Arsenic is more potent than cadmium or manganese in disrupting the INS-1 beta cell microRNA landscape

Abstract

Diabetes is a metabolic disorder characterized by fasting hyperglycemia and impaired glucose tolerance. Laboratory and population studies have shown inorganic arsenic (iAs) can impair these pathways. Other metals including cadmium (Cd) and manganese (Mn) have also been linked to diabetes phenotypes. MicroRNAs, short non-coding RNAs that regulate gene expression, have emerged as potential drivers of metabolic dysfunction. MicroRNAs responsive to metal exposures in vitro have also been reported in independent studies to regulate insulin secretion in vivo. We hypothesize that microRNA dysregulation may associate with and possibly contribute to insulin secretion impairment upon exposure to iAs, Cd, or Mn.

We exposed insulin secreting rat insulinoma cells to non-cytotoxic concentrations of iAs (1 μM), Cd (5 μM), and Mn (25 μM) for 24 hours followed by small RNA sequencing to identify dysregulated microRNAs. RNA sequencing was then performed to further investigate changes in gene expression caused by iAs exposure.

While all three metals significantly inhibited glucose-stimulated insulin secretion, high-throughput sequencing revealed distinct microRNA profiles specific to each exposure. One of the most significantly upregulated microRNAs post-iAs treatment is miR-146a (~+2-fold), which is known to be activated by nuclear factor κB (NF-κB) signaling. Accordingly, we found by RNA-seq analysis that genes up-regulated by iAs exposure are enriched in the NF-κB signaling pathway and genes down-regulated by iAs exposure are enriched in miR-146a binding sites and are involved in regulating beta cell function. Notably, iAs exposure caused a significant decrease in the expression of Camk2a, a calcium-dependent protein kinase that regulates insulin secretion, has been implicated in type 2 diabetes, and is a likely target of miR-146a. Further studies are needed to elucidate potential interactions among NF-kB, miR-146a, and Camk2a in the context of iAs exposure.

Introduction

Diabetes mellitus, a complex metabolic disorder, is a growing epidemic currently affecting approximately 450 million individuals globally (2017). The International Diabetes Federation (IDF) predicts this number will more than double by the year 2025 (2017). There are three major types of diabetes: type 1 diabetes, type 2 diabetes, and gestational diabetes; but all three types are characterized by hyperglycemia and/or impaired glucose tolerance. Classical type 2 diabetes (T2D) is often preceded by insulin resistance, but ultimately is due to pancreatic beta cell dysfunction. The causes of beta cell dysfunction are thought to be complex and multifactorial, with genetic susceptibility, obesity, exercise, and environmental exposures playing roles. Though beta cells make up the majority of the pancreatic islets, their capacity to proliferate is much lower than the other cell types (Bouwens and Rooman 2005). Furthermore, beta cells are extremely sensitive to environmental pollutants. In particular, exposures to estrogenic compounds, organophosphorus compounds, persistent organic pollutants and heavy metals have been shown to contribute to pancreatic beta cell dysfunction in both in vivo and in vitro laboratory experiments (Hectors et al. 2011; Makaji et al. 2011). Exposures to metal contaminants have been shown to increase the risk for developing dysglycemia (Fabricio et al. 2016). Thus, more research is needed to understand and combat the effects of environmental exposures that may potentially lead to the development of diabetes.

Inorganic arsenic (iAs), a drinking water and food contaminant, is one of the most widely studied environmental toxicants linked to diabetes. iAs is naturally occurring and can be found in both pentavalent (iAsV) and trivalent (iAsIII) forms, though iAsIII is a more potent disruptor of glucose homeostasis. Several mechanisms have been proposed to explain this disruption including inhibition of insulin signaling (Paul et al. 2008), stimulation of gluconeogenesis (Liu et al. 2014), and inhibition of glucose-stimulated insulin secretion (GSIS) (Douillet et al. 2013; Dover et al. 2017, 2018). Ingestion is the most common route of human exposure to iAs and once in the body it undergoes a series of reactions in which two methyl groups are added in succession forming methylarsenic (MAs) and dimethylarsenic (DMAs) species. Higher percentages of DMAs in urine have been associated with an increased incidence of T2D in human studies (Kuo et al. 2015); however, in vitro laboratory studies showed that both MAs and DMAs, when in the trivalent state (MAsIII, DMAsIII), impair insulin secretion and mitochondrial metabolism in beta cells (Dover et al. 2017), and also impair insulin secretion in isolated murine pancreatic islets (Douillet et al. 2013).

Cadmium (Cd) and Manganese (Mn) have also been studied in the context of T2D, albeit with contradicting results. Cd can accumulate in beta cells and cause dysfunction, which has been shown to increase insulin release in rats (Treviño et al. 2015) while being associated with lower insulin secretion in humans as observed in the Third National Health and Nutrition Examination Survey (NHANES III) cohort (Schwartz et al. 2003). Mn deficiency has been implicated in fasting hyperglycemia and T2D risk in epidemiological studies (Liu et al. 2016; Li et al. 2017). Though each metal may be influencing GSIS in a unique way, iAs, Cd and Mn have all been shown to impair beta cell mitochondrial function, which is an essential component of GSIS (Dover et al. 2018).

MicroRNAs (miRNA) are short, non-coding, single-stranded RNA molecules. They negatively regulate expression of target genes at the post-transcriptional level by binding to 3′ untranslated regions of target messenger RNA (mRNA). miRNAs are sensitive to environmental stimuli (Vrijens et al. 2015) which makes them excellent candidate biomarkers of exposure. Several circulating plasma miRNAs were recently found to be associated with the levels of MAs, a toxic metabolite of iAs, in the plasma of individuals exposed to arsenic through drinking water (Beck et al. 2018).

Many miRNAs play important roles in the control of metabolic processes associated with impaired glucose homeostasis and T2D, such as gluconeogenesis in the liver (Yang et al. 2017), insulin secretion (Poy et al. 2004; Bagge et al. 2012; Jo et al. 2018), insulin sensitivity (Trajkovski et al. 2011), and pancreatic beta cell survival (Belgardt et al. 2015).

In this study, we examined the effects of iAs, MAsIII, Cd, and Mn on miRNA expression in an insulin secreting rat insulinoma cell line. This comparative analysis of miRNA profiles in beta cells in response to different contaminants, each of which have known associations with T2D, has never before been performed. Results of this study provide the first step toward the characterization of the potential roles of miRNAs in metal-induced beta cell dysfunction and potentially differentiate these from the mechanisms involved in classical T2D.

Materials and methods

Cell culture and treatments

Rat insulinoma cells expressing human pro-insulin, INS-1 832/ 13 (Hohmeier et al. 2000b), passage numbers 50–59, were cultured at 5% CO2, 37°C in RPMI 1640 medium (Gibco, Waltham, MA) supplemented with 10% FBS, 10mM HEPES (Gibco, Waltham, MA), 2mM L-glutamine (Gibco, Waltham, MA), 1mM sodium pyruvate (Gibco, Waltham, MA), 100 U/ml penicillin (Gibco, Waltham, MA), 100μg/ml streptomycin (Gibco, Waltham, MA), and 0.05mM β-mercaptoethanol (Sigma, St. Louis, MO).

INS-1 832/13 cells were exposed to iAs (iAs^III; sodium arsenite, > 99% pure; Sigma-Aldrich, St. Louis, MO), MAs^III (methylarsine oxide, > 98% pure), CdCl2 (100% Pure; Sigma Aldrich, St. Louis, MO), or MnCl2 (≥ 99% Pure; Sigma Aldrich, St. Louis, MO) for 24h prior to GSIS, cell viability testing, or RNA isolation.

GSIS

INS-1 832/13 cells were seeded at a density of 1,000,000 cells/well in 12-well tissue culture plate 24 hours prior to exposure. Cells were then treated with iAs, MAsIII, Cd, or Mn for 24 hours. Following this 24-hour exposure, cells were placed in a secretion assay buffer (SAB) comprised of 114 mM NaCl, 4.7 mM KCl, 1.2 mM KH2PO4, 1.16 mM MgSO4, 20 mM Hepes, 2.5 mM CaCl2, 0.2% bovine serum albumin, 25.5 mM NaHCO3, and 0 mM glucose for 40 min. After 40 min in glucose-free medium, cells were incubated in SAB containing 2.5 mM glucose for 60 min and 16.7 mM glucose for 2 h. Metal exposures were maintained during this time by adding 1 μM iAs, 0.5 μM MAs, 5 μM Cd, or 25 μM Mn alongside each SAB incubation.

Aliquots of media were collected for the 2.5 and 16.7mM glucose incubations and cells were lysed for protein analysis. The amount of insulin secreted from cells into the media was determined using a Rat/Mouse Insulin ELISA (Crystal Chem) and normalized for cellular protein.

Cell Viability Assay

INS-1 832/13 cells were seeded at a density of 100,000 cells/ well in a 96 well plate. 24 h after plating, arsenicals were added along with CellTox™ Green Dye using the express, no-step addition at dosing method for the CellTox™ Green Cytotoxicity Assay (Promega, Madison, WI). Fluorescence at 485Ex/520EM was measured at 24 h following addition of metals in a Synergy HT plate reader. Fluorescence indicates a loss of membrane integrity occurring as a function of cytotoxicity, and loss of cell membrane integrity has been linked to cell death.

Pancreatic islets from mice exposed to iAs in vivo

Pancreatic islets from a recently completed study were used in this experiment. All procedures involving mice were approved by the University of North Carolina Institutional Animal Care and Use Committee. Male and female C57BL/6J WT mice were obtained from Jackson Laboratories and let to acclimatize at the UNC Animal Facilities for one week. Mice were housed under controlled conditions with 12-h light/dark cycle at 22 ± 1 °C and 50 ± 10% relative humidity. Mice were randomly assigned treatment groups with 1–3 animals per group. Both male and female mice drank for 20 weeks (ad libitum) either deionized water or deionized water containing sodium arsenite (AsNaO₂, ≥ 99% pure; Sigma-Aldrich, St. Louis, MO, USA) at final concentrations of 0.1 mg As/L (100 parts per billion). Water with sodium arsenite was prepared weekly to minimize oxidation of iAs^III to As^V. Mice were sacrificed by cervical dislocation and pancreatic islets were isolated using previously described procedures (Douillet et al. 2013). Briefly, the pancreas was infused in situ with collagenase P (1 mg/mL, Roche Diagnostics Corp., Indianapolis, IN) via the common bile duct. Pancreas was then removed and digested in the collagenase solution for 12 min at 37 °C. The digestate was washed and islets were purified by centrifugation in a gradient of Ficoll PM 400 (GE Healthcare, Uppsala, Sweden). The islets were stored at −80°C.

Statistical analysis

All data derived from INS-1 832/13 cells are presented as technical replicates from 3 or more biological replicates as specified in the figure legends. Boxplots represent the 25th (bottom), 50th (middle), and 75th (top) percentiles of the data. The protein normalized insulin values were analyzed using a student’s t test. Data are presented as ng insulin/ μg protein, mean ± standard error of the mean (SEM) or as mean ± standard deviation (SD) as indicated; p values less than 0.05 are considered statistically significant.

RNA isolation, Small RNA Sequencing, RNA Sequencing, and qPCR

Following a 24-hour exposure in INS-1 cells or immediately after isolation of islets, total RNA was isolated using the Total RNA Purification kit (Norgen; Ontario, Canada). RNA concentrations were examined with a Nanodrop spectrophotometer, RNA yield and integrity were measured by Agilent 2100 Bioanalyzer (Santa Clara, CA).

We performed small RNA sequencing as described previously (Beck et al. 2018), using Illumina HiSeq, which resulted in an average of 32 million reads per sample.

RNA sequencing libraries were prepared using the Kapa mRNA Library Prep Kit (Kapa Biosystems). Paired end sequencing was performed on the Highseq4000 platform at the UNC High Throughput Sequencing Facility (University of North Carolina, Chapel Hill, NC).

For RT-qPCR, one μg of total RNA was used for reverse transcription with the High Capacity RNA to cDNA kit (Life Technologies; Grand Island, NY) and 200 ng RNA was used for reverse transcription with the TaqMan microRNA Reverse Transcription kit (Life Technologies). MiRNA qPCR was performed using TaqMan Universal PCR Master Mix (Life Technologies) per the manufacturer’s protocol, on a Roche LightCycler 480 II PCR Detection System (Roche; Basel, Switzerland). Reactions were performed in triplicate using U6 as the normalizer.

Bioinformatic Analyses

For small RNA the alignment tool miRquant 2.048 was used for adapter trimming, and alignment of reads to the rn6 version of the rat genome using two different mapping programs (Bowtie and SHRiMP), before quantifying miRNAs and miRNA isoforms (termed isomiRs). On average, 80% of the trimmed reads mapped to the rat genome, of which ~50% corresponded to microRNA loci. (Supplemental Table 1) The length distribution of mapped reads revealed a clear peak at the size range 21–24 nucleotides (Fig. 1), which matches the expectation for miRNAs. MiRNAs with greater than 25 reads per million mapped to miRNAs (RPMMM) in at least one sample were included for further analysis. This threshold yielded a set of 406 detected miRNAs across all samples. Differential expression analysis for miRNAs was performed using DESeq2.

Figure 1.

Read length distribution from small RNA-seq in INS-1 832/13 cells after 24-hour exposure to 1 μM iAs, 5 μM Cd, or 25 μM Mn.

RNA-seq data were mapped to the rn6 genome with STAR, and transcripts were quantified with Salmon. Normalization and differential analysis was performed using DESeq2. Genes were classified as differentially expressed if they had a fold change ≥1.5 in either direction, and false discovery rate (FDR) < 0.1.

Functional Analysis of Differentially Expressed Genes

Functional analyses were performed using Enrichr (http://amp.pharm.mssm.edu/Enrichr). Databases used for protein interactions include the CORUM database, a collection of experimentally verified mammalian protein complexes (Ruepp et al. 2008); the ChIP-x Enrichment Analysis (ChEA) database, a gene-set library with transcription factors labeling sets of putative target genes curated from published experiments (Lachmann et al. 2010); and a protein-protein interaction (PPI) network (Chen et al. 2012). The Kyoto Encyclopedia of Genes and Genomes (KEGG) was used to identify enriched biological pathways (Ogata et al.).

Results

Effects of metal exposure on insulin secretion and cell viability

First, we examined the effects of 24-hour exposure to 1 μM iAs, 5 μM Cd, or 25 μM Mn on insulin secretion by maintaining INS-1 832/13 cells for 1 hour in low glucose containing medium followed by stimulating for 2 hours with high glucose medium. Metal exposure was maintained for the duration of the glucose stimulation. We found that exposures to all three metals significantly decrease insulin secretion in cells stimulated with 16.7 mM glucose (Fig. 2a); iAsIII by 61%, Cd by 68%, and Mn by 59%. To examine if the diminished GSIS was due to loss of cells we normalized the amount of secreted insulin for cellular protein. We also measured cell viability using CellTox™ Green. Here, cells treated with digitonin were used as positive controls. The 24-hour exposure to iAs, or Cd had no effects on cell viability. Cells exposed to 25 μM Mn exhibited a minor, but statistically significant reduction in cell viability (Fig. 2b).

Figure 2.

Glucose stimulated insulin secretion by INS-1 832/13 cells (a) and cell viability (b) after 24-hour exposure to 1 μM iAs, 5 μM Cd, or 25 μM Mn. Data are represented as mean of 6–8 technical replicates + SE for 3 or more biological replicates. **p < 0.01; ***p < 0.001 treatment versus control high glucose stimulation or treatment versus control fluorescence.

Identification of robustly expressed miRNAs significantly dysregulated by metal exposure

To identify metal-responsive miRNAs, we performed high throughput sequencing on untreated cells as well as those exposed to iAs, Cd, and Mn. Reads between 18–24 nucleotides in length comprised 58–62% of mapped reads across samples (Fig. 1). Principle Component Analysis (PCA) based on the 50 most variably expressed miRNAs revealed highly distinct miRNA profiles among the treatment groups (Fig. 3a). PC loading analysis revealed the most variable miRNAs among the different conditions (Table 1). Sequencing data was further analyzed using DESeq2. Setting a fold change threshold of >=1.5 in either direction, and an adjusted p-value <0.05, we found that 12 miRNAs are significantly altered by one or more metal exposures. Of the metals, iAs exposure had the greatest effect on the miRNA landscape, resulting in the dysregulation of 9 miRNAs, as compared to 1 and 2 miRNAs altered by Cd and Mn exposure, respectively (Fig. 3b, Table 2).

Figure 3.

(a) PCA plot of the top 50 most variable miRNAs across all samples and (b) volcano plot of miRNA expression changes after exposure to iAsIII, Cd, or Mn. Vertical lines represent a fold change of +/− 1.5, and the horizontal line is an adjusted p-value of 0.05; N=3 per treatment group (c) RT-qPCR validation of sequencing results for two robustly expressed miRNAs altered by metal exposure, miR-146a and miR-708. Data are represented as mean of technical replicates (N=12 for each exposure) + SE for 3 or more biological replicates. *p <0.05; **p < 0.01; ***p < 0.001 treatment versus control relative expression.

Table 1:

PC2 Loadings

	PC2 Loadings
miR	Value
rno-mir-134-5p	0.072594738
rno-mir-27b-5p	0.072163351
rno-mir-382-5p	0.071162951
rno-mir-421-3p	0.07110006
rno-mir-22-3p	−0.070975602
rno-mir-29a-3p_-_1	−0.070370273
rno-mir-146a-5p	−0.069618672
rno-mir-132-5p_-_1	−0.069344043
rno-mir-29a-3p_+_6	−0.069080702
rno-mir-708-5p	0.068978647

Table 2:

Differentially altered miRNAs

Exposure	miRNA	Fold Change	P-value	P-adj
iAs	rno-mir-708-5p	1.5566526	6.71E-13	2.24E-10
	rno-mir-146a-5p	1.7613383	1.98E-12	4.97E-10
	rno-mir-29b-1-5p_-_5	1.7117298	2.20E-10	2.45E-08
	rno-mir-369-3p	1.5683211	3.12E-06	0.000174
	rno-mir-541-5p	−1.6156638	3.48E-10	3.49E-08
	rno-mir-409a-3p_-_1	−1.8185392	1.20E-09	1.09E-07
	rno-mir-410-3p	−1.7233563	6.84E-09	5.27E-07
	rno-mir-409a-3p_-_2	−1.5879385	7.62E-06	0.000333
	rno-mir-134-5p	−1.5890461	8.97E-06	0.000374
Cd	rno-mir-195-3p	−1.5995832	2.56E-08	5.14E-06
Mn	rno-mir-582-3p_-_1	−1.5280725	3.04E-11	1.54E-08
	rno-mir-125b-1-3p	1.520848	8.60E-06	0.000507

Two miRNAs, miRs-146a and 708, were selected for validation by quantitative RT-PCR (qRT-PCR) based on their established roles in beta cell health (Roggli et al. 2010) and function (Rodríguez-Comas et al. 2017). This analysis revealed an almost two-fold increase in miR-146a upon exposure to iAsIII whereas Mn had the opposite effect, and no change was seen with Cd exposure. Also, exposure to iAsIII caused a nearly two-fold decrease in the expression of miR708, while Cd and Mn exposure both had non-significant effects. These data confirm the results obtained from high-throughput sequencing (Fig. 3c).

To obtain preliminary data on miR-146a expression after in vivo exposure to iAs, we performed small RNA-seq on islets previously isolated and archived as part of another study in which mice were exposed to 100 ppb iAs in drinking water. Islets from only a total of N=6 males and N=3 females across both treated and untreated conditions were available to us. The results suggest that iAs effects on miRNA-146a expression may be sex-specific, because we observed an upregulation of miR-146a only in the female mice (Supplementary Fig 1). However, we stress that this particular finding is very preliminary given the low sample size; therefore, well-powered and rigorously-controlled in vivo studies with both sexes included are required in the future.

Upregulation of miR-146a is iAs-specific

Since we found that iAs is the greatest disruptor of miRNA expression patterns, we focused our attention on iAs and one of its metabolites, MAsIII. First, we demonstrated a dose-dependent suppression of GSIS by iAs (Fig. 4a), concomitant with a dose-dependent increase in miR-146a (Fig. 4b), as determined by RT-qPCR. These data indicate that miR-146a is inversely correlated with GSIS during iAs exposure. Next, we tested the effect of MAsIII. Although 0.5 uM MAsIII exerted an effect on GSIS similar to that of 1 uM iAs, it had no effect at all on miR-146a levels (Fig 4c, d). These data suggest that miR-146a is sensitive only to iAs and not its methylated metabolites.

Figure 4.

GSIS (a) and miR-146a (b) expression (as determined by RT-qPCR) after 24-hour exposure to 0.5 or 1 μM iAs (miR-146a, 0 μM iAs N=27, 0.5 μM iAs N=15, 1 μM iAs N=22). Also, GSIS (c) and miR-146a (d) expression after 24-hour exposure to 1 μM iAs or 0.5 μM MAs (miR-146a, 0 μM iAs N=12, 0.5 μM iAs N=9, 1 μM iAs N=12). Data are represented as mean of technical replicates + SE for 3 or more biological replicates. *p < 0.05; **p < 0.01; ***p < 0.001 treatment versus control high glucose stimulation or treatment versus control relative expression.

Genes involved in beta cell function are significantly suppressed by iAs exposure, and several are validated targets of miR-146a-5p

We next sought to investigate the effect of iAs on the gene expression landscape. We profiled genes by RNA-seq in INS-1 832/13 untreated cells (control) and in cells exposed to iAs for 24 hours. After DESeq2 analysis, we found 1288 significantly altered genes (610 genes downregulated and 678 upregulated; fold-change > 1.5 and adjusted p < 0.05) in exposed cells (Fig 5). Using the Enrichr analysis of the ChEA, PPI and CORUM databases, we found that upregulated genes are significantly over-represented in the Nfkb signaling pathway (Fig. 6). Increased Nfkb activity is consistent with the elevation in miR-146a expression (Taganov et al. 2006). These data are indicative of a change in the Nfkb/miR-146 axis in response to iAs.

Figure 5.

(a) A heatmap of Euclidean distance between samples based on DESeq2 normalized gene expression. (b) Volcano plot displaying genes altered by 24-hr iAs exposure. The vertical lines represent a fold change of +/− 1.5, and the horizontal line shows adjusted p=0.05. Control, N = 5; iAs, N=4.

Figure 6.

(a) Results of pathway enrichment analysis for upregulated genes (n=678) using the CORUM database. (b) Results of ChIP-X enrichment analysis for upregulated genes. (c) Results of protein-protein interaction enrichment analysis for upregulated genes. Dashed vertical line represents adjusted p=0.05.

To uncover candidate miR-146a targets that are affected by iAs exposure, we focused on down-regulated genes, since miRNAs are negative regulators of gene expression. Using the Enrichr tool, we found that the cAMP and Wnt pathways (Yajima et al. 1999; Abiola et al. 2009), both of which are known to promote GSIS in beta cells, are enriched among downregulated genes (KEGG; adjusted p=0.068 and 0.069, respectively). Moreover, downregulated genes are also significantly enriched in the regulatory networks of ISL1 and FOXO1 (ChEA; adjusted p=0.000046 and 0.05, respectively), two transcription factors known to be important for controlling beta cell identity and function (Kitamura et al. 2005; Ediger et al. 2014). Lastly, we observed from the miRTarBase resource that several down-regulated genes are likely direct targets of miR-146a, most notably, Ca2+/calmodulin-dependent protein kinase IIα (Camk2a), an established regulator of insulin secretion (Dixit et al. 2013; Santos et al. 2014).

Discussion

Exposure to iAs and other metals have been associated with an increased risk for the development of T2D and impaired pancreatic beta cell function. However, the genetic and molecular mechanisms underlying this risk remain unresolved. In this work, we’ve carried out the first step to explore this area. Specifically, we have performed a comparative assessment across different metals on GSIS and miRNA profiles, identified the most significantly dysregulated miRNAs in response to iAs exposure, and performed comprehensive gene expression profiling to uncover regulatory pathways that may contribute to iAs-associated defects.

Several studies have reported that exposure to heavy metals might adversely affect pancreatic islet function and lead to the development of T2D. A 2012 study observed an inhibition of GSIS in a murine pancreatic beta cell line (MIN6) exposed to 1 μM Cd for 48 hours (El Muayed et al. 2012). Chang et al report significant cytotoxicity and impaired insulin secretion in rat RIN-m5F cells after 24-hour exposure to 3 or 5 μM Cd, respectively (Chang et al. 2013). Similarly, previous research from our lab using the INS-1 832/13 rat insulinoma cell line observed an inhibition of GSIS after only 24 hours of exposure to 5 μM Cd (Dover et al. 2018) suggesting the effects of Cd on beta cell function may be cell-type dependent.

Manganese is a well characterized essential trace metal. Mn serves as a cofactor for many reactions involved in the maintenance of glucose homeostasis, including reactions of insulin synthesis, and secretion (Korc 1983; Baly et al. 1984). As such, Mn deficiency has been implicated in diabetes development, potentially through an altered metabolism of the metal. However, results of a human cohort study by Shan et al suggested that both low and high levels of plasma manganese are associated with diabetes (Shan et al. 2016). An in vitro study by Dover et al demonstrated that exposures to 12.5 μM Mn for 24 hours in INS-1 cells causes an inhibition of the mitochondrial oxygen consumption rate as well as inhibition of GSIS (Dover et al. 2018).

Numerous cross-sectional and prospective epidemiological studies from Taiwan, Bangladesh, Mexico, the U.S. and other areas have shown that chronic exposure to iAs is associated with diabetes (Maull et al. 2012; Wang et al. 2014; Sung et al. 2015). Interestingly, human exposure to iAsIII is thought to cause an impairment of pancreatic beta cell function rather than insulin resistance (Del Razo et al. 2011; Gribble et al. 2012; Rhee et al. 2013). In vitro studies have expounded on this by showing that 24-hour exposure to non-cytotoxic concentrations of iAsIII results in an impairment of GSIS in INS-1 cells without affecting cell viability (Dover et al. 2017). Whole pancreatic islets exposed to iAsIII also exhibit impaired GSIS yet no detrimental effects on insulin content of the islets or islet viability were observed (Douillet et al. 2013). Our current study confirms the impairment of GSIS after iAsIII exposure, and offers insight into potential transcriptional mechanisms leading to dysregulated GSIS.

In our current study, iAs, Cd, and Mn exerted significant and comparable negative effects on GSIS. However, relative to iAs, Cd and Mn led to only modest effects on the miRNA landscape, which could suggest that these metals perturb GSIS through different molecular mechanisms. Published literature identified a role for miR-195 in pancreatic development in mice. Downregulation of miR-195 in our study could suggest a decreased regeneration ability in beta cells exposed to Cd. Mn exposure led to a downregulation of miR-582, whose target genes are greatly enriched in the Wnt signaling pathway (Fang et al. 2015). Activation of Wnt/β-catenin signaling partially mediates the upregulation of glucose transporters in the cell membrane as a response to insulin signaling, which allows the cell to uptake glucose (Wang et al. 2017). A downregulation of miR-582 could suggest a compensatory mechanism by which the cell is attempting to increase Wnt signaling, and thus increase glucose uptake. Lastly, miR-125b was upregulated in response to Mn. A study by Bloomston et al shows miR-125b to be one of 7 miRNAs overexpressed in chronic pancreatitis which could mean that a long-term exposure to Mn may be carcinogenic (Bloomston et al. 2007).

A previous study of beta cell function and inflammation identified miR-146a as an important player in cytokine-mediated beta cell failure in MIN6 cells (Roggli et al. 2010). Newer studies using mouse models have confirmed the role of miR-146a as an important regulator of inflammatory cytokines (Bhatt et al. 2016; Chen et al. 2017). A recent meta-analysis of 12 published studies found an association between T2D susceptibility and a downregulation of miR-146a (Alipoor et al. 2017). Notably, iAs exposure has been established to increase inflammation and oxidative damage within the pancreas. Interestingly, while exposure to MAs did affect GSIS, it did not elicit the same upregulation of miR-146a, which may suggest that iAs and MAs act on different targets to affect insulin secretion. The upregulation of miR-146a in INS-1 cells may also represent a compensatory mechanism to protect against the inflammatory effects of iAs exposure. Our findings support the connections identified between miR-146a, iAs exposure, and oxidative stress that are worth investigating further.

Camk2a is one of the main effector enzymes involved in calcium signaling, a critical mechanism regulating GSIS in beta cells (Dadi et al. 2014). It also serves as a mediator of the inflammatory response to oxidative stress (Waldsee et al. 2014; Zhang 2017; Zhong and Huang 2017). Moreover, we and others have previously linked iAs exposure to decreased calcium influx and ultimately impaired GSIS (Díaz-Villaseñor et al. 2008; Huang et al. 2019). Several transcriptome-wide studies over the past decade suggest that miR-146a likely directly targets the 3’ UTR of the Camk2a gene (Chi et al. 2009; Leung et al. 2011; Schug et al. 2013). Overall, our findings suggest that iAs-induced suppression of GSIS could be explained in part by the inhibition of Camk2a and that this inhibition may be mediated by miR-146a. Future studies should focus on further functional characterization of Camk2a to determine whether its suppression contributes to reduced GSIS in the context of iAs exposure.

Limitations

In the present study, we used the INS-1 832/13 cell line, which has been widely used in studies examining insulin secretion and the underlying mechanisms (Hohmeier et al. 2000a; Park et al. 2007; Skelin et al. 2010; Lorenz et al. 2013; Dover et al. 2017; Jo et al. 2018). In our studies, inhibition of glucose-stimulated secretion in this cell line after in vitro exposure to iAs and its metabolites (Dover et al. 2017) was consistent with beta-cell dysfunction reported in epidemiologic studies carried out in populations exposed to iAs (Del Razo et al. 2011; Mendez et al. 2016). Still, translatability of our current findings to in vivo conditions remains uncertain. Results of our pilot in vivo experiment suggest that miR-146a is also overexpressed in mouse pancreatic islets in response to in vivo iAs exposure, although this effect may be sex-specific. Our previous published studies show that susceptibility to the diabetogenic effects of iAs exposure differ between male and female mice (Huang et al. 2016, 2018; Douillet et al. 2017). Future studies should further elucidate the potential role of miR-146a, and other miRNAs, in diabetes associated with metal exposure (specifically iAs) with sex as an important variable.

Conclusion

The main findings in this work are four-fold: (1) the definition of metal-specific miRNA profiles; (2) the determination that iAs influences miRNA expression patterns much more robustly than Cd or Mn; and (3) the identification of miR-146a-5p as the most upregulated miRNA upon iAs treatment, and (4) the involvement of genes down-regulated by iAs in beta cell function, including Camk2a, a predicted target of miR-146a. Both miR-146a and Camk2a have been associated with insulin secretion, and therefore merit further functional investigation. To our knowledge, this study is the first to identify and report the potential relevance of the relationship between miR-146a-5p, Camk2a, and Nfkb signaling in the context of iAs-induced beta cell dysfunction and diabetes.