Research ArticleHEALTH AND MEDICINE

Glutathione dynamics determine the therapeutic efficacy of mesenchymal stem cells for graft-versus-host disease via CREB1-NRF2 pathway

See allHide authors and affiliations

Science Advances  15 Apr 2020:
Vol. 6, no. 16, eaba1334
DOI: 10.1126/sciadv.aba1334

Abstract

Glutathione (GSH), the most abundant nonprotein thiol functioning as an antioxidant, plays critical roles in maintaining the core functions of mesenchymal stem cells (MSCs), which are used as a cellular immunotherapy for graft-versus-host disease (GVHD). However, the role of GSH dynamics in MSCs remains elusive. Genome-wide gene expression profiling and high-throughput live-cell imaging assays revealed that CREB1 enforced the GSH-recovering capacity (GRC) of MSCs through NRF2 by directly up-regulating NRF2 target genes responsible for GSH synthesis and redox cycling. MSCs with enhanced GSH levels and GRC mediated by CREB1-NRF2 have improved self-renewal, migratory, anti-inflammatory, and T cell suppression capacities. Administration of MSCs overexpressing CREB1-NRF2 target genes alleviated GVHD in a humanized mouse model, resulting in improved survival, decreased weight loss, and reduced histopathologic damages in GVHD target organs. Collectively, these findings demonstrate the molecular and functional importance of the CREB1-NRF2 pathway in maintaining MSC GSH dynamics, determining therapeutic outcomes for GVHD treatment.

INTRODUCTION

Allogeneic hematopoietic stem cell transplantation (allo-HSCT) is a therapy for patients with leukemia and other disorders, such as lymphoma, aplastic anemia, immune deficiency disorders, and some solid tumor cancers, which has been used successfully since 1968 (1). Despite advances that have improved survival after allo-HSCT, graft-versus-host disease (GVHD) causes substantial posttransplant morbidity and mortality (2, 3). GVHD is a multiorgan disorder characterized by immune dysregulation with donor T cell activation, culminating in a cytokine storm. Activated effector T cells, together with other immune cells, cause end-organ injury by releasing inflammatory cytokines, including interleukin-1 (IL-1), IL-2, tumor necrosis factor–α (TNF-α), and interferon-γ (IFN-γ) (1, 4). Standard prophylaxis and treatment for GVHD are immunosuppressive drugs with profound inhibitory effects on allogeneic T cells, but these interventions increase the risk of potentially life-threatening infections and tumor relapse (5, 6). About 50% of patients do not respond to first-line treatment, and patients with steroid-refractory acute GVHD have a poor prognosis, with long-term morality rates approaching 90% (1). Therefore, there is an urgent unmet clinical need to develop innovative therapies for GVHD.

A number of immune modulatory cells have been identified as potential targets for cellular immunotherapies in GVHD (7). Clinical trials have demonstrated that infusion of ex vivo– or in vivo–derived regulatory T (Treg) cells substantially reduces the incidence of acute GVHD (8, 9). Clinical trials have also investigated the efficacy of natural killer (NK) cell infusion and in vivo NK activation to promote the deletion of alloreactive T cells and the induction of Treg cells (10). In addition, the therapeutic efficacies of FoxP3negIL-10+ (FoxP3-negative) regulatory T and NK T cells, as well as various myeloid suppressor populations of hematopoietic (myeloid-derived suppressor cells) and stromal origin [mesenchymal stem/stromal cells (MSCs)], have been evaluated in preclinical and clinical settings (7, 11).

In particular, cell therapy based on MSCs derived from umbilical and adult tissues is a promising strategy for treatment-refractory GVHD (1214), as MSCs have a wide range of immunosuppressive and immunomodulatory properties on innate and adaptive immune cells (15). Recently, we reported that human MSCs manufactured using a process termed small cells primed with hypoxia and calcium ions (SHC) exhibited enhanced therapeutic potency in a humanized mouse model of GVHD (16). However, the infusion of MSCs for GVHD treatment yielded seemingly conflicting results in preclinical and clinical trials (17). These conflicting data strongly point to the need for improved understanding of cell-intrinsic and cell-extrinsic factors that dictate the therapeutic efficacy of MSCs in GVHD.

Recently, we demonstrated that glutathione (GSH), an abundant cellular thiol and major determinant of cellular redox equilibrium, is essential to maintain the stemness and migration capacities of MSCs, determining their therapeutic efficiency in asthma (18). Furthermore, perturbation of GSH homeostasis in HSCT recipients contributes to GVHD pathogenesis, particularly in the initiation stage (19).

In the present study, we used genome-wide gene expression profiling and high-throughput live-cell GSH monitoring assays to identify that cyclic adenosine monophosphate (cAMP) response element–binding protein 1 (CREB1) was crucial to the maintenance of GSH dynamics in MSCs, mediated by nuclear factor erythroid 2 like 2 (NRF2), a well-known master regulator of redox homeostasis (20). NRF2 regulated GSH homeostasis by transcribing target genes involved in GSH synthesis and recycling. Last, the CREB1-NRF2 signaling cascade enforced the functionality of MSCs, which determined the therapeutic potency of MSCs in a humanized mouse model of GVHD. Together, these findings suggest that supporting proper GSH dynamics in MSCs through the CREB1-NRF2 pathway, together with promoting GSH synthesis, could provide therapeutic advantages for treatment of allogeneic conflicts, including GVHD.

RESULTS

CREB1 activity was up-regulated in MSCs with high GSH levels

We recently reported the first reversible chemical probe for GSH, FreSHtracer (fluorescent real-time thiol tracer), which enables real-time monitoring of GSH levels in living cells due to its fast-reversible and ratiometric reaction with GSH (18, 21, 22). In the present study, we sorted human MSCs derived from embryonic stem cells (ESCs) (hES-MSCs) (23) by flow cytometry and divided them into GSHHigh and GSHLow subpopulations based on the fluorescence reaction of FreSHtracer with GSH to elucidate the molecular nature of redox homeostasis in MSCs (Fig. 1A). As reported, the ratio of fluorescence intensity at 510 nm to that at 580 nm [F510/F580 ratio; fluorescence ratio (FR)] was correlated with GSH concentration (18). Transcriptomic analysis of GSHHigh and GSHLow hES-MSCs identified 149 differentially expressed genes (70 up-regulated and 79 down-regulated in GSHHigh versus GSHLow hES-MSCs) (fig. S1A). Gene Ontology (GO) analysis using the MetaCore pathway method (data file S1) identified that genes involved in pathways and processes related to hematopoietic stem cell mobilization, cell adhesion, and the inflammatory response were differentially expressed in GSHHigh and GSHLow cells (Fig. 1B and fig. S1B). MetaCore gene network analysis identified that CREB1-associated (Fig. 1C) or ubiquitin- and RelA-associated (fig. S1C) gene networks were characteristically represented in GSHHigh hES-MSCs.

Fig. 1 CREB1 was activated in MSCs with high GSH levels.

(A) Overview for transcriptome analysis of hES-MSCs that were fluorescence-activated cell sorting (FACS)–sorted on the basis of GSH level using FreSHtracer. (B and C) Ten most highly represented pathway maps (B) and CREB1-associated gene networks (C) in MetaCore transcriptome analysis of GSHHigh and GSHLow hES-MSCs. (D) Enrichment plots of GSEA analysis, focusing on gene sets of CREB1-binding targets that were enriched in GSHHigh hES-MSCs. (E and F) Real-time quantitative polymerase chain reaction (qPCR) assay (n = 4) (E) for CREB1-associated genes or Western blot analysis (n = 3) (F) of total (t-) or phosphorylated (p-) proteins in naive or FACS-sorted GSHHigh and GSHLow hES-MSCs. Quantification data for transcript and protein levels are represented as ratios relative to the value in the GSHLow and GSHHigh cells, respectively. The expression level of indicated proteins was normalized with that of β-ACTIN, which was used as a loading control [(F), right]. *Nonspecific band. (G and H) Increased activity of a luciferase reporter with eight tandem CREB1 response elements (8× CRE-Luc) (n = 4) (G) and CREB1 upstream activators (n = 4) (H), including cAMP levels and protein kinase A (PKA) activity, in GSHHigh and GSHLow hES-MSCs. All quantification data are presented as means ± SEM. Statistical analyses were performed using one-way (E) or two-way analysis of variance (ANOVA) (F) with Bonferroni post hoc tests and nonparametric Mann-Whitney U test (G and H). *P < 0.05, **P < 0.01, and ***P < 0.001. ns, not significant.

To further investigate the role of CREB1 in the regulation of GSH levels, we curated the gene sets (data file S1) by collecting CREB1-binding target genes from a genome-wide study (24) and then compared the transcriptome of hES-MSCs with different GSH levels based on these curated CREB1 targets. Gene set enrichment analysis (GSEA) revealed that the transcriptomes of GSHHigh hES-MSC samples were positively enriched with CREB1 targets involved in DNA metabolism, replication, and repair processes (Fig. 1D and fig. S1D). In particular, the cAMP-dependent transcription factor ATF2 (activating transcription factor 2) and the transcription factor complex activating protein 1, including JUN, JUNB, and FOS like-1 (FOSL1), were identified as differentially regulated components of the CREB1 gene networks (Fig. 1C). Transcript levels of these genes (Fig. 1E) and phosphorylated active proteins (Fig. 1F and fig. S1E) were increased in GSHHigh hES-MSCs. Accordingly, phosphorylated CREB1 protein (Fig. 1F) and the activity of a CREB1 reporter containing eight tandem cAMP response elements (CREs) (Fig. 1G and fig. S1F) (25) were increased in GSHHigh hES-MSCs. Consistent with these results, cAMP levels and cAMP-dependent protein kinase A (PKA) activity, upstream activators of CREB1, were increased in GSHHigh hES-MSCs (Fig. 1H), demonstrating CREB1 activation in hES-MSCs with high GSH levels.

Real-time measurement of GSH-recovering capacity in living cells

To investigate the biological relevance of the CREB1 gene network, we advanced the FreSHtracer-based GSH assay to enable real-time tracing of GSH change in every single cell using an Operetta high-content imaging analysis system. By quantifying the pattern of GSH dynamics for approximately 1 hour after exposure to oxidative insults such as diamide, a thiol-specific oxidant, we evaluated and quantified MSC GSH-recovering capacity (GRC) under different conditions (Fig. 2A). When hES-MSCs were treated with diamide after equilibration with 2 μM FreSHtracer for 2 hours, the FR rapidly decreased after diamide addition and gradually recovered to the untreated level in a dose-dependent manner (fig. S2, A and B). From these data, we identified 200 μM diamide as the optimal dosage for further GRC assays. As expected, the GRC in hES-MSCs was severely impaired by pretreatment with buthionine sulfoximine, an inhibitor of GSH synthetase (Fig. 2B), and 3-bis(2-chloroethyl)-1-nitrosourea, an inhibitor of GSH reductase (fig. S2, C and D). However, GSH ethyl ester (GSH-EE), a cell-permeable GSH, increased GSH levels (Fig. 2C), indicating that the GRC assay provided an accurate assessment of cellular capacity for GSH homeostasis.

Fig. 2 High-throughput real-time measurement of GSH dynamics in living cells.

(A) Experimental overview for measuring GRC and basal GSH level using FreSHtracer in response to exposure to 200 μM diamide (arrow). GSH dynamics index (GI) of each sample was quantified on the basis of both initial FR (for baseline of total GSH) and slope after diamide treatment (for GRC). (B and C) F510/F580 FR plot (left) and representative images of F510 and F580 fluorescence (right) in hES-MSCs treated with 160 μM buthionine sulfoximine (BSO) (B) for 24 hours or 1 mM GSH-EE (C) for 4 hours. GI values are depicted below each FR plot. (D to G) FR plots of hES-MSCs carrying short hairpin RNA (shRNA) for cJUN (D), JUNB (E), FOSL1 (F), or ATF2 (G). Two independent shRNAs were used. (H) FR plots of human UC-MSCs cultivated using the SHC procedure (SHC-MSCs). Naive cultured cells were used as a control. All FR plots are presented as means ± SEM [n = 3 for (B) and (C); n = 6 for (D) to (H)]. ***P < 0.001 (two-way ANOVA). (I and J) Flow cytometry assay for measuring cell size (I) and GSEA enrichment plot (J) for a set of genes up-regulated in SHC-MSCs in GSHHigh and GSHLow hES-MSCs. Data are presented as ratios relative to the value in the GSHLow cells and are expressed as means ± SEM (n = 6) (I). **P < 0.01 (nonparametric Mann-Whitney U test). Values for the GI for each plot are presented in data file S3.

Next, we used the high-throughput GRC assay to investigate the roles of the CREB1 target genes JUN, JUNB, FOSL1, and ATF2, which were first identified as being transcriptionally up-regulated in GSHHigh hES-MSCs and later confirmed to be more active at the protein level in GSHHigh hES-MSCs (Fig. 1F and fig. S1E). Silencing of each gene severely decreased the basal level of GSH of hES-MSCs and impaired GRC following diamide treatment (Fig. 2, D to G). We also examined the utility of the GRC assay to evaluate the primitiveness of MSCs produced by the SHC process, which supported the therapeutic efficacy in GVHD for MSCs generated using this approach (16). Human umbilical cord (UC)–derived MSCs (UC-MSCs) cultivated using the SHC procedure (SHC-MSCs) had higher basal GSH levels and GRC activity than naively cultured (naive) UC-MSCs (Fig. 2H). Consistent with this result, fluorescence-activated cell sorting (FACS)–sorted GSHHigh hES-MSCs were smaller in size than GSHLow cells (Fig. 2I), and the transcriptomes of GSHHigh hES-MSCs were characteristically enriched with a gene set composed of genes up-regulated in SHC-MSCs (Fig. 2J). GSHHigh hES-MSCs exhibited increased expression of a subset of biomarkers for SHC-MSCs (fig. S2E) (16). As a result of these studies, we developed a powerful tool to evaluate and quantify GSH dynamics in living cells by combining a reversible chemical GSH probe with a high-content live-cell imaging assay.

CREB1 modulated redox homeostasis and immune modulation activity of MSCs

Using the GRC method, we examined the functional role of CREB1 in the modulation of GSH dynamics in MSCs. Silencing CREB1 severely decreased basal GSH level and GRC in hES-MSCs (Fig. 3A), and similarly defective GSH dynamics were observed with overexpression of the dominant-negative (DN) inhibitor of CREB1 (Fig. 3B), which prevents DNA binding of B-ZIP proteins in a dimerization domain-dependent manner (26). CREB1-DN overexpressing cells had only minimal CREB1 luciferase reporter activity (fig. S3A). Transfection with CREB1 short hairpin RNA (shRNA) (shCREB1) or expression of CREB1-DN depleted GSH levels and impaired GRC in UC-MSCs, suggesting a critical role for CREB in the regulation of GSH dynamics in this context (fig. S3, B and C). Furthermore, pretreatment with forskolin (FSK), a CREB1 activator, increased basal GSH level and GRC in both MSC types (Fig. 3C and fig. S3D), supporting the crucial role of CREB1 in maintaining MSC GSH dynamics.

Fig. 3 FSK priming enhanced hES-MSC GSH dynamics and immunomodulatory functions.

(A to C) FR plots of hES-MSCs carrying CREB1 shRNA (shCREB1) (A) or DN inhibitor (CREB1-DN) (B) and FR plots of cells treated with 2 μM FSK for 12 hours (C). FR plots are presented as means ± SEM (n = 6). ***P < 0.001 (two-way ANOVA). GI values are depicted below each FR plot. (D to H) CFU-F (n = 6) (D); chemotaxis (n = 7) (E) to PDGF-AA (10 ng/ml) in the absence or presence of STI571 (0.5 μg/ml), a PDGFR inhibitor; chemotaxis in the response to (n = 7) (F) SDF1α (150 ng/ml) (left) or SDF1β (right) in the absence or presence of AMD3100 (10 μM), a CXCR4 antagonist; in vitro Matrigel tube formation (n = 7) (G); and immune modulation (n = 4) (H) assays in FSK (2 μM for 12 hours)–primed hES-MSCs. Representative results for each assay are presented in the left [×200 magnification; scale bars, 100 μm; (E and F)] or bottom panel [×40 magnification; scale bars, 200 μm; (G)] of quantitative data. For the Matrigel tube formation assay, CM was prepared from FSK-primed or NT cells. For negative and positive controls, saline and recombinant human vascular endothelial growth factor A (VEGF-A) (50 ng/ml) were used, respectively. The immune modulatory capacity (H) of MSCs were evaluated by suppression of T cell proliferation (CFSE/CD3+) in human PBMCs stimulated by PHA. Data are presented as ratios relative to the value in the NT cells and are expressed as means ± SEM. Statistical analyses were performed using nonparametric Mann-Whitney U test (D), one-way ANOVA (E to G), or two-way ANOVA (H). **P < 0.01 and ***P < 0.001 relative to NT group. ##P < 0.01 and ###P < 0.001.

In this regard, hES-MSCs pretreated with 2 μM FSK for 12 hours (FSK-hES-MSCs) exhibited significantly higher colony-forming unit fibroblast (CFU-F) activity (Fig. 3D), an indicator of clonogenic progenitor cells, than nontreated (NT) control cells. FSK priming minimally affected the multipotency of hES-MSCs (fig. S3E). FSK-hES-MSCs exhibited greater chemoattraction to platelet-derived growth factor (PDGF), which was significantly blocked by a PDGF receptor (PDGFR) inhibitor, STI571 (Fig. 3E). In addition, FSK priming enhanced the chemoattraction to stromal-derived factor 1 (SDF1), which is a major chemokine modulating the mobilization and homing of adult stem cells (27). The enhanced responsiveness to SDF1 chemokines in FSK-hES-MSCs was significantly impaired by an SDF1 receptor CXCR4 antagonist, AMD3100, indicating the dependence on CXCR4 signaling activity (Fig. 3F). The beneficial effects of FSK on chemotaxis in response to PDGF-AA and SDF1 were also observed in human UC-MSCs pretreated with 2 μM FSK for 12 hours (fig. S3, F to H). Furthermore, conditioned medium (CM) from FSK-hES-MSCs improved the angiogenic potency of endothelial cells in a Matrigel tube formation assay (Fig. 3G). Prior reports demonstrated that MSCs suppress in vitro proliferation of human peripheral blood mononuclear cells (PBMCs) by stimulating phytohemagglutinin (PHA) (16). In our study, FSK-hES-MSCs repressed mitogen-driven proliferation of CD3+ T cells in PBMCs (Fig. 3H).

We further examined the effect of FSK on human bone marrow (BM)–derived MSCs (BM-MSCs). Consistent with the above data, BM-MSCs pretreated with 2 μM FSK for 12 hours (FSK-BM-MSCs) showed increased intracellular GSH level and GRC (fig. S4A). In addition, FSK priming stimulated the functionality of BM-MSCs with respect to CFU-F activity, chemotaxis toward PDGF-AA, or SDF1 and immune modulatory capacity (fig. S4, B to F). Together, these data demonstrate that CREB1 plays a crucial role in maintaining the core functions of MSCs derived from different sources including human ESCs, UC, and BM. Consistent with these findings, silencing JUN, another target validated from gene expression and GRC assays, severely impaired the core functions of MSCs, including multipotency, proliferation, CFU-F, and chemotaxis toward PDGF (fig. S5), further supporting the involvement of the CREB1 gene network in MSC function.

NRF2 was a critical effector of CREB1-mediated GSH dynamics

Next, we investigated the function of a downstream CREB effector in MSCs. CREB1 has been reported to bind the NRF2 promoter and regulate NRF2 transcription via the cAMP signaling pathway (24, 28, 29). Because NRF2 is a master regulator of antioxidant responses, including GSH pathway genes, we determined whether NRF2 played a role in CREB1-mediated GSH dynamics. FSK priming increased NRF2 transcript levels (Fig. 4A), leading to a dose-dependent increase of NRF2 protein levels (Fig. 4B) in hES-MSCs. The majority of NRF2 protein in FSK-hES-MSCs was localized to the nucleus, indicating activation of the NRF2 pathway (Fig. 4C). The induction and activation of NRF2 by FSK priming were also detected in UC- and BM-MSCs (fig. S4, G and H). CREB1 silencing and CREB1-DN expression both prevented FSK-mediated up-regulation and nuclear translocation of NRF2 protein, suggesting that the effects of FSK were CREB1 dependent (Fig. 4, B and C). hES-MSCs carrying NRF2 shRNA (shNRF2) exhibited remarkable defects in basal GSH level and GRC activity (Fig. 4D), emphasizing the key role of NRF2 in MSC GSH homeostasis.

Fig. 4 NRF2 was a critical effector of CREB1-mediated GSH dynamics.

(A to C) Real-time qPCR (n = 6) (A), Western blot (B), and immunostaining (C) assays for NRF2 expression (green) in FSK-primed hES-MSCs in the absence or presence of CREB1 silencing (shCREB1) or expression of CREB1-DN inhibitor. The expression level of the indicated proteins was normalized with that of β-ACTIN, which was used as a loading control (n = 3) [(B), right]. (C) Representative images for confocal microscopy are shown with ×1000 magnification. Scale bars, 10 μm. Nuclei were stained with DAPI (4′,6-diamidino-2-phenylindole) (blue). (D) FR plots of hES-MSCs carrying control (shCTR) or two independent NRF2 shRNA (shNRF2) constructs (n = 6). (E and F) Real-time qPCR assays evaluating the induction (E) or recruitment of NRF2 (F) in the promoter regions of the indicated NRF2 target genes by FSK priming (n = 4). (G and H) Real-time qPCR assay of NRF2 target genes in FSK-primed hES-MSCs in the absence or presence of NRF2 (G) or CREB1 (H) silencing (n = 4). (I and J) Real-time qPCR (n = 4) (I) and Western blot (J) assays for the expression of CREB1-NRF2–dependent GSH synthesis (GCLM and GCLC) and redox cycling (GSR and PRDX1) genes. The expression level of the indicated proteins was normalized with that of β-ACTIN (n = 3) [(J), right]. *Nonspecific band. (K) FR plot in NRF2 silenced (shNRF2_#1) hES-MSCs, which were rescued by ectopic expression of the indicated CREB1-NRF2 target genes (n = 6). GI values are depicted in the right panel of the FR plot. Data are presented as ratios relative to the value in the control groups and are expressed as means ± SEM. *P < 0.05, **P < 0.01, and ***P < 0.001 relative to NT or control group. Statistical analyses were performed using nonparametric Mann-Whitney U test (A), one-way ANOVA (E and F), or two-way ANOVA (B, D, and G to K) with Bonferroni post hoc tests.

NRF2 directly regulates transcription of several genes modulating redox homeostasis (30). FSK treatment up-regulated a subset of NRF2 redox homeostasis targets, including GCLC, GCLM, GSR, PRDX1, PRDX2, PRDX3, MGST2, and ALAS1 in both hES-MSCs (Fig. 4E) and UC-MSCs (fig. S4I). A chromatin immunoprecipitation (ChIP) assay revealed that the promoters of these genes were bound by NRF2 in hES-MSCs (fig. S6A) and that FSK priming further stimulated NRF2 binding in these loci (Fig. 4F). Among these genes of interest, FSK induction of genes involved in GSH synthesis (GCLM and GCLC) and redox cycling (GSR and PRDX1) was dependent on the expression of NRF2 (Fig. 4G) and was impaired by the expression of both shCREB1 and CREB1-DN (Fig. 4H). Furthermore, NRF2 silencing in hES-MSCs decreased expression of these genes at both the transcript (Fig. 4I) and protein (Fig. 4J) levels. These CREB1-NRF2 targets were up-regulated in FSK-BM-MSCs (fig. S4J). On the basis of these findings, we selected the CREB1-NRF2 targets GCLM, GCLC, GSR, and PRDX1 for further evaluation.

Ectopic expression of these CREB1-NRF2 target genes enforced intracellular GSH level and GRC (fig. S6, B to D), which were severely impaired by silencing of each gene (fig. S6, E and F), suggesting that these genes determined the redox homeostasis of hES-MSCs. Defective GRC in NRF2-silenced hES-MSCs was sufficiently rescued by ectopic expression of each CREB1-NRF2 target (Fig. 4K). Together, these data demonstrate that CREB1-NRF2 cascade modulates MSC GSH dynamics by directly targeting genes responsible for GSH synthesis and redox cycling.

Overexpression of CREB1-NRF2 targets induced primitiveness in MSCs

We next explored the role of the CREB1-NRF2 targets in MSC functionality. hES-MSCs overexpressing each target gene, including GCLM, GCLC, PRDX1, and GSR, exhibited significantly higher CFU-F (fig. S6, G and H) and PDGF-responsive chemotaxis (fig. S6, I and J) capacities than control MSCs expressing an empty construct, although cell proliferation was not significantly affected in overexpressing cells (fig. S6K).

We observed similar roles for the CREB1-NRF2 targets in MSCs of a different tissue origin. When we silenced or overexpressed GCLM, GCLC, PRDX1, and GSR in UC-MSCs (fig. S7, A to C), the intracellular GSH level and GRC were severely impaired by silencing of each target gene (Fig. 5A) and reinforced by overexpression of each target gene (Fig. 5B). Similar to hES-MSCs, ectopic expression of each of these genes rescued NRF2 silencing–induced defects in total GSH level and GRC (Fig. 5C). In this regard, ectopic expression of GCLM, GCLC, PRDX1, and GSR made UC-MSCs resistant to oxidative stress–induced cell death (fig. S7D).

Fig. 5 Overexpression of CREB1-NRF2–dependent GSH modulators enhanced MSC immunomodulatory functions.

(A to C) FR plots of human UC-MSCs with silencing (A) or ectopic expression (B) of GCLM, GCLC, PRDX1, and GSR. Defective GSH dynamics in NRF2-silenced (shNRF2_#1) UC-MSCs were restored by ectopic expression of each CREB1-NRF2 target gene (C). FR plots are presented as means ± SEM (n = 3). GI values are depicted in the right panel of each FR plot. (D and E) Suppression of T cell proliferation (CFSE/CD3+) in PHA-stimulated PBMCs by coculture with human UC-MSCs overexpressing GCLM, GCLC, PRDX1, or GSR. (F and G) Representative FACS cytograms (F) and quantification results (G) for the proliferation of human T cells in the MLR assay. Single and allogeneic PBMCs (PA + PB) were cocultured with human UC-MSCs carrying an empty control vector or a CREB1-NRF2 target gene. Percentage of proliferating T cells (CFSE/CD3+) is expressed as mean ± SEM (n = 4). (H) Anti-inflammatory assays using CM prepared from the indicated cells or IMR90, a normal primary fibroblast for control. Expression levels of Tnf-α transcript are represented as ratios relative to empty control and are quantified and presented as means ± SEM (n = 4). Statistical analyses were performed using one-way [(G) and (H)] or two-way [(A) to (E)] ANOVA with Bonferroni post hoc tests. *P < 0.05, **P < 0.01, and ***P < 0.001 relative to empty control groups. ###P < 0.001 relative to control groups for each assay [(E), PBMC treated with only PHA; (G), PA + PB; (H), IMR90].

In functional assays, UC-MSCs overexpressing each CREB1-NRF2 target exhibited significantly increased CFU-F, proliferation, and PDGF chemotaxis capacities (fig. S7, E to J). Similar to FSK priming, overexpression of GCLM, GCLC, PRDX1, and GSR stimulated the immune modulatory activities of UC-MSCs, repressing PHA-driven proliferation of CD3+ T cells in PBMCs (Fig. 5, D and E). We evaluated the immunomodulatory properties of these cells in response to allogeneic stimulation using an allogeneic mixed lymphocyte reaction (MLR) assay, revealing that UC-MSCs overexpressing target genes strongly repressed the proliferation of PBMCs upon allogeneic stimulation (Fig. 5, F and G). Furthermore, CM from UC-MSCs carrying each CREB1-NRF2 target gene strongly repressed TNF-α induction in MH-S cells, an alveolar macrophage cell line, following stimulation with lipopolysaccharide (Fig. 5H), indicating increased anti-inflammatory potency in these cells. Together, these in vitro functional assays demonstrated that GCLM, GCLC, PRDX1, and GSR are key downstream effectors for preserving redox homeostasis in MSCs, which promotes proliferative, self-renewal, migratory, proangiogenic, anti-inflammatory, and immunomodulatory capacities, as well as cell viability under oxidative microenvironments, which are crucial for their therapeutic potency.

Enhanced therapeutic efficacy of MSCs with high GSH and GRC

To examine the in vivo implication of our findings, we compared the therapeutic outcomes of human UC-MSCs carrying empty control–, GCLM-, or PRDX1-expressing constructs by injecting these cells into a humanized GVHD mouse model transplanted with human PBMCs (14, 16). At 8 weeks after transplantation, all mice transplanted with human PBMCs alone (GVHD) died, while the majority of GVHD mice transplanted with PBMCs and injected with UC-MSCs expressing empty control (60%), GCLM (90%), and PRDX1 (90%) survived (Fig. 6A). Mice in the untreated GVHD group showed severe weight loss, which was alleviated by UC-MSC treatment (Fig. 6B). GVHD mice injected with GCLM- and PRDX1-overexpressing UC-MSCs showed better survival and less weight loss than GVHD mice injected with empty control MSCs (Fig. 6B). Clinical scoring and histological analyses of representative GVHD target organs, including the small intestine, lung, kidney, and liver, indicated that injection of GCLM- and PRDX1-overexpressing UC-MSCs decreased immune cell infiltration and characteristic tissue injuries in GVHD mice, such as the sloughing of villi in the small intestine and fibroses in the lung and liver (Fig. 6, C and D). Accordingly, tissue repair was improved in mice injected with GCLM- and PRDX1-overexpressing UC-MSCs compared with mice injected with empty control MSCs (Fig. 6D).

Fig. 6 High-GRC MSCs had enhanced therapeutic efficacy in GVHD.

(A to D) Survival rate (A), body weight (B), histological analysis of the indicated organs with hematoxylin and eosin staining (×200 magnification; scale bars, 100 μm) (C), and quantification of histological disease scoring (D) in humanized GVHD mice injected with 2.5 × 106 human PBMCs, followed by administration of 5 × 105 human UC-MSCs carrying empty control or GCLM-, GCLC-, PRDX1-, and GSR-expressing constructs. Phosphate-buffered saline (PBS) was injected in the sham group instead of PBMCs. Data are presented as means ± SEM from 10 independent animals in each group. *P < 0.05, **P < 0.01, and ***P < 0.001 relative to empty vector group. ###P < 0.001 relative to GVHD group. Statistical analyses were performed using one- or two-way ANOVA with Bonferroni post hoc test. (E and F) Representative fluorescent microscopic images (×400 magnification; scale bars, 100 μm) (E) and quantification (n = 10) (F) for immunostaining for human β2-microglobulin (green) in GVHD target tissues of humanized GVHD mice. Nuclei were stained with DAPI (blue). (G) Multiplex analysis of 28 human cytokines and chemokines in sera from the indicated NSG mice 5 weeks after infusion of PBMCs. Data are presented as dot plots of means ± SEM (n = 10). Details for quantification of other human cytokines are presented in fig. S8 (B to D). (H) Suppression of T cell proliferation in GVHD mice infused with human UC-MSCs ectopically expressing CREB1-NRF2 target genes. Representative flow cytometric data in the left panel show the percentage of human (h) and mouse (m) CD45+ cells (top) or hCD3- or hCD4-expressing human T cells (bottom) in peripheral blood of GVHD mice from the indicated groups. Quantitative data are presented as means ± SEM (n = 10). *P < 0.05, **P < 0.01, and ***P < 0.001 relative to empty vector group. #P < 0.05, ##P < 0.01, and ###P < 0.001 relative to GVHD group. Statistical analyses were performed using one- or two-way ANOVA with Bonferroni post hoc test.

Mechanistically, immunofluorescence staining of human β2-microglobin revealed that homing and engraftment of GCLM- and PRDX1-overexpressing UC-MSCs into GVHD target organs were improved relative to empty control cells (Fig. 6, E and F, and fig. S8A). Serum was harvested from NSG mice at the time of GVHD development, and cytokine levels were measured by multiplex cytokine analysis. Consistent with the anti-inflammatory and immunomodulatory activities observed in vitro (Fig. 5, D to H), human chemokines and inflammatory cytokines, including TNF-α, IFN-γ, granulocyte-macrophage colony-stimulating factor (GM-CSF), CXCL-10, and fibroblast growth factor (FGF) acidic, were more effectively reduced by injection of GCLM- and PRDX1-overexpressing UC-MSCs than by injection of empty control MSCs (Fig. 6G). All MSC therapy groups exhibited decreased levels of chemokines and cytokines, such as C-C motif chemokine ligand 2 (CCL2), CCL-4, IL-1RA, IL-6, IL-8, IL-10, and IL-23 (fig. S8, B to D), compared with untreated GVHD mice. Furthermore, flow cytometric analysis of peripheral blood revealed that injection of UC-MSCs decreased the number of human CD45+, CD3+, and CD3+CD4+ cells in GVHD mice and that injection of GCLM- and PRDX1-overexpressing UC-MSCs further decreased these human cells in GVHD mice (Fig. 6H). Together, these findings demonstrated that improved UC-MSC GSH dynamics modulated by GCLM and PRDX1 enhanced the therapeutic potency of MSCs in GVHD.

DISCUSSION

A number of preclinical and clinical trials investigating the therapeutic potential of MSCs are currently in progress worldwide. However, the clinical application of these cells is hampered by incomplete understanding of the cell-intrinsic and cell-extrinsic factors that determine or predict the therapeutic outcomes of MSCs. The present study demonstrated that redox homeostasis status, regulated by CREB1-NRF2–mediated GSH synthesis and redox cycling, is an important cellular factor in determining the core and therapeutic functions of MSCs for GVHD treatment in vivo (fig. S9).

High levels of reactive oxygen species (ROS) and resultant oxidative stress play an important role in exacerbating GVHD. Oxidative stress caused by preexisting disease conditions and the requirement for conditioning regimens induces tissue injuries, and oxidatively modified molecules function as ligands for innate immune cell activation, facilitating alloantigen presentation and donor T cell activation, which are required for GVHD initiation (31). The oxidative stress following allo-HSCT could also be exacerbated by impaired antioxidant defense capacity, including GSH (19). In GVHD experimental models, alloantigen-activated T cells exhibit higher cellular mitochondrial ROS generation and contain less GSH (32). Oxidation of the host GSH-regulated redox system and failure of compensatory GSH synthesis enzyme up-regulation precedes GVHD initiation, as established by an increase in circulating TNF-α (19). The importance of GSH in modulating inflammation has also been well-established in several immunological disorders (33). Collectively, these studies suggest an important role for GSH perturbation in immune cells and target tissue cells in GVHD pathogenesis, indicating that GSH can also function as a critical cell-extrinsic factor for immune modulatory cell therapies targeting GVHD. Moreover, UC-MSCs with ectopic expression of each CREB1-NRF2 target were more resistant to oxidative stress–induced cell death than control cells (fig. S7D). Therefore, the precise roles of GSH as a cell-intrinsic and cell-extrinsic factor, or the interplay between these roles, should be further investigated to advance the clinical application of GVHD targeted cell therapies.

Mechanistically, our transcriptomic analysis demonstrated that CREB1-associated or transcription target genes were up-regulated in GSHHigh hES-MSCs (Fig. 1). CREB1 activity was increased in GSHHigh hES-MSCs, and FSK treatment enhanced the proliferative, clonogenic, migratory, anti-inflammatory, and immunomodulatory activities of MSCs with different sources including human ESCs (Fig. 3), UC (fig. S3), and BM (fig. S4). Previously, we reported that the CREB1 pathway is involved in induction of naive pluripotency in murine ESCs and primitiveness of human MSCs by ascorbic acid 2-glucoside (AA2G), a stable vitamin C derivative (34). The importance of CREB1 in naive pluripotency is further supported by accumulating evidence that FSK induces the naive state of human ESCs (35, 36) and stimulates somatic cell reprogramming (37). In hES-MSCs and UC, treatment of AA2G promotes their self-renewal (CFU-F) activity, PDGF-responsive cell migration, and anti-inflammatory potency, leading to the improved therapeutic outcomes in an asthma mouse model (34). CREB1 silencing severely impairs the beneficial effects of the vitamin C derivative on hES-MSCs (34). These findings strongly suggest that CREB1 is a central mediator in preserving the naive or primitive states of stem cells, which are preserved in vivo in specific niches but are extremely difficult to stabilize in vitro, largely because of the accumulation of epigenetic abnormalities (38) and oxidative stress provoked by supraphysiological stimulation (18). Therefore, identifying the upstream and downstream factors modulating CREB1 in specific types of stem cells is an important future study not only to advance our understanding of the molecular nature of stemness but also to develop optimal culture conditions for in vitro preservation of naive or primitive stem cell states and for the optimized and safe use of stem cells for basic research and therapeutic purposes.

Accordingly, in the present study, we identified that NRF2, a well-known master regulator of redox homeostasis, functioned as a critical effector for CREB1 in the maintenance of MSC GSH dynamics. CREB1 activated NRF2, which up-regulated NRF2 target genes involved in GSH synthesis (GCLM and GCLC) and redox cycling (GSR and PRDX1) through NRF2 binding to target gene protomer regions. This CREB1-NRF2 cascade was critical for the functionality of MSCs on the basis of the findings that ectopic expression of these CREB1-NRF2 targets in MSCs induced similar beneficial effects to FSK priming of MSCs (Fig. 5). In addition, the enforced GSH dynamics and up-regulation of NRF2 and related CREB1-NRF2 targets were observed in the aforementioned AA2G-primed hES- and hUC-MSCs (fig. S10, A to C), which showed the improved therapeutic outcomes in an asthma mouse model by promoting their self-renewal, engraftment, and anti-inflammatory capacities (34). Therefore, the CREB1-NRF2 cascade is a common mechanism underlying functional improvement by enhanced redox hemostasis (e.g., a high content of GSH and priming with FSK or AA2G) in human MSCs derived from different sources including ESCs, UC, and adult BM (fig. S10D).

The enhanced immunomodulatory functions of MSCs overexpressing CREB1-NRF2 target genes were confirmed in a humanized mouse model of GVHD (Fig. 6). In line with these findings, MSCs overexpressing each CREB1-NRF2 target exhibited enhanced immunomodulatory activity to the allogeneic response in an in vitro MLR assay (Fig. 5, F and G). However, it should be recognized that the immune reaction in the xenogeneic GVHD animal model used in the present study may be distinct from the allogeneic response of human patients with GVHD. In addition, the long-term therapeutic effects of these MSCs with enhanced GSH dynamics were not examined in this study. Therefore, further studies regarding these limitations and the clinical relevance of GSH dynamics–related mechanisms in specific types of MSCs are required to successfully translate our promising preclinical findings into clinical practice.

GVHD is a leading cause of late morbidity and mortality after allo-HSCT. However, there are urgent unmet clinical needs for this disorder, including lack of effective therapies and second-line therapy and dependence on steroids for first-line therapy (1). A number of immune modulatory cells, including MSC-based therapies, are being developed and have yielded exciting results (7, 12, 13). To successfully adapt these cell therapies as first-line treatments for established GVHD, it is necessary not only to evaluate cell functionality using cost-effective and standardized preparation procedures but also to develop methods for real-time monitoring and prediction of the therapeutic potency of living cells.

FreSHtracer is a nondestructive, integrated, and image-based probe that is ideal for assessing cellular property (39). In this study, we integrated FreSHtracer into a high-content imaging analysis system to monitor MSC GSH dynamics in real time under various cellular and environmental conditions. Quantitative readouts of basal GSH levels and GRC provided by this assay were highly correlated with in vitro and in vivo functionality of MSCs. Thus, this GSH dynamics assay could be a suitable tool to achieve a breakthrough in evaluating the quality and functionality of MSCs. Given the important role of GSH in several other cell types, this assay might also be applicable to other immune and regenerative cell therapies.

In summary, we demonstrated the molecular and functional importance of the CREB1-NRF2 pathway in MSC GSH dynamics, which determined therapeutic outcomes in a mouse model of GVHD. We also developed a reliable method to provide qualitative and quantitative analyses of GSH dynamics, which are predictive for core MSC functions, including self-renewal, migration, and immunomodulatory capacities. This will accelerate the clinical use of MSCs to treat GVHD, which remains a leading cause of late morbidity and mortality because of the lack of effective therapeutic strategies. Furthermore, our findings can be applied to other immune modulatory cells and disorders associated with immune deregulation.

MATERIALS AND METHODS

Study approval

All animal experiments were approved by the Institutional Animal Care and Use Committee of the University of Ulsan College of Medicine (IACUC-2018-12-184). Human UC samples were obtained from healthy, normal, full-term newborns after obtaining written informed parental consent in accordance with guidelines approved by the Ethics Committee on the Use of Human Subjects at Asan Medical Center. Informed consent was obtained from all pregnant mothers before UC collection.

Cell culture and in vitro characterization of MSCs

hES-MSCs differentiated from H9-hESCs were cultured in EGM-2 MV medium (Lonza, San Diego, CA, USA) on plates coated with rat tail collagen type I (Sigma-Aldrich, St. Louis, MO, USA) as previously described (23, 40). Human UC-MSCs were grown in low-glucose Dulbecco’s modified Eagle’s medium (DMEM) containing 10% heat-inactivated fetal bovine serum (FBS), human epidermal growth factor (5 ng/ml; Sigma-Aldrich), basic FGF (10 ng/ml), and long-R3 insulin-like growth factor-1 (50 ng/ml; ProSpec, Rehovot, Israel) as previously described (41). Human BM-MSCs purchased from Lonza (Basel, Switzerland) were cultured following the manufacturer’s instructions. All MSCs used in the present study were maintained in a humidified atmosphere with 5% CO2 at 37°C.

An MTT assay (Sigma-Aldrich) was used to evaluate cell viability. Cell proliferation, self-renewal (CFU-F), multipotency, surface marker expression, transwell migration, quantification of angiogenesis in vitro using Matrigel, in vitro anti-inflammation, immune modulation, and allogeneic MLR assays of MSCs were performed as previously described (16, 34). Chemotactic activity was measured using an 8-μm pore Boyden chamber (Corning) with PDGF-AA (10 ng/ml; R&D Systems, Minneapolis, MN, USA) or SDF1 chemokines (150 ng/ml; R&D Systems) in the lower chamber. Cell function assays were quantified by digital image analysis using Image-Pro 5.0 software (Media Cybernetics, Rockville, MD, USA).

MLR assay

Different lots of human PBMCs were purchased from STEMCELL Technologies. Stimulator human PBMCs were inactivated at 5.0 gray (Gy) using the X-RAD 320 x-ray irradiator (Precision X-Ray Inc., North Branford, CT, USA). Inactivated PBMCs (1 × 105) were added to each well of a 96-well culture plate containing human MSCs (2 × 104) and responder PBMCs (1 × 105). The mixtures were further incubated for 5 days at 37°C in 5% CO2. Cell proliferation activity was analyzed using the CFSE Cell Division Tracker Kit (#423801; BioLegend, San Diego, CA, USA), and the percentage of CD3+ T cells (#300412; BD Biosciences) was analyzed using the BD FACSCanto II flow cytometer (BD Biosciences). Analysis of FACS data was performed using FlowJo software 7.6.5 (FlowJo LLC, Ashland, OR, USA).

Humanized GVHD animal model

The animal model of GVHD was generated as described previously (14, 16). Briefly, 9-week-old male NOD/ShiLtJ-Prkdcem1AMCIl2rgem1AMC (NSG) mice (26 to 29 g) were purchased from JOONGAH BIO (JA BIO, Suwon-si, Gyeonggi-do, Republic of Korea), irradiated at 2.0 Gy using an X-RAD 320 x-ray irradiator (Precision X-Ray Inc.), and injected with 2.5 × 106 human PBMCs (#70025; STEMCELL Technologies, Vancouver, Canada) or the same volume of phosphate-buffered saline (PBS; sham group) via the tail vein within 24 hours of irradiation. After 18 days, 5 × 105 human UC-MSCs carrying empty control–, GCLM-, or PRDX1-expressing constructs were suspended in 100 μl of PBS and injected via the tail vein. For sham and GVHD groups, PBS alone was injected as a control. Clinical symptoms of GVHD were evaluated daily by examining body weight loss, survival, hunched back, and fur texture and were recorded every second day. Five mice per group were used in two independent experiments. Six weeks after MSC administration, target organs (lung, liver, kidney, and small intestine) and peripheral blood were harvested from mice in all groups for histological, donor T cell proliferation, and multiplex cytokine analyses, as previously described (16). Mice were randomly allocated to treatment groups, and the order of irradiation, cell transplantation or vehicle injection, and daily examination were randomized. Treatment groups were masked to investigators who were involved in GVHD induction procedures. All GVHD symptoms, histological, cytokine, and peripheral blood cell assessments were conducted by blinded investigators.

Assessment of GVHD animals

Histological assessment of GVHD target organs was performed, as described previously (16). For donor T cell proliferation, cells isolated from NSG mouse blood were resuspended in 500 μl of PBS and incubated with fluorescein isothiocyanate– or phycoerythrin-conjugated primary antibodies for 30 min at room temperature. Primary antibodies against human antigens CD3 (#300412), CD4 (#300512), and CD45 (#555483) and mouse antigen CD45 (#553079) were purchased from BD Biosciences. The fluorescence intensity of cells was analyzed using the BD FACSCanto II flow cytometer (BD Biosciences). Data were analyzed using FlowJo software 7.6.5 (FlowJo).

For multiplex cytokine analysis, sera from humanized GVHD mice were analyzed using the Magnetic Luminex Screening Assay Human Premixed Multi-Analyte Kit (#LXSAHM-28, R&D Systems). Measuring and quantification of the 28-plex human cytokines were requested from KOMA BIOTECH Inc. (Seoul, Korea) and performed using the Varioskan Flash Reader (Invitrogen/Thermo Fisher Scientific), according to the manufacturer’s instructions.

Real-time monitoring of GRC in living cells

Human MSCs were plated into Greiner CELLSTAR 96-well plates (#655090, Greiner Bio-One, Kremsmünster, Austria) at a density of 1 × 103 cells per well. Following further incubation for 12 hours, the cells were labeled in the culture media containing 2 μM FreSHtracer (Cell2in Inc., Seoul, Korea) at 37°C in 5% CO2 incubator for 2 hours. Time-lapse imaging of the cells was recorded from 3 min before to 45 min after treatment of 200 μM diamide (Sigma-Aldrich) using the Operetta High-Content Imaging System (HH12000000, PerkinElmer, Waltham, MA, USA) with ×200 or ×400 magnifications. Fluorescence emissions were detected at 510 and 580 nm, exciting at 430 and 520 nm, respectively, to obtain the FR values of FreSHtracer (18). The fluorescence signals were analyzed using Harmony High-Content Imaging and Analysis Software 3.1 (PerkinElmer) with confocal mode. Values for GSH dynamics index (GI) and related initial FR (for baseline of total GSH) and slope after diamide treatment (for GRC) for each plot are presented in data file S3.

Transcriptome analysis

The hES-ESCs labeled with 2 μM FreSHtracer for 2 hours in culture medium were sorted into GSHHigh and GSHLow fractions according to the ratio of fluorescence intensity at 510 nm to that at 580 nm (F510/F580 ratio; FR) of FreSHtracer using an Aria III flow cytometer system (BD Biosciences, Franklin Lakes, NJ, USA), as described previously (18). Total RNA from FACS-sorted cells was isolated using the RNeasy Mini Kit (QIAGEN, Valencia, CA, USA), including treatment with DNase I (QIAGEN). One microgram of total RNA was subjected to analysis using the Affymetrix GeneChip Human 2.0 ST Array (Affymetrix, Santa Clara, CA, USA). Microarray image data were processed on a GeneChip GCS3000 Scanner and Command Console software (Affymetrix). After importing the CEL files of six samples (three independent samples from each group), the data were summarized and normalized using the robust multi-average method implemented in Affymetrix Expression Console Software.

MetaCore microarray software (Clarivate Analytics, Philadelphia, PA, USA) or GSEA (Broad Institute, Cambridge, MA, USA) was used for functional analysis of the transcriptome databases for gene networks, biofunctions, and canonical pathways with the default settings. In MetaCore analysis, genes up- or down-regulated ≥1.5-fold with P < 0.05 were defined as significantly changed. Details of the statistical values and significant genes for biological processes and pathways in the MetaCore analysis are described in data file S1. For GSEA, gene sets were obtained from published literature or filtered from a curated functional gene set (C2) database. Significant differences were determined on the basis of a false discovery rate of <0.25. A detailed list of gene sets with the corresponding genes and references is described in data file S2.

Gene expression analysis

Quantitative assessment of the mRNA levels of the target genes was performed using 50 ng of total RNA, as described previously (18). The relative expression level of the target genes was determined using the 2−ΔΔCt method, and GAPDH was used as the endogenous control gene.

Western blotting

Cell extracts (30 μg) were prepared in radioimmunoprecipitation assay (RIPA) lysis buffer (Santa Cruz Biotechnology, Santa Cruz, CA, USA) and separated on SDS–polyacrylamide gel electrophoresis gels. The expression level of the indicated proteins was assessed by probing with specific antibodies, which are indicated in Table 1. The density of signals for the indicated proteins was measured and quantified using National Institutes of Health ImageJ software.

Table 1 Key resource table.

View this table:

Immunostaining

To detect NRF2 protein, human MSCs were fixed with 4% paraformaldehyde (Sigma-Aldrich) for 30 min, stained with an antibody specific to NRF2 (#ab62352; Abcam), and then labeled with an Alexa Fluor 488–conjugated anti-rabbit antibody (#A11008, Thermo Fisher Scientific, Waltham, MA, USA). Nuclei were counterstained with 4′,6-diamidino-2-phenylindole (Sigma-Aldrich). Images were acquired using a ZEISS LSM 800 confocal microscope system (Carl Zeiss, Munich, Germany).

ChIP assay

ChIP analysis was performed using a Magna ChIP G kit (Millipore, Billerica, MA, USA) according to the manufacturer’s instructions. The detailed procedures were performed as reported previously (38, 42) using 3 μg of ChIP-grade antibody against NRF2 (#ab62352, Abcam).

Ectopic expression and RNA interference

Open reading frame (ORF) constructs for human GCLM, GCLC, PRDX1, and GSR genes were purchased (Dharmacon, Lafayette, CO, USA; GCLM, #OHS6085-213573329; GCLC, #OHS6085-213573328; PRDX1, #OHS1770-202318473; GSR, #OHS1770-202312884) and cloned into the pLX304 lentiviral vector (Addgene plasmid 25890) using Gateway Technology (Invitrogen/Thermo Fisher Scientific, Waltham, MA, USA). For silencing of the indicated genes, two independent shRNAs were designed for each target and were cloned into the pLenti6/BLOCK-iT lentiviral vector (Invitrogen/Thermo Fisher Scientific). Lentiviruses containing the ORFs or shRNA constructs were produced as previously described (16, 42). In vitro and in vivo assays were performed 4 days after lentiviral infection. The sequences of oligonucleotides in each shRNA are indicated in Table 1.

Quantification of the amount of cAMP or activities of PKA and CREB1

The luciferase reporter containing eight tandem CRE (8× CRE-Luc), which was provided by Y. Song, University of Ulsan College of Medicine, Seoul, Korea (25), was used to measure the biological activity of CREB1, as reported previously (34). The intracellular amount of cAMP was measured in 20 μg of cell extract using the Parameter cAMP Assay Kit (R&D Systems), and the PKA kinase activity was measured using cell extracts prepared from 5 × 105 cells using the PKA Kinase Activity Assay Kit (Abcam). Assays were performed according to the manufacturer’s instructions.

Statistical analysis

Data were statistically analyzed using a nonparametric Mann-Whitney test or analysis of variance (ANOVA) (one- or two-way) with a Bonferroni post hoc test. All analyses were performed using GraphPad Prism 7.0 software (GraphPad Software, La Jolla, CA, USA). P < 0.05 was considered statistically significant.

SUPPLEMENTARY MATERIALS

Supplementary material for this article is available at http://advances.sciencemag.org/cgi/content/full/6/16/eaba1334/DC1

This is an open-access article distributed under the terms of the Creative Commons Attribution-NonCommercial license, which permits use, distribution, and reproduction in any medium, so long as the resultant use is not for commercial advantage and provided the original work is properly cited.

REFERENCES AND NOTES

Acknowledgments: Funding: This research was supported by the National Research Foundation of Korea (NRF-2017M3A9B4061890, NRF-2018R1A2B2001392, and NRF-2020R1C1C1006494), an NRF MRC grant funded by the Korean government (MSIP) (NRF-2018R1A5A2020732), the Ministry of Education (NRF-2017R1D1A1B03035059 and NRF-2018R1D1A1B07047450), and a grant from the Korean Health Technology R&D Project, Ministry of Health & Welfare, Republic of Korea (HI18C2391). Author contributions: Conceptualization: D.-M.S., E.M.J., K.C., and I.-G.K. Methodology: D.-M.S., J.L., J.H., J.-W.S., E.M.J., and K.C. Investigation: J.L., J.H., H.J., Y.K., S.L., H.Y.Y., C.-M.R., H.Y., S.S., J.-S.R., E.M.J., and D.-M.S. Writing (original draft): D.-M.S., E.M.J., J.L., and J.H. Writing (review and editing): D.-M.S., E.M.J., and J.L. Funding acquisition: D.-M.S., E.M.J., and I.-G.K. Resources: K.-S.H., H.-M.C., H.-R.K., J.-S.R., and K.C. Data curation: D.-M.S., E.M.J., J.L., J.H., and H.J. Supervision: D.-M.S. and E.M.J. Competing interests: I.-G.K., K.C., D.-M.S., E.M.J., and J.-W.S. cofounded Cell2in, a company focused on developing FreSHtracer-based assays. The other authors have no competing interests to declare. Data and materials availability: The transcriptome data described in this study have been deposited in the Gene Expression Omnibus (GEO) of the NCBI and are accessible under GEO series accession number GSE135998. Functional analysis of transcriptomes and core analyses of gene networks, biofunctions, and canonical pathways were performed using MetaCore or GSEA microarray software with default settings. MetaCore is an integrated software that requires users to purchase a license. Details of the statistical values and significant genes for biological processes and pathways in the MetaCore analysis are described in data file S1. A detailed list of gene sets with the corresponding genes and references used for GSEA, a publicly accessible software, is provided in data file S2. Values for the GI for each plot are presented in data file S3. Information about all reagents, resources, and oligonucleotides is described in the Table 1.
View Abstract

Stay Connected to Science Advances

Navigate This Article