NRF2 Is a Potential Modulator of Hyperresistance to Arsenic Toxicity in Stem-Like Keratinocytes

Arsenic is a well-known human carcinogen. Stem cells are indicated to be involved in arsenic carcinogenesis and have a survival selection advantage during arsenic exposure with underlying mechanisms undefined. In the present study, we demonstrated that CD34high-enriched cells derived from HaCaT human keratinocytes showed stem-like phenotypes. These cells were more resistant to arsenic toxicity and had higher arsenic efflux ability than their mature compartments. The master transcription factor in antioxidant defense, nuclear factor erythroid 2-related factor 2 (NRF2) with its downstream genes, was highly expressed in CD34high-enriched cells. Stable knockdown of NRF2 abolished the hyperresistance to arsenic toxicity and holoclone-forming ability of CD34high-enriched cells. Our results suggest that skin epithelial stem/progenitor cells are more resistant to arsenic toxicity than mature cells, which is associated with the high innate expression of NRF2 in skin epithelial stem/progenitor cells.


Introduction
Arsenic is naturally distributed in the environment as a component of water and soil and is also a contaminating metalloid dispersed by human activity [1]. Although arsenic has been identified as a human carcinogen with the skin as one of the targets [2], its carcinogenic mechanisms remain elusive. Transplacental carcinogenic activity of arsenic observed in humans [3,4] and in experimental mouse models [5,6] suggests that the fetal stem cell is a key target during carcinogenesis of this metalloid. Chronic exposure to arsenic has been shown to induce overabundance of stem cells or cancer stem cells during malignant transformation in various human cell lines [7][8][9], indicating a stem cell survival selection advantage in response to arsenic toxicity regardless of the tissue origin. With respect to the skin, in vitro models observed that short-term arsenic exposure maintained epidermal cells in a geminative state, blocking differentiation of the putative stem/progenitor cells [10]; chronic arsenic exposure induced an elevation in CD34-positive stem-like cancer cells during the process of malignant transformation of HaCaT cells [7]. In vivo study observed that in utero exposure to arsenite in mice led to disrupted homeostasis of skin stem cell [11] and increased risk of skin tumors with an accumulation of skin cancer stem cells as marked by CD34 [5]. However, no direct evidence for the effect of arsenic on skin stem cells is available.
Considering the observed overabundance of stem cells after arsenic exposure and the active role of NRF2 in chemical resistance [24][25][26][27], it is reasonable to assume that NRF2 is involved in stem cell biology and stem cell response to arsenic. Previous studies indicate that NRF2 regulates stem cell maintenance and differentiation; however, the reports are not consistent. For example, a reduction of high NRF2 activity induced cell differentiation and inhibited pluripotency in human embryonic stem cells (hESCs) [28], whereas in muscle and neural stem/progenitor cells, more evidence supported that NRF2 was not highly expressed or activated until proliferation or differentiation occurred and therefore played roles in tissue regeneration [29,30]. It appears that the function of NRF2 in homeostasis of stem cells depends on cell types and specific physical/pathological conditions. The role of NRF2 in stem cells, especially when they are under insult or stress, requires to be further elucidated.
CD34 is a surface marker for stem cells and cancer stem cells in the skin [31,32]. In the present study, CD34 highenriched cells isolated from HaCaT human keratinocytes showed stem-like phenotypes and were used as a stem/progenitor cell model. Then, we tested our hypothesis that skin stem/progenitor cells possess a survival selection advantage in response to arsenic exposure. Furthermore, we found that NRF2 was more active in these stem/progenitor cells compared with the mature compartments and contributed to the hyperresistance to arsenic toxicity in these stem/ progenitor cells.

Cell
Viability. Cells were seeded in the 96-well plate at a density of 1.5 × 10 4 cells per well and incubated at 37°C with 5% CO 2 overnight. After treated with sodium arsenite at indicated concentrations for 24 h, cell viability was assayed using a CellTiter 96 AQueous One Solution Cell Proliferation Assay (MTS) Kit (Promega, Madison, WI) according to the manufacturer's instruction. This method measures the production of formazan catalyzed by mitochondrial dehydrogenase of viable cells. Formazan production was measured at a wavelength of 490 nm using a Synergy H1 microplate reader (Biotek, Vermont, USA).

Apoptosis
Detected by Flow Cytometry. 8 × 10 5 cells per well were plated in the 6-well plate and allowed to attach overnight. Then, cells were treated with 20 μM or 40 μM sodium arsenite for 24 h, harvested with trypsinization, and centrifuged at 300 ×g for 5 min. Apoptosis was detected with TACS Annexin V Kit (4830-250-K, Trevigen) according to the manufacturer's protocol. Annexin V-FITC and propidium iodide-(PI-) staining cells were quantified with Canto II flow cytometer (Becton-Dickinson) equipped with Cell-Quest software (Becton-Dickinson).

Reverse Transcriptase Quantitative Polymerase Chain
Reaction (RT-qPCR). RT-qPCR was performed as previously described [9]. Briefly, total RNA was isolated using TRIzol reagent (Invitrogen, Carlsbad, CA), purified with RNeasy mini kit columns (Qiagen, Valencia, CA), and reverse transcribed to cDNA using Prime Script RT reagent kit (TaKaRa, Dalian, China). SYBR Premix Ex Taq Kit (TaKaRa) was used to assess cDNA amplifications. Data were analyzed using the delta-delta cycle time method. All primers were designed with Primer-BLAST online (http://www.ncbi.nlm.nih.gov/ tools/primer-blast) and obtained from Sigma-Aldrich. Primer sequences are provided in supplemental material Table S1 available online at https://doi.org/10.1155/2017/ 7417694.
2.10. Cellular Arsenic Accumulation and Export. Determination of cellular arsenic accumulation and export was modified from a previous report [35]. Cells were grown in arsenic-free medium to approximately 70% confluence. The medium was replaced with either fresh medium or medium containing 10 μM sodium arsenite. After 24 h, to determine arsenic accumulation, cells were washed three times with PBS, harvested with trypsin, counted, and digested overnight with 50% perchloric acid : nitric acid (2 : 1) at 70°C; to determine arsenic export, PBS-washed cells were incubated in fresh arsenic-free medium for another 24 h. Then, medium was collected, and cells were trypsinized and counted. Arsenic concentration in the cell or medium was determined by graphite furnace atomic absorption spectrophotometry using a Perkin Elmer AAnalyst 100. Arsenic concentrations were normalized to cell number.

Statistical
Analysis. Data were analyzed with GraphPad Prism 5 (GraphPad Software, San Diego, CA) and presented as mean ± standard deviation (SD). Unpaired Student's ttests were used to compare the difference between Veh. and arsenic-exposed cells, as well as between scramble and NRF2-KD cells. Bonferroni posttests after one-way or twoway analysis of variance (ANOVA) were applied to other comparisons when it was appropriate. p < 0 05 was considered significant. All the experiments were carried out at least in triplicate.

Survival Potential of CD34 high Subpopulation in HaCaT
Cells in Response to Arsenic Exposure. CD34 is a surface marker for stem cells in the skin [31,32]. We first assessed the presence of the cells with high expression of CD34 surface marker in HaCaT cells with flow cytometry (Figure 1(a)). CD34 high subpopulation is quite a small fraction, approximately 0.2-0.7% in HaCaT cells. After arsenic challenge (50 μM, 24 h), CD34 high subpopulation was markedly increased to 8.9-11.1% (Figure 1(b)). Stem cells are resistant to chemical insult and are suggested to act as a "strategic" adaption in arsenic carcinogenesis [27]. One possible explanation for the observed increase of CD34 high subpopulation is the hyperresistance of CD34 high cells to arsenic-induced cytotoxicity. We tested this idea in the following study.

Stem-Like Characteristics of CD34 high -Enriched Cells.
Subsequently, CD34 high fraction was separated by positive MACS sorting and a CD34 high -enriched population was obtained (Figure 2(a)). Markers for undifferentiated epithelial cells such as p63, K5, and K14 and those for stem cells such as CD34, OCT4, and SHH were consistently expressed at higher mRNA levels and protein levels in CD34 highenriched cells than CD34 low -expressing and parent cells (Figures 2(b) and 2(c)). Holoclone formation was used to indicate potential self-renewal characteristics of stem cell. Freshly isolated CD34 high -enriched cells formed more holoclones, which are large and round colonies with a smooth boundary mostly containing small cells, than CD34 lowexpressing and parent cells in 2 weeks (Figure 2(d)). Moreover, holoclones formed by CD34 high -enriched cells appeared larger than those generated by CD34 low -expressing and parent cells (Figure 2(d)). Holoclones formed in all cell groups could be repeatedly subcultured and generate more holoclones. Though no specific in vitro assay can confirm the identity of stem cells, the above evidence supports that CD34 high -enriched subpopulation isolated from HaCaT cells exhibits stem-like phenotypes.

Increased Resistance of CD34 high -Enriched Cells to
Arsenic Toxicity. To test cellular defense against arsenic toxicity, CD34 high -enriched cells, CD34 low -expressing cells, and parent cells were acutely (24 h) exposed to sodium arsenite. As shown in Figure 3(a), CD34 high -enriched cells showed significantly increased viability compared with CD34 low -expressing and parent cells in response to arsenic from 20 μM to 80 μM. The lethal concentration 50 (LC50) was 88.1 μM for CD34 high -enriched cells, 58.8 μM for CD34 low -expressing cells, and 56.7 μM for parent cells. Then, arsenic-induced apoptosis was measured with Annexin V-FITC and PI double staining method. Consistent with the results of cell viability, less apoptotic cells were observed in high-level arsenic-exposed CD34 highenriched cells compared with CD34 low -expressing and parent cells (Figures 3(b) and 3(c)).

High
Activity of NRF2 and ROS Levels in CD34 high -Enriched Cells. Western blot showed higher innate and arsenic-induced NRF2 protein levels in CD34 high -enriched cells than CD34 low -expressing and parent cells (Figure 4(a)). Consistently, a panel of NRF2 downstream genes, including those encoding multidrug resistance proteins (MRPs) such as ABCG2 and ABCC1, and antioxidant enzymes HMOX1 and GCLC, was increased at the transcript level in CD34 high -enriched cells compared with CD34 low -expressing and parent cells (Figure 4(b)). The intracellular ROS levels in CD34 high -enriched cells were surprisingly higher than CD34 low -expressing and parent cells under basal conditions (Figures 4(c) and 4(d)). Interestingly, ROS levels were increased to 1.5-2 fold of Veh. in CD34 low -expressing and parent cells in response to arsenic, but remained stable in CD34 high -enriched cells (Figures 4(c) and 4(d)).

Increased Export of Arsenic from CD34 high -Enriched
Cells. Some NRF2 downstream genes, including MRPs and GCLC, are involved in intracellular arsenic export, and their high expression may contribute to arsenic resistance [35]. Thus, we evaluated arsenic accumulation and export in CD34 high -enriched cells after acute arsenic exposure. CD34 high -enriched cells showed significantly lower intracellular arsenic levels than CD34 low -expressing cells after 24 h arsenic exposure ( Figure 5(a)). The reduced intracellular arsenic accumulation may be caused by enhanced arsenic export. As shown in Figure 5(b), after acute exposure to the same level of arsenic, CD34 highenriched cells eliminated more accumulated arsenic into the medium in the next 24 h compared with CD34 lowexpressing and parent cells.
3.6. Abrogation of Hyperresistance to Arsenic Toxicity in CD34 high -Enriched Cells due to NRF2 Silencing. Since resistance to arsenic toxicity and increased cellular arsenic export of CD34 high -enriched cells can be attributed to high NRF2 activity, we subsequently examined what if NRF2 was silenced. NRF2-KD did not change the proportion of CD34 high cells in HaCaT cells (Figure 6(a)), suggesting that reduced NRF2 expression does not affect expression of CD34. CD34 high -enriched cells isolated from NRF2-KD cells showed significant decreases in NRF2 mRNA levels (Figure 6(b)), which were approximately 50% of scramble. Western blot showed that NRF2 protein expression was obviously reduced in NRF2-KD cells than scramble cells, especially when cells were treated with 20 μM sodium arsenite (Figure 6(c)). Decreased cell viability in NRF2-KD CD34 high -enriched cells compared with HaCaT cells were exposed to vehicle (Veh., same volume of PBS to arsenic solution); right: HaCaT cells were exposed to 50 μM sodium arsenite for 24 h. (b) Quantification of CD34 high population in HaCaT cells exposed to Veh. and arsenic. 1 × 10 5 cells were counted for each sample. n = 3. * p < 0 05 compared with Veh.  scramble compartments in response to 10 μM to 80 μM arsenic, ranging from 9% to 29%, was observed. (Figure 6(d)). In addition, NRF2-KD in CD34 high -enriched cells did not affect the number of holoclones but reduced the size of holoclones compared with scramble ( Figure 6(e)). Meanwhile, cells in NRF2-KD holoclones were not as small or as compact as in scramble (Figure 6(e)).

Discussion
In the present study, we found that CD34 high -enriched cells purified from the HaCaT cell line exhibited a stem-like phenotype. In addition, the CD34 high stem-like cells showed hyperresistance to arsenic toxicity compared with the mature cells. This hyperresistance was associated with higher innate  expression of NRF2 and its downstream genes in CD34 high cells. These data support the hypothesis that skin stem/progenitor cells have a survival selection advantage in response to arsenic exposure, which serves as a possible mechanism for the observed overabundance of stem cells during arsenic carcinogenesis [7][8][9]. Our results are consistent with the previous report by Tokar et al. [27], which found that the WPE human prostate stem cells showed innate resistance and hyperadaptability to arsenic with higher transcript expression of NRF2, ABCC1, and GST-pi than the parent RWPE-1 cells.
Although the role of stem cells in arsenic adaptation is still poorly understood, stem cells are usually resistant to chemical toxicity and act as a "strategic" reserve for wound repair from chemical insult [26,37]. Resistance to toxins of stem cells is mostly observed in cancer stem cells and is largely attributed to overexpression of antioxidant proteins and low intracellular levels of reactive oxygen species. As a master transcription factor for antioxidant proteins, NRF2 is likely to account for the toxic-resistance feature of normal stem cells and cancer stem cells [38]. With respect to normal stem cells, exogenous NRF2 transduction or pharmacological NRF2 activation (e.g., tBHQ or melatonin) prevented toxininduced cell death in human or mouse neural stem cells [30,39,40]. In rat cardiac stem cells, NRF2 upregulation induced by resveratrol was found to enhance cell survival and improve cardiac function [41]. In murine mesenchymal stem cells, NRF2 upregulation reduced oxidative stressinduced apoptosis and cytotoxicity [42]. A normal human prostate stem cell line, the WPE-stem cell, showed innate resistance to arsenic cytotoxicity and high transcript expression of antiapopototic, antistress, and drug adaption genes including NRF2 [27]. As for cancer stem cells, higher expression or activity of NRF2 has been demonstrated in glioma, lung, esophageal, breast ovarian, and colon stem cells [43][44][45][46]. Our study provides the direct evidence that skin stem/progenitor cells preserve high NRF2 activity. The difference in cell viability between scramble and NRF2-KD CD34 high -enriched cells ranged from 9% to 29%, which is relatively mild and indicates that other mechanisms independent of NRF2 are involved in cell resistance to arsenic. However, this difference is close to what we have observed in comparison of cell viability between CD34 low -expressing and CD34 high -enriched cells. Thus, the relatively high NRF2 expression/activity may play a key role in hyperresistance to arsenic of skin stem/progenitor cells than their mature compartments.
NRF2 is considered to facilitate chemical resistance through several mechanisms: (1) upregulating downstream antioxidants and detoxifying enzymes such as GCLC, GCLM, and HOMX1 [47]; (2) inducing MRPs that export chemicals from the cell [18][19][20]; (3) interacting with antiapoptotic pathway resulting in attenuation of chemicalinduced apoptosis [48]; and (4) enhancing proteasome activity leading to activation of the survival-related pathway [49]. In the present study, a panel of NRF2 downstream genes was highly expressed in CD34 high -enriched cells. High transcript levels of GCLC indicate a potential enhanced glutathione (GSH) synthesis. In addition to antioxidant property, GSH mitigates arsenic toxicity by increasing arsenic export through formation of arsenic-GSH complexes [50]. These complexes can be formed either nonenzymatically [51] or enzymatically by GST-pi [52]. GSH also affects arsenic resistance by participating in arsenic methylation through reduction of pentavalent arsenic species to trivalent ones [53]. However, HaCaT cells are arsenic-methylation deficient. MRPs are export pumps in xenobiotic metabolism and  showed higher transcript levels in CD34 high -enriched cells. All are consistent with the observed low accumulation but high export of arsenic in CD34 high -enriched cells. Therefore, in addition to the well-known antioxidant activation, elevated arsenic export associated with high NRF2 activity may be involved in hyperresistance to arsenic in CD34 high stem-like cells.
NRF2 also plays an important role in stem cell biology; however, the reports are controversial. The homolog of NRF2 in Drosophila, CncC, was continuously active in intestinal stem cells of Drosophila and maintained quiescence of stem cells [54]. CncC inhibition mediated by KEAP1 was required for proliferation of these stem cells [54]. In human embryonic stem cells (hESCs), NRF2 was highly expressed compared with differentiated cells, while a reduction of NRF2 activity induced cell differentiation and inhibited pluripotency of the stem cells [28]. In human airway basal stem cells, NRF2 activation promoted cell proliferation by regulating NOTCH1 and antioxidant program [55]. In skeletal muscle stem/progenitor cells, NRF2 supported cell proliferation by transcriptional upregulating MyoD and facilitated muscle regeneration [29]. Increased NRF2 activity enhanced both proliferation and differentiation of neuronal progenitor cells [30]. In hematopoietic stem cells (HSCs), NRF2 activation was required for maintenance of quiescence [56,57]. Meanwhile, NRF2 activation determined differentiation fate of HSCs, driving the myeloid differentiation but suppressing erythoid or lymphoid differentiation [58]. Though most studies support that NRF2 is involved in pluripotency, proliferation, and differentiation of stem cells, it appears that the exact role of NRF2 in homeostasis of stem/progenitor cells is tissue and cell-type specific. In the present study, NRF2 silencing does not affect population size of stem/progenitor cells under basal conditions. However, NRF2-KD CD34 highenriched cells formed smaller holoclones compared with scramble compartments, indicative of the role of NRF2 in self-renewal in these cells. NRF2 activity is suggested to be closely associated with cell cycle regulation. Cell cycle was arrested in G1 phase in NRF2-knockdown human breast cells with mitoquinone treatment [59]. While G2/M arrest resulted from NRF2 deficiency was observed in mouse alveolar epithelia [60] and human glioma stem cells [61]. Therefore, our data are consistent with previous findings and provide new information on the role of NRF2 in skin stem/ progenitor cells.
Stem cells undergoing self-renewal are indicated to possess low intracellular ROS levels, while elevated ROS levels switch stem cell self-renewal to differentiation [62]. As a potent regulator of stem cell self-renewal and differentiation, NRF2 is suggested to function at least in part due to its antioxidant capacity. However, the functions of NRF2 through regulating ROS in stem cells are diversely reported and presumably context dependent. In some cases, NRF2 may regulate proliferation of stem cells independent upon ROS [56], but directly bind to ARE in promoter region of some stem cell self-renewal genes, such as CXCR4 [57] and NOTCH1 [55]. It is quite interesting to observe that CD34 high -enriched cells presented with higher NRF2 expression/activity, as well as markedly elevated intracellular ROS levels, than mature compartments under basal conditions. This unexpected high intracellular ROS levels in the putative skin stem-like cells may suggest diverse redox balance in different cell/tissue types and even in cells of the same origin with different differentiating stages. However, the underlying mechanisms cannot be clearly clarified in the present study. Arsenic exposure is well-known to induce enhanced NRF2 protein levels as we have seen in the present study (Figure 4(a)). This NRF2 induction was accompanied with increased ROS levels in the mature compartments (CD34 low -expressing and parent cells), but without alteration of ROS levels in CD34 highenriched cells (Figures 4(c) and 4(d)). One possible explanation for this phenomenon is that innate high expression of NRF2 and antioxidant proteins in CD34 high -enriched cells veiled the alteration of ROS at the determination time point. Also, arsenic can activate NRF2 through other mechanisms such as NRF2 stabilization by p62-mediated sequestration of KEAP1 resulting from autophagy inhibition [63].