|Home | About | Journals | Submit | Contact Us | Français|
ALS is a devastating neurodegenerative disease primarily affecting motor neurons. Mutations in TDP-43 cause some forms of the disease, and cytoplasmic TDP-43 aggregates accumulate in degenerating neurons of most ALS patients. Thus, strategies aimed at targeting the toxicity of cytoplasmic TDP-43 aggregates may be effective. Here we report results from two genome-wide loss-of-function TDP-43 toxicity suppressor screens in yeast. The strongest suppressor of TDP-43 toxicity was deletion of Dbr1, which encodes RNA lariat debranching enzyme. We show that in the absence of Dbr1 enzymatic activity intronic lariats accumulate in the cytoplasm and likely act as decoys to sequester TDP-43 away from interfering with essential cellular RNAs and RNA-binding proteins. Knockdown of Dbr1 in a human neuronal cell line or in primary rodent neurons is also sufficient to rescue TDP-43 toxicity. Our findings provide insight into TDP-43 cytotoxicity and suggest decreasing Dbr1 activity could be a potential therapeutic approach for ALS.
Amyotrophic lateral sclerosis (ALS), also known as Lou Gehrig’s disease, is a late-onset neurodegenerative disease characterized by loss of motor neurons, progressive weakness and eventually paralysis and death within 3–5 years1. Most ALS cases are sporadic (SALS), but 10% are familial (FALS), of which ~20% result from mutations in the Cu/Zn superoxide dismutase 1 gene (SOD1)2. SOD1 mutations are thought to cause disease by a toxic gain-of-function3, and thus, strategies to lower SOD1 levels are being pursued4. However, SOD1 mutations account for only a small percentage of ALS cases. Additional therapeutic strategies are needed.
RNA-binding proteins and RNA-processing pathways have recently been implicated in ALS5,6. The RNA-binding protein TDP-43 was found in cytoplasmic inclusions in spinal cord neurons of most non-SOD1 ALS cases7,8 and mutations in TDP-43 were identified in FALS and SALS cases9–13. Mutations in another RNA-binding protein, FUS/TLS, were also found in some ALS patients14–18. Therefore, therapies targeting TDP-43 and/or FUS could be effective in cases not caused by SOD1 mutations. Interestingly, TDP-43 inclusions occur in many frontotemporal dementia cases, including, and targeting TDP-43 may be effective for these patients. Efforts are underway to define the mechanisms by which TDP-43 and FUS and defects in RNA processing pathways contribute to ALS.
We have used yeast models to illuminate mechanisms underpinning TDP-43 and FUS aggregation and cellular toxicity19–21. TDP-43 forms aggregates in the cytoplasm of yeast cells and is toxic, recapitulating two key features of TDP-43 relevant to human disease19. We used a genome-wide plasmid overexpression screen to identify modifiers of aggregation and toxicity. One modifier of toxicity was Pbp1, whose human homolog, ataxin 2, harbors a polyglutamine tract that is expanded (>34 glutamines) in spinocerebellar ataxia type 2. We found that intermediate-length polyQ expansions (~27–33 glutamines) in ataxin 2 are a significant genetic risk factor for ALS22,23. Additional genetic modifiers of TDP-43 toxicity in yeast might provide more insight into pathogenic mechanisms and could represent novel therapeutic targets.
Here we report results from a genome-wide loss-of-function screen to identify yeast genes that modify TDP-43 toxicity. In our previous screen (22 and A.D.G. unpublished), we used a library of yeast overexpression plasmids. Here, we interrogated collections of non-essential yeast knockouts and knockdowns of essential genes. Loss-of-function screens may be useful to identify therapeutic targets for which inhibitors could be developed. Among the most potent knockout suppressors of TDP-43 toxicity was dbr1Δ. Dbr1 catalyzes the debranching of lariat introns that are formed during pre-mRNA splicing. We show that inhibiting Dbr1’s debranching enzymatic activity is sufficient to rescue TDP-43 toxicity. We further provide evidence that intronic lariat species that accumulate in the cytoplasm of dbr1Δ cells act as decoys to sequester toxic cytoplasmic TDP-43, possibly preventing it from interfering with other essential cellular RNA targets and other RNA-binding proteins. Knockdown of Dbr1 in a human neuronal cell line and in primary rodent neurons is also sufficient to rescue TDP-43 toxicity, suggesting that the effect of Dbr1 on TDP-43 toxicity is conserved from yeast to mammals. We propose that the debranching enzymatic activity of Dbr1 could be a potential therapeutic target for ALS and related TDP-43 proteinopathies.
To identify genes that modify TDP-43 toxicity, two independent unbiased genome-wide yeast deletion screens were performed in two laboratories (Supplementary Fig. 1). Similar approaches have been used to discover modifiers of neurodegenerative disease proteins α-synuclein, huntingtin, and FUS/TLS21,24,25. In the first screen, we used synthetic genetic array (SGA) analysis26 to introduce a galactose-inducible TDP-43 expression plasmid into each non-essential yeast deletion strain by mating (Supplementary Fig. 1a). The second screen included the deletion genes and a library of essential genes generated by decreased abundance by mRNA perturbation (DAmP) technology27 (Supplementary Fig. 1 b–e). We selected yeast deletion strains in which TDP-43 was more (enhancer) or less (suppressor) toxic than in wild-type (WT) cells. Screen 1 was repeated three independent times and only hits that were reproduced all three times were selected. Screen 2 contained three biological and two technical replicates (six replicates total), which allowed the calculation of an interaction score. To provide independent evidence that the gene deletions identified from the screens modify TDP-43 toxicity, we generated gene deletions for the hits from screen 1 and selected genes from screen 2 in an independent strain background and confirmed the effects on TDP-43 toxicity. In screen 1, we identified 14 yeast deletion strains that modified TDP-43 toxicity (Table 1), six enhancers and eight suppressors. The more sensitive and quantitative analysis of screen 2 allowed us to identify a total of 2,581 potential enhancers and 2,056 potential suppressors (Supplementary Data Set 1). In analyzing these results, we chose to focus here on suppressors of TDP-43 toxicity because these could represent attractive therapeutic targets for small molecule inhibitors or RNA interference (see Supplementary Note for discussion of additional yeast TDP-43 toxicity modifier genes).
One of the most effective deletion suppressors of TDP-43 toxicity identified independently in both screens was dbr1Δ, and we focused further efforts on it. Dbr1 deletion suppressed the toxicity of WT TDP-43 as well as an ALS-linked mutant form of TDP-43 (Q331K, Fig. 1a and Supplementary Fig. 2). Immunoblotting confirmed that toxicity was not suppressed because of decreased TDP-43 expression in the dbr1Δ strain (Fig. 1b). Deletion of Dbr1 did not suppress the toxicity of two other neurodegenerative disease proteins, α-synuclein and mutant huntingtin. Indeed, huntingtin toxicity was slightly enhanced in the dbr1Δ strain (Fig. 1a). However, Dbr1 deletion suppressed toxicity of another RNA-binding protein linked to ALS, FUS/TLS (Fig. 1a), demonstrating specificity of the Dbr1 genetic interaction for ALS-linked RNA-binding proteins TDP-43 and FUS/TLS.
Identification of Dbr1 as a potent modifier of TDP-43 toxicity suggests the possibility that inhibiting Dbr1 enzymatic activity (Fig. 1c) could be a promising therapeutic strategy for ALS, especially since TDP-43 is thought to contribute broadly to almost all non-SOD1 ALS cases7. Therefore, we tested if inhibiting Dbr1 function could rescue TDP-43 toxicity in mammalian cells. We generated a stable human M17 neuroblastoma cell line that inducibly expresses epitope-tagged ALS-linked mutant TDP-43 (Q331K), under control of a doxycycline-regulated promoter (Fig. 2a). Inducible expression of TDP-43 Q331K was confirmed by immunofluorescence and immunoblotting (Fig. 2b and data not shown). Upregulating mutant TDP-43 was toxic in these cells (Fig. 2c). To determine if inhibiting Dbr1 rescues TDP-43 toxicity, we transfected these cell lines with an siRNA against human DBR1 or a non-targeting control siRNA (Fig. 2d). Neither siRNA affected TDP-43 expression levels in these cells (data not shown). Whereas the non-targeting siRNA had no effect on toxicity, transfection with the DBR1-specific siRNA significantly reduced TDP-43 toxicity (Fig. 2d).
We next asked whether knockdown of Dbr1 prevents TDP-43-mediated toxicity in primary neurons. We showed previously that overexpression of TDP-43 in primary cortical neurons induces neurodegeneration with cytoplasmic mislocalization and aggregation of TDP-43, recapitulating several key pathologic features of TDP-43 proteinopathies in vitro28. To determine if Dbr1 knockdown protects neurons from TDP-43 cytotoxicity, we used automated microscopy and longitudinal analysis, a technique that identifies and characterizes variables that significantly contribute to neuronal survival29. We isolated and cultured rodent primary cortical neurons and transfected them with three constructs: (1) mApple, (2) enhanced green fluorescent protein (EGFP) or TDP-43 fused to EGFP (TDP-43-EGFP), and (3) siRNA directed against Dbr1 (sR-Dbr1) or scrambled siRNA (sR-Sc). With automated microscopy, hundreds of individual neurons were imaged from each cohort at regular, 24-hr intervals for 8 days (Fig. 3a). In this case, mApple serves as a sensitive survival marker, since the loss of mApple fluorescence, cell blebbing, or disruption of the cell membrane signifies cell death29. These criteria are at least as specific as traditional markers of particular cell death pathways, but have the advantage of detecting all forms of cell death in a single assay and are therefore more sensitive. Because each neuron in a population can be followed longitudinally for an extended period of time, the analysis is analogous to that used in a prospective clinical trial. Powerful statistical tools developed for human epidemiological studies, including Kaplan-Meier survival analysis and Cox proportional hazards analysis, are used to describe the data that are generated from these experiments.
We initially performed a small pilot experiment to determine the optimal amount of Dbr1 siRNA (Supplementary Fig. 3), and found that the most effective dose was 30 nM Dbr1 siRNA. Based on the effect size that we noted in the pilot experiment, subsequent experiments were scaled to a population size to achieve the power necessary to detect a significant effect on neuronal survival. As expected, the risk of death in primary cortical neurons was significantly greater in neurons expressing TDP-43-EGFP than in control neurons expressing EGFP alone (Fig. 3b). The relative risk or hazard ratio (HR) calculated by Cox proportional hazards analysis was 2.23 (p < 1×10−9), signifying an approximately twofold increase in the risk of death. Knockdown of Dbr1 in neurons expressing TDP-43-EGFP reduced the risk of death by nearly 20%, with an HR of 0.81 (p 0.04). In contrast, knocking down Dbr1 had no effect upon the survival of the control cohort expressing EGFP alone (HR 1.2, p 0.3). Using a Kaplan-Meier survival curve to display the same data demonstrates a similar effect on neuron survival (Fig. 3c). Interestingly, this improvement in survival did not depend on decreased cytoplasmic TDP-43. In fact, we observed significantly more primary neurons with cytoplasmic TDP-43 when treated with Dbr1 siRNA than with control siRNA (Fig. 3d). These data suggest that neurons can survive even in the face of cytoplasmic TDP-43. These results in primary neurons expressing human TDP-43 extend the protective effect of Dbr1 knockdown that we observed in yeast and in neuroblastoma cells, and indicate that the ability of Dbr1 inhibition to suppress the toxic effects of TDP-43 accumulation is conserved from yeast to mammalian cells, organisms separated by a billion years of evolution.
DBR1 encodes an RNA debranching enzyme, highly conserved from yeast to man, that cleaves the 2′-5′ phosphodiester bonds of lariat introns that are formed during pre-mRNA splicing30,31 (Fig. 1c). After transcription, primary pre-mRNA transcripts are processed by the spliceosome to remove introns, which are released as lariats32. Lariats are formed during splicing when the 5′-exon attacks the 3′-splice site in a nucleophilic manner. RNA nucleases within the cell can digest the 3′ tail of the lariat introns but the lariat structure is resistant to digestion until Dbr1, a specialized debranching enzyme, cleaves the 2′-hydroxyl-5′-phosphate bond (Fig. 1c). Thus, debranching appears to be a rate-limiting step for the turnover of intronic RNA because the steady-state levels of lariat introns are greatly increased in dbr1Δ cells31,33.
Deletion of Dbr1 reduced TDP-43 toxicity in yeast (Fig. 1), mammalian cells (Fig. 2) and primary neurons (Fig. 3), but it was unclear whether the rescue was because of loss of debranching enzymatic activity or potentially another Dbr1 function. To test this directly, we used “debranching activity dead” mutants of Dbr1. We co-transformed dbr1Δ cells with expression plasmids for TDP-43 and either WT Dbr1 or two independent Dbr1 point mutants (D40A or N85A) that lack debranching enzymatic activity in vitro and in vivo33. Immunoblotting confirmed that the WT and mutant Dbr1 proteins were expressed at equivalent levels (Fig. 4a). Whereas expressing WT Dbr1 restored TDP-43 toxicity to dbr1Δ cells, expressing either of the debranching-inactive Dbr1 point mutants did not (Fig. 4b). Moreover, we tested a panel of 16 Dbr1 mutants, some of which retain full debranching activity others that have lost all activity, and others with partial debranching activity33. There was a perfect correlation between debranching activity and ability to restore TDP-43 toxicity to dbr1Δ cells: WT Dbr1 fully restored TDP-43 toxicity, whereas completely inactive Dbr1 had no effect, and partially active Dbr1 partially restored TDP-43 toxicity (Supplementary Table 1). Thus, inhibiting Dbr1 debranching enzymatic activity is sufficient to rescue TDP-43 toxicity in yeast cells. Finally, expressing mammalian Dbr1 (mDbr1) in dbr1Δ cells was also sufficient to restore TDP-43 toxicity (Fig. 4b), indicating the debranching function of Dbr1 is conserved from yeast to mammals.
In the absence of Dbr1, steady-state levels of lariat introns are greatly increased 33. This suggested an intriguing possibility. Perhaps the accumulating lariat introns in dbr1Δ cells act as a kind of decoy for TDP-43, sequestering it away from important cellular RNAs and other RNA-binding proteins. First, we determined if the rescue of TDP-43 toxicity was specific to lariat intron accumulation or if accumulation of any non-specific RNA species rescued toxicity. We expressed TDP-43 in three other yeast deletion strains, upf1Δ, upf2Δ, and xrn1Δ, that each result in accumulation of RNA species within the cell owing to alterations in the nonsense-mediated decay (NMD) pathway34–38. Whereas, deleting Dbr1 rescued TDP-43 toxicity, TDP-43 toxicity was not suppressed in upf1Δ, upf2Δ, or xrn1Δ cells and was even slightly enhanced in upf1Δ and upf2Δ cells (Fig. 4c), suggesting that the rescue of TDP-43 toxicity is specific to intronic lariat accumulation.
We next sought to determine the mechanism by which intronic lariats suppress TDP-43 toxicity in dbr1Δ cells. We and others showed TDP-43 is toxic in the cytoplasm in multiple cellular and animal model systems19,28,39,40. Perhaps intronic lariats accumulate in the nucleus of dbr1Δ cells, sequester TDP-43 there and prevent it from aggregating in the cytoplasm, and thereby suppress toxicity. In this model, sequestering TDP-43 in the cytoplasm should overcome the toxicity suppression by dbr1Δ. To test this, we retained TDP-43 in the cytoplasm by mutating the nuclear localization signal (NLS)22,41. We reasoned that, if dbr1Δ suppresses TDP-43 toxicity in the nucleus, the ΔNLS mutant TDP-43 ould still be toxic. Surprisingly, Dbr1 deletion still suppressed toxicity when TDP-43 was sequestered in the cytoplasm (Fig. 5a), suggesting the site of action of intronic lariats is the cytoplasm.
The above genetic result suggested that intronic lariats act in the cytoplasm to suppress TDP-43 toxicity. To test this directly, we developed a method to visualize, in living cells, the endogenous intronic lariats that accumulate in dbr1Δ cells. We reasoned that, if we could visualize the intronic lariats, we could determine 1) if they localized to the cytoplasm or nucleus and 2) if they co-localized with TDP-43 inclusions. A technique to visualize the localization of endogenous mRNAs in living yeast cells was recently developed42. m-TAG uses homologous recombination to insert binding sites for the RNA-binding MS2 coat protein (MS2-CP) between the coding region and 3′UTR of any yeast gene. Expression of MS2-CP fused to GFP enables the visualization of any yeast mRNA that has been tagged with MS2 binding sites42–44. We modified m-TAG to visualize intronic lariats. Instead of tagging the 3′UTR of a target yeast gene, we used homologous recombination to integrate MS2 binding sites in the intron of the ACT1 gene (Fig. 5b), which has previously been shown by northern blot analysis to accumulate in the absence of Dbr1 activity33.
We transformed WT and dbr1Δ yeast cells, each harboring MS2 binding sites integrated in the intron of the ACT1 gene (Fig. 5c), with a plasmid containing a 3XGFP-tagged MS2-CP RNA-binding protein under control of the inducible MET25 promoter. We analyzed the localization of MS2-CP-GFP by fluorescence microscopy. As reported43, because the MET25 promoter is somewhat leaky, we detected the GFP fusion protein even without induction (by growth in the presence of methionine). In WT cells, in which no intronic lariats should accumulate33, we detected faint MS2-CP-GFP expression in a diffuse pattern throughout the cell (Figs. 5c and Supplementary Fig. 4). In contrast, in dbr1Δ cells, MS2-CP-GFP accumulated in one or two bright foci per cell and these were always located in the cytoplasm (Fig. 5c). We observed similar results by tagging an independent intron from another yeast gene, CYH2 (data not shown). These foci could represent the excised lariat introns that accumulate in dbr1Δ cells or a splicing intermediate45 (also see Supplementary Note). However, since we did not see them in WT cells in which ACT1 or CYH2 introns were tagged with MS2 sequences (Figs. 5c and Supplementary Fig. 4), these foci are likely to be lariat introns that accumulate in the absence of Dbr1 activity.
Having established that intronic lariats accumulate in the cytoplasm of dbr1Δ cells, we next tested our hypothesis that the accumulating lariats in dbr1Δ cells act as a decoy for TDP-43 by preventing it from interacting with other important cellular RNAs and RNA-binding proteins. We expressed untagged TDP-43 in WT and dbr1Δ cells and visualized its localization by immunocytochemistry with a TDP-43-specific antibody (Fig. 5d, e). We visualized tagged intronic lariats by fluorescence microscopy. Remarkably, in every dbr1Δ cell analyzed, TDP-43 co-localized with intronic lariats (Fig. 5e). Consistent with our genetic results (Fig. 5a) and fluorescence microscopy experiments (Fig. 5c), TDP-43 was not retained in the nucleus in dbr1Δ cells and accumulated in the cytoplasm (Fig. 5e). However, the size and shape of TDP-43 inclusions were profoundly different in dbr1Δ and WT cells. In WT cells, TDP-43 formed multiple small, irregular shaped cytoplasmic foci (Fig. 5d), whereas in dbr1Δ cells, TDP-43 always formed at least one large, perfectly round focus (Fig. 5d), which always co-localized with intronic lariats (Fig. 5e).
We also assessed whether the TDP-43 ribonucleoprotein complex contains lariats by biochemical means. We expressed TDP-43 containing a Flag epitope tag in WT and dbr1Δ cells and immunoprecipitated TDP-43 from cell lysates with an anti-Flag antibody. The associated RNA was separated by two-dimensional electrophoresis. Linear RNA molecules migrate in a diagonal on the two-dimensional gels, but lariat and lariat breakdown products migrate in a separate arc owing to the reduced mobility conferred by their altered structure (Supplementary Fig. 6a). As expected, in WT and dbr1Δ cells, TDP-43 associated with RNAs. However, in dbr1Δ cells, the TDP-43 ribonucleoprotein complex contained additional RNA species, consistent with intronic lariats.
Finally, to directly test our hypothesis that intronic lariats formed in dbr1Δ cells act as decoys for TDP-43 by competing with it for binding to other cellular RNAs, we performed in vitro electrophoretic mobility shift assays. Incubation of a radioactively labeled RNA probe known to bind TDP-43 (UGAAAAUUAAUGUGUGUGUGUGGAAAAUU)46 with recombinant TDP-43 results in a gel shift (Supplementary Fig. 6 b, c). We isolated total RNA from WT and dbr1Δ cells and tested the ability of these RNAs to compete for binding to TDP-43. As expected, RNAs from WT and dbr1Δ cells competed for TDP-43 binding (Supplementary Fig. 6b), consistent with TDP-43 binding a large number of RNA targets47. To specifically test the ability of intronic lariats to compete for binding to TDP-43 (i.e., to act as decoys), we treated RNAs isolated from WT and dbr1Δ cells with RNase R, which degrades all RNAs except for branched lariats48. When these RNase R-treated samples were incubated with the TDP-43 binding reaction, only the RNase R-treated RNAs isolated from dbr1Δ cells competed TDP-43 away from its binding site (Supplementary Fig. 6c). Thus, dbr1Δ cells contain an RNase R-resistant species, likely intronic lariats, that compete for binding to TDP-43.
Using two unbiased genetic screens in yeast, we discovered Dbr1 as a potent modifier of TDP-43 toxicity. We provide evidence that in the absence of Dbr1, intronic lariats accumulate in the cytoplasm, and we propose that these act as decoys to sequester TDP-43 away from interfering with essential RNAs and RNA-binding proteins. Thus, we suggest that the debranching enzymatic activity of Dbr1 could be a novel therapeutic target to combat toxic effects from cytoplasmic TDP-43 aggregation in ALS. The additional genetic modifiers of TDP-43 toxicity uncovered from these two yeast screens (Table 1 and Supplementary Dataset 1) will hopefully provide even more insight into disease mechanisms and open novel avenues for therapeutic intervention.
Recent studies indicate that RNA-binding is an important component of TDP-43 toxicity in cellular and animal models19,22,49,50. One potential mechanism of neurotoxicity is that TDP-43 aggregates in the cytoplasm, sequestering essential cellular RNAs and RNA-binding proteins away from their normal function. This could lead to catastrophic changes in RNA metabolism, owing to defects in RNA stability, splicing, transport and/or translation of essential RNAs. This effect might be more deleterious to motor neurons if motor neuron-specific transcripts were preferentially sequestered by TDP-43 or even if motor neurons were simply more sensitive to subtle alterations in any of these RNA metabolic pathways. TDP-43 loss-of-function likely also contributes to disease. TDP-43 is depleted from nuclei of affected neuronal populations in ALS patients8, and down-regulation of TDP-43 in mouse leads to widespread dysregulation of splicing47. Finally, TDP-43 mutations might act in a dominant negative manner, interfering with the function of WT TDP-4351,52. These three pathogenic mechanisms (gain-of-toxic properties in the cytoplasm, loss-of-function from the nucleus, and dominant negative effects) are almost certainly not mutually exclusive and likely all contribute to ALS pathogenesis.
Inhibiting Dbr1 function might overcome or prevent cytoplasmic TDP-43 (and FUS/TLS) aggregates from wreaking havoc on critical RNAs and RNA-binding proteins. We propose the accumulated intronic lariat species in dbr1Δ cells suppress TDP-43 toxicity by acting as decoys, causing TDP-43 to bind to them rather than to important cellular RNA targets (Fig. 5f). The catalog of genome-wide RNA targets for TDP-43 is emerging47,53-55. Intriguingly, TDP-43 appears to preferentially bind to long intronic pre-mRNA sites compared to coding exonic regions or UTRs47. This might explain why the intronic RNA sequences (lariats) that accumulate in dbr1Δ cells sequester TDP-43 away from other RNA species in the cell and suppress TDP-43 toxicity. TDP-43 might also preferentially recognize the lariat branch point structure or binding to the lariat promotes nucleation or other alterations to TDP-43 conformation.
For Dbr1 to be a therapeutic target, several questions must be addressed. Wouldn’t inhibition of Dbr1 debranching activity be toxic itself? Indeed, the highest concentration of Dbr1 siRNA (50 nM) in primary rat neurons increased the risk of death (Supplementary Fig. 3), suggesting too much Dbr1 inhibition could be deleterious. However, in budding yeast, Dbr1 deletion is tolerated (Fig. 1a), and siRNA knockdown in M17 cells does not result in growth inhibition (M.A. and A.D.G. unpublished and56). Nevertheless, it will be important to determine what level of Dbr1 inhibition is tolerated and if an appropriate therapeutic index can be achieved. An additional consideration for this approach in mammalian cells is the non-coding RNAs in intronic regions that depend on Dbr1 activity. Expression of these non-coding RNAs may be dysregulated by Dbr1 inhibition, and the consequences of this must be assessed. Furthermore, since abnormal accumulation of RNA is hypothesized to trigger autoimmunity and there would likely be a large pool of lariats in human cells, the effects of Dbr1 inhibition on autoimmuity should be analyzed. Although, to our knowledge, Dbr1 inhibitors are not available, our novel method to visualize intronic lariats in living cells (Fig. 4) will facilitate screening of chemical compound libraries for small molecule Dbr1 inhibitors.
There are currently no TDP-43-directed therapies for ALS or related TDP-43 proteinopathies. Antisense oligonucleotides and RNA interference approaches are emerging as attractive therapeutic strategies in neurodegenerative diseases in which decreasing levels of a toxic mutant protein might be efficacious57–60. Indeed, treating a mouse model of inherited ALS (caused by a mutation in SOD1) with antisense oligonucleotides to SOD1 significantly slowed disease progression4. This approach offers tremendous promise for treating ALS patients with SOD1 mutations, but since mutations in SOD1 account for only ~2–5% of all ALS cases, additional therapeutic strategies are needed. Because TDP-43 appears to contribute to ALS pathogenesis broadly7, targeting TDP-43 with antisense approaches could be effective. However, it is unknown if TDP-43 contributes to disease via loss- or gain-of-function mechanisms61, mandating caution in pursuing TDP-43-lowering approaches. We present an alternative strategy: by targeting Dbr1, by small-molecule inhibitors or antisense oligonucleotides, it might be possible to antagonize TDP-43’s toxic effects in the cytoplasm, without interfering with potential critical functions in the nucleus.
The dbr1Δ strain was obtained by replacing the DBR1 coding region with a KanMX4 cassette in the BY4741 or W303 strains and verified by colony PCR. CEN and 2-micron galactose-inducible TDP-4328, 2-micron ΔNLS-TDP-43-YFP22, FUS21, α-synuclein62, and Htt103Q expression plasmids were as described63.The yeast Dbr1 mutant expression plasmids 33 and m-TAG plasmids pLOXHIS5MS2L, pSH47 and pMS2-CP-GFP (x3) were as described 44.
We used the synthetic genetic array (SGA) technique to screen the collection of non-essential only (Screen 1, performed in A.D.G. laboratory) yeast knockout strains or a collection of non-essential yeast knockout strains combined with essential genes reduced in expression by DAmP27 (Screen 2, performed in the laboratories of R.V.F. and N.K.) for modifiers of TDP-43 toxicity. These screens were performed as described26,64,65 with some modifications66, using a Singer RoToR HDA (Singer Instruments, Somerset, UK). For screen 1, the galactose-inducible TDP-43 expression construct (pAG416Gal-TDP-43) was introduced into MATα strain Y7092 to generate the query strain. This query strain was mated to the yeast haploid deletion collection of non-essential genes (MATa, each gene deleted with KanMX cassette; Fig. 1A). Haploid mutants harboring the TDP-43 expression plasmid were grown in the presence of glucose (TDP-43 expression “off”) or galactose (TDP-43 expression “on”). After growth at 30°C for 2 days, plates were photographed and colony sizes measured by ImageJ image analysis software, based on67. The entire screen was repeated three times and only hits that reproduced all three times were selected for further validation by random spore analysis on DNA sequencing of deletion strain bar codes.
For screen 2, a control query strain containing the empty vector was generated by transforming the yeast strain BY5563 (Mat α) with p415 Gal GFP and the experimental query strain was generated by transforming BY5563 (Mat α) with p415 Gal TDP-43-GFP. Both strains were mated with the combined yeast knockout-DAmP collection (Mat a) and haploid yeast containing the deletion/DAmP yeast gene with the control or experimental plasmid were selected as described65. Three biological replicates were isolated by selecting for haploid yeast after sporulation on three plates. Each biological replicate was duplicated to generate six replicates (three biological and two technical replicates per biological replicate). Plates were imaged at several time points after final inoculation on glucose or galactose. Colony size was assessed as described67.
Rat cortical neurons were isolated from embryonic d20–21 rat (Sprague Dawley) pups, cultured in serum-free media in 96 well plates for 4 days in vitro, and then transfected using Lipofectamine2000 (Invitrogen; full protocols at http://www.gladstone.ucsf.edu/gladstone/site/finkbeiner/). Neurons were transfected with plasmids (pGW1 backbone) encoding mApple and either TDP-43-EGFP or EGFP, and with siRNA (SMARTpool, Dharmacon) directed against DBR1 or scrambled siRNA.
For neuronal survival analysis, we used a robotic imaging system as described68,69. Briefly, approximately 24–48 hours after transfection, images were obtained with an inverted Nikon microscope (TE2000) equipped with PerfectFocus, an extra-long working distance (ELWD) 20× objective lens, and an Andor Clara 16-bit, ultra-cooled camera. Illumination was provided by an adjustable, high-intensity, and long-lasting xenon lamp and a liquid light guide to maximize the signal-noise ratio. The stage and shutter movements, fluorescence excitation and emission filters, focusing, and imaging acquisition are fully automated and controlled through a 64-bit host computer running a combination of proprietary software (ImagePro, Media Cybernetics) and custom-designed algorithms.
Digitized images from the RFP (mApple) and GFP (TDP-43-EGFP) channels were assembled into montages with ImagePro (Media Cybernetics), and cells were prospectively counted in the RFP channel by original code developed in MatLab. Cell death was marked by loss of fluorescence, membrane rupture, or neurite retraction. The time of death for each neuron was determined by the last time that the neuron was seen. Kaplan-Meier and cumulative risk of death curves were generated using the survival package in R. Statistical significance of survival differences between cohorts of neurons was determined by the log-rank test, and Cox proportional hazards analysis was used to measure the relative change in the risk of death, or hazard ratio.
Nucleic acid pellets from immunoprecipitation were resuspended in 5 μl of DEPC water and 5 μl of 2X nucleic acid loading dye (0.025 g xylene cyanol, 0.025 g bromophenol blue in 100% formamide) and heated at 85°C for 5 min. Nucleic acids were separated in the first dimension by loading samples into a prerun (300 V for 10 min) 5% polyacrylamide gel (19:1, acrylamide:bis-acrylamide) containing 7 M urea and 1X TBE (90 mM Tris base, 90 mM boric acid, 2 mM EDTA, pH 8.0) and electrophoresing samples at 300 V. Nucleic acids were separated in the second dimension by placing the entire lane from the first dimension gel above the second dimension gel. Second dimension gel was exactly the same as the first dimension gel except the polyacrylamide percentage is 10%. Samples were electrophoresed at 300 V until xylene cyanol dye reached the bottom of the second dimension gel. Two-dimensional nucleic gels were washed briefly in water and then placed in 1X Sybr Gold (Invitrogen) solution prepared in water. Gels were protected from light and gently agitated at 23–25°C in 1X Sybr Gold solution for 45 min. Images of the gel were captured on an ultraviolet light box at an excitation wavelength of 302 nm and a 4-sec exposure time.
To visualize intronic lariats in dbr1Δ cells, we adapted the m-TAG protocol42 for use with introns. We used PCR-based chromosomal gene tagging, using homologous recombination to integrate the loxP::Sphis5+::loxP::MS2L cassette into the intron of either the ACT1 gene or the CYH2 gene in WT or dbr1Δ cells. For ACT1, we inserted the MS2L cassette at bp 159 of the ACT1 genomic DNA sequence (ACT1/YFL039C on chromosome VI from coordinates 54696 to 53260). For CYH2, we inserted the MS2L cassette at bp 303 of the CYH2 genomic DNA sequence (RPL28/YGL103W on chromosome VII from coordinates 310967 to 311927). To excise the loxP-flanked Sphis5 cassette, we transformed cells with plasmid pSH47, which expresses Cre recombinase under control of the galactose-regulated promoter. Proper integration of MS2L sequences was verified by colony PCR and DNA sequencing.
We visualized MS2-tagged intronic lariats in WT and dbr1Δ cells as described in42. We transformed WT and dbr1Δ cells, harboring an MS2-tagged ACT1 intron, with pMS2-CP-GFP(x3), which expresses MS2-CP fused to three copies of GFP under the control of the MET25 promoter. Yeast cells were fixed with 70% ethanol and stained with DAPI in Vectashield mountaing medium. Colocalization of an MS2L-tagged intronic lariat with TDP-43 was determined using fluorescence microscopy to detect lariats (GFP) and immunocytochemistry with a TDP-43-specific antibody to detect untagged TDP-43.
To generate a lentiviral vector for the conditional expression of TDP-43 Q331K, an N-terminally FLAG- and C-terminally myc-tagged TDP-43Q331K Gateway entry clone was used in a Gateway LR reaction to transfer it to the pSLIK-Neo lentiviral destination vector. Recombinant lentiviruses were produced as described in70. For lentiviral transduction, human M17 neuroblastoma cells (ATCC) were incubated with lentiviral supernatants for 6 h at 37°C. Stable cell lines were selected by culturing transduced cells for 5 days in growth medium (minimum essential medium (MEM) alpha /F-12 (Invitrogen) containing 10% heat-inactivated fetal bovine serum, 100 U/ml penicillin, and 100 μg/ml streptomycin (complete medium) with 100 μg/ml Geneticin). Individual clones were selected and analyzed for inducible expression by immunoblotting and immunofluorescence analysis.
We assessed the effect of TDP-43 on cell proliferation in the MTT assay. M17 cells stably transduced with either an empty vector or with the TDP-43 Q331K vector were seeded in 96-well plates. To induce TDP-43Q331K expression, doxycycline (0.1, 1, or 2 μg/ml) was added to the growth medium, and cells were incubated for 3 days. To measure proliferation, MTT (3-(4,5-dimethylthiazol-2-yl)-2,5-diphenyltetrazolium bromide) was added to each well and incubated for 4 h at 37°C. Acidic isoproponal (40 mM HCl) was then added to each well to solubilize the blue formazan crystals. Absorbance at 570 nm was read with a Tecan Safire II plate reader. Absorbance measurements were normalized to the absorbance of cells not treated with doxycycline and used to calculate a percent viability for each condition. Two-way ANOVA was used to evaluate differences in means between two groups, and P < 0.05 was considered statistically significant. Each condition was replicated at least four times and each experiment was independently repeated at least three times.
siRNAs targeting human Dbr1 and non-targeting control siRNAs were purchased from Thermoscientific (Dbr1, catalog #J-008290-08-0005; non-targeting, catalog #D-001810-01-05). siRNA knockdown was validated by immunoblotting with a Dbr1-specific antibody (Proteintech, rabbit polyclonal catalog #16019-1-AP). The M17 cell line with inducible TDP-43 Q331K expression was transfected with 50 nM of hDbr1 siRNAs or non-targeting control siRNAs using RNAi Lipofectamine (Invitrogen). Two days later cells were treated with media containing doxycycline (0 and 1 μg/ml) to induce expression of TDP-43 and then toxicity was assessed by MTT assay after 2 days (day 4). TDP-43 toxicity in M17 cells treated with Dbr1 siRNAs was compared to toxicity in cells treated with non-targeting control siRNAs or empty vector.
GST-TDP-43 was purified from E. coli as described20. RNA-binding assays were performed as described71. Briefly, competitor RNA was purified from WT W303 yeast or dbr1Δ W303 yeast using an RNeasy Mini Kit (Qiagen) and was either left untreated or treated with RNase R to degrade any linear RNA structures while leaving lariat RNA intact. After RNase R treatment, RNA was purified using an RNeasy MinElute Cleanup Kit (Qiagen). A radioactively labeled RNA probe, UG6 (5′-GAAAAUUAAUGUGUGUGUGUGGAAAAUU-3′), which is known to bind TDP-4346, was generated by T7 polymerase-catalyzed transcription of a DNA template in the presence of 32P-labeled UTP. Standard binding reactions were carried out in 10 μl, with a final concentration of 4 mM MgCl2, 25 mM phosphocreatine, 1.25 mM ATP, 1.3% polyvinyl alcohol, 25 ng of yeast tRNA, 0.8 mg of BSA, 1 mM DTT, 0.1 μl Rnasin (Promega, 40 U/ml), 75 mM KCl, 10 mM Tris, pH 7.5, 0.1 mM EDTA, and 10% glycerol. Binding reactions containing GST-TDP-43 (0.5μM) plus 32P-labeled UG6 (1 μl) were incubated for 15 min at 30°C in the presence of various dilutions of competitor RNA (0–12.9 μg). After binding, reactions were transferred to ice and heparin was added to a final concentration of 0.5 μg/μl; reactions were analyzed on a 4.5% native gel (acrylamide/bis 29:1, BioRad).
We thank Kristen Lynch and Sarah Smith for helpful suggestions and discussions about RNA and splicing; Tadashi Nakaya for advice and assistance with lentivirus transduction experiments and analysis; Cornelia Kurischko for advice and assistance with visualizing P-bodies and stress granules; Quinn Mitrovich and Alex Plocik for advice with running the two-dimensional nucleic acid gels; Sean Collins and Dale Cameron for useful advice and assistance in performing the data analysis for screen 2; Brittany Hodges, Divya Hosangadi, Priya Patel, Pierce Nathanson, and Carolin Mrejen for assistance with yeast experiments; Jon Epstein and Alya Raphael for critical comments on the manuscript and helpful suggestions; and Gary Howard for editorial assistance. Andrew Elden helped with initial stages of this project. We are grateful to Dr. Jeffrey Gerst (Weizmann Institute, Israel) for providing the yeast m-TAG plasmids, to Dr. Roy Parker (University of Arizona) for sharing the P-body and stress granule marker plasmids, and to Jimena Weibezahn in Dr. Jonathan Weissman’s laboratory (University of California San Francisco) for providing temperature sensitive CDC48 (cdc48-3, SM 4783) and wild type CDC48 isogenic (SM 5124) yeast strains. We thank Dr. Beate Schwer for providing the yeast mutant Dbr1 expression plasmids. We thank Dr. Charlie Boone for the MATα strain Y7092. This work was supported by NIH Director’s New Innovator Awards 1DP2OD004417 (to A.D.G.) and 1DP2OD002177 (to J.S.), NIH grants NS065317 and NS065317 (to A.D.G.), NS067354 (to J.S.), and GM084448, GM084279, GM081879 and GM098101 (to N.J.K.), a New Scholar in Aging Award from the Ellison Medical Foundation (to J.S.), a grant from the Packard Center for ALS Research at Johns Hopkins (A.D.G. and J.S.), and a grant from the Consortium for Frontotemporal Research (to R.V.F.), NIH grant 2P01AG02074 (to S.F.), a grant from the ALS Association (to S.F.), and the Taube-Koret Center and Hellman Family Foundation (to S.F.). A.D.G. is a Pew Scholar in the Biomedical Sciences, supported by The Pew Charitable Trusts, and a Rita Allen Foundation Scholar. N.J.K. is a Searle Scholar and a Keck Young Investigator. R.V.F. is an Investigator of the Gladstone Institutes. The J. David Gladstone Institutes received support from a National Center for Research Resources Grant RR18928.
AUTHOR CONTRIBUTIONSM.A., M.J.H., M.D.F., S.J.B., J.S., N.J.K., S.F., R.V.F., and A.D.G. designed the experiments. M.A., M.J.H., M.D.F., S.J.B., J.S., E.A.S., Z.D., X.F., and A.D.G. performed the research. M.A., M.J.H., M.D.F., S.J.B., J.S., N.J.K., S.F., R.V.F., and A.D.G. analyzed and interpreted data. M.A., M.J.H., M.D.F., S.J.B., S.F., R.V.F., and A.D.G. wrote the paper with contributions from all authors.
COMPETING FINANCIAL INTERESTS
A.D.G. is an inventor on patents and patent applications that have been licensed to FoldRx.