MicroRNAs as Potential Signatures of Environmental Exposure or Effect: A Systematic Review

1Most common diseases result from the combined effect of genes and environmental factors and the interactions between them. Epigenetic effects and non-coding gene products have gained research focus over the last two decades because protein-coding genes cannot account for all observed genomic effects. Here we focus on microRNAs (miRNAs) as key regulators of health and disease. miRNAs are endogenous, single-stranded, short non-coding RNA sequences (~ 22 nucleotides) that regulate gene expression at the posttranscriptional level. Since the first discovery of miRNAs in Caenorhabditis elegans, hundreds of miRNAs in eukaryotes have been identified to influence physiological processes such as development, growth, differentiation, immune reaction, and adaptation to stress. Diverse disease states, such as cancer and heart failure, are associated with distinct miRNA signatures, suggesting that specific miRNA programs are activated in pathophysiological processes.

2Recent advances in molecular biology opened the opportunity for new approaches in population-based studies, in which exposures to a broad spectrum of environmental pollutants are evaluated in concert with biological systems, a concept proposed as the “exposome”. From this viewpoint, miRNAs could potentially be novel biomarkers of exposure. For the purpose of this review, we focused on the response of miRNAs to environmental exposures.

3miRNA characteristics. miRNA-mediated gene silencing is accomplished by base pairing of the 5´ region of miRNAs with the target mRNA sequence, leading to translational repression and/or mRNA degradation. miRNAs have been paradoxically shown to up-regulate gene expression by enhancing translation under specific conditions. The effect of miRNA expression on gene expression is not linear, as multiple miRNAs may target the same mRNA, and the majority of mRNAs contain multiple binding sites for miRNAs, generating a highly complex regulatory network system. For details on miRNA synthesis, biogenesis, miRNA mechanism of action, see Figure 1 and reviews by and.

4miRNA nomenclature. miRNAs are named using the “miR” prefix and a unique identifying number (e.g., miR-1, miR-2). The identifying numbers are assigned sequentially, with identical miRNAs having the same number, regardless of organism. Paralogous sequences whose mature miRNAs differ at only one or two positions are given lettered suffixes: for example, miR-10a and miR-10b. Distinct hairpin loci that give rise to identical mature miRNAs have numbered suffixes (e.g., mir-281-1, mir-281-2). The mature sequences are designated “miR,” whereas the precursor hairpins are labeled “mir.” The -3p and -5p suffixes that sometimes appear within an miR name refer to the arm from which the mature sequence comes. For nomenclature guidelines, see.

5miRNA analysis techniques suitable for large epidemiological studies. In recent years, miRNA expression changes following exposure to environmental toxicants, even before disease onset, have gained researchers’ interest. The measure of miRNAs in large epidemiological studies needs to be high throughput and sensitive enough to detect small changes in healthy subjects. At the same time, techniques need to be affordable in order to be conducted in large population studies. Moreover, given the complexity of phenomena induced by exposure but not fully explained by an effect on a single transcript, current research is going toward genome-wide techniques. Another challenge is tissue specificity of miRNAs: The availability of only noninvasive samples in epidemiological studies conducted on healthy populations limits our capability to investigate target tissues and opens important questions on the meaning of those markers in surrogate tissues. In epidemiological research, free and exosomal miRNAs in body fluids are interesting study objects because of their potential to serve as a proxy for tissue-specific miRNAs. A limitation of this approach is that these miRNAs differ between different body fluids, and their function is not clear. Although miRNAs hold promise as exposure biomarkers, recent studies have been primarily disease focused [reviewed by ].

6Genome-wide miRNA analysis can be achieved by amplification-based [real-time quantitative reverse-transcriptase polymerase chain reaction (qRT-PCR)], hybridization-based (microarrays), and sequencing-based [next-generation sequencing (NGS)] technologies. Method selection depends on the type of sample to be analyzed and the RNA preparation protocol used. qRT-PCR is considered the gold standard because of its sensitivity, specificity, accuracy, and simple protocols. qRT-PCR can evaluate candidate miRNA expression or array plates that include a large number of miRNAs in one reaction, to OpenArray® (Applied Biosystems, Life Technologies), which allows the simultaneous amplification of a very large panel of miRNAs using nanoscale volumes. In a recent review, compared qRT-PCR to different array-based platforms used to study mRNAs/miRNAs.

7Several miRNA microarray chip platforms that are commercially available [e.g., Affymetrix GeneChip® 3.0 miRNA array (Affymetrix Inc.), Agilent Human miRNA Microarray system (Agilent Technologies), Exiqon miRCURY LNA™ microarray (Exiqon Inc.)] differ in probe design and detection stringency. The limitation of this microarray chip method is the availability and stringency of probes on the chip platform that pair with miRNAs of interest. Microarrays have the advantage of being easily correlated to mRNA expression data, thus providing functional information. Furthermore, unlike other current miRNA analysis techniques, microarrays allow fast analysis of miRNAs without an arbitrary preselection step. However, the large amount of data produced can generate false-positive results, and the time-consuming step of validation by qRT-PCR is almost necessary.

8NGS strategies based on deep sequencing overcome some of the technical drawbacks of probe-based methodologies, especially the ability to detect only previously known sequences. As miRNAs are sequenced directly, information about sequence variations or posttranscriptional RNA editing becomes available for further analysis. The newly developed Nanostring nCounter 27 (Nanostring Technologies Inc.) uses two sequence-specific capture probes to allow for discrimination between similar variants of a single miRNA. NGS technologies [e.g., Illumina/Solexa (Illumina Inc.), GA Roche/454 GS FLX Titanium (Roche Diagnostics Corp.), and ABI/SOLID (Applied Biosystems)] allow complete “miRnomes” to be sequenced and allow for the discovery of novel miRNAs and isoforms. Another benefit of NGS technology is that it can identify precursor and primary miRNAs as well as their mature forms. NGS will likely become the gold standard for miRNA analysis because of its ability to sequence short fragments in a high-throughput mode. The choice between these methods is a key factor in establishing the possibility of success of any epidemiological study. Each method has pros and cons and should be evaluated based on the specific research.

9Search strategy and selection criteria. To identify the articles relevant to this topic, we undertook a comprehensive search of the PubMed (http://www.ncbi.nlm.nih.gov/pubmed) and ScienceDirect (http://www.sciencedirect.com/) databases initially using “microRNA” and “environmental exposure” as key terms. We did additional searches in which we replaced “microRNA” with “mir,” “miRNA,” or “epigenetic changes” and we substituted “environmental exposure” with “smoking,” “passive smoking,” “cigarette smoke,” “air pollution,” “nanoparticle exposure,” “bisphenol A,” “endocrine disruptors,” or “chemical exposure” in every possible combination. We also considered review articles as well as references found in our literature search. We excluded articles not written in English. The PubMed search covered 1 January 1980 to 1 June 2014. Articles dealing only with the description of single nucleotide polymorphisms (SNPs) in miRNA genes were disregarded, as were those articles dealing only with the description of miRNAs in nonmammalian species. A flowchart detailing the search strategy is presented in Figure 2. For miRNAs differentially expressed in response to more than one personal or environmental exposure, we researched disease phenotypes correlated with them by searching each of these miRNAs on the Human microRNA Disease Database (HMDD; http://202.38.126.151/hmdd/mirna/md/) and the miR2Disease Base (http://www.mir2disease.org/). Results of these searches are presented in Table 1, including the direction of regulation (up or down) of the miRNA and the ensuing phenotype.

10Smoking-induced changes in miRNA expression. The most studied environmental factor in relation to epigenetics is smoking; it was among the first factors shown to affect the miRNA machinery in humans. Results of in vitro studies concerning smoking and miRNAs are summarized in Table 2.

11analyzed miRNA expression patterns in the lungs of mice exposed to passive cigarette smoke, and they established life-course–related miRNA expression changes by comparing miRNA expression in lungs from unexposed newborn, postweaning, and adult mice. These researchers observed developmental-stage–specific miRNA expression profiles in which miRNAs that were highly expressed in newborns tended to be less expressed in adult mice and vice versa, whereas miRNA expression in postweaning mice was intermediate. Results from in vivo studies concerning smoking and miRNAs are shown in Table 3.

12Two studies reported a comparison between mRNA and miRNA whole genome expression patterns for smokers and nonsmokers. reported that quitting smoking altered the plasma miRNA profiles to resemble those of nonsmokers. In addition, Let-7c and miR-150 could be of importance in the initiation of smoke-induced decline of lung function, because genes that were associated with lung function impairment in genome-wide association studies have been reported to be significantly enriched in binding sites for these miRNAs, namely STAT3 and TNFR-II.

13The effect of in utero exposures on health during childhood and later in life is a growing area of research interest with major public health implications. An adaptive response in the fetus to in utero exposures can result in persistent changes into adulthood. miRNA expression levels in placenta can affect health later in life. Studies on miRNA expression and human exposure at different stages of life (in utero, adult) are included in Table 4.

14Not surprisingly, miRNAs that are frequently observed to be down-regulated in response to smoking have also been identified as down-regulated in lung, pancreatic, and stomach cancer. Development of cardiovascular disease is associated with up-regulation of miR-206, and this miRNA has significantly higher expression levels in smokers than in nonsmokers. Furthermore, two miRNAs that are frequently down-regulated in association with cigarette smoke (i.e., miR-21 and miR-146a) have lower expression levels in individuals with type 2 diabetes compared with healthy controls. Therefore, these miRNAs could support the observation that smoking is an independent risk factor for type 2 diabetes. Human studies concerning smoking-induced changes of miRNA expression are summarized in Table 4. Figure 3 is a Venn diagram displaying the common and distinct miRNAs from in vitro, in vivo, and human studies on smoking-induced miRNA alterations. miR-125b and miR-21, identified in in vivo and human studies, respectively, were also reported in in vitro studies. Furthermore, several miRNAs were identified in multiple studies, such as miR-34b and miR-146a.

15Table 1 summarizes miRNAs with altered expression in response to environmental and/or personal exposures reported in at least two independent studies, along with their known roles in disease etiology. miRNAs observed in association with either environmental or personal exposures are often associated with cancer; in particular, breast and lung cancer and leukemia have been frequently reported (Table 1). Furthermore, many aberrations in the cardiovascular system have been reported, such as hypertension, heart failure, myocardial infarct, and atherosclerosis. Exposures such as air pollution and smoking can cause cardiovascular disease and cancer; however, the data shown in Table 1 indicate that the listed miRNAs play a causative role in disease etiology, rather than being merely a marker of exposure.

16Air pollution exposure and miRNA expression. Particulate matter (PM) is a complex mixture of small particles and liquid droplets. Particle pollution is made up of a number of components, including acids, organic chemicals, metals, and soil or dust particles. The size of particles is directly linked to their potential to cause health problems. Although the clinical effects of PM exposure are obvious, the underlying mechanism of disease initiation and progression is less well understood. miRNAs play a pivotal role in maintaining healthy lungs. Because the lungs are an important target site for PM, we suggest that miRNAs could underlie the observed health effects of PM exposure. In vitro studies on air pollution and miRNAs are summarized in Table 5.

17In a cohort study of steel plant workers, examined the effect of PM exposure on miRNA expression. Blood samples were collected at the beginning of the working week (“preexposure”) and at the end of the working week (“postexposure”). PM mass and metal components measured in the plant were correlated with miRNA expression analyses of blood samples. Urinary 8-hydroxy-2´-deoxyguanosine (8-OH-dG) levels were measured as a readout of oxidative stress. Both miR-222 and miR-21 were significantly increased in post- versus preexposure samples, and only miR-21 expression levels were positively correlated with 8-OH-dG. Oxidative stress has been reported to induce miR-21 expression; thus, the association between 8-OH-dG and miR-21 might simply reflect the response of miR-21 to production of reactive oxygen species (ROS) in the blood due to the PM-induced increase in oxidative stress (Table 6).

18The cardiovascular anomalies observed in association with air pollution exposure have often been attributed to the generation of oxidative stress. MiR-21 is up-regulated in response to diesel exhaust particles and metal-rich PM and is highly expressed in the cardiovascular system, where it plays an important role in vascular cell proliferation and apoptosis and disease [reviewed by ]. Therefore, miR-21 expression could be an important mechanistic link explaining the association between air pollution exposure and cardiovascular disease.

19observed distinct miRNA expression profiles in patients with asthma compared with controls after subway exposure. Current epidemiological studies have identified the first miRNAs associated with air pollution exposure, and provide a list of putative biomarkers. Table 6 summarizes the human studies on air pollution and miRNAs. A Venn diagram displays the common and distinct miRNAs from in vitro and human studies on air pollution–induced miRNA alterations (Figure 4). The only miRNAs identified in both in vitro and human studies in association with air pollution exposure are miR-10b and miR-128. Furthermore, miRNAs -9, -21, -143, -155, -222, -223, and -338 were identified in at least two independent studies on air pollution and miRNA.

20Nanoparticles. Nanoparticles are emitted from natural and anthropogenic sources and are produced via nanotechnology. Fast propagation of nanotechnologies into different industries and consumer products is causing exponential growth of nanomaterial production. Hence, increasing amounts of nanoparticles reach occupational settings and the indoor and outdoor environments, thus representing a potentially serious hazard to human health. Nanoparticles are also able to cross cell membranes, and their interactions with biological systems are relatively unknown. Table 7 includes the studies on nanoparticle-induced changes in miRNA expression, all of which were performed in animal models.

21Chemical exposure-induced changes in miRNA. Formaldehyde. Formaldehyde is an air toxic present in the atmosphere due to emission from anthropogenic and biogenic sources. Ninety-five percent of inhaled formaldehyde is absorbed within the respiratory tract. Formaldehyde has been reported to change gene expression patterns in nasal and lung cells. The miRNAs reported to be down-regulated in association with formaldehyde exposure have been reported to be involved in the development of diverse tumors (e.g., breast and gastrointestinal cancer, melanoma) as well as heart failure (Table 1). Given the capability of formaldehyde to pass deep into lung tissue and enter systemic circulation, the link with cardiovascular disease and cancer has been widely discussed [reviewed by ]. Interestingly, miR-181a, one of the miRNAs down-regulated after formaldehyde exposure, was reported to affect the DNA damage response in breast cancer, enabling the identification of aggressive breast tumors based on increased miR-181a expression.

22Endocrine disruptors. Organochlorine pesticides and plasticizing agents are ubiquitous environmental endocrine-disrupting compounds that impact human health. Bisphenol A (BPA) is an industrial plasticizer often used as a coating in food cans and in plastic bottles. Dichlorodiphenyltrichloroethane (DDT) is a well-known organochlorine pesticide. Because DDT is very persistent in the environment, accumulates in fatty tissues, and can travel long distances in the upper atmosphere, residues from historical use remain a current threat to human health.

23DDT and BPA have been reported to interfere with endogenous estrogens and thyroid hormone, leading to aberrations of the reproductive, immune, and central nervous systems. DDT and BPA cross the placental barrier and can induce in utero effects that could lead to detrimental effects later in life.

24reported that prenatal exposure to BPA can alter mammary development and lead to breast cancer in humans. From a clinical perspective, it is interesting that decreased expression of let-7f has been associated with increased breast cancer risk, and treatment of MCF-7 breast cancer cells with BPA resulted in reduced let-7f expression. Furthermore, miR-146a has been proposed to induce an Alzheimer’s disease pathway and is up-regulated after BPA exposure (Table 1). Therefore, the neurodegenerative consequences of BPA exposure could at least partially be attributed to miR-146a. In vitro studies could provide researchers with interesting miRNAs that have potential to be used as biomarkers for chemical exposure.

25Polychlorinated biphenyls (PCBs) were widely used organic chemicals until their production was banned because of environmental concerns. PCBs are stable compounds that bioaccumulate in fatty tissues; they have been reported to cause systemic changes in gene expression, suggesting that miRNA regulation could be involved in this process. reported an association between maternal PCB exposure and fetal toxicity, impaired fetal growth, and pregnancy loss.

262,3,7,8-Tetrachlorodibenzo-p-dioxin (TCDD) has been reported to adversely affect the immune system in rats. In addition, reported that TCDD exposure of pregnant mice affected the immune system of fetuses by suppressing T-cell function. Given the regulatory role miRNAs play in the immune system, it can be expected that miRNAs are important in regulating the detrimental health effects observed after exposure to TCDD and PCBs.

27Arsenic. Environmental exposure to arsenic, especially to trivalent inorganic arsenic (As3+), is a health concern given the high concentrations present in groundwater across the world. Exposure to arsenic has been associated with increased risk of cancer due to genomic instability, and long-term arsenic exposure has been observed to induce peripheral vascular injury. A Venn diagram showing the common and distinct miRNAs from in vitro and human studies on arsenic-induced miRNA alterations is presented in Figure 5. Only miRNA-21 was associated with arsenic exposure in in vitro model systems and in human studies. Three miRNAs were identified by at least two independent studies on arsenic exposure and miRNA expression, namely, miR-26b, miR-181a, and miR-222.

28Aluminum sulfates. Aluminum is the most widely distributed metal in the environment and is extensively used in daily life. Chronic exposure of animals to aluminum is associated with behavioral and neuropathological changes. Epidemiological studies have shown poor performance in cognitive tests and a higher abundance of neurological symptoms in workers occupationally exposed to aluminum.

29Hexahydro-1,3,5-trinitro-s-triazine (RDX). The polynitramine explosive RDX is a heavily used second-generation high explosive, and its use can result in the contamination of soils, sediments, and water. RDX exposure has been reported to be toxic to the neural and immune systems and to increase tumor incidence in several cancers.

30Diethylstilbestrol (DES). The synthetic estrogen DES was prescribed to pregnant women from the 1940s to the 1960s in order to prevent miscarriages; however, DES was later reported to be responsible for increasing breast cancer in the mothers and gynecologic tumor incidence in their exposed daughters.

31Perfluorooctanoic acid (PFOA). Perfluoroalkyl chemicals (PFCs) are highly stable and widely used in industrialized countries. PFCs are both lipophobic and hydrophobic; thus, after absorption they will bind to proteins in serum and liver rather than accumulate in lipids. PFOA is one of the most commonly used PFCs.

32The studies we reviewed on chemical-induced changes in miRNA expression are summarized in Tables 8–10 by type of study: in vitro (Table 8), in vivo (Table 9), and human (Table 10) studies.

33miRNAs are omnipresent in the genome and are important regulators of gene expression in response to intracellular as well as environmental cues. In this review, we examined the response of the miRNA machinery to personal and environmental exposures, including air pollution, cigarette smoking, and chemicals such as endocrine disruptors. miRNAs have been proposed as biomarkers for disease; however, the literature also reveals their potential to be used as biomarkers of environmental exposure.

34In different studies on the same environmental pollutant, overall the identified miRNAs showed similar patterns of expression regulation. In studies where smoking-induced changes were investigated, the general observation was a down-regulation of expression. For example, miR-125b was down-regulated in response to cigarette smoke in both primary human bronchial epithelial cells and mouse lung tissue. However, when unique miRNAs had altered expression patterns in response to different environmental exposures, their direction of regulation could be the same (10/25 miRNAs) or the opposite (15/25 miRNAs; 60%). The different exposures we discussed here have their own unique health effects, so one would not expect them to have the same effect on the miRNA machinery. However, there is sometimes a discrepancy when looking at the same exposure indicator; for example, in response to smoking, miR-21 has been reported to be up-regulated in some studies and down-regulated in others (Table 4). Part of the discrepancy can be explained by the different exposure models that were used.

35In general, different in vitro studies show little overlap, potentially because of the complex miRNA–mRNA networks that underlie the observations and the differences in exposure used across studies. In studies of the same environmental pollutant in vitro, in vivo, or in humans, identified miRNAs were quite distinct (Figures 3–5). This can be explained in part by the observation that animal models do not always reflect genomic responses that occur in humans. Discrepancy between different studies might also stem from differences in exposure duration. For example, in a study in rats, the duration of exposure uniquely influenced expression patterns of the individual miRNAs.

36Human epidemiological studies are necessary to observe exposure-related effects on miRNAs. Understanding the exposome requires putting together pieces of a complex puzzle. Epidemiological studies need input from experimental studies to identify good candidate biomarkers, and results from epidemiological studies often need follow-up by experimental studies to investigate mechanisms of action and to study tissue dependency of effects because human studies are most often performed in easily accessible tissues such as blood and saliva as a surrogate for the actual target tissues.

37Currently, epidemiological studies on microRNA often involve free or exosomal miRNAs present in saliva or other body fluids. However, it is not clear whether these observed miRNA changes are a true reflection of the body’s response and can really predict health effects. In blood, miRNAs within the exosomes have been shown to overlap with cellular miRNA profiles: observed that exosomes derived from blood were enriched for miRNAs and that miRNA profiles between blood cells and the cell-free exosomal fraction showed important overlap.

38Because miRNAs can regulate mRNA expression in both a negative manner and a positive manner and because many miRNAs can bind the same mRNA, it is difficult to draw conclusions from miRNA studies without infomation on the concurrent mRNA(s) expression pattern. However, this information is rare in current reports on epidemiological studies of miRNAs. The findings of this review underscore the complex networks that are built by miRNAs and the mRNAs they regulate because one miRNA can influence many mRNAs according to the timing and pattern of expression.

39Many of the reviewed studies used large-scale microarray profiling, but follow-up and validation with more quantitative approaches often lags behind. This delay is understandable because of the cost and labor intensity inherent to these techniques; however, it is important to confirm the miRNAs that are responsive to environmental exposures.

40Researchers are currently publishing extensive lists of miRNAs that are responsive to environmental exposures and showing their utility as biomarkers of effect. Future research should focus on identifying the molecular mechanism behind miRNA expression changes in response to exposure to determine whether the changes in miRNA expression are merely a symptom of the (patho)physiological processes the organism undergoes after exposure, or whether miRNAs are the drivers responsible for these changes. recently reviewed the putative mechanisms of action behind miRNAs’ response to environmental exposure. However, the effect of the identified miRNAs on putative mRNA targets should also be studied to determine whether the change in miRNA expression has functional consequences and which mRNAs are true miRNA targets under the given circumstances.

41At present, little is known about whether environmental exposures induce long-term changes in human miRNA expression or whether these have a transient character. To address this problem, more longitudinal studies should be conducted to examine the long-term effects of exposure. Results from animal studies suggest that miRNA expression changes in response to formaldehyde exposure are transient and revert to normal levels after recovery from exposure, but reported that miRNA profiles in target organs did not recover 1 week after cessation of long-term cigarette smoke exposure. In a study in humans, observed that miRNA expression profiles of individuals who quit smoking resembled those of nonsmokers.

42Follow-up in future generations is necessary to determine the heritability of the miRNA expression changes. It would be very interesting to examine the effect of in utero environmental exposures on development of disease in later life and the role miRNAs play in inducing these health effects. Furthermore, long-term longitudinal studies would allow us to distinguish between cause and effect of miRNA expression and environmental exposure, and would also allow us to estimate the contribution of miRNAs to disease development. Studies have shown that miRNAs can be used as biomarkers of disease as well as biomarkers for environmental exposure and that miRNAs hold great potential to explain disease etiology.