Transcriptional Modulation of the ERK1/2 MAPK and NF-κB Pathways in Human Urothelial Cells After Trivalent Arsenical Exposure: Implications for Urinary Bladder Cancer

Abstract

Chronic exposure to drinking water contaminated with inorganic arsenic (iAs) is associated with an increased risk of urinary bladder (UB) cancers in humans. The exact role of specific iAs metabolite(s) in As-mediated carcinogenesis remains largely unknown. Experimental evidence suggests that trivalent arsenicals, namely arsenite (iAs^III) and two of its metabolites, monomethylarsonous acid (MMA^III) and dimethylarsinous acid (DMA^III), are possible proximate UB carcinogens. Here, we used a transcriptomics approach to examine perturbed molecular pathways in a human urothelial cell line (UROtsa) after short-term exposure to iAs^III, MMA^III and DMA^III. Molecular pathways containing genes that encode proteins implicated in UB cancer development were perturbed by both MMA^III and DMA^III. These pathways included those of the extracellular signal-regulated kinase 1/2 mitogen-activated protein kinase (ERK 1/2 MAPK) and nuclear factor kappa beta (NF-κB). Together, these results may inform the current understanding of effects in the UB induced by acute As exposure and the relationship of these effects with As-mediated carcinogenesis.

INTRODUCTION

Chronic arsenic (As) exposure is associated with the development of cardiovascular disease, diabetes and various cancers [1]. The most common source of chronic As exposure worldwide is drinking water contaminated with inorganic forms of As (iAs) from natural, geologic sources [2, 3]. Inorganic As exists in drinking water as a mixture of two oxidation states, namely trivalent arsenite (iAs^III) and pentavalent arsenate (iAs^V), which predominate in anaerobic and aerobic conditions, respectively [4]. Millions of people worldwide and an estimated tens of millions in Bangladesh alone are exposed to iAs levels in drinking water that far exceed the World Health Organization’s recommended limit of 10 ppb [5-7].

Although iAs is classified as a known human carcinogen with primary targets of the skin, lung, and urinary bladder (UB) [8], the mode of action (MOA) of iAs carcinogenesis remains largely undefined [9-11]. Experimental evidence supports the generation of oxidative stress as a major factor, which can contribute to carcinogenesis by causing macromolecule (e.g. DNA) damage and stimulating signaling pathways that control processes such as proliferation, differentiation and apoptosis [11-14]. Arsenic can exert other potentially carcinogenic effects on cells that are not necessarily associated with oxidative stress. For instance, while they are not point mutagens, arsenicals can induce other genetic alterations such as chromosomal aberrations [15, 16], can bind and inhibit enzymes [17], and can influence gene expression by altering the epigenome [18].

Determining which of these effects is a key event in As carcinogenesis is complicated by iAs^III/V metabolism. Arsenite and arsenate are biotransformed in humans and rodents to yield several trivalent and pentavalent methylated arsenical metabolites. During this process, individual methyl groups are added to iAs in a stepwise, unidirectional manner [19]. Chronic exposure to iAs^III/V results in six major arsenicals in human urine in which 10-30% of the total As content exists as trivalent and pentavalent iAs (iAs^III+V), and 10-30% exists as the monomethylated forms monomethylarsonous acid (MMA^III) and monomethylarsonic acid (MMA^V). The remaining 60-80% exists as the trivalent/pentavalent dimethylated species dimethylarsinous acid (DMA^III) and dimethylarsinic acid (DMA^V) [20-22]. One or more of these methylated arsenicals may contribute to the carcinogenic effect of chronic iAs exposure [23]. Not all methylated arsenicals have equivalent biological effects. For example, in general the trivalent arsenicals MMA^III and DMA^III are more cytotoxic in vitro than their pentavalent counterparts and iAs^III/V [24] and have been considered likely carcinogenic arsenicals in vivo [25].

Recently, we examined the gene expression profiles of normal human epidermal keratinocytes (NHEKs) after short-term exposure (24 h) to iAs^III, MMA^III and DMA^III [26]. Several trends in gene expression were observed after exposure to iAs^III and especially MMA^III that were consistent with several proposed mechanisms of As-mediated skin carcinogenesis. For example, increased mRNA levels of oxidative stress response genes, keratinocyte growth factors, and genes in the extracellular-signal regulated mitogen-activated protein kinase (ERK1/2 MAPK) and nuclear factor kappa beta (NF-κB) pathways were observed. Curiously, we observed a minimal transcriptional response in NHEKs after exposure to multiple doses of DMA^III, and these DMA^III-induced transcriptional profiles were distinct from those induced by iAs^III or MMA^III. These results suggested that, unlike iAs^III and MMA^III, DMA^III is not a potent transcriptional inducer of genes that elicit carcinogenic responses in the skin, at least after short-term exposure.

Here we set out to examine the transcriptional responses of human urothelial cells (UROtsas), another major target cell type of As carcinogenesis, to determine if short-term exposure to iAs^III, MMA^III and DMA^III is associated with modulation of genes implicated in UB cancer development. We found that both MMA^III and DMA^III perturbed molecular networks involving signaling pathways and genes implicated in UB cancer development, namely the ERK1/2 MAPK and NF-κB pathways. Together, these results suggest that even after short-term exposure to trivalent arsenicals, transcriptional alterations have occurred that may provide insight into the MOA of As-induced carcinogenesis in the UB.

MATERIALS AND METHODS

Arsenicals

The arsenicals used in this study have been previously described [26]. Briefly, working solutions of sodium m-arsenite (NaAsO₂; herein referred to as iAs^III), monomethylarsonous acid in the form of (CH₃)AsI₂ (MMA^III), and dimethylarsinous acid in the form of (CH₃)₂AsI (DMA^III) were prepared in sterile phosphate buffered saline, pH 7.4 (PBS) immediately before use.

UROtsa Growth Conditions

UROtsas, a human urothelial cell line transformed by the large T antigen of SV40 [27], were obtained from Dr. Zuzana Drobna (Department of Nutrition, UNC Gillings School of Global Public Health, University of North Carolina at Chapel Hill). UROtsas do not exhibit anchorage-independent cell growth and do not develop tumors in immunocompromised mice if maintained at low passage numbers [27, 28]. Routine UROtsa growth and subculturing conditions were as described by Rossi et al. [29]. UROtsas were grown in UROtsa growth medium which contained 1:1 DMEM/Ham’s F12 media (GIBCO/Invitrogen, Carlsbad, CA) supplemented with 5% fetal bovine serum (FBS; Sigma-Aldrich, St. Louis, MO) [29]. Prior to arsenical treatment, the growth medium was removed, cells were washed briefly with PBS and the growth medium was replaced with 1:1 DMEM/Ham’s F12 without FBS (UROtsa dosing medium). Cells were grown in a 37°C, 5% CO₂ humid atmosphere in the absence of antibiotic and antifungal agents and matching cell line passages were used for all endpoint analyses.

Arsenical Treatment for Cytotoxicity Determination and Microarray Analysis

The UROtsas utilized in this study have been previously shown to be a suitable cell line to study the effects of individual arsenicals as they readily absorb them but have little [30] or no [24, 31, 32] capacity to further methylate them. For cytotoxicity assays, UROtsas were seeded into 96 well plates at a density of 10,000 cells/cm² and grown in 250 μl/well of their respective growth medium as described above until they reached ~70% confluence (48 h). For each well, the culture medium was then removed, the cells were briefly rinsed with PBS, and 125 μl of UROtsa dosing medium was added. An additional 125 μl of UROtsa dosing medium containing PBS (vehicle-treated controls) or 125 μl PBS containing appropriate concentrations of iAs^III, MMA^III or DMA^III was added to each well to yield final arsenical concentrations of 0, 0.01, 0.03, 0.3, 1, 3, 10, 30 and 100 μM. Following incubation for 24 h at 37°C, 5% CO₂, all media were removed and replaced with fresh UROtsa dosing medium containing 0.003% neutral red dye (Sigma #N6264). After 3 h incubation at 37°C, 5% CO₂, dye taken up by viable cells was extracted with a 50% ethanol/1% acetic acid solution and quantified spectrophotometrically at 540 nm. For each arsenical dose, data were averaged from 18 wells (from three independent experiments containing 6 wells each) and normalized to vehicle-treated controls.

For RNA isolation, UROtsas were seeded at a density of 3500 cells/cm² in 25 cm² flasks and grown in UROtsa growth medium as described above until they were ~70% confluent (48 h). The cells were rinsed briefly with PBS followed by exposure to 1.0 μM iAs^III, MMA^III, DMA^III or PBS (vehicle-treated controls) in UROtsa dosing medium for 24 h at 37°C, 5% CO₂. For each flask, the dosing medium was removed and the cells were lysed with 3 ml TRIzol Reagent (Invitrogen cat# 15596-026) according to the manufacturer’s instructions. Cell lysates were stored at −70°C until RNA was extracted.

RNA Extraction and RNA Processing

RNA was prepared from each lysed cell sample according to the TRIzol RNA extraction protocol (Invitrogen) followed by the RNeasy Mini Kit (Qiagen, Valencia, CA). RNA concentrations were determined spectrophotometrically (A₂₆₀). RNA quality was determined spectrophotometrically (A₂₆₀/A₂₈₀) and by using the RNA 6000 Nano Assay with the 2100 Bioanalyzer (Agilent, Palo Alto, CA). For microarray analysis, 15 μg of total RNA from each sample was converted to biotinylated cRNA using the Affymetrix 3′ IVT Express Kit (cat #901228) as described by the manufacturer (Affymetrix, Santa Clara, CA). Total cRNA was then quantified using a NanoDrop ND-1000 Spectrophotometer (NanoDrop Technologies, Wilmington, DE) and evaluated for quality after fragmentation using a 2100 Bioanalyzer. The cRNA was hybridized to Affymetrix GeneChip® Human U133 Plus 2.0 arrays overnight at 45°C in an Affymetrix GeneChip Hybridization Oven 640. Each chip was hybridized with cRNA obtained from one biological replicate. The hybridized arrays were washed and stained using a Fluidics Station 450 and scanned using a GeneChip Scanner 3000 7G according to the manufacturer’s instructions (Affymetrix). After scanning, raw data (.cel files) were obtained using Affymetrix GeneChip® Command Console software (v. 3.0).

Microarray Analyses

Three biological replicates were used for each of the arsenical treatment groups and vehicle-treated controls. Robust Multiarray Average (RMA) was used for normalization and probe-level summarization [33] using Partek Genomics Suite ™ (version 6.5) software (Partek, Inc. St. Louis, MO). Differentially expressed genes (DEGs) after arsenical treatment compared to controls were identified using the following filters in Partek: 1-way ANOVA (p<0.05) followed by Benjamini-Hochberg FDR (q<0.05) correction and >1.5 or <−1.5-fold change in expression. Hierarchical clustering was performed on genes that were differentially expressed in at least one arsenical treatment group relative to controls.

IPA software (version 8.8, Ingenuity® Systems, www.ingenuity.com) was used to analyze for enriched biological functions and molecular interactions associated with DEGs. The Ingenuity Knowledge Base within IPA is a literature-based database of known molecular interactions and functional annotations based on known relationships between cells, cellular components, drugs and diseases. For network generation, DEGs from each group were mapped to their corresponding object within the Ingenuity Knowledge Base. These objects were then overlaid on the global network of known molecular interactions, and smaller networks of interacting molecules within the global network were algorithmically generated based on their connectivity. Statistically significant associations of DEGs with biological functions and diseases (p<0.005) and canonical pathways (p<0.05) were also determined using IPA. All p values were calculated in IPA using a right-tailed Fisher’s exact test, which calculates the probability that associated functions, pathways and network interactions were generated due to chance alone.

Quantitative Reverse Transcription-PCR (qRT-PCR)

The same total RNA preparations were used for microarray and qRT-PCR analyses. Quantitative RT-PCR was performed using the LightCycler® 480 Real-Time PCR System (Roche Applied Science, Indianapolis, IN) using Qiagen’s QuantiTect SYBR Green RT-PCR kit (cat. #204243) and QuantiTect® Primer Assays according to the One Step RT PCR Protocol. Each 50 μl reaction contained 10 ng total RNA from a single biological sample, and three biological replicates were used for each treatment group and vehicle-treated controls. Beta actin (ACTB) was chosen for a reference target for each treatment group due to its stable expression in microarrays across all treatment groups. Expression levels of each target gene were normalized against ACTB expression within each treatment group and fold change values of the normalized target gene expression in arsenical treatment groups vs. vehicle-treated controls were calculated using the 2^−ΔΔCT method [34]. Statistically significant changes in target expression levels between in arsenical treatment groups compared to vehicle-treated controls was determined using a Student’s t-test. The genes and QuantiTect primer assays that detect them were: ACTB as a reference target (Hs_ACTB_2_SG); dehydrogenase/reductase (SDR family) member 2 (DHRS2; Hs_DHRS2_1_SG) for the iAs^III group; matrix metalloproteinase 1 (MMP1; Hs_MMP1_1_SG) and sodium channel, nonvoltagegated 1 alpha (SCNN1A; Hs_SCNN1A_1_SG) for MMA^III group and MMP1 (Hs_MMP1_1_SG) and prostate transmembrane protein, androgen induced 1 (PMEPA1; Hs_PMEPA1_1_SG) for the DMA^III group.

RESULTS

Transcriptional Response after Trivalent Arsenical Exposure

Arsenical concentrations that do not induce considerable cytoxicity/apoptosis have been shown to be relevant for studying perturbations in carcinogenic signaling pathways [14, 26, 35]. We examined the cytotoxic effects of varying doses of iAs^III, MMA^III and DMA^III in UROtsas after 24 h exposure (Figure 1, Supplemental Material). The IC50s for each of the arsenicals were 43 μM, 4.1 μM and 3.7 μM for iAs^III, MMA^III and DMA^III, respectively. For the gene expression analyses, arsenical concentrations that resulted in >75% UROtsa viability were selected. The 1 μM dose met this criterion for each arsenical. This dose is particularly relevant for iAs^III as long-term exposure of UROtsas at this dose results in malignant transformation [36].

Affymetrix microarrays that contained >54,000 probe sets representing >38,500 well-characterized human genes were used for the gene expression analyses. The number of unique UROtsa genes that were differentially expressed in each treatment group relative to controls was: 2 (iAs^III), 763 (MMA^III) and 176 (DMA^III) with some overlap among the genes altered by the arsenicals (Figure 1A; Table 1, Supplementary Material). Of these genes, 830 unique genes were differentially expressed in at least one treatment group (Figure 1B). A total of 108 genes were common between the MMA^III and DMA^III groups, representing 16% and 61% of the total DEGs in these treatment groups, respectively. All of the shared DEGs between the MMA^III and DMA^III groups had a concordant direction of modulation between the two groups.

Figure 1.

(A). Venn diagrams displaying shared and unique UROtsa DEGs among the iAs^III, MMA^III and DMA^III treatment groups. The total number of DEGs in each treatment group is in parentheses. (B). Hierarchical clustering of the 830 unique genes in UROtsas that were differentially expressed vs. controls after 24 h exposure to at least one trivalent arsenical. Z scores of fold change values are shown.

Transcriptional trends observed in microarray experiments were verified by qRT-PCR for five selected targets for the iAs^III (DHRS2), MMA^III (MMP1, SCNN1A) and DMA^III (MMP1, PMEPA1) exposure groups (Figure 2). The targets chosen collectively represent a large range of expression changes in treatment groups relative to controls. Each of the five targets selected for qRT-PCR showed the same direction of modulation compared to controls as observed in microarray experiments. All but PMEPA1 in the DMA^III dose group had a statistically significant change in expression (p<0.01) in the treatment group vs. controls as determined by qRT-PCR (Figure 2).

Figure 2.

Comparison of gene expression changes determined by microarrays and quantitative reverse transcription-PCR (qRT-PCR) of selected genes. Statistically significant changes in gene expression compared to vehicle-treated controls are indicated as determined by microarrays (*) and qRT-PCR (**).

Overview of Statistically Significant Biological Functions and Canonical Pathways Associated with DEGs

DEGs from each of the arsenical treatment groups were analyzed for associated biological functions, diseases, and canonical pathways. The low number of DEGs in the iAs^III treatment group did not result in statistically significant results for these analyses. The five most significant biological functions/diseases associated with the DEGs for the MMA^III exposure group were: cancer (1.12 × 10⁻¹¹), lipid metabolism (2.13 × 10⁻⁹), small molecule biochemistry (2.13 × 10⁻⁹), vitamin and mineral metabolism (2.13 × 10⁻⁰⁹), and cell death (1.76 × 10^μ7). For the DMA^III group, the five most significant biological functions/diseases were: cancer (3.37 × 10⁻⁹), reproductive system disease (3.37 × 10⁻⁹), cell-to-cell signaling and interaction (2.2 × 10⁻⁸), tissue development (2.2 × 10⁻⁸), and cellular development (1.72 × 10⁻⁷).

The majority of the altered canonical pathways in the MMA^III and DMA^III groups are associated with one or more of the following functions: lipid metabolism, stress response/oxidative stress response, and inflammatory response (Table 2, Supplemental Material). Genes involved in stress response/oxidative stress response were generally upregulated in both the MMA^III and DMA^III groups. Oxidative stress response genes in the MMA^III group included NAD(P)H dehydrogenase, quinine 1 (NQO1), aldehyde oxididase 1 (AOX1), ferritin, heavy polypeptide 1 (FTH), and ferritin, light polypeptide (FTL), all known to play a role in the NRF2-mediated oxidative stress response pathway (Table 2A, Supplemental Material). Upregulated oxidative stress response genes in the MMA^III group also included glutamate cysteine ligase, catalytic subunit (GCLC), glutathione-S-transferase mu 3 (GSTM3), and glutathione reductase (GSR) in the glutathione metabolism pathway (Table 2A, Supplemental Material). The generalized stress response genes in the DMA^III group included tumor necrosis factor (TNF; aryl hydrocarbon receptor signaling pathway), heat shock 27 kDa protein (HSP27 aka HSPB1; P38 MAPK pathway), junction mediating and regulatory protein, tumor protein 53 (TP53) cofactor (JMY; TP53 pathway) (Table 2B, Supplemental Material).

Genes that play a role in immune response were particularly enriched in the DMA^III group. These pathways contained many of the same upregulated pro-inflammatory genes such as interleukin 1, alpha (IL1A), interleukin 6 (IL6), interleukin 8 (IL8) and TNF (Table 2B, Supplemental Material).

Molecular Networks Associated with DEGs

To identify genes that encode proteins that interact in the cell and therefore may represent a coordinated response to arsenical exposure, DEGs from each group were analyzed for molecular networks. The 763 DEGs in the MMA^III group encode proteins that interact in 25 sub-networks (Table 3, Supplemental Material). Eighteen out of the 25 MMA^III sub-networks had overlapping proteins which generated a large, interacting protein network (interactome; p<10⁻¹⁸) of 473 total proteins, of which 293 are encoded by MMA^III DEGs (Figure 3A). A total of 14 sub-networks were associated with 176 DMA^III DEGs (Table 3, Supplemental Material). Four out of the 14 DMA^III sub-networks were integrated and part of a larger interactome (p<10⁻¹⁷) of 137 total proteins including 59 encoded by DMA^III-altered DEGs (Figure 3B).

Figure 3.

A-B. Large interacting molecular networks associated with the MMA^III and DMA^III treatment groups, respectively; (C). MMA^III–induced ERK 1/2 MAPK network; (D). DMA^III –induced ERK 1/2 MAPK network; (E). MMA^III–induced NF-κB network; (F). DMA^III–induced NF-κB network. Upregulated and downregulated genes are displayed in red and green, respectively and interacting genes that are not differentially expressed are not shaded.

We performed a detailed analysis of the most significant molecular networks (p≤10⁻²⁴) generated in the MMA^III and DMA^III treatment groups. Networks in both groups contained signaling proteins that play a role in pathways implicated in cancer development that have been previously shown to be modulated by arsenical exposure, namely those involving ERK 1/2 MAPK and the transcription factor NF-κB (Figure 3C-D and 3E-F, respectively). There were no common modulated genes among any of these ERK1/2 MAPK and NF-κ networks.

The ERK 1/2 MAPK-containing networks in the MMA^III and DMA^III groups were comprised of proteins that have associated functions related to cellular development, i.e. cellular development, cellular growth and proliferation and metabolism (MMA^III) and cellular development and tissue development (DMA^III) (Figure 3C-D; Table 3, Supplemental Material). The NF-κB-containing molecular networks had primary functions associated with drug/lipid metabolism, small molecule biochemistry (MMA^III) and cellular movement, hematological system function and development, and immune cell trafficking (DMA^III) (Figure 3).

Identification of UB Cancer-Associated DEGs

We examined the DEGs in the MMA^III and DMA^III groups for genes with known associations with human cancers, particularly of the UB. In particular, we set out to identify genes that are transcriptionally altered here and have previously been shown to have altered mRNA expression or protein levels in UB cancers compared to normal UB. Several genes that are commonly overexpressed in UB cancers had increased mRNA levels in UROtsas after acute MMA^III and/or DMA^III exposure (Table 4, Supplemental Material). These genes include growth and differentiation factor 15 (GDF15), heparin binding epidermal growth factor (HBEGF), matrix metalloproteinase 1 (MMP1), and IL8 in both the MMA^III and DMA^III groups; mitogen activated protein kinase 1 (MAPK1), heat shock 70 kDa protein 1 (HSPA1A), heat shock 70 kDa protein 1B (HSPA1B), macrophage migration inhibitory factor (MIF), and glucose-phosphate isomerase (GPI) in the MMA^III group; and IL1A, IL6, and tumor necrosis factor (TNF) in the DMA^III group [37-46]. Functions associated with these genes include the promotion of angiogenesis (GPI), growth (HBEGF), and tissue invasiveness (MMP1) [41, 43, 47]. In addition, many of these genes have pro-inflammatory roles, namely MIF, PTGS1, IL1A, IL6, IL8 and TNF [48-51], whereas GDF15 is immunosuppressive [52]. Several of the UB cancer-associated genes enriched amongst the ERK1/2 MAPK or NF-κB-containing molecular networks of the MMA^III (GPI, MAPK1, MIF, PTGS1) and DMA^III (GDF15, HBEGF, IL8, TNF) groups (Figure 3C-F). In addition, several of the pro-inflammatory genes with known associations with UB cancers were among those that had elevated protein levels or were implicated as having important roles in UROtsa malignant transformation after prolonged MMA^III exposure, namely MAPK1, IL8, MIF, PTGS1 in the MMA^III group and IL1A, IL6, PTGS1, TNF in the DMA^III group [53-55].

DISCUSSION

In this work, we set out to investigate the transcriptional responses of UROtsas after acute exposure to iAs^III, MMA^III and DMA^III. Importantly, epidemiological evidence indicates iAs is a human carcinogen with the UB as a major target organ [56]. The MOA of iAs-carcinogenesis in the UB is not well understood, but MMA^III and DMA^III are considered likely key carcinogenic forms. Specifically, MMA^III and DMA^III are often more cytotoxic [24], more potent enzyme inhibitors [57], cause more DNA strand breaks [58], and are more clastogenic [59] than iAs^III or pentavalent arsenicals in vitro. As further evidence of the potential for trivalent arsenicals to induce cancer, adoption of a cancer phenotype in vitro is observed much quicker in rat and human cells after chronic iAs^III exposure if they are capable of biotranforming iAs^III to methylated metabolites [60]. In addition, a recent publication has indicated that MMA^III can act as a multi-organ transplacental carcinogen in mice [61].

In the UB, there is evidence that iAs^III, MMA^III and DMA^III may all play an important role in cancer development. For example, in rats, DMA^III is the likely arsenical responsible for inducing carcinogenic effects in the UB upon exposure to DMA^V [62-64]. In addition, UROtsas undergo malignant transformation after long-term exposure to either iAs^III or MMA^III [36, 65]. Together, these results suggest that iAs^III, MMA^III and DMA^III have the potential to elicit carcinogenic effects in urothelial cells.

We previously determined that short-term exposure of NHEKs to trivalent arsenicals in vitro caused arsenical-specific effects in terms of the transcriptional modulation of genes implicated in the development of non-melanoma skin cancers (NMSCs) [26]. Here, we expanded this research effort to determine if genes associated with UB cancer development were modulated in UROtsas after short-term trivalent arsenical exposure. We report two key findings: (1) while the magnitude of the transcriptional responses were different between the arsenical groups, short-term exposure to both MMA^III and DMA^III resulted in modulation of ERK 1/2 MAPK- and NF-κB-associated signaling pathways; (2) along with other genes, these modulated signaling pathways in the MMA^III/DMA^III groups are implicated in UB cancer development and also are implicated as important drivers of MMA^III-mediated UROtsa transformation.

Our data highlight differences between the transcriptional profiles for each of the arsenical treatment groups. While DMA^III was the most cytotoxic arsenical (DMA^III≥MMA^III>iAs^III), the greatest transcriptional response was induced by MMA^III exposure (MMA^III>DMA^III>iAs^III) with a 4-fold greater number of genes modulated by MMA^III than DMA^III. Some similarities existed between the MMA^III and DMA^III groups. All of the shared genes between these groups had a concordant direction of modulation compared to controls. Moreover, both groups were enriched for molecular networks containing signaling pathways associated with cancer development, namely the ERK1/2 MAPK and NF-κB pathways. Carcinogenic effects associated with ERK 1/2 MAPK activation include stimulating proliferation, promoting cell migration and invasion, increasing genomic instability, and preventing apoptosis [66, 67]. NF-κB can promote cancer development by stimulating proliferation, angiogenesis, and inflammation [13, 68]. NF-κB and ERK 1/2 can be activated by reactive oxygen species (ROS) and are believed to play important roles in oxidative stress-related cancers associated with metal/metalloid (e.g. As) exposure [13]. Activation of these pathways has long been associated with As exposure in a variety of cell types and is believed to play major roles in As-mediated carcinogenesis through the modulation of genes involved in proliferation, development and inflammation [14, 35, 69]. Importantly, these biological functions (cellular proliferation/development and inflammation) are among the most significant functions associated with the ERK MAPK 1/2 and/or NF-κB-containing molecular networks in the MMA^III and DMA^III groups. These results indicate that short-term exposure to 1 μM MMA^III and DMA^III modulate signaling pathways implicated in cancer development, including As-mediated cancers.

These data also indicate that short-term exposure to MMA^III and DMA^III perturbed signaling pathways and genes implicated in UB cancer development. Chronic inflammatory conditions and signaling pathways that converge on MAPKs, phosphatidylinositol 3-kinase (PI3K) and NF-κB are believed to play important roles in the development of many UB cancers [70-73]. For example, in UB cancer cell lines, ERK1/2 MAPK activation is associated with sustained cellular proliferation [74] and sustained ERK1/2 MAPK is associated with poor prognosis in UB cancer patients [39]. Additionally, specific inhibitors of NF-κB [75] and inhibitors that target both NF-κB and ERK1/2 MAPK [76] have been associated with the induction of apoptosis and reduced invasion/migration in human UB cell lines, respectively. Pro-inflammatory signaling is associated with the development of many UB cancers, even though UB cancer is a heterologous, multi-stage disease that likely develops via multiple pathways [77]. Most of the UB cancer-associated genes modulated in the MMA^III and/or DMA^III groups have pro-inflammatory roles and are upregulated in response to arsenical exposure relative to controls, namely IL8, PTGS1 observed for both MMA^III and DMA^III; IL1A, IL6, TNF observed in response to DMA^III and MIF observed in response to MMA^III. These results indicate that short-term MMA^III and DMA^III exposure in UROtsas modulates signaling pathways and inflammatory genes implicated as important mediators of some UB cancers.

We also observed similarities between the transcriptional profiles induced by short-term MMA^III and DMA^III exposure and processes implicated in MMA^III-mediated UROtsa transformation. Activated signaling pathways associated with MMA^III-mediated malignant transformation have been characterized extensively, in which chronic inflammation secondary to sustained activation of ERK 1/2 MAPK has been implicated as a potential important mechanism of carcinogenesis. In particular, the upregulation of proinflammatory cytokines/chemokines (e.g. IL1, IL6, IL8 and TNF) and signaling through pathways that involve epidermal growth factor receptor (EGFR), RAS, MAPK, PI3K, NF-κB and overexpression of certain targets of these pathways, e.g. endoperoxide synthase 1 and 2 (PTGS1/2) are believed to be important for the establishment and/or maintenance of the malignant phenotype of MMA^III-transformed UROtsas [53-55, 78, 79]. Both iAs^III- and MMA^III-transformed UROtsas have characteristics that are consistent with UB cancers that form under chronic inflammatory conditions. In particular, the development of SCCs (as opposed to transitional cell carcinomas, TCCs), is favored under sustained inflammatory conditions in the UB [80], and both iAs^III- and MMA^III-transformed UROtas have squamous characteristics and form SCCs in immunocompromised mice [36, 65]. As previously mentioned, transcript levels of several genes implicated as important mediators of MMA^III-induced UROtsa transformation were increased in this study, namely MAPK1, IL8, MIF, PTGS1 in the MMA^III group and IL1A, IL6, IL8, PTGS1, TNF in the DMA^III group, and the expression/overexpression of these genes or the proteins they encode have been previously associated with UB cancers [38, 39, 42, 44-46, 81]. Taken together, these results not only indicate that pro-inflammatory signaling that occurs concurrently with MMA^III-mediated UROtsa transformation in vitro also occurs in UROtsas after short-term exposure to MMA^III or DMA^III, but that these modulated signaling pathways and pro-inflammatory genes are consistent with those believed to play important roles in the development of UB cancers.

Of note, we observed that both MMA^III and DMA^III were capable of inducing pathways and genes that are implicated in UB cancer development and MMA^III-mediated malignant transformation of UROtsas. Few studies have investigated the gene expression effects in human urothelial cells after DMA^III exposure [82] and to our knowledge, the capacity of DMA^III to induce urothelial cell transformation in vitro is unknown. In our previous study, genes implicated in NMSC development, including those that signal through ERK1/2 MAPK and NF-κB, were strongly associated with MMA^III exposure but not associated with DMA^III exposure [26]. MMA^III is often viewed as a particularly likely carcinogenic metabolite of iAs as it is often the most cytotoxic arsenical observed in vitro [24] and has recently been shown to be a multi-organ transplacental carcinogen [61]. Also, As-exposed populations with increased urinary MMA^III+V/DMA^III+V ratios have been associated with increased susceptibility to As-associated cancers, including NMSCs and UB cancers [83-85]. It has been clearly established that DMA^III can induce carcinogenic effects such as the production of ROS and DNA strand breaks and is the likely reactive metabolite contributing to UB cancers in DMA^V-exposed rats [86-88]. Here, we demonstrate that unlike the DMA^III-induced transcriptional response observed in NHEKs, DMA^III exposure is associated with carcinogenic transcriptional alterations in human urothelial cells, suggesting cell-specific responses to DMA^III may exist. Individuals have different capacities to biotransform iAs to MMAs and DMAs, resulting in varying proportions of MMAs and DMAs in urine and tissues [21, 89]. Therefore, taken together, these results suggest cell-specific responses to arsenicals and interindividual differences in iAs metabolism likely play important roles in the susceptibility to As-associated disease.

In summary, this study reveals that acute MMA^III and DMA^III exposures in UROtsas result in transcriptional alterations of pathways implicated in the development of UB cancers and MMA^III-mediated UROtsa transformation in vitro. Future studies should help inform which arsenical(s) exert carcinogenic effects in the UB and inform the relationship of molecular events that drive UB cancers of different etiologies.

ABBREVIATIONS

As

arsenic

DMA^III

dimethylarsinous acid

DMA^V

dimethylarsinic acid

DEGs

differentially expressed genes

ERK1/2 MAPK

extracellular signal-regulated kinase 1/2 mitogen-activated protein kinase

iAs^III

arsenite

iAs^V

arsenate

IC50

half maximal inhibitory concentration

MMA^III

monomethylarsonous acid

MMA^V

monomethylarsonic acid

MOA

mode of action

NF-κB

nuclear factor kappa beta

NRF2

nuclear factor-E2-related factor 2

qRT-PCR

quantitative reverse transcription polymerase chain reaction

UB

urinary bladder