|Home | About | Journals | Submit | Contact Us | Français|
Amyotrophic lateral sclerosis (ALS) is a fatal motor neuron disorder. In motor neurons of ALS, TAR DNA binding protein-43 (TDP-43), a nuclear protein encoded by TARDBP, is absent from the nucleus and forms cytoplasmic inclusions. TDP-43 auto-regulates the amount by regulating the TARDBP mRNA, which has three polyadenylation signals (PASs) and three additional alternative introns within the last exon. However, it is still unclear how the autoregulatory mechanism works and how the status of autoregulation in ALS motor neurons without nuclear TDP-43 is. Here we show that TDP-43 inhibits the selection of the most proximal PAS and induces splicing of multiple alternative introns in TARDBP mRNA to decrease the amount of cytoplasmic TARDBP mRNA by nonsense-mediated mRNA decay. When TDP-43 is depleted, the TARDBP mRNA uses the most proximal PAS and is increased in the cytoplasm. Finally, we have demonstrated that in ALS motor neurons—especially neurons with mislocalized TDP-43—the amount of TARDBP mRNA is increased in the cytoplasm. Our observations indicate that nuclear TDP-43 contributes to the autoregulation and suggests that the absence of nuclear TDP-43 induces an abnormal autoregulation and increases the amount of TARDBP mRNA. The vicious cycle might accelerate the disease progression of ALS.
In amyotrophic lateral sclerosis (ALS) and frontotemporal lobar degeneration, TAR DNA binding protein-43 (TDP-43), a nuclear RNA-binding protein, forms inclusion bodies in the cytoplasm and disappears from the nuclei of affected neurons and glial cells (1–4). The altered intracellular distribution of TDP-43 induces the alteration of TDP-43 function, which might underlie the molecular pathogenesis of ALS. Regulating its own amount is one of the functions of TDP-43 (5–8). Thus, it has been speculated that, in ALS, the mislocalization of TDP-43 influences the amount of TDP-43 through an autoregulatory mechanism (2,9). The homeostasis of the amount of TDP-43 is crucial for cell function and survival. Excess TDP-43 forms inclusion bodies, resulting in cell dysfunction and death in animal models (10–12), whereas a reduced amount of TDP-43 induces widespread dysregulation of mRNA metabolism in cultured cells and mouse brain (8,13–16). Two hypotheses have been proposed for the status of the TDP-43 autoregulation in ALS (2). If nuclear TDP-43 contributes to the autoregulation, loss of nuclear TDP-43 leads to increasing its biosynthesis and then TDP-43 accumulates, resulting in the formation of inclusion bodies. If cytoplasmic TDP-43 contributes to the autoregulation, ectopic cytoplasmic localization of TDP-43 decreases the TDP-43 synthesis and finally resulting in a disappearance of TDP-43 from the nucleus. However, it is not clear whether nuclear or cytoplasmic TDP-43 regulates the amount of TDP-43 (2). Moreover, it is unclear whether the the autoregulation of TDP-43 is preserved in the affected motor neurons with ALS. To resolve these issues, we have to elucidate the autoregulatory mechanism of TDP-43.
The autoregulation of TDP-43 is conducted by the TARDBP mRNA, which has three polyadenylation signals (PASs) (pA1, pA2 and pA4) and three additional alternative introns (intron 6, 7 and 8) within the last exon (Figure (Figure1A1A and Supplementary Table S1) (5,6,8). For the downregulatory mechanism, two hypotheses have been proposed and still controversial whether a nonsense-mediated mRNA decay pathway contributes to the autoregulation (5–8). Using minigene and RT-PCR analysis, Polymenidou et al. have proposed that an isoform lacking introns 6 and 7 was degraded through a nonsense-mediated mRNA decay pathway (8). Meanwhile Baralle et al. have proposed that the spliceosomal assembly and/or constitutive processing of intron 7 lead to degradation of the TARDBP mRNA through an unknown mechanism (6,7). They have also proposed a mechanism that TDP-43 inhibits transport of TARDBP mRNA from the nucleus to the cytoplasm (7). Regarding the upregulatory mechanism of TDP-43, the mechanism has not been investigated, while it has been reported that mice with a half dose of the TARDBP gene, which encodes TDP-43, displayed normal levels of TDP-43 (17). Here we performed an in-depth analysis of the autoregulatory mechanism of TDP-43 with a focus on the upregulatory mechanism in response to TDP-43 depletion. Moreover, we investigated its autoregulatory status in ALS motor neurons without nuclear TDP-43.
The full-length coding region of isoform I or isoform II of human TDP-43 cDNA (Figure (Figure1E)1E) was isolated from a human cDNA library (Clontech) and subcloned into a pcDNA DEST-40 vector (Invitrogen) (14,18). The original plasmid encoding the minigene was provided by Dr Hurng-Yi (National Taiwan University) (19). The wild-type exon 6 of TDP-43 and the enhanced-green fluorescence protein (EGFP)-fused minigene were amplified by polymerase chain reaction (PCR) with specific primers from the cDNA library and subcloned into the original minigene construct or the pEGFP-C3 vector (Clontech). The Δintron 6 or Δintron 7 minigene, AU-rich element mutant construct and spliced site mutant constructs were produced using the GeneArt site-directed mutagenesis system (Invitrogen). The primer sequences for construction of these plasmids are listed in Supplementary Table S2.
Full-length human TDP-43 coding regions tagged with myc were amplified by primers 5′-AAAGGATCCATGTCTGAATATATTCGGGTA-3′ and 5′-TTTCTCGAGCTACAGATCCTCTTCTGAGATGAGTTTTTGTTCCATTCCCCAGCCAGAAGACTT-3′, followed by subcloning into the pcDNA5/FRT/TO vector and co-transfection into Flp-In T-REx™ HEK293 cells (Invitrogen) with Flp recombinase (encoding plasmid pOG-44). We cultured Flp-In T-REx™ HEK293, SH-SY5Y and HEK293T cells in Dulbecco's modified Eagle's medium (Gibco) supplemented with 10% fetal bovine serum. For Flp-In T-REx™ HEK293 selection, Hygromycin- and blasticidin-resistant colonies were pooled and expanded. Transgene expression was induced with 1 μg/ml doxycycline (Sigma). X-tremeGene HP DNA transfection reagent (Roche) was used for transfections. Cycloheximide (Sigma), actinomycin D (Sigma) and caffeine (Sigma) was used at a final concentration of 100 μg/ml, 5 μg/ml and 10 mM, respectively. Cells were transfected with small interfering RNA (siRNA) using Lipofectamine RNAi MAX (Invitrogen) according to the manufacturer's protocol. siRNA probes used in this study are summarized in Supplementary Table S2.
Cells were lysed with RIPA buffer (25 mM Tris–HCl, pH 7.6, 150 mM NaCl, 1% NP-40, 1% sodium deoxycholate and 0.1% sodium dodecyl sulphate) supplemented with protease inhibitor cocktail (Sigma) and centrifuged at 12,000 ×g for 10 min at 4°C. The supernatant was resuspended in Laemmli Sample Buffer (BioRad). Anti-TDP-43 antibody (1:5000; Abnova), anti-TDP-43 (amino acids 3-12) antibody (for Figure Figure1I;1I; 1:5000; Cosmobio), anti-Lamin B1 antibody (1:5000; MBL), anti-myc antibody (1:10,000; Millipore), anti-GFP antibody (1:5000; MBL), anti-DsRed polyclonal antibody (1:500; Clontech), anti-Actin antibody (1:5000; Santa Cruz) or anti-GAPDH antibody (1:10,000; MBL) was used as the primary antibody and subsequently detected with HRP-conjugated secondary antibodies (1:20,000; Dako). StartingBlock T20 blocking buffer (Thermo scientific) was used for Figure Figure1I1I.
TARDBP mRNA was detected using the QuantiGene® ViewRNA ISH cell assay kit (Affymetrix) according to the manufacturer's instructions. TDP-43 tet-inducible Flp-In T-REx™ HEK293 cells were cultured on MAS-coated glass slides (MATSUNAMI glass) in the presence or absence of doxycycline for 48 h. Cells were fixed in 4% formaldehyde solution, permeabilized, treated with protease and hybridized with custom-designed QuantiGene® ViewRNA Type 4 (488 nm) probes targeting nucleotides 2055–2774 (probe set A) or nucleotides 373–1327 (probe set C) and Type 1 (550 nm) probes targeting nucleotides 2795–4198 (probe set B) of the human TARDBP mRNA (NM_007375). To visualize the hybridized probes, cells were incubated with pre-amplifier, amplifier and labeled probe. The cells were observed using a Zeiss LSM710 confocal laser-scanning microscope.
Supplementary Table S6 shows details of the patients and controls included in this study. For sporadic ALS (SALS) patients, the mutations in TARDBP and SOD1 and G4C2 expansion in C9ORF72 were excluded. This study was approved by the Institutional Ethical Review Board of Niigata University, written informed consent was obtained from the families and all clinical investigations were conducted according to the principles expressed in the Declaration of Helsinki. In situ hybridization was performed on a formalin-fixed paraffin-embedded 4-μm-thick lumbar section using a QuantiGene® ViewRNA ISH Tissue Assay kit (Affymetrix). Sections were fixed in 10% formaldehyde for 1 h, deparaffinized using Histoclear solution (National Diagnostics), boiled in pre-treatment solution (Affymetrix) for 20 min and digested with protease QF (Affymetrix) for 10 min. Sections were hybridized for 3 h at 40°C with custom-designed QuantiGene® ViewRNA Type 1 probes targeting nucleotides 373–1327 of human TARDBP mRNA (NM_007375; probe set C; Affymetrix) (Supplementary Figure S2A). Bound probes were then amplified using PreAmp1 QF and Amp1 QF molecules. Label Probe conjugated with alkaline phosphatase (Label Probe-AP) was then added, and Fast Red Substrate was used to produce signal. Slides were then counterstained with hematoxylin. Bright field and fluorescent microscopy images for each cell were taken by fluorescence microscopy (Keyence; BioRevo BZ-9000). We selected images of motor neurons in which a nucleolus was observable. The numbers of dot signals within the nucleus and the cytoplasm and the areas of the nucleus and cytoplasm were measured using Imaris software (Bitplane AG).
To examine the nuclear localization of TDP-43 and the intracellular distribution of TARDBP mRNA in the same cells, immunohistochemistry and in situ hybridization were performed using serial sections of spinal cord from SALS cases. For immunohistochemical analysis, heat retrieval using an autoclave for 10 min in 10 mM citrate buffer (pH 6.0) was applied to the sections. Immunofluorescence detection of TDP-43 was performed using a rabbit polyclonal antibody against TDP-43 (10782-2-AP, 1:3000; Proteintech). The secondary antibody used was Alexa Fluor 488 goat anti-rabbit IgG (1:300; Molecular Probes). The immunolabeled sections were counterstained with hematoxylin.
Cell fractionation was performed as described previously (20). Total RNA was extracted (Nucleospin RNAII; Takara Bio) and poly(A)+ RNA was purified (NucleoTrap mRNA Mini kit; Takara Bio) according to the manufacturer's instructions. The RNA samples were loaded on 1.5% gels (NorthernMax-Gly Kit; Life Technologies), transferred to Hybond N+ nylon membranes (GE Healthcare) and probed with internally DIG-labeled sequences following pre-hybridization in ULTRAhyb hybridization buffer (Ambion). DIG-labeled RNA probes were prepared from template DNA amplified by specific primers (Supplementary Table S2) using a DIG Northern Starter Kit (Roche). Visualization of transcripts was performed with a CDP-Star reagent (Roche). Signals were detected by an LAS4000 mini biomolecular imager (GE Healthcare). The densitometric analysis of each band was performed by ImageQuant TL analysis software (GE Healthcare).
For cultured cells, first-strand cDNA was synthesized from total RNA by ReverTra Ace (TOYOBO) with random primers and an oligo(dT) primer (for standard qRT-PCR), or with P7-t25-vn primer (for 3′-end qRT-PCR) (21). For human materials, total RNA was extracted from frozen human spinal cords using the mirVana™ miRNA isolation kit (Ambion). Supplementary Table S5 shows details of the patients and controls included in this study. RNA quality was assessed based on the RNA integrity number, as determined using the Agilent 2100 Bioanalyzer (Agilent Technologies,). First-strand cDNA synthesis was performed using the SuperScript® VILO™ cDNA Synthesis Kit (Invitrogen). Informed consent was obtained from relatives prior to the study, and ethical approval was obtained from research ethics committees of Niigata University. qRT-PCR was performed with SYBR Green Premix ExTaq II on a TP-850 Real-Time PCR Detection system (Takara Bio) using a gene-specific primer pair (for standard qRT-PCR) or a gene-specific forward primer and P7 primer (for 3′-end qRT-PCR; Supplementary Table S2). PCR conditions were 30 s at 95°C for initial denaturing, followed by 40 cycles of 5 s at 95°C and 30 s at 60°C. Expression levels of the target gene relative to RPLP1 and RPLP2 were determined using the ΔΔCT method. Primer specificity was evaluated using the BLAST program (http://blast.ncbi.nlm.nih.gov/Blast.cgi), dissociation temperature measurements and agarose gel electrophoresis analysis. The sequences of primers are listed in Supplementary Table S2.
Cell fractionation and native RIP were performed with the EZ-Magna Nuclear RIP™ (Native) Nuclear RNA-Binding Protein Immunoprecipitation Kit (Millipore) according to the manufacturer's instructions. Immunoprecipitation was performed using Dynabeads M-280 Sheep anti-Mouse IgG (catalog no. 11201D; Life Technologies), mouse anti-TDP-43 monoclonal antibodies (catalog no. 60019-2-Ig; Proteintech, or catalog no. H00023435-M01; Abnova) or mouse IgG (catalog no. CS200621; Millipore). The co-precipitated RNA were extracted with Nucleospin RNA XS (Takara Bio) and reverse-transcribed with SuperScript III (Invitrogen) using oligo(dT) primers. Synthesized cDNA was amplified with AmpliTaq Gold® 360 (Life Technologies) using specific primers (Supplementary Table S2).
To investigate the autoregulatory mechanism of TDP-43, we generated a stable Flp-In T-REx™ HEK293 cell line that contains a doxycycline-responsive transgene for the human TDP-43 complementary DNA (cDNA) coding region with a myc sequence. After 72 h of induction, expression of endogenous TDP-43 was completely suppressed (Figure (Figure1B1B and Supplementary Figure S1A). RT-PCR analysis showed that the amount of TARDBP mRNA decreased (upper panel in Figure Figure1C),1C), whereas the amount of pre-mRNA was not decreased (middle panel in Figure Figure1C),1C), indicating that the transcriptional activity of TARDBP pre-mRNA is not affected. To investigate the alteration of endogenous TARDBP mRNA, including its intracellular distribution (5,7), we extracted poly(A)+ RNA from nuclear and cytoplasmic fractions (Supplementary Figure S1B). We performed northern blot analysis using three probes: one for exons 2–4 (probe A) and two located in introns (intron 6 for probe B and intron 7 for probe C) (Figure (Figure1D1D and E). The northern blot analysis and sequencing of the RT-PCR products revealed that intron 6 had three donor sites (nucleotide positions 769, 833 and 842 in cDNA); isoform III lacked introns 6, 7 and 8; and isoform IV lacked introns 6 and 7 (Figure (Figure1E;1E; Supplementary Figure S1C and D). Isoforms III and IV have additional termination codons located more than 55 bp upstream from the last exon–exon junction (Figure (Figure1E).1E). This condition fulfills the criteria for nonsense-mediated mRNA decay (8,22).
The isoform using pA1 was the most abundant product (isoform II in lanes 1 and 5 of Figure Figure1D),1D), and the isoform using pA4 was the second most abundant product in the cytoplasm and the nucleus (I in lanes 1 and 5; Figure Figure1D).1D). When TDP-43 was ectopically expressed, isoforms I and II of the endogenous TARDBP mRNA disappeared from the cytoplasm (Figure (Figure1D1D and F). In the nucleus, isoform I showed no obvious change, whereas isoform II disappeared (Figure (Figure1D1D and F). The results of RT-PCR analysis consistent with the results of northern blotting analysis (Supplementary Figure S1C and D). In addition, we have performed RT-PCR analysis in human neuroblastoma SH-SY5Y cells with transiently expressed TDP-43 (Supplementary Figure S1E and F). The results of RT-PCR analysis also consistent with the results obtained from Flp-In T-REx™ HEK293 cell (Supplementary Figure S1E and G).
Next, we investigated whether the cytoplasmic TARDBP mRNA was degraded by nonsense-mediated mRNA decay. We performed northern blot analysis in the presence of the nonsense-mediated mRNA decay inhibitor cycloheximide (CHX) (Figure (Figure1D1D and F). With ectopic expression of TDP-43, the abundance of isoform IV was significantly increased in the cytoplasm (lane 4 in Figure Figure1D1D and F). By qRT-PCR analysis, we confirmed that the splicing isoform lacking both introns (not just intron 7) was more abundant in this condition (Supplementary Figure S2A and B). When UPF1, a central component of the nonsense-mediated mRNA decay pathway was depleted (Supplementary Figure S2C), the isoform lacking both introns was increased (Supplementary Figure S2D and E). Moreover, we treated the cells with caffeine which is a SMG1 kinase inhibitor and attenuate a nonsense-mediated mRNA decay by inhibiting the UPF1 phosphorylation cycle (23). The caffeine is more effective than UPF1 depletion on increasing in the isoform lacking both introns (Figure (Figure11 G–I; Supplementary Figure S2F and G). Finally, under these conditions, we detected the product of truncated forms of TDP-43 thought to use additional termination codons, which are translated from nonsense-mediated mRNA decay sensitive mRNAs, isoforms III and IV (Figure (Figure1J).1J). Although the previous report discounted the contribution of a nonsense-mediated mRNA decay mechanism and emphasized that the splicing of intron 7 is enough for autoregulation (5,6,8), our analysis revealed the contribution of a nonsense-mediated mRNA decay mechanism for autoregulation and that both introns are spliced in the process.
In a northern blot analysis, we found that the amount of TARDBP mRNA using pA4 relative to that using pA1 was higher in the nucleus than in cytoplasm (I and II in Figure Figure1D1D and F). To investigate the intracellular distribution of TARDBP mRNA variants, we applied quantitative in situ RNA expression technology using QuantiGene® ViewRNA (Affymetrix). This method enabled us to quantitate the amount of mRNA by the number of spots (24). We performed a double in situ hybridization study using probes targeting intron 7 (probe set A) and pA2 to pA4 (probe set B) (Figure (Figure2A).2A). The spots that are positive for both probes indicate isoform I, and the spots that are only positive for probe set A indicate isoform II. The spots positive for probe set B were more abundant in the nucleus than the spots positive for probe set A (Dox-, Figure Figure2B),2B), indicating that TARDBP mRNA using pA4 exists predominantly in the nucleus. This result was confirmed by a double in situ hybridization study using the probe set for exons 3–6 and probe set B (Supplementary Figure S3A and B). After 48 h of ectopic expression of TDP-43, TARDBP mRNA completely disappeared from the cytoplasm but not from the nucleus (Dox+, Figure Figure2B),2B), as shown in northern blot analysis. Most of these spots in the nucleus were positive for both probe sets (merged panel in Figure Figure2B2B Dox+). We further investigated the levels of TARDBP mRNA isoforms at several time points after induction of ectopic TDP-43 expression. The ectopically expressed TDP-43 (TDP-43-myc) was detected at 6 h after induction, and a decrement in endogenous TDP-43 was observed at 24 h after induction (Figure (Figure2C2C and D). The abundance of cytoplasmic TARDBP mRNA was markedly decreased at 6 h, whereas the abundance of the nuclear pA4 isoform was unchanged during the observed time points (Figure (Figure2D2D).
Next, to investigate the alteration of TARDBP mRNA in conditions of TDP-43 depletion, we used minigene constructs fused with exons from the survival of the motor neuron (SMN) gene and exon 6 of the human TARDBP gene (SMN-TDPwt) (Figure (Figure3A).3A). The transcripts derived from the minigene lack the Kozak consensus sequence and initiation codon. Therefore, a pioneer round translation, which is essential for nonsense-mediated mRNA decay, is not induced (25). By using this system, we were able to investigate the splicing variants in conditions free of nonsense-mediated mRNA decay and with depletion of the only endogenous TARDBP mRNA. We transfected the minigene into HEK293T cells under conditions of increased or decreased amount of TDP-43 (Figure (Figure3B)3B) and performed northern blot analysis using poly(A)+ RNA from nuclear and cytoplasmic fractions (Figure (Figure3C).3C). The structures of each isoform were determined by sequencing of RT-PCR and 3′-RACE products (Figure (Figure3D).3D). Ectopic expression of TDP-43 resulted in decreased abundances of isoform II and increased abundances isoforms III and IV in the cytoplasm (Figure (Figure3C,3C, left panel, and E). In the nucleus, the abundance of isoform II was decreased, but the abundance of isoform IV was not significantly increased. These results were consistent with those observed with endogenous TARDBP mRNA, suggesting that the RNA metabolism of the minigene is similar to that of endogenous TARDBP mRNA. The depletion of endogenous TDP-43 increased the abundance of isoform II and decreased the abundances of isoforms III and IV in the cytoplasm (Figure (Figure3C,3C, right panel, and F). In the nucleus, the abundance of isoform II increased and the abundance of isoforms I and III decreased. The qRT-PCR analysis confirmed that the abundances of isoforms III and IV were increased by ectopic expression of TDP-43 and decreased by depletion of TDP-43 (Supplementary Figure S4A and B).
Our results show that the splicing of introns 6 and 7 is regulated during the autoregulatory processes of TDP-43. We then investigated whether the splicing of intron 6 influences the amount of TDP-43. We used minigenes containing a part of exon 6 of the TARDBP fused with EGFP. The minigene uses the stop codon and PAS for the TARDBP gene (Figure (Figure4A),4A), allowing us to investigate the effect of 3′-UTR sequences of the TARDBP gene on the amount of EGFP fusion protein. To inhibit the splicing of the intron 6, we created substitutions of the nucleotide sequences at three donor sites and one acceptor site of intron 6 (m-intron 6) (Figure (Figure4A).4A). In this minigene, the splicing of intron 6 induced by ectopic expression of TDP-43 was inhibited (Figure (Figure4B).4B). The ectopic expression of TDP-43 reduces the abundance of EGFP fusion mRNA (Figure (Figure4B4B and C) and protein (Figure (Figure4D4D–F) in both minigenes. However, the reduction in the abundance of EGFP fusion mRNA and protein after ectopic expression of TDP-43 was significantly smaller with the m-intron 6 minigene than with the wild-type minigene (Figure (Figure4C4C and F) indicating that splicing of intron 6 enhances the reduction in the abundance of the product.
Then, we investigated whether splicing of intron 7 is necessary for splicing of intron 6. First, we investigated the correlation between the splicing of introns 6 and 7 using minigenes lacking each intron (Supplementary Figure S5A). We found that TDP-43 efficiently induced the splicing of intron 7 in the minigene lacking intron 6 (Δintron 6a and b). In contrast, the splicing of intron 6 was not induced in the minigene lacking intron 7 (Δintron 7) (Supplementary Figure S5B). Next, we inhibited the splicing of intron 7 by substitution of the nucleotide sequences at the splicing donor site for intron 7 (Figure (Figure4G).4G). In this minigene, splicing of intron 6 was also inhibited (Figure (Figure4H),4H), indicating that splicing of intron 7 is essential for splicing of intron 6.
These results indicate that splicing of intron 7 is a primary event in the autoregulatory mechanism. For splicing of intron 7, which includes pA1, the usage of pA1 should be inhibited by TDP-43, as reported by Avendano-Vazquez et al. (5). To investigate this possibility, we compared the proportion of transcripts using pA1 and pA4 in whole unspliced isoforms upon expression or depletion of TDP-43 in cells transfected with SMN-TDP minigenes (Figure (Figure3A).3A). We selectively amplified the isoform using pA1 by using 3′-end qRT-PCR with a reverse primer targeting the polyadenylation site and a forward primer in intron 7 (Figure (Figure5A)5A) (21). The ectopic expression of TDP-43 markedly decreased the proportion of the isoform using pA1 and increased the proportion of the isoform using pA4 (Figure (Figure5B,5B, left panel). In contrast, the depletion of TDP-43 increased the proportion of the isoform using pA1 and decreased the proportion of the isoform using pA4 (Figure (Figure5B,5B, right panel). Next, to exclude the possibility that splicing of intron 7, in which pA1 exists, influence the selection of pA1, we inhibited the splicing of intron 7 by substituting the splicing site of intron 7 (Figures (Figures4G4G and H; ;5C).5C). Because the 3′-end qRT-PCR method could not distinguish between isoforms using pA1 and those using pA2, we compared the proportion of isoforms extending beyond pA1 in whole unspliced isoforms upon expression or depletion of TDP-43. As expected, the ectopic expression of TDP-43 increased the proportion of isoforms extending beyond pA1 (Figure (Figure5D,5D, left panel), whereas the depletion of TDP-43 decreased it, indicating that TDP-43 directly inhibits the usage of pA1 (Figure (Figure5D,5D, right panel).
TDP-43 increased the abundance of mRNA using pA4 and decreased the abundance of cytoplasmic TARDBP mRNA. Because the 3′-UTR sequence contains a target sequence for miRNA or a putative adenylate uridylate-rich element that contributes to the stability of mRNA (25), the mRNA using pA4, which has a long 3′-UTR sequence, might be less stable than that using pA1. To explore this possibility, we compared the stability of mRNA using each PAS by halting mRNA transcription with actinomycin D in TDP-43 tet-inducible Flp-In T-REx™ HEK293 cells. We found that the rates of degradation of TARDBP mRNA using pA4 were not significantly different from those of TARDBP mRNA using pA1, regardless of the presence or absence of doxycycline (Supplementary Figure S6A). We further investigated the effect of the putative adenylate uridylate-rich element by introducing nucleotide substitutions in this element (26); however, the stability of TARDBP mRNA was unchanged (Supplementary Figure S6B). These results indicate that the stability of TARDBP mRNA is not regulated by the 3′-UTR sequences.
We found that a substantial amount of TARDBP mRNA exists in the nucleus. TDP-43 is a nuclear protein that binds to TARDBP mRNA via the TDP-43-binding region, which is located within intron 7 after the 3′-UTR sequence (6,8,15). Indeed, TARDBP mRNA using pA4 was co-isolated with TDP-43 from the nuclear fraction of HEK293T cells (Figure (Figure6A).6A). This result raises the possibility that depletion of TDP-43 facilitates the transport of TARDBP mRNA from the nucleus to the cytoplasm. By using northern blot analysis of poly(A)+ RNA derived from HEK293T cells transfected with a SMN-TDP43-wt minigene, we found that TDP-43 depletion increased the amount of cytoplasmic mRNA using pA4 relative to that of cytoplasmic mRNA using pA1 (Figures (Figures3C3C and F; ;6B,6B, left panel). The qRT-PCR assay also confirmed these results (Figure (Figure6B,6B, right panel). To investigate a possible role for the TDP-43-binding region, we compared the relative amount of cytoplasmic mRNA between the minigene lacking the TDP-43-binding region (Δintron 7) and that of wild-type cytoplasmic mRNA (Figure (Figure6C,6C, left panel). The cytoplasmic-to-nuclear ratio of RNA was higher in mRNA lacking the TDP-43-binding region (Figure (Figure6C,6C, middle and right panel).
Although TDP-43 loses the nuclear localization in affected neurons in ALS, the autoregulatory status of TDP-43 in these neurons have not been investigated. We then addressed this issue using postmortem spinal cord tissue from ALS patients. At first, we investigated the expression level of total TARDBP mRNA and TARDBP mRNA lacking both intron 6 and intron 7 in RNAs obtained from spinal cords of SALS or control. Although the amount of TARDBP mRNA lacking both intron 6 and intron 7 decreased in ALS as expected, the amount of total TARDBP mRNA did not increase as previously reported (Figure (Figure7A7A and B) (27).
Our data indicate that a depletion of nuclear TDP-43 increases cytoplasmic TARDBP mRNA, which should be translated into TDP-43 and decreases the nuclear TARDBP mRNA. This system does not just increase the total amount of mRNA, but redistributes the mRNA to increase the mRNA translated into the protein. Moreover, the number, size and status of neurons may influence the amount of the mRNA. Therefore, we attempted to investigate whether the cytoplasmic—and not nuclear—TARDBP mRNAs was increased in each spinal motor neurons. Thus, we applied quantitative in situ hybridization technology using QuantiGene® ViewRNA. We counted the number of spots for TARDBP mRNA in the nucleus and cytoplasm of spinal motor neurons from control, SALS and familial ALS patients (Figure (Figure7A7A and C) (24). Familial ALS patients included ALS1 (caused by a mutation in the SOD1 gene and shows no TDP-43 pathology) and ALS10 (caused by a mutation in the TARDBP gene and shows TDP-43 pathology) patients (4,18,28,29).
The TARDBP mRNA spots were observed in the nucleus and cytoplasm of spinal motor neurons (Figure (Figure7C7C and Supplementary Figure S7). The mean number of TARDBP mRNA spots in neurons from SALS patients decreased in the nucleus and increased in the cytoplasm compared to controls; however, the differences were not significant (Supplementary Figure S7). Because cell size is positively associated with the amount of TARDBP mRNA (Supplementary Table S4), and because it has been reported that cell size is smaller in SALS than in control neurons (30), we investigated the proportion of TARDBP mRNA in the cytoplasm. The proportion of TARDBP mRNA in the cytoplasm showed a Gaussian distribution in the controls; however, in SALS neurons, the distribution of the proportion was markedly shifted to 100 (Figure (Figure7D7D and Supplementary Table S3). The mean values of the proportion of cytoplasmic TARDBP mRNA in SALS and ALS10 were significantly larger, by 121 and 131%, respectively, than those in controls. In contrast, the mean values of the proportion of cytoplasmic TARDBP mRNA in ALS1 were similar to those in controls. We performed a stepwise multiple linear regression analysis using cell size and ALS status (control or SALS) as independent variables (Table (Table1).1). The status of SALS was selected as a negative factor for the number of nuclear TARDBP mRNA spots and a positive factor for the number of cytoplasmic or total TARDBP mRNA spots.
Finally, we compared the number of TARDBP mRNA spots between neurons with nuclear TDP-43 (unaffected neurons) and those with cytoplasmic TDP-43 inclusions (affected neurons) in SALS (Figure (Figure8A).8A). The TARDBP mRNA spots did not co-localize with cytoplasmic TDP-43 inclusions (Figure (Figure8A,8A, bottom panel). The number of TARDBP mRNA spots was not significantly different between affected neurons and unaffected neurons (Figure (Figure8B).8B). The mean value of the proportion of cytoplasmic TARDBP mRNA in affected neurons was significantly larger, by 134%, than in unaffected neurons (Figure (Figure8C8C).
We demonstrated the autoregulatory mechanism of TDP-43 under conditions of increased or depleted TDP-43 by northern blot analysis of endogenous TARDBP mRNA and using a minigene reporter assay. A model for the autoregulatory mechanism of TDP-43 is proposed in Figure Figure9A.9A. Autoregulation is performed by the coordination of selection of the polyadenylation signal and alternative splicing accompanied by regulation of intracellular distribution of TARDBP mRNA. In conditions of excess nuclear TDP-43, TDP-43 inhibits the selection of pA1 by binding to intron 7 of TARDBP pre-mRNA (5), resulting in elongation of transcripts beyond pA1. The elongated transcripts present the acceptor site of intron 7. Then, the transcript undergoes intron 7 splicing, followed by intron 6 splicing. The isoform is exported to the cytoplasm and degraded by nonsense-mediated mRNA decay (8), resulting in a decrement in the cytoplasmic TARDBP mRNA. Some transcripts that escape intron 7 splicing select pA4. The mRNA molecules that use pA4 tend to be retained in the nucleus with nuclear TDP-43 through the binding of TDP-43 to the 3′-UTR sequence beyond pA1 (5,8,15). When nuclear TDP-43 is depleted, TARDBP transcripts select pA1 and are transported to the cytoplasm, resulting in an increase in the amount of cytoplasmic TARDBP mRNA.
Two hypotheses have been proposed to explain how the autoregulatory mechanism of TDP-43 contributes to the molecular pathogenesis of ALS (2). These hypotheses differ as to whether nuclear or cytoplasmic TDP-43 contributes to the autoregulation. Our results from cultured cells indicate that the amount of TDP-43 is regulated by the amount of cytoplasmic TARDBP mRNA. In ALS, the nuclear TDP-43 decreases and cytoplasmic TDP-43 increases as inclusions. Thus, if nuclear TDP-43 contributes to the autoregulation, then the amount of cytoplasmic TARDBP mRNA increases. On the contrary, if cytoplasmic TDP-43 contributes to the autoregulation, then the amount of cytoplasmic TARDBP mRNA decreases. We found that the amount of cytoplasmic TARDBP mRNA increased in affected neurons from SALS patients, indicating that nuclear TDP-43 contributes to the autoregulation. An increased amount of cytoplasmic TARDBP mRNA may lead to an increase in TDP-43 in affected neurons in ALS. If the newly synthesized TDP-43 is recruited into cytoplasmic inclusions, then nuclear TDP-43 is consistently depleted. The continued depletion of nuclear TDP-43 in ALS would continuously upregulate the synthesis of TDP-43. This would be a vicious cycle that may accelerate the disease progression in ALS proposed by Lee et al. (2).
The quantification of the amount of mRNA in a single cell is challenging. The previous analysis using a microarray assay did not show a difference of TARDBP mRNA in spinal motor neurons between ALS and control samples (27). The difference between the previous result and ours might be due to the microarray analysis not being sensitive enough to detect the alteration of the amount of TARDBP mRNA and/or to the fact that the size of the neurons was not considered in the previous report. We applied quantitative in situ hybridization technology using QuantiGene® ViewRNA to investigate the amount and intracellular distribution on mRNA in the human spinal motor neurons (24). The result for ALS1, in which the amount of cytoplasmic TARDBP mRNA did not change, further confirmed that the findings are not simply a result of dysfunction of spinal motor neurons.
Regarding the downregulatory mechanism of TDP-43, our results confirmed the hypothesis proposed by Polymenidou et al., which is that the splicing of introns 6 and 7 leads to degradation of mRNA through a nonsense-mediated mRNA decay mechanism (8). Although they proposed the hypothesis based on the result of an RT-PCR analysis of minigene-derived RNA (8), we demonstrated by northern blot analysis that the endogenous mRNA transcripts were susceptible to nonsense-mediated mRNA decay. The autoregulatory mechanism of alternative splicing associated with nonsense-mediated mRNA decay has been reported in RNA-binding proteins (31,32). These proteins have several characteristic features in their last exon: 1) large size; 2) a nucleotide sequence that is highly conserved among species; and 3) amino acid sequences that have been predicted to have a disordered structure (31,33). TDP-43 is an RNA-binding protein, and the last exon of the TARDBP gene has these characteristic features (34).
Baralle et al. have reported that the transcripts susceptible to nonsense-mediated mRNA decay were not observed by northern blot analysis. Thus they have disagreed about the contribution of mRNA transcripts lacking intron 6 and of nonsense-mediated mRNA decay (6). However, we were able to find the transcripts susceptible to nonsense-mediated mRNA decay by northern blot analysis for several reasons. First, we used TARDBP exons 2–4 for the probe in the northern blot analysis, whereas Baralle's group used intron 7 or a unique fragment lacking both introns 6 and 7 (6). Their probes could not detect the isoforms that contribute to autoregulation because intron 7 is excised when TDP-43 is downregulated, and intron 6 shows marked diversity. Second, we used poly(A)+ RNA from the cytoplasmic or nuclear fraction to increase sensitivity, whereas they used total RNA from whole cell extracts. Moreover, instead, based on results from experiments with minigenes containing a part of TARDBP, they proposed that an isoform lacking only intron 7 is retained in the nucleus (5,7). However, we found that the isoform lacking only intron 7 is not a major isoform in the nucleus and cytoplasm (Figure (Figure1D1D and F).
We have shown the importance of splicing intron 6 for the autoregulation of TDP-43. Although the splicing of intron 7 fulfills the criteria for nonsense-mediated mRNA decay, both introns were spliced out in most transcripts degraded by a nonsense-mediated mRNA decay pathway. The mutated minigene analysis revealed that the splicing of intron 7 is important for the splicing of intron 6 and not just for binding TDP-43. Moreover, inhibiting the splicing of intron 6 resulted in a smaller decrease in the amount of protein product derived from the TARDBP exon 6-fused minigene after ectopic TDP-43 expression. Interestingly, the mutations observed in ALS10 patients cluster within intron 6 (34). We found an increased amount of TARDBP mRNA in spinal motor neurons in ALS10 patients. In addition, an increased amount of TARDBP mRNA has been reported in iPS-derived neurons of ALS10 patients (35). Therefore, it would be interesting to investigate how the mutations observed in ALS10 patients affect the splicing of introns 6 and 7.
We demonstrated that TDP-43 is autoregulated by the amount of cytoplasmic TARDBP mRNA. The process is regulated by coordination between the use of different PASs and alternative splicing of TARDBP mRNA, accompanied by nonsense-mediated mRNA decay. Our observations demonstrate a role for nuclear TDP-43 in TDP-43 autoregulation and indicate that once the mislocalization of TDP-43 occurs, TDP-43 synthesis is upregulated by an autoregulatory mechanism in affected motor neurons in ALS patients. The vicious cycle might accelerate disease progression by inducing the formation of cytoplasmic inclusions (10–12) and/or the dysregulation of RNA metabolism (8,13–16). Breaking the vicious cycle of TDP-43 production by modulating the autoregulatory mechanism might provide a new therapeutic strategy for halting the progression of ALS.
Supplementary Data are available at NAR Online.
Grant-in-aid for Scientific Research on Innovative Areas (Brain Protein Aging and Dementia Control) , Foundation of Synapse and Neurocircuit Pathology ; MEXT; Grants-in-aid for Scientific Research [26250017, 25253065]; Grant-in-aid for Young Scientists [15K19479] (Japan Society for the Promotion of Science); Grant-in-aid from the Research Committee of CNS Degenerative Diseases and Comprehensive Research on Disability, Health, and Welfare from the Japanese Ministry of Health, Labor, and Welfare, Japan ; Grant-in-aid from Takeda Science Foundation; Grant-in-aid from the Tsubaki Memorial Foundation; Grant-in-aid from JSPS Fellows from the Ministry of Education, Culture, Sports, Science, and Technology of Japan. Funding for open access charge: Scientific Research on Innovative Areas (Brain Protein Aging and Dementia Control) .
Conflict of interest statement. None declared.