Introduction

More than 150 million people in 70 countries around the world are affected by drinking arsenic-contaminated groundwater with arsenic content much above the permissible limit laid down by WHO (WHO 1996). Among six affected states in India, West Bengal is the worst affected state where more than 26 million people (Chakraborti et al. 2009) in 9 districts are affected by drinking arsenic-contaminated water. Arsenic-induced premalignant and malignant skin lesions are hallmarks of chronic arsenic toxicity which includes hyperkeratosis, Bowen’s disease (BD), squamous cell carcinoma (SCC), and basal cell carcinoma (BCC). Interestingly, only 15–20% of the total population shows arsenic-induced skin lesions (arsenic-susceptible group) and the rest do not (resistant group), even with similar levels of arsenic exposure through drinking water. Arsenic exposure has also been previously associated with several deleterious health effects including cancers of internal organs, peripheral neuropathy, respiratory distress, conjunctivitis of the eyes, cardiovascular disease, diabetes mellitus, and liver diseases (Das et al. 2012; Guha Mazumder 2008; Paul et al. 2015).

Altered microRNA (miRNA) expression profiles have been associated with various deleterious health outcomes including cancers (Calin and Croce 2006; Sonkoly et al. 2008). Only very recently prenatal arsenic exposure has been associated with miRNA expression from human cord blood in new born children (Rager et al. 2014). It has also been found that inorganic arsenic exposure triggers a shift in miRNA expression in Jurkat cell lines (Sturchio et al. 2014) but such studies involving adult humans who are chronically exposed to arsenic have hardly been done. So, we recognized the need to explore the role of miRNA in arsenic-induced health effects in our study population, the chronically exposed individuals of West Bengal.

In this study, plasma miRNA microarray was done in arsenic-exposed individuals to find out whether the expression pattern varies in the two exposed groups and whether miRNA contributes to arsenic susceptibility. The results were validated with quantitative real-time PCR in 60 individual samples for some selected miRNAs. Bioinformatic analyses were done to predict the downstream target molecules and different putative pathways that are affected due to chronic arsenic toxicity. Association study was also done between the commonly occurring non-dermatological health effects and miRNA expressions in these study subjects.

Materials and methods

Study site and participants

A case-control study was conducted involving 60 individuals from highly arsenic-affected Murshidabad district of West Bengal. The study participants were divided into two age-sex–matched groups as follows: 30 arsenic-exposed individuals with skin lesions (cases) and 30 arsenic-exposed individuals without skin lesions (controls). Arsenic content in their drinking water and urine samples were measured for exposure assessment. Results are shown in Table 1. The arsenic content in their drinking water was much above the permissible limits laid down by WHO (10 μg/L). However, there was no significant difference in arsenic content in either drinking water or urine samples between the skin lesion and no skin lesion groups. Only non-smoker individuals were recruited as study participants. Urine, water, and blood samples were collected only from those subjects who provided informed consent to participate in the study. The details of the study site selection and sample collection are described previously (Banerjee et al. 2011). This study was conducted in accordance with the Helsinki II Declaration and approved by the Institutional Human Ethical Committee of CSIR-Indian Institute of Chemical Biology.

Table 1 Demographic characteristics of the study participants
Full size table

Arsenic estimation in water and urine samples

Drinking water and first morning voids of urine samples were collected from the study participants following standard protocols. Immediately after collection, both the samples were stored in salt-ice mixture and brought to the laboratory and stored at − 20 °C, before estimation of arsenic was done. Arsenic measurement by an atomic absorption spectrometer (Shimadzu AA-7000) coupled to a graphite furnace atomizer (Shimadzu) using the AA Wizard software and arsenic lamp (lamp current 380 mA).

Collection of blood and plasma

Venous blood was drawn by vein-puncture method, collected in EDTA-vacutainer tubes, and immediately put on ice. Blood samples were brought to the laboratory on ice within 2 h after collection where further work was done. Plasma samples (about 500 μl) were collected by centrifugation of whole blood for 5 min at 1500 rpm at room temperature.

Plasma microRNA profiling and data analyses

Plasma RNA was isolated from 24 individuals (12 from each group of the following groups: arsenic-exposed individuals with and without skin lesions). These were pooled into two separate samples respectively. RNA was quantified in a nanodrop analyzer and the samples were then sent to the iLife Technologies, Bangalore, for microarray analysis using the Affymetrix Gene Chip miRNA 2.0 platform. A cutoff of fold change more than or equal to 2.0 was employed to identify the differentially regulated miRNAs. Each sample was analyzed in triplicate. Microarray data was processed and analyzed using R programming platform and Bioconductor packages. Background correction was carried out using the normexp method implemented in the Bioconductor package limma. Normalization between the arrays was carried out using quantile method. In order to identify genes that are differentially expressed between samples, an empirical Bayes moderated t test, as implemented in limma, was implemented. To identify genes that are differentially expressed and further mining, a p-value threshold of 0.001 and a fold change cutoff (FC) of > 2 for upregulated and ≤ 0.5 for downregulated were applied.

Quantitative real-time PCR

In order to validate the microarray data, 12 differentially expressed miRNAs were selected for the downstream work based on their target association with cancers of skin and other internal organs (as found from KEGG pathway analysis and Pubmed search). miR-21, miR-23a, mir-124, miR-126, miR-619, miR-3613, miR-1282, miR-4530, miR-17, miR-18a, miR-27a, and miR-122 were identified as important candidates. For each of the 24 samples employed for profiling work, cDNA conversion and q-real-time PCR were carried out. Additionally plasma samples from 36 more individuals (18 individuals from each of the groups) were also employed for the purpose.

miRNA was isolated from plasma samples using the Qiagen miRNeasy Serum/Plasma Kit (Qiagen, Valencia, CA) following manufacturer’s instructions. The quantity (ng/μl) and purity (A260/280 ratio) was measured by the Nanodrop 2000 spectrophotometer (Thermoscientific, Waltham, Massachusetts, USA)

Total RNA was converted to cDNA using ReverseAid H minus First strand cDNA synthesis kit (Thermo Fisher Scientific, USA) following manufacturer’s instructions. The primers used were as follows:

  • miR21: 5′-CTCAACTGGTGTCGTGGAGTCGGCAATTCAGTTGAGTCAACATC-3′

  • miR23a: 5′-CTCAACTGGTGTCGTGGAGTCGGCAATTCAGTTGAGGGAAATCC-3′

  • mir124: 5′-CTCAACTGGTGTCGTGGAGTCGGCAATTCAGTTGAGGGCATTCA-3′

  • miR126: 5′-CTCAACTGGTGTCGTGGAGTCGGCAATTCAGTTGAGCGCATTAT-3′

  • miR619: 5′-CTCAACTGGTGTCGTGGAGTCGGCAATTCAGTTGAGCATGAGCC-3′

  • miR3613: 5′-CTCAACTGGTGTCGTGGAGTCGGCAATTCAGTTGAGGAAGGGTT-3′

  • miR1282: 5′-CTCAACTGGTGTCGTGGAGTCGGCAATTCAGTTGAGAAGCAGAA-3′

  • mir4530: 5′-CTC AACTGGTGTCGTGGAGTCGGCAATTCAGTT GAGCGCTCCCG-3′

  • miR17: 5′-CTCAACTGGTGTCGTGGAGTCGGCAATTCAGTTGAGCTACCTGC-3′

  • miR18a: 5′-CTCAACTGGTGTCGTGGAGTCGGCAATTCAGTTGAGCTATCTGC-3′

  • miR27a: 5′-CTCAACTGGTGTCGTGGAGTCGGCAATTCAGTTGAGTGCTCACA-3′

  • miR122: 5′-CTCAACTGGTGTCGTGGAGTCGGCAATTCAGTTGAGCAAACACC-′

  • U6 snRNA: 5′-AAAATATGGAACGCTTCACG-3′

for cDNA conversion. The miRNA expression level was determined by quantitative real-time PCR (Agilent technologies: Stratagene MX3000P) by using MESA GREEN qPCR Master MIX Plus SYBR assay I Low Rox Kit (Eurogentec), following the manufacturer’s instructions and normalized using the 2−ΔΔCT method relative to the U6 small nuclear RNA.

The forward primers used were as follows:

  • miR-21-F: 5′-ACACTC CAGCTGGGTAGCTTATCAGACTGA-3′

  • miR-23a-F: 5′-ACACTCCAGCTGGGATCACATTGCCAGGGA-3′;

  • mir-124-F: 5′-ACACTCCAGCTGGGTAAGGCACGCGGTGAA-3′

  • miR126-F: 5′-ACACTCCAGCTGGGTCGTACCGTGAGTAAT-3′

  • miR-619-F: 5′-ACACTCCAGCTGGGGCTGGGATTACAGGCA-3′;

  • miR-3613-F: 5′-ACACTCCAGCTGGGACAAAAAAAAAAGCCC-3′;

  • miR-1282-F: 5′-ACACTCCAGCTGGGTCGTTTGCCTTTTTCT-3′

  • miR-4530-F: 5′-ACACTCCAGCTGGGCCCAGCAGGACGGGAG-3′

  • miR17-F: 5′-ACACTCCAGCTGGGTAAAGTGCTTACAGTG-3′

  • miR18a-F: 5′-ACACTCCAGCTGGGTAAGGTGCATCTAGTG-3′

  • miR27a-F: 5′-ACACTCCAGCTGGGAGGGCTTAGCTGCTTG-3′

  • miR122-F:5′-ACACTCCAGCTGGGTGGAGTGTGACAATGG-3′

  • URP: 5′TGGTGTCGTGGAGTCG-3′

  • U6-F: 5′-CGCTTCGGCAGCACATATACTAAAATTGGAAC-3′;

  • U6-R: 5′-GCTTCACGAATTTGCGTGTCATCCTTGC-3′.

All the reactions were carried out in triplicate (URP, universal reverse primer).

Western blot analyses

Cells were lysed with RIPA lysis buffer (Thermo Fisher Scientific, Rockford, USA), for protein isolation from the PBMC of the study participants as per manufacturer’s instructions. Equal amounts of proteins were subjected to 12% SDS-PAGE separation and Western blot analyses were done in as previously described (Banerjee et al. 2017). Image J software was used to for densitometric analysis of the bands obtained. β-actin was used as the loading control.

miRNA target prediction and pathway enrichment analyses

MiRNA target prediction and function analysis was done by bioinformatic approaches. We used TargetScan (http://www.targetscan.org/,TargetScan Human 6.2) to predict the targets of differentially expressed miRNAs. In TargetScan, biological targets of miRNAs were predicted by searching for the presence of conserved 8 mer and 7 mer sites that match the seed region of each miRNA. We predicted only those as target genes who were targeted by at least 20 differentially expressed miRNAs as potential functional targets in arsenicosis. Functional annotation and pathway enrichment of those predicted genes were done by employing the KEGG (Kyoto Encyclopedia of Genes and Genomes) database using a DAVID (the Database for Annotation, Visualization and Integrated Discovery) online analysis software (http://david.abcc.ncifcrf.gov/tools.jsp).

Identification of non-dermatological health effects in the study population

Identification of the non-dermatological health effects like respiratory problems, peripheral neuropathy, and conjunctival irritations of the eyes was done by expert physicians in the relevant field as explained previously (Banerjee et al. 2011, 2017). We have divided our entire study population in such a way that any individual either have any of the above disease or do not. Then, we found out differential miRNA expression patterns between both the groups (with and without the disease) in terms of fold change.

Statistical analyses

Data are expressed as mean ± SD. Chi-square test was employed to compare the distribution of gender among the groups. One-way ANOVA with the Tukey–Kramer multiple comparison post test was done to test for significant differences in age, arsenic content in urine, and water samples. Microsoft Excel and Graph Pad Instat3 (San Diego, CA) software were used for the purpose. Image J software was used for the densitometric analysis of the Western blot bands.

Results

Demographic characteristics of the study participants

Demographic characteristics of the study participants are shown in Table 1. The study participants were matched with respect to their age, sex, and socio-economic conditions. Only non-smoker individuals were recruited for the study. Arsenic exposure assessment showed that arsenic content in their drinking water was much above the permissible limits laid down by WHO (> 10 μg/L). However, there was no significant difference in arsenic concentration in the drinking water or urine samples between the exposed group with skin lesions and the exposed group without skin lesions (cases and controls), indicating that both groups had similar levels of arsenic exposure.

Plasma microRNA profiling of the study participants

Microarray data shows that 202 microRNAs are differentially expressed in the plasma of the skin lesion group as compared with the no skin lesion group, where 199 miRNAs were upregulated and 3 were downregulated. The entire results are shown as heat map in Fig. 1 and in electronic supplementary information Table 1. The data were analyzed with miRNA QCTool (Affymetrix, Santa Clara, California, USA) using the Affymetrix default analysis settings and quantile as the normalization method.

Fig. 1
figure 1

Heat map showing miRNA profile of arsenic exposed individuals with skin lesions (symptomatic) vs exposed individuals without skin lesions (asymptomatic), where among 202 differentially expressed miRNAs, 199 have been upregulated and 3 have been downregulated in the former group

Full size image

Validation by quantitative real-time PCR

Since performing quantitative real-time PCR reactions for all the differentially expressed miRNAs were beyond our scope, so we focused on important miRNAs based on whether they were closely associated with different cancers including that of the skin, chosen by KEGG pathway analysis and Pubmed search. We did quantitative real-time PCR analysis for miR-21, miR-23a, miR-124, miR-126, miR-619, miR-3613, miR-1282, and miR-4530. Results are shown in Fig. 2a which shows that miR-21, miR-23a, miR-619, miR-126, and miR-3613 were significantly upregulated (4.49-, 2.26-, 4.64-, 2.98-, 3.73-folds respectively) and miR-1282 and miR-4530 (− 3.11 and − 4.26 respectively) were significantly downregulated in the skin lesion group compared with the no skin lesion group; however, miR-124 showed no significant difference in the expression patterns. When matched with the microarray data (electronic supplementary information Table 2), similar trend in expression levels were found.

Fig. 2
figure 2

Quantitative real-time PCR data showing fold change in expression levels of different microRNAs in the exposed group with skin lesions compared to the exposed group without skin lesions

Full size image

miRNA target prediction and bioinformatic analyses

Since miRNAs regulate the expression of their target genes, so, the next step was to screen the potential functional targets of the differentially expressed miRNAs. By using the target prediction program TARGET SCAN, we identified 1275 potential functional targets that may participate in arsenic susceptibility. Pathway enrichment analysis was done as described under the “Methods” section and the results are represented in Table 2. Pathway enrichment analysis revealed 7 different pathways corresponding to the target genes when compared between the arsenic-exposed skin lesions and no skin lesion groups (p < 0.05, Table 2). The pathways that have the potential functional targets are mainly those which are dealing with glycerophospholipid, metabolism, lipid biosynthesis, colorectal cancer, T cell receptor signaling, progesterone-mediated oocyte maturation, neurotrophin signaling, and insulin signaling pathways, among which glycerophospholipid metabolism pathway (Fig. S2) was found to be the most enriched pathway with 10 potential functional targets. Detailed study of the pathways in cancer revealed that these miRNAs have targets in basal cell carcinoma, melanoma, colorectal cancer, bladder cancer, lung cancer, endometrial cancer, and prostate cancer; most of which have been associated with chronic arsenic exposure.

Table 2 Enriched pathways for potential functional targets in arsenic susceptibility
Full size table

Literature survey showed that miR-17, miR-18a, miR-27a and miR-122 are important contenders in the enriched pathways as found above. Therefore, we did quantitative real-time PCR to find their expression status in our study population. Results are shown in Fig. 2b. We found significant upregulation of miR-27a, miR-122 in the cases as compared with the controls. miR-17 and miR-18a showed no significant change in expression levels (Fig. 3).

Fig. 3
figure 3

Representative Western blots. Relative intensity of each band after normalization with the intensity of β-actin in a blot (below each Western blot) was measured. Image J software was used for the purpose

Full size image

Protein expression study

Since most of the microRNAs under consideration like miR21, miR23a, and others are established oncomers, and arsenic exposure is associated with cancerous outcomes; therefore, we checked whether the expressions of the survival proteins, pAKT and PI3K, were also affected. PTEN is a tumor suppressor as well as a target of miR21. So, we did Western blot analyses to study their expression at protein levels in the two groups. Results showed that expression of both pAKT and PI3K are greater in the exposed individuals with skin lesions compared with the individuals without the same. While, PTEN expression was greater in the no skin lesion group compared with the skin lesion group.

Association of miRNA expressions with non-dermatological health effects

Commonly occurring non-dermatological health effects in our study population include respiratory problems, peripheral neuropathy and conjunctival irritation of the eyes. We found that miRNAs were differentially associated with the commonly occurring non-dermatological health effects (Table 3). miR-21 and miR-3613 expression levels were found to be associated with respiratory distress and conjunctival irritation of the eyes (> 2-fold change). miR-23a expression was higher for the individuals with respiratory problems compared with those who did not have the disease. miR-124 levels were significantly elevated in the individuals with eye problems compared with those who did not have the disease.

Table 3 Association of miRNA expression levels with non-dermatological health effects
Full size table

Discussion

Among over 26 million arsenic-exposed people, only 15–20% shows arsenic-induced skin lesions but the rest do not, but until date there is no report on the probable role played by the microRNAs in this aspect. Since epigenetic modifications have been found to play important role in arsenic-induced health effects, we hypothesized that altered miRNA expressions might contribute to arsenic susceptibility as well. For that reason, we studied the plasma microRNA expression patterns in the arsenic-exposed individuals with and without skin lesions followed by downstream validation by qRT-PCR. Arsenic exposure levels were similar in both the study groups.

The microarray data shows that 199 miRNAs are upregulated in the exposed group with skin lesions compared with the no skin lesion group. Most of the upregulated microRNAs are either oncogenic miRNAs or miRNAs associated with inflammation (like miR-21, miR-23a, miR-3613, let-7 family, mir-17, miR-27a, and mir-122). These miRNAs and their increased expressions have been previously associated with different types of cancers and inflammatory conditions (Banerjee et al. 2017; Knyazev et al. 2016 ;Roufayel and Kadry 2017; Suresh et al. 2015; Wu et al. 2015, Wang et al. 2018; Zhao et al. 2017). When verified by real-time PCR in our laboratory conditions, similar changes in microarray expressions were also found. This marked difference in miRNA expression patterns in the two groups might make the individuals with skin lesions “more susceptible” compared with the “more resistant” no skin lesion group and they develop premalignant and malignant skin lesions, even with similar levels of arsenic exposure. From this, the increased prevalence of cancers of other internal organs in the exposed individuals with skin lesions might also be explained. pAKT and PI3K are two molecules involved generally in cell proliferation and are associated with various cancerous outcomes. PTEN is a tumor suppressor and direct downstream target of miR21. Increased expression of pAKT and pI3K and decreased expression of PTEN in the cases as compared with the controls shows that they are at higher risk of developing different forms of cancers.

We also came across previous results where miR-21 and miR-27a and miR-122 were found to be associated with glycerophospholipid metabolism (Yang et al. 2015). Another evidence show that miR-122 and miR-33 are found to have a significant impact on lipid homeostasis and are potential therapeutic targets for treating lipid disorders (Horie et al. 2014). In our results, we found that both miR-27a and miR-122 were upregulated in the exposed individuals with skin lesions compared with the exposed individuals without skin lesions. Our results are consistent with previous findings where serum levels of miR-122 have found to increase significantly in patients with non-alcoholic liver diseases and chronic hepatitis C infections (Cermelli et al. 2011; Pirola et al. 2015). Since arsenic exposure has previously been associated with cancers of internal organs including that of the liver (Guha Mazumder 2008), our results might indicate a link between the same.

Pathway enrichment analysis of the microarray data revealed that 7 different pathways had targets of the candidate miRNAs namely glycerophospholipid, colorectal cancer, T cell receptor signaling, progesterone-mediated oocyte maturation, neurotrophin signaling, and insulin signaling pathways. Glycerophospholipid metabolism pathway was found to be the most enriched pathway with 10 potential functional targets. Glycosphingolipid biosynthesis is another important pathway involving 4 potential targets. In this context, it is worthwhile to discuss that human skin consist three layers of protein-lipid-protein and any defects in lipid biosynthesis or metabolism could affect their integrity which might lead to the development of skin lesions. Since cholesterol molecules play key role in converting normal keratinocytes to BCC (KEGG Pathway), defects in lipid metabolism/biosynthesis might also contribute to the increased incidences of skin lesions/cancers in the skin lesion group. Our hypothesis may be strengthened by previous studies which relate defects in lipid metabolism or biosynthesis with different types of cancers (Liao et al. 2017; Vergara et al. 2017). Another aspect is that these miRNAs have also targets in T cell receptor signaling and colorectal cancer pathways. Our previous studies have shown that the exposed individuals with skin lesions have impaired immune system with reduced T cell proliferation and cytokine secretion, impaired macrophage functions, and increased immune cell death due to apoptosis in the exposed individuals with skin lesions (Banerjee et al. 2009, 2011; Biswas et al. 2008), thereby supporting this result. Again, colorectal cancers involves a number of pathways like WNT signaling, Ras signaling, and p53 signaling pathways (Colussi et al. 2013), many of which are targeted by the miRNAs differentially expressed in this study. This might make one group more susceptible than the other.

Differential miRNA expression patterns were also found to be associated with the commonly occurring non-dermatological health effects in the study population. Significant association was found between expressions miRNA of the miR-21, miR-23a, miR-124, miR-126, and miR-3613 with respiratory distress and conjunctival irritation of the eyes, while no association was found between miR-619, miR-1282, and miR-4530 levels with any of the disease conditions. This may be due to the fact that the diseases mentioned are complex diseases and such outcomes might be as a result of various gene-gene and gene-environment interactions.

Our results are in accordance with previous findings where one or more of these miRNAs have been associated with eye problems or respiratory distressed conditions (Liu et al. 2016; Li et al. 2016; Kupczyk and Kuna 2014; Zhang et al. 2017). The increased expressions of different miRNAs like miR21, 23a, and others may contribute to increased inflammatory conditions which contribute not only to the precancerous and cancerous skin lesions but also to the respiratory diseases and conjunctival irritation of the eyes in the study subjects.

Conclusion

This study for the first time shows that microRNA expression patterns vary greatly among the arsenic-exposed individuals with and without skin lesions. These miRNAs have targets in different biochemical pathways which might lead to different physiological responses in the two exposed groups. This also leads us to the involvement of new possible mechanisms and of arsenic-induced toxic outcomes involving lipid molecules, which requires immediate attention. Further detailed molecular study on the expression of downstream molecules in the enriched pathways is required to establish our findings. To our knowledge, this is the first report of its kind in an arsenic-exposed population of West Bengal in India.