The effects of microRNA on the absorption, distribution, metabolism and excretion of drugs

Abstract

The importance of genetic factors (e.g. sequence variation) in the absorption, distribution, metabolism, excretion (ADME) and overall efficacy of therapeutic agents is well established. Our ability to identify, interpret and utilize these factors is the subject of much clinical investigation and therapeutic development. However, drug ADME and efficacy are also heavily influenced by epigenetic factors such as DNA/histone methylation and non-coding RNAs [especially microRNAs (miRNAs)]. Results from studies using tools, such as in silico miRNA target prediction, in vitro functional assays, nucleic acid profiling/sequencing and high-throughput proteomics, are rapidly expanding our knowledge of these factors and their effects on drug metabolism. Although these studies reveal a complex regulation of drug ADME, an increased understanding of the molecular interplay between the genome, epigenome and transcriptome has the potential to provide practically useful strategies to facilitate drug development, optimize therapeutic efficacy, circumvent adverse effects, yield novel diagnostics and ultimately become an integral component of personalized medicine.

Linked Articles

This article is part of a themed section on Epigenetics and Therapy. To view the other articles in this section visit http://dx.doi.org/10.1111/bph.2015.172.issue-11

Tables of Links

TARGETS
Enzymesa	Transportersc
CYP1A1	ABCA1
CYP1B1	ABCB1, MDR1/P-gp
CYP2E1	ABCC1
CYP3A4	ABCC2
DNMT1, DNA methyltransferase 1	ABCC3
DNMT3A, DNA methyltransferase 3α	ABCC4
HDAC	ABCC5
Histone methyltransferase EZH2	ABCC10
Nuclear hormone receptorsb	ABCE1
ESR1, oestrogen receptor α, NR3A1	ABCG2
Glucocorticoid receptor, NR3C1	Cystine/glutamate exchanger, SLC7A11
LXRA, liver X receptor α, NR1H3	SERT, 5-HT transporter, SLC6A4
PPARα, NR1C1	SLC15
PXR, pregnane X receptor	vGLUT1, vesicular glutamate transporter 1, SLC17A7
RXR, retinoid receptors

LIGANDS
17β-oestradiol	Raloxifene
Acetaminophen, paracetamol	Tamoxifen
Cholesterol	Testosterone
Cisplatin	Topotecan
Cocaine	Vinblastine
Dexamethasone
Dox, doxorubicin
Fluoxetine
Gemcitabine
GSH, glutathione
Halothane
Isoflurane
Methadone
Mitoxantrone

These Tables list key protein targets and ligands in this article which are hyperlinked to corresponding entries in http://www.guidetopharmacology.org, the common portal for data from the IUPHAR/BPS Guide to PHARMACOLOGY (Pawson et al., 2014) and are permanently archived in the Concise Guide to PHARMACOLOGY 2013/14 (^a,b,cAlexander et al., 2013a,b,c).

The proper control of absorption, distribution, metabolism and excretion (ADME) of xenobiotics is essential for living organisms to obtain energy, acquire necessary building blocks (e.g. essential amino acids) and maintain homeostasis in a complex chemical environment. Genes involved in ADME activities encode various receptor/transporters, biotransformation enzymes and accessory proteins (PharmaADME, http://pharmaadme.org/joomla/). These proteins include membrane transporters responsible for the absorption and excretion of specific molecules and enzymes to convert xenobiotics for excretion. To date, over 300 transporters and enzymes directly involved in ADME process have been described. This long list of components makes the study of ADME inherently complex, as transporters and enzymes work in concert to respond dynamically to diverse external factors. Despite the formidable complexity of the field, an understanding of ADME is critical for drug development in order to increase therapeutic efficacy and reduce adverse effects (Caldwell et al., 2009; Emoto et al., 2010; Bell and Wang, 2012).

It is well known that there are considerable inter-individual variations in response to drug treatment (Alomar, 2014). Besides polymorphisms within the sequence of therapeutic targets, genetic variations in ADME-related genes have been shown to affect therapeutic response through drug transportation and metabolism (Nakajima and Aoyama, 2000; Evrard and Mbatchi, 2012). However, genetic variations cannot completely explain the differences observed between individuals. Recent studies using high-throughput global molecular profiling technologies have shown that ADME activities have also been affected by a number of epigenetic factors, including DNA methylation, post-translational histone modifications and non-coding regulatory RNAs, which directly or indirectly alter the expression of ADME genes (Ivanov et al., 2012). For example, microRNAs (miRNAs) are short, non-coding regulatory RNAs involved in modulating ADME-related gene expression and protein translation levels (Liu and Pan, 2014). The study of epigenetic factors on drug metabolism, pharmacoepigenetics, is a rapidly growing field and plays an important role in therapeutic efficacy by affecting the pharmacokinetic (PK) and pharmacodynamic properties of drugs. In this review, we summarize the current understanding on the involvement of miRNAs in ADME.

MicroRNA

More than 60% of human genomic DNA is transcribed into RNA, yet protein-coding genes account for only less than 3% of the genome (Bernstein et al., 2012; Kellis et al., 2014). Non-coding RNAs (ncRNAs) are RNAs that are not translated into proteins. Some ncRNAs, such as ribosomal RNAs and transfer RNAs, have well-established cellular roles. However, there are other classes of ncRNAs that have been recently discovered (Esteller, 2011). Among them, miRNAs are probably the best studied due to their important roles in the post-transcriptional regulation of messenger RNAs (mRNAs). miRNAs were originally discovered in the nematode Caenorhabditis elegans as mediators of temporal pattern formation (Ambros, 2001; Lagos-Quintana et al., 2001). These small regulatory RNAs were found to be conserved in metazoan species and have also been observed in several virus families (Pfeffer et al., 2004). miRNAs are involved in a broad range of cellular activities, including proliferation, development, homeostasis, immunity, metabolism and apoptosis (Kloosterman and Plasterk, 2006; Stefani and Slack, 2008). Because of their important regulatory roles in these diverse biological processes, miRNAs display characteristic expression patterns in cells and tissues, different developmental stages and disease processes.

miRNA genes are found in diverse locations throughout the genome (Rodriguez et al., 2004) and often cluster together (Lagos-Quintana et al., 2001; Seitz et al., 2004). Most of the sequences that encode miRNAs are transcribed by RNA polymerase II (Lee et al., 2004). Many miRNAs may be transcribed on the same primary transcript (pri-miRNA). This pri-miRNA is then processed into a smaller hairpin structure (pre-miRNA) by the microprocessor complex, RNase III enzyme Drosha/DGCR8. Following nuclear export via the RanGTP-dependent double-strand RNA binding protein exportin-5, the pre-RNA is then cleaved by another RNase III enzyme, Dicer, to yield an approximately 22 nucleotide-long RNA duplex (Kim et al., 2009). Unwinding by an RNA helicase allows a mature miRNA strand to be loaded onto the protein Argonaute to form the RNA-induced silencing complex (RISC). The RISC typically interacts with the 3’ untranslated region (UTR) of mRNA targets, resulting in the translational repression or degradation of the transcript. Unlike other regulatory molecules, miRNA target recognition is relatively flexible, as the miRNA sequence itself is short and the interaction between miRNA and mRNA is based upon partial sequence complementarity (Bartel, 2009). This permits a single miRNA to interact with many mRNA targets. Additionally, a single mRNA can also contain many different miRNA binding sites. It has been estimated that the majority of the human mRNAs are targeted by one or more miRNAs (Friedman et al., 2009). These loose constraints governing miRNA–mRNA interactions increase the diversity of possible interactions and make miRNA-mediated regulatory networks often difficult to unravel.

At present, over 2500 human mature miRNAs have been deposited into the online repository miRBase (v20, www.mirbase.org) (Griffiths-Jones et al., 2008). To decipher the interactions between miRNA and mRNA, a number of informatic tools, such as TargetScan (www.targetscan.org) (Lewis et al., 2005) and miRanda (www.microrna.org) (Enright et al., 2003), have been developed to predict miRNA–mRNA interactions based upon sequence complementation, binding stability and degree of sequence conservation. A typical approach to evaluate the specificity of miRNA–target interactions is the use of gene reporter assays in cell culture. A specific miRNA of interest is transfected into a cell line expressing a transgenic construct containing a reporter gene (e.g. luciferase) and the 3′-UTR region of a putative mRNA target. However, this approach relies upon overexpressed components and may ultimately have limited physiological relevance. The strongest support for in vivo relevance is derived from a combination of experiential methods (reviewed in Thomson et al., 2011). Although thousands of miRNA–mRNA interactions have been predicted, few mRNA–miRNA interactions have been empirically validated. These validated interactions are archived in online databases, such as miRTarBase (http://mirtarbase.mbc.nctu.edu.tw/), Tarbase (http://diana.imis.athena-innovation.gr/DianaTools/index.php?r=tarbase/index) and miRecords (http://mirecords.biolead.org/).

MicroRNAs regulate the expression of ADME phase I and II enzymes

Increasing evidence has indicated that a substantial number of transporters and drug metabolizing enzymes are regulated by miRNAs (Yu, 2007; 2009,) (Table 1 and Figure 1). We collected all miRNA–ADME interactions from literature and constructed a miRNA-mediated regulatory network for ADME (Figure 1). The network contains 189 interactions among the 381 nodes, which includes 120 miRNAs and 261 ADME genes.

Table 1.

List of interactions between miRNA and drug metabolism-related genes

Gene ID	Function	Interacting miRNA	References
CYP1A1	Phase I enzyme	miR-18b, -20b	Wang et al., 2009; Glubb and Innocenti, 2011
CYP1B1	Phase I enzyme	miR-27b	Tsuchiya et al., 2006; Chuturgoon et al., 2014
CYP2A3	Phase I enzyme	miR-126*, -34	Kalscheuer et al., 2008
CYP2E1	Phase I enzyme	miR-378	Mohri et al., 2010; Takahashi et al., 2014
CYP2J2	Phase I enzyme	Let-7b	F Chen et al., 2012a
CYP3A4	Phase I enzyme	miR-27b, -577, -1, -532-3p, -627, -223, -148	Takagi et al., 2008; Pan et al., 2009a; Takahashi et al., 2014; Wei et al., 2014
CYP7A1	Phase I enzyme	miR-122a, miR-422a	Song et al., 2010
CYP19A1	Phase I enzyme	miR-19b, -106a, let-7f	Shibahara et al., 2012; Kumar et al., 2013
CYP24A1	Phase I enzyme	miR-125b	Komagata et al., 2009
GSS	Phase I enzyme	miR-125b	Tili et al., 2012
GPX7	Phase I enzyme	miR-122	Akinc et al., 2008
SULF1	Phase I Enzyme	miR-516a	Takei et al., 2011
GSTP1	Phase II enzyme	miR-133a, -513a-3p	Zhang et al., 2012; Uchida et al., 2013
SULT1A1	Phase II enzyme	miR-631	Yu et al., 2010
UGT1A	Phase II enzyme	miR-491-3p	Dluzen et al., 2014
ABCA1	Transporter	miR-33, -33*, -758, -106b, -613, -27a, -27b, -145, -101, -135b	Ramirez et al., 2011; Borel et al., 2012; Kim et al., 2012; Goedeke et al., 2013; Kang et al., 2013; Rottiers et al., 2013; Zhang et al., 2014; Zhao et al., 2014
ABCB1/P-gp/MDR1	Transporter	miR-451, -27a, -298, -354, -7, -200c, -19a/b	Kovalchuk et al., 2008; Zhu et al., 2008; Pogribny et al., 2010; Bitarte et al., 2011; Bao et al., 2012; J Chen et al., 2012b; F Wang et al., 2013a
ABCB11	Transporter	miR-33	Allen et al., 2012
ABCC1/MRP-1	Transporter	miR-134, -326, -1291, -199a/b, -296	Guo et al., 2010; Liang et al., 2010; Borel et al., 2012; Pan et al., 2013
ABCC2	Transporter	miR-379, let-7c	Haenisch et al., 2011; Zhan et al., 2013
ABCC3	Transporter	miR-9*	Jeon et al., 2011
ABCC4/MRP4	Transporter	miR-124a, -506, -125a/b	Borel et al., 2012; Markova and Kroetz, 2014
ABCC5	Transporter	miR-128	Zhu et al., 2011
ABCC6	Transporter	miR-9*	Jeon et al., 2011
ABCC7/CFTR	Transporter	miR-145, -494, -1246, -509-3p	Gillen et al., 2011; Ramachandran et al., 2013
ABCC10	Transporter	Let-7a/e	Borel et al., 2012
ABCE1	Transporter	miR-124, -203, -26a, -135b, -145	Furuta et al., 2010; Borel et al., 2012
ABCG1	Transporter	miR-33	Marquart et al., 2010; Rayner et al., 2010
ABCG2/BCRP	Transporter	miR-520h, -519c, -328, -487a, -181a	Liao et al., 2008; Pan et al., 2009b; To et al., 2009; Wang et al., 2010; Li et al., 2011; Padmanabhan et al., 2012; Jiao et al., 2013; Ma et al., 2013
SLC6A4	Transporter	miR-16, -15a	Baudry et al., 2010; Tamarapu Parthasarathy et al., 2012; Moya et al., 2013
SLC7A5	Transporter	miR-126	Miko et al., 2011
SLC7A11	Transporter	miR-27a	Drayton et al., 2014
SLC15A1/PEPT1	Transporter	miR-92a	Dalmasso et al., 2011b
SLC12A2	Transporter	miR-384, -494, -1246	Gillen et al., 2011
SLC16A1/MCT1	Transporter	miR-124, -29a/b	KK Li et al., 2009b; Pullen et al., 2011
NR1I2/PX receptor	Nuclear Receptor	miR-148a	Takagi et al., 2008
NR3C1/GR	Nuclear Receptor	miR-18, -124a	Vreugdenhil et al., 2009
NR2B1/RXRA	Nuclear Receptor	miR-27a, -27b	Ji et al., 2009; Komagata et al., 2009; Mohri et al., 2009
NR1I1/VD receptor	Nuclear Receptor	miR-27b, -125b	Mohri et al., 2009; Pan et al., 2009a
NR1C1	Nuclear Receptor	miR-10b, -506, -21, -27b	Zheng et al., 2010; Kida et al., 2011; Tong et al., 2011; Zhou et al., 2011
NR1C3	Nuclear Receptor	miR-130, -27b, -27a	Jennewein et al., 2010; Kim et al., 2010; Lee et al., 2012
NR1C2	Nuclear Receptor	miR-15a, -199a, -214, -9	Yin et al., 2010; el Azzouzi et al., 2013; Thulin et al., 2013
NR1H3/LXRA	Nuclear Receptor	miR-613, -206, -1	Ou et al., 2011; Zhong et al., 2013; Vinod et al., 2014
NR2A1/HNF4A	Nuclear Receptor	miR-24, -34a	Takagi et al., 2010
NR3A1	Nuclear Receptor	miR-221, -222, let-7a, miR-130a, -22, -206	Iorio et al., 2005; Adams et al., 2007; Zhao et al., 2008; 2011b,; Pandey and Picard, 2009; Xiong et al., 2010; Tang et al., 2011
RXRA	Modifier	miR-27a/b	Ji et al., 2009

Figure 1.

A hypothetical network view on the interactions between miRNA and ADME transcripts. Molecular interactions used were obtained from KEGG pathway and literature. The network contains 189 interactions between miRNA and its target genes, and 381 nodes including120 miRNAs (squares) and 261 ADME genes (circles). Different colours of circles represent different categories of ADME genes as indicated. The visualization of the network was carried our by Cytoscape.

The biotransformation of xenobiotics is generally separated into two phases: phase I includes the addition or exposure of chemically reactive functional groups such as amine (–NH₂), hydroxyl (–OH), sulfhydryl (–SH) or carboxyl (–COOH) group by enzymes, in preparation for phase II, where the phase I products are further modified with water-soluble or charged chemical motif(s), converting them into water-soluble structures for excretion. Cytochrome P450 (CYP450, a class of monooxygenases) is the largest group of phase I enzymes. In humans, there are 57 functional CYP450 genes, which can be grouped into 18 different families based upon sequence similarity (Lewis, 2004). A number of miRNAs have been shown to be involved in the post-transcriptional regulation of cytochrome P450 genes (Table 1 and Figure 1). For example, the expression level of CYP1A1 was correlated with the levels of miR-18b and miR-20b (Wang et al., 2009; Glubb and Innocenti, 2011). CYP1B1 and CYP3A4 were found to be regulated by miR-27b (Tsuchiya et al., 2006; Pan et al., 2009a), and CYP2E1 by miR-378 (Mohri et al., 2010).

A few phase I enzymes have gained more attention due to their involvement in metabolizing important therapeutic agents. CYP3A4, a member of the CYP3 family predominantly expressed in liver and intestine, metabolizes more than 50% of the therapeutic drugs currently on the market (Plant and Gibson, 2003). It is also involved in the synthesis and metabolism of a number of endogenous molecules, including cholesterol, steroid hormones (e.g. testosterone), bile acids and vitamin D (Wang et al., 2013c). Because of its important role in drug metabolism, changes in CYP3A4 activity (either through altered gene expression or inhibition of enzyme activity) cause a number of adverse effects associated with drug usage. CYP3A4 can be induced by ligands that bind and activate the pregnane X (PX) receptor (Lehmann et al., 1998). Experimental data also indicate that the level of CYP3A4 is affected by epigenetic factors, including histone acetylation and methylation (Dannenberg and Edenberg, 2006; Kacevska et al., 2012) and also miRNAs.

A recent study based upon several different miRNA target prediction programmes identified 105 miRNAs that may potentially interact with CYP3A4 mRNA (Wei et al., 2014). Among the 105 miRNA candidates, 14 of them were experimentally verified to interact directly with the 3′-UTR of CYP3A4 transcript in HEK293T cells. Analysis of the miRNAs identified in vitro and CYP3A4 transcript and protein levels in human liver samples suggested that four of these miRNAs (miR-1, -532-3p, -577 and -627) attenuate the translation of CYP3A4 in vivo. Besides showing that miRNAs affect both transcript and protein levels in vivo, results from this study also illustrated our limited understanding of how miRNAs select their targets in cells, since only about 10% of the predicted interactions could be verified experimentally.

CYP2E1 is involved in the metabolism of anaesthetics, such as isoflurane and halothane, organic solvents, such as ethanol and benzene, acetaminophen (paracetamol) and other compounds. Computational analysis predicted a potential miR-378 binding site within the 3′-UTR of CYP2E1 (Mohri et al., 2010), the functionality of which was verified by in vitro reporter assays. In addition, an inverse correlation between CYP2E1 protein levels and miR-378 abundances was observed in a panel of 25 human liver specimens, providing further support for the possible significance of this interaction in vivo (Mohri et al., 2010).

CYP1B1 is overexpressed in diverse cancers (e.g. breast, colon, lung, oesophagus, skin, lymph node, brain and testis) relative to their corresponding normal tissues (Murray et al., 1997) and is known to activate various pro-carcinogens (Shimada et al., 1996). In addition, CYP1B1 enhances the conversion of 17β-oestradiol into 4-hydroxyestradiol (Han and Liehr, 1994; Newbold and Liehr, 2000), a metabolite known to cause DNA damage through free radical production and which may promote the development of oestrogen-dependent cancers. Examination of the 3′-UTR of CYP1B1 transcript revealed a high-quality target site for miR-27b (Tsuchiya et al., 2006), which was confirmed using luciferase reporter assays. Ablation of miR-27b targeting via an antisense oligonucleotide increased the abundance and activity of endogenous CYP1B1 in MCF-7 cells. In addition, the amount of miR-27b was inversely correlated with CYP1B1 protein level in most of the oestrogen receptor/progesterone receptor positive breast cancer samples examined (Tsuchiya et al., 2006).

GSH is synthesized from glutamylcysteine and glycine by glutathione synthetase (GSS) in an ATP-dependent process (Snoke and Bloch, 1955). GSH has many cellular functions, including the scavenging of peroxides and free radical species, and the detoxification of xenobiotics (Pastore et al., 2003). Overexpression of miR-125b in chronic lymphocytic leukaemia-derived cell lines reduced GSS protein abundance and led to a decrease of GSH activity (Tili et al., 2012). Using a luciferase reporter assay, miR-125b was verified to directly interact with the 3′-UTR of the GSS transcript.

Glutathione peroxidase 7 (GPX7) counters oxidative stress generated from polyunsaturated fatty acid metabolism (Utomo et al., 2004) and can neutralize hydrogen peroxide in the absence of glutathione (Peng et al., 2012). Knockdown of GPX7 expression resulted in an increase in reactive oxygen species, DNA damage and apoptosis in oesophageal squamous epithelial cells upon exposure to bile acids in vitro. GPX7 level was found to be up-regulated in the livers of mice treated with antagomiR-122 (a miR-122 antagonist). This suggests a possible interaction between miR-122 and GPX7 transcript (Krutzfeldt et al., 2005; Akinc et al., 2008).

UDP-glucuronosyltransferases (UGTs) are responsible for the phase II processing of many endogenous (e.g. bilirubin, steroid hormones) (Bosma et al., 1994; Belanger et al., 2003) and exogenous (e.g. drugs, chemotherapeutics and carcinogens) compounds (recent review in Rowland et al., 2013). This family of enzymes is composed of two subfamilies (UGT1A and 2B) and catalyses the addition of carbohydrate groups (e.g. glucuronide) to its substrates for excretion (Mackenzie et al., 2005; Nagar and Remmel, 2006). The UGT1A family has nine alternatively spliced protein isoforms sharing a common 3′-UTR. In silico analysis of the UGT1A 3′-UTR identified a potential miR-491-3p target sequence (Dluzen et al., 2014). Introducing a miR-491-3p mimic into HuH-7 cells reduced UGT1A1 mRNA abundance by 48% relative to non-targeting control mimics and direct interaction between miR-491-3p and UGT1A1 3′-UTR was also confirmed by reporter assays. This reduction in UGT1A1 mRNA was complemented with a significant reduction in UGTA1 enzymatic activity, as measured by the conversion of raloxifene into glucuronidated metabolites (raloxifene-6-glucuronide and -4′-glucuronide). Knockdown of the endogenous miR-491-3p level in HepG2 cells increased the abundance of these metabolites, although without reducing UGT1A1 mRNA levels (Dluzen et al., 2014). An inverse correlation between miR-491-3p abundance and the quantity of UGTA3 and A6 isoforms (sharing the same 3′-UTR as UGT1A1) was also observed in samples of normal human liver (Dluzen et al., 2014).

MicroRNAs regulate the expression of transporter proteins

miRNAs have also been observed to regulate two major classes of drug transporters: ATP-binding cassette (ABC) and solute carrier (SLC) proteins (Table 1 and Figure 1). As these membrane transporters facilitate the uptake and excretion of many endo- and xenobiotics, alterations in their levels may affect the PK properties of drugs.

A recent study reported that the levels of a number of ABC transporters were increased in hepatocellular carcinoma and that the increase is associated with the decrease of several miRNAs (Borel et al., 2012). Using luciferase reporter assays, Borel et al. confirmed the interactions of some ABC transporter transcripts and miRNAs including ABCA1 (interaction with miR-101 and miR-135b), ABCC1 (miR-199a/b and miR-296), ABCC4 (miR-125a/b), ABCC5 (miR-101, -125a and let-7a), ABCC10 (let-7a/e) and ABCE1 (miR-26a, -135b and -145) (Borel et al., 2012). In addition to miR-101 and miR-135b, experimental data also suggest an interaction between ABCA1 and miR-33, since administration of a locked nucleic acid miRNA antagonist (antimiR) of miR-33 into a non-human primate model resulted in functional de-repression of the ABCA1 level. This finding demonstrates the possibility of modulating the interaction between specific miRNA and mRNA using antimiR, and provides proof of principle for this antimiR-based therapeutic modality (Rottiers et al., 2013).

Besides miR-199a/b and miR-296, ABCC1 transcript is also targeted by miR-326. Overexpressing miR-326 reversed the multidrug resistance phenotype, sensitizing the drug-resistant MCF-7/VP cells to doxorubicin (DOX) (Liang et al., 2010). The ABCC1 transcript was also observed to be targeted by miR-134 and the ABCC1 protein level is decreased by the increased level of miR-134 in drug-resistant H69AR cells (Guo et al., 2010).

Other miRNA–ABC transporter interactions have also been reported. For example, miR-379 interacts with the 3′-UTR of ABCC2 and suppresses ABCC2 transcript level in HepG2 cells (Haenisch et al., 2011). miR-298 overexpression in vitro down-regulated the level of ABCB1, also known as the drug transporter multidrug resistance protein 1/P-glycoprotein (MDR1/P-gp), which leads to breast cancer cell sensitivity to DOX (i.e. decreased the efflux of DOX from cells). miR-298 was found to directly interact with 3′-UTR of ABCB1 transcript (Bao et al., 2011). MiR-451 and miR-27a have also been observed to suppress the expression of ABCB1 and sensitize breast cancer cells to DOX (Kovalchuk et al., 2008). In a display of the complexity of miRNA–mRNA regulatory networks, overexpression of miR-19a/b has been shown to increase the abundance of the ABCB1 transcript, which is likely to be mediated through indirect interactions with other gene expression regulatory factors (Wang et al., 2013a).

Cancer cells can become resistant to chemotherapy by evading miRNA-based control of ADME genes. For example, ABCG2 plays an important role in cellular transport of anti-cancer drugs, such as mitoxantrone, doxorubicin and topotecan, and has been observed to acquire a truncated 3′-UTR in drug-resistant cells relative to the parental cell line (To et al., 2008). The truncated mRNA omits a miR-519c binding site, which mediates a decrease in ABCG2 expression level by miR-519c. This adaptive truncation of the 3′-UTR was later found to be present in several ABCG2-overexpressing cell lines (To et al., 2009). In addition to miR-519c, other miRNAs such as miR-520h and miR-328 have also been found to interact with ABCG2 transcript. The levels of miR-520h and miR-328 are lower in cells expressing ABCG2 relative to ABCG2-negative cells (Wang et al., 2010). Results from a comparative study indicated stronger inhibition of ABCG2 expression in human breast cancer cells by miR-519c and miR-328 than miR-520h (Li et al., 2011).

miRNAs have also been found to affect the absorption and excretion of small molecules through SLC proteins. Folate, nucleoside and amino acid transporters all belong to the SLC transporter family and mediate the uptake of hydrophilic drugs such as gemcitabine and other nucleoside analogues (Candelaria et al., 2010). These levels of membrane transporters are also regulated by miRNA. For example, the expression of SLC15 was found to inversely correlate with miR-92b, which was then demonstrated to reduce SLC15 mRNA, protein and transport activity levels in Caco2-BBE cells (Dalmasso et al., 2011b). Another SLC transporter, serotonin transporter 1 (SERT1, SLC6A4), has been found to be suppressed by miR-16 (Baudry et al., 2010). In silico target prediction identified miR-16 as a potential regulator of SLC6A4. Overexpression of miR-16 in 1C11 cells reduced the SLC6A4 level. Reduction of miR-16 by an anti-miR-16 oligonucleotide resulted in an increase of SLC6A4 level. This interaction has also been demonstrated in vivo: in mouse studies, administration of the selective serotonin reuptake inhibitor fluoxetine was found to increase miR-16 abundance in serotonergic raphe nuclei with a reduced SLC6A4 level (Baudry et al., 2010).

miRNA profiling data identified miR-27a as down-regulated in cisplatin-resistant bladder cancer cell lines, relative to the cisplatin-sensitive lines from which they were derived (Drayton et al., 2014). MiR-27a was found to directly target the cystine/glutamate exchanger – SLC7A11, which forms a heterodimer with SLC3A2 to assemble the x_c-cystine-glutamate transporter. This protein complex imports cysteine (as cystine) for the synthesis of GSH, which enables cisplatin detoxification. miR-27a expression was found to inversely correlate with SLC7A11 level in patient samples. Low miR-27a and high SLC7A11 were found to correlate with poor patient prognosis.

MicroRNAs regulate the expression of nuclear receptors

Nuclear receptors play important roles in cellular responses towards environmental stimulation by activating or inactivating the expression of genes, including those that encode drug transporters and biotransformation enzymes. For instance, the retinoid X receptor (RXR) heterodimerizes with the steroid family of orphan nuclear receptors, the constitutive androstane receptor (CAR) or the pregnane X receptor (PXR) to participate in the xenobiotic-mediated transcriptional activation of CYP2B and CYP3A. This type of interaction increases the complexity of the role of miRNAs in regulating ADME activities, since miRNAs may affect the expression of ADME genes via indirect targeting the 3′-UTRs of nuclear receptors (Table 1 and Figure 1). For example, the interactions between miR-27b and miR-125b with the vitamin D (VD) receptor/RXRA alter CYP3A4 expression (Komagata et al., 2009; Mohri et al., 2009). The level of CYP3A4 is also affected by miR-148 through its interaction with PXRs (Takagi et al., 2008).

The PPARα (NR1C1) regulates the expression of a number of ADME genes and its abundance is affected by the levels of miR-21or miR-27b (Kida et al., 2011). An independent study showed that the overexpression of miR-506 also suppresses NR1C1expression, which leads to hydroxycamptothecin resistance in a colon cancer cell line (Tong et al., 2011). Another miRNA-regulated nuclear receptor is liver X receptor α (LXRA/NR1H3), which is closely related to RXR and PPAR, and can be targeted by miR-613. The activation of LXRA resulted in an increased level of miR-613 (Ou et al., 2011). This type of feedback loop probably plays an important part in maintaining proper ADME activity, which is critical for endo- and xenobiotic transport and metabolism.

The oestrogen receptor α (ESR1) and the glucocorticoid receptor (NR3C1) are known to affect the expression of CYP3A4 and ABCG2. MiR-221 and -222 inhibit ESR1 expression. Overexpression of miR-221/222 in a breast cancer cell line resulted in the cells becoming resistant to tamoxifen (an antagonist of the oestrogen receptor) treatment, while knocking down the level of miR-221/222 sensitized the cells (Zhao et al., 2008). A similar effect was also observed with let-7b and let-7i (Y Xhao et al., 2011b). The ESR1 transcript has also been observed to be targeted by miR-206, and ESR1 mRNA abundance negatively correlates with the miR-206 level (Iorio et al., 2005; Adams et al., 2007). To further understand the interaction of miRNA and ESR1 transcript, Picard et al. performed a study on 14 miRNAs, which might be involved in ESR1expression and found that miR-22 exerted the strongest inhibition of oestrogen signalling through targeting ESR mRNA (Pandey and Picard, 2009).

Effects of epigenetic factors on miRNA expression

Even though numerous studies showed the effects of specific miRNAs on ADME-related transcripts, the levels of individual miRNA sequences are affected by intrinsic and extrinsic factors. Like other transcripts, the expression of miRNAs can be affected by epigenetic factors, such as DNA methylation. For example, rapid changes to miRNA levels have been observed upon treating cells with histone deacetylase (HDAC) inhibitors (Scott et al., 2006). Alterations to miRNA promoter methylation status are associated with disease conditions. For example, the promoter of the tumour suppressor miR-125b is hyper-methylated in human breast cancer, leading to a lower level of miR-125b and the de-repression of its target transcription factor, ETS1 (V-Ets avian erythroblastosis virus E26 oncogene homologue 1) (Zhang et al., 2011). In addition, the promoter of the metastasis suppressor miR-355 is also hyper-methylated in breast cancer (Png et al., 2011). Hypo-methylation has also been observed to alter miRNA expression, as let-7a-3 has been found to be heavily methylated in normal human tissue, but hypo-methylated in some human lung cancers (Brueckner et al., 2007).

In a display of further complexity, miRNAs can reciprocally influence epigenetic states by targeting transcripts that encode enzymes involved in DNA methylation and histone modifications (Table 2). Several HDACs have been observed by reporter assays to be regulated by different miRNAs (Scott et al., 2006; H Li et al., 2009a; Sato et al., 2011; Wang et al., 2013b). In addition, some polycomb proteins that are involved in the remodelling of chromatin structure are also regulated by miRNAs. For example, the expression of the histone methyltransferase EZH2 is repressed by miR-101 (Varambally et al., 2008).

Table 2.

List of microRNAs affecting enzymes involved in epigenetic modifications

Gene ID	Gene description	Interacting miRNA	References
DNMT1	DNA (cytosine-5-)-methyltransferase 1	miR-126, -152, -185, -148, -140, -342	Wang et al., 2011; S Zhao et al., 2011a; Takata et al., 2013; Azizi et al., 2014; Xiang et al., 2014
DNMT3A	DNA (cytosine-5-)-methyltransferase 3 alpha	miR-29b, -143, -199a, -370	Fabbri et al., 2007; Garzon et al., 2009; Ng et al., 2009; 2014,; Qi et al., 2013; Chen et al., 2014
DNMT3B	DNA (cytosine-5-)-methyltransferase 3 beta	miR-29, -148, -495	Fabbri et al., 2007; Duursma et al., 2008; Garzon et al., 2009; Yang et al., 2014
HDAC1	Histone deacetylase 1	miR-34, -449, -520h	Noonan et al., 2009; Zhao et al., 2013; Shen et al., 2014
HDAC2	Histone deacetylase 2	miR-145	Noh et al., 2013
HDAC4	Histone deacetylase 4	miR-1, -155, -365	Chen et al., 2006; Guan et al., 2011; Sandhu et al., 2012
HDAC5	Histone deacetylase 5	miR-2861	H Li et al., 2009a
HDAC6	Histone deacetylase 6	miR-433	Simon et al., 2010
EZH2	Enhancer of zeste homologue 2	miR-26a, -26b, -101, -214, -124, -181a, -210, -424, -138	Sander et al., 2008; Varambally et al., 2008; Juan et al., 2009; Zhang et al., 2013; Overhoff et al., 2014; Xie et al., 2014
BMI1	B lymphoma Mo-MLV insertion region 1 homologue	miR-200c, -203 -183, -128	Godlewski et al., 2008; Wellner et al., 2009

Influence of xenobiotic drugs on miRNA expression

miRNA expression also can be affected by exposure to xenobiotics. For example, the abundances of a number of miRNAs (miR-27a, -148a, -124a and miR-451) in MCF-7, Caco2, SH-SY5Y and BE(2)-M17 cell lines were affected by exposing the cells to common pharmacological agents, including dexamethasone, vinblastine, bilobalide and cocaine (Rodrigues et al., 2011). As miRNAs regulate drug metabolizing enzymes and transporters, this might lead to considerable changes in the PK properties of the drug itself.

Suppression of miRNAs that control drug metabolism and disposition may explain changes in the expression of efflux transporters (Yu, 2009). For instance, the induction of CYP3A4 and ABCB1 by dexamethasone functions at least partially through the suppression of miR-27b, -451 and -148a, which may interact with CYP3A4 and ABCB1 transcripts (Rodrigues et al., 2011). Vinblastine reduces the levels of miR-27a/b, -324-3p, -328, -148a and -451, which may cause the increase of ABCC1 expression in MCF-7 cells (Schrenk et al., 2001), ABCB1 in LS-180 cells (Harmsen et al., 2010) and CYP3A4 in HepG2 cells (Smith et al., 2010). Another example is the brain-specific miR-124a, a major regulator of neuronal identity (Conaco et al., 2006; Maisel et al., 2010). The abundance of this miRNA was reduced by psychoactive drugs (cocaine, methadone and fluoxetine), which might present a mechanism of neuroplasticity in response to xenobiotic agents.

Xenobiotic drugs may act on other proteins responsible for miRNA processing, leading to an altered expression of mature miRNAs (Shan et al., 2008; Melo et al., 2011). For example, enoxacin, a broad-spectrum antibacterial fluoroquinolone, enhanced endogenous miRNA production, including tumour suppressor miRNAs, by targeting the pre-miRNA processing protein TAR RNA-binding protein 2 (Melo et al., 2011).

Host miRNA expression changes in response to the gut microbiome

The gastrointestinal tract contains a broad spectrum of microbial species that modulate the utilization of xenobiotics beyond the capability encoded in the host genome (Kau et al., 2011; Haiser and Turnbaugh, 2013). Alterations to the microbiome population can influence the therapeutic efficacy of drugs (Viaud et al., 2013). In addition, microbes can also contribute to dose-limiting toxicity of other agents, an undesirable effect that can be relieved by inhibition of bacterial β-glucuronidases (Wallace et al., 2010).

The gut microbiome can also affect host miRNA expression, which may modulate the levels of various ADME genes and affect the ADME system. Comparison of the colon miRNA profiles of germ-free and control mice identified several dys-regulated miRNA species, including miR-128, -200C*, -342-5p, -465c-5p, -466d-3p, -466d-5p, -665 and -683 (Dalmasso et al., 2011a). Using various experimental tools including luciferase assays and immunoblotting, it has been shown that the ABCC3, a cell surface transporter, is a direct target of miR-665. Microbiome inoculation down-regulated miR-665 level, which, in turn, increased the level of ABCC3 in enterocytes at both the mRNA and the protein levels (Dalmasso et al., 2011a). Although the exact effects of the changes of ABCC3 level are yet to be determined, the biliary transport and excretion of organic anions are most likely to be affected. This finding demonstrates the complex interactions between gut microbiome and xenobiotic metabolism. The interaction is not just directly at chemical levels but also involves complex gene/protein networks through regulatory factors such as miRNA and other epigenetic factors.

Conclusions and future prospects

Variation between individuals in response to therapeutic agents is often attributed to the differences in host genetic factors. However, our knowledge of other contributors including epigenetic factors and microbiome dimensions is rapidly increasing. miRNA plays an important role in various physiopathological processes and substantial evidence is accumulating for its involvement in ADME. However, miRNA-mediated regulatory process is complex; although a number of computational algorithms have been developed to predict miRNA–mRNA interactions (Enright et al., 2003; Lewis et al., 2005), the accuracy of these predicted interactions is limited. In addition, most biological studies have been performed in vitro with only limited in vivo complementation. The precise physiologically relevant effects of miRNAs on ADME remains unclear and further study is required to generate detailed, highly substantiated empirical interaction networks to realize their diagnostic and therapeutic potential.

The recent realization of the complexity of the gut microbiome and its capacity to manipulate xenobiotics provides a new front in the study of drug metabolism and its effects on miRNA expression. Even though germ-free animal models provide some important insights on the effect of gut microbiome on host gene and miRNA expression, the host–microbiome interaction is complex and remains to be deciphered. Systems biology seeks to integrate results from different high-throughput profiling technologies to understand the dynamic changes of a biological system and predict its responses to various inputs. Using this approach to study the effects of epigenetic factors and microbiome on drug metabolism would provide a more comprehensive view on how different parts of the system interact with each other. A better understanding of factors affecting drug metabolism will be integral to personalized medicine, as it would provide guidance on drug development, increase the therapeutic efficacy, tailor specific treatment strategies for individuals and reduce adverse effects.

Glossary

ABC

ATP-binding cassette

ABCB1/MDR1/P-gp

multidrug resistance protein 1/P-glycoprotein

ADME

absorption, distribution, metabolism, excretion

antimiR

miRNA antagonist

CYP450

cytochrome P450

DOX

doxorubicin

ESR1

oestrogen receptor alpha

ETS1

V-Ets avian erythroblastosis virus E26 oncogene homologue 1

GPX7

glutathione peroxidase 7

GSS

glutathione synthetase

HDAC

histone deacetylase

LXRA

liver X receptor α

miRNA

microRNA

mRNA

messenger RNA

ncRNA

non-coding RNA

NR1C1

PPARα

PK

pharmacokinetic

PXR

pregnane X receptor

RISC

RNA-induced silencing complex

RXR

retinoid X receptor

SLC

solute carrier

UGT

UDP-glucuronosyltransferase

UTR

untranslated region

VD receptor

vitamin D receptor

Conflict of interest

The authors declare no conflict of interest.