Sulforaphane reactivates cellular antioxidant defense by inducing Nrf2/ARE/Prdx6 activity during aging and oxidative stress

Upon oxidative stress and aging, Nrf2 (NFE2-related factor2) triggers antioxidant defense genes to defends against homeostatic failure. Using human(h) or rat(r) lens epithelial cells (LECs) and aging human lenses, we showed that a progressive increase in oxidative load during aging was linked to a decline in Prdx6 expression. DNA binding experiments using gel-shift and ChIP assays demonstrated a progressive reduction in Nrf2/ARE binding (−357/−349) of Prdx6 promoter. The promoter (−918) with ARE showed a marked reduction in young vs aged hLECs, which was directly correlated to decreased Nrf2/ARE binding. A Nrf2 activator, Sulforaphane (SFN), augmented Prdx6, catalase and GSTπ expression in dose-dependent fashion, and halted Nrf2 dysregulation of these antioxidants. SFN reinforced Nrf2/DNA binding and increased promoter activities by enhancing expression and facilitating Nrf2 translocalization in nucleus. Conversely, promoter mutated at ARE site did not respond to SFN, validating the SFN-mediated restoration of Nrf2/ARE signaling. Furthermore, SFN rescued cells from UVB-induced toxicity in dose-dependent fashion, which was consistent with SFN’s dose-dependent activation of Nrf2/ARE interaction. Importantly, knockdown of Prdx6 revealed that Prdx6 expression was prerequisite for SFN-mediated cytoprotection. Collectively, our results suggest that loss of Prdx6 caused by dysregulation of ARE/Nrf2 can be attenuated through a SFN, to combat diseases associated with aging.


Results
Age-related increased oxidative load in LECs was linked to progressive decline in Nrf2, Cat and Prdx6 expression. To identify age-related changes in ROS production and the connection between expression of Prdx6 and its regulator Nrf2, an antioxidant defense pathway, we monitored the intracellular redox-state of primary hLECs of different ages cultured in 96 well plate by using H2-DCF-DA dye 8,43 . Quantification by staining with H2-DCFH-DA dye revealed an age-dependent progressive increase in ROS levels (Fig. 1A), and a higher abundance of ROS was noted in aged hLECs (Fig. 1A, 52 y onward) 44 . Figure 1A reflects the ROS levels in pooled samples of LECs derived from lenses of different age groups as described in the Methods and figures Legends section. This result prompted us to manipulate experiments to maximize the limited supply of primary hLECs. To discern if the apparent increase in ROS levels during aging was due to loss of Nrf2/Prdx6, mRNA from the same group of lenses/hLECs of different ages was isolated and was quantified with qPCR. Data analysis revealed that lens/hLECs mRNA expression of Prdx6, Cat and Nrf2 declined with aging, and this loss was more significant in aged cells (Fig. 1B-D). As expected, we found a significant inverse correlation between expression of Nrf2/Prdx6 and increased ROS levels during aging.
Binding of Nrf2 to ARE in Prdx6 was functionally dysregulated with aging. We next examined whether the age-related decline of Prdx6 mRNA is associated with loss of Nrf2 binding to ARE present in the Prdx6 promoter. We carried out gel-shift assay with nuclear fraction of hLECs directly detached from lens (to avoid cell culture effects) selected from the same age group of lenses that were used in the previous experiment (Fig. 1A). As shown in Fig. 2A, we found progressively reduced Nrf2 binding to 32 p-oligonucleotide containing ARE during aging. The lowest level of binding was displayed with nuclear fraction from older hLECs ( Fig. 2A). Furthermore, Nrf2 depletion experiment revealed reduced or no binding to probe in gel-shift experiments (Fig. 2B, right panel; ages 26 y, 52 y and 66 y) compared to the control (Fig. 2B, left panel; 26 y, 52 y and 66 y), demonstrating that Nrf2 specifically bound to probe and formed complex (Nrf2/DNA). Next we tested the functionality of Nrf2 binding by using transactivation assay. Cultured primary hLECs of different ages were transfected with Prdx6 promoter containing ARE sequences (Fig. 2C, top panel). We observed a significant decline in Prdx6 promoter activity in aging cells (Fig. 2C, gray vs black bar), which was directly related to the decline in Nrf2 binding to ARE. Collectively, our results demonstrated the functional loss of Nrf2's activity in aging.

Sulforaphane induced Nrf2-dependent ARE-antioxidant gene transcripts in LECs.
Based on the decline in Nrf2′s expression and DNA binding ability (Figs 1 and 2), we sought to determine whether SFN would stimulate basal levels of antioxidant gene expression in LECs. Because using primary hLECs was cumbersome due to their limited availability, we utilized the SRA-hLECs and, to generalize our findings, we included primary rLECs as a model system. We first determined an effective noncytotoxic concentration of SFN as indicated in Figs 3A and 4A. Cell growth assessed at 24 h of treatment showed that concentrations of 3 µM and 6 µM and 2.4 µM and 4.8 μM had better effects on SRA-hLECs and rLECs growth, respectively. Thus, these doses were used throughout the study unless otherwise stated.
To examine the efficacy of SFN in inducing expression of antioxidants Prdx6, Cat and phase 2 detoxifying enzyme GSTπ from their basal expression in LECs, SRA-hLECs treated with SFN for 6 h and 24 h were processed for qPCR. Basal transcription of these genes was dramatically increased in SFN-treated cells, as evidenced by increased mRNA levels (Fig. 3B,D,F; Open vs gray and black bars; gray vs black bar). In another set of SFN-treated SRA-hLECs, cellular extracts immunoblotted with anti-Prdx6, anti-Cat and anti-GSTπ antibodies revealed significantly increased expression of all three proteins, The maximum expression level was detected at 6 µM of SFN concentration (Fig. 3C,E,G; Black bars), consistent with increased expression of mRNA ( Fig. 3B,D,F).
Antioxidant response can differ in cell types of different genetic backgrounds. Thus, we next examined whether the results obtained in SRA-hLECs were reproducible in primary rLECs. We found that rLECs treated with SFN (2.4 µM and 4.8 µM) for 6 h and 24 h had similar increased expression patterns of transcripts of all three molecules (Fig. 4B,D,F) as observed with SRA-hLECs. Next we examined the levels of Prdx6, Cat and GSTπ protein in rLECs treated with SFN. Immunoblot analysis with their corresponding specific antibodies showed increased protein expression in SFN-treated cells (Fig. 4C,E,G). These data demonstrate that SFN activated the genes expression by enhancing their transcription in both SRA-hLECs and rLECs. Aging/aged hLECs displayed increased accumulation of ROS, which was associated with progressive decline in Prdx6, Cat and Nrf2 expression. (A) Excessive accumulation of ROS in aging/aged hLECs. Primary hLECs isolated from lenses of different ages were divided into six groups: 16-21 y (n = 6); 24-26 y (n = 6); 34-36 y (n = 4); 52-58 y (n = 6); 62-68 y (n = 12); 75 y (n = 4). Cells were cultured in 96 well plate (5000/ well), and ROS were quantified using H2-DCF-DA dye assay as shown. Data represent the mean ± S.D. of two independent experiments. 16-21 y vs 24-26 y, 34-36 y, 52-58 y, 62-68 y and 75 y (aging samples); *p < 0.001. (B-D) Aging/aged hLECs showing a significant loss of Prdx6, Cat and Nrf2. Total RNA was isolated from hLECs and human lenses of different ages as indicated and was processed for real-time PCR analysis. # LECs directly detached from lenses and were used for assays to avoid cell culture effects. The data represent the mean ± S.D. from three independent experiments. p values were determined for younger vs aging samples. *p < 0.001.

SFN activated Nrf2 transcription and reinforced its translocalization into nucleus.
To establish the molecular mechanism of the Nrf2 activation in LECs, we examined the time-and dose-dependent effect of SFN regulation of Nrf2 expression by using the same concentrations of SFN and durations of treatment which had been found effective in activating antioxidant genes (Figs 3 and 4). SRA-hLECs were treated with SFN as shown in Fig. 5A and qPCR was conducted. The mRNA levels of Nrf2 increased with SFN treatment (Fig. 5A), emphasizing that Nrf2 can be an activator of its own transcription as previously reported 45 . Next we examined the Nrf2 protein level in cytosolic and nuclear extracts of SRA-hLECs treated with different concentrations of SFN for 6 h (the time at which mRNA was at its peak). Immunoblot data using anti-Nrf2 antibody revealed that Nrf2 migrated at approximately 110 kDa-in SDS-PAGE, which was enriched in nuclear extract of SRA-hLECs, and maximum accumulation occurred at 6 μM of SFN concentration as shown in Fig. 5B. Conversely, cytosolic extract had a residual minimal amount of Nrf2 protein. However, Western analysis revealed more than two faint bands. A knockdown experiment (shNrf2) coupled with immunoblotting with anti-Nrf2 antibody ( Supplementary Fig.  S1A) revealed that the band with strong density shown in Fig. 5B was specific to Nrf2. We also observed the presence of Nrf2 in nuclear fraction of untreated control SRA-hLECs. This argues that a low level of Nrf2 in nuclear fraction of LECs may be necessary for basal expression of protective genes in favor of maintaining cellular activity.
Next, to discern activation of the antioxidant response after SFN treatment, we examined cellular and subcellular changes in Nrf2 disposition in rLECs. In untreated control cells, Nrf2 was present at very low levels in whole cell extract and was barely detectable, demonstrating that ongoing proteasomal degradation machinery was active during normal physiological conditions (Fig. 5C, panel a and b) 18,46 . We observed that cells treated with Figure 2. Aging hLECs displayed a significant loss in Nrf2 binding to ARE and in transactivating Prdx6 promoter activity. (A) Gel-shift with nuclear extract from lenses of variable ages shows age-related loss of Nrf2 binding to ARE in Prdx6 promoter. Nuclear fraction directly isolated from hLECs of different ages containing equal amounts of protein were incubated with 32 p-labeled wild-type ARE probe from Prdx6 promoter and processed for gel-shift assay. An apparent age-related reduction in Nrf2/ARE binding was observed (A, Nrf2/ DNA). (B) Nrf2-specific antibody depletion assay revealing depletion of Nrf2/ARE complex, demonstrating specificity of Nrf2 binding to ARE probe. Equal amounts of nuclear protein were incubated with antibody specific to Nrf2 to deplete Nrf2 SFN showed Nrf2 accumulation in total cell extracts within 6 h at each concentration (Fig. 5C, panel; a), which is consistent with SFN-mediated inactivation of Keap1 as noted in Introduction section. Because both concentrations of SFN were effective, we chose only the higher concentration, 4.8 μM, to treat cells for shorter time (2 h) to examine how quickly Nrf2 translocated/accumulated into nucleus. Nuclear fraction isolated from SFN-treated and -untreated rLECs were immunoblotted as shown in Fig. 5C, panel; b. A significant accumulation of Nrf2 was detected in nuclear fraction of SFN-treated rLECs when observed at 2 h (Fig. 5C, panel; b), suggesting this initial lag period may be necessary for translational synthesis of new Nrf2 protein 47 , and also that the time period of within 2 h may represent the time critical for nuclear translocation and ARE-mediated gene transcription. As a whole, the data indicated increased cellular abundance of Nrf2 but nuclear accumulation, a basic phenomenon occurring in SRA-hLECs and rLECs during SFN induction of Nrf2 as described previously for other cells 26,48 . Upregulation of antioxidant genes in SRA-hLECs/rLECs was largely derived from SFN-induced augmented Nrf2 binding to ARE. To determine whether SFN activation of Prdx6 transcription in SRA-hLECs resulted from a gain in DNA binding activity of Nrf2 to ARE, nuclear fraction from SFN treated SRA-hLECs (0, 3 μM, 6 μM, 8 μM) for 24 h were tested on gel-shift assay. We synthesized the oligonucleotides derived from Prdx6 promoter containing ARE ( −357 nTGACCGAGCn −349 ) and its mutant containing GT binding sites (Fig. 6). Nuclear fraction from SFN-treated cells showed enhanced binding to ARE and formed a shifted  (Fig. 6A, lane1). The increase of binding was related to increased concentrations of SFN (Fig. 6). The shifted Nrf2/DNA complex that appeared in lanes was diminished when the Nrf2-depleted nuclear extract was used for binding assay (Fig. 6A, lanes, 5 to 8). Nonetheless, there was mild interaction between Nrf2-depleted nuclear extract to probe, which may have occurred because antibody concentration was not optimal for absolute depletion of Nrf2 (Lane 5 vs 6, 7 and 8). In addition, nuclear extract did not interact with the mutant probe (Fig. 6A, Mut probe; lanes, 9 to 12), verifying the specificity of ARE/Nrf2 binding.

SFN enhanced interaction of Nrf2/ARE in SRA-hLECs in a time-and concentration-dependent manner.
Given the apparent inductive response of antioxidant genes (Figs 3 and 4) to SFN, we sought to determine how effectively SFN activated Nrf2/ARE interaction in SRA-hLECs. Using gel-shift assay, we determined the SFN-induced kinetics of Nrf2/ARE interaction. We treated cells with two concentrations, 3 μM and 8 μM of SFN, based upon our previous finding. Gel-shift assay with the same ARE probe and its mutant as shown in Fig. 6A demonstrated that the higher concentration of SFN enhanced Nrf2 binding and formed Nrf2/DNA complex (Fig. 6B, lane 3) compared to the lower concentration (Fig. 6B, lanes, 1 and 2; respectively). In contrast, with the mutant probe, nuclear fraction of SRA-hLECs did not show the Nrf2/ARE complex (lanes 4, 5 and 6), indicating specificity.
Because a 3 μM concentration of SFN for 1 h did not affect Nrf2/ARE interaction significantly, we examined the effect of duration of SFN treatment on the interaction. SRA-hLECs treated with 3 μM and 6 μM of SFN for 4 h and 8 h were processed for gel-shift assay. Figure 6C shows a time-dependent increase of Nrf2/ARE binding (Fig. 6C). A closer observation of Nrf2/ARE complex revealed that maximum binding occurred in nuclear fraction of SRA-hLECs treated with either concentration for 8 h (Fig. 6C, lanes 1 and 2 vs 3 and lanes 4 and 5 vs 6). Primary culture of rLECs treated with different concentrations of SFN were processed for extraction of total cell extract as well as cytosolic and nuclear fractions at predefined time intervals as indicated. Cellular extract (C, a) or nuclear fraction (C, b) containing equal amounts of protein were immunoblotted with anti-Nrf2 antibody. β-actin was used as loading control. A significant accumulation of Nrf2 in nucleus was observed when examined at 2 h and onwards compared to basal levels (untreated control).
In vivo DNA protein binding assay revealed that SFN enhanced Nrf2 enrichment at ARE sequences present in the Prdx6 promoter. Careful analysis of in vitro data on SFN-induced Nrf2/DNA interaction showed that Nrf2 exclusively bound to ARE (Fig. 7A). Next, to determine if increased activation of Nuclear fraction extracted from SRA-hLECs was incubated with 32 p radiolabeled wild-type or mutant probes containing ARE sites. SFN concentration-dependent binding activity of Nrf2 to ARE (Nrf2/DNA; lanes 1 vs 2, 3 and 4) compared to mutant probe (lanes 9, 10, 11 and 12). Antibody depletion assay showed disruption of Nrf2/DNA complex (lanes 5, 6, 7 and 8), suggesting that Nrf2 in nuclear extract selectively bound to ARE. (However, antibody did not entirely deplete Nrf2 in nuclear faction of SRA-hLECs, so some residual interaction can be seen in all lanes.) (B) SFN rapidly stimulated Nrf2 binding activity to ARE present in Prdx6 human promoter. SRA-hLECs were cultured in the presence of DMSO (control vehicle) or with different concentrations of SFN for 1 h. Nuclear fractions were isolated and processed for gel-shift assay. A strong Nrf2/DNA complex was formed with SRA-hLECs treated with 8 μM of SFN for 1 h (B, lane 1 vs 2 vs 3). In contrast, mutant probe did not act similarly, validating that the Nrf2/DNA complex on gel-shift was specific. (C) Gel-shift and antibody depletion assay showed SFN amelioration of Nrf2 binding activity to ARE in the Prdx6 promoter in concentration-and time-dependent fashion. SRA-hLECs were treated with different concentrations of SFN for different time periods. Nuclear extracts containing equal amounts of proteins were incubated with radiolabeled ARE probe. A relative modulation in Nrf2/DNA complex intensity was observed, and was related to concentration and time of exposure as shown in a representative figure (C, lane 1 vs 2 and 3; Lane 2 vs 3 and lane 4 vs 5 and 6; lane 5 vs 6). In contrast, antibody depletion assay showed reduced band intensity or ablation of DNA/Nrf2 complex (C, lanes 7, 8, 9, 10, 11 and 12). Bold bases represent mutation sites; mutated base(s) as shown and underlined denote core ARE sequences in Prdx6 promoter. NS denotes nonspecific band.
SCiENtifiC RepoRtS | 7: 14130 | DOI:10.1038/s41598-017-14520-8 Nrf2 occurred via a direct mechanism in vivo, we employed chromatin immunoprecipitation (ChIP) assay to measure the occupancy of Nrf2 on ARE of hPrdx6 gene promoter. SRA-hLECs treated with SFN (0 μM, 3 μM and 6 μM) for 24 h were processed for ChIP assay with anti Nrf2 antibody (Fig. 7A) as described in the Methods section 49 . Figure 7B shows that the Prdx6 promoter containing ARE sequences was occupied by Nrf2, and increased enrichment of Nrf2 to the sequences was SFN concentration-dependent. No amplicon was observed with control IgG, pointing to specificity of Nrf2 antibody. These data demonstrate that SFN enhanced Nrf2 enrichment at ARE sequences, and explain the mechanism of SFN-dependent increased Prdx6 transcription.
To test the efficacy of SFN in activating Nrf2 in aging/aged hLECs, we performed ChIP assay. Aging hLECs treated with SFN (0 μM, 3 μM and 6 μM) for 24 h were processed for ChIP assay with Prdx6 promoter as mentioned above. As shown in Fig. 7C, the enrichment of Nrf2 at ARE sequences in Prdx6 promoter was significantly increased in SFN-treated older/aged LECs in a concentration-dependent fashion (Fig. 7C, open vs gray and black bars; gray vs black bar). However, younger LECs were relatively more responsive to SFN treatment. Thus it appears that the aging hLECs retained Nrf2 activity when exposed to SFN. However, we did not perform Western analysis of SFN-treated cells to examine the nuclear or cytosolic levels of Nrf2; nonetheless ChIP experiments directly provided evidence of a concentration-dependent enrichment of Nrf2 at ARE site.
SFN's failure to activate mutant Prdx6 promoter demonstrated that transactivation was largely derived from direct binding of Nrf2 to ARE in Prdx6 promoter in vivo. To examine the consequences of the SFN-induced changes in Nrf2 binding to ARE on Prdx6 transcription, we transfected SRA-hLECs with WT-Prdx6 promoter-CAT construct containing ARE or its mutant (Fig. 7D, Top drawing) along with GFP plasmid. These transfectants were treated with SFN (DMSO or 6 μM) for 24 h. Transactivation assay with mutant construct showed significant inhibition in CAT activity, and SFN failed to activate it (Fig. 7D). Conversely, wild-type promoter displayed robust promotion of CAT activity in response to SFN (Fig. 7D, WT; gray vs black bar), suggesting that SFN upregulated Prdx6 transcription through ARE. However, data revealed that mutation at ARE site did not completely abolish Prdx6 promoter activity, indicating the possible involvement of other transcriptional proteins or pathways.
To examine whether SFN restored Nrf2 dysregulation of Prdx6 transcription in aging hLECs, we transfected hLECs with WT-Prdx6-CAT (Fig. 7E). SFN significantly enhanced transcriptional activity of Prdx6 from basal activity levels in all aging cells (gray vs black bar). Younger cells were more responsive than aged cells, and the response was directly related to Nrf2/ARE interaction shown in Figs 6 and 7.

Prdx6-knockdown disclosed that SFN-treated LECs gained resistance against UVB-induced cellular insults though Prdx6.
With the goal of developing transcription-based "inductive therapy" to reinforce the endogenous Prdx6, we chose SFN because of its effectiveness in cytoprotection and in treating/postponing oxidative/ aging disorders 25,48,50,51 . Eyes are maximally exposed to UVB radiation. Therefore, we examined whether treatment with SFN would abate the cellular injuries evoked by UVB stress. We used antisense of Prdx6 (As-Prdx6) to knock down Prdx6 in SRA-hLECs as reported previously 7 . These transfectants were treated with SFN and then exposed to UVB and measured for viability and ROS production (Fig. 8A,B; lined bars). Cell viability assay revealed that SFN was significantly less effective in protecting SRA-hLECs having As-Prdx6. Also quantification of ROS levels in these SRA-hLECs showed that SFN did not lower ROS expression significantly (Fig. 8B, lined bars), suggesting that SFN acted mainly through Prdx6. In experiments to examine the cytoprotective ability of SFN against UVB-induced LECs injuries, we used SRA-hLECs, hLECs of different age groups and rLECs and pretreated them with SFN as indicated. Figure 8A,C and E show enhanced viability of SRA-hLECs, hLECs and rLECs (open vs gray and black bars; gray vs black bar) and reduced expression of ROS (Fig. 8B,D,F, open vs gray and black bars; gray vs black bar) with variable levels of UVB exposure (400 J/m 2 or 800 J/m 2 ) after SFN treatment. Data were normalized with absorbance of untreated controls. At concentration of 6 μM for hLECs and 2.4 μM and 4.8 μM for rLECs, SFN was effective in protecting LECs (Fig. 8). Moreover, none of the concentrations provided absolute protection, suggesting the involvement of other antioxidants augmented by SFN. Because other antioxidants did not protect hLECs against UVB stress significantly, we think that their protective role in lens/LECs may be minor compared to that of Prdx6.

Discussion
In this study, we showed for the first time that, in aging, increased oxidative stress in lenses and lens cells is associated with failure of protective response due to dysregulation of Nrf2 and its target antioxidant gene Prdx6, and that this process was attenuated by application of SFN. Our work also revealed that ROS increase progressively during aging, and in aged cells become even more substantially increased (Fig. 1A), in a process directly related to progressive reduction in Nrf2,Prdx6 and Cat expression (Fig. 1B-D). Our data are consistent with results reported in other model systems 25,52,53 , showing the negative effects of aging on DNA binding activity of Nrf2 54 . This is thought to be caused by impaired cytoplasmic-nuclear shuttling of Nrf2 and age-related reduction in cellular abundance and activity of Nrf2 13,55-58 . Moreover, recently it was reported that Nrf2 levels are regulated by glycogen synthase kinase 3 (GSK-3) in Keap-independent pathways. Under strong oxidative stress, GSK-3 targets Nrf2 for β-TrCP-mediated proteasomal degradation, and thus the stability of Nrf2 is controlled via GSK-3/β-TrCPd 46,59 . We believe that strong activation of GSK-3 during normal aging or increased oxidative stress can lead to β-TrCP-mediated degradation of Nrf2, followed by repression of its target cytoprotective genes. Our DNA binding experiment with ARE probe derived from hPrdx6 promoter showed a significant reduction in Nrf2/ARE interaction in aging, and we noticed a dramatic reduction in Nrf2/ARE binding with nuclear fraction of elderly hLECs ( Fig. 2A,B). Promoter assay also demonstrated that reduced binding of Nrf2 negatively affected Prdx6 transcription (Fig. 2C). Unfortunately, due to a scarcity of human samples (lenses) and limited proliferation of LECs in culture, we were not able to use the same hLECs for all the experiments, nor could we obtain samples with matched ages. Nonetheless, our study demonstrated an underlying molecular mechanism of Prdx6 repression that can be associated with a decline in Nrf2 expression/activity in aging lenses. Chromatin samples prepared from SRA-hLECs treated with varying concentrations (0, 3 µM and 6 µM) of SFN for 24 h were subjected to ChIP assay with a ChIP grade antibody, anti-Nrf2 (black bars) and control IgG (gray bars). The DNA fragments were used as templates for qPCR by using primers designed to amplify −400 to −305 region of the human Prdx6 promoter bearing Nrf2/ARE sites as shown. Histogram shows the amplified DNA band visualized with real-time PCR analysis. DMSO (0) vs 3 µM and 6 µM SFN and 3 µM vs 6 µM SFN treatment; *p < 0.001. (C) SFN reinforced the enrichment of Nrf2 to its responsive element, ARE, present in Prdx6 gene promoter in aging/aged primary hLECs. ChIP assay was conducted using anti-Nrf2 antibody. Immunoprecipitated DNA fragments were purified and processed for qPCR analysis using primers indicated above and in the Methods section, but in primary hLECs of variable ages treated with different concentrations In previous studies, we found that Prdx6 depletion causes increased susceptibility to UVB-or H 2 O 2 -induced cell death. We also observed that other antioxidants were not effective at protecting LECs/lenses [7][8][9] , suggesting that Prdx6 expression is critical for protection of eye lenses. Prdx6 provides cytoprotection by removing ROS through its GSH peroxidase activity. ROS are produced in cells continuously through nonenzymatic and enzymatic reactions such as superoxide-dismutase (SOD)-catalyzed disproportionation of the superoxide radicals (O 2 •− ) to H 2 O 2 as well as by redox cycling. Continuous exposure to oxidants can also contribute substantially to the cellular steady state levels of H 2 O 2 and O 2 •− . However, constitutive generation of H 2 O 2 is derived mostly from mitochondria dependent upon NADH 60,61 and activity of nicotinamide adenine dinucleotide phosphate (NADPH) oxidase (Nox) enzymes dependent on NADPH 62,63 . The primary function of NADPH oxidases is to produce ROS. These enzymes have been suggested to contribute to initiation of many diseases linked to oxidative stress 64 . Levels of ROS can also be raised during increased activity of cells and oxidative stress. Nonetheless, H 2 O 2 is produced not only by mitochondria, but also by endoplasmic reticulum, peroxisomes and plasma membrane. Prdx6 is localized in these organelles, indicating the importance of this molecule in controlling the redox active state of cells.
Previously we found that Prdx6 expression declines with aging 1 . We surmised that the best strategy in this situation might be to use a natural activator such as SFN to restore the activity of Nrf2 and its target gene Prdx6. SFN preparations have been approved as clinically safe for use in healthy volunteers 65 . To ascertain whether SFN would induce the Nrf2/Prdx6 pathway in LECs, we examined levels of three antioxidants, Prdx6, Cat and GSTπ, in SRA-hLECs and rLECs. We specifically selected GSTπ as it is an electron donor to Prdx6, which controls the redox active state of Prdx6 66 . Figures 3 and 4 show that SFN induced expression of all three genes, Prdx6, Cat and GSTπ in dose-dependent fashion in both SRA-hLECs and rLECs. Closer in silico analysis of rat Prdx6 promoter (−10K from ATG sequences) revealed that it has several ARE-like sequences (data not shown), demonstrating that upregulation in antioxidant genes in rLECs occurred through the Nrf2 pathway. Surprisingly, our study revealed that SFN-treated SRA-hLECs displayed increased expression of GSTπ. This result, however, was not in agreement with a previously published report 67,68 . Our knockdown experiment with Sh-Nrf2 showed that SFN does activate GSTπ expression as shown, but not through Nrf2 (Supplementary Fig. 1). The different outcomes may be related to different cell types, or possibly SFN regulated GSTπ through pathways other than Nrf2/Keap1. The latter situation might be anticipated, since SFN has been shown to affect a number of pathways aside from Nrf2/Keap1 signaling 69 . There have been reports that GSTπ promoter bears ARE, Sp1 and ARE/TRE sites, which are required for its transcription 70,71 . Possibly, failure of Nrf2 to activate GSTπ in hLECs is due to the existence of Bach1 repressive signaling; Bach1 inactivation is required for GSTπ expression 72 . Furthermore, other transcription factors may be involved, like Sp1 72,73 , which may be activated through SFN 74 and modulate GSTπ in hLECs. Nevertheless, further study is warranted to examine these possibilities.
Intriguingly, SFN-treated LECs displayed increased expression of Nrf2 mRNA and increased abundance of nuclear Nrf2 (Fig. 5), and this cellular response was time-and concentration-dependent. This observation is in agreement with a previous study showing that Nrf2 gene expression itself is regulated via ARE/Nrf2 mechanism 45 . Furthermore, our experiment demonstrated that SFN reinforced the nuclear accumulation of Nrf2 in LECs (Fig. 5B,C), and thereby enhanced Nrf2/ARE binding (Figs 6 and 7). We found that increased Nrf2/ARE activity was associated with increased promoter activity of Prdx6. However, mutant Prdx6 promoter retained some activity. The modulation in ARE-dependent gene transcription may be affected by Nrf2 interacting with proteins as it interacts with Jun/Fos family, Fra, small Maf, and ATF4 13,47 . These factors may modify the transcription potential of Nrf2 in activating ARE-mediated transcription.
UVB is a major culprit for inducing oxidative damage of eye lens/LECs. We found that rLECs and SRA-hLECs pretreated with SFN showed resistance against UVB injuries (Fig. 8A-D). In addition, SFN protected aging hLECs against UVB stress. We think that SFN does so by activation of the Nrf2/ARE pathway as evidenced by Figs 6 and 7. Examining the contribution of Prdx6 in rescuing SFN-treated LECs, our Prdx6-knockdown experiment revealed that SFN without Prdx6 became significantly less effective in protecting LECs facing UVB (Fig. 8A,B). Other antioxidants that were reinforced by SFN failed to protect LECs, indicating that Prdx6 is essential to protect LECs against UVB. Moreover, we also recognize that oxygen levels in the eye are generally low, as maintaining lens clarity over a prolonged time is essential 75,76 . Nevertheless, hLECs are metabolically highly active, with high concentrations of mitochondria. ROS generation due to oxygen or reductive stress-induced ROS (in the hypoxic range) is slowed through an enriched antioxidant defense system of LECs that can normalize functioning and thereby maintains lens homeostasis (adaptive response). Thus we think that LECs cultured in vitro behave similarly to other cells due to adaptive responses. Several published studies as well as our own study have tested antioxidant activity of biochemical reagents with in vitro model systems (20% O 2 ) by applying exogenous stresses, and these activities have been reproduced in vivo 8,[77][78][79][80][81] . However, the in vitro study conducted in the current research should clarify Prdx6′s ability to protect LECs and its regulation by SFN.
In summary, we have shown that Nrf2 and its mediated genes are dysregulated in aging LECs and lenses. Importantly, activation of Nrf2 can be reinforced by treating aged lens cells with SFN. The study also detailed the molecular mechanism that occurs during aging, at least in lens/LECs, i.e., the increased accumulation of oxidative load due to failure of antioxidant response, and found that Prdx6 expression is required to reverse the pathogenic process in LECs. Based upon this work, we propose a chemopreventive strategy of using small molecules like SFN to block/delay cataractogenesis or etiopathogenesis in eye lens.
with the same wild-type Prdx6 promoter ARE site (upper panel). Lower panel, relative CAT activity of the wildtype promoter in SFN-treated aging/aged hLECs. All data are presented as mean ± S.D. values derived from three independent experiments. DMSO vs SFN treated samples; younger (21 y old) vs aging samples; *p < 0.001.

Methods
Cell culture and treatments. Primary rat LECs (rLECs) were isolated from 6-week-old Sprague-Dawley albino rats (n = 8) as described previously 8 . The rLECs were maintained in Dulbecco's Modified Eagle's Media (DMEM; Life Technologies, Carlsbad, CA, USA) with 10% fetal bovine serum (FBS; Sigma, St. Louis, MO, USA). rLECs reaching 80 to 90 percent confluence were harvested and used for assays. All the experiments on rLECs were conducted at passages (P) 3 to 5. Isolation of rLECs from animals was approved by the Kanazawa Medical University, and procedures were conducted in accordance with the National Institutes of Health Guidelines for Laboratory Animals at the Kanazawa Medical University, Japan.
Human LECs used were of two types: (1) a cell line (SRA01/04) immortalized with SV40, and (2) primary human LECs isolated from deceased persons of different ages. To avoid confusion, the remaining text will designate the immortalized LECs as SRA-hLECs, and the primary human (h) LECs as primary hLECs or hLECs The SRA-hLECs were derived from 12 infants who underwent surgery for retinopathy of prematurity 82 (a kind gift of Dr. Venkat N. Reddy, Eye Research Institute, Oakland University, Rochester, MI, USA). These cells were maintained in DMEM with 15% FBS, 100 µg/ml streptomycin, and 100 µg/ml penicillin in 5% CO 2 environment at 37 °C as described previously 83,84 . Isolation and generation of hLECs. Primary hLECs were isolated from normal eye lenses of deceased persons or healthy donors of different ages (16,18,21,24,26,34,36,52,56,58 8,9,85,86 . Capsules were spread by forceps with cell layers upward on the surface of plastic petri dishes. Culture explants were trypsinized and re-cultured. Cell cultures attaining 90 to 100 percent confluence were trypsinized and used for experiments 49,84,87 . Western analysis was used to validate the presence of αA-crystallin, a specific marker for LEC identity (data not shown). Cell survival assay (MTS assay). A colorimetric MTS assay (Promega, Madison, WI, USA) was performed as described earlier 8,9,88 . This assay of cellular viability uses 3-(4,5-dimethylthiazol-2-yl)-5-(3-carb oxymethoxyphenyl)-2 to 4-(sulphophenyl) 2H-tetrazolium salt. The A 490 nm (O.D.) value was measured after 2 h with a plate reader, Spectra Max Gemini EM (Mol. Devices, Sunnyvale, CA). Results were normalized with absorbance of the untreated control(s). Quantitation of ROS levels by H2-DCF-DA assay. SRA-hLECs or primary hLECs or rLECs were cultured in 96 well plates (5 × 10 3 /well) in the presence or absence of SFN. At predefined times these cells were subjected to UVB stress. After eight hours levels of ROS were measured by using fluorescent dye dichlorofluorescin diacetate (H2-DCF-DA), a nonpolar compound that is converted into a polar derivative (dichlorofluorescein) by cellular esterase after incorporation into cells 7 . Levels of ROS (intracellular fluorescence) were detected at excitation (Ex) 485 nm/emission (Em) 530 nm by Spectra Max Gemini EM (Mol. Devices, Sunnyvale, CA).

Real-Time Reverse Transcriptase-Polymerase Chain Reaction (RT-PCR).
The total RNA from the SRA-hLECs, rLECs and primary hLECs directly detached from lenses (to avoid cell culture effect) was extracted using RNeasy Mini Kit (Qiagen, Valencia, CA) following the manufacturer's protocol. Total RNA from lenses of variable ages was extracted to examine the levels of Nrf2 and Prdx6. From 0.5 to 2 micrograms of total RNA was reverse-transcribed with High Capacity cDNA Reverse Transcription Kit following the manufacturer's instructions. For rLECs, a Gene Amp PCR System 9700 (Applied Bio Systems, Foster City, CA, USA) was used; for hLECs and SRA-hLECs the SYBR Green Master Mix (Roche Diagnostic Corporation, Indianapolis, IN) in a Roche ® LC480 Sequence detector system (Roche Diagnostic Corporation) was employed. GSTπ, Catalase, and Prdx6 as well as Nrf2 gene expressions were analyzed with RT-PCR on 7300 Real Time PCR System (Applied Biosystems) using the primers designed for each molecule of rat genes (TaqMan; ratPrdx6 probe ID: Rn01759191_g1; rat catalase probe ID: Rn00560930_m1; rat GSTπ probe ID: Rn00561378_gH) or human genes (Universal probe library for human; Prdx6 probe ID: NM_004905.2; catalase probe ID:NM_001752.3; GSTπ probe ID:NM_000852.3; Nrf2 probe ID: NM_001145413 and β-actin probe ID: NM_001101.3). 18 S ribosomal RNA (Applied Biosystems), β-actin as an endogenous control, and/or both were used to normalize the expression of GSTπ, Catalase and Prdx6 in each group. The relative quantity of mRNA was obtained using the comparative CT method.
Protein expression analysis. Cell extract of LECs were prepared in ice-cold radioimmune precipitation buffer and protein blot analysis was performed as described previously 1, 89 . The membranes were probed with Anti-Prdx6 antibody (Ab) (Abcam ® , Cambridge, MA, USA and Lab Frontier, Seoul, Korea), anti-catalase monitor the levels of ROS and cell viability, and the percentage of ROS and cell survival levels was then calculated for each group, respectively.

Construction of Prdx6 antisense.
A human LEC cDNA library was used to isolate Prdx6 cDNA having a full-length open reading frame. A full-length Prdx6 antisense (Prdx6-As) construct was made by sub-cloning Prdx6 cDNA into a pcDNA3.1/NT-GFP-TOPO vector in reverse orientation. Plasmid was amplified following TOP 10 bacterial cells transformation as described earlier 9 .
Statistical methods. 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 P value of < 0.05 and 0.001 for two or more independent experiments.