Sulforaphane-Induced Klf9/Prdx6 Axis Acts as a Molecular Switch to Control Redox Signaling and Determines Fate of Cells

Sulforaphane (SFN), an activator of transcription factor Nrf2 (NFE2-related factor), modulates antioxidant defense by Nrf2-mediated regulation of antioxidant genes like Peroxiredoxin 6 (Prdx6) and affects cellular homeostasis. We previously observed that dose levels of SFN are crucial in determining life or death of lens epithelial cells (LECs). Herein, we demonstrated that higher doses of SFN (>6 μM) activated death signaling by overstimulation of Nrf2/ARE (antioxidant response element)-mediated Kruppel-like factor (Klf9) repression of Prdx6 expression, which increased reactive oxygen species (ROS) load and cell death. Mechanistically, Klf9 bound to its repressive Klf9 binding elements (RKBE; 5-CA/GCCC-3) in the Prdx6 promoter, and repressed Prdx6 transcription. Under the condition of higher dose of SFN, excessive Nrf2 abundance caused death signaling by enforcing Klf9 activation through ARE (5-RTGAYnnnGC-3) in Klf9 promoter that suppress antioxidant genes such as Prdx6 via a Klf9-dependent fashion. Klf9-depletion showed that Klf9 independently caused ROS reduction and subsequent cell survival, demonstrating that Klf9 upregulation caused cell death. Our work revealed the molecular mechanism of dose-dependent altered activity of SFN in LECs, and demonstrated that SFN activity was linked to levels of Nrf2/Klf9/Prdx6 axis. We proposed that in the development of therapeutic interventions for aging/oxidative disorders, combinations of Klf9-ShRNA and Nrf2 inducers may prove to be a promising strategy.


Introduction
Substantial experimental evidence indicates that etiopathobiology of age-related disorders is associated with increased levels of reactive oxygen species (ROS) and reduced expression and activity of antioxidant pathway, leading to molecular damage that result in pathobiology and disease. To cope with internal and external oxidative stresses, cells have evolved defensive mechanisms that integrate the coordinated stimulation of a group of phase I and phase II antioxidant enzymes, such as NAD(p)H quinone oxidoreductase (NQO1), Thioredoxin reductases (Txnrds) Glutathione peroxidase (GPx), Catalase (Cat), Superoxide dismutases (SODs) and peroxiredoxin 6 (Prdx6). These enzymes are upregulated by Nrf2 (NFE2-related factor 2), a universal transregulator of antioxidant response genes, to revive cellular redox homeostasis [1][2][3][4]. The Nrf2-mediated protective pathway is highly sensitive to oxidative stress in cells [5]. To maintain homeostasis, Nrf2 binds to antioxidant response elements (ARE) present in antioxidant protein genes and upregulates their transcription. Dysregulation of Nrf2 during aging or oxidative stress results in failure of antioxidant response-mediated cellular homeostasis [1].
SFN is a naturally occurring organosulfur found in cruciferous vegetables, with high levels detected in broccoli and broccoli sprouts. Recent evidence shows SFN s diversified activities, ranging from cell survival and cytoprotection to cytotoxicity and growth inhibition; these activities are determined by concentrations of SFN and cellular background [1,[63][64][65][66][67]. SFN exerts its function via activating the redox-sensitive Nrf2/ARE antioxidant pathway and interacting with other mammalian biochemical pathways [63,68,69]. Activation of Nrf2 by SFN occurs through increased cytosolic transcription and facilitation of Nrf2 translocation in nucleus by inducing cysteine modifications to Keap1 [30,70]. A greater understanding of how SFN produces its bimodal effects and directs survival or death signaling pathways in LECs or redox active cells is needed. In the research reported here, we used human LECs as a model system to reveal the dose-dependent molecular mechanism of SFN in regulation of Nrf2/ARE-mediated protective pathway. We found that the cytotoxic effect of SFN at high doses was linked to repression of Prdx6 and dramatic increase in ROS levels with aberrant expression and nuclear accumulation of Nrf2. This accumulation of Nrf2 led to its binding to ARE present in the promoter of unfavorable target Klf9 and enhanced its expression and nuclear abundance. We found that Klf9 bound to its Repressive Klf9 Binding Elements (RKBE) of the Prdx6 promoter and repressed the Prdx6 expression that in turn resulted in increased ROS-induced cell injury, suggesting an advent of dominant Klf9-mediated repressive signaling during high doses of SFN. Klf9 depletion experiments showed that Klf9 dramatically reduced ROS levels and cell injury. Aberrant Klf9 expression induced by higher doses of SFN was a major culprit in acceleration of ROS generation and cell death, while lower doses of SFN did not affect Klf9 expression, and led to SFN-induced Nrf2/ARE-mediated cellular protection.

Cell Culture
hLECs were derived from 12 infants who underwent surgery for retinopathy of prematurity [71] (a kind gift of Dr. Venkat N. Reddy, Eye Research Institute, Oakland University, Rochester, MI, USA). These LECs were immortalized with SV40 and were maintained in Dulbecco s Modified Eagle Medium (DMEM; Invitrogen, Waltham, MA, USA) with 15% fetal bovine serum (FBS; Atlanta Biologicals, Atlanta, GA, USA), 100 µg/mL streptomycin, and 100 µg/mL penicillin in 5% CO 2 environment at 37 • C as described previously [7,57]. To examine the effect of Sulforaphane (SFN), cells were treated with different concentrations (DMSO (dimethyl sulfoxide; Sigma Aldrich, St. Louis, MO, USA), 6, 12, 18 and 24 µM in complete medium) of SFN. A stock solution of SFN (50 mM) was prepared in DMSO and diluted in culture medium keeping the final DMSO concentration at < 0.05% and same concentration of DMSO was used as vehicle control. DL-SFN (Catalog no. S4441) and tBOOH (tert Butyl hydroperoxide; Catalog no. 458139) were purchased from Sigma Aldrich (St. Louis, MO, USA).

Quantitation of Intracellular ROS Level by H2-DCF-DA and CellROX ® Deep Red Reagent
Intracellular ROS level was measured by use of fluorescent dye dichlorofluorescin diacetate (H2DCFDA), a nonpolar compound that is converted into a polar derivative (dichlorofluorescein) by cellular esterase after incorporation into cells [7,52,72,73]. On the day of the experiment, the medium was replaced with Hank s solution containing 10 µM H2DCFDA dye and cells were incubated at 37 • C. After 30 min, intracellular fluorescence was detected with excitation (Ex) at 485 nm and emission (Em) at 530 nm by a Spectra Max Gemini EM (Mol. Devices, Sunnyvale, CA, USA).
ROS level were measured according to the company s protocol (CellROX ® Deep Red Oxidative Stress Reagent, Catalog No. C10422) and described in previous protocol [74]. In brief, LV Sh-control and LV Sh-Klf9 hLECs (5 × 10 3 ) cells were seeded in 96 well plate and 24 h later treated with different A colorimetric MTS assay (Promega, Madison, WI, USA) was performed as described earlier [7,75,76]. This assay of cellular viability uses MTS and an electron coupling reagent (Phenazine ethosulfate; PES). PES has enhanced chemical stability, which allows it to be combined with MTS to form a stable solution. Assay are performed by adding MTS reagent directly to culture cells, incubating for 1-4 h and then recording absorbance at 490 nm with a 96-well plate reader, Spectra Max Gemini EM (Mol. Devices, Sunnyvale, CA). Results were normalized with absorbance of the untreated control(s).
The relative quantity of the mRNA was obtained using the comparative threshold cycle (CT) method. The expression levels of target genes were normalized to the levels of β-actin as an endogenous control in each group.

Extraction of Nuclear and Cytosolic Fraction and Total Cell Lysates
Nuclear extract was prepared following as described earlier [52,77]. Briefly, LECs (1 × 10 6 ) were cultured in 100-mm plates. hLECs were treated with DMSO control or different concentrations of SFN (6, 12, 18 and 24 µM) for 24 h. The cells were washed gently with chilled phosphate-buffered saline (pH 7.4). Cells were collected by centrifugation using a micro-centrifuge and resuspended in 5 pellet volumes of cytoplasmic extract buffer ((10 mM 4-(2-hydroxyethyl)-1-piperazineethanesulfonic acid (HEPES, adjusted pH at 7.9), 10 mM KCl, 0.1 mM EDTA (Ethylenediaminetetraacetic acid), 0.4% (v/v) Nonidet P-40, 0.5 mM phenylmethylsulfonyl fluoride (PMSF), 1 mM DTT (Dithiothreitol) and Protease inhibitor). After a short incubation on ice and centrifugation (4 • C) at 10000 rpm for 10 min, the cytoplasmic extract was transfer in fresh tube from the pellet. Following careful washing with cytoplasmic extract without detergent (Nonidet P-40), the fragile nuclei were resuspended in nuclear extract buffer ((20 mM HEPES (adjusted pH at 7.9), 0.4 M NaCl, 1 mM EDTA, 10% (v/v) glycerol, 1 mM DTT, 0.5 mM PMSF and Protease Inhibitor) and incubated for 2 h at 4 • C with continuous vortexing. Finally, the extract was spun at 14,000 rpm for 15 min to pellet the nuclei. After centrifugation, the nuclear extract was transferred and aliquoted in fresh tubes, and individual aliquots were stored at −70 • C to avoid repeated freezing and thawing of the preparation.
Total cell lysates of LECs were prepared in ice-cold radioimmune precipitation buffer (RIPA buffer) [58]. Briefly, LECs (1 × 10 6 ) were cultured in 100-mm plates. The cells were washed gently with chilled phosphate-buffered saline (pH 7.4). Cells were collected by centrifugation using a micro-centrifuge and resuspended in three pellet volumes of RIPA buffer *(1% IGEPAL (CA-630; Sigma Aldrich, St. Louis, MO, USA) 0.5% sodium deoxycholate, 0.1% sodium dodecyl sulfate (SDS; Catalog No. 15553027, Thermo Fisher Scientific, Waltham, MA, USA), 0.5 mM phenylmethylsulfonyl fluoride (PMSF), and Protease inhibitor). After 30 min of incubation on ice and centrifugation (4 • C) at 10,000 rpm for 10 min, the total cell lysates was transfer in fresh tube from the pellet. Protein level was estimated according to the Bradford protein assay and/or Pierce™ BCA Protein assay methods and extract was used for experiment as required.

Transcription Factor Nrf2 Activation Assay
Nrf2 activation assay was performed according to manufacturer s protocol (TransAM Nrf2 Transcription Factor Assay Kit, Cat No 50296, Active motif, Carlsland, California, USA). In brief, 10 µg of nuclear extract (up to 10 µL diluted with complete lysis buffer) prepared from SFN treated hLECs added to the strips well, following the addition of 40 µL complete binding buffer contains 20 pmol of the wild-type and/or mutated consensus oligonucleotide to each sample well. For blank well, 10 µL of complete lysis buffer were used. The plate was incubated for 1 h at room temperature (RT) with mild agitation. 100 µL primary antibody (1:1000 in 1× antibody binding buffer) added after three washes with 1× washing buffer and incubated at RT for 1 h without agitation. 100 µL of diluted anti-rabbit HRP (horseradish peroxidase)-conjugated antibody (1:1000 dilution in 1× antibody binding buffer) after three washing was added and incubated for 1 h at RT. 100 µL of developing solution was added to wells after four washing and incubated at RT in dark for 2 to 10 min. Finally, by addition of 100 µL of stop solution, optical density (O.D.) was recorded at absorbance 450 nm.

Chromatin Immunoprecipitation (ChIP) Assay
ChIP was performed using the ChIP-IT ® Express (Cat. No. 53008; Active Motif, Carlsbad, CA, USA) and ChIP-IT ® qPCR analysis kit (Cat. No. 53029; Active Motif, Carlsbad, CA, USA) following the manufacturer s protocol and as described earlier [1,7]. The following antibodies were used: control IgG and antibody specific to Klf9 (Catalog No. 701888, Thermo Scientific, Waltham, MA, USA) and Nrf2 (Catalog No. Ab180845, Abcam). RT-PCR amplification was carried out using 5 µL of DNA sample with primers as indicated below. The program for quantification amplification was 3 min 94 • C, 20 sec at 95 • C, 30 sec at 59 • C and 30 sec at 72 • C for 36 cycles in 25 µL reaction volume (RT-PCR). Data obtained with RT-PCR run on 1% agarose gel and visualized band under UV and image captured.

Construction of Human Prdx6 Promoter-Chloramphenicol Acetyltransferase (CAT) Reporter Vector
The 5 -flanking region spanning from −1559 to +30 bp was isolated from human genomic DNA by using an Advantage ® Genomic PCR Kit (Cat. No. 639103 & 639104, Clontech Laboratories, Inc, Mountain View, CA 94043). PCR product was cleaned and verified by sequencing as described previously [80]. A construct containing −1559 to +30 bp was engineered by ligating it to basic pCAT vector (Promega) using the SacI and XhoI sites. Primers were as follows: Sense; 5 -GACAGAGTTGAGCTCCACACAG-3 ; and antisense; 5 -CACGTCCTCGAGAAGCAGAC-3 .

Lentiviral (LV) Infection
CopGFP control lentiviral particle (LV Sh-Control, Sc-108084) and BTEB1 (Klf9)/GFP ShRNA (LV Sh-Klf9, sc-37716-VS) were purchased from Santa Cruz Biotechnology and hLECs were infected following the Company s protocol. Briefly, hLECs were cultured in 12 well plate in complete medium. After 24 h, media were removed and then substituted with 1 mL medium containing polybrene (sc-134220, Santa Cruz) at a final concentration of 5 µg/mL. Cells were infected by adding the Sh-Control and Sh-Klf9 lentiviral particles to the culture, mixed by swirling and incubated overnight. At 24 h post infection, polybrene-containing medium were removed and fresh complete medium (without polybrene) were added. Infected cells were split and incubated in complete medium for a further 24-48 h. For stable selection, the infectants were treated with puromycin dihydrochloride (sc-108071) selection marker. These stably infected hLECs with LV Sh-Control or LV Sh-Klf9 were used for the current studies.

Statistical Analysis
For all quantitative data collected, statistical analysis was conducted by Student s t test and /or one-way ANOVA when appropriate, and was presented as mean ± S.D. of the indicated number of experiments. A significant difference between control and treatment group was defined as a p value of <0.05 and 0.001 for two or more independent experiments.

High Doses of SFN Increased Generation of Intracellular ROS and Were Linked to LEC Death
We recently showed [1] that SFN activates Nrf2/ARE antioxidant pathway of LECs. During this study we observed increased LECs death with increased SFN concentrations (> 6µM). Because SFN is known to protect cells against oxidative stress through upregulation of the Nrf2/ARE antioxidant pathway, we sought to determine the mechanism by which high doses of SFN regulate ROS levels in LECs. We first identified an effective noncytotoxic dose of SFN, finding that doses of less than 6 µM were effective noncytotoxic concentrations, while doses of more than 6 µM inhibited LEC growth and increased cell death. We hypothesized that higher doses of SFN may induce increased levels of ROS that in turn lead to death of LECs. We treated hLECs with different concentrations of SFN (6 µM non-cytotoxic and 12 µM, 18 µM and 24 µM cytotoxic) and measured ROS and cell viability using H2-DCF-DA dye and MTS assays, respectively. With higher concentrations, we observed dose-dependent significant increases in accumulation of ROS ( Figure 1A) and cell death ( Figure 1B). In contrast, noncytotoxic concentrations of SFN did not alter ROS ( Figure 1A), and increased cell growth ( Figure 1B). These observations served as a basis for study of the molecular mechanism of SFN-mediated bimodal function. Reverse primer: 5′-GACCCAGAAGTATGGAAGACCTTATGTCAGAG-3

Lentiviral (LV) Infection
CopGFP control lentiviral particle (LV Sh-Control, Sc-108084) and BTEB1 (Klf9)/GFP ShRNA (LV Sh-Klf9, sc-37716-VS) were purchased from Santa Cruz Biotechnology and hLECs were infected following the Company′s protocol. Briefly, hLECs were cultured in 12 well plate in complete medium. After 24 h, media were removed and then substituted with 1 mL medium containing polybrene (sc-134220, Santa Cruz) at a final concentration of 5 μg/mL. Cells were infected by adding the Sh-Control and Sh-Klf9 lentiviral particles to the culture, mixed by swirling and incubated overnight. At 24 h post infection, polybrene-containing medium were removed and fresh complete medium (without polybrene) were added. Infected cells were split and incubated in complete medium for a further 24-48 h. For stable selection, the infectants were treated with puromycin dihydrochloride (sc-108071) selection marker. These stably infected hLECs with LV Sh-Control or LV Sh-Klf9 were used for the current studies.

Statistical Analysis
For all quantitative data collected, statistical analysis was conducted by Student′s t test and /or one-way ANOVA when appropriate, and was presented as mean ± S.D. of the indicated number of experiments. A significant difference between control and treatment group was defined as a p value of <0.05 and 0.001 for two or more independent experiments.

High Doses of SFN Increased Generation of Intracellular ROS and Were Linked to LEC Death
We recently showed [1] that SFN activates Nrf2/ARE antioxidant pathway of LECs. During this study we observed increased LECs death with increased SFN concentrations (> 6μM). Because SFN is known to protect cells against oxidative stress through upregulation of the Nrf2/ARE antioxidant pathway, we sought to determine the mechanism by which high doses of SFN regulate ROS levels in LECs. We first identified an effective noncytotoxic dose of SFN, finding that doses of less than 6 μM were effective noncytotoxic concentrations, while doses of more than 6 μM inhibited LEC growth and increased cell death. We hypothesized that higher doses of SFN may induce increased levels of ROS that in turn lead to death of LECs. We treated hLECs with different concentrations of SFN (6 μM non-cytotoxic and 12 μM, 18 μM and 24 μM cytotoxic) and measured ROS and cell viability using H2-DCF-DA dye and MTS assays, respectively. With higher concentrations, we observed dosedependent significant increases in accumulation of ROS ( Figure 1A) and cell death ( Figure 1B). In contrast, noncytotoxic concentrations of SFN did not alter ROS ( Figure 1A), and increased cell growth ( Figure 1B). These observations served as a basis for study of the molecular mechanism of SFNmediated bimodal function.

High Doses of SFN Induced a Dramatic Rise in Nrf2/Klf9 Expression with Reduced Prdx6 Expression and Increased Accumulation of Nrf2/Klf9 in the Nucleus
Based on these findings (Figure 1), we surmised that the Nrf2/ARE antioxidant pathway in LECs may have been suppressed, as previously reported in other cell types [26]. To determine the molecular mechanism by which ROS was increased ( Figure 1A), we examined the dose-dependent effect of SFN on Nrf2/Klf9 mRNA and protein levels by using the same concentrations shown in Figure 1. We treated hLECs for 24 h with increasing amounts of Nrf2 activator SFN [1,81], as shown in Figure 2, and conducted RT-qPCR. Levels of Nrf2 and its target gene, Prdx6 mRNA expression, were increased in response to 6 µM of SFN treatment, but Klf9 levels were not upregulated ( Figure 2A, gray bar). Treatment with doses higher than 6 µM of SFN resulted in dose-dependent Klf9 upregulation, which was correlated with increased Nrf2 expression and reduced Prdx6 expression (Figure 2A, black bars).

High Doses of SFN Induced a Dramatic Rise in Nrf2/Klf9 Expression with Reduced Prdx6 Expression and Increased Accumulation of Nrf2/Klf9 in the Nucleus
Based on these findings (Figure 1), we surmised that the Nrf2/ARE antioxidant pathway in LECs may have been suppressed, as previously reported in other cell types [26]. To determine the molecular mechanism by which ROS was increased ( Figure 1A), we examined the dose-dependent effect of SFN on Nrf2/Klf9 mRNA and protein levels by using the same concentrations shown in Figure 1. We treated hLECs for 24 h with increasing amounts of Nrf2 activator SFN [1,81], as shown in Figure 2, and conducted RT-qPCR. Levels of Nrf2 and its target gene, Prdx6 mRNA expression, were increased in response to 6 μM of SFN treatment, but Klf9 levels were not upregulated ( Figure  2A, gray bar). Treatment with doses higher than 6 μM of SFN resulted in dose-dependent Klf9 upregulation, which was correlated with increased Nrf2 expression and reduced Prdx6 expression ( Figure 2A, black bars).  Cells were untreated or treated with increasing doses of SFN; non-toxic amount of SFN (<6 µM) and higher doses of SFN (>12 µM) as indicated. After 24 h, RT-qPCR was conducted using primers specific to Nrf2 (AI), Klf9 (AII) and Prdx6 (AIII). Experiments were conducted three times and data is represented as mean ± S.D. * p < 0.05, ** p < 0.001; SFN treated versus untreated (DMSO) control. (B) Nrf2 and Klf9 are overly accumulated in nucleus and inversely related to reduce abundance of Prdx6 protein in response to excessive amount of SFN. hLECs were treated or untreated with different doses of SFN as indicated. Nuclear (Nrf2 and Klf9) and cytosolic (for Prdx6) fractions were isolated. Samples containing equal amounts of protein were immunoblotted using an antibody specific to Nrf2, Klf9 and Prdx6, as shown. A dramatic nuclear accumulation of Nrf2 and Klf9 (BI) and reduction in Prdx6 expression (BII) were observed in response to increased concentrations of SFN treatment. Right adjacent panel reveals densitometry analysis of each protein bands. (C) SFN treated hLECs displayed significant increased Nrf2 activity. hLECs were treated with DMSO control or different concentration of SFN for 24 h, and nuclear extracts were analyzed for Nrf2-ARE (antioxidant responsive element) binding by ELISA (enzyme-linked immunosorbent assay). Nuclear extracts containing equal amounts of protein were processed and assayed for Nrf2 activity using a commercially available kit (Active motif) as described in Materials and Methods. The data represent the mean ± S.D. from three independent experiments. p values were determined SFN treated versus untreated (DMSO) control. * p < 0.05, ** p < 0.001.
To examine intracellular fate and status of nuclear Nrf2, Klf9 and cytoplasmic Prdx6 protein in hLECs in response to SFN treatment, we analyzed the nuclear and cytosolic fractions of SFN-treated hLECs for 24 h using immunoblotting ( Figure 2B). Immunoblot results using Nrf2 or Klf9 antibodies revealed that Nrf2 or Klf9 migrated at approximately 110 kDa or 32 kDa on SDS-PAGE(sodium dodecyl sulphate-polyacrylamide gel electrophoresis). Enrichment of both molecules in nuclear extracts of hLECs was increased with increases of SFN dose, and maximum accumulation was observed at 24 µM of SFN ( Figure 2B, I and II). This increased nuclear accumulation of Nrf2/Klf9 was directly related to reduction in the cellular abundance of Prdx6 protein, indicating that Klf9 may be involved in suppression of Prdx6 in hLECs treated with higher doses of SFN, as reported previously in another model system.
We found a dramatic inverse correlation between expression of Nrf2 and its target gene Prdx6 transcripts in cells treated with higher doses of SFN (Figure 2A), which was unexpected, as Nrf2 is a transactivator of Prdx6. We postulated that the inverse correlation may be related to dysregulation of DNA binding efficiency/activity in Nrf2, or other adverse signaling may be involved. To investigate that possibility, we examined Nrf2 transactivation capacity. Nuclear fraction isolated from hLECs treated with increasing doses of SFN having equal amounts of protein was employed to quantify the activity of Nrf2 by TransAM Nrf2 transcription factor assay (Active Motif) as shown in Figure 2C. Data revealed that Nrf2-DNA binding activity progressively increased with higher doses of SFN, indicating that DNA binding activity of Nrf2 was not attenuated by higher doses of SFN or SFN-induced increased ROS production. Collectively, these experiments indicate that some other factor may be involved in the repression of Prdx6 transcription. Since Prdx6 was suppressed at mRNA levels, we focused our further experimentation on transcriptional mechanisms by conducting in silico analysis of Klf9 and Prdx6 promoters for putative binding sites of Nrf2 and Klf9, respectively.

Bioinformatics Analyses Revealed Putative Nrf2 Activating and Klf9 Repressive Binding Sites in Klf9 (A) and Prdx6 (B) Promoters, Respectively
To study the molecular mechanism underlying suppression of Nrf2/ARE-mediated antioxidant protective pathway in hLECs treated with higher doses of SFN, we utilized tools for Motif discovery, MEME. This in silico analysis identified putative Nrf2 binding sites and repressive Klf9 binding sites in Klf9 ( Figure 3A) and hPrdx6 ( Figure 3B) promoters, respectively. We found that Klf9 promoter, spanning from 5 −5856 to +1 bp (TSS), contained four well-conserved antioxidant response elements (ARE, 5 -RGGAYnnnGC-3 ) [13,26], a known binding site for transcription factor, Nrf2 ( Figure 3A). Intriguingly, analysis of Prdx6 promoter, ranging from 5 -1559 to transcription start site (+1) bp, showed that it had five conserved Repressive Klf9 Binding Elements (RKBE; 5 -C A/G CCC-3 ) ( Figure 3B) in the regulatory region of Prdx6 promoter. Moreover, the presence of a functional Nrf2/ARE binding site in Prdx6 promoter has been reported by our group and others [1,82].

In Vivo DNA and Protein Interaction Disclosed that Klf9 Was Transcriptional Target for Nrf2, and SFN Treatment Enhanced Nrf2/ARE Binding in Dose-Dependent Fashion
Nrf2 is not a bona fide regulator of Klf9, as are antioxidant genes. However, it has been shown that Klf9 gene is transcriptionally regulated by Nrf2 during higher levels of oxidative stress [26]. Given the significant induction of Klf9 mRNA and increased nuclear localization of Nrf2 with higher doses of SFN in hLECs, we sought to validate the previous findings [26] and determine whether Nrf2 directly bound to ARE sites present in regulatory region of Klf9 promoter as predicted (Figure3A) in hLECs treated with cytotoxic doses of SFN. We carried out ChIP assay to confirm the occupancy of Nrf2 on ARE of Klf9 gene promoter as designated in Figure 4A. Human LECs treated with SFN (DMSO or 12-24 μM) for 24 h were processed for ChIP assay using antibody specific to Nrf2 or control IgG antibodies as described in Materials and Methods. The immunoprecipitated products were processed for RT-qPCR with primers flanking Nrf2 binding sites (Sites 3 and 4) in Klf9′s regulatory regions ( Figure 4A). As shown in Figures 4B and 4C, we observed a progressive increase in enrichment of Nrf2 on both sites, ARE 3 and ARE 4 sequences, and found that enrichment was SFN dose-dependent, suggesting that higher doses of SFN (beyond survival doses) resulted in the aberrant and unphysiological accumulation of Nrf2 in nucleus, leading to Nrf2 binding to ARE of Klf9 promoter as evidenced by ChIP-qPCR ( Figure 4B and 4C, Black bars). Unexpectedly, Nrf2 failed to bind to predicted sites 1 and 2 ( Figure 4A, data not shown), indicating that these sequences may be not be active sites. No binding was detected with control IgG ( Figure 4B and 4C, gray bars), validating the specificity of Nrf2 antibody used in the experiments. These data indicated that higher doses of SFN augmented enrichment of Nrf2 at ARE sequences, and may be the molecular mechanism underlying upregulation of Klf9 in hLECs treated with toxic doses of SFN, if Nrf2/ARE binding was functional.

TSS +1
XhoI (+30 bps)   Nrf2 is not a bona fide regulator of Klf9, as are antioxidant genes. However, it has been shown that Klf9 gene is transcriptionally regulated by Nrf2 during higher levels of oxidative stress [26]. Given the significant induction of Klf9 mRNA and increased nuclear localization of Nrf2 with higher doses of SFN in hLECs, we sought to validate the previous findings [26] and determine whether Nrf2 directly bound to ARE sites present in regulatory region of Klf9 promoter as predicted ( Figure 3A) in hLECs treated with cytotoxic doses of SFN. We carried out ChIP assay to confirm the occupancy of Nrf2 on ARE of Klf9 gene promoter as designated in Figure 4A. Human LECs treated with SFN (DMSO or 12-24 µM) for 24 h were processed for ChIP assay using antibody specific to Nrf2 or control IgG antibodies as described in Materials and Methods. The immunoprecipitated products were processed for RT-qPCR with primers flanking Nrf2 binding sites (Sites 3 and 4) in Klf9 s regulatory regions ( Figure 4A). As shown in Figure 4B,C, we observed a progressive increase in enrichment of Nrf2 on both sites, ARE 3 and ARE 4 sequences, and found that enrichment was SFN dose-dependent, suggesting that higher doses of SFN (beyond survival doses) resulted in the aberrant and unphysiological accumulation of Nrf2 in nucleus, leading to Nrf2 binding to ARE of Klf9 promoter as evidenced by ChIP-qPCR ( Figure 4B,C, Black bars). Unexpectedly, Nrf2 failed to bind to predicted sites 1 and 2 ( Figure 4A, data not shown), indicating that these sequences may be not be active sites. No binding was detected with control IgG (Figure 4B,C, gray bars), validating the specificity of Nrf2 antibody used in the experiments. These data indicated that higher doses of SFN augmented enrichment of Nrf2 at ARE sequences, and may be the molecular mechanism underlying upregulation of Klf9 in hLECs treated with toxic doses of SFN, if Nrf2/ARE binding was functional. for 24 h, as indicated. ChIP assay was carried out with Nrf2 and control IgG antibodies. Pulled DNA fragments were subjected to qPCR analysis. DNA fragments present in the immunoprecipitation were amplified with primers that specifically recognize a fragment of Klf9 promoter containing ARE sites as indicated. As a negative control, ChIP with control IgG was used. * p < 0.05, ** p < 0.001; SFN treated versus DMSO control. Φ and ΦΦ denotes primer pairs used for ChIP-qPCR. Note: Predicted sites ARE1 and ARE2 by in silico analysis are false positive sites (current study and [26]).

Higher Doses of SFN Activated Klf9 Transcriptional Activity Through Nrf2/ARE Interaction
We next sought to determine whether the observed Nrf2 binding to ARE present in the regulatory region that occurred following excessive doses of SFN in hLECs was functional and could activate Klf9 transcription. The results from in vivo DNA binding (Figure 4) led us to engineer four CAT reporter plasmids of Klf9 promoter: (i) pCAT-Klf9 wild-type (WT); (ii) pCAT-Klf9 mutant (mut) at site 3 (ARE3-mut, TGAYnnnGC; T to G); (iii) pCAT-Klf9 mutant at site 4 (ARE4-mut; T to G) (iv) pCAT-Klf9 mutant at both sites (ARE3-mut + ARE4-mut). To examine the functional consequences of SFN-induced changes on Klf9 transcription, we transfected hLECs with a pCAT-Klf9 WT reporter plasmid (ARE; RTGAYnnnGC) or its mutant plasmids (RGGAYnnnGC) as noted above, along with a GFP plasmid as indicated in Figure 5. These transfectants were treated with SFN (DMSO or 12-24 μM) for 24 h. We found that the transactional activity of mutant plasmids mutated at ARE was significantly reduced compared to wild-type plasmids in response to increasing amount of SFN ( Figure 5B); however, plasmids having mutation at both ARE sites of Klf9 promoter showed robust loss in transcription activity compared to wild-type promoter′s transcriptional activity in response to higher doses of SFN ( Figure 5B). Together, data showed that both ARE sites of Klf9 promoter played a part in activation of Klf9 transcription in response to higher doses of SFN. It was interesting to observe that SFN failed to activate Klf9 transcription at lower concentrations (data not shown) at which SFN activates Nrf2-mediated antioxidant pathway [1]. We surmised that ARE3 and ARE4 of Klf9 have lower affinity than ARE of Prdx6 promoter; however, exploring this phenomenon would ChIP assay was carried out with Nrf2 and control IgG antibodies. Pulled DNA fragments were subjected to qPCR analysis. DNA fragments present in the immunoprecipitation were amplified with primers that specifically recognize a fragment of Klf9 promoter containing ARE sites as indicated. As a negative control, ChIP with control IgG was used. * p < 0.05, ** p < 0.001; SFN treated versus DMSO control. Φ and ΦΦ denotes primer pairs used for ChIP-qPCR. Note: Predicted sites ARE1 and ARE2 by in silico analysis are false positive sites (current study and [26]).

Higher Doses of SFN Activated Klf9 Transcriptional Activity through Nrf2/ARE Interaction
We next sought to determine whether the observed Nrf2 binding to ARE present in the regulatory region that occurred following excessive doses of SFN in hLECs was functional and could activate Klf9 transcription. The results from in vivo DNA binding ( Figure 4) led us to engineer four CAT reporter plasmids of Klf9 promoter: (i) pCAT-Klf9 wild-type (WT); (ii) pCAT-Klf9 mutant (mut) at site 3 (ARE3-mut, TGAYnnnGC; T to G); (iii) pCAT-Klf9 mutant at site 4 (ARE4-mut; T to G) (iv) pCAT-Klf9 mutant at both sites (ARE3-mut + ARE4-mut). To examine the functional consequences of SFN-induced changes on Klf9 transcription, we transfected hLECs with a pCAT-Klf9 WT reporter plasmid (ARE; RTGAYnnnGC) or its mutant plasmids (RGGAYnnnGC) as noted above, along with a GFP plasmid as indicated in Figure 5. These transfectants were treated with SFN (DMSO or 12-24 µM) for 24 h. We found that the transactional activity of mutant plasmids mutated at ARE was significantly reduced compared to wild-type plasmids in response to increasing amount of SFN ( Figure 5B); however, plasmids having mutation at both ARE sites of Klf9 promoter showed robust loss in transcription activity compared to wild-type promoter s transcriptional activity in response to higher doses of SFN ( Figure 5B). Together, data showed that both ARE sites of Klf9 promoter played a part in activation of Klf9 transcription in response to higher doses of SFN. It was interesting to observe that SFN failed to activate Klf9 transcription at lower concentrations (data not shown) at which SFN activates Nrf2-mediated antioxidant pathway [1]. We surmised that ARE3 and ARE4 of Klf9 have lower affinity than ARE of Prdx6 promoter; however, exploring this phenomenon would require a separate line of investigation. Furthermore, to determine whether upregulation of Klf9 is activated only by SFN, or whether other Nrf2 activators at high doses may have the same effect, we treated transfectants containing wild-type pCAT-Klf9 and its ARE mutant at both sites for 6 h with a high dose of tBOOH (200 µM), a well-known activator of Nrf2 [83]. Similar to SFN, treatment with the high dose of tBOOH resulted in a robust increase in Klf9 transcriptional activity, in contrast to mutant plasmid ( Figure 5C), suggesting that the upregulation of Klf9 is not specific to SFN, but can occur with other Nrf2 activators applied at higher or cytotoxic doses.  [83]. Similar to SFN, treatment with the high dose of tBOOH resulted in a robust increase in Klf9 transcriptional activity, in contrast to mutant plasmid ( Figure 5C), suggesting that the upregulation of Klf9 is not specific to SFN, but can occur with other Nrf2 activators applied at higher or cytotoxic doses. Promoter activity was measured. * p < 0.05, ** p < 0.001; SFN treated versus untreated control, # p < 0.05, ## p < 0.001; ARE3-mut or ARE4mut or ARE3-mut+ARE4-mut versus WT, $ p < 0.05, $$ p < 0.001; ARE3-mut+ARE4-mut versus ARE3mut or ARE4-mut.

Treatment of hLECs With Increased Amounts of SFN Mobilized the Klf9 Enrichment on Its Binding Sites Present in Prdx6 Promoter
In silico analyses spotted out the presence of five RKBEs (a single GC boxes) [26,33] in Prdx6 promoter. Recent report has shown that aberrant expression of Klf9 and nuclear accumulation leads to repression of antioxidant genes [26]. Based on this information, we posit if the observed inverse correlation between Prdx6 and aberrant expression and increased accumulation of Klf9 (Figures 2) in response to lethal doses of SFN can be connected to repression of Prdx6 transcription in hLECs; thus, we carried out an in vivo DNA binding experiment. We treated hLECs with increasing amounts of SFN (6 to 24 μM) and performed ChIP assay to measure the occupancy of Klf9 at Prdx6 gene promoter (1559 bp upstream of it transcription start site). As shown in Figure 6, Klf9 occupied the Prdx6 promoter at RKBEs (5′-C A/G CCC-3′), and the increased enrichment of Klf9 to the sequences was SFN concentration-dependent, while no enrichment of Klf9 occurred at 6 μM of SFN. Moreover, no amplicon was detected with control IgG, validating the specificity of Klf9 antibody used in the experiment. Thus, the experiment revealed that Klf9 interacted with all RKBEs of Prdx6 promoter in hLECs were transfected with wild-type Klf9 promoter (−5856/+71) or its mutant at ARE3 (ARE3-mut) or at ARE4 (ARE4-mut) and at both ARE3 plus ARE4 (ARE3-mut+ARE4-mut) sites. After 48 h, transfectants were treated with DMSO control or different concentration of SFN (12, 18 and 24 µM) for 24 h (B) and tBOOH (200 µM) for 6 h (C) as shown, and Prdx6 Promoter activity was measured. * p < 0.05, ** p < 0.001; SFN treated versus untreated control, # p < 0.05, ## p < 0.001; ARE3-mut or ARE4-mut or ARE3-mut+ARE4-mut versus WT, $ p < 0.05, $$ p < 0.001; ARE3-mut+ARE4-mut versus ARE3-mut or ARE4-mut.

Treatment of hLECs with Increased Amounts of SFN Mobilized the Klf9 Enrichment on Its Binding Sites Present in Prdx6 Promoter
In silico analyses spotted out the presence of five RKBEs (a single GC boxes) [26,33] in Prdx6 promoter. Recent report has shown that aberrant expression of Klf9 and nuclear accumulation leads to repression of antioxidant genes [26]. Based on this information, we posit if the observed inverse correlation between Prdx6 and aberrant expression and increased accumulation of Klf9 ( Figure 2) in response to lethal doses of SFN can be connected to repression of Prdx6 transcription in hLECs; thus, we carried out an in vivo DNA binding experiment. We treated hLECs with increasing amounts of SFN (6 to 24 µM) and performed ChIP assay to measure the occupancy of Klf9 at Prdx6 gene promoter (1559 bp upstream of it transcription start site). As shown in Figure 6, Klf9 occupied the Prdx6 promoter at RKBEs (5 -C A/G CCC-3 ), and the increased enrichment of Klf9 to the sequences was SFN concentration-dependent, while no enrichment of Klf9 occurred at 6 µM of SFN. Moreover, no amplicon was detected with control IgG, validating the specificity of Klf9 antibody used in the experiment. Thus, the experiment revealed that Klf9 interacted with all RKBEs of Prdx6 promoter in vivo as predicted through bioinformatics tool, and the progressive increase in the binding of Klf9/RKBE was SFN dose-dependent, as indicated in Figure 6. vivo as predicted through bioinformatics tool, and the progressive increase in the binding of Klf9/RKBE was SFN dose-dependent, as indicated in Figure 6.

Transcriptional Assay Disclosed that Klf9/RKBE Interaction Transcriptionally Repressed Its New Target Gene, Prdx6 during High Doses of SFN
To examine the functionality of the dose-dependent SFN-induced increased Klf9 binding to RKBE on Prdx6 transcription, we performed transactivation assay in hLECs containing Prdx6 promoter fused to reporter gene CAT, and treated cells with the same increasing doses of SFN as shown in Figure 6. We constructed Prdx6 promoter linked to CAT reporter gene plasmid having Klf9 sites ( Figure 7A), and mutated sites, RKBE1, RKBE3 and RKBE1 + RKBE3 and mutated them using site directed mutagenesis (SDM) to test the functionality of binding shown in Figure 6. pCAT-Prdx6 Wild type (WT) or its mutants at RKBE1 (RKBE1-mut) or RKBE3 (RKBE3-mut) or at both sites (RKBE1-mut+ RKBE3-mut) of Prdx6 promoter plasmids were transfected into hLECs. 48h later transfectants were treated with different concentrations of SFN (for 24h) or tBOOH (6h) as shown in Figure 7. Transcriptional activity of pCAT-Prdx6 WT was dramatically reduced with increasing doses of SFN treatment, but was significantly restored in mutant plasmids, and pCAT-Prdx6-RKBE1-mut plus RKBE3-mut plasmid showed a further increase in transcriptional activity than pCAT-Prdx6-RKBE1-mut or RKBE3-mut as indicated in Figure 7. Those observations suggest that observed Klf9/RKBE binding in Prdx6 promoter ( Figure 6) were functional, and loss of Prdx6 expression in hLECs treated with toxic doses of SFN was caused by Klf9/RKBE present in regulatory region of Prdx6 promoter. Next, we tested the effect of another known Nrf2 activator, where tBOOH acts through ARE, on the transcriptional activity of Prdx6. Similar to the SFN experiments noted above, transfectants containing mutant reporter plasmids mutated either at one RKBE site (RKBE1-mut or RKBE3-mut) or both sites (RKBE1-mut plus RKBE3-mut) and provided a similar pattern of Prdx6 transcription as observed in SFN experiments in cells treated with toxic dose of tBOOH (200μM for 6h) ( Figure 7C, gray bars versus black bars), showing a significant revival of transcriptional activity in mutant Prdx6 promoters mutated at RKBE as seen in Figure 7C. Nevertheless, data demonstrated that mutation at the above-noted two sites did not completely eliminate SFN-driven Klf9-mediated suppression of Prdx6 promoter activity, pointing to the contribution of three other sites (RKBE2, RKBE4 and RKBE5) in Prdx6 repression. Collectively, data revealed that RKBE negatively regulated Prdx6 expression through repressing its transcription in cells facing SFN-induced toxicity. This also

Transcriptional Assay Disclosed that Klf9/RKBE Interaction Transcriptionally Repressed Its New Target Gene, Prdx6 during High Doses of SFN
To examine the functionality of the dose-dependent SFN-induced increased Klf9 binding to RKBE on Prdx6 transcription, we performed transactivation assay in hLECs containing Prdx6 promoter fused to reporter gene CAT, and treated cells with the same increasing doses of SFN as shown in Figure 6. We constructed Prdx6 promoter linked to CAT reporter gene plasmid having Klf9 sites ( Figure 7A), and mutated sites, RKBE1, RKBE3 and RKBE1 + RKBE3 and mutated them using site directed mutagenesis (SDM) to test the functionality of binding shown in Figure 6. pCAT-Prdx6 Wild type (WT) or its mutants at RKBE1 (RKBE1-mut) or RKBE3 (RKBE3-mut) or at both sites (RKBE1-mut+ RKBE3-mut) of Prdx6 promoter plasmids were transfected into hLECs. 48h later transfectants were treated with different concentrations of SFN (for 24h) or tBOOH (6h) as shown in Figure 7. Transcriptional activity of pCAT-Prdx6 WT was dramatically reduced with increasing doses of SFN treatment, but was significantly restored in mutant plasmids, and pCAT-Prdx6-RKBE1-mut plus RKBE3-mut plasmid showed a further increase in transcriptional activity than pCAT-Prdx6-RKBE1-mut or RKBE3-mut as indicated in Figure 7. Those observations suggest that observed Klf9/RKBE binding in Prdx6 promoter ( Figure 6) were functional, and loss of Prdx6 expression in hLECs treated with toxic doses of SFN was caused by Klf9/RKBE present in regulatory region of Prdx6 promoter. Next, we tested the effect of another known Nrf2 activator, where tBOOH acts through ARE, on the transcriptional activity of Prdx6. Similar to the SFN experiments noted above, transfectants containing mutant reporter plasmids mutated either at one RKBE site (RKBE1-mut or RKBE3-mut) or both sites (RKBE1-mut plus RKBE3-mut) and provided a similar pattern of Prdx6 transcription as observed in SFN experiments in cells treated with toxic dose of tBOOH (200µM for 6h) ( Figure 7C, gray bars versus black bars), showing a significant revival of transcriptional activity in mutant Prdx6 promoters mutated at RKBE as seen in Figure 7C. Nevertheless, data demonstrated that mutation at the above-noted two sites did not completely eliminate SFN-driven Klf9-mediated suppression of Prdx6 promoter activity, pointing to the contribution of three other sites (RKBE2, RKBE4 and RKBE5) in Prdx6 repression. Collectively, data revealed that RKBE negatively regulated Prdx6 expression through repressing its transcription in cells facing SFN-induced toxicity. This also suggests that the advent of Klf9-mediated dominant repressive signaling in hLECs in response to high doses of SFN that surpasses Nrf2-mediated activation of Prdx6. Results unveiled the molecular mechanism involved in SFN-mediated cytotoxicity, which is linked to attenuation of Prdx6 by Nrf2/Klf9 axis as shown in Figures 1 and 2. suggests that the advent of Klf9-mediated dominant repressive signaling in hLECs in response to high doses of SFN that surpasses Nrf2-mediated activation of Prdx6. Results unveiled the molecular mechanism involved in SFN-mediated cytotoxicity, which is linked to attenuation of Prdx6 by Nrf2/Klf9 axis as shown in Figures 1 and 2. (RKBE1-mut) or at RKBE3 (RKBE3-mut) or RKBE1 plus RKBE3 (RKBE1-mut+ RKBE3-mut). After 48 h, transfectants were treated with different concentration of SFN for 24 h (B) and tBOOH (C) for 6 h, as indicated. Promoter activity is measured and data represent mean ± S.D. values of three independent experiments. *p < 0.05, **p < 0.001; SFN treated versus untreated control, # p < 0.05, ## p < 0.001; RKBE1-mut or RKBE3-mut or RKBE1-mut+ RKBE3-mut versus WT, $$ p < 0.001; RKBE1-mut+ RKBE3-mut versus RKBE1-mut or RKBE3-mut.

Klf9-deficient hLECs Showed Resistance against Treatment with Toxic Doses of SFN
SFN is a major activator with diversified functions, including a cytoprotective effect. Its protective activity depends on the amount of SFN applied to cells and on cell types [1,63,67]. Results from the current study revealed that the Nrf2/Klf9 axis played an adverse role by repressing the protective antioxidant gene Prdx6 in response to high doses of SFN. Recently, several reports have noted that SFN application can halt or abate several diseases and their complications [1,[84][85][86][87][88][89][90][91][92][93], indicating that a dose of SFN is indispensable to Nrf2/ARE-driven protective pathway. Because of its effectiveness in cytoprotection and the cytotoxicity of its higher doses via aberrant expression of Klf9, we used a Klf9 knockdown strategy to examine if Klf9-depleted cells can resist increasing doses of SFN. Our goal was to provide a proof of concept for a potential basis for combination therapy (SFN plus Klf9-ShRNA). As described in Materials and Methods, we stably infected hLECs using lentiviral (LV) Sh-Klf9 or LV Sh-control ( Figure 8A), and examined the effect of Klf9 depletion on SFNmediated cytotoxicity. As shown in Figure 8BI, BII, and 8C, RT-qPCR and immunoblot analyses revealed that LV Sh-Klf9 significantly knocked down the Klf9 expression in hLECs infected with LV Sh-Klf9. It was intriguing to observe that deficiency of Klf9 was directly linked to upregulation of Prdx6 mRNA and protein, corroborating that Klf9 has played a part in repression of Prdx6 expression. Next, we treated these tranfectants with increasing toxic doses of SFN for 24 h (Figure 8D and 8E) and assessed the levels of ROS and cell survival. ROS assay and viability assay revealed that toxic doses of SFN were significantly less effective at inducing ROS-evoked toxicity in Klf9-deficient cells, and survival of these cells was increased ( Figure 8E, black bars versus gray bars) with significant reduction in ROS levels, demonstrating that aberrant expression of Klf9 was a major cause of SFNmediated cytotoxicity. Thus, we propose that SFN application along with Klf9, ShRNA should be hLECs were transfected with Prdx6 (−1559 / + 30) wild type (WT) plasmid construct or its mutants at RKBE1 (RKBE1-mut) or at RKBE3 (RKBE3-mut) or RKBE1 plus RKBE3 (RKBE1-mut+ RKBE3-mut). After 48 h, transfectants were treated with different concentration of SFN for 24 h (B) and tBOOH (C) for 6 h, as indicated. Promoter activity is measured and data represent mean ± S.D. values of three independent experiments. * p < 0.05, ** p < 0.001; SFN treated versus untreated control, # p < 0.05, ## p < 0.001; RKBE1-mut or RKBE3-mut or RKBE1-mut+ RKBE3-mut versus WT, $$ p < 0.001; RKBE1-mut+ RKBE3-mut versus RKBE1-mut or RKBE3-mut.

Klf9-Deficient hLECs Showed Resistance against Treatment with Toxic Doses of SFN
SFN is a major activator with diversified functions, including a cytoprotective effect. Its protective activity depends on the amount of SFN applied to cells and on cell types [1,63,67]. Results from the current study revealed that the Nrf2/Klf9 axis played an adverse role by repressing the protective antioxidant gene Prdx6 in response to high doses of SFN. Recently, several reports have noted that SFN application can halt or abate several diseases and their complications [1,[84][85][86][87][88][89][90][91][92][93], indicating that a dose of SFN is indispensable to Nrf2/ARE-driven protective pathway. Because of its effectiveness in cytoprotection and the cytotoxicity of its higher doses via aberrant expression of Klf9, we used a Klf9 knockdown strategy to examine if Klf9-depleted cells can resist increasing doses of SFN. Our goal was to provide a proof of concept for a potential basis for combination therapy (SFN plus Klf9-ShRNA). As described in Materials and Methods, we stably infected hLECs using lentiviral (LV) Sh-Klf9 or LV Sh-control ( Figure 8A), and examined the effect of Klf9 depletion on SFN-mediated cytotoxicity. As shown in Figure 8B(I,II),C, RT-qPCR and immunoblot analyses revealed that LV Sh-Klf9 significantly knocked down the Klf9 expression in hLECs infected with LV Sh-Klf9. It was intriguing to observe that deficiency of Klf9 was directly linked to upregulation of Prdx6 mRNA and protein, corroborating that Klf9 has played a part in repression of Prdx6 expression. Next, we treated these tranfectants with increasing toxic doses of SFN for 24 h (Figure 8D,E) and assessed the levels of ROS and cell survival. ROS assay and viability assay revealed that toxic doses of SFN were significantly less effective at inducing ROS-evoked toxicity in Klf9-deficient cells, and survival of these cells was increased ( Figure 8E, black bars versus gray bars) with significant reduction in ROS levels, demonstrating that aberrant expression of Klf9 was a major cause of SFN-mediated cytotoxicity. Thus, we propose that SFN application along with Klf9, ShRNA should be considered as a novel strategy to treat Nrf2/ARE dysregulation-mediated diseases during aging or oxidative stress.

Discussion
The redox stress-sensitive, Nrf2-antioxidant response element (ARE)-mediated antioxidant pathway is the major cellular defense against several internal and external environmental stresses. Oxidative stress and many compounds with electrophilic properties, such as heavy metals, tBOOH and naturally occurring phytochemicals such as SFN, have been reported to yield therapeutic benefits by modulating Nrf2-ARE pathway [1,27,67,[94][95][96][97][98]. At physiological conditions, Nrf2 is controlled via Keap1, an adopter protein for CULLI3-based ubiquitin E3 ligase that ubiquitinates Nrf2 for proteasomal degradation. During exposure to electrophiles (such as SFN) Keap1 is inactivated due to binding of SFN and that subsequently leads to dislodging of Nrf2 from Keap1, and allows Nrf2 to escape from degradation and translocalizes into the nucleus, where it activates antioxidant genes [5,99]. In recent years, several studies have shown that SFN has a cytoprotective function with

Discussion
The redox stress-sensitive, Nrf2-antioxidant response element (ARE)-mediated antioxidant pathway is the major cellular defense against several internal and external environmental stresses. Oxidative stress and many compounds with electrophilic properties, such as heavy metals, tBOOH and naturally occurring phytochemicals such as SFN, have been reported to yield therapeutic benefits by modulating Nrf2-ARE pathway [1,27,67,[94][95][96][97][98]. At physiological conditions, Nrf2 is controlled via Keap1, an adopter protein for CULLI3-based ubiquitin E3 ligase that ubiquitinates Nrf2 for proteasomal degradation. During exposure to electrophiles (such as SFN) Keap1 is inactivated due to binding of SFN and that subsequently leads to dislodging of Nrf2 from Keap1, and allows Nrf2 to escape from degradation and translocalizes into the nucleus, where it activates antioxidant genes [5,99]. In recent years, several studies have shown that SFN has a cytoprotective function with diversified activities [26,63]. Nevertheless, studies have shown that SFN does not act as direct antioxidant, but acts via upregulating the Nrf2/ARE pathway [27,100]. SFN may also induce the phosphorylation of Nrf2 by activating various kinases, where it can alter nuclear and cytoplasmic trafficking as well as Nrf2 integrity [101]. In maintaining redox-homeostasis in the cellular microenvironment, Nrf2 expression and activity are tightly regulated [21]. However, aberrant expression and activation of Nrf2, in response to increased doses of electrophiles or oxidative load, are known to induce adverse signaling, failure of redox homeostasis and cell death [1,26,63,[102][103][104]. This suggests that the diversified activity of Nrf2/ARE-mediated response is connected to its regulation in the cellular microenvironment and doses of Nrf2 activators; at low doses, most effectors of Nrf2/ARE pathway provide cytoprotection and higher doses are cytotoxic under the conditions of cell background and treatment regimens. However, studies of the molecular mechanisms involved in the dose-dependent switch from cell survival to cell death have been limited, and the mechanisms have not been investigated in lens cells. In recent work, we observed that increase amounts of SFN induced cytotoxicity, while nontoxic doses enhanced cell survival by activating Nrf2/ARE/Prdx6 pathway (current study and [1]). However, higher doses of SFN are known to cause cell death in various cell types and to inhibit tumor cell growth [98,[105][106][107][108]. Using hLECs and Nrf2/Prdx6 axis as a model system, in the present study, we found that toxic doses of SFN dramatically enhanced Nrf2 expression and its nuclear accumulation, which was correlated with increased expression and nuclear accumulation of Klf9 and reduced expression of Prdx6 transcripts and protein (Figures 1 and 2). We thought this might be associated with reduced DNA binding efficiency of Nrf2 to Prdx6 gene promoter [1,9,109], but experimentation ruled out that possibility. Conversely, data showed that Nrf2 s DNA binding efficacy increased with increased doses of SFN ( Figure 2C). Zucker et al. [26] demonstrated that over-accumulation of Nrf2 in nucleus targets Klf9 and significantly upregulates its transcription by binding to ARE present in Klf9 promoter, which in turn binds to GC box sequences present in some of antioxidant genes and represses their transcription. Intriguingly, in silico analysis of Prdx6 gene promoter ( Figure 3B) coupled with DNA binding and transactivation assays disclosed that repression of Prdx6 is linked to Klf9 in cells treated with increased amounts SFN (Figures 4-8), emphasizing Klf9-mediated repressive signaling is dominant over Nrf2-mediated activation of Prdx6 in response to toxic doses of SFN. Klf9 is a ubiquitously expressed protein, with functions, which have been little explored in lens cells. Our data indicated that Klf9 could cause transcription repression of Prdx6 gene in hLECs exposed to toxic amount of SFN. Prdx6 is a powerful antioxidant protective protein, which plays a critical role in the antioxidant defense system [1,7,43,44,52,57,58,72,73,[76][77][78][79]. Prdx6 protects cells by regulating ROS levels and thereby maintains cellular homeostasis [1,52]. Prdx6 is abundantly expressed in eye, lungs, brain and other organs [1,6,7,[56][57][58]79]. The protein maintains cell integrity by reducing peroxidized phospholipids in cell membrane, protecting DNA damage and controlling survival signaling during aging and oxidative stress [110]. Prdx6 localizes in almost all ROS producing organelles, such as mitochondria, endoplasmic, lysosome and plasma membrane, suggesting its importance in the cellular antioxidant defense system.
In a previous publication, we reported that SFN, a known classic activator of Nrf2, enhances antioxidant proteins such as Prdx6, catalase and GSTπ in human, mouse and rat LECs in a concentration-dependent manner [1]. Other studies found that pretreatment of mouse fibroblasts or LECs with nontoxic concentrations of SFN (3-6 µM) increased their resistance against oxidative stress caused by paraquat, hydrogen peroxide or UVB (ultra violet-B) in an Nrf2-dependent manner [1,81]. As shown in Figure 1B, a nontoxic (6 µM) concentration of SFN enhanced hLECs viability, while toxic/lethal concentrations (12-24 µM) of SFN caused cell death that was directly associated with increased ROS generation ( Figure 1A, black bars). Doses above 10 µM of SFN concentration has been reported to reduce cell viability and cause cell death in FHL124 cells [111]. Up to 10 µM of SFN enhances the viability of N9 cells, while 20 µM and 50 µM concentrations increase cell death [112]. Moreover, toxic doses of SFN have been shown to increase Nrf2 activity [112] leading to Klf9 upregulation. Our current data is consistent with a previous report [26] revealing that increased nuclear abundance of Nrf2 is connected to increased Klf9 levels and reduced expression of Prdx6 mRNA and protein in hLECs in response to high doses of SFN (Figure 2A,B). SFN has been found to enhance Nrf2/ARE-mediated antioxidant genes mRNA and protein expression in cells from C57BL/6J mice and humans, and in mouse, rat and human LECs [1,113]. SFN at high concentrations is known to cause cell death due to oxidative stress [26,98]. In this work, we sought to determine whether similar Nrf2/Klf9-mediated injurious signaling occurs in lens cells in treated with increased doses of SFN. Published reports and the current study have shown that low doses of SFN increase the levels of antioxidant molecules without inducing Klf9 [81]. A Klf9-dependent feed-forward regulatory loop has been shown to trigger ROS-evoked cell death [114,115]. Collectively, our data revealed a model of Klf9-dependent feed-forward regulation of ROS that occurred through Nrf2 upregulation of Klf9 and Klf9-mediated repression of Prdx6 in LECs treated with increased amount of SFN. At low concentrations of SFN, Nrf2-dependent antioxidant is activated, but at higher concentrations, Nrf2 continues to accumulate in the nucleus, leading to Nrf2 binding to the Klf9 promoter, increased Klf9 transcription, and further increases in ROS levels (Figures 1 and 8). Previously, we have reported that Prdx6 depletion causes increased vulnerability to oxidative stressors-induced cell death. In addition, we observed that other cellular antioxidants were ineffective at protecting LECs [73,76,116], demonstrating the importance of Prdx6 for cellular survival and cytoprotection. Nrf2 activation of Prdx6 provides cytoprotection by limiting ROS. Intracellular ROS are generated in cells by nonenzymatic and enzymatic process such as superoxide-dismutase (SOD)-catalyzed disproportionation of the superoxide radicals to H 2 O 2 and by redox cycling. We believe that exposure of a toxic dose of SFN or oxidants can contribute substantially to the cellular steady levels of H 2 O 2 and the super oxide radicals due to reduced availability of Prdx6 caused by Nrf2/Klf9 axis-mediated Prdx6 suppression. Recently, it has been reported that Nrf2 can modulate ROS generation [117]. However, how redox balance is maintained following electrophiles exposure is cumbersome to explain at this time point, and thus, further work is warranted. Nonetheless, constitutive production of H 2 O 2 is derived from mitochondria dependent on nicotinamide adenine dinucleotide (NADH) [118] and activity of nicotinamide adenine dinucleotide phosphate (NADPH) oxidase (NOX) enzymes dependent upon NADPH [119,120]. Aberrant activity of these enzymes has been suggested to contribute to the onset of many diseases linked to oxidative stress. Prdx6 is localized in ROS-producing organelles, such as mitochondria, endoplasmic reticulum and peroxisomes, indicating that Prdx6 deficiency in these organelles can disorient ROS-mediated survival signaling leading to oxidative stress-induced cellular damage or cell death. Furthermore, recently, Klf9 has been identified as a new target gene for Nrf2 [26,121,122]. The Klf9 regulatory region bearing active ARE3 and ARE4 ( Figures 3A, 4 and 5) has lesser affinity for Nrf2 than the ARE present in antioxidant genes [26]. We observed similar phenomena, as Figures 4 and 5 show increased functional binding of Nrf2 to ARE of Klf9 in response to lethal doses of SFN.
Klf9 belongs to the KLF transcription factor family and has diversified roles in oncogenesis, cell differentiation and development and cell death [26,38]. Klf9 acts as both a transcription activator and repressor, and in doing so, controls gene expression. Klf9 activates promoters having repeated GC boxes like Sp1, but represses transcription of genes containing a single GC box sequence [33]. Our in vivo DNA-binding and transactivation assays revealed that, indeed, Klf9 acted as repressor of Prdx6 expression. We found that Prdx6 genes bear five Klf9 repressive binding sites (RKBE; C A/G CCC), and these sites were not repeated GC boxes, but were single GC box sequences, demonstrating that Klf9 acted as repressor for Prdx6 by binding to RKBE 1, 2, 3, 4 and 5 present in its gene promoter during toxic doses of SFN (Figures 6 and 7). Interestingly, using knockdown assay, we found that Klf9 deficiency in hLECs enhanced Prdx6 expression and reduced levels of ROS, and these cells survived well in response to increased doses of SFN ( Figure 8). These data reveal the molecular mechanism of SFN-mediated diversified function, and demonstrate that Klf9 s cellular level is the major factor determining dose-dependent SFN-mediated activity. Furthermore, SFN has been shown to influence the activity and expression of various transcription factors such as NF-кB [123], Ap1 [124], Sp1 [125,126] and Nrf2 [127] and so forth, by modulating cellular signaling, including antioxidant expression. However, our study found that increased ROS-induced toxic doses of SFN play a part in repression of Prdx6 via the Nrf2/Klf9 axis as shown in Figure 9.

Conclusions
In conclusion, our work revealed, for the first time, the novel molecular mechanism underlying how SFN-induced Nrf2-mediated protective signaling goes awry in hLECs in response to increased doses of Nrf2 activator, SFN. We showed that during increased doses of SFN, Klf9-mediated repression of Prdx6 expression is a major event, resulting in acceleration of ROS accumulation and cell death. Thus, dose levels of SFN determine whether it has beneficial or harmful effects, with Nrf2/Klf9/Prdx6 axis being pivotal for SFN activities. Based upon our data, it should be of prime importance to evaluate doses of SFN or other Nrf2 activators, which do not evoke Klf9/Prdx6 axis (mediated adverse signaling). We propose that a therapy combining SFN with a Klf9 ShRNA would be an ideal approach to abating or treating disorders linked to aging and oxidative stressors.
Author Contributions: D.P.S. designed and conceived the experiments. D.P.S., B.C. and E.K. performed the experiments and D.P.S., B.C. and E.K. conducted data analyses. B.C. and D.P.S. wrote the main manuscript and E.K. reviewed and provided comments to make editing. D.P.S. supervised the work and provided funding. All authors read and provided inputs on the paper, and approved the final version.
Funding: This study was supported by the National Eye Institute, NIH EY024589 (to D.P.S.). Klf9 represses Prdx6 expression by binding to repressive binding element(s) (RKBE) in the Prdx6 gene promoter, resulting in increased ROS-mediated cell death. (C) Klf9 can be depleted by use of its Sh-Klf9, eliminating the adverse signaling caused by lethal doses of SFN. This suggests that Klf9 upregulation is an important molecular mechanism underlying SFN-induced ROS-mediated cell injury.

Conclusions
In conclusion, our work revealed, for the first time, the novel molecular mechanism underlying how SFN-induced Nrf2-mediated protective signaling goes awry in hLECs in response to increased doses of Nrf2 activator, SFN. We showed that during increased doses of SFN, Klf9-mediated repression of Prdx6 expression is a major event, resulting in acceleration of ROS accumulation and cell death. Thus, dose levels of SFN determine whether it has beneficial or harmful effects, with Nrf2/Klf9/Prdx6 axis being pivotal for SFN activities. Based upon our data, it should be of prime importance to evaluate doses of SFN or other Nrf2 activators, which do not evoke Klf9/Prdx6 axis (-mediated adverse signaling). We propose that a therapy combining SFN with a Klf9 ShRNA would be an ideal approach to abating or treating disorders linked to aging and oxidative stressors.
Author Contributions: D.P.S. designed and conceived the experiments. D.P.S., B.C. and E.K. performed the experiments and D.P.S., B.C. and E.K. conducted data analyses. B.C. and D.P.S. wrote the main manuscript and E.K. reviewed and provided comments to make editing. D.P.S. supervised the work and provided funding. All authors read and provided inputs on the paper, and approved the final version.
Funding: This study was supported by the National Eye Institute, NIH EY024589 (to D.P.S.).