**Emilia Kozlowska, Wlodzimierz J. Krzyzosiak \* and Edyta Koscianska \***

Department of Molecular Biomedicine, Institute of Bioorganic Chemistry, Polish Academy of Sciences, Noskowskiego 12/14 Str., 61-704 Poznan, Poland; E-Mail: emiliak@ibch.poznan.pl

**\*** Authors to whom correspondence should be addressed; E-Mails: wlodkrzy@ibch.poznan.pl (W.J.K.); edytak@ibch.poznan.pl (E.K.); Tel.: +48-61-852-8503, Fax: +48-61-852-0532.

*Received: 30 May 2013; in revised form: 6 August 2013 / Accepted: 8 August 2013 / Published: 19 August 2013*

**Abstract:** With the advent of deep sequencing technology, a variety of miRNA length and sequence variants, termed isomiRNAs (isomiRs), have been discovered. However, the functional roles of these commonly detected isomiRs remain unknown. In this paper, we demonstrated that miRNAs regulate the expression of the *HTT* gene, whose mutation leads to Huntington's disease (HD), a hereditary degenerative disorder. Specifically, we validated the interactions of canonical miRNAs, miR-137, miR-214, and miR-148a, with the HTT 3'UTR using a luciferase assay. Moreover, we applied synthetic miRNA mimics to examine whether a slight shifting of miRNA seed regions might alter the regulation of the *HTT* transcript. We also examined miR-137, miR-214, and miR-148a isomiRs and showed the activity of these isoforms on reporter constructs bearing appropriate sequences from the HTT 3'UTR. Hence, we demonstrated that certain 5'-end variants of miRNAs might be functional for the regulation of the same targets as canonical miRNAs.

**Keywords:** miRNA; isomiR; target validation; luciferase assay; huntingtin; Huntington's disease

#### **1. Introduction**

MicroRNAs (miRNAs) are 21- to 24-nucleotide noncoding RNAs that fine-tune gene expression. These molecules act at the posttranscriptional level through modulation of translational efficiency and/or destabilization of target transcripts (reviewed in [1]). miRNAs exert their functions through imperfect pairing with the 3' untranslated region (UTR) of target mRNAs. Nucleotides 2 through 8 of the miRNA, termed the "seed" sequence, are essential for target recognition and binding [2].

The canonical pathway of animal miRNA biogenesis includes two subsequent cleavages (reviewed in [3–6]). Briefly, precursor miRNAs (~60-nt pre-miRNAs) are generated from primary transcripts (pri-miRNAs) through cleavage with the ribonuclease Drosha and exported to the cytoplasm by Exportin-5. Then, ~22-nt miRNA duplexes are generated through cleavage with the ribonuclease Dicer. Only one miRNA strand (the guide strand) of the duplex induces Argonaute proteins (AGO) to form the programmed RNA-induced silencing complex (RISC); the other strand (the passenger strand, or miRNA\*) is released and degraded. The thermodynamic stability of the ends of the miRNA duplexes plays a crucial role in miRNA strand selection.

Currently, more than 2000 mature human miRNAs have been deposited in the miRNA repository (miRBase, Release 19) [7]. The deep sequencing of short RNAs has not only enabled the identification of novel miRNAs but also revealed that miRNAs are heterogeneous and differ in length. Heterogeneous miRNA variants are referred to as isomiRNAs (isomiRs) [8]. The primary source of the heterogeneity of miRNA length is imprecise cleavage by the ribonucleases Drosha and Dicer [8–11], which can be further biased at the AGO2 binding step [12]. However, miRNA length variation might also reflect various downstream effects, such as limited miRNA degradation by exonucleases, the addition of extra nucleotides [13–15], and miRNA sequencing artifacts [16,17]. It has recently been shown that the human trans-activation response (TAR) RNA-binding protein (TRBP), a molecular partner of Dicer, might also contribute to miRNA length heterogeneity. Specifically, TRBP triggered the production of isomiRs that were longer at the 5' strand than the canonical miRNAs by a single nucleotide. As a result, different mRNAs were targeted due to changes in guide-strand selection [18]. It has also been reported in *Drosophila* that the Nibbler (Nbr) 3'–5' exonuclease trims the 3'ends of miR-34 generating isomiRs shorter than the canonical sequence [19]; however, there is no evidence for similar exonuclease activity in vertebrates.

miRNAs control the expression of the majority of human genes [20], and these molecules are involved in many physiological and pathological processes. The alteration of miRNA expression has been associated with numerous diseases, including neurodegenerative disorders, such as Huntington's disease (HD). HD is the most common fatal polyglutamine (polyQ) disorder and results from the expansion of a CAG repeat in exon 1 of the huntingtin (*HTT*) gene. The precise mechanism of HD pathogenesis is not fully understood, but both the mutant protein (reviewed in [21]) and mutant transcript might be toxic to cells (reviewed in [22]). Of particular interest is the potential involvement of miRNA in the regulation of the *HTT* gene. The global deregulation of miRNAs in samples obtained from HD patients was demonstrated using Illumina massively parallel sequencing [23]. Most importantly, miRNA of varying lengths and/or sequences (isomiRs) were observed for the vast majority of miRNAs detected in two forebrain areas, the frontal cortex (FC) and striatum (ST), of both healthy individuals and HD patients [23].

In general, the miRNA heterogeneity observed in deep sequencing might have important functional implications. Most importantly, miRNAs with shifted 5'-ends have different seed sequences responsible for the recognition of a complementary sequence and the binding to mRNA. Therefore, it is assumed that heterogeneous 5' isomiRs might regulate different targets [10,15,24,25]. Moreover, both 5' and 3' isomiRs might exhibit modified turnover properties [24,26] and altered strand selection within the RISC because strand selection is influenced by the extent of the 3' overhang and the degree of pairing for any miRNA-miRNA\* duplex [27,28].

An early evidence supporting the hypothesis of isomiRs functionality comes from an experiment that showed a difference in target cleavage between miR-142-5p and its variant, which contained two extra nucleotides at the 5'-end [29]. A putative functional role for isomiRs has been suggested in many reports because isomiRs actively associate with the RISC and translational machinery [24,30–32] (reviewed in [33]). This assumption was further supported by the observation that isomiRs exhibit differential expression across tissues and developmental stages [26,34,35]. Nevertheless, the real biological significance of isomiRs is not fully understood because few studies concerning isomiR regulation at the cellular level have been reported, and thus far, only variants of miR-133, miR-101, and miR-31 have been experimentally examined. Specifically, it was shown that 5'-isomiR-101, which is highly expressed in the brain, associates with AGO2 immunocomplexes and decreases the expression of five validated miR-101 targets but to a lesser degree than the canonical miR-101 [35]. Differential mRNA targeting was demonstrated in the case of two prevalent 5' isomiRs of the key cardiac regulator miR-133a [31]. Three miR-31 isoforms that differed only slightly in their 5'- and/or 3'-end sequences were compared (namely, hsa-miR-31, ptr-miR-31, and mmu-miR-31), implicating isomiR-31s in the concordant and discordant regulation of six known target genes [36].

In this paper, we validated miRNA-mRNA interactions that might be involved in the regulation of the HTT transcript. Specifically, we experimentally assessed the validity of three predicted interactions and demonstrated that the canonical miR-137, miR-214, and miR-148a bind to the 3'UTR of the *HTT* gene. These results provide the first evidence that miR-137 and miR-148a regulate the expression of huntingtin and confirm that this regulation is also mediated by miR-214, as previously reported [37]. Moreover, using luciferase reporter assays, we investigated the regulation of huntingtin using select miRNA isoforms. We focused on 5'-end isomiRs with the shifted seed sequence that is the primary determinant of mRNA target recognition. Here, we showed that certain 5'-end isomiRs of miR-214 are functional for the downregulation of huntingtin expression.

#### **2. Results and Discussion**

#### *2.1. Prediction of miRNA Interactions with the HTT Transcript*

In a previous study, we predicted potential miRNA interactions with mRNAs derived from genes triggering hereditary neurological disorders known as trinucleotide repeat expansion diseases (TREDs), including Huntington's disease (HD) [38]. The results of this in-depth *in silico* analysis prompted further research on the potential miRNA-mediated regulation of the HTT transcript in the context of the pathogenesis and therapy of HD. We compared different target prediction algorithms and verified our predictions using the available data gathered in various databases dedicated to miRNA target prediction (e.g., miRWalk database [39] and miRTarBase [40]). We selected interactions with miR-137, miR-214, and miR-148a for experimental verification. The deregulation of the expression of these miRNAs in HD patients or in cellular models of HD has been reported. Specifically, miR-137 was downregulated in the striatum of HD patients [23], while both miR-214 and miR-148a were upregulated in ST*HdhQ111/HdhQ111* cells [41]. Moreover, miR-137 is highly expressed in the nervous system, suggesting the involvement/potential role of this miRNA in the pathogenesis of HD. miR-137 has also been recently identified as a direct target of the repressor element-1 (RE-1) silencing transcription factor (REST) [42]. The second candidate, miR-214, has been positively verified in previous studies; miR-214, along with three other miRNAs (miRs 150, 146a, and 125b), downregulated the expression of huntingtin [37]. The same study also showed that these miRNAs affect the formation of mutHTT aggregates, the toxicity induced by mutHTT, and the expression of brain-derived neurotrophic factor (BDNF), thereby collectively contributing to HD pathogenesis.

The candidate miRNAs (miRs 137, 214, and 148a) ranked high in the results generated by either algorithm based on conservation criteria, *i.e.*, Diana-micro T [43], miRanda [44], or PicTar [45]. However, our prediction was primarily based on the use of the TargetScanHuman algorithm (Release 6.2) [46]. According to TargetScan, a site for miR-137 is highly conserved among vertebrates, and sites for miRs 214 and 148a are poorly conserved among mammals or vertebrates. In addition, the miR-137 and miR-148a sites were 8mers (defined as exact matches to positions 2–8 of the mature miRNA, followed by an adenine), while the selected miR-214 site was a 7mer-m8 (an exact match to positions 2–8 of the mature miRNA). The positions of the miR-137, miR-214, and miR-148a binding sites in the 3'UTR of the huntingtin transcript and the base pairing of these miRNAs with target sequences are presented in Figure 1. The binding parameters of these miRNAs met the recommended bioinformatics criteria, and their experimental validation was of particular interest in the light of current knowledge of potential involvement of miRNAs in neurodegeneration and the entire competing endogenous RNA (ceRNA) activity network [47], which recently has been shown to be implicated in neurodegenerative diseases including HD [48,49].

**Figure 1.** Graphical presentation of selected miRNA target site distribution in the 3' untranslated region (3'UTR) of the huntingtin transcript. To predict miRNAs that potentially target the HTT 3'UTR, the TargetScanHuman algorithm (Release 6.2) [46] was used. (**A**) Regions of interaction for the miRNAs selected for experimental validation; (**B**) miRNA base pairing with an appropriate target sequence is schematically presented.

*2.2. Canonical miR-137*, *miR-214*, *and miR-148a Regulate the Expression of the HTT Gene*

For the experimental validation of the predicted binding of the selected canonical miRNAs (miRs 137, 214, and 148a) to their target sites in the HTT 3'UTR, experiments using reporter constructs and luciferase assays were performed as described previously [50]. However, sequences carrying binding sites for the appropriate miRNAs were cloned into pmirGLO vector (Promega), which is considered optimal for miRNA-mRNA interaction studies. Constructs bearing single miRNA binding sites were generated and defined as wild-type reporters (WT). Constructs with mutations that disrupted native pairing within the binding region (5' seed site) of the candidate miRNAs (MUT) and constructs that showed perfect complementarity (PM) to these sites were also generated to provide negative and positive controls, respectively (details in the Experimental section).

We transfected HEK293T cells with either reporter carrying potential miRNA binding sites. Four constructs were transfected into cells and tested in parallel. To determine whether the miRNAs of interest were expressed in the HEK293T cells, we performed northern blot analysis. The expression of miR-137 was not detected in the HEK293T cells. miR-214 and miR-148a were expressed at low and moderate levels (Figure 2), respectively, consistent with the available deep sequencing results. Therefore, our experimental system required miRNA overexpression, and we used miRNA-coding plasmid vectors (System Biosciences, Open Biosystems) for this purpose (compare endogenous miRNA levels and those expressed from vectors in Figure 2).

**Figure 2.** Endogenous expression and overexpression of miR-137, miR-214, and miR-148a in HEK293T cells. Northern blot detection of miRs 137, 214, and 148a in non-treated HEK293T cells and cells transfected with miRNA-coding plasmids (System Biosciences, Open Biosystems). M denotes the size marker, end-labeled 17, 19, 21, 23, and 25-nt oligoribonucleotides. En and Ex indicate the miRNA levels, endogenous and expressed from appropriate vectors, respectively. Hybridization to U6 RNA provides a loading control.

In the luciferase assays, we obtained considerable repression of the luciferase expression after the transfection of reporter constructs for the three miRNAs tested (Figure 3A). Specifically, we observed a significant reduction in luciferase activity when reporter constructs bearing binding sites for miRs 137 and 214 were used (reductions to 83% and 79%, respectively) and a slightly weaker but reproducible and statistically significant suppression of the luciferase activity in the case of miR-148a (suppression to 87%). The luciferase activity for all of the MUT constructs showed efficient de-repression nearly equal to that in the control experiment; the positive controls (PMs) repressed luciferase at low levels, ranging from 17% to 33% (for miR-148a PM and miR-214 PM, respectively) of the empty reporter construct. These results verify the reliability of the experimental system used.

**Figure 3.** Regulation of the huntingtin (HTT) expression by canonical miRNAs. (**A**) Relative repression of the luciferase expression. Reporter constructs carrying a single binding site for miR-137, miR-214, and miR-148a were tested. For each luciferase. experiment, the miRNA activity on four constructs was measured in parallel: an empty pmirGlO vector (Control), a wild-type potential binding site for the appropriate miRNA (WT), a mutated binding site (MUT), and a site with full complementarity (PM). The firefly luciferase activity was normalized against *Renilla* luciferase activity*.* An average result from at least three independent experiments is shown (details in the text); (**B**) Relative HTT mRNA levels. Real-time PCR performed 48 h after transfection of HEK293T cells with miR-137, miR-214, and miR-148a. The bar graphs show the quantification of the HTT mRNA levels normalized to actin mRNAs based on data collected from three independent experiments; (**C**) Relative HTT protein levels. Western blot analysis of the cellular levels of HTT protein 72 h after transfection of HEK293T cells with miR-137, miR-214, and miR-148a. The bar graphs show the quantification of the protein levels detected in three western blot experiments. A representative blot is shown. The asterisks indicate statistical significance; a single asterisk at *p*-value < 0.05 and a double asterisk at *p*-value < 0.01.

We also monitored huntingtin expression at the mRNA and protein levels following the transfection of HEK293T cells with miRNA-coding plasmids. Real-time PCR performed 48 h after transfection with miR-137, miR-214, or miR-148a showed a strong decrease in the HTT mRNA level (Figure 3B). Similarly, the HTT protein level was significantly reduced 72 h posttransfection in cells overexpressing any of the miRNAs (Figure 3C, Figure S1). This observation is consistent with the finding that miRNA binding reduces the cellular levels of targeted transcripts [51,52]. However, other studies have reported that no or minimal changes in the respective mRNA levels were observed or that these changes were only reported for certain targets [35]. Overall, miR-137, -214 and -148a were positively verified as negative regulators of the *HTT* gene. The lack of regulation of the huntingtin expression, demonstrated in both luciferase assays and western blotting, was observed for the other miRNA (miR-107) and shown for comparison as supplementary data (Figure S2). The strongest reduction in the luciferase activity and the greatest and second-greatest repression at the mRNA and protein levels were observed with miR-214. Thus, this study provides further support for the regulatory potential of miR-214, which was previously validated in a different experimental system [37]. Moreover, this study provides the first evidence of HTT regulation by miR-137 and miR-148a.
