Arsenic leads to autophagy of keratinocytes by increasing aquaporin 3 expression

Exposure to arsenic, a ubiquitous metalloid on Earth, results in human cancers. Skin cancer is the most common arsenical cancers. Both autophagy and aquaporin pathway are known to promote carcinogenesis. However, the mechanisms by which arsenic regulates aquaporin and autophagy in arsenical skin cancers remain elusive. This study aims to address how arsenic regulates aquaporin-3, the predominant aquaporin in epidermal keratinocytes, and how this process would induce autophagy. Quantitative real-time PCR and immunofluorescence were used to measure the expression of aquaporin 3 in arsenical skin cancers and arsenic-treated keratinocytes. Beclin-1 expression and autophagy were measured. We examined if blocking aquaporin 3 could interfere arsenic-induced autophagy in keratinocytes. Expression of aquaporin 3 is increased in arsenical cancers and in arsenic-treated keratinocytes. Arsenic induced autophagy in primary human keratinocytes. Notably, the arsenic-induced autophagy was inhibited by pretreatment of keratinocytes with aquaporin inhibitors Auphen or AgNO3, or RNA interference against aquaporin 3. The data indicates that the aquaporin 3 is an important cell membrane channel to mediate arsenic uptake and contributes to the arsenic-induced autophagy.

Another mechanism underlying As-BD is mitochondrial biogenesis 10,11 . In As-BD, mitochondrial biogenesisrelated genes are overexpressed. Enhanced mitochondrial biogenesis leads to aberrant keratinocyte proliferation, and blocking of mitochondrial function abrogates arsenic-induced keratinocyte proliferation 10 . Epigenetic modification also plays a role in arsenic-induced carcinogenesis 12 . In an skin-equivalent organotypic culture model consisting of keratinocytes, fibroblasts, and peripheral blood mononuclear cells, arsenic treatment leads to pathognomonic characteristics of As-BD. By using this model, we identified arsenic induces epigenetic modification of E2F1 promotor, which leads to centromere amplification and subsequent caspase-8-mediated apoptosis of keratinocytes 12 .
An important pathognomic feature of As-BD is the coexistence of uncontrolled cell proliferation, abnormal differentiation, along with individual cell death among the epidermal keratinocytes. Autophagy is reported to be involved in carcinogenesis of skin. A report showed autophagy gene ATG7 regulates ultraviolet radiation-induced inflammation and skin tumorigenesis 13 . For biological samples, beclin-1 protein is a well-established biomarker to measure the development of autophagy 14 . In addition to beclin-1 expression, the autophagic flux is used to detect autophagy in various diseases 15 . LC3 proteins are cleaved by Atg4 to become LC3-I, which then conjugates to phosphatidylethanolamine to form LC3-II during autophagy 15 . Thus, LC3 decreases while LC3-II increases in autophagy induction 16 . Hence, the ratio of LC3-II/LC3-I could be another useful marker for autophagic flux. On the other hand, p62 is a link between LC3 and ubiquitinated substrates and is degraded during autophagy. Therefore, autophagic activation correlates with a decreased p62 level, while autophagic suppression correlates with an increased p62 level 17 . While autophagy is regarded as an important tumor facilitator, the exact role of autophagy in As-BD remains to be elucidated. Another mechanism that may contribute to skin carcinogenesis is transportation of arsenic. Transport systems responsible for uptake and toxicity of inorganic arsenic can be divided into trivalent arsenite uptake and pentavalent arsenate uptake 18 . Because arsenic is not required for life, there seem no arsenic-specific uptake systems. Pentavalent arsenate is taken into most cells adventitiously by phosphate uptake systems while trivalent arsenite is taken into most cells primarily via aquaglyceroporins or sugar permeases. Trivalent arsenites are more toxic than the pentavalent arsentes. The first trivalent arsenite uptake system was identified in 1997 19 . As(III) and trivalent antimony, Sb(III), were demonstrated to be transported into cells of Escherichia coli by the glycerol channel GlpF, which belongs to the aquaporin superfamily 19,20 .
Since mechanisms contributing to As-BD such as epigenetic regulation 21 and mitochondria 22 are also related to aquaporin 3 (AQP3), it is intriguing to explore if AQP3 plays a role in the pathogenesis of As-BD. More specifically, we aim at investigating if arsenic is transported into keratinocytes through AQP3 in As-BD and, if so, how this process would induce autophagy to facilitate carcinogenesis. Herein, we showed expression of aquaporin 3, the most common aquaporin in keratinocytes, is up-regulated in arsenical skin cancers and arsenictreated keratinocytes. Arsenic induces autophagy in keratinocytes, and blocking AQP3 interferes arsenic-induced autophagy. These data indicate that AQP3 is an important membrane protein to mediate arsenic uptake in keratinocytes and that AQP3 contributes to the arsenic-induced autophagy.

Results
Increased AQP3 expression in As-BD. The human As-BD tissues showed typical full-layer epidermal dysplasia, epidermal hyperplasia, and individual cell apoptosis. Because AQP3 may facilitate arsenic absorption into keratinocytes, we asked whether AQP3 was increased in As-BD skin lesions (Fig. 1). The result of quantitative real-time PCR showed that mRNA of AQP3 is increased in lesional and perilesional As-BD than that in normal control skin (Fig. 1a). We further examined if increased AQP3 expression could be reproduced in protein level. By using immunofluorescence, the results showed that samples harvested from lesional and perilesional As-BD showed positive immunofluorescence of AQP3 (Fig. 1b). These experiments confirmed that expression of AQP3 is increased in As-BD.
Increased AQP3 expression in arsenic-treated keratinocytes. To further investigate the mechanisms how expression of AQP3 may contribute to pathogenesis of As-BD, we used arsenic-treated keratinocytes to measure the mRNA and protein expression of AQP3 in vitro (Fig. 2). Consistent with what was observed in As-BD, real-time PCR analysis showed slight upregulated AQP3 in keratinocytes treated with arsenic at 0.1 µM or 1 µM. The AQP3 upregulation was less prominent in keratinocytes pretreated with 5 µM (Fig. 2a). Interestingly, the protein expression of AQP3 was increased in keratinocytes treated with arsenic at 0.1 µM and 1 µM at 24, 48, and 72 h as shown by immunofluorescent exam. For keratinocytes treated with arsenic at 5 µM, immunofluorescence exam showed an increased AQP3 expression at 24 h while the intensity of immunofluorescence was lesser at 48 and 72 h, suggesting a partial toxicity of arsenic or other compensatory mechanisms after 48 h treatment at this concentration (Fig. 2b). While immunofluorescence (Fig. 2b) and quantitative real-time PCR (Fig. 2a) both showed increased AQP3 expression, western blotting (Fig. 2c,d; Supplementary Fig. S1-S4) from the whole cell lysate did not show significant difference of AQP3 expression. This discrepancy may result from the increased AQP3 expressions mainly on cell membrane that were detected better by immunofluorescence than by western blot. These data suggested that arsenic may induce expression of AQP3 on keratinocyte surfaces while its biological effect remains to be clarified. AQP3 siRNA transfection partially inhibited arsenic-induced expression of AQP3 and autophagy. We then asked whether AQP3 mediated autophagy induced by arsenic. To do so, we transfected cells with AQP3 siRNA and see whether blocking AQP3 would inhibit arsenic-induced autophagy. To avoid off-target effects due to long incubation and transfection process, we examined results at 6 h with higher arsenic concentration (5 µM). As expected, transfection of AQP3 siRNA significantly reduced AQP3 mRNA expression in keratinocytes treated with arsenic at 5 µM for 6 h as compared to control siRNA ( Fig. 3a)  www.nature.com/scientificreports/ Immunofluorescence exam showed arsenic induced the expression of AQP3 in a small fraction of cells, while AQP3 siRNA transfection inhibited the surface expression of AQP3 in arsenic-treated keratinocytes (Fig. 3b). Autophagy assay kit also showed that arsenic induced autophagy in keratinocytes. Reduced autophagy was observed in keratinocytes transfected with RNA interference against AQP3 (Fig. 3c). These data indicated that arsenic-induced autophagy is mediated, at least partially, by AQP3.
Arsenic-induced autophagy was inhibited by pretreatment with AgNO 3 , a chemical inhibitor for AQP3. The cell toxicity of AgNO 3 is low below 50 µM (Fig. 4a). To compile further evidences whether AQP3 might mediate arsenic-induced autophagy in keratinocytes, we pretreated cells with AgNO 3 , a chemical inhibitor for AQP3, and investigated whether arsenic-induced autophagy could be inhibited. The autophagy was measured by an autophagy assay kit, showing arsenic at 5 µM significantly induced autophagy (Fig. 4b). Notably, after pretreatment with AgNO 3 , the percentages of autophagic keratinocytes were decreased significantly, particularly in those pretreated with AgNO 3 at 1 and 10 µM (Fig. 4b). Quantized data from the autophagy assay kit showed the inhibition effect of AgNO 3 on arsenic-induced autophagy was dose-dependent ( Fig. 4c).
Arsenic-induced AQP3 expression was reduced by pretreatment of Auphen, another chemical inhibitor for AQP3. Since AgNO 3 might have off-targeted effects other than AQP3 inhibition, we synthesized Auphen, another chemical inhibitor, that was known to inhibit AQP3 23 , and tested whether this compound could inhibit autophagy induced by arsenic. Our result showed that cell viability of Auphen was low (Fig. 5a), with a concentration leading to 50% cell viability (EC50) lies between 10 and 50 µM. In cells treated with arsenic at 1 µM, the beclin-1 expression as measured by immunofluorescence was not significantly different whether cells were pretreated with Auphen or not. In cells treated with arsenic at 5 µM, Auphen pretreatment for 1 h reduced the expression of beclin-1 at 24 h after arsenic treatment (Fig. 5b). We next determined whether arsenicinduced autophagy could be rescued by Auphen. To do that, we measured the expression of autophagy-related markers, including beclin-1, LC3, and p62 by western blot. Consistent with immunofluorescence findings, western blot showed that enhanced expression of beclin-1 as well as AQP3 by arsenic was inhibited by pretreatment of Auphen and the inhibition occurred as early as 1 h after arsenic treatment (Fig. 5c,d; Supplementary  Fig. S5-S9). LC3 was induced by arsenic, however, no significant decreases were found in cells pretreated with Auphen. Intriguingly, p62 expression was not changed significantly in arsenic-treated keratinocytes up to 3 h and its expression was even robustly increased in Auphen-pretreated keratinocytes 24 h after arsenic treatment (Fig. 5d).

Discussion
AQP3 plays a role in hydration in epidermis, and the expression of AQP3 is changed in skin diseases such as atopic dermatitis and psoriasis 24,25 . In the present study, we show AQP3 expression is increased in As-BD skin lesions and arsenic-treated keratinocytes. While arsenic induces autophagy in keratinocytes, this can be reserved by RNA interference against AQP3 as well as chemical inhibitors for AQP3, including AgNO 3 and Auphen. www.nature.com/scientificreports/ Considering that arsenic exposure due to drinking water contamination remains to be a global issue not only in undeveloped countries but also in developed countries such as USA 2 , any interventions that can reduced uptake of arsenic into keratinocytes provides potential therapeutics to prevent As-BD and possibly arsenicrelated cancers.
In views of that drinking water with potential arsenic contamination cannot be rapidly changed due to various issues such as government infrastructural construction and unclear and/or undisclosed quantification of arsenic concentration of drinking water in various areas of the world, the development of a compound or medication that can prevent uptake of arsenic into cells is a practical and useful strategy to reduce arsenic-related cancers. In the present study, we identify that Auphen, also known as 1,10-phenanthroline gold(iii) dichloride, inhibits arsenic-induced autophagy. This compound or other chemicals may be used as a medication to prevent reuptake of arsenic into human cells and subsequent arsenical cancers, although further preclinical studies need to be done to further support this prevention strategy, including short term toxicity, long term safety, and animal experiments.
Our study showed that both the chemical inhibitors for aquaporin, AgNO 3 and Auphen, could inhibit the autophagy or beclin-1 expression in arsenic-treated keratinocytes. However, in Auphen pretreatment experiments (Fig. 5), although LC3 was induced by arsenic, this induction was not decreased in cells pretreated with Auphen. More intriguingly, p62 expression was dramatically increased in Auphen-pretreated keratinocytes 24 h after arsenic treatment. It is possible that measurement of p62 expression and LC3 by western blot may not be the optimal way to measure autophagy. The autophagic flux is at best confirmed by GFP-mRFP-LC3, not by western-blotting of LC3 and p62. In addition to its role in autophagy, p62 may mediate keratinocyte proliferation in psoriasis, a disease with hyperproliferative epidermis by keratinocytes 26 . P62 is involved in the arsenic-induced keratinocyte proliferation too 27 . Therefore, the induction of p62 expression in Auphen-pretreated keratinocytes at 24 h after arsenic treatment may have biological roles other than autophagy induced by arsenic. Auphen may exert biological effects other than inhibition of AQP3. The bottom line is that we showed Auphen, a chemical inhibitor for AQP3, inhibited arsenic-induced expression of beclin-1 and autophagy.
There are limitations of the present study. In this study, we didn't measure arsenic concentrations of tissue samples from the patients with As-BD. Although, technically, arsenic concentrations can be measured from www.nature.com/scientificreports/ human samples such as hairs and nails, the linear correlations between tissue concentrations and skin lesions are not high 28 . In addition, arsenic concentrations in well water vary, and it is difficult to quantify arsenic exposure in each patient. Therefore, we use primary keratinocytes to perform experiments to quantify arsenic exposure and examine autophagy markers to clarify the effects of arsenic on keratinocytes. Finally, as mentioned previously, the autophagic flux is at best confirmed by GFP-mRFP-LC3, not by western-blotting of LC3 and p62.
In conclusions, we demonstrate that a new mechanism how AQP3 facilitates arsenic uptake and may lead to As-BD and that this mechanism can be inhibited by an Au compound. The present study highlights a potential therapeutic for arsenical cancers. More pre-clinical and human clinical studies are warranted to validate the role of Au compound to reduce incidence of As-BD and other arsenical cancers.

Methods
Study subjects. Twenty-five biopsy-proven As-BD tissue samples were obtained from 15 patients (10 males and 5 females, age range: 62-82 years). The As-BD was diagnosed based on the typical pathological findings from human subjects, which showed full-layer epidermal dysplasia, epidermal proliferation, and individual cell apoptosis. All patients have arsenic well water exposure history since childhood. Because vitiligo is a confounder of AQP3 expression, and therefore patients with vitiligo were excluded in our present study 29 . Control skin of As-BD patients was harvested from patients' normal skin by minimal skin biopsy (for routine skin surgery to remove benign skin tumors). The procedures were described in the IRB, and all patients signed informed consent. Normal human primary keratinocytes were obtained from adult foreskin through routine circumcision. Written informed consent was obtained from all participants to obtain As-BD tissues and primary keratinocytes. This work has received approval for research ethics from Institutional Review Board at Chang Gung Memorial Hospital (104-9380A3) and a proof/certificate of approval is available upon request. All methods were performed in accordance with the relevant guidelines and regulations.
Primary keratinocyte culture. The method for keratinocyte cultivation is described as previously reported 7 . Briefly, skin specimens were washed with phosphate-based saline (PBS), cut into small pieces, and incubated in medium containing 0.25% trypsin (Gibco, Grand Island, NY) overnight at 4℃. The epidermal sheet was lifted from the dermis using fine forceps. The epidermal cells were centrifuged (500 g, 10 min) and the pellets dispersed into individual cells by repeated aspiration. The keratinocytes were gently resuspended in keratinocyte-SFM (serum-free medium) (Gibco) containing 25 lg/ml bovine pituitary extract (BPE) and 5 ng/ ml recombinant human epidermal growth factor (rhEGF). The medium was changed every two days. Keratinocytes at the third passage were then grown in keratinocyte-SFM medium free of supplements 24 h before the experiments. www.nature.com/scientificreports/ Quantitative real-time PCR. Quantification and purification of the RNA were as previously reported 30 .
Briefly, the RNA was measured by A260/A280 absorption (NanoDrop spectrophotometer; Thermo Fisher Scientific, Waltham, MA, USA). RNA samples with ratios greater than 1.7 were stored at − 70 °C for further analysis. Extracted RNA (1 μl) was then subjected to PCR amplification using MPCR kits (Maxim Biotech, San Francisco, CA, USA) according to the manufacturer's instructions 30 . Quantitative real-time PCR was performed using LightCycler 96 real-time PCR system (Roche, Mannheim, Germany). The AQP3 and reference gene β-actin primers were obtained from Genomics (New Taipei City   Inhibition of AQP3 by small interfering RNA (siRNA). The siRNA protocol was adapted from the previously reported 32 . Briefly, cells were seeded to be 70-90% confluent before transfection. Cells were transfected with AQP3 siRNA or negative control (NC) siRNA (both from Dharmacon, Lafayette, CO, USA) using Lipofectamine 3000 Transfection Reagent according to instructions from the manufacturer (Thermo Fisher Scientific). Lipofectamine 3000 Reagent was diluted in siRNA-containing serum-free medium for 15 min at room temperature. The siRNA-lipid complex was then added to the cells for 6 h. Thereafter, the siRNA-lipid complex www.nature.com/scientificreports/ was replaced and incubated with FBS at 37 °C in a humid atmosphere with 5% CO2 before the subsequent western blot and quantitative PCR.
Inhibition by AgNO 3 . Keratinocytes were treated with AgNO 3 , an aquaporin 3 (AQP3) inhibitor, to examine if AgNO 3 suppresses arsenic-induced autophagy. Keratinocytes were pretreated with AgNO 3 at a concentration of 0.1, 1, or 10 µM for 15 min. Keratinocytes were divided into 2 groups based on pretreatment or without pretreatment of AgNO 3 . Arsenic at a concentration of 5 µM or 0 µM (control group) was used to treated keratinocytes with or without AgNO 3 pretreatment to examine the effect of AgNO 3 on arsenic-induced autophagy.
Inhibition by Auphen (C 12 H 8 AuCl 2 N 2 ). Next, we used Auphen, an Au chemical compound that inhibits arsenic-induced AQP3 expression, to examine if chemical inhibition has similar results with siRNA inhibition and AgNO 3 inhibition. Auphen is synthesized accordingly 33,34 . Among gold-based metal compounds, Auphen has the most active binding activity on AQP3 23 . While other AQP3 inhibitors may be toxic, Auphen shows a non-toxic inhibitory effect and therefore has more potential for clinical use 23 . Auphen used in this experiments was synthesized by PJ and JJW with purity more than 99%.
Statistical analysis. All data are expressed as mean ± standard deviation (SD). For multiple-group (more than two groups) comparisons, one-way analysis of variance (ANOVA) with Tukey post hoc test was used for multiple comparisons. A P-value less than 0.05 was considered statistically significant.