*2.3. 5'-End Variants of miRNAs Are Functional and Might Regulate the Same Targets as Canonical miRNAs*

Although many reports suggest isomiR functionality [24,30–32], there is still little research that address this issue experimentally. Specifically, one variant of miR-101 [35] and two isomiRs of miR-133 [31] and miR-31 [36] have been investigated. In these reports, the isomiRs were less effective than their canonical analogs [35] or exhibited differences in effectiveness depending on the regulated target [31].

Here, we determined whether the 5'-end variants of three miRNAs (5'-end isomiRs), namely miRs 137, 214 and 148a, might function in the same experimental system (*i.e.*, whether these miRs reduce the luciferase activity when appropriate reporter constructs are used). We designed and synthesized miRNA variants with seed sequences shifted by −1, +1, or +2 nt (Integrated DNA Technologies) (Figure 4A). We selected miRNA 5' isoforms that are relatively highly represented in deep sequencing data because we considered sequence abundance a prerequisite for the functionality of these molecules. We based this selection on the sequencing data gathered in the YM 500 database [53] but we also evaluated the expression levels of isomiRs in other sources [32]. The only exception was isomiR-137+1, whose sequence is barely detectable using deep sequencing. This isomiR variant was added to the analysis to examine the same miRNA seed shifts for all isomiRs tested. Moreover, trimming variants that affect the 5' end of miRNAs were reported to be abundant species, and the vast majority of these 5' isomiRs affected a single nucleotide upstream of the reference miRNA [35]. A strong correlation between the expression of miRNAs and isomiRs was also observed [30].

According to the TargetScanHuman Custom (Release 5.2) [46] prediction, none of the selected isomiRs targeted HTT (Figure 4A); thus, we verified the targeting of these molecules experimentally. In addition, we assessed *in silico* how the overall number of genes targeted by the analyzed miRNAs and isomiRs might vary due to the change introduced into their seed regions. Potential targets for the 5'-end variants of miR-137, miR-214, and miR-148a were predicted using the TargetScan Custom 5.2 algorithm [46] and are shown in the Venn diagrams by overlaps (Figure 4B). Specifically, targets for the canonical miRNAs were compared with the targets of the miRNAs with seed regions shifted by −1, +1, and +2 nt. This analysis revealed that the number of predicted targets changed, but apart from unique targets, many genes were still predicted as targets for both miRNAs and isomiRs, confirming that isomiRs might share certain common mRNA targets but not all mRNA targets [36]. These results are also consistent with the suggestion that isomiRs function cooperatively to target common biological pathways [30]. However, distinct functions for miRs and isomiRs have also been suggested [31,35].

**Figure 4.** Graphical presentation of selected isomiR variants and their potential to target different genes. (**A**) Nucleotide sequences of miR-137, miR-214, miR-148a, and their isoforms. miRNA sequences are marked in red, and isomiR sequences are shown in blue, green, and violet for −1-, +1-, and +2-nt seed shifting, respectively. The miRNA seed sequences are labeled with black rectangles. Information on the miRNA lengths, as well as their potential for targeting the *HTT* gene and isomiR expression levels, is also provided. ( \*) Ability to interact with the HTT 3'UTR, as predicted by the TargetScanHuman algorithm (Release 6.2) for miRNAs and the TargetScan Custom (Release 5.2) for isomiRs [46], ( \*\*) isomiR read number according to the YM500 database [53]; (**B**) Venn diagrams showing the predicted miRNA targets for selected isomiRs. Potential targets for the 5'-end variants of miR-137, miR-214, and miR-148a were predicted using the TargetScan Custom algorithm (Release 5.2) and are shown as overlaps in the Venn diagrams. Targets for the canonical miRNAs are compared with the targets for the miRNAs with the shifted seed regions and are depicted in the same colors as in panel A. The numbers inside the circles denote the numbers of potential targets predicted for the appropriate miRNA variants.


To validate the regulation of the HTT transcript by canonical miRNAs in a luciferase assay, we overexpressed the desired miRNAs from plasmid vectors. To study the interactions of the HTT transcript with isomiRs, appropriate isomiR sequences had to be introduced into cells as synthetic oligonucleotides. Thus, we transfected HEK293T cells with both the miR-137 mimic and miR-137 vector (System Biosciences) to determine whether these two experimental systems generate the same results (Figure 5A). Moreover, we examined miRNA mimic activities at different final concentrations (10, 30, and 50 nM) to determine the optimal concentration for these experiments (Figure 5B). A clear correlation between the results of the luciferase experiments with the miR-137-coding plasmid and the synthetic miR-137 mimic was observed; thus, we further investigated the functionality of our 5'-end isomiRs using appropriate miRNA mimics. In the luciferase assays, we obtained considerable and significant repression of the luciferase expression after the transfection of the reporter constructs and all three miR-214 isomiR mimics, namely, isomiR-214+1, isomiR-214+2, and isomiR-214-1 (luciferase repression equal to 71%, 80%, and 79%, respectively). Moreover, this reduction in the luciferase activity was comparable to the reduction induced by the canonical miR-214 mimic (71%) (Figure 6A). In contrast, the luciferase activity was not reduced when miR-137 isomiRs were used, in the case of neither isomiR-137+1 nor isomiR-137-1, compared with the considerable repression observed using the canonical miR-137 mimic (79%) (Figure 6B). Similarly, in the case of isomiR-148a+1 and isomiR-148a-1, the activity of luciferase was slightly reduced (9% and 6%, respectively), while the reduction obtained for the canonical miR-148a mimic was much stronger (80%) (Figure 6C). The observed difference in the functionality of the analyzed isomiRs raises the question when miRNA-mRNA pairing conforms to strict rules and when some flexibility in the miRNA seed region is permitted, and which additional mechanisms other than the base paring of the seed region might affect target genes repression by isomiRs.

Several factors influence the recognition of a target site by miRNA, e.g., the sequence composition of the 3'-UTR [54], the immediate environment of the putative target site [55], and the structural accessibility of the target site [2,56]. Moreover, endogenous natural antisense transcripts transcribed from the opposite strand of a protein-coding gene or a non-protein coding gene [34] and the RNA-binding proteins [57] could directly bind to mRNA, thereby masking the miRNA binding site of a target gene and preventing the inhibitory effects of the miRNA on target gene translation. These factors, however, are of importance to canonical miRNA binding. Here, we examined several 5' isomiRs of slightly different lengths that previously demonstrated canonical miRNA targeting. Therefore, the structural features and genomic context of these molecules did not significantly differ between the canonical miRNAs and their isomiRs or between the isomiRs themselves.

A distinct feature of the functional isomiR-214 variants and the two other isomiRs examined in this study was the fact that miR-214 is a 7mer with compensatory base pairing at the 3' end (see Figure 1). Although canonical miRNA-target specificity is primarily triggered by complementarity within the seed region, non-canonical interactions depend also on 3' compensatory sites [2,58], which might be important for miR-214 and its variants. The miRNA/isomiR length was also suggested as a factor that might affect functionality. In a study of isomiRs, the analysis of two miR-133a mimics (22/23 nt) was performed, followed by the analysis of two other variants that represented the respective other length for each miR-133a variant. However, the luciferase repression did not depend on mimic length within this range [31]. Therefore, alterations to the 3' end of the miR-133a mimic did not affect the level of mRNA repression, suggesting that the 3' end is not essential for efficient target binding in this case. Another important factor that might account for the disparate functioning of isomiRs is differential binding capacity with the Argonaute complex (affinity of a given miRNA to AGO). Previous studies have shown that some miRNA variants were differentially loaded onto AGOs, and the 5'-end nucleotide of small RNA was critical for its interaction with AGO proteins [12,59–61]. However, miR-101 was more efficiently loaded into the RISC than its isomiR [35], and the 5'-end nucleotide of isomiR-31s was not a rigorous criterion for AGO complex loading [36]. In this study, in the case of the most effective miRNA, namely miR-214, all variants were functional regardless of the different nucleotides at their 5' end (Figure 2). Small changes in the miRNA sequence profoundly affected the functional asymmetry of the miRNA duplex, altering which strand of a miRNA duplex functions in mRNA silencing [18]. Therefore, it cannot be ruled out that, in the case of the nonfunctional isomiRs of miR-137 and miR-148a, the passenger strands were incorporated into the RISC and did not target their binding sites.

**Figure 5.** Correlation between the results of the luciferase experiments conducted with miR-137-coding plasmid and synthetic miR-137 mimics. (**A**) Relative repression of the luciferase expression. Reporter constructs carrying a single binding site for miR-137 were tested; miRNA activity on four constructs was measured in parallel (Control, WT, MUT, and PM), as described in Figure 3. Left—miRNA expression from the synthetic oligonucleotide (miR-137 mimic), right—miRNA overexpressed from the miR-137 vector. The firefly luciferase activity was normalized against *Renilla* luciferase activity*.* The standard errors are calculated from three independent experiments; (**B**) The relative repression of the luciferase expression resulted from the miRNA mimic activity. Four reporter constructs were tested (Control, WT, MUT, and PM) but with the addition of miR-137 mimic at different final concentrations, specifically 10, 30, and 50 nM, as denoted in the figure. The standard errors were calculated from one experiment performed in triplicate.

**Figure 6.** Regulation of the huntingtin expression by isomiRs. Relative repression of the luciferase expression for miR-214, miR-137, miR-148a, and their isomiRs (+1, +2, or −1). Reporter constructs carrying single binding sites for the appropriate miRNAs were tested, namely miR-137 (**A**), miR-214 (**B**), and miR-148a (**C**), as depicted in the figure. For each luciferase experiment, the miRNA activity on four constructs (Control, WT, MUT and PM) was measured in parallel, as described in Figures 3 and 5. The firefly luciferase activity was normalized against *Renilla* luciferase activity*.* The standard errors were calculated from three independent experiments. The asterisks indicate statistical significance; a single asterisk at *p*-value < 0.05 and a double asterisk at *p*-value < 0.01.

#### **3. Experimental Section**

#### *3.1. Cell Culture*

HEK293T cells were obtained from the American Type Culture Collection (ATCC) and grown in Dulbecco's Modified Eagle's Medium (DMEM, Lonza, Wakersville, MD, USA) supplemented with 8% fetal bovine serum (FBS) (Sigma-Aldrich, St. Louis, MO, USA), 2 mM L-glutamine, and an antibiotic-antimycotic solution (Sigma-Aldrich, St. Louis, MO, USA) at 37 °C in a humidified atmosphere of 5% CO2. At 24 h prior to transfection, the HEK293T cells were plated in 12-well or 6-well dishes in DMEM growing medium and harvested 24, 48, and 72 h post-transfection for the luciferase assay, real-time PCR, and western blot analyses, respectively.

#### *3.2. Plasmid Constructs and Synthetic miRNA Oligonucleotides*

To generate reporter constructs bearing miRNA-binding sites, the pmirGLO Dual-Luciferase miRNA Target Expression Vector was used (Promega, Madison, WI, USA). This vector is based on Promega dual-luciferase technology, with firefly luciferase (*luc2*) as the primary reporter for monitoring mRNA regulation and *Renilla* luciferase (*hRluc-neo*) as a control reporter for normalization and selection. Specific oligonucleotides with *Dra*I and *Xba*I ends containing single binding sites for the analyzed miRNA (HTT b.s. for miRs 214, 137, and 148a) were synthesized (IBB Warsaw). The appropriate oligos were annealed by boiling and gradual cooling and subsequently phosphorylated and cloned into the pmirGLO vector, previously digested with *Dra*I (Fermentas, St.-Leon-Rot, Germany) and *Xba*I (Fermentas, St.-Leon-Rot, Germany) restriction enzymes, downstream of the *luc2* gene. For all miRNAs, three types of constructs were prepared, namely wild type (WT), carrying mutations (MUT) and perfect match (PM) constructs (for sequences refer to Table S1), which all have 10-nucleotide flanking sequences, as described previously [50].

For miRNA overexpression, commercial plasmid constructs expressing miRNA precursors (pri-miR-148a (Open Biosystems, Huntsville, AL, USA), pri-miR-137, or pri-miR-214 (System Biosciences, Mountain View, CA, USA)) were used. These plasmids contain pri-miRNA sequences in their natural genome context to ensure biologically relevant interactions with the endogenous processing machinery.

Synthetic miRNA mimics (miR-137, miR-214, and miR-148a mimics) and their length variants were chemically synthesized (Integrated DNA Technologies). The following modifications were introduced: (1) 2'-*O*-methyl modification on positions 1 and 2 and a two-nucleotide UU overhang on the 3' end of the miRNA mimic sense strand, (2) 5' phosphorylation and a two-nucleotide overhang based on nucleotide types found in natural pre-miRNAs on the 3' end of the miRNA mimic antisense strand. All sequences are presented as supplementary data (Table S2).

#### *3.3. Cell Transfection*

HEK293T cells were transfected using Lipofectamine 2000 (Invitrogen, Carlsbad, CA, USA) according to the manufacturer's protocols. For luciferase assays, the cells were transfected in 12-well plates at ~80% confluence. For each transfection experiment, 200 ng of the appropriate reporter construct and either 250 ng of the appropriate miRNA-coding vector or 30 nM of miRNA mimic were used. The cells were harvested 24 h after transfection and assayed for luciferase activity. For miRNA overexpression required for real-time PCR and western blot analyses, the cells were grown to 80% and 60% confluence, respectively, transfected in 6-well plates with 1 μg/mL pri-miRNA plasmid vectors, and harvested at 48 and 72 h, respectively.

#### *3.4. Luciferase Reporter Assay*

After harvesting, the cells were lysed in a passive lysis buffer (Promega, Madison, WI, USA). The luciferase activity was measured using a Dual-Luciferase Reporter Assay System (Promega, Madison, WI, USA) according to the manufacturer's instructions with a Centro LB 960 luminometer (Berthold Technologies, Oak Ridge, TN, USA).

#### *3.5. RNA Isolation and Real-Time PCR*

Total RNA from HEK293T cells was isolated using TRI Reagent (MRC, Inc., BioShop, Cincinnati, OH, USA) according to the manufacturer's instructions. The RNA concentration was estimated using a NanoDrop spectrophotometer. cDNA was obtained from 500 ng of total RNA using Superscript III (Life Technologies, Carlsbad, CA, USA) and random hexamer primers (Promega, Madison, WI, USA). For subsequent quantitative real-time analyses, 50 ng of cDNA was used. Real Time PCR was performed on a LightCycler 480 II system (Roche Diagnostics, Mannheim, Germany) using TaqMan Gene Expression Assays and TaqMan Universal Master Mix II (Applied Biosystems, Foster City, CA, USA). The results obtained for the assessment of huntingtin mRNA levels were normalized to the levels of actin mRNA.

#### *3.6. Northern Blotting*

High-resolution northern blotting was performed as previously described [62,63]. Briefly, 25 μg of total RNA was extracted from HEK293T cells and resolved on a 12% denaturing polyacrylamide gel in 0.5× TBE. The RNA was transferred to a GeneScreen Plus hybridization membrane (PerkinElmer, Spokane, WA, USA) using semi-dry electroblotting (Sigma-Aldrich, St. Louis, MO, USA), immobilized by subsequent UV irradiation (120 mJ/cm2 ) (UVP), and baked in an oven at 80 °C for 30 min. The membranes were probed with specific DNA oligonucleotides (Table S3) complementary to the annotated human miRNAs miR-137-3P, miR-214-3P, and miR-148a-3P (miRBase). The probes were labeled with [γ32P] ATP (5000 Ci/mmol; Hartmann Analytics, Braunschweig, Germany) using USB OptiKinase (Affymetrix, Cleveland, OH, USA). The hybridizations were performed at 37 °C overnight in a PerfectHyb buffer (Sigma-Aldrich, St. Louis, MO, USA). The marker lanes contained a mixture of radiolabeled RNA oligonucleotides (17-, 19-, 21-, 23-, and 25-nt in length). Hybridizations to U6 RNA provided loading controls. Radioactive signals were quantified by phosphorimaging (Multi Gauge v3.0; Fujifilm).

#### *3.7. Western Blotting*

A total of 15 µg of protein was diluted in sample buffer containing 2-mercaptoethanol, denatured for 5 min, and separated on 3%–8% gradient Tris-Acetate gels (Invitrogen, Carlsbad, CA, USA) in XT Tricine Buffer (BioRad, Hercules, CA, USA). After electrophoresis, the proteins were electrotransferred onto a nitrocellulose membrane (Sigma, St. Louis, MO, USA). All immunodetection steps were performed on a SNAPid (Millipore, Billerica, MA, USA) in PBS buffer containing 0.25% nonfat milk and 0.1% Tween 20, and the membranes were washed in PBS/Tween. For huntingtin and tubulin detection, the blots were probed with the primary anti-huntingtin (1:500, Millipore, Billerica, MA, USA) and anti-alpha-tubulin (1:5000, Covance, Emeryville, CA, USA) antibodies, respectively, and subsequently probed with HRP-conjugated secondary antibodies (1:500, Sigma, St. Louis, MO, USA). The immunoreaction was detected using Western Bright Quantum (Advansta, CA, USA). The protein amounts were quantified using GelPro 3.1 software (Media Cybernetics, Bethesda, MD, USA).

#### *3.8. Statistical Analysis*

All experiments were repeated at least three times. Graphs were generated using GraphPad Prism 5 (GraphPad Software). The figures for the luciferase assays were generated after averaging the results from the repeat experiments for a particular construct. The values for error bars (mean with SD) and the statistical significance were calculated using GraphPad Prism 5. The statistical significance of the luciferase reduction in the case of transfection with constructs carrying miRNA-binding sites was assessed using a one-sample *t*-test with a hypothetical value of 1 assigned to cells transfected with a control empty vector. *p*-values < 0.05 (two-tailed) were considered significant.

#### **4. Conclusions**

This study presents new evidence that *HTT* gene expression is regulated by miRNAs and, most importantly, demonstrates that certain isomiRs are functional and regulate the same target as canonical miRNAs.

IsomiRs are commonly reported in deep-sequencing studies and have been described in all studied organisms and tissues. The existence of miRNA variants might contribute considerably to the complexity of target regulation by miRNAs and strongly increase the regulatory potential of these molecules. The presence of isomiRs could have far-reaching implications for miRNA therapeutic applications; it must be taken into account in various diagnostic tests as well as in the design of miRNA mimics or anti-miRs as therapeutic agents. Therefore, of particular importance is to identify factors that determine the biological relevance of isomiRs.

#### **Acknowledgments**

This work was supported by funding from the Polish Ministry of Science and Higher Education (N N301 523038), the National Science Centre (2011/03/B/NZ1/03259), and the European Regional Development Fund within the Innovative Economy Programme (POIG.01.03.01-00-098/08). The real-time PCR analyses were performed on a LightCycler 480 II system (Roche) in the Laboratory of Subcellular Structures Analysis at the Institute of Bioorganic Chemistry, PAS, in Poznan.

#### **Conflicts of Interest**

The authors declare no conflict of interest.

#### **References**


Reprinted from *IJMS*. Cite as: Ma, J.; Yu, S.; Wang, F.; Bai, L.; Xiao, J.; Jiang, Y.; Chen, L.; Wang, J.; Jiang, A.; Li, M.; Li, X. MicroRNA Transcriptomes Relate Intermuscular Adipose Tissue to Metabolic Risk. *Int. J. Mol. Sci.* **2013**, *14*, 8611-8624.
