Arsenic-induced Changes in miRNA Expression in Cancer and Other Diseases

Abstract

miRNAs (miRNA) are essential players regulating gene expression affecting cellular processes contributing to disease development. Dysregulated miRNA expression has been observed in numerous diseases including hepatitis, cardiovascular diseases and cancers. In cardiovascular diseases, several miRNAs function as mediators of pathogenic stress-related signaling pathways that may lead to an excessive extracellular matrix production and collagen deposition causing cardiac stress resulting in fibrosis. In cancers, many miRNAs function as oncogenes or tumor suppressors facilitating tumor growth, invasion and angiogenesis. Furthermore, the association between distinct miRNA profile and tumor development, progression and treatment response has identified miRNAs as potential biomarkers for disease diagnosis and prognosis. Growing evidence demonstrates changes in miRNA expression levels in experimental settings or observational studies associated with environmental chemical exposures such as arsenic. Arsenic is one of the most well-known human carcinogens. Long-term exposure through drinking water increases risk of developing skin, lung and urinary bladder cancers, as well as cardiovascular disease. The mechanism(s) by which arsenic causes disease remains elusive. Proposed mechanisms include miRNA dysregulation. Epidemiological studies identified differential miRNA expression between arsenic-exposed and non-exposed individuals from India, Bangladesh, China and Mexico. In vivo and in vitro studies have shown that miRNAs are critically involved in arsenic-induced malignant transformation. Few studies analyzed miRNAs in other diseases associated with arsenic exposure. Importantly, there is no consensus on a consistent miRNA profile for arsenic-induced cancers because most studies analyze only particular miRNAs. Identifying miRNA expression changes common among humans, rodents and cell lines might guide future miRNA investigations.

Introduction

In recent years, numerous studies have indicated that miRNA dysregulation can play a role in chronic disease. Much of the focus has been on cancer, although there have been a few studies investigating miRNA dysregulation in non-cancer diseases. Consequently, there is substantial information available about the role of miRNA dysregulation in cancer, but little about the role in other disease processes.

miRNAs can act as oncogenes, tumor suppressor genes, or both, depending upon the tissue type [1–3]. In miRNA regulation of chemical carcinogenesis, one miRNA can have many target mRNAs, and many miRNAs can target one mRNA to form a complex network for fine regulation of gene expression [4]. Different miRNAs can play several roles in pathological processes. miRNAs are involved in the regulation of cell cycle, differentiation, proliferation, apoptosis, stress tolerance, and immune response [1, 2]. Thus, the carcinogenic potential may be dependent on the balance between tumor-suppressor miRNAs and oncogenic miRNAs [4]. The miRNA profile readily changes in the cancer-target tissues after exposure to carcinogens. Epidemiological and experimental studies have shown that miRNAs are critically involved in arsenic-induced carcinogenesis.

Here, we summarize the recent epidemiological and experimental data on miRNA expression changes induced by arsenic. A total of 200 studies were identified by searching PubMed in May 2020 using the terms “miRNA AND arsenic”. Seventy-four papers were removed as they did not meet the screening criteria 1 (e.g studies not written in English and/or analyzing other metals along with arsenic) (Figure 1). An additional 47 studies were excluded as they were not related to carcinogenesis or other diseases. The remaining 85 manuscripts were further analyzed in detail. The in vivo and in vitro studies analyzing miRNA dysregulation in the whole genome using RNA sequencing or microarrays were included in this review. For human studies, any manuscript analyzing a single or few miRNAs were included. A selection of 22 studies including 11 epidemiological studies, 4 in vivo studies and 7 in vitro studies were ultimately examined in the review (Figure 1).

FIGURE 1.

Diagram of literature search and selection process of the studies. *Others category included studies on bioinformatics analysis, plants, and non-English written. The sum of all 5 categories (epidemiological, in vivo, in vitro, review and others) in the initial set is > 200 because some studies assessed more than one category.

Arsenic

Arsenic contamination in drinking water is a global concern because of its toxicity and persistence in the environment [5]. Due to its potential threat to human health, arsenic has been ranked number one in the Agency for Toxic Substances and Disease Registry Substance Priority List for more than 20 years [6]. Although both developed and developing countries face this problem, it is particularly a challenge for developing nations with limited economic and technical capacities to remove arsenic from contaminated groundwater [7]. Several reports have identified arsenic levels in more than 50 countries, including Bangladesh, Taiwan, Ghana, India, Chile, Mexico, Argentina, China, and the United States [8], affecting more than 220 million people at concentrations that exceed the Maximum Contaminant Level of 10 μg/L recommended by the World Health Organization and the US Environmental Protection Agency [9]. In particular, Bangladesh is the most affected region by far, with an estimated 50 million people being at risk of exposure and approximately 78% of its districts with arsenic levels greater than 50 μg/L [10, 11]. Arsenic is classified as a Class I human carcinogen by the International Agency for Research on Cancer (IARC) [12] and induces diverse chronic health effects such as urinary bladder, lung and skin cancer along with other non-neoplastic toxicities [13, 14]. One of the recently suggested mechanisms by which arsenic causes cancer is miRNA dysregulation [15–17], although the precise mechanism remains unclear.

miRNA and Cancer

miRNAs are a large family of small non-coding RNAs (~21–22 nucleotide-long) considered as key molecular components of both normal and pathologic cells [18]. A recent estimate extrapolated 2300 true human mature miRNAs [19] that can target cytosolic mRNAs by binding to their complementary sequences in the 3’ untranslated region (UTR) leading to decreased translation, deadenylation, or degradation of the mRNA. The specific outcome is dependent upon the degree of base-pairing complementarity between the target mRNA and the “seed region” at the 5’ end of the miRNA (Figure 2) [20, 21].

FIGURE 2.

The fate of the targeting mRNA depends on the complementarity of the seed sequence (nucleotides 2 to 8 on the 5’ end of the miRNA) to the mRNA. A continuous base-pairing promotes mRNA deadenylation or degradation. Imperfect base-pairing inhibits translation [21]. Adapted from [98].

The involvement of miRNAs in numerous diseases, including cancer, has been investigated in a variety of reports. Microarray expression analysis in different types of cancer tissues/cells such as breast, colon, stomach, lung and prostate have shown distinct aberrant miRNA expression [22, 23]. The underlying mechanisms for such dysregulation include defects in the miRNA biogenesis machinery, chromosomal abnormalities, transcriptional control changes and epigenetic changes. As a result, abnormal miRNA expression can confer to the cell sustained proliferative signaling, apoptosis resistance, acquisition of mesenchymal and metastatic characteristics, and angiogenesis induction [22]. As indicated above, each tumor type has a distinct miRNA signature that discriminates it from other neoplastic types and normal tissue [20]; however, different types of cancer present commonly dysregulated miRNAs. For example, miRNA-21 is considered an oncogene that is upregulated and promotes metastasis in several cancers including lung, breast and leukemia [24–26]. The let-7 family of miRNAs maintains differentiation and acts as a tumor suppressor. Thus, downregulation of let-7 family miRNAs is seen in various types of cancers such as hepatocellular carcinoma, gastric adenocarcinoma, pancreatic cancer, ovarian cancer, prostate cancer, and renal cell carcinoma among others [27]. Importantly, tumors export miRNAs into the circulation [28, 29] and these circulating miRNAs can serve as biomarkers for diagnosis and prognosis, and to monitor treatment response. In addition, miRNA dysregulation is implicated in chemical carcinogenesis since abnormal miRNA expression is observed in association with several environmental toxicants [30, 31].

miRNA and Environmental Toxicants

During their life span, humans are exposed to a wide variety of both naturally occurring and anthropogenic chemical toxicants that can contribute to the development of chronic diseases such as cancer [32, 33]. Variation in gene expression plays a prominent role in tumorigenesis. Environmental toxicant exposures may alter gene expression in part through changes in the miRNA profile. In recent years, studies have indicated that certain human carcinogenic chemicals, including metals, organic pollutants, cigarette smoke, pesticides, and drugs are capable of altering miRNA expression [34, 35]. Krauskopf et al. documented the first evidence of alterations in the miRNA and transcriptomic profiles upon environmental exposure to carcinogenic persistent organic pollutants (POPs) in a population-based study [36]. Air pollution, associated with increased risk of lung and breast cancer, has been shown to induce changes in miRNAs [37]. Plasma levels of miR-145-5p were decreased in an epidemiological study [38]. This specific miRNA inhibits the proliferation of non-small cell lung cancer cells by targeting the oncogene c-Myc [39]. Heavy metal exposure, another matter of global concern, is considered a major threat to human health leading to several cancerous and non-cancerous outcomes. In vitro data showed cadmium inducing or suppressing a variety of miRNAs with different cellular functions, resulting in cell transformation [30, 40, 41]. Additional miRNA expression profile alterations can be epidemiologically and experimentally observed in arsenic-induced carcinogenesis [30, 31, 42]. Despite extensive research, the mechanisms and the precise role of miRNA dysregulation in the etiology of arsenic-induced cancers remain fairly uncharacterized.

Arsenic Induced Changes in miRNA Expression – In Vitro Evidence

Several in vitro studies examined the relationship between arsenic exposure and dysregulated miRNA expression, but only a select few utilized microarray-based analyses to assess whole-genome miRNA expression (Table 1). Studies using HaCaT cells (immortalized human keratinocytes) showed differential expression levels in miRNAs after arsenic exposure [43, 44]. Following a 28-week treatment of 100 nM sodium arsenite, the oncogenic transformation was confirmed by anchorage-independent growth [43]. Upon microarray analysis of the arsenic-treated cells, 12 miRNAs were up-regulated and 14 were down-regulated. To validate the microarray results, RT-qPCR was performed on 6 selected miRNAs, observing upregulation in miR-6739-5p, mir-4521, miR-181b-5p, miR-100-5p and miR-3919, and downregulation in miR-513a-5p. In addition to investigating the microRNAome of arsenic-transformed cells, Zhou et al. [43] also incorporated proteomics and metabolomics to show the mechanistic changes in HaCaT cells after arsenic exposure. The utilization of databases to build molecular networks impacted in arsenic-treated HaCaT cells displayed 14 miRNAs associated with oxidative stress and redox metabolism in arsenic-induced skin carcinogenesis [43]. These investigations provide further evidence of epigenetic and metabolic alterations in arsenic-induced malignant transformation. Gonzalez et al. [44] malignantly transformed HaCaT cells after continuous exposure to 500 nM sodium arsenite for 4 weeks. The expression of miRNA was evaluated in the arsenic-treated cells using the Locked Nucleic Acid (LNA) miRNA array. In total, 30 miRNAs were found to be differentially expressed when compared to the control group with 21 miRNAs up-regulated and 9 down-regulated [44]. For RT-qPCR analysis, five miRNAs were selected, four of which are linked to carcinogenesis (miR-21, miR-34b, miR-200a and miR-141) and one (miRPLUS-A1087) that had the lowest fold change when compared to control [44]. Increased expression levels were seen in miR-21, miR-200a and miR-141 with decreased expression levels in miRPLUS-A1087, confirming their LNA miRNA array results. However, mir-34b did not have consistent results between the microarray (downregulated) and RT-qPCR (upregulated). Modulations in these miRNAs appear to have an impact on signaling pathways that are prominent in skin carcinogenesis. In silico analysis revealed miR-21, miR-200a, and miR-141 potentially altering expression of target genes in melanoma and mitogen-activated protein kinase (MAPK) signaling pathways [44]. Interesting findings on early miRNA expression changes related to the transformation process of HaCaT cells were reported by Al-Eryani et al. [45]. After 3 and 7 weeks of 100 nM arsenite exposure, 293 and 373 small RNAs (including snoRNAs, stem-loop and mature miRNAs) were differentially expressed, respectively. Eleven of them were differentially expressed at both time points: miR-339, miR-1228, miR-4309, miR-4692, miR-548au, miR-548a-3p, miR-645, miR-1254, miR-2682-5p, miR-3618 and miR-8083. Further analysis revealed 38 differentially expressed mRNAs in arsenic-exposed cells at both time points that were predicted to be targeted by these 11 miRNAs. Mouse double minute 2 homolog (MDM2), high mobility group box 1protein (HMGB1) and TP53 mRNA and protein expression levels were validated by RT-qPCR and Western blot, respectively, confirming that early changes in miRNAs and target mRNAs may contribute to arsenic-induced carcinogenesis.

Table 1:

Dysregulated miRNAs induced by arsenic in in vitro models.

Cell Line	Arsenic Concentration & Time of Exposure	Results	miRNA Findings	Reference
HaCaT	100 nM, 28 weeks	Presence of giant multinuclear cells and exhibition of anchorage-independent growth through colony formation after arsenic exposure	Up: miR-7–5p, miR-4521, miR-6739–5p, miR-181b-5p, miR-100–5p, miR-20a-5p, miR-146a-5p, miR-3919, miR-125b-5p, miR-378i, miR-494–3p, miR-140–3p Down: miR-1973, miR-23a-3p, miR-4787–5p, miR-3178, miR-513a-5p, miR-7704, miR-3196, miR-4497, miR-31–5p, miR-3960, miR-1273g-3p, miR-203a, miR-4284, miR-4508 Validated-Up: mir-6739–5p, mir-4521, mir-181b-5p, mir-100–5p and hmir-3919 Down: hsa-mir-513a-5p	[43]
HaCaT	500 nM, 4 weeks	Alterations in miRNAs linked to carcinogenesis: miR-22, miR-21, miR-34b, miR-141, miR-200a, miR-27b and miR-23b	Up: miR-22, miR-21, miR-34a, miR-205, miR-141, miR-1260, miR-720, miR-1280, miR-200a, miR-19b, miR-30e, miR-27b, miR-23b, miR-27a, miR-1274a, miR-181a, miR-1469, miR-19a, miR-184, miR-101, miR-29b Down: miRPlus-E1012, miRPlus-G1246–3p, miR-1285, miRPlus-E1247, miRPlus-F1147, miRPlus-F1231, miR-34b, miR-F1017, miRPlus-A1087 Validated-Up: miR-21, miR-141 and miR-200a Down: miR-PLUS-A1087	[44]
HaCaT	100 nM, 3 and 7 weeks	Total TP53 and TP53-S15-phosphorylation induction. However, TP53-K382-hypoacetylation suggested that the induced TP53 is inactive in arsenic exposed cells	Up (3 weeks): miR-339, miR-4309, miR-645, miR-2682–5p Down (3 weeks): miR-1228, miR-4692, miR-548au, miR-548a-3p, miR-1254, miR-3618, miR-8083 Up (7 weeks): miR-339, miR-1228, miR-4692, miR-645, miR-2682–5p Down (7 weeks): miR-4309, miR-548au, miR-548a-3p, miR-1254, miR-3618, miR-8083	[45]
RWPE-1 WPE-stem cells	5 μM, 29 weeks (CAsE-PE cells) 18 weeks (As-CSC cells)	Multiple in vitro signs of malignant transformation and production of tumor xenografts in nude mice	Up (CAsE-PE): miR-9, miR-96, miR-183 Down (CAsE-PE): miR-134, miR-127–5p, miR-373, miR-34c-5p, miR-146b-5p, miR-135b, miR-222, miR-155, miR-138, miR-205, miR-218, miR-10b, miR-181d, miR-125a-5p, let-7b, miR-181b, miR-98, let-7i, miR-34a, miR-196a, miR-181a, let-7e, miR-181c, let-7c, miR-125b, miR-126 As-CSC Up: miR-34a, let-29b, miR-193b, miR-7 Down: miR-9, miR-34c-5p, miR-135b, miR-138, miR-205, miR-218, miR-143, miR-355, miR-148a Validated (CAsPE)-Down: miR-134, miR-373, miR-155, miR-138, miR-205, miR-181d, miR-181c, let-7b, let-7i, let-7e, and let-7c Validated (As-CSC)-Down: miR-143, miR-34c-5p, and miR-205	[46]
Jurkat	2 μM, 24 hours 144 hours	Formation of large cell aggregates and increased percentage of G2/M population of cells following 24 h arsenic exposure with return to pretreatment conditions at 144h	Up (24h): miR-150 miR-181a, miR-142–5p, miR-222, miR-663, miR-638, miR-30d, miR-130b, miR-378, miR-181c, miR-142–5p, miR-625, miR-744, miR-629, miR-140–3p, miR-575, miR-7–1, let-7d, mir-221, miR-186, miR-425, miR-107, miR-361–5p, miR-532–5p, miR-149, miR-222, miR-663, miR-638 Down(24h): miR-29c, miR-150, miR-200c, miR-18a, miR-342–5p, miR-223, let-7e, let-7a, let-7c, miR-21, miR-574–5p, miR-181b, Up (144h): miR-575, miR-574–5p, miR-7–1, let-7d, mir-221, miR-186, miR-361–5p, miR-181b, miR-532–5p, miR-149, miR-222, miR-663, miR-638 Down (144h): miR-29c, miR-130b, miR-378, miR-150, miR-200c, miR-18a, miR-342–5p, miR-181c, miR-142–5p, miR-625, miR-744, miR-629, miR-140–3p, miR-181a, miR-223, miR-34c-3p, let-7e, let-7a, let-7c, miR-21, miR-425, miR-107 Validated (both)-Up:miR-663, miR-221, miR-222, miR-638 Down:miR-150, Up (24):miR-30d, miR-142–5p, miR-181a Down(144):miR-142–5p, miR-181a,	[48]
HepG-2	4 μM, 24 hours	Apoptotic and growth inhibitory effects of ATO on HepG-2 cells were significantly enhanced by miR-29a	Up: miR-24, miR-29a, miR-30a, miR-210 and miR-866–3p Down: miR-744, miR-296–5p, miR-663, and miR-675 Validated-Up: miR-24, miR-29a, miR-30a and miR-210	[49]
HUVECs	20 μM, 24 hours	Alterations in miRNAs implicated in hypertrophic heart (hsa-miR-19b and hsa-miR-29b) and regulation of fatty acid metabolism and insulin signaling (hsa-miR-301a, and hsa-miR-33a) Alterations in miRNAs targeted key cellular functions: phosphoproteins/ genes in alternative splicing, transcription regulation, RNA metabolic process and transcription factor activity	Up: miR-21, let-7i, miR-130a, miR-103, miR-107, miR-132, miR-16, miR-182, miR-193a-3p, miR-194, miR-196b, miR-19b, miR-200a, miR-215, miR-221, miR-23b, miR-26a, miR-29b, miR-29c, miR-335, miR-365, miR-493, miR-151–5p, miR-138, miR-301a, miR-96, miR-429, miR-10a, miR-542–3p, miR-487b, miR-361–5p, miR-15b, miR-24, miR-30b, miR-425, miR-532–5p, let-7a, miRPlus-A1087, miR-20b, miR-100, miR-148b, miR-17, miR-30d, miR-15a, miR-93, miR-125b, miR-101, miR-92a, miR-1184, miRPlus-E1060, miR-10b, miR-145, miR-128, miR-362–5p, miR-937, miR-140–3p, miR-20a, miR-25, miR-33a, miR-940, miR-886–3p, miR-874, miR-28–5p, miR-181a, miR-339–3p, miR-29a, miR-30c, miRPlus-E1136, miRPlus-E1141, miRPlus-F1147, let-7f, let-7d, let-7g, miR-1287, miRPlus-E1070, miRPlus-E1016, miR-27a, miRPlus-E1088, miRPlus-E1196, miRPlus-E1100, miR-375, miR-31, miR-191	[50]
			Down: miR-299–3p, miR-325, miR-200c, miR-622, miR-585, miR-934, miR-183, miRPlus-C1115, miRPlus-D1058, miR-761, miR-892b, miRPlus-D1036, miR-122, miR-135a, miR-640, miRPlus-A1072, miR-549, miR-508–5p, miR-638, miRPlus-F1155, miR-1249, miRPlus-F1035, miRPlus-F1243, miRPlus-E1045, miRPlus-F1170, miRPlus-F1149, miRPlus-E1133, miRPlus-F1218, miRPlus-E1074, miRPlus-F1225, miR-1275, miRPlus-E1015, miRPlus-E1101, miRPlus-F1215, miRPlus-E1211, miRPlus-E1245, miRPlus-E1205, miR-1299, miRPlus-F1127, miRPlus-F1141, miRPlus-F1239, miRPlus-F1066, miRPlus-E1071, miRPlus-E1209, miR-205, miR-1252, miR-548n, miR-198, miR-617, miRPlus-E1063, miR-890, miR-542–5p Validated-Up: miR-19b, miR-29b, miR-33a, miR-874, miR-21 Down: miR-508–5p, miR-1252, miR-198

One study examined 84 cancer-linked miRNAs in prostate epithelial cells (RWPE-1) and isogenic stem cells (WPE-stem cells) exposed to 5 μM sodium arsenite [46]. Mouse xenografts indicated RWPE-1 cells were malignantly transformed after 29 weeks (CAsE-PE cells) and WPE-stem cells after 18 weeks (As-CSC cells) of arsenic exposure. Expression levels of 29 and 13 miRNAs were altered in CAsE-PE cells and As-CSC cells, respectively. In comparison, the two malignant phenotypes had more downregulated miRNAs in common (miR-34c-5p, miR-135b, miR-138, miR-205 and miR-218) compared to upregulated miRNAs [46]. Predicted targets of differentially expressed miRNAs were also analyzed to show the inverse correlation of target genes and miRNA expression. Ngalame et al. [46] further investigated the miRNAs involved in KRAS expression, which is generally overexpressed in malignant prostate cells [47]. Downregulation of miR-134, miR-373, miR-155, miR-138, miR-205, 181d, 181c, let-7b, let-7i, let-7e and let-7c in CAsE-PE cells correlated with marked increased expression of target RAS oncogenes, validating that miRNA expression inversely correlated with RAS gene expression [46]. It is important to note that both cell types (CAsE-PE and AS-CSC) were exposed to the same arsenic concentrations for different times; therefore, a distinct duration of exposure could play a role in why certain miRNAs did not change in a specific subpopulation versus the other.

Sturchio et al. [48] exposed human Jurkat leukemic T cells to 2 μM sodium arsenite for 0, 24 and 144 hours. Compared to the untreated cells, the Jurkat cells treated with arsenic showed increased cell proliferation and large cell aggregation, but no significant apoptotic effect. Of the 93 miRNAs analyzed, 36 were differentially expressed at different time points. Eight miRNAs (miR-663, miR-150, miR-221, miR-222, miR-142-5p, miR-181a, miR-30d and miR-638) were selected for validation. For both time points, up-regulated levels of miR-663, miR-222 and miR-638 were found, when compared to untreated Jurkat cells. There was a clear distinction between the time points and miR-150, miR-181a, and miR-142-5p, which had high expression levels at 24 h and low at 144 h. This study also examined potential gene targets of these miRNAs and showed upregulated miR-221, miR-222 and miR-638 related to decreased expression of the target, Ring Finger Protein 4 (RNF4) [48]. By examining the predicted and confirmed targets of these differentially expressed miRNAs, the regulatory role of miRNAs during arsenic exposure was validated.

Although much of the focus has been on arsenic-induced carcinogenesis, there are also studies examining the effects of arsenic trioxide (ATO) as a potential chemotherapeutic. Meng et al. [49] examined 677 human miRNAs in HepG-2 cells (human hepatocellular carcinoma cell line) treated with 4 μM ATO for 24 h. Nine miRNAs were found to have altered expression, with five being upregulated (miR-24, miR-29a, miR-30a and miR-210) and 4 downregulated (miR-744, miR-296-5p, miR-663, and miR-675). To validate the miRNA results, RT-qPCR analysis was performed for miR-24, miR-29a, miR-30a and miR-210 due to their role in cancer. Further investigation of miR-29a showed a synergistic effect with ATO in inducing apoptosis and inhibiting HepG-2 cell growth, compared to the other miRNAs. This effect was demonstrated by transfecting the cells with miR-29a and examining the effects of ATO in a dose-dependent manner (1–5 μM). With the use of CI-isobologram analysis, arsenic trioxide concentrations 3 μM or less showed synergistic effects between ATO and miR-29a. In the cell viability assay, miR-29a had an inhibition rate of 22.6% compared to its counterparts which were under 10% [49]. The expression levels of key miRNAs can play a role in the susceptibility to ATO-induced cell death.

Arsenic can target multiple organs, which leads to different diseases. The role of arsenic in vascular injury was examined by exposing HUVECs (human umbilical vein endothelial cells) to 20 μM arsenite for 24 h [50]. Microarray analysis showed that 85 human miRNAs were up-regulated and 52 were down-regulated after arsenic exposure. miRNAs associated with hypertrophic heart (miR-19b and miR-29b) were upregulated by ten-fold. Nine miRNAs (miR-19b, miR-29b, miR-33a, miR-874, miR-2, miR-24, miR-508-5p, miR-1252, and miR-198) were selected for RT-qPCR analysis, with only miR-24 not being validated [50]. The remaining miRNAs displayed similar expression changes seen in the microarray results. Interestingly, this study looked at how cellular functions such as transcription, splicing and posttranslational modification are all altered after arsenic treatment in HUVECs. This study suggests that arsenic exposure can cause alterations in miRNA expression profiles in HUVECs, which could explain the link between arsenic and peripheral vascular injury [51, 52]. Nevertheless, a limitation seen in this study is the use of very high levels of arsenic exposure (20 μM) compared to toxicologically relevant human internal arsenic exposure (100 nM) [53–55]. The use of concentrations that mimic human internal exposure is critical to the reliability and relevance of the data. Thus, such limitation could misidentify important players or regulators of arsenic toxicity.

Arsenic and Changes in miRNA Expression – In vivo Evidence

The main barrier to improving our understanding of the mode of action for arsenic carcinogenesis has been the lack of a reliable animal model for any of its targets. Many studies have used in vivo systems, only a few have evaluated miRNA dysregulation (Table 2). Daily oral administration of sodium arsenite at 5 mg/kg body weight was given to Wistar rats for 14 or 28 days [56]. RNA was purified from liver tissue and miRNA expression was analyzed. After 14 days, the level of miR-let7a expression was downregulated compared with the control group (0.9% NaCl daily oral administration). Inversely, this miRNA level was upregulated by 28 days. miR-146a levels were downregulated at 14 and 28 days. Some other adverse effects observed in this study include liver toxicity and increased oxidative stress. Ren et al. [57] analyzed miRNA levels extracted from the liver of Sprague-Dawley rats exposed via drinking water containing 0, 0.1, 1, 10, or 100 mg/L sodium arsenite for 60 days. Multiple miRNAs were differentially expressed in liver tissue in response to arsenic exposure by Hi-Seq high-throughput sequencing. The top four upregulated miRNAs (miR-151, miR-183, miR-148b and miR-192) and the top two downregulated miRNAs (miR-26a, miR-423) were subjected to RT-qPCR validation. The Hi-Seq and RT-qPCR profiles were highly correlated for five of the six miRNAs (miR-151, miR-183, miR-26a, miR-423 and miR-148b). These results were observed along with the alteration of hepatic glutamate-cysteine ligase activity and glutathione levels which can impact cellular antioxidant capacity [57, 58]. A very recent study assessed whether in utero arsenic exposure induced miRNA alterations that could lead to greater postnatal susceptibility to cancer [59]. Pregnant C3H mice received sodium arsenite at 42.5 and 85 ppm in drinking water (exposures known to cause liver cancer) from gestation day 8–19, and the livers from male fetal mice were collected for analysis. MicroRNA array showed significant differential expression of 50 and 140 miRNAs induced by exposures of 42.5 or 85 ppm respectively, out of the 718 analyzed miRNAs. The increased expression of miR-205, miR-203, miR-215, miR-34a, and decreased expression of miR-217 were confirmed by RT-qPCR. These results affirm the likely contribution of aberrant miRNA expression to adult adverse outcomes including liver cancer [60, 61].

Table 2:

Arsenic-induced miRNA changes in in vivo models

Animal Strain, Tissue	Arsenic Concentration & Time of Exposure	Results	miRNA Findings	Reference
Wistar, liver	5 mg/kg bw, 14 and 28 days	Liver toxicity and increased oxidative stress	Up: miR-let7a, after 14 days Down: miR-let7a, after 28 days, miR-146a	[56]
Sprague-Dawley, liver	0.1, 1, 10, or 100 mg/L sodium arsenite, 60 days	Alteration of hepatic glutamate–cysteine ligase activity and glutathione levels	Up: miR-151, miR-183, miR-148b and miR-192 Down: miR-26a, miR-423	[57]
C3H mice, liver	Transplacental exposure to 42.5 and 85 ppm, gestational days 8–19	Differential DNA methylation	50 miRNAs altered at 42.5 ppm and 140 at 85 ppm Validated-Up: miR-205, miR-203, miR-215, miR-34a Down: miR-217	[59]
ApoE−/− mice, liver	Transplacental exposure to 85 mg/L, gestational day 8 to birth	Increased expression of inflammatory cytokine and heat shock protein 70. Elevated SREBP1 expression, and liver enzymes in plasma	PND1 – Up: miR-361, miR-148a, miR-467a Down: miR-let-7d, miR-719, miR-679, miR-592, miR-222, miR-497, miR-211, miR-188 PND70 – Up: miR-211, miR-291a-5p-291b-5p, miR-294, miR-302, miR-464 Down: miR-1, miR-10a, miR-15b, miR-214a, miR-130a, miR-149, miR-186, miR-193, miR-218, miR-337, miR-376a, miR-412, miR-476b, miR-681, miR-715	[63]

In contrast to the absence of a reliable animal model to assess arsenic-induced carcinogenesis, the use of Apolipoprotein E-knockout (ApoE^−/−) mice exposed to arsenic via placenta from gestational day 8 to birth (dams were exposed to 85 mg/L arsenite via drinking water) is an analogous model for in utero arsenic exposure induction of accelerated atherosclerosis [62]. By using this model, States et al. [63] evaluated the miRNA transcriptome profile in liver from pups euthanized on the day of birth (PND1) and at 10-weeks of age (PND70). miR-361, miR-148a and miR-467a were induced and miR-let-7d, miR-719, miR-679, miR-592, miR-222, miR-497, miR-211 and miR-188 were suppressed in livers of arsenic exposed PND1 mice. In livers of arsenic exposed PND70 mice, miR-211, miR-291a-5p-291b-5p, miR-294, miR-302 and miR-464 were induced and miR-1, miR-10a, miR-15b, miR-214a, miR-130a, miR-149, miR-186, miR-193, miR-218, miR-337, miR-376a, miR-412, miR-476b, miR-681 and miR-715 were suppressed. Differentially expressed miRNAs were then scanned for their target mRNAs using miRBase. These results were further crossed with the mRNAs found differentially expressed in livers from PND1 and PND70 mice. Gene ontology analysis of the differentially expressed mRNAs targeted by differentially expressed miRNAs enriched terms such as fundamental metabolic processes (glycolysis/gluconeogenesis), protein metabolism (protein export, ribosome) and inflammatory processes (complement, antigen processing, coagulation) [63]. These findings suggest that transplacental arsenic exposure leads to a proinflammatory response postnatally that may contribute to early onset of atherosclerosis [63].

Arsenic and Changes in miRNA Expression – Epidemiological Evidence

A limited number of human population-based studies examined the impact of arsenic on miRNA expression and its association with carcinogenesis or other diseases (Table 3). Serum expression levels of miR-126, miR-155, and miR-145 were evaluated in Mexican women exposed to inorganic arsenic via drinking water in San Luis Potosí State [64]. Expression levels of these miRNAs have been associated with cardiovascular alterations such as coronary heart disease, atherosclerosis and aortopathy [64–66]. The mean urinary arsenic level quantified in urine samples (n=105) was 19.5 ± 14.0 μg/g creatinine, which is lower than the tolerable limit proposed by the Centers for Disease Control and Prevention (CDC) of 50 μg/g of creatinine [67]. A significant increase in miR-155 levels was seen in women presenting urinary arsenic levels >15 μg/g creatinine. In contrast, the levels of miR-126 were significantly decreased. These two miRNA levels were also assessed in children (aged 6–12 years; n = 73) exposed to arsenic from the same area [68]. The mean urinary arsenic level found in these children was 30.5 ± 25.5 μg/g of creatinine. However, 40% of the children presented urinary arsenic levels >40 μg/g of creatinine. A negative association between urinary arsenic concentrations and plasma miR-126 levels was found, the same association observed by Ruíz-Vera’s study [64] discussed earlier. However, the levels of miR-155 remained unchanged. The limitations of both studies include small sample size and lack of analysis of other environmental chemical compounds that could have a significant effect on the miRNA levels.

Table 3:

Impact of arsenic on miRNA expression in human population-based studies.

Country	Sample	Results or adverse effect	miRNA Findings	Reference
Mexico	Plasma	Association between arsenic and miR-155 and miR-126, proposed as predictive cardiovascular (CDV) disease biomarkers	Up: miR-155 Down: miR-126	[64]
Mexico	Plasma	Negative association between arsenic and plasma miR-126 levels (early biomarker of CVD risk)	Down: miR-126	[68]
China	Plasma	Dysregulated miRNAs might impact immune inflammation, oxidative stress and DNA damage repair	Up: 56 miRNAs Down: 18 miRNAs Validated-Up: miR-21, miR-145, miR-155, and miR-191	[69]
China	Plasma	Skin, liver and kidney damage	Up: miR-21, miR-145, miR-155, and miR-191	[70]
India	Plasma	Precancerous and cancerous skin lesions	Up: 199 miRNAs Down: 3 miRNAs Validated-Up: miR-21, miR-23a, miR-619, miR-126, miR-3613 Down: miR-1282 and miR-4530	[71]
India	Plasma	Malignant squamous cell carcinoma and basal cell carcinoma	Up: mir-21	[72]
Bangladesh	Urine	Inverse relationship of miR-200c and miR-205, EMT related miRNAs, with arsenic	Down: miR-200c and miR-205	[73]
India	Skin Lesions	Premalignant hyperkeratosis HK and malignant basal cell carcinoma and 3 squamous cell carcinoma	Up: miR-425–5p and miR-433 in BCC and SCC compared to HK; miR-184 and miR-576–3p in SCC relative to BCC and HK Down: miR-29c, miR-381, miR-452, miR-487b, miR-494 and miR-590–5p in BCC relative to SCC and HK	[75]
Mexico	Newborn cord blood	Dysregulated miRNAs have roles in cancer and inflammatory response	Up: let-7a, miR-107, miR-126, miR-16, miR-17, miR-195, miR-20a, miR-20b, miR-26b, miR-454, miR-96, and miR-98	[76]
Mexico	Plasma	Dysregulated miRNAs are linked to cardiovascular disease and diabetes	MMA correlated with miR-423–5p, miR-142–5p −2, miR-423–5p +1, - 320c-1, −320c-2, and −454–5p	[16]
China	Urine	Changes in genes related to apoptosis, cell cycle inhibition, or carcinogenesis	iAs, MMA and DMA negatively correlated with miR-548c-3p	[77]

Circulating miRNAs involved in arsenic carcinogenesis were screened through miRNA arrays in a population from the Chinese village Jiaole, an endemic area for arsenic poisoning caused by coal-burning. In the first part of the study, Sun et al. [69] identified miRNAs differentially expressed between the exposed group (n=10) and a non-exposed (n=10) group that came from an arsenic-free area. Of 754 miRNAs analyzed, 56 were upregulated and 18 were downregulated in the exposed group compared with the control group. Among these, miR-21, miR-141, miR-148a, miR-145, miR-155, miR-191, miR-218 and miR-491 levels showed the largest fold change increase, while miR-200b, miR-200c, miR-26, and miR-34c levels had the most decrease in expression. The levels of miR-21, miR-145, miR-155, miR-191 and miR-200b were validated by RT-qPCR. In the second part of the study, RT-qPCR assay was used to evaluate the levels of these five miRNAs in another study population, composed of 50 exposed individuals and 50 control individuals. The results confirmed increased circulating levels of miR-21, miR-145, miR-155 and miR-191 in arsenic exposed individuals, but no changes in miR-200b. These five miRNAs were further analyzed in a bigger population-based study in the same village in China that recruited a total of 457 participants (arsenic exposure group n=371, reference group n=86) [70]. The mean levels of urinary arsenic were 27.85 μg/L in the exposed group and 21.36 μg/L in the reference group respectively. Again, miR-21, miR-145, miR-155 and miR-191 serum levels were increased in the exposed group compared with the reference group, and there was no difference among the levels of miR-200b between the two groups. They further divided all 457 participants into sub-groups of skin, liver and kidney damage and analyzed the levels of the five miRNAs. The skin damage group was divided into normal, suspicious, mild, moderate, and severe skin damage. Liver damage included fatty liver, diffuse liver echo changes, hepatic hemangioma, hepatic cysts, cirrhosis, and liver cancer. Kidney damage comprised renal cyst, kidney abscess, nephritis, hydronephrosis, and kidney cancer. The levels of miR-21, miR-145, miR-155, and miR-191 in the skin damage group were higher than the skin non-damage group. Likewise, miR-21 and miR-145 were increased in the liver damage group relative to the liver non-damage group. Only miR-191 was upregulated in the kidney damage group compared to the kidney non-damage group. Thus, these dysregulated miRNAs could be associated respectively with skin, liver, and kidney damage caused by arsenic.

Banerjee et al. [71] also analyzed global plasma miRNA profiling of 12 arsenic exposed individuals with skin lesions (precancerous and cancerous) versus 12 exposed individuals without skin lesions, but in West Bengal, India. The arsenic level in water was much higher than 10 μg/L in both groups, and the urinary arsenic levels were similar between groups. Microarray analysis showed 202 miRNAs differentially expressed in the plasma of the skin lesion group compared with the no skin lesion group, 199 miRNAs were upregulated and 3 were downregulated. Based on skin cancer relevance, some miRNAs were selected for RT-qPCR validation: miR-21, miR-23a, miR-619, miR-126, and miR-3613 were upregulated and miR-1282 and miR-4530 were downregulated in the skin lesion group compared with the no skin lesion group; miR-124 showed no changes. Thus, except for miR-124, the microarray data were validated by real-time PCR. Despite its low sample size, this is the first study investigating global changes in miRNA expression induced by arsenic in a West Bengali population, one of the worst arsenic-endemic areas in the world. The level of miR-21 was also investigated in another study performed in West Bengal [72]. In this study, 83 exposed individuals (45 with arsenic-induced malignant squamous cell carcinoma and basal cell carcinoma and 38 exposed individuals without skin lesions) and 40 unexposed individuals were recruited. Circulating miR-21 expression level was upregulated in the exposed individuals compared to the unexposed. Within the exposed group, miR-21 expression level was much higher in the skin lesion group compared to the no skin lesion group. Interestingly, there was no difference in the miR-21 expression patterns between the unexposed group and the exposed group without skin lesions, while the expression in the skin lesion group was around 6 fold higher compared to the arsenic unexposed group. These findings support the role of miR-21 in contributing to arsenic-induced skin cancer.

An in vitro study, where some of the epithelial mesenchymal transition (EMT) related miRNAs (miR-200a, miR-200b and miR-200c) were found dysregulated after arsenic treatment, served as a basis for a population-based study [73]. A total of 110 urine samples from subjects exposed to different levels of arsenic in Bangladesh and 67 urine samples collected from an arsenic-safe area in Baltimore were analyzed. An inverse relationship of miR-200c and miR-205 with arsenic exposure was observed. The miR-200 family members are important in combating tumor cell invasion, EMT and metastasis, and have been described as a new class of biomarkers for tumor prognosis [74]. It is important to mention that the control group included in this cohort study is not from the same source population or geographic region, which could contribute to bias or confounding in the study.

The only available study in the literature that assessed the miRNA expression profiles in arsenic-induced skin lesions was performed by Al-Eryani et al. [75] in individuals from West Bengal, India. Keratinocytes were harvested from premalignant (3 hyperkeratosis (HK)) and malignant (3 basal cell carcinoma (BCC) and 3 squamous cell carcinoma (SCC)) lesions, RNA was purified from keratinocytes isolated from the lesions by laser capture microdissection and analyzed in RT-qPCR arrays. Thirty-five miRNAs were differentially expressed among the three lesion types analyzed. Interestingly, miR-425-5p and miR-433 were upregulated in both BCC and SCC compared to HK suggestive of a cancer-associated role. miR-184 and miR-576-3p were induced in SCC relative to both BCC and HK indicating their potential involvement with metastasis. miR-29c, miR-381, miR-452, miR-487b, miR-494 and miR-590-5p were decreased in BCC relative to both SCC and HK. These findings indicate that the expression of some miRNAs was both phenotype- and stage-related and highlights the involvement of miRNAs in arsenic-induced transformation.

Prenatal exposure to arsenic was examined in a pregnancy cohort study [76]. Pregnant women in Mexico were recruited for this study. The levels of arsenic in drinking water ranged from below detectable to 236 μg/L (mean 51.7 μg/L) and in maternal urine from 6.2 to 319.7 μg/L (mean 64.5 μg/L). RNA was isolated from 40 newborn cord blood samples and miRNA microarray analysis was performed. Genome-wide miRNA analysis showed maternal urinary arsenic is associated with increased expression in cord blood of let-7a, miR-107, miR-126, miR-16, miR-17, miR-195, miR-20a, miR-20b, miR-26b, miR-454, miR-96, and miR-98, many of which have known roles in cancer and inflammatory response [76].

Two studies aimed to identify circulating miRNA profiles as potential biomarkers of arsenic exposure. Beck et al. [16] analyzed 82 plasma samples for miRNAs and iAs and its methylated metabolites in individuals chronically exposed to inorganic arsenic (iAs) via drinking water in Mexico. Plasma monomethylarsonous acid (MMA) significantly correlated with miR-423-5p, −142-5p −2, −423-5p +1, −320c-1, −320c-2, and −454-5p. No associations were found for plasma iAs or dimethylarsinic acid (DMA). On the other hand, the concentrations of iAs, MMA and DMA in urine were negatively correlated with miR-548c-3p expression levels in plasma of Chinese workers in plants producing arsenic trioxide [77].

Discussion

Arsenic exposure mainly via drinking water is a major public health concern, especially due to its association with increased risk of developing cancerous and non-cancerous chronic diseases. A number of studies have addressed arsenic carcinogenesis; a few have addressed atherogenesis; however, the underlying mechanisms are not yet established. To date, proposed mechanisms involve epigenetic variations such as alterations in DNA methylation, histone modifications and miRNA expression. miRNAs are estimated to modulate expression of up to 60% of all protein-coding genes and a single miRNA may regulate 400 different target mRNAs [78]. As a consequence, miRNAs play a crucial role in the regulation of biological processes and are key regulators in pathogenesis, particularly in carcinogenesis [20, 22, 79].

Although the roles of miRNAs in carcinogenesis have been studied, there remain significant areas needing investigation to understand the full range of influence. Specifically, when examining miRNA expression profiles in different cell lines and tissue types, similarities and differences are found in expression levels of the miRNAs examined. Taking into account the diversity of cancer genomes and phenotypes, analysis of genome-wide miRNA expression signatures aids in identifying key miRNAs and networks that significantly impact the progression of the disease. In the Sun et al. [69] study, around 754 miRNAs were analyzed, and the findings showed 56 upregulated and 18 downregulated in the arsenic exposed group compared to control. In considering the West Bengal study that conducted miRNA profiles on 12 arsenic exposed individuals, there were 199 miRNAs upregulated and 3 downregulated compared to the controls [71] (Table 3). These two epidemiological studies indicate that the use of the whole genome profiling approach allows for the identification of large numbers of miRNAs that potentially have an impact on disease pathogenesis. Even with the in vitro and especially the in vivo studies, still not enough experiments have been conducted that examine the effects of arsenic in the whole genome. Shifting the focus from only a few miRNAs to a more thorough miRNA analysis is necessary, as miRNA expression profiles might enlighten mechanisms of action and improve our understanding of the complex biology of cancers and other diseases.

The toxicity of arsenic is dependent upon how it is metabolized [80]. As arsenic exposure increases or accumulates over time, it leads to alterations in cellular growth, apoptotic behavior and genomic signaling pathways, which drive carcinogenesis [81]. However, at higher concentrations of arsenic, increased apoptosis due to excessive instability in gene expression is observed [81]. It has been reported that arsenic concentrations of 0.1–10 μM result in genomic and proteomic changes that promote oxidative stress and proliferative signaling, which are key mechanisms in carcinogenesis [82]. Although apoptosis still occurs in this concentration range, it becomes more dominant at higher arsenic concentrations, which explains arsenic use as a therapeutic agent in treating acute promyelocytic leukemia [82, 83]. When taking into consideration arsenic exposure times, it is imperative that the effects that occur in humans are emulated. Experimental designs, especially those working with in vitro models, should effectively reflect chronic arsenic exposure because disease progression can occur in a span of years/decades [73]. Within the various studies listed in this review, there are major differences in arsenic concentration and exposure (Table 2). For example, Zhou et al. [43] and Gonzalez et al. [44] utilized HaCaT keratinocytes and selected their concentrations and exposure times based on previous studies that showed malignant transformation in keratinocytes after long-term low arsenic exposure (20–28 weeks, 100 nM to 500 nM [84–86]). However, no similar dysregulated miRNAs were observed when both studies were compared likely due to differences in the experimental design. This presents a key disadvantage in determining which miRNAs can be potential biomarkers. With other studies using high arsenic concentrations such as 20 μM in Li et al. [50], it is questionable whether such a high concentration with a low exposure time of 24 h is best to represent physiological changes. It is important to note that a toxicologically relevant arsenic exposure is considered 100 nM, as it is consistent with the blood arsenic levels in populations chronically exposed to arsenic in China [53] and in Mexico [54, 55]. In human studies, Zeng et al. [70] identified effects of arsenic on multiple organs with different miRNA changes associated with kidney, skin and liver damage. Arsenic exposure induces a variety of human diseases, affecting different tissues and cells. Therefore, future experiments must consider that the adverse effects experimentally induced by arsenic are dependent on dose and duration of exposure, as well as animal strain or cell lines.

The much needed evaluation of altered global miRNA expression upon arsenic exposure will identify new candidates for molecular studies on enriched pathways or molecules targeted by these miRNAs. Studies analyzing lesions or tumor samples from arsenic exposed individuals enable the identification of distinct alterations that might differ from non-arsenic related malignancies [75, 87]. Importantly, this characterization provides critical insights on diagnostic and therapeutic targets in arsenic induced-carcinogenesis. On the other hand, studies investigating arsenic-related alterations in miRNA levels only in blood or urine for example, lack confidence in extrapolating and linking these changes with cancers or other diseases. Another key factor in investigating malignant transformation is the use of an appropriate cell line. Malignantly transformed cell lines such as Jurkat and HepG-2 might be well characterized and easy to culture. However, studies using these types of cells should be carefully analyzed since they are already malignantly transformed and do not fully represent what is occurring in humans during carcinogenesis. Thus, by using an already transformed cell line, important carcinogenic events that would occur might be intrinsic to the cell line and not seen as an effect of arsenic exposure. Factoring in these above measures as well as including optimal arsenic concentrations/exposure times will prove advantageous for future studies investigating miRNA involvement in arsenic-induced disease.

Although the whole genome profiling approach identifies hundreds of miRNA expression changes, changes in certain miRNAs were consistent between the studies reviewed here. Only one miRNA was found dysregulated in all three types of studies (in vitro, animal models, and human population-based): miR-205, a highly conserved miRNA, with homologs discovered in multiple species. Depending on the tissue, miR-205 acts either as a tumor suppressor inhibiting proliferation and invasion or as an oncogene facilitating tumor initiation and proliferation [88, 89]. In arsenic carcinogenesis, miR-205 also appears to exert both oncogenic and tumor suppressor functions. In HaCaT cells exposed to 500 nM arsenite for 4 weeks [44] and in livers of transplacental arsenic exposed C3H mice [59], miR-205 was upregulated. On the other hand, miR-205 was downregulated in CAsE-PE [46], AS-CSC [46] and HUVEC [50] cells, and in the urine from a population in Bangladesh [73]. These data suggest that miR-205 impacts arsenic-induced carcinogenesis, although it is not clear how miR-205 contributes to malignancy. MiR-205 is predicted to target components of the cell cycle, the mTOR signaling pathway, and the ERBB signaling pathway, among others [90]. Thus, changes in its levels likely impair important pathways that can contribute to cancer development. Future studies are needed to investigate the role of miR-205 in arsenic carcinogenesis. miR-21 is another miRNA with expression frequently found altered upon arsenic exposure. The changes in miR-21 expression were more consistent between the analyzed studies, with the exception of the downregulation seen in Jurkat cells [48]. miR-21 levels were increased in HaCaT [44] and HUVEC cells [50], and in the plasma of people exposed to arsenic in China [69, 70] and India [71, 72]. Interestingly, no changes in this miRNA were seen in animal models. miR-21 is evolutionarily conserved across a wide range of vertebrate species and it has been investigated extensively. [91, 92]. In cancers, this miRNA is frequently upregulated, supporting its oncogenic activity. However, the role of miR-21 in arsenic-induced carcinogenesis remains controversial [91, 93, 94]. A recent systematic review and meta-analysis evaluated the effects of arsenic on miR-21 [93]. After analyzing 17 studies, Liu et al. [93] suggested that arsenic exposure increases miRNA-21 levels, reduces phosphatase and tensin homolog (PTEN), and protein sprouty homolog 1 (Spry1) levels. Subsequently, the protein levels of the epithelial markers E-cadherin and N-cadherin decrease and the protein levels of the mesenchymal marker vimentin increases, ultimately leading to malignant transformation [93]. Granting this miRNA has been linked with arsenic carcinogenesis, specifically arsenic-induced skin carcinogenesis, its role in cutaneous melanoma has also been described. In melanomas, there is an increase in miR-21 expression compared to benign nevi. In borderline lesions, miR-21 was overexpressed and associated with thickness and mitotic activity [95]. Considering that melanomas are linked with sunlight exposure rather than arsenic exposure [96], these observations may raise questions regarding the role of miR-21 specifically in arsenic-induced skin carcinogenesis. It should be highlighted that some of the miRNAs were differentially expressed only in animal models and in cell lines (miR-183, - 146a, −151, −148b, −26a, −203, 215 and 34a), while others changed only in human samples and in cell lines (let-7a, miR-16, −17, −20a, −20b, −96, −98, −107, −155, −191, −23a, −126, −200c, −425, −184, −29c and 487b). miR-184, for example, was induced in SCC compared to BCC and HK in arsenic-induced skin lesions [75], and it was upregulated in HaCaT cells exposed to 500 nM arsenic for 4 weeks [44], and there is evidence of its overexpression in psoriatic keratinocytes [97], but no evidence of involvement in sunlight-induced skin cancers. Thus, this miRNA could be induced specifically in skin cancers caused by arsenic exposure. The miRNA alterations induced by arsenic observed in three different types of studies examined in this review highlight that the parallel between in vitro, in vivo and humans provides meaningful insights on the role of miRNAs in arsenic carcinogenesis.

In summary, this review discusses the impact of arsenic on miRNA dysregulation in humans, animal models and cell lines. These alterations might affect important targets in a range of pathways, contributing to disease development. Considerable progress has been made on the understanding of the effects of arsenic in miRNA expression and cancer development. Unfortunately, there is not a consensus on a consistent miRNA profile for arsenic-induced cancers, especially because the majority of the studies analyze particular miRNAs rather than the whole genome. In contrast, the role of miRNAs in arsenic-induced non-cancerous diseases has received little attention, a critical area that should be explored in future research. A distinct miRNA profile might be useful as a biomarker for exposure, diagnosis and therapeutic efficacy. Furthermore, this review highlights miRNA expression changes common among humans, rodents and cell lines, which might serve as guidance on selecting miRNA for future studies.