SEARCH

SEARCH BY CITATION

Keywords:

  • amyotrophic lateral sclerosis;
  • let-7b;
  • microRNAs;
  • miR-663;
  • TDP-43

Abstract

  1. Top of page
  2. Abstract
  3. Introduction
  4. Results
  5. Discussion
  6. Materials and methods
  7. Acknowledgements
  8. References
  9. Supporting Information

TDP-43 has recently been described as the major component of the inclusions found in the brain of patients with a variety of neurodegenerative diseases, such as frontotemporal lobar degeneration and amyotrophic lateral sclerosis. TDP-43 is a ubiquitous protein whose specific functions are probably crucial to establishing its pathogenic role. Apart from its involvement in transcription, splicing and mRNA stability, TDP-43 has also been described as a Drosha-associated protein. However, our knowledge of the role of TDP-43 in the microRNA (miRNA) synthesis pathway is limited to the association mentioned above. Here we report for the first time which changes occur in the total miRNA population following TDP-43 knockdown in culture cells. In particular, we have observed that let-7b and miR-663 expression levels are down- and upregulated, respectively. Interestingly, both miRNAs are capable of binding directly to TDP-43 in different positions: within the miRNA sequence itself (let-7b) or in the hairpin precursor (miR-663). Using microarray data and real-time PCR we have also identified several candidate transcripts whose expression levels are selectively affected by these TDP-43–miRNA interactions.


Abbreviations
DYRK-1A

dual-specificity tyrosine-(Y)-phosphorylation regulated kinase 1A

EPHX1

epoxide hydrolase

GAPDH

glyceraldehyde-3-phosphate dehydrogenase

GST

glutathione S-transferase

LAMC1

laminin, gamma 1 (formerly LAMB2)

miRNA

microRNA

siRNA

short inhibitory RNA

STX3

syntaxin 3

VAMP3

vesicle-associated membrane protein 3

Introduction

  1. Top of page
  2. Abstract
  3. Introduction
  4. Results
  5. Discussion
  6. Materials and methods
  7. Acknowledgements
  8. References
  9. Supporting Information

TDP-43 is a protein belonging to the hnRNP class of nuclear factors that has been described to play a role in a variety of cellular processes, including gene transcription, pre-mRNA splicing and mRNA stability [1]. Recently, it has been identified as the major protein component of neuronal inclusions in neurodegenerative diseases such as frontotemporal dementias and amyotrophic lateral sclerosis [2]. The impact of TDP-43 in the neurodegeneration field has been so pervasive that disease nomenclature consensus is currently being modified to reflect the new clinical and pathological findings originating from recent research better [3,4]. This finding has promoted studies to characterize better the functional role(s) played by this protein inside the cell. As a result, apart from its historical involvement in splicing and transcription [5–7], several recent observations have successfully highlighted new biological characteristics of this protein, such as acting as a neuronal response activity factor and an in vitro mRNA translational repressor [8], an mRNA stability factor for neurofilaments [9,10] and as a regulator of Rho family GTPase expression [11] and HDAC6 [12]. All of these observations may be conducive to understanding the potentially pathogenic role of TDP-43 in neurodegeneration.

With regards to the wider biological properties of TDP-43, a new indication has been provided by its presence in both the human and the mouse microprocessor complexes, suggesting a potential involvement in microRNA (miRNA) biogenesis [13,14]. Further support for a role in miRNA biogenesis for TDP-43 is its localization in perichromatin fibres [15], a nuclear region specifically associated with this process [16]. The Drosha nuclear complex is one of the key enzymes involved in the biogenesis of miRNAs and has the function of converting pri-miRNA molecules to ∼70 nucleotide-long pre-miRNA molecules, which are then exported to the cytoplasm and further processed in mature miRNAs by Dicer [17]. These small RNA molecules can then bind to their target mRNAs through sequence complementarity and affect gene expression by regulating either mRNA levels or translation [18–21]. Recently, hnRNP proteins were shown to be involved in miRNA processing [22,23]. It was therefore of interest to investigate the consequences on the cellular miRNA population of removing TDP-43.

Results

  1. Top of page
  2. Abstract
  3. Introduction
  4. Results
  5. Discussion
  6. Materials and methods
  7. Acknowledgements
  8. References
  9. Supporting Information

An analysis of Drosha levels by western blot in TDP-43-depleted Hep-3B cells did not reveal any significant changes in Drosha migration pattern or signal intensity with respect to mock-treated cells (Fig. 1A), pointing to specific miRNA targets for TDP-43. To investigate this possibility, miRNA profiling in TDP-43-depleted Hep-3B cells from three independent samples was performed by Exiqon (Vedbaek, Denmark). The microarray experiment tested for 607 known and proprietary miRNA sequences (438 and 169, respectively). In this triplicate experiment, 146 miRNA sequences could be detected in our samples and 90 of these miRNA signatures could be quantitatively tested in all three short interfering RNA (siRNA) and control experiments (a list of the 67 registered ones can be found in Fig. S1). The eight miRNAs that were either down- or upregulated in a statistical significant manner following depletion of TDP-43 in Hep-3B cells are shown in Fig. 1B. For the three most statistically significant miRNAs (let-7b, miR-663 and miR-744), the results were validated using the commercial miRvana kit, which is based on a hybridization procedure with small radioactive probes based on the miRNA of interest (Fig. 1C, D). In this experiment, the changes in these miRNA expression levels as detected by the microarray experiment were confirmed in three cell lines: HeLa (adenocarcinoma), Hep-3B (hepatocarcinoma) and SH-S-5Y (neuroblastoma).

image

Figure 1.  Effect of TDP-43 depletion on Drosha and selected miRNA expression levels in Hep-3B cells. (A) Western blot assay of Hep-3B cells treated with a control siRNA (mock) and a specific TDP-43 siRNA (siRNA). The protein extracts were normalized by Coomassie intensity (lower panel) and hybridized with a polyclonal antibody against TDP-43 and a rabbit polyclonal antibody against Drosha. (B) Heat map showing all the miRNAs (< 0.05) differentially expressed in TDP-43-depleted Hep-3B cells with respect to mock-siRNA-treated cells. The blue labels indicate downregulated miRNAs, the red labels indicate upregulated ones. The clustering is reported as log2(Hy3/Hy5) ratios. (C) TDP-43 knockdown levels achieved in three cell lines: HeLa, Hep-3B and SH-SY-5Y cells. (D) Quantification of let-7b, miR-663 and miR-744 expression levels in HeLa, Hep-3B and SH-SY-5Y cell lines using the commercial miRvana kit. Undigested probe (p).

Download figure to PowerPoint

As microarray experiments represent an indirect way of measuring TDP-43 effects on the general miRNA population, it was not possible, on the basis of these data alone, to rule out the possibility that a lack of TDP-43 may have affected the levels or activity of another factor involved in miRNA processing (for example, hnRNP A1 or other miRNA processing factors). Therefore, in order to establish a direct link between TDP-43 and any of these miRNAs, we focused on TDP-43 RNA binding properties that have been previously characterized in our laboratory [24,25].

Looking at the miRNA sequences it was interesting to note that let-7b contained in its sequence a discrete number of (GU)n repeats, the preferred target sequence of TDP-43 [24] (Fig. 1C). A band shift analysis performed using recombinant GST–TDP-43 confirmed that both the let-7b and the let-7b hairpin sequence (Fig. 2A) could bind these sequences (Fig. 2B). Most interestingly, variations in the levels of both let-7a and let-7c did not appear to be statistically significant in the microarray assay (Fig. 2C). By comparing the let-7a, -7b and -7c sequences (Fig. 2D, upper panel) we observed that a critical guanosine residue in the let-7b sequence at position +17 had the effect of creating a new GU repeat, suggesting that this miRNA could be particularly sensitive to TDP-43 cellular levels as opposed to the other let-7 family members. A band shift experiment using labelled let-7a, -7b and -7c sequences confirmed that recombinant GST–TDP-43 could only bind the let-7b sequence (Fig. 2D, lower panel). The critical importance of the +17 residue is highlighted by the observation that introducing a + 17a > g substitution in the let-7a sequence can promote TDP-43 binding (Fig. 2D, lower panel). It should be noted that the importance of this critical residue has also been confirmed using pulldown analysis by immobilizing these miRNA sequences on adipic acid dehydrazide beads and incubating with total HeLa nuclear extracts. The results of this experiment confirmed that introducing a + 17a > g nucleotide in the let-7a sequence gave it the ability to bind TDP-43, even in the presence of all other nuclear competing proteins (Fig. S2).

image

Figure 2.  Specific interaction of TDP-43 with let-7b. (A) Schematic diagram of the let-7b miRNA sequence and of its precursor hairpin. (B) Band shift assay with recombinant GST–TDP-43 using the labelled let-7b sequence itself (left) and the let-7b hairpin element (right). (C) Heat map profile for all detected members of the let-7 family found in our assay, together with their statistical significance. (D) The upper panel shows the sequence comparison (the GU dinucleotides are highlighted in bold), the lower panel shows a band shift analysis of labelled let-7a, let-7a+17a>g, let-7b and -7c miRNA sequences incubated with recombinant GST–TDP-43.

Download figure to PowerPoint

We also examined the sequences of all the other miRNAs, and noted that within the sequence of the miR-663 precursor (the second most statistically affected miRNA after let-7b) there was an almost perfect GU repeated sequence localized in the apical portion of the hairpin (Fig. 3A). A band shift analysis with recombinant TDP-43 confirmed binding to the precursor hairpin, but not to the miR-663 sequence itself (Fig. 3B, left and central panels, respectively). Deletion of the GU-rich sequence in the hairpin also abolished TDP-43 binding (Fig. 3B, right panel). Finally, neither the miR-744 sequence and its hairpin (Fig. S3) nor all the other identified miRNA sequences could bind TDP-43 in band shift analyses (data not shown). These data are consistent with the observation that the sequence of this miRNA does not contain a sufficient number of (ug)n repeats.

image

Figure 3.  Interaction of TDP-43 with miR-663 and functional analysis. (A) Potential TDP-43 binding site to the miR-663 precursor hairpin element (highlighted in bold). (B) Band shift assay with recombinant GST–TDP-43 using the labelled miR-663 sequence itself (left), the miR-663 hairpin element (middle) and a miR-663 gucugugu-deleted sequence (right). (C) Potential TDP-43 binding site to the miR-574-5p sequence and the sequence of its precursor hairpin element. (D) Band shift assay with recombinant GST–TDP-43 using the labelled miR-574-5p sequence itself (left) and the miR-574-5p hairpin element (right).

Download figure to PowerPoint

From a TDP-43–miRNA interaction point of view, these results also suggest that there may be several other potential miRNA targets of TDP-43 that could not be detected in our analysis because they were not expressed at sufficient levels (or at all) in Hep-3B cells. In order to obtain some indication in this regard, we examined the primary sequence of all known miRNAs present in miRBase for GU-repeated regions. This analysis identified two other miRNAs that could potentially bind TDP-43: miR-574-5p in the miRNA sequence itself (Fig. 3C) and miR-558 in the hairpin element (Fig. 4A). Nothing is known regarding the expression profile or importance of these miRNAs, with the exception of miR-558, which has been described to be transiently upregulated in fibroblasts following high doses of radiation [26]. Band shift assays confirmed that TDP-43 could bind efficiently to the miR-574-5p sequence (Fig. 3D, left) but, unlike let-7b, could not bind anymore to the miR-574-5p sequence when it was embedded in the RNA secondary structure (compare Figs 2B and 3D, right). The reason for this probably resides in the inability of TDP-43 to compete for RNA secondary structure formation in the miR-574-5p sequence. This structure, in fact, is more extended and GC-rich than the corresponding let-7b structure element. As expected, TDP-43 could bind to the miR-558 hairpin sequence, but not to the miR-558 miRNA (Fig. 4B).

image

Figure 4.  Interaction of TDP-43 with miR-558 and miR-574-5p. (A) Potential TDP-43 binding site to the miR-558 sequence and the sequence precursor hairpin element. (B) Band shift assay with recombinant GST–TDP-43 using the labelled miR-558 sequence itself (left) and the miR-558 hairpin element (right). (C) Schematic diagrams of the constructs pGL3, pGL3-mir-let-7b and pGL3-mir-663. Each construct contained four copies of the complementary target sequence of let-7b and miR-663, respectively. (D) Results of a luciferase assay performed on TDP-43-depleted and mock-depleted Hep-3B cells following transfection of these constructs. In this type of experiment, the level of the interaction between the endogenous let-7b and miR-663 and the expression vector determined the levels of luciferase expression. Transfection efficiencies were normalized using the Renilla luciferase internal control. Standard deviation values from three independent experiments are indicated.

Download figure to PowerPoint

In order to confirm the functional significance of the TDP-43 let-7b/miR-663 interactions we then used a heterologous assay based on a luciferase reporter. Four complementary target sequences for let-7b and miR-663 were subcloned in the pGL3 vector, to obtain pGL3-mir-let-7b and pGL3-mir-663 (Fig. 4C). Both constructs, together with a pRL-TK Renilla luciferase vector, were transiently transfected into Hep-3B cells and assayed for luciferase activity in both the presence (mock) or the absence (siRNA) of TDP-43 according to the manufacturer’s instructions. The results were normalized according to the firefly/Renilla luciferase ratios obtained in each sample. As expected, no significant difference could be detected in the firefly/Renilla ratios of the pGL3 empty vector following knockdown of TDP-43 in Hep-3B cells (Fig. 4D, left). However, a significant increase in reporter gene activity was observed following transfection of the pGL3-mir-let-7b sequence following TDP-43 knockdown (Fig. 4D, centre). This is the result that should have been expected if depletion of TDP-43 was associated with lower expression levels of let-7b (as this would have meant lower translational inhibition on the pGL3-mir-let-7b construct). Exactly the opposite effect was observed when we transfected the pGL3-mir-663 construct in depleted or control cells (Fig. 4D, right). Also, this result was completely consistent with increased miR-663 expression following TDP-43 depletion, as such an outcome would have caused a higher translational inhibition on the pGL3-mir-663 construct. One important issue that should be mentioned is the fact that these two GU-rich regions in the let-7b miRNA and miR-663 do not exactly match the optimal TDP-43 binding consensus represented by perfect GU-repeated sequences and this may well explain why in both cases TDP-43 has only modulating effects on their expression rather than an all or nothing phenomena.

Most importantly, it was interesting to determine the potential consequences of these changes in terms of cellular transcript alterations. It was originally thought, in fact, that miRNA-mediated regulation was mainly at the level of translation and not at the level of mRNA degradation. It is now clear, however, that this view is only partially correct and that, depending on a variety of factors still only partially understood, many miRNA targets are regulated by degradation (as recently reviewed by Nilsen [20]). This has enabled the identification of miRNA targets by mRNA microarray analysis but, of course, it still remains very difficult to determine the proportion of mRNA targets affected in this way as opposed to strictly translation regulatory pathways (at least until large-scale proteomic approaches reach the level of sensitivity now available for mRNA microarray approaches).

Keeping in mind these limitations, we took advantage of our previously determined microarray evaluation of the cellular transcripts that were either down- or upregulated following TDP-43 knockdown in HeLa cells [27]. These transcripts (a total of 786) were compared with a set of transcripts (numbering 838) that have been observed to be downregulated following let-7b overexpression in a culture of primary human fibroblasts and which contained a let-7b seed target region in their 3′ UTRs [28]. The 23 common hits between the two lists are reported in Table 1. First of all, it should be noted that in the microarray experiment, 16 of the 23 hits were upregulated following TDP-43 removal. This situation was therefore largely consistent with the downregulatory effect on let-7b expression levels following TDP-43 removal (Fig. 1B). More interestingly, among the most upregulated transcripts were several with a potentially important function in neuronal and synapse development: the dual-specificity tyrosine-(Y)-phosphorylation regulated kinase 1A (DYRK-1A), syntaxin 3 (STX3), the vesicle-associated membrane protein 3 (cellubrevin; VAMP3) and laminin, gamma 1 (LAMC1, formerly LAMB2). Interestingly, this list also contained the enzyme cyclin-dependent kinase 6, which we previously found to be upregulated following TDP-43 removal [27]. Upregulation of these transcripts was confirmed by real-time PCR (Figs 5A, 6A) using six independent siRNA knockdown and siRNA control batches. The results showed that all these transcripts were significantly upregulated from a minimum of 1.7- to 3-fold following TDP-43 removal (Fig. 5A). In parallel to this analysis we wanted to rule out the possibility that upregulation of these transcripts could be due to changes in their mRNA splicing profiles owing to the presence of several putative TDP-43 binding sites in their intronic elements (Fig. 5B). Normal RT-PCR analysis of the coding regions, however, also ruled out this possibility by showing that the splicing profile of these transcripts did not specifically change following TDP-43 removal (Fig. 5C).

Table 1.   List of altered cellular transcripts in TDP-43 knockdown experiments that have also been found to be downregulated following let-7b overexpression.
GeneAccession numberFull nameMicroarray variationa
  1. Fold expression difference according to Ayala et al. [27].

ADRB2NM_000024Adrenergic, beta-2-, receptor, surface+1.6
IGFBP3NM_000598Insulin-like growth factor binding protein 3+2.3
IL6NM_000600Interleukin 6 (interferon, beta 2)−1.1
IGF1RNM_000875Insulin-like growth factor 1 receptor+1.2
CDK6NM_001259Cyclin-dependent kinase 6+10.0
DAB2NM_001343Disabled homolog 2, mitogen-responsive phosphoprotein (Dros.)+1.6
DYRK1ANM_001396Dual-specificity tyrosine-(Y)-phosphorylation regulated kinase 1A+4.6
CSNK1ENM_001894Casein kinase 1, epsilon−1.5
TNPO1NM_002270Transportin 1+1.1
LAMC1NM_002293Laminin, gamma 1 (formerly LAMB2)+3.1
STX3NM_004177Syntaxin 3+10.5
CALD1NM_004342Caldesmon 1+1.1
VAMP3NM_004781Vesicle-associated membrane protein 3 (cellubrevin)+1.4
SMC1ANM_006306Structural maintenance of chromosomes 1A−1.2
CAP2NM_006366CAP, adenylate cyclase-associated protein, 2 (yeast)+1.3
KIAA0152NM_014730KIAA0152−1.4
PHF16NM_014735PHD finger protein 16+1.1
RHOBTB3NM_014899Rho-related BTB domain containing 3+1.5
HSD17B11NM_016245Hydroxysteroid (17-beta) dehydrogenase 11−2.6
TOB2NM_016272Transducer of ERBB2, 2+1.8
CDV3NM_017548CDV3 homolog (mouse)+1.5
SLC5A6NM_021095Solute carrier family 5 (sodium-dependent vitamin transp.), mem 6−3.2
ZC3H12ANM_025079Zinc finger CCCH-type containing 12A−1.2
image

Figure 5.  Real-time PCR levels of let-7b regulated transcripts. (A) Real-time PCR quantification analysis of the DYRK1A, LAMC1, STX3 and VAMP3 transcript levels following TDP-43 knockdown in HeLa cells based on the results of Table 1. Six independent experiments were analysed and both standard deviations and P-values are shown for each transcript. (B) Schematic diagram of the intron/exon architecture of these genes with the presence of potential TDP-43 binding motifs, (ug)6, indicated. (C) Standard RT-PCR of each transcript to rule out the effects of TDP-43 on their RNA splicing process.

Download figure to PowerPoint

image

Figure 6.  Real-time PCR levels of let-7b and miR-663 regulated transcripts. (A) Real-time PCR quantification analysis of the EPHX1 transcript levels following TDP-43 knockdown in HeLa cells based on the results of Table 2. Six independent experiments were analysed and both standard deviations and P-values are shown for each transcript. (B) Schematic diagram of the intron/exon architecture of these genes with the presence of potential TDP-43 binding motifs indicated. (C) Standard RT-PCR of each transcript to rule out the effects of TDP-43 on their RNA splicing process. (D) Measurement by real-time PCR of the let-7b and miR-663 precursor levels following TDP-43 depletion and mock depletion in HeLa cells. Standard deviations are shown above each bar and P-values are indicated.

Download figure to PowerPoint

In the case of miR-663, no data are currently available regarding the variation in cellular transcripts following its overexpression/removal. In order to find an alternative solution, our list of microarray targets following TDP-43 removal was compared with a list of more than 1000 putative miR-663 targets obtained using the miranda software and downloaded from miRBase (http://microRNA.sanger.ac.uk/). Only three putative common transcripts were identified through this comparison (Table 2). It can be seen that in this reduced sample obtained by indirect methods we had two cases that showed the expected decrease in transcript levels that could follow miR-633 increase due to TDP-43 depletion. We have analysed in more detail the enzyme epoxide hydrolase (EPHX1) because of its putative role as an antagonist of oxidative stress [29]. The decrease in EPHX1 levels was confirmed by real-time PCR (Fig. 6A) and RT-PCR ruled out any effect of TDP-43 removal on the splicing process of this enzyme (Fig. 6B–C).

Table 2.   List of altered cellular transcripts in TDP-43 knockdown experiments that also represent putative miR-663 targets.
GeneAccession numberFull nameMicroarray variationa
  1. Fold expression difference according to Ayala et al. [27].

EPHX1NM_000120Epoxide hydrolase 1−2.1
CDANM_001785Cytidine deaminase+2.6
AAMPNM_001087Angio-associated, migratory cell protein−2.3

Finally, we also began to investigate the regulatory pathways that may be controlled by TDP-43. At least for TDP-43, we decided to measure the pre-miRNA levels in TDP-43-depleted and mock-depleted cells. For this reason, we measured the levels of pri-let-7b miRNAs according to established protocols [30]. As shown in Fig. 6D, upper panel, following TDP-43 removal, the levels of pri-let-7b were significantly increased to a level that was comparable with the loss of mature let-7b miRNA within the cell. Moreover, these changes were statistically significant. These results demonstrate that TDP-43 actively participates in the Drosha processing mechanisms and its absence in the case of let-7b leads to a block in the maturation of pri-let-7b miRNA. Finally, we also measured the levels of pri-miR-663 using a similar procedure. In this case, however, the difference in miR-663 precursor levels did not reach statistical significance (Fig. 6D, lower panel).

Discussion

  1. Top of page
  2. Abstract
  3. Introduction
  4. Results
  5. Discussion
  6. Materials and methods
  7. Acknowledgements
  8. References
  9. Supporting Information

The biological function of TDP-43 in the eukaryotic cell is far from being fully understood. Even more obscure is its role in the pathogenesis of amyotrophic lateral sclerosis/frontotemporal lobar degeneration and other neurodegenerative diseases. In particular, several gain- or loss-of-function mechanisms have been put forward in recent times. The gain-of-function mechanisms focus on the generation of potentially toxic C-terminal fragments [31–33], its toxicity in a yeast cellular model [34] and increased aggregation properties in the presence of missense mutations in the C-terminal region [35]. On the other hand, loss-of-function mechanisms are supported by indications that TDP-43 may be playing a fundamental role in a variety of nuclear processes, such as splicing regulation [5], transcription [36], chromatin organization [37] and a variety of other processes, such as cell death and nuclear shape [27]. Loss-of-function mechanisms are also supported by a recent Drosophila animal model that has shown that removal of the fly homologue of TDP-43 can recapitulate several features of motoneuron disease [38]. These two different pathophysiological mechanisms are not mutually exclusive and may indeed take place at the same time, although determining their relative importance may be especially important with regards to planning and developing successful therapeutic strategies.

To understand these pathological processes better, it is of course important to define TDP-43 functional properties as much as possible. In this regard, the effects of TDP-43 on the miRNA population are particularly interesting, considering previous observation that TDP-43 itself is a minor component of the Drosha enzyme complex [13] and the increasing role played by aberrant miRNA expression in a variety of neurodegenerative diseases, as recently reviewed in several publications [39–43].

However, to date no studies are yet available regarding the potential role played by TDP-43 in miRNA processing. In general, Drosha-associated factors are required to help or inhibit the processing of particular subsets of miRNA molecules. Indeed, this has been shown to be the case for the p68 and p72 helicases [14] and, more recently, for the KH-type splicing regulatory protein (KSRP) protein [44]. Of course, this regulatory role is not solely confined to Drosha-associated proteins. Indeed, one of the best characterized example of miRNA regulatory proteins is represented by Lin-28, which can regulate let-7 processing [45–48] by inducing uridylation of its precursor and cause its degradation [49]. In a situation probably more similar to TDP-43, miRNA regulating properties have also been described for the well-known splicing factor hnRNP A1. This protein has been shown to regulate the expression of miR-18a by binding to the loop of pri-miR-18a and inducing a relaxation at the stem, creating a more favourable cleavage site for Drosha [22,23,50]. Our results have shown that TDP-43 has the potential to affect the levels of four miRNAs, let-7b, miR-663, miR-574-5p and miR-558, by potentially binding to their sequence and/or precursor elements (schematically summarized in Fig. 7). With regards to the potential importance of the interaction between TDP-43 and miRs 574-5p/558 a cautionary note is represented by the fact that, owing to the lack of cell lines expressing these miRNAs, we were unable to functionally validate them. Therefore, this is an issue that will have to be addressed in future studies.

image

Figure 7.  Schematic diagram of TDP-43–miRNA interactions. This figure shows a summary of TDP-43 interactions with the various miRNA sequences and precursors identified in the present study. Moreover, it summarizes the effects of its removal on miRNA expression levels and on potentially important transcripts for neuronal development or degeneration.

Download figure to PowerPoint

We then asked what kind of processing steps in the biogenesis of these miRNAs may be affected. In the case of the let-7b family, the data that let-7a, which originates from the same precursor as let-7b, is not affected by TDP-43 support that the regulation is post-transcriptional. In particular, the observation that TDP-43 depletion leads to an increase in pri-let-7b levels suggests that for this miRNA, TDP-43 helps to keep/recruit the pri-miRNA sequences in place during Drosha processing. In the case of miR-663, we should consider the fact that for several miRNAs, such as miR-30 and miR-21, efficient processing is dependent on the presence of a terminal loop more than 10 nucleotides long [51]. However, the measurement of miR-663 precursor levels in TDP-43 minus and mock-depleted cells has failed to find a statistically significant difference. This suggests that miR-663 regulation by TDP-43 may take place in steps subsequent to Drosha cleavage, an observation that may be consistent with the opposite effect of TDP-43 on miR-663 levels (upregulated) as opposed to let-7b (downregulated).

The function of these different up- or downregulatory mechanisms is, of course, still an open question. The most probable explanation is that there might be two sets of transcripts whose expression has to be upregulated (in the case of let-7b) and downregulated (in the case of miR-663) at the same time to achieve a functionally specific effect. At the moment, identifying these hypothetical effects is hampered by our incomplete knowledge of TDP-43 general functions and its expression regulation within the cell (especially in normal, nonpathological conditions).

With regards to the miRNA we have identified, nothing is known about the functions of miR-663, miR-558 and miR-574-5p. On the other hand, the let-7b family is an abundant, highly conserved family of miRNAs that are important in cellular differentiation processes and their misregulation may lead to cancer formation, as recently reviewed by Roush and Slack [52]. However, Drosophila let-7 has been described as being essential for correct neuromuscular development in the transition from larva to adult [53], suggesting that members of this family may also participate in neuronal and developmental processes.

In keeping with this hypothesis, we provide evidence that the removal of TDP-43 from the cell nucleus causes specific downregulation of let-7b, and this can in turn influence the expression levels of several potentially important transcripts involved in neurodegeneration and synapse formation (Fig. 7). These transcripts include DYRK1A, a kinase that has been found to be upregulated in patients affected by Down syndrome and whose increased expression correlates with the neuronal defects [54,55]. They also include components of synapse formation, such as STX3, which is important for the growth of neurite processes [56], and VAMP3, which can functionally substitute for synaptobrevin in synaptic exocytosis [57]. The upregulation of LAMC1, on the other hand, is particularly interesting in light of previous observations that dysmorphic nuclear shape phenotypes are produced upon removal of TDP-43 [27]. Finally, another interesting transcript that is downregulated following TDP-43 knockdown (but this time due to miR-663 upregulation) is represented by the EPHX1 enzyme, a detoxifying enzyme that functions to regulate oxidative stress and has been previously shown to be significantly elevated in the hippocampal region of patients suffering from Alzheimer’s disease [29].

Taken together, these results provide an experimental basis suggesting that TDP-43 can play a role in miRNA expression pathways. Of course, how these changes relate to TDP-43′s other normal biological properties (splicing, transcription, mRNA export/translation) and, most importantly, to an eventual disease context, will require future analyses. Finally, as TDP-43 is also a splicing factor, it will also be interesting to explore the potential role of TDP-43 in Drosha-free miRNA synthesis (miRtrons) [58]. At the moment, going through the list of miRtron genes recently compiled by Berezikov et al. [59], the consensus sequences of the small introns responsible for miRtron formation in vertebrates display a G-rich sequence at the 5′ end and a U/C-rich sequence at the 3′ end. None of these two sequences contains a number of GU repeats that may resemble (at least visually) potentially strong TDP-43 binding sites. However, it is a possibility that warrants experimental testing in the future.

Materials and methods

  1. Top of page
  2. Abstract
  3. Introduction
  4. Results
  5. Discussion
  6. Materials and methods
  7. Acknowledgements
  8. References
  9. Supporting Information

Cell culture and siRNA transfection

Hep-3B cells (ATCC, Manassas, VA, USA) were grown in Dulbecco’s modified Eagle’s medium (Gibco, Rockville, MD, USA) supplemented with 10% fetal bovine serum (Gibco), glutamine, 5% glucose and antibiotic antimytotic (Sigma, St Louis, MO, USA) in 5% CO2 at 37 °C. Two transfections using 0.1 nmol siRNA were carried out at intervals of 24–48 h and cells were collected after 24 or 48 h from the second transfection. Western blot against Drosha was performed using a commercial rabbit polyclonal antibody (Abcam, Cambridge, MA, USA).

Microarray and direct miRNA analysis

Total RNA from TDP-43 siRNA, control siRNA-treated (siCONTROL nontargeting siRNA #2) and untreated Hep-3B cells were obtained using Trizol (Invitrogen, Carlsbad, CA, USA) and cleaned up using the miRNAeasy Kit (Qiagen, Valencia, CA, USA) according to the manufacturer’s instructions. Three independent RNA batches from each category of treated and untreated cells were prepared. The RNA samples were then sent for microarray analysis to Exiqon (Denmark) [60]. The results are reported as a heat map diagram according to Eisen et al. [61]. The false discovery rate method was used for the interpretation of microarray results (607 miRNAs were analysed, at a P-value <0.05). A direct miRNA expression level analysis was carried out using the miRvana kit (Ambion, Austin, TX, USA) according to the manufacturer’s instructions.

Real-time expression profiling of miRNA precursors

In order to analyse the expression levels of hsa-mir-663, a specific TaqMan® pri-miRNA assay (Applied Biosystems, Foster City, CA, USA) was used according to the manufacturer’s instructions. Primers were designed to amplify specifically the primary precursor molecule for hsa-mir-let-7b, as described previously [30]. Sequences of primers to the hairpin-containing precursor were let-7b_for, 5′-tgaggtagtaggttgtgtggtt-3′ and let-7b_rev, 5′-gggaaggcagtaggttgtatag-3′. The TaqMan minor groove binder (MGB) probe, let-7b 5′-S-carboxyfluorescein (FAM)-agtgatgttgcccc-MGB 3′, was designed to have a 5′ FAM and an MGB at the 3′ end. The TaqMan MGB probe was synthesized by Applied Biosystems. To normalize the results, the housekeeping gene glyceraldehyde-3-phosphate dehydrogenase (GAPDH) was used. Real-time PCR was performed on a CFX96TM real-time PCR detection system (Bio-Rad, Hercules, CA, USA). PCR was performed for 15 s at 95 °C and 1 min at 60 °C for 45 cycles followed by the thermal denaturation protocol, as described previously [30]. The expression levels of hsa-mir-let-7b relative to GAPDH RNA were determined using the 2−ΔΔCT method [62].

Band shift analysis

Each miRNA sequence obtained from the miRBase resource [63] was cloned in the SacI-BamH1 restriction sites of Bls KS+ sites using sense and antisense oligonucleotides (sequences available upon request from E.B., ICGEB). The BamH1 linearized plasmids were in vitro transcribed according to standard protocols in the presence α-32P-UTP (Perkin-Elmer, Boston, MA, USA). Binding reactions with 300 ng purified GST–TDP-43 were performed in 1 × bind shift binding buffer (20 mm Hepes pH 7.9, 72 mm KCl, 1.5 mm MgCl2, 0.78 mm magnesium acetate, 0.52 mm dithiothreitol, 3.8% glycerol, 0.75 mm ATP and 1 mm GTP) and electrophoresed on a 5% polyacrylamide gel at 100 V for 1 h in 0.5 × Tris borate EDTA (TBE) buffer at 4 °C. The gel was then dried and exposed with X-OMAT autoradiographic film (Kodak, Rochester, NY, USA) for 24 h at −80 °C.

pGL3 luciferase gene reporter constructs and assays

Four complementary target sequences for the let-7b and 663 miRNAs were cloned in the XbaI site of the pGL3.1-basic vector (Promega, Madison, WI, USA) (to obtain plasmids pGL3-mir-let-7b and pGL3-mir-663, respectively). Hep-3B cells were plated in 24-well culture plates 24 h prior to TDP-43 siRNA or control siRNA treatment. Cells were cotransfected with 120 ng each reporter construct and 80 ng pRL-TK Renilla luciferase vector (Promega) using oligofectamine (Invitrogen) for each transfection. pRL-TK Renilla luciferase activity was used to control for transfection efficiency. Twelve hours post-transfection the cells were washed twice with phosphate-buffered saline and harvested using passive lysis buffer, as described by the manufacturer. Samples were analysed for both firefly and Renilla luciferase activity by luminometry (Turner Biosystems, Sunnyvale, CA, USA, 20/20n luminometer) using dual-luciferase reporter assay reagents according to the manufacturer’s protocol (Promega) and normalized to Renilla luciferase expression. For each construct, three independent transfection experiments were performed (using triplicate samples for each experiment).

Quantitative real-time PCR analysis

Total RNA was extracted from luciferase and TDP-43 siRNAs-treated HeLa cells using Trizol reagent (Invitrogen) according to the manufacturer’s instructions. The cDNA synthesis was performed starting from 1 μg of each RNA sample using Moloney murine leukaemia virus reverse transcriptase (Invitrogen) and exameric random primers. In order to detect any genomic DNA contamination, parallel reactions for each RNA sample were performed in the absence of reverse transcriptase. The quantification of gene expression levels was performed by real-time PCR using SYBR green technology. Specific primers for DYRK1A, STX3, VAMP3, EPHX1, LAMC1 and GAPDH genes were designed using beacon designer software (Bio-Rad) (sequence available upon request from E.B., ICGEB). The housekeeping gene GAPDH was amplified and used to normalize the results. All amplifications were performed on a CFX96™ real-time PCR detection system (Bio-Rad). The relative expression levels were calculated according to the following equations: ΔCT = CT(target) − CT(normilizer). Comparative expression level (i.e. difference between luciferase and TDP-43 siRNA-treated HeLa cells) = 2−ΔΔCT.

Acknowledgements

  1. Top of page
  2. Abstract
  3. Introduction
  4. Results
  5. Discussion
  6. Materials and methods
  7. Acknowledgements
  8. References
  9. Supporting Information

The authors wish to thank Samdhutta Dhir for help with the bioinformatics analysis. This work was supported by the Telethon Onlus Foundation (Italy) and by a European community grant (EURASNET- LSHG-CT-2005-518238).

References

  1. Top of page
  2. Abstract
  3. Introduction
  4. Results
  5. Discussion
  6. Materials and methods
  7. Acknowledgements
  8. References
  9. Supporting Information

Supporting Information

  1. Top of page
  2. Abstract
  3. Introduction
  4. Results
  5. Discussion
  6. Materials and methods
  7. Acknowledgements
  8. References
  9. Supporting Information

Fig. S1. A full list of the known miRNAs identified in the microarray screening of Hep-3B TDP-43-depleted cells.

Fig. S2. Affinity pull down analysis of various miRNA sequences.

Fig. S3. Lack of interaction between TDP-43 and miR-744 sequences.

Please note: As a service to our authors and readers, this journal provides supporting information supplied by the authors. Such materials are peer-reviewed and may be re-organized for online delivery, but are not copy-edited or typeset. Technical support issues arising from supporting information (other than missing files) should be addressed to the authors.

FilenameFormatSizeDescription
FEBS_7643_sm_FigsS1-S3.pdf387KSupporting info item

Please note: Wiley Blackwell is not responsible for the content or functionality of any supporting information supplied by the authors. Any queries (other than missing content) should be directed to the corresponding author for the article.