Long Isoforms of NRF1 Contribute to Arsenic-Induced Antioxidant Response in Human Keratinocytes

Background Human exposure to inorganic arsenic (iAs), a potent oxidative stressor, causes various dermal disorders, including hyperkeratosis and skin cancer. Nuclear factor–erythroid 2–related factor 1 (NRF1, also called NFE2L1) plays a critical role in regulating the expression of many antioxidant response element (ARE)-dependent genes. Objectives We investigated the role of NRF1 in arsenic-induced antioxidant response and cytotoxicity in human keratinocytes. Results In cultured human keratinocyte HaCaT cells, inorganic arsenite (iAs3+) enhanced the protein accumulation of long isoforms (120–140 kDa) of NRF1 in a dose- and time-dependent fashion. These isoforms accumulated mainly in the nuclei of HaCaT cells. Selective deficiency of NRF1 by lentiviral short-hairpin RNAs in HaCaT cells [NRF1-knockdown (KD)] led to decreased expression of γ-glutamate cysteine ligase catalytic subunit (GCLC) and regulatory subunit (GCLM) and a reduced level of intracellular glutathione. In response to acute iAs3+ exposure, induction of some ARE-dependent genes, including NAD(P)H:quinone oxidoreductase 1 (NQO1), GCLC, and GCLM, was significantly attenuated in NRF1-KD cells. However, the iAs3-induced expression of heme oxygenase 1 (HMOX-1) was unaltered by silencing NRF1, suggesting that HMOX-1 is not regulated by NRF1. In addition, the lack of NRF1 in HaCaT cells did not disturb iAs3+-induced NRF2 accumulation but noticeably decreased Kelch-like ECH-associated protein 1 (KEAP1) levels under basal and iAs3+-exposed conditions, suggesting a potential interaction between NRF1 and KEAP1. Consistent with the critical role of NRF1 in the transcriptional regulation of some ARE-bearing genes, knockdown of NRF1 significantly increased iAs3+-induced cytotoxicity and apoptosis. Conclusions Here, we demonstrate for the first time that long isoforms of NRF1 contribute to arsenic-induced antioxidant response in human keratinocytes and protect the cells from acute arsenic cytotoxicity.

volume 119 | number 1 | January 2011 • Environmental Health Perspectives Research Chronic exposure to high levels of inorganic arsenic (iAs) is associated with a wide range of human ailments, including cancer, arterio sclerosis, hypertension, type 2 diabetes, and a variety of skin disorders (Pi et al. 2000;Yoshida et al. 2004). The skin is one of the organs most sensitive to iAs toxicity. This is potentially due to the high affinity of arsenic for sulf hydryl groups, which leads to arsenic accumulation and retention in keratinrich skin tissue. Arsenicinduced non malignant skin lesions, including hyper keratosis and pig mentation disorders, are some of the most common and earliest signs of chronic iAs exposure (Pi et al. 2000;Yoshida et al. 2004). The proliferative skin lesions associated with human iAs exposure include Bowen's disease and squamous cell or basal cell carcinoma [International Agency for Research on Cancer (IARC) 1987; Wong et al. 1998]. Although iAs is a confirmed human skin toxi cant, the under lying molecular mechanism(s) is still unclear. Accumulating evidence suggests that oxidative stress occurs in response to iAs expo sure (Pi et al. 2002(Pi et al. , 2003 and may be one important factor in dermal arsenic toxicity, including carcino genesis. Indeed, evidence of arsenicinduced oxidative DNA damage has been observed in cellbased systems (Kojima et al. 2009;Pi et al. 2005) and in the biologi cal samples of rodents and humans (Piao et al. 2005;Yamauchi et al. 2004).
The nuclear factor-erythroid2-related factors (NRFs) belong to the cap'n'collar (CNC) subfamily of basicregion leucine zip per (bZIP) transcription factors, which include NRF1 (NFE2L1/LCRF1/TCF11), NRF2 (NFE2L2), NRF3 (NFE2L3), and the nuclear factor-erythroid 2 p45 subunit, as well as more distantly related factors such as BTB and CNC homology 1 (BACH1) and BACH2 proteins (Motohashi et al. 2002). Both NRF1 and NRF2 form hetero dimers with small Maf or other bZIP proteins and bind to cis acting element(s) termed anti oxidant or elec trophile response elements (AREs; also known as EpREs) in the proxi mal promoters of target genes (Motohashi et al. 2002), leading to acti vation of transcription (Biswas and Chan 2010;Venugopal and Jaiswal 1998). Although NRF3 can hetero dimerize with MafK or MafG and bind AREs (Chenais et al. 2005;Kobayashi et al. 1999), the role of NRF3 in the regula tion of AREresponsive genes remains elusive. NRF1 (Wang and Chan 2006) and NRF3 (Nouhi et al. 2007;Zhang et al. 2009a) are targeted to the endo plasmic reticulum (ER), whereas NRF2 is localized primarily to the nucleo plasm and cyto plasm. Supporting the importance of NRF1 in the develop mental process is the finding that loss of NRF1 func tion in mice results in lategestational embry onic lethality (Chan et al. 1998). Liverspecific disruption of Nrf1 results in the development of steato hepatitis and hepatic neoplasms (Xu et al. 2005). In contrast, Nrf2deficient mice are viable but show a higher susceptibility to both oxidative damage and chemical carcino genesis Chan et al. 2001;RamosGomez et al. 2001), whereas Nrf3null mice develop normally and reveal no obvious pheno type (Derjuga et al. 2004). Fibroblasts derived from Nrf1mutant embryos showed decreased gluta thione (GSH) levels and enhanced sensitivity to the toxic effects of oxidants (Chan and Kwong 2000;Kwong et al. 1999), suggesting critical roles for NRF1 in cellu lar oxidative defense.
Previous studies (Aono et al. 2003;Du et al. 2008), including our own (Pi et al. 2003;Fu et al. 2010), have demonstrated that NRF2 is a key player in the cellular adaptive response to inorganic arsenite (iAs 3+ )induced oxidative stress. In contrast, the role of NRF1 in arsenic induced anti oxidant response and cyto toxicity has not been established. In the present study, we examined the distinctive roles of NRF1 in iAs 3+ induced anti oxidant response, cytotox icity, and apoptosis, as well as the inter play between NRF1 and NRF2, in response to iAs 3+ exposure, using HaCaT cells, a human keratino cyte cell line that models the skin as a target of iAs. In this study, we found direct evidence that iAs 3+ activates both the NRF1 and NRF2mediated anti oxidant responses, which protects the cells from acute arsenic cyto toxicity. This indicates for the first time that NRF1 is a novel target of iAs 3+ exposure. The results of this study provide important insights into the initial molecular response to iAs 3+ in the target cells of arsenic toxicity and carcino genicity.

Materials and Methods
Reagents and cell culture. We purchased sodium arsenite, sulforaphane (SFN), and tert-butyl hydro quinone (tBHQ) from Sigma Chemical Co. (St. Louis, MO, USA) and tunica mycin (TU), thapsigargin (TG), and brefeldin A (BFA) from Calbiochem (San Diego, CA, USA). HaCaT cells, a sponta neously immortalized human epithelial cell line developed by Boukamp et al. (1988) were obtained from N.E. Fusening, German Cancer Research Center, Heidelberg, Germany. The cells were cultured in Dulbecco's modified Eagle's medium supplemented with 10% fetal bovine serum, 100 U penicillin/mL, and 100 µg streptomycin/mL, as previously described (Pi et al. 2003). Cultures were main tained at 37°C in a humidified 5% CO 2 atmo sphere. Culture media, fetal bovine serum, and supplements were purchased from Invitrogen (Carlsbad, CA, USA). The stock solutions of chemicals used in the current study were pre pared in culture medium or 0.5% dimethyl sulfoxide (DMSO) in medium (vehicle).
Lentiviral-based short-hairpin RNA (shRNA) transduction. We obtained MISSION shRNA lentiviral particles from Sigma. Lentiviral transduction of HaCaT cells with particles for shRNAs targeting NRF1 (SHVRSNM_003204) or scrambled non target negative control (shScr; SHC002V) was performed as described previously (Woods et al. 2009). The cells were maintained in medium containing 1.0 µg/mL puromycin.
Chromatin immunoprecipitation assay. We performed ChIP analyses using the EZ ChIP kit (Upstate Biotechnology, Temecula, CA, USA) according to the manufacture's protocol. PCR amplification was carried out for 40 cycles with 5 µL of sample DNA solution, and PCR products were separated on 3% agarose gels in Tris-acetate-EDTA buffer. Two primers were used to amplify the segment flanking an active ARE site on NAD(P)H:quinone oxido reductase 1 (NQO1) promoter with forward primer 5´attacctgccttgaggagca3´ and reverse primer 5´cggattactgtggtgcccta3´, which generate a 206bp product.
Acute cytotoxicity assay. A minimum of five replicates of 10,000 cells/well were plated in 96well plates and allowed to adhere to the plate for 24 hr, at which time the medium was removed and replaced with fresh serum free medium containing arsenic compounds. Cells were then incubated for an additional 24 hr, and cell viability was determined using the CellTiter NonRadioactive Cell Proliferation Assay Kit with MTT [3(4,5 dimethyl thiazol2yl)2,5diphenyltetrazolium bromide] (Promega, Madison, WI, USA). Measurements are expressed as a percentage of the untreated control of corresponding cells. The LC 50 (concentration lethal to 50% of cells) values were determined from analysis of the loglinear phase of the curves.
Determination of apoptosis by flow cytometry. Cells were seeded in a sixwell plate and grown to approximately 80% confluence. After 20 hr of iAs exposure, the floating and attached cells were harvested for apoptosis analysis. We detected phosphatidyl serine on the outer leaflet of apoptotic cells using the TACS Annexin VFITC (fluorescein isothio cyanate) Apoptosis Detection Kit (Trevigen, Gaithersburg, MD, USA) as described pre viously (Pi et al. 2005). For each sample, 10,000 cells were examined by flow cytom etry (Becton Dickinson FACSVantage; BD Biosciences, San Jose, CA, USA). We deter mined the percentage of apoptotic cells by sta tistical analysis of the various dot plots using CellQuest software (BD Biosciences).
ARE reporter assay. We obtained Cignal Lenti ARE reporter, which expresses a luciferase gene driven by multiple ARE (TCACAGTGACTCAGCAAAATT) repeats, from SABiosciences (Frederick, MD, USA). Lentiviral transduction of HaCaT cells was performed as described previously (Woods et al. 2009). Cells were grown to approximately 90% confluency and sub cultured in medium containing 1.0 µg/mL puromycin. The luciferase activity was measured by Luciferase Reporter Assay System (Promega) according to the manu facturer's protocol. The luciferase activity was normalized to cell viability that was determined using the NonRadioactive CellProliferation Assay Kit (Promega).
Statistical analyses. We performed all sta tistical analyses using GraphPad Prism, ver sion 5 (GraphPad Software, San Diego, CA, USA), with p < 0.05 taken as significant. More specific indices of statistical significance are indicated in individual figure legends. Data are expressed as mean ± SE. For comparisons among groups, we performed oneway analysis of variance with Bonferroni post hoc testing.

iAs 3+ increases nuclear NRF1 accumula tion.
Based on the Ensembl database (Wellcome Trust Sanger Institute/European Bioinformatics Institute 2010), the human NRF1 gene con tains six exons, transcribes three splice vari ants, and translates into three proteins, NRF11, NRF12, and NRF13, with 742, 772, and 791 amino acids, respectively [see Supplemental Material, Table 2 (doi:10.1289/ ehp.1002304)]. The predicted molecular weights (MWs) of NRF11, NRF12, and NRF13 are 81.5, 84.7, and 86.9 kDa, respec tively. However, our immuno blots (Figure 1), using an antibody developed against an epitope volume 119 | number 1 | January 2011 • Environmental Health Perspectives corresponding to amino acids 191-475 map ping near the Nterminus of human NRF1, showed that multiple bands with apparent MWs approximately 120-140 kDa were dra matically diminished by knockdown (KD) of NRF1 using lentiviral shRNA targeting human NRF1 in HaCaT cells (NRF1KD), suggest ing that these immuno reactive bands repre sent endogenous human NRF1. Interestingly, these NRF1 protein bands significantly increased in response to iAs 3+ but only margin ally responded to tBHQ and SFN exposure. In addition, in response to iAs 3+ treatment, a 78kDa protein exhibited a pattern similar to that of the bands at 120-140 kDa ( Figure 1). However, this protein was not detectable in nuclear fractions (data not shown), suggesting that this protein, if it represents an isoform of NRF1, is not associated with NRF1 tran scriptional activity. Although we also observed multiple bands between 22 and 78 kDa on the blot, these bands lack correspondence to NRF1 silencing and were unaltered by iAs 3+ exposure, suggesting that they may represent non specific binding of the antibody used for Western blot analysis. A 65kDa isoform of mouse NRF1 has previously been identified and shown to potentially function as a domi nant negative inhibitor of AREmediated tran scription (Wang et al. 2007).
To investigate the involvement of NRF1 in iAs 3+ induced antioxidant response in human keratinocytes, we meas ured the dose response and time course of iAs 3+ induced NRF1 accumulation. As shown in Figure 2A, exposure to iAs 3+ resulted in NRF1 protein accumulation in HaCaT cells in a time and dose dependent fashion that reached a peak at 6 hr. Consistent with our previous study (Pi et al. 2003), the same iAs 3+ treatment also concomitantly induced NRF2 protein accu mulation in a pattern similar to that of NRF1 ( Figure 2A). Because nuclear accumulation is essential for a nuclear factor's transcriptional activity, we determined the levels of NRF1 and NRF2 in sub cellular fractions after iAs 3+ expo sure. We detected iAs 3+ induced NRF1 and NRF2 mainly in nuclear fractions ( Figure 2B), suggesting that NRF1 functions as a tran scription factor, as does NRF2, in response to iAs 3+ exposure. To determine the transcrip tional activity of NRF1 and NRF2, we assessed the activity of the Cignal Lenti ARE reporter, which is designed to monitor the activity of the anti oxidant response signal transduction pathway in cultured cells. HaCaT cells sta bly transduced with the ARE reporter showed a dose and timedependent induction of luciferase activity after tBHQ and SFN treat ment, confirming that the cells are respon sive to ARE activation [ Figure 2C; see also Supplemental Material, Figure 1 (doi:10.1289/ ehp.1002304)]. Consistent with the finding that iAs 3+ strongly induced nuclear accumula tion of both NRF1 and NRF2 (Figure 2A,B) Figure 1. Representative image of NRF1 immunoblots with whole-cell lysates derived from NRF1-KD and Scr control HaCaT cells. Cells were treated with vehicle (medium), 20 µM iAs 3+ , 50 µM tBHQ, or 7.5 µM SFN for 6 hr. Whole-cell lysates (50 µg protein) were separated on 4-12% Tris-glycine gels and detected using anti-NRF1. β-Actin was used as a loading control, and SeeBlue Plus2 (Invitrogen) was used as an MW marker.    (Figures 1 and 2B), we found iAs 3+ to be a more potent activator for the ARE reporter than tBHQ and SFN ( Figure 2C; see also Supplemental Material, Figure 1). To further confirm that NRF1 can bind to ARE, we per formed a ChIP assay targeting an active ARE site on NQO1 promoter (Dhakshinamoorthy and Jaiswal 2000). As shown in Figure 2D, acute iAs 3+ exposure increased the binding of NRF1 with the ARE site of NQO1 promoter.
Effect of ER stressors on NRF1 protein modification. Previous studies have reported that NRF1 is a glycosylated protein sequestered in the ER and that ER stressors, including TU, BFA, and TG, have been found to affect the glycosylation status of recombinant human or murine NRF1 (Wang and Chan 2006;Zhang et al. 2009b). To study whether endogenous human NRF1 is regulated by the same mecha nism, we investigated the effect of ER stressors on the migration of iAs 3+ induced NRF1 using SDSPAGE. As shown in Figure 3A, treat ment of HaCaT cells with TU, an inhibitor of Nlinked protein glycosylation (Shang et al. 2002), resulted in a faster migration of NRF1 proteins. In contrast, BFA, which blocks pro tein transport from the ER to Golgi (Li et al. 2006), led to accumulation of slower migrat ing NRF1, whereas TG, which blocks ER uptake of calcium by inhibiting sarcoplasmic/ endoplasmic Ca 2+ ATPase (Thastrup et al. 1990), did not affect NRF1 migration but slightly decreased iAs 3+ induced NRF1 accu mulation. To evaluate the effects of NRF1 modification by ER stressors on its transcrip tional activity, we assessed nuclear NRF1 accu mulation and AREreporter activity in HaCaT cells exposed to iAs 3+ with TU, BFA, or TG. We observed the 120-140 kDa forms of NRF1 mainly in nuclear fractions ( Figure 3B), suggesting that these forms may retain tran scriptional activity. In contrast, we detected the slower migrating NRF1 induced by BFA mostly in cyto solic fractions ( Figure 3C). Although TU + iAs 3+ -induced fastermigrat ing forms of NRF1 were detectable in nuclear fractions, the levels of these forms were much lower than those of the 120-140 kDa forms in nuclear fractions induced by iAs 3+ alone or by TG + iAs 3+ ( Figure 3B). Consistent with the findings in immuno blots, all three ER stressors, which induced ER stress response at the con centrations used [see Supplemental Material, Figure 2 (doi:10.1289/ehp.1002304)], signifi cantly reduced basal and iAs 3+ induced ARE reporter activity ( Figure 3D).
Because NRF2 is another important tran scription factor for ARE activation (Pi et al. 2003), we determined the effect of ER stres sors on NRF2 expression. In contrast to the varied effects on NRF1, the three ER stressors had no obvious effect on NRF2 migration on SDSPAGE ( Figure 3A-C), suggesting that no protein modification occurred in NRF2. However, TU and BFA slightly enhanced basal NRF2 protein level, whereas TG decreased it ( Figure 3A). Under iAs 3+ exposed conditions, TU and TG obviously reduced NRF2 lev els in wholecell lysates and nuclear fractions, whereas BFA had little effect ( Figure 3B).

Involvement of NRF1 in iAs 3+ -induced anti oxidant response.
To study the role of NRF1 in iAs 3+ induced antioxidant response and cyto toxicity, we performed lenti viral shRNAmediated knockdown of NRF1 in HaCaT cells, using five shRNAs against NRF1 for transduction [see Supplemental Material, Table 3 and Figure 3 (doi:10.1289/ ehp.1002304)]. One of the constructs (sh NRF1-5) markedly silenced NRF1 expres sion compared with Scr (shScr), whereas the other four constructs had a moderate silencing effect. Because the level of NRF1 protein is barely detectable in untreated cells and even in tBHQ-or SFN-challenged cells (Figure 1), the efficiency of knockdown by shNRF1-5 (NRF1KD cells) was confirmed by notably diminished induction of NRF1 caused by iAs 3+ exposure (Figure 1; see also Supplemental Material, Figure 3B). Furthermore, the expres sion of NRF1 downstream targets GCLC and GCLM were attenuated (see Supplemental Material, Figure 3C,D), indicating that NRF1 activity is suppressed in NRF1KD cells. Along with the reduction of GCLC and GCLM, the intra cellular GSH level was significantly reduced by silence of NRF1 (see Supplemental Material, Figure 3E), confirming that NRF1 is critical in regulation of GSH synthesis.
To define the molecular basis for how NRF1 is involved in cellular oxidative defense against acute iAs 3+ toxicity, NRF1KD and Scr cells were acutely exposed to iAs 3+ ; we then determined the inducible expression of AREdependent genes, including HMOX1, NQO1, SRX, GCLC, GCLM, and NRF1, at mRNA ( Figure 4) and protein levels [see . In Scr cells, iAs 3+ dose and timedependently increased NRF1 protein levels ( Figure 5) and enhanced the mRNA levels of AREdependent genes (Figure 4). Knockdown of NRF1 substantially decreased NRF1 accumulation ( Figure 5) and the expres sion of NQO1, GCLC, and GCLM under basal and iAs 3+ exposed conditions (Figure 4). Interestingly, induction of HMOX1 caused by iAs 3+ did not depend on NRF1 (Figure 4).
A previous study (Leung et al. 2003) revealed that NRF1 and NRF2 have over lapping roles in regulating basal expression of AREdependent genes. Thus, we studied the cross talk of NRF1 with NRF2, as well as with KEAP1, a wellknown negative regula tor of NRF2 transcriptional activity (Hayes and McMahon 2009). As shown in Figure 5, silencing of NRF1 in HaCaT cells did not disturb iAs 3+ induced NRF2 accumulation. However, lack of NRF1 decreased protein lev els of KEAP1 under basal and iAs 3+ challenged conditions, although KEAP1 was not affected by iAs 3+ treatment.

iAs 3+ -induced cytotoxicity and apoptosis in NRF1-deficient HaCaT cells.
To inves tigate the roles of NRF1 in iAs 3+ induced cyto toxicity, we measured the acute (24 hr) effect of iAs 3+ on cell metabolic integrity in NRF1KD cells. Selective deficiency of NRF1 in HaCaTs significantly enhanced the sensi tivity to iAs 3+ toxicity ( Figure 6A). The LC 50 value (mean ± SE) was 28.62 ± 3.06 µM in NRF1KD cells, whereas it was 35.99 ± 2.11 µM in Scr cells. To further substantiate these findings, we meas ured iAs 3+ induced apop tosis and necrosis using flow cytometry with Annexin VFITC and propidium iodide double staining. Consistent with the results of cytotoxicity, the knockdown of NRF1 in HaCaT cells significantly enhanced the sensi tivity to iAs 3+ induced apoptosis [ Figure 6B; see also Supplemental Material, Figure 4 (doi:10.1289/ehp.1002304)].

Discussion
NRF1 is a ubiquitously expressed transcrip tion factor that occurs in a wide range of tissues (Biswas and Chan 2010;Luna et al. 1994). Skin is a major target organ for the chronic toxic and carcinogenic effects of iAs ). Our previous stud ies revealed that chronic induction of ARE dependent genes may be linked to acquired apoptotic resistance and malignant transfor mation of keratino cytes following iAs 3+ expo sure, whereas NRF2 has been recognized as a key transcription factor in iAs 3+ induced anti oxidant response (Pi et al. 2003(Pi et al. , 2007(Pi et al. , 2008. The present study provides the first demon stration that long isoforms (120-140 kDa) of NRF1 also contribute to iAs 3+ induced anti oxidant response in human keratino cytes and suggests that activation of NRF1 is potentially involved in chronic dermal arsenic toxicity.
Hyperkeratosis and cancer are the most common human skin disorders caused by chronic iAs exposure (IARC 1987;Pi et al. 2000;Wong et al. 1998;Yoshida et al. 2004). However, the under lying mechanism is unclear. It has been reported that disrup tion of Keap1 in mice leads to skin hyper keratosis, most likely because of constitutive activation of NRF2 and aberrant expression of some AREdependent cytokeratins . In humans, increased expres sion of AREdependent genes, resulting from mutations in KEAP1 and/or NRF2, has been linked to a malignant pheno type in the lung and other organs Kensler 2009. Padmanabhan et al. 2006;Shibata et al. 2008aShibata et al. , 2008bSingh et al. 2006;Stacy et al. 2006). Given the importance of NRF1 ( Figure 4) and NRF2 (Pi et al. 2003) in regulating the expression of AREdependent genes induced by iAs 3+ , it is highly possible that NRF1 and/ or NRF2 activation plays a pathogenic role in skin dis orders chronically induced by arsenic exposure, including carcinogenesis, although additional research is required to confirm this.
Apoptosis normally functions to control the integrity of cell populations by eliminating aberrant clones, whereas failure of apoptosis likely is a key contributor to tumor initiation and progression, as well as drug resistance in skin cancer and cancer in general (Guzman et al. 2003;Hanahan and Weinberg 2000). Thus, an acquired, generalized apoptotic resis tance is an important event in the process of arsenicinduced malignant transformation (Pi et al. 2008). Our previous data indicated that HaCaT cells chronically treated with iAs 3+ show a generalized resistance to apoptosis and malignant transformation, which may be associated with enhanced basal NRF2 activ ity (Pi et al. 2005(Pi et al. , 2008. Here, for the first time we report that NRF1 also contributes to iAs 3+ induced AREdependent gene expres sion and protects cells from acute arsenic tox icity, suggesting that NRF1 may be another key transcription factor in arsenic carcino genesis. However, whether NRF1 activation is involved in acquired apoptotic resistance in malignant transformation induced by chronic iAs 3+ exposure needs further study. It has been predicted that human NRF1 gene may transcribe at least four different transcripts with alterative first exons, differ ential splicing, and alterative poly adenylation (Biswas and Chan 2010). In addition to the long isoforms as we observed in HaCaT cells, a 65kDa isoform of mouse NRF1 has been identified and shown to potentially function Figure 5. Effect of iAs 3+ on protein expression of NRF1, NRF2, and KEAP1 in Scr and NRF1-KD cells. HaCaT cells were exposed to vehicle (medium) or 10 µM iAs 3+ for indicated times (A) or to vehicle (medium) or indicated concentrations of iAs 3+ for 6 hr (B). Vehicle Vehicle as a dominant negative inhibitor of ARE mediated transcription (Wang et al. 2007). Although we observed two bands close to 65 kDa on Western blots in the present study (Figure 1), neither corresponded to NRF1 silencing or were altered by iAs 3+ exposure, suggesting that they may represent non specific binding of the anti body used for analysis. This discrepancy, which could be due to differences in cell types, treatment, and anti bodies used for immuno blotting, needs further study.

NRF1-KD
Previous studies have suggested that NRF1 is sequestered in ER and that oxidative stress activates NRF1 by permitting accumulation into the nucleus (Biswas and Chan 2010). The ER is a central organelle as the place of lipid synthesis, protein folding, and protein maturation (Banhegyi et al. 2007). As a major intra cellular calcium storage compartment, the ER also plays a critical role in maintenance of cellu lar calcium homeo stasis (Li et al. 2006). ER stress (conditions interfering with the func tion of ER) can be induced by accumulation of unfolded proteins and excessive protein traf fic (Banhegyi et al. 2007;Li et al. 2006). ER stress could also be elicited in the cell culture system by pharmaco logi cal agents, including TU, BFA, and TG, through distinct molecular mechanisms (Li et al. 2006;Shang et al. 2002;Thastrup et al. 1990). Consistent with previous studies using recombinant human or murine NRF1 (Wang and Chan 2006;Zhang et al. 2009b), treatment of HaCaT cells with TU, an inhibitor of Nlinked protein glycosylation (Shang et al. 2002), resulted in faster migration of NRF1 isoforms on SDSPAGE, suggest ing that long isoforms of endogenous human NRF1 are glycosylated proteins. In contrast, BFA, which blocks protein transport from ER to Golgi (Li et al. 2006), led to accumulation of slower migrating NRF1 proteins, suggesting that NRF1 may be further glycosylated in ER if its transportation to Golgi is blocked. TG, which blocks ER uptake of calcium by inhibit ing sarcoplasmic/endoplasmic Ca 2+ ATPase (Thastrup et al. 1990), slightly decreased iAs 3+ induced NRF1 accumulation but did not affect migration on SDSPAGE. The find ing that ER stressors TU, BFA, and TG affect NRF1 migration on SDSPAGE differently suggests that ER stress may not be a common mechanism for NRF1 modification. Although ER is an important organelle for NRF1 post translational modification and may be involved in NRF1mediated anti oxidant response, the molecular basis for how ER participates in NRF1 activation needs further investigation. Biswas and Chan (2010) have reported that NRF1 and NRF2 have over lapping roles in regulating basal expression of ARE dependent genes. In the present study we found that basal and inducible expression of some AREdriven genes, such as GCLC, GCLM, and NQO1, are highly dependent on NRF1. However, the induction of HMOX1 by high concentrations of iAs 3+ was indepen dent of NRF1, suggesting that HMOX1 is not regulated by NRF1. It should be noted that NRF1/NRF2independent mechanisms for iAs 3+ induced expression of GCLC and GCLM have been demon strated in murine hepato cytes and mouse embryo fibro blasts (Thompson et al. 2009). This inconsistency with the present study suggests that forms of human and mouse NRF1 behave differently or, more likely, reflects differences between the cell types evaluated in in vitro assays. As with NRF2, NRF1 has been postulated to inter act with KEAP1 (Biswas and Chan 2010), although the biological significance of this reaction is poorly charac terized. In the present work, lack of NRF1 in HaCaT cells did not disturb iAs 3+ induced NRF2 accumulation but noticeably decreased KEAP1 protein lev els under basal and iAs 3+ exposed conditions, suggesting a potential inter action between NRF1 and KEAP1. If KEAP1 could serve as a negative regulator of NRF1, decreased KEAP1 expression caused by NRF1 silencing may rep resent a compensation mechanism to maintain the overall cellular ARE activity. However, this hypothesis needs further investigation.
In the present study, we found convinc ing evidence that NRF1 is involved in the regulation of the ARE gene battery induced by iAs 3+ and contributes to the resistance against iAs 3+ induced cyto toxicity and apoptosis. Importantly, we demonstrated arsenic activa tion of NRF1 in a human skin cell line, impli cating an NRF1mediated oxidative stress response cascade as an important event in a potential target cell of arsenic carcinogenesis. Given the potential importance of oxidative stress in arsenic dermal toxicity and carcino genicity, as well as the critical role of NRF1 in the defense against oxidative damage, our findings provide an important insight into the mechanism of chronic arsenic dermal toxicity.