TBPH/TDP-43 modulates translation of Drosophila futsch mRNA through an UG-rich sequence within its 5′UTR

Nuclear factor TDP-43 is an evolutionarily conserved multifunctional RNA-binding protein associated with frontotemporal lobar degeneration (FTLD) and amyotrophic lateral sclerosis (ALS). In recent years, Drosophila models of ALS based on TDP-43 knockdown/overexpression have allowed to find several connections with disease. Among these, we have previously described that silencing the expression of its fly ortholog (TBPH) can alter the expression of the neuronal microtubule-associated protein Futsch leading to alterations of neuromuscular junction (NMJ) organization. In particular, TBPH knocked out flies displayed a significant reduction of Futsch protein levels, although minimal variation in the futsch mRNA content was observed. These conclusions were recently validated in an independent study. Together, these observations strongly support the hypothesis that TBPH might regulate the translation of futsch mRNA. However, the mechanism of TBPH interference in futsch mRNA translation is still unknown. In this work, we use EMSA experiments coupled with RNA-protein co-immunprecipitations and luciferase assays to show that TBPH interacts with a stretch of UG within the 5'UTR of futsch mRNA and translation is positively modulated by this binding. Most importantly, this function is also conserved in human TDP-43. This result can therefore represent the first step in elucidating the relationship between TDP-43, protein translation, and eventual disease onset or progression. This article is part of a Special Issue entitled SI:RNA Metabolism in Disease.


Introduction
Transactive response DNA binding protein 43 kDa (TDP-43) is a multifunctional nuclear factor that regulates expression, splicing, transport, and mRNA stability of numerous cellular genes, including its own transcript (Ayala et al., 2011;Buratti and Baralle, 2012).
Since this discovery, many animal models have been developed that aim to reproduce at least some of the aspects of these diseases by overexpressing either the wild type or mutated forms of this protein.
Importantly, all these studies have taken advantage of the fact that TDP-43 structure and function is highly conserved through evolution (Ayala et al., 2005). Accordingly, several animal models show that both knockdown and overexpression of this factor recapitulate different aspects of ALS pathology such as neuronal degeneration and locomotor alterations (Belzil et al., 2013;Lee et al., 2012;Liu et al., 2013;Wegorzewska and Baloh, 2011).
In particular, several studies have used Drosophila as a model system, because the fly TDP-43 ortholog (TBPH) is structurally and functionally homolog to the human nuclear factor (Ayala et al., 2005;Buratti et al., 2004;Romano et al., 2014b). In general, therefore, Drosophila TDP-43-related models suggest that TBPH controls phenotypes that can be associated with human ALS pathology (Romano et al., 2012).
For this reason, characterization of TBPH physiological and pathological functions can provide important insights into ALS pathophysiology. In particular, among the Drosophila genes whose expression is regulated by TBPH, there are several genes known to code for different pre-synaptic proteins. Specifically, in the presynaptic compartment the expression levels of Futsch, Syntaxin 1A (syx), Synapsin (syn) and Cysteine string protein (csp) proteins become downregulated after neuronal suppression of TBPH. This suggests that TBPH can influence synaptic transmission through direct or indirect modulation of proteins controlling microtubule network or synaptogenesis (Feiguin et al., 2009;Romano et al., 2014a).
Among these factors, the futsch gene (CG34387; FBgn0259108) was the first gene identified as mis-regulated in TBPH-null flies (Feiguin et al., 2009). In addition to the fact that futsch is the fly homolog of human MAP1B, this gene was particularly interesting because TBPH-null flies showed a significant reduction in futsch protein expression without modification of futsch mRNA levels. However, co-immunopreciptation assays demonstrated that TBPH protein can strongly interact with futsch mRNA (Feiguin et al., 2009;Godena et al., 2011). These observations suggested a direct role of TBPH in the post-transcriptional regulation of futsch expression. In this case, therefore, TBPH might be acting similarly to what demonstrated for the Drosophila Fragile X-related protein, another RNA binding protein, that works as a translational repressor of futsch to finely tune synaptic growth (Zhang et al., 2001). In addition to these data, a recent report has further supported a role of TBPH in futsch protein translation (Coyne et al., 2014). Interestingly, this study confirmed that TDP-43 interacts in vivo with futsch mRNA and showed that this interaction modulates the intra-cellular transport/localization and translation of this mRNA in fly motor neurons (Coyne et al., 2014). This study also suggested that TBPH might sequester futsch mRNA into RNP complexes and regulate its expression in motor neuron cell bodies through its controlled release (Coyne et al., 2014). Notwithstanding these concordant lines of evidence connecting TBPH/TDP-43 with the translation process, it is not yet clear whether this action is direct or not. Therefore, the aim of this work was to better characterize this process by mapping the yet unknown binding site of TBPH/TDP-43 to the futsch mRNA and validating its functional importance in the translation process.

Results
2.1. Mapping in vitro the interaction of TBPH/TDP-43 with the 5′UTR of futsch mRNA Considering the importance of 5′UTRs sequences for translational regulation of gene expression (Araujo et al., 2012), it was very likely that TBPH could play a role in the translation of the futsch mRNA by binding to this particular region.
More specifically, we observed that futsch mRNA 5′UTR contains a stretch of UGs (Fig. 1A). Considering that both TBPH and TDP-43 share the ability to bind to UG-rich sequences with high affinity, we then tested whether TBPH/TDP-43 could bind this region. To this end, we performed EMSA analysis (Electrophoretic Mobility Shift Assay) using a labeled probe containing this sequence and purified recombinant TBPH/TDP-43. This experiment clearly showed that both the fly and human TDP-43 proteins can Both wild type and ΔUG futsch 5′-UTRs including the UG-rich tract (boldfaced) were amplified by PCR (gray shaded sequence). The sequence of the forward oligo (underlined) includes the T7 promoter used for either the cold or 32 P-labeled RNA transcription. (B) EMSA analysis with both recombinant GST-TDP-43 and GST-TBPH proteins shows that the proteins can bind UG-rich RNAs at comparable levels. (C) TDP-43 binds specifically the UG-rich-tract within the 5′UTR futsch RNA. The interaction of GST-TDP-43 with the 32 P-labeled UG-sequence of futsch 5′UTR (lane 2) can be successfully competed with cold wild type futsch 5′UTR RNA (lanes 3 and 4), but not by cold ΔUG futsch 5′UTR RNA (lanes 5 and 6). (C) Co-immunoprecipitation of the 5′UTR futsch RNA by TDP-43. This experiment shows the RT-qPCR analysis of 5′UTR futsch RNA immunoprecipitated by Flag-tagged TDP-43. The enrichment-fold is referred to an unrelated protein (EGFP) or to the mutant TDP-43 F/L that is unable to bind RNA (TDP-43 F/L). Significant levels of enrichment were observed for the wild type but not for ΔUG futsch 5′UTR RNA. This analysis confirms the interaction of TDP-43 with the UG-rich tract of futsch 5′UTR mRNA. bind the 5′UTR futsch RNA probe, and that such an interaction was abolished by the deletion of the UG-rich tract (Fig. 1B).
The binding specificity of TBPH for the 5′UTR of futsch was also confirmed by competition studies where increasing amounts of unlabeled wild type or ΔUG 5′UTR futsch RNA were used to compete with radioactive wild type 5′UTR futsch RNA for GST-TBPH binding (Fig. 1C). As shown in this figure, increasing amounts of wild type cold 5′UTR RNA were able to effectively interfere with the interaction between TBPH and the labeled probe. However, when ΔUG cold 5′UTR futsch RNA was used as a competitor, the band intensity corresponding to the complex formed by the GST-TBPH protein and the labeled wild type 5′UTR futsch RNA did not decrease (Fig. 1C).
Taken together, these experiments support the hypothesis that the interaction between TBPH/TDP-43 and the 5′UTR futsch RNA is specific occurs through this UG-rich sequence.

RNA-protein pull down
In order to verify the interaction between both fly and human TDP-43 orthologs with futsch mRNA 5′UTR, we then performed RNA-protein pull down assays.
In the first experiment, GST-TBPH or GST proteins were initially incubated with Glutathione resin. Then, in vitro transcribed wild type 5′UTR (futsch 5′UTR wt) or ΔUG 5′UTR (futsch 5′UTR ΔUG) RNAs were added. After several washes followed by RNA purification, RT-qPCR was used to quantify the amount of wild type or ΔUG 5′UTR RNA pulled down by TBPH. The enrichment-fold was normalized to the GST protein. As shown in Fig. 2A, significant levels of enrichment (30x) were observed for the wild type RNA sequence and not for ΔUG RNA sequence.
Subsequently, this specific interaction was also tested using the human TDP-43 ortholog (Fig. 2B,upper panel). In this experiment, HEK293 cells were transfected with pFLAG-TDP43, pFLAG-TDP43 F/L or pEGFP-N1 vectors. Then, cell extracts expressing comparable levels of recombinant proteins were used for RNA-protein co-immunoprecipitation assays (Fig. 2B, lower panel). In this case, the enrichment-fold was referred to an unrelated protein (EGFP) or to the mutant TDP-43 F/L (TDP-43 F/L). This mutant was used because it carries two F/L substitutions (F147L/F149L) into the first RNA recognition motif (RRM1) that abolish the TDP-43 RNAbinding ability (Buratti and Baralle, 2001). Also in this case, significant levels of enrichment were observed for only the wild type 5′UTR RNA sequence and not for the ΔUG 5′UTR RNA (Fig. 2B). As expected, no enrichment was observed for an unrelated gene (RPL13a mRNA). On the other hand, when the preferred TDP-43/ TBPH (UG) 9 RNA sequence was added to extract samples a high level of enrichment was observed for the TDP-43 wild type protein as opposed to the TDP-43 F/L mutant (data not shown).
Taken together, these experiments support the hypothesis that both Drosophila and human TDP-43 orthologs interact specifically with the UG-rich tract of the futsch 5′UTR RNA.

Futsch 5′UTR modulates translational efficiency
Next, we sought to understand whether the 5′UTR UG-rich tract of futsch might directly influence translational efficiency. This was performed using a cell-based luciferase assay. To set this up, we cloned in the inducible pGL4.11 vector the Drosophila melanogaster Hsp70Ba gene promoter (in order to get a controlled and uniform luciferase expression) with the futsch 5′UTR sequence cloned immediately upstream of the luciferase ATG-start codon. As part of this system, we generated two vectors carrying either the wild type or the ΔUG 5′UTR sequence, as negative control (Fig. 3A).
Initially, these constructs were transfected in the S2 Drosophila cell line and preliminary setup experiments showed that the peak of Hsp70Ba promoter activity was observed 6 h after thermal shock (not shown). Therefore, after transfection, the induction of luciferase expression was obtained by exposing the cells to thermal shock (20 min at 37°C) followed by a 6 h restoration period at 25°C, before assaying luciferase activity. First of all, the results of the initial transfection showed that luciferase activity of the futsch 5′-UTR ΔUG construct resulted significantly lower (27%) of that from the 5′-UTR wt (Fig. 3B). This finding suggested that the (UG)rich tract could influence the translational efficiency of the downstream open reading frame. Subsequently, the wild type 5′ UTR and the ΔUG 5′UTR futsch constructs were transfected into HeLa cells along with a TBPH-wild type cDNA pFLAG-expression vector. Then, 24 h after the end of the transfection cells were heatshocked for 1 h at 42°C and firefly/renilla luciferase activities were measured after 6 h, as previously described. In this experiment, the relative ratio of firefly/renilla luciferase activity determined from cells transfected with the plasmid carrying the wild type futsch 5′UTR was set at 1 for normalization purposes. As shown in Fig. 3C, TBPH overexpression induced a statistically significant increase (22%) in the luciferase activity, and this occurred only in presence of the wild type 5′UTR futsch sequence. In fact, TBPH overexpression did not influence luciferase activity of the ΔUG 5′ UTR futsch construct (that remained significantly lower than that the wild type 5′UTR construct without TBPH overexpression, Fig. 3C).

Discussion
In this work, we report that both fly and human TDP-43 orthologs interact specifically with an UG-rich sequence within the futsch 5′ UTR mRNA and that overexpression of Drosophila TBPH can positively modulate the translational efficiency only in presence of this UG-rich sequence. In fact, using an inducible promoter (in order to modulate gene expression more finely compared to constitutive promoters), we have found that futsch 5′UTR The relative ratio of firefly/renilla luciferase activity determined from cells transfected with the plasmid carrying the wild type futsch 5′UTR was set at 1 to normalize results. The relative ratio of firefly/renilla luciferase activity was significantly lower (p o 0.05) for the ΔUG than the wild type futsch 5′UTR RNA. (C) Luciferase activity in HeLa cells. Left panel: Wild type and ΔUG 5′UTR constructs were transfected into HeLa cells, along with a TBPH-wild type cDNA pFLAG-expression vector. 24 h after transfection, cells were heat shocked for 1 h at 42°C and the luciferase activities were measured after 6 h-restore at 37°C. The relative ratio of firefly/renilla luciferase activity determined from cells transfected with the plasmid carrying the wild type futsch 5′UTR sequence was set at 1 to normalize results. All experiments were performed in triplicate. Right panel: Western blot to determine flag-TBPH overexpression. Anti-Tubulin (Tubulin) and anti-Flag (TBPH) antibodies were used to verify the levels of TBPH overexpression in cells cotransfected with wild type and the ΔUG futsch 5′UTR constructs (left panel). Optical densitometry (ratio TBPH/Tubulin bands) was used to compare the levels of TBPH expression. All experiments were performed in triplicate.
can positively regulate the expression the reporter gene only if the UG-rich sequence is present. Therefore, our data confirm and extend the hypothesis of the TBPH-dependent translational regulation of futsch gene expression and define the sequence context required to affect the translational efficiency of futsch mRNA. With respect to the human homolog of futsch, MAP1B gene, it has to be noted that the UG sequence is not conserved. However, MAP1B has been found to be associated with TDP-43-containing RNP complexes in mouse models (Sephton et al., 2011), suggesting that TBPH/TDP-43 might also play a role in the translational regulation of this transcript (although further work will be required to clarify this issue).
Most importantly, however, this property of TBPH/TDP-43 is probably not confined to just this transcript and it is very likely that several additional targets exist which might be affected at the translational level. For example, at least another presynaptic protein, Cysteine string protein (csp) (Dawson-Scully et al., 2007), might be regulated in this manner. In fact, in TBPH null flies, this synaptic marker has been found to be downregulated only at the protein level and csp mRNA levels were not modified notwithstanding its co-immunopreciptation by TBPH (Romano et al., 2014a).
Also in this respect, a recent report has identified TDP-43 as a general component of mRNP transport granules in neurons, with its involvement in the anterograde axonal transport of target mRNAs from the soma to distal axonal compartments, such as NMJ (Alami et al., 2014). This finding, along with observation that some TDP-43 mutations can alter trafficking of TDP-43 cognate mRNAs in ALS patients, has suggested that TDP-43 might support spatially appropriate translation of target mRNAs and that the alteration of this TDP-43 activity might contribute to neurodegeneration (Alami et al., 2014). In summary, altogether these observations support the hypothesis that TDP-43 might play a role in translation regulation, that is conserved through evolution and whose alteration might contribute to the ALS pathogenesis. Consistently, futsch/ MAP1B localization seems to be altered in the spinal cord of motor neurons of ALS patients (Coyne et al., 2014).
Therefore, further studies with patients affected by TDP-43 proteinopathy will be useful to better define the contribution of the translational properties of this factor in ALS pathogenesis and to test whether this activity can be modulated by TDP-43 diseasecausing mutations.
All PCR amplicons were initially cloned in pGEM-T easy vector (Promega, Madison, WI, USA) and fully sequenced.
After sonication, each sample was divided in aliquots and in vitro transcribed wild type or ΔUG 5′UTR futsch RNAs (0.1 ng) or 0.1 ng of UG 9 RNA were added. Samples (150 ml) were incubated 2 h at 4°C with 0.5 μg/ml of anti-FLAG M2 mouse monoclonal or monoclonal Anti-Green Fluorescent Protein (Sigma, St. Louis, MO, USA) antibody in PBS containing 0.1% Tween 20 and 3% Bovine Serum Albumin (Sigma, St. Louis, MO, USA). Then, 30 ml of A/G Plus agarose beads (Santa Cruz Biotechnology, Santa Cruz, CA) were added to each co-immunoprecipitation sample (2 h at 4°C in rotating wheel). After this incubation, five wash cycles (5 min/each at 4°C in rotating wheel; centrifugation at 5000 Â g for 10 min) were carried with HEGN buffer containing 0.2% DOC and 0.5 M urea. RNA extraction and reverse transcription were carried out as described in the RNA-GST pull down section.

Quantitative real-time PCR analysis
All quantitative PCRs were performed on a CFX96 real-time PCR detection system (Bio-Rad, Hercules, CA, USA), with primers specific for futsch 5′UTR or UG 9 amplicons (Godena et al., 2011). All amplifications were performed on a CFX96 Real-Time PCR Detection System (Bio-Rad, CA, USA). In order to calculate the enrichment fold, all data were initially normalized to the respective inputs. The signal was measured and represented according to fold increases compared to the control signal. The enrichment-fold is referred to pull down experiments with GST protein. The results derived from three independent immunoprecipitation experiments and error bars were used to calculate standard deviations on the normalized ratios.