Altered microRNA expression profile in amyotrophic lateral sclerosis: a role in the regulation of NFL mRNA levels

To characterize the expression pattern of known human miRNAs in sALS SC, RT-qPCR was performed using the TaqMan Low Density Human MicroRNA assay kit as described in the methods below. This allowed for the analysis of a total of 664 miRNAs from 3 cases of controls and 5 cases of sALS.

We observed 4 miRNAs expressed exclusively in sALS but not in controls [miR-558 (p < 0.001); miR-16-2*, miR-146a* (p < 0.01); miR-508-5p (p < 0.05)] (Table 1). In addition, 7 miRNAs were expressed in controls and not in sALS [miR-624 (p < 0.001); miR-520e, miR-524-5p, miR-548a-5p, miR-606, miR-612, miR-647(p < 0.05)] (Table 2). From 245 miRNAs expressed in both sALS and controls but at statistically different levels, 6 were expressed in sALS at higher levels than controls [miR-373*, miR-551a (p < 0.01); miR-506, miR-518a-5p, miR-518e*, and miR-890 (p < 0.05)] (Table 3) and all the remaining (239) were down-regulated in the sALS SC tissue (Additional file 1: Table S1). Only 10 miRNAs of the potential 664 miRNAs had no expression in sALS or control lumbar SC (Additional file 1: Table S2). To validate these observations, we randomly selected miRNAs from the results of the TaqMan array and subjected these to RT-qPCR using SYBR Green. We confirmed the significant alteration of these miRNAs in the SC in sALS (data not shown).

These results demonstrate that a significantly altered miRNA expression profile exists in the SC in sALS.

To identify networks and biological functions related to the target genes of miRNAs differentially expressed in sALS versus controls, we used Ingenuity Pathway Analysis (IPA). A group of 256 miRNAs were analyzed. From these, 10 miRNAs were up-regulated in the SC in sALS (4 miRNAs expressed only in sALS and 6 miRNAs expressed in both sALS and controls but at higher levels in sALS) and all the remaining were down-regulated (7 miRNAs expressed only in controls and 239 miRNAs expressed in both sALS and controls but at higher levels in controls). The most enriched associated networks and biological functions amongst targets of dysregulated miRNAs are shown in the Table 4. MiRNAs up-regulated in sALS are associated with nervous system development and function. In addition, we identified inflammatory response within the top biological functions of the predicted targets of these miRNAs. As well, within the top network and biological functions of predicted targets of down-regulated miRNAs in sALS we found cell death, cellular growth and proliferation, inflammatory disease and nervous system development and function (Table 4).

These results indicate that miRNAs dysregulated in sALS could potentially be involved in the biological functions affected in the ALS pathology.

We used two computational algorithms (microRNA.org, TargetScan) to develop a predicted panel of miRNAs that have differential expression in sALS and MREs within the human NFL mRNA 3′UTR (Table 5). This focus on NFL mRNA was driven by our previous observations of a selective suppression of NFL mRNA in spinal motor neurons in ALS [13‐15]. For predicting MREs we considered the three NFL 3′UTRs reported: NFL 3′UTR-Short (−S; 286 b), NFL 3′UTR-Medium (−M; 1380 b) and NFL 3′UTR-Long (−L; 1838 b). We took into account only miRNAs with perfect pairing of their seed regions to their respective MREs within the NFL 3′UTR, and either conserved or not conserved MREs but always with good mirSVR scores and the most favorable Gibbs free energies (ΔGs) required for the interaction.

Using the above restrictions, the only miRNA that had an MRE in the NFL mRNA 3′UTR and was up-regulated in sALS was miR-146a*. All miRNAs down-regulated in sALS had only one recognition element within the NFL mRNA 3′UTR. Although the sequence identity between NFL mRNA 3′UTRs among Homo sapiens and other mammals is over 90% (Figure 1A), we found that some miRNA recognition sites are conserved but others are not (Table 5). It is important to note that some miRNAs have MREs within the three NFL 3′UTRs reported, -S, -M and –L (e.g. miR-23a), others within NFL 3′UTR-M and –L (e.g. miR-146a*) and one miRNA has a single recognition element within the NFL 3′UTR-L (miR-524-5p; Table 5, Figure 1B).

These results show that not only is the miRNA expression profile altered in ALS, but that miRNA that are predicted to target NFL mRNA 3′UTR are amongst those that are altered.

To study the functional relevance of those miRNA with altered expression in sALS on the regulation of the NFL mRNA 3′UTR, we used luciferase reporter gene assays and relative quantitative RT-PCR. We chose to examine all three lengths of 3′UTR to provide information about the availability of MREs to interact in the context of distinct mRNA structures. We transfected HEK293T cells with pre-miRNAs and a reporter vector harboring one of the three human NFL mRNA 3′UTR (−S, -M or -L).

In the reporter assay, we focused on two patterns: miRNAs up-regulated in sALS that induced down-regulation of the firefly luciferase activity, and miRNAs down-regulated in sALS that induced up-regulation of the luciferase activity. Both cases could explain the selective suppression of NFL mRNA observed in the SC in sALS.

Reporter gene assays showed that the activity of firefly luciferase was significantly decreased with miR-146a* (p < 0.001; up-regulated in sALS; Table 1) when the reporter gene was bound to NFL mRNA 3′UTR-M (Figure 2). Interestingly, this down-regulatory effect was not seen in the case of miR-146a* on the reporter containing the NFL 3′UTR-L. In contrast, miR-23a, miR-23b and miR-30b (down-regulated in sALS; Additional file 1: Table S1) induced an increase of the luciferase activity when the reporter was linked only to NFL 3′UTR-S (p < 0.001; Figure 2). In addition, miR-193a-5p and miR-556-5p were also down-regulated in sALS (Additional file 1: Table S1) and showed an up-regulatory effect on the NFL 3′UTR-M (p < 0.001) and -L (p < 0.001 and p < 0.01, respectively; Figure 2). Finally, miR-524-5p and miR-582-3p (down-regulated in sALS; Table 2 and Additional file 1: Table S1) induced an up-regulatory effect on the NFL 3′UTR-L (p < 0.001), and NFL 3′UTR-S (p < 0.001), -M (p < 0.001) and –L (p < 0.01; Figure 2), respectively. MiR-30c, miR-192 and miR-215 did not show correlation between the results of the reporter gene assay and the expression levels observed in sALS that could explain the suppression of NFL mRNA observed in sALS SC (data not shown). No relative change in the levels of the reporter linked to the NFL 3′UTR was found in cells transfected with miR-let-7a (negative control; Figure 2).

To elucidate whether the group of functional miRNAs were acting on NFL mRNA stability or translational regulation, we studied changes on the luciferase mRNA levels caused by NFL 3′UTR regulation through miRNAs. Relative quantitative RT-PCR showed that miR-146a* was able to decrease the level of firefly luciferase mRNA fused with NFL 3′UTR-M (p < 0.01; Figure 3), while miR-524-5p and 582-3p were able to increase the level of luciferase mRNA coupled to NFL 3′UTR-L (p < 0.01) and NFL 3′UTR-S (p < 0.001), -M (p < 0.001) and –L (p < 0.001), respectively (Figure 3). The remaining miRNAs are not presented because they did not show significant differences in the RT-PCR, or showed RT-PCR results in disagreement with the reporter gene assay that could be explained by an additional miRNA effect (direct or indirect) on translation (data not shown).

Therefore, miR-146a*, miR-524-5p and miR-582-3p showed consistent results in the reporter gene assay and the relative quantitative RT-PCR, making them prime candidates for the negative regulation of NFL mRNA expression in the SC in ALS.

To study whether the regulatory effect of miR-146a*, miR-524-5p and miR-582-3p on the NFL 3′UTR was direct or indirect, we developed 3′UTR mutants of each MRE to test them with the reporter gene assay. We mutated only the first two nucleotides of the 3′ end of each miRNA recognition element thereby preventing broad changes in the NFL 3′UTR secondary structure that could interfere with the interpretation of our results (Figure 4A). Previously, it has been reported that mismatches at these positions significantly reduced the magnitude of target regulation [36]. In fact, two mutations within any of the three MREs analyzed (miR-146a*, miR-524-5p and miR-582-3p) further destabilized target recognition, yielding less favorable ΔGs (Figure 4B).

The reporter gene assay showed that all NFL 3′UTR MRE mutants exhibited a significant decrease in the effect of each miRNA compared with the wild type. Our data demonstrate that miR-146a*, miR-524-5p and miR-582-3p are able to directly regulate the NFL 3′UTR (Figure 4C).

All of our results demonstrate that a pool of three functionally relevant miRNAs capable of interacting with NFL mRNA 3′UTR, miR-146a*, miR-524-5p and miR-582-3p, are dysregulated in sALS, suggesting their participation in the regulation of NFL mRNA expression in mammalian cells. It is interesting to note that not all the miRNA found to be expressed at different levels in sALS served to solely down- or up-regulate the reporter linked to the NFL mRNA 3′UTR. These data strongly suggest that it will likely be a net effect of action of all miRNA involved in NFL expression, and not the action of one single miRNA, that is important for NFL regulation.

We selected 3 cases of non-neurological disease controls and 5 cases of sALS (Table 6). All ALS cases were both clinically and neuropathologically confirmed using the El Escorial Criteria (World Federation of Neurology Research Group on Neuromuscular Disease, 1994). Written consent for tissue donation at autopsy was obtained from either the patient (ante-mortem) or the spouse (post-mortem) in accordance with the London Health Sciences Centre ethics policies. All cases were genotyped and confirmed to have no known mutations in SOD1, TARDBP, FUS/TLS or expanded repeats on C9ORF72 by Dr. Rosa Rademakers (Mayo Clinic, Jacksonville, Florida). No cases were associated with cognitive impairment or frontotemporal degeneration.

MiRNAs were extracted using the mirVana™ miRNA isolation kit (Life Technologies Inc., Ambion, Burlington, ON, Canada) according to the manufacturer’s instructions. Briefly, archival ventral lumbar SC tissue stored at −80°C was placed in 10 volumes of lysis/binding buffer on ice and homogenized with a Polytron until all visible clumps were dispersed. After organic extraction with acid-phenol:chloroform, miRNA isolation was performed by precipitating with ethanol and purifying over two consecutive glass-fiber filters. This procedure yielded a size-selected RNA fraction consisting of RNA species of less than 200 bases. Yield and purity of the miRNA enriched fraction was measured by UV absorbance at OD260/280. Quality of RNA samples was confirmed by BioAnalyzer (Agilent).

The stem-loop RT-PCR based TaqMan Human MicroRNA Arrays (Life Technologies Inc., Applied Biosystems, Burlington, ON, Canada) were used representing 664 mature miRNA in the Sanger miRBase v12 in a two-card set of arrays (Array A and B). Array A contained 3 positive controls and 1 negative control while Array B contained 6 positive controls and 1 negative control. RT-PCR reactions were performed according to the manufacturer’s instructions. Briefly, 300 ng of enriched miRNA (extracted as described above) was reverse transcribed using Megaplex RT Primers and the TaqMan miRNA reverse transcription kit (Life Technologies Inc., Applied Biosystems). For pre-amplification of miRNA cDNA, the RT product was pre-amplified using Megaplex PreAmp Primers and TaqMan PreAmp Master Mix as per manufacturer’s protocols (Life Technologies Inc., Applied Biosystems). The pre-amplified cDNA was diluted with 0.1 × TE (pH 8.0) to 100 μl and then 10 μl of the cDNA was used in each plate for RT-qPCR reaction.

Quantitative real-time PCR (RT-qPCR) was performed using the Applied Biosystems 7900HT system and TaqMan Universal Master Mix (Life Technologies Inc., Applied Biosystems). Cycle threshold (Ct) values were calculated using the SDS software v.2.3. Ct values > 40 were considered to be below the detection level of the assay, therefore, only the miRNAs with Ct ≤ 40 were included in the analyses.

For the data analysis the Ct value of the endogenous mammalian U6 control was subtracted from the corresponding Ct value for the target gene resulting in the normalized ΔCt value. The Student’s t test was used to determine significant differences among ΔCt values of control and sALS groups. The differential expression (the difference between the average of normalized ΔCt value of the target sample (sALS) and the average of normalized ΔCt value for the corresponding calibrator sample (control); ΔΔCT), was used to calculate the expression fold value as follows:

Where Log₁₀RQ correlates directly with up- (positive value) and down-regulation (negative value).

For control cases, the results were deemed negative if all cases were negative for expression. For sALS cases, agreement amongst 4 or more cases was required to consider the result either positive or negative.

We used Ingenuity Pathway Analysis (IPA, 368 Ingenuity® Systems, Redwood City, CA) to analyze dysregulated miRNAs in the SC in ALS. IPA allows us to identify over-represented networks and biological functions of miRNA target genes. The networks are scored with a numerical value and a score ≥ 2 was considered significant. The score is based on the number of Network Eligible molecules in the network, the size of the network, the Network Eligible molecules in the given dataset and the number of molecules in the IPA dataset that could potentially be included in the network. The p-value associated with a biological process is calculated with the Fisher’s exact test, considering the number of functions/pathways/lists of eligible molecules that participate in the annotation, the total number of knowledge base molecules known to be associated with that function, the total number of functions/pathways/lists eligible molecules and the total number of genes in the Reference Set (IPA tutorial).

To determine if miRNAs differentially expressed in sALS have miRNA recognition elements within the NFL mRNA 3′UTR we used two miRNA target prediction sites: microRNA.org (http://​www.​microrna.​org/​microrna/​getGeneForm.​do) [49] and TargetScan (http://​www.​targetscan.​org/​) [45].

Three reporter plasmids were constructed by inserting human NFL 3′UTR of different lengths between NheI and SalI sites downstream of the firefly luciferase gene in the pmirGLO vector (Promega, Madison, WI, USA; pmirGLO-NFL 3′UTR). NFL 3′UTR-Short (−S) encoded the shortest predicted isoform homologous to the murine NFL mRNA (GenBank NM_010910) consisting of 1–286 b of the 3′UTR. NFL 3′UTR-Medium (−M) included bases 1–1380 (GenBank BC039237), and NFL 3′UTR-Long (−L) comprised bases 1–1838 (GenBank NM_006158). Mutations in two nucleotides of each MRE of the NFL 3′UTR were introduced using QuikChange Lightning Site-Directed Mutagenesis kit (Agilent Technologies Canada Inc., Mississauga, ON, Canada) according to the manufacturer’s instructions. Mutagenesis was designed carefully preventing broad changes in the NFL 3′UTR secondary structure using the mfold web server (http://​mfold.​rit.​albany.​edu/​?​q=​mfold/​RNA-Folding-Form) [50]. ΔG values for the interaction among each miRNA and MREs were calculated using RNAhybrid server [51].

To perform functional analysis, HEK293T cells were maintained in Dulbecco’s modified Eagle’s medium (DMEM) containing 10% FBS and plated 24 h prior to transfection in 96-well plates at 9 × 10³ cells/well. Cells were co-transfected with 100 nM of pre-miRNAs (Life Technologies Inc., Ambion) and 3.47 fmol of pmirGLO-NFL 3′UTR (−S, -M or -L) using Lipofectamine 2000 reagent (Life Technologies Inc., Invitrogen, Burlington, ON, Canada), according to the manufacturer’s instructions.

24 h after transfection total RNA was isolated using TRIzol reagent (Life Technologies Inc., Ambion) according to the manufacturer’s instructions. After the reverse transcription, relative quantitative PCR was carried out co-amplifying Firefly and Renilla luciferases encoded in the same pmirGLO vector using the following primers: fLuc: forward 5′ CAA GAC TAT AAG ATT CAA TCT GCC CTG CTG 3′ and reverse 5′ GAT GTT GGG GTG TTG CAG CAG GAT 3′; rLuc: forward 5′ GAG CAA CGC AAA CGC ATG ATC ACT G 3′ and reverse 5′ TTC AGC AGC TCG AAC CAA GCG GT 3′.