield1 (to express and stabilize DD-DAO, respectively), chased for one day in the same medium lacking doxycycline and Shield1 (to degrade the residual cytosolic pool of DD-DAO that has not yet been imported into peroxisomes), and incubated for 1 h in DPBS containing 10 mM 3-AT and 10 mM L-Ala or D-Ala, or (C,D) incubated for 10 min in rMEMα supplemented or not (b, basal state) with 500 µM AT-4 and chased in the same medium without AT-4 for 0, 1, and 3 h. Next, the cells were treated with 10 mM NEM (to block free thiol groups), and total cell lysates were processed for SDS-PAGE under nonreducing conditions and immunoblot analysis with antibodies specific for GSTK1 and SQSTM1. The migration points of relevant molecular mass markers (expressed in kDa) are shown on the left. The arrows and arrowheads mark nonoxidatively and oxidatively modified immunoreactive protein bands, respectively. The asterisks mark bands of unknown nature.

Discussion
Proteins inside the peroxisome lumen must be constantly safeguarded against the potentially damaging effects of H 2 O 2 [73], and one potential protective mechanism may include the glutathione redox system [7]. However, although previous studies with genetically encoded redox sensors have demonstrated that the peroxisomal glutathione pool is maintained in a reduced state [15,74], virtually nothing is known about the role and metabolism of this low molecular weight antioxidant inside peroxisomes. In this study, we set out to examine the function of the peroxisomal pool of GSTK1, the only glutathione metabolism-related enzyme discovered in the mammalian peroxisomal proteome to date [17,18].
We first confirmed that peroxisomes in HEK-293 cells do indeed contain GSTK1 (Figure 1), an observation in line with the subcellular staining patterns visible in images generated in the context of the Human Protein Atlas project (https://www.proteinatlas. org/ENSG00000197448-GSTK1/subcellular; accessed on 26 April 2023).
Next, we examined how GSTK1 inactivation affected the expression levels of a specific set of peroxisomal and mitochondrial proteins (Figures 2, 3A and S1), mitochondrial respiration ( Figure 3B), and peroxisomal and cytosolic redox levels, but no significant differences could be identified in cells cultured under normal conditions ( Figure 4). However, this may not come as a surprise considering that GSTK1 −/− mice had normal amounts of reduced and total glutathione in their livers and kidneys [34]. As others have shown that loss of GSTK1 exacerbates oxidative stress and tubular apoptosis in the kidneys from streptozotocin-induced diabetic mice [37], we also investigated its potential role in peroxisomal oxidative stress responses. From these studies, it is clear that GSTK1 enhances the thiol-disulfide exchange between po-roGFP2 and the glutathione redox pair following oxidative insult withdrawal (Figures 5 and 6).
As the po-roGFP2 recovery phenotype cannot be observed in ∆GSTK1 cells expressing po-GRX1-roGFP2 (Figure 7), our findings indicate that the peroxisomal pool of GSTK1 possesses glutaredoxin-like activity. Given that (i) GRXs catalyze disulfide bond reduction via a dithiol (CXXC-requiring) or monothiol (CXXS-dependent) mechanism [5,75,76], and (ii) the active center of GSTK1 (S 16 XXS 19 ; Ser 16 ) is lacking a cysteine [77], this may appear counterintuitive. However, here it is crucial to note that also other proteins with thiolindependent GRX activity have been identified. Examples include the bacterial disulfidebond oxidoreductase YfcG and the yeast transcriptional regulator Ure2. YfcG exhibits very robust glutathione (GSH)-dependent disulfide-bond reductase activity toward the model substrate 2-hydroxyethyl disulfide [78] and Ure2 is a multifunctional protein with GRX activity toward small molecule disulfides (or GSH mixed disulfide bonds) and protein disulfides [79]. In addition, it has been proposed that these thiol-independent GRX-like activities are physiologically important under conditions of acute oxidative stress, where active site cysteines may be rapidly oxidized [79].
In summary, this study provides evidence that the peroxisomal pool of GSTK1 possesses GRX-like activity. This implies that, after the GST-Omega class [80], the GST-Kappa class becomes the second subfamily of human GSTs with such activity. It is noteworthy that the true endogenous substrates and mode of action of GSTK1 have yet to be identified. In the BioGRID database [81], only two peroxisomal proteins (alkyldihydroxyacetonephosphate synthase and CAT) have been reported to be physical interactors of GSTK1, a fact that may be explained by the transient nature of the GSTK1-protein substrate interactions or the absence of suitable interaction conditions. Another intriguing but unresolved question is why the recovery kinetics of po-roGFP2 and po-GRX1-roGFP2 are slower than that of c-roGFP2. However, given that little is known about how glutathione is transported across the peroxisomal membrane and how GSSG is regenerated to GSH within the peroxisome lumen, this is a challenging question. Nonetheless, given that (i) peroxisomes appear to be closed structures under in vivo conditions [82,83], and (ii) the transport of glutathione across the peroxisomal membrane requires a specific transporter in yeast [84], it is tempting to speculate that the delayed recovery phenotype may be due to a reduced replenishment of the peroxisomal GSH content following a strong oxidative insult.

Conclusions
In this study, we investigated the role of GSTK1 in peroxisomal redox metabolism in HEK-293 cells. We provide evidence that, while the protein appears to be unnecessary for peroxisomal redox homeostasis under basal conditions, the peroxisomal pool of GSTK1 possesses a glutaredoxin-like activity towards the glutathione redox sensor roGFP2. The discovery of this new biochemical function of GSTK1 opens up new avenues to advance our understanding of how changes in its expression contribute to aging and disease.