Tri-snRNP activity modulates tauopathy phenotypes

Abstract

Alzheimer’s disease (AD) and other tauopathies are neurodegenerative disorders with devastating consequences for cognition and memory. Pathogenic accumulation of tau can be modeled in Caenorhabditis elegans, which recapitulate human neurodegeneration including aging-dependent accumulation of phosphorylated tau, tau aggregation, neuronal dysfunction, and neuron degeneration. Using forward genetic screens to identify genes modulating tau pathology, we identified a single point mutation in dib-1 that ameliorates tau-driven behavioral defects, prevents neurodegeneration, and decreases tau protein levels. The dib-1 gene encodes a small, highly conserved protein, known as TXNL4A in humans, and participates in mRNA splicing via the U4/U6.U5 tri-snRNP. Notably, heterozygous loss of prp-8, a neighboring protein within the tri-snRNP, also prevents tau-driven neurodegeneration. RNA sequencing of dib-1 mutants demonstrates widespread intron retention consistent with disruption of tri-snRNP functions. Disruption of nonsense-mediated decay further rescues tau-driven phenotypes only in the presence of the dib-1 mutation. TXNL4A levels are decreased in AD in human frontal cortex, demonstrating the translational relevance of dib-1. Taken together, these findings suggest pathological tau impacts splicing function, and spliceosomal activity modulation can ameliorate tauopathy.

Graphical Abstract

Graphical Abstract.

Introduction

Alzheimer’s disease (AD) is a progressive neurodegenerative disease and the most common cause of dementia. In AD, intracellular tau polymerizes as neurofibrillary tangles and extracellular deposits of β-amyloid aggregate to form plaques. Accumulation of tau correlates well with cognitive decline in AD [1]. Tau binds microtubules in normal neurons, but in disease, tau dissociates from microtubules, initiating a pathologic cascade culminating in tau fibrilization; tau fibrils form the defining neuropathological hallmark of tauopathies, including progressive supranuclear palsy and frontotemporal dementia. Competing theories about the causal neurotoxic event in this cascade posit the toxic effects of tau in vivo: formation of intracellular tau aggregates, novel toxic functions of hyperphosphorylated tau, or prevention of normal microtubule binding due to hyperphosphorylation [2].

The optically transparent and genetically tractable nematode worm Caenorhabditis elegans (C. elegans) presents a cost-effective and translationally relevant platform to model human disease in behaving and rapidly aging animals [3, 4]. C. elegans has served as an excellent model system for investigating molecular and cellular mechanisms of neurodegeneration due to their high degree of genetic homology and conservation of nervous system cell types with vertebrate animals [3, 5, 6]. To dissect the molecular mechanism of tau accumulation, we have employed transgenes to express human tau in C. elegans neurons, leading to behavioral deficits, accumulation of insoluble tau, degenerative changes in axons, loss of neurons, and decreased lifespan [7]. Using this transgenic model, we have leveraged established forward genetic screening approaches to investigate the molecular genetic underpinnings of tauopathy [3, 8].

Genetic experiments using C. elegans identified sut-2/MSUT2, a genetic suppressor of tau, that also modulates tau in AD mouse models and exhibits depletion in human AD cases [9]. Using a similar forward genetic approach, we isolated a novel mutation in dib-1, a gene encoding a small, highly conserved component of the spliceosome. DIB-1 was first identified as a necessary protein for mitosis in yeast [10, 11], a role that was later found to be due to its central function in messenger RNA (mRNA) splicing [12]. S. pombe Dib1 and human TXNL4A share 79% homology yet DIB-1 does not share domains or sequence similarity with other proteins [13]. The dib-1 gene and its homologs are essential in C. elegans,S. cerevisiae, and S. pombe [13, 14]. Human spliceosomopathies result from mutations in the numerous splicing components and factors. Most of the syndromic features of these genetic variants result from increased sensitivity of neural crest cells to disturbances in splicing patterns during development [15]. TXNL4A mutations cause Burn-McKeown syndrome, identified by bilateral choanal atresia, hearing loss, cleft lip or palate, craniofacial dysmorphisms, and cardiac defects [16]. A large genetic study found that the most common genetic underpinning for the syndrome is a loss of function in one allele and a low-frequency 34 bp deletion in the core promoter on the other allele [16, 17].

TXNL4A/DIB-1/DIM-1 functions in the spliceosome, a large cellular machine made up of >300 proteins and 5 RNAs, catalyzing splicing of pre-mRNA into mature mRNA. TXNL4A associates with the U4/U6.U5 tri-snRNP and interacts with the N-terminus of splicing factor PRP6 [18]. The tri-snRNP is a 1.5 megadalton complex that is pre-assembled and forms part of the spliceosome following compositional and conformational changes. The tri-snRNP consists of three small nuclear RNAs (snRNAs) and >30 proteins [19]. U5 snRNA, the main RNA associated with TXNL4A, is the only common complex in both the major and minor spliceosome. Absence of TXNL4A leads to defective tri-snRNP assembly, and TXNL4A may prevent premature spliceosome activation [18]. DIB-1 exhibits flexible binding site activity and can tolerate significant mutations, even in locations thought to be essential for folding [14]. To test whether generalized reduction in U4/U6.U5 tri-snRNP function might rescue tauopathy, we also investigated prp-8 as a mediator of tauopathy. PRP-8 is a large, highly conserved protein, with 66% sequence homology from yeast to humans [20, 21], in the catalytic core of the spliceosome [22]. Similarly to DIB-1, PRP-8 serves as a component of the U5 snRNP and the U5/U4.U6 tri-snRNP, but its presence in different conformations of the spliceosome indicates that it may have numerous functions [23]. To date, the role of TXNL4A and PRPF8 in AD remains unknown. In this paper, we characterize the molecular mechanism by which dib-1 and prp-8 suppress tau mediated neurodegeneration.

Materials and methods

Ethics approval and consent to participate

Informed consent for research brain donation was obtained according to protocols approved by the UW and KPWHRI Institutional Review Boards.

C. elegans strains

C. elegans strains were maintained at 20°C on nematode growth medium plates containing OP50 Escherichia coli as previously described [24]. Before collecting samples for protein and RNA extraction, strains were grown on media plates with five times more peptone to increase population density. Strains used are listed in Supplementary Table S1. The wild-type control strain used was Bristol N2 [24]. The bk1847 C. elegans allele was generated via chemical mutagenesis as in previous publications [25, 26], and the S2L dib-1 mutant alleles bk3111 and bk3112 were constructed by CRISPR–Cas9 genome editing using ALT-R reagents (Integrated DNA Technologies) as in prior work [27–29]. DIB-1 wild-type and S2L neuronal overexpression transgenic lines, CK2827 and CK4114, respectively, were constructed as previously described [30]. In brief, the DIB-1 coding sequence expressed under the pan neuronal snb-1 promotor was introduced into the C. elegans genome as a mix containing Psnb-1::dib-1 (25 ng/μl) or Psnb-1::dib-1(S2L) (75 ng/μl) and co-injection marker Pmyo-3::mCherry (30 or 20 ng/μl, respectively) microinjected into N2. Progeny were screened for mCherry in the body wall muscle as sign of transmission of extrachromosomal arrays. Extrachromosomal arrays were integrated by sub-lethal doses of 312 nm UV irradiation. Integrated lines were then backcrossed with N2 thrice before use in experiments. Transgenic tau strains used include CK144 (bkIs144[Paex-3::h4R1Ntau Pmyo-2::GFP]), CK1443 (bkIs1443[Paex-3::h4R1Ntau Pmyo-2::dsRED]), CK1441 (blks1441[Paex-3:: h4R1Ntau Pmyo-2::dsRED]), and CK10 (bkIs10[Paex-3::Tau V337M Pmyo-2::GFP]) [26, 31]. For strains without obvious phenotypes or fluorescent markers, genotypes were confirmed by polymerase chain reaction (PCR) and gel electrophoresis. Additional strains used are from the C. elegans Genetics Center or sent from personal collections.

C. elegans behavior

C. elegans swimming behavior measurements were carried out as in previous publications [7, 9]. C. elegans were developmentally synchronized by selecting L4 animals 24 h prior to the scheduled experiment. Plates were washed with 1 ml of M9 buffer for 30 s and the liquid transferred to a smaller behavior plate. After 60 s in M9, a 60-s video was recorded of the plate and worms were tracked using the WormTracker system. One thrash is defined as a bend larger than 20 degrees lasting longer than 2 frames (14 frames per second) as measured from the counterpoint to the quarter point of the worm (i.e. point between head and center and point between tail and center). Movement was quantified as the number of turns for the total tracking time and averaged over three biological replicates [28].

Lifespan assay

C. elegans were developmentally synchronized by selecting L4 animals 24 h prior to the scheduled experiment. L4 individuals were place onto 35 mm NGM plates that were previously seeded with 10× concentrated OP50 and treated with 10 mg/ml 5-flourodeoxyuridine to prevent viable egg production for the duration of the experiment. Around 100 worms per strain were inspected daily for indications of life including movement or eating behavior. If a C. elegans failed to move on visual inspection, they were lightly tapped with a platinum pick to assess response to touch. Once the animal failed to respond to touch, they were removed from the plate and marked as dead. C. elegans that showed indications of bursting were removed from the analysis [28, 32].

Protein extraction

Protein was extracted from bulk C. elegans samples using a lysis approach [32]. C. elegans are developmentally synchronized using hypochlorite treatment and then grown to adulthood at 20°C on 5× PEP plates. Plates were then washed to remove worms using M9 buffer, and worms were washed another two times in M9 buffer to remove excess E. coli. Samples were pelleted by 1-min centrifugation at 400 × g, snap-frozen in liquid nitrogen and stored at −80°C until further processing. 2× SDS protein sample buffer (0.046 M Tris, 0.005 M EDTA, 0.2 M dithiothreitol, 50% sucrose, 5% sodium dodecyl sulfate, and 0.05% bromophenol blue) was added to the samples at four times the volume (μl) of buffer to the weight of the pellet (mg), homogenized via sonication three times at 70% power for 15 s, boiled for 10 min at 95°C, centrifuged at 13 000 × g for 2 min, and stored at −20°C.

Immunoblot

Protein immunoblotting was performed using the Criterion apparatus (Bio-Rad) as suggested by the manufacturer (Bio-Rad Laboratories, Hercules, CA, USA). 10 or 6 μl of protein sample was used per well depending on total well number and loaded in a precast 4%–15% gradient sodium dodecyl sulfate polyacrylamide (SDS–PAGE) gel. The ladder used was the Precision Plus Protein Standards (Bio-Rad). Gels were run for one hour or until the dye had reached the bottom of the gel before being transferred to a polyvinylidene difluoride (PVDF) membrane as recommended by the manufacturer (Bio-Rad Laboratories, Hercules, CA, USA). Membranes were blocked in 5% milk in phosphate-buffered saline (PBS) and then incubated overnight with primary antibody in 5% milk in PBS in a 4°C refrigerator while shaking. Membranes were then washed with PBS with 0.05% Tween three times for 15 min, incubated at room temperature with horseradish peroxidase (HRP)-coupled secondary antibody for 2 h, and washed again with PBS with 0.05% Tween three times. The primary antibodies used were rabbit monoclonal anti-tau SP70 antibody (Rockland) at 1:5000, mouse anti-tubulin antibody E7 (Developmental Studies Hybridoma Bank) at 1:5000, mouse anti-tau CP13 antibody (Peter Davies) at 1:1000, mouse anti-tau PHF-1 antibody (Peter Davies) at 1:1000, and mouse anti-Dim1 antibody (Santa Cruz) at 1:500. Secondary antibodies used were anti-rabbit HRP (Jackson Immuno Research) and anti-mouse HRP (Jackson Immuno Research) at 1:5000 [33, 34]. Blots were imaged and quantified using the Odyssey Fc 2800 imager system (LI-COR, Lincoln, NE, USA). Three biological replicates were analyzed for each group and compared using a one-way ANOVA with Tukey’s multiple comparison tests.

RNA extraction

C. elegans were developmentally synchronized using hypochlorite treatment and then grown to adulthood at 20°C on 5× PEP plates. Plates were then washed to remove worms using M9 buffer, and worms were washed another four times in M9 buffer to remove excess E. coli. Samples were pelleted by 1-min centrifugation at 400 × g, snap-frozen in liquid nitrogen, and stored at −80°C until further processing. RNA was purified using TRIzol Reagent as recommended by the manufacturer (Thermo Fisher Scientific, Inc.). RNA was resuspended in 50 μl sterile water and concentration assessed using a NanoPhotometer^® NP80 spectrophotometer (Implen GmbH, Munich, Germany).

Quantitative RT-PCR

Complementary DNA (cDNA) was prepared using iScriptTM Reverse Transcription Supermix for RT-qPCR (Bio-Rad Laboratories, Inc., Hercules, CA, USA) as instructed by the manufacturer. In order to measure the human MAPT gene, the primer set used was (forward primer: 5′-GTGTGGCTCATTAGGCAACATCC-3′, reverse primer: 5′-CGTTCTCGCGGAAGGTCAG-3′). In order to normalize, a primer set designed to detect C. elegans rpl-32 gene was used (forward primer: 5′-GGTCGTCAAGAAGAAGCTCACCAA-3′, reverse primer: 5′-TCTGCGGA-CACGGTTATCAATTCC-3′). Three biological replicates and three technical replicates were used to assess statistical significance. Quantitative PCR (qPCR) was performed using the iTaqTM Universal SYBR^® Green Supermix kit (Bio-Rad Laboratories, Inc., Hercules, CA, USA) in a 384-well plate on a 7900HT Fast Real-Time PCR System (Applied Biosystems, Foster City, CA, USA). Data were normalized to rpl-32, an internal reference control gene, after using the delta-delta CT method [35].

Neurodegeneration

Neurodegeneration was assessed using a transgenic reporter strain with fluorescently marked GABAergic inhibitory motor neurons. C. elegans was crossed with the CZ1200 strain (juIs76[unc25p::GFP]) to allow visualization of the neurons via fluorescent microscopy. C. elegans was developmentally synchronized by selecting L4 animals 24 h prior to the scheduled experiment and grown at 20°C until reaching day one of adulthood. Animals were mounted on glass slides with a 2% agarose gel pad and 10 μl of 0.1% sodium azide. The number of ventral nerve cord GABAergic D-type motor neurons was counted under fluorescence microscopy on a DeltaVision Elite (GE, Issaquah, WA, USA) imaging system using an Olympus ×60 oil objective. Representative images were taken using the same objective. The three most rostral neurons were not included as their fluorescence could not be reliably determined due to fluorescence of the pharynx in tau transgenic animals, leaving 16 neurons for counting. Three biological replicates with at least 10 animals per replicate were analyzed for statistical significance using a one-way ANOVA with Tukey’s multiple comparisons test. Representative images were created using ImageJ to overlay GFP and POL channels [36].

Transcriptomic analysis

PolyA-enriched RNA from stage-matched day 1 adult C. elegans populations (n = 16) was sequenced on a NovaSeq X Plus (Novogene Co, Ltd., Davis, CA, USA) for ∼25 million 300 bp pair-end reads. FASTQ files were pre-processed with FASTP (version 0.24.0) [37] and aligned with STAR two-pass-mapping (version 2.7.11b) [38] against the C. elegans genome WBcel235 (version 113) [39]. Differential mRNA splicing was calculated for both novel and annotated splicing sites using SpliceWiz with default filters (version 1.8.0) [40] and DESeq2 (version 1.46.0) [41] in R (version 4.4.2). DESeq2 in R was also used to calculate differential gene expression and normalized read counts following featureCounts (version 2.20.0) in R (version 4.4.2) without multi-mapping reads [42]. Gene ontology analysis of biological processes over-represented by genes with significant intron retention (padj < 0.01 & ΔPSI > 0.1) and subject to nonsense-mediated decay (NMD) was determined by DAVID (version 2025_1) [43, 44].

Human post-mortem tissue

Samples of post-mortem brain tissue were obtained from the University of Washington Alzheimer’s Disease Research Center (ADRC) Neuropathology Core via the UW BioRepository and Integrated Neuropathology (BRaIN) Laboratory. Informed consent for research brain donation was obtained according to approved protocols. AD brain donors were chosen based upon the clinical diagnosis of dementia and neuropathologically confirmed Alzheimer’s disease neuropathologic change (ADNC) sufficient to explain dementia (intermediate or high). Brain tissues used as controls for this study were derived from age-matched cognitively normal research participants with neuropathologically confirmed low levels of ADNC. Fixation of donor brains occurred by immersion in 10% neutral buffered formalin for at least 2 weeks. The frontal cortex and medial temporal gyrus were processed and embedded in paraffin and sectioned at 5 μm thickness according to routine protocols for neuropathological analysis as described below.

Immunohistochemical evaluation

Immunohistochemistry (IHC) of TXNL4A/DIB-1/DIM-1 was performed on frontal cortex tissue (7 control cases and 21 AD cases), and IHC of PRPF8 was performed on medial temporal gyrus tissue (7 control cases and 12 AD cases). Brain sections were deparaffinized, rehydrated through ethanols, and autoclaved (TXNL4A) or microwaved (PRPF8) in citrate buffer for antigen retrieval. Sections were treated for endogenous peroxidases, blocked in 5% milk, and incubated with the following Santa Cruz Biotechnology antibodies: mouse monoclonal antibody Dim1 B-11(sc-514093; 1:50) or PRPF8 E-5 (sc-55533; 1:200) overnight at 4°C, followed by biotinylated mouse secondary antibody. Sections were incubated with an avidin-biotin complex (Vector, Vectastain Elite ABC kit) and the reaction product was visualized with 0.05% diaminobenzidine (DAB)/0.01% hydrogen peroxide. Digital images were obtained using a Leica DM6 microscope with a DFC 7000 digital camera (Leica Microsystems) and imported into Adobe Photoshop.

Quantitative analysis of immunohistochemistry

HALO digital image software (Indica Labs) was used to quantify TXNL4A and PRPF8 immunoreactivity in human brain. Brain sections were manually annotated around the regions of interest (frontal cortex or medial temporal gyrus), average staining intensity was determined to allow quantification without contribution of background staining and a common threshold was then applied to all sections for that assay. Data represent the area of positive immunoreactivity within the region of interest divided by the total annotated area. This value was multiplied by the average optical density (OD) to yield the final normalized immunoreactive area × OD. Data are displayed as the mean ± SEM. A two-tailed Student’s t-test was used to assess differences in immunoreactivity between experimental groups. Statistical analysis and graphing were performed using the Prism V8.3 software package (GraphPad).

Results

Mutations in dib-1 ameliorate tau-driven behavioral deficits

To understand potential molecular drivers of tau mediated neurodegeneration, we conducted a forward genetic screen using a previously described transgenic C. elegans model of the tau pathology in AD/ADRD [7]. Due to their rapid reproduction and ease of genetic manipulation, C. elegans have proven ideal for large, forward genetic screens [3, 8]. From a screen for strong suppressors of the tauopathy motility phenotype, we isolated an early missense mutation in the dib-1 gene that appeared to strongly suppress tauopathy phenotypes (Fig. 1A and Supplementary Fig. S1A). To confirm the causal mutation, we re-constructed the identical molecular lesions (S2L) using CRISPR in two independent alleles (bk3111andbk3112) in an otherwise un-mutagenized wild-type (N2) background. The re-constructed single point mutation shows significant amelioration of the behavioral phenotype, as quantified through movement, a well validated proxy for neurodegeneration and behavioral defects [7] in C. elegans with expression of wild-type tau (Fig. 1B) and expression of mutant tau (Fig. 1C), indicating that dib-1 mediates tau induced toxicity and neurodegeneration.

Figure 1.

Mutations in dib-1 ameliorate tau-driven motor deficits. (A) DIB-1 is a small, highly conserved protein. The mutation identified in this paper leads to a serine to leucine change (S2L) early in the amino acid sequence. Predicted protein folding from AlphaFold2. (B) Thrashing assays for wild-type (N2), dib-1 mutant, tau (high) transgenic (Tg), and dib-1 S2L; tau (high) Tg Caenorhabditis elegans (C. elegans). Tau (high) Tg animals express high levels of tau protein. Videos of one day old worms were recorded for 1 min in buffer and WormTracker was used to count the number of turns made per second; N = 3, n = 142–167. Ordinary one-way ANOVA with multiple comparisons; ns = not significant, **P< 0.01, ****P< 0.0001. (C). Mutation of dib-1 also prevents motor related deficits in C. elegans expressing mutant (V337M) tau; N = 3, n = 78–81. Ordinary one-way ANOVA with multiple comparisons; ****P< 0.0001. (D). Representative immunoblots with tau (high) Tg C. elegans, as well as two dib-1 S2L alleles; tau (high) Tg C. elegans, for DIB-1 protein. (E) Quantification of three biological replicates in D for DIB-1 protein. Ordinary one-way ANOVA with multiple comparisons. No significant differences between groups. (F) Overexpression of wild-type DIB-1 protein exacerbates behavioral defects in transgenic C. elegans expressing low levels of tau protein (Tau (low) Tg). N = 3, n = 79–120. Ordinary one-way ANOVA with multiple comparisons; ****P< 0.0001. (G) Overexpression of mutant S2L DIB-1 protein ameliorates the tau-driven behavioral defect in Tau (low) Tg C. elegans; N = 3, n = 77–103. Ordinary one-way ANOVA with multiple comparisons; ****P< 0.0001. (H) Heterozygous loss of DIB-1 using genetic balancer chromosome tmC25[unc-5(tmIs1241)] rescues behavioral defects in Tau (low) Tg C. elegans; N = 3, n = 61–103. Ordinary one-way ANOVA with multiple comparisons; *P< 0.05, **P< 0.01, ****P< 0.0001.

The high degree of homology in DIB-1 amino acid sequences [13] allowed us to take advantage of antibody cross-reactivity between species and use an antibody raised against human TXNL4A to detect the levels of DIB-1 protein in C. elegans whole animal lysates. Despite the behavioral rescue by the dib-1 S2L mutation, DIB-1 protein levels were unchanged in tau transgenic strains with and without dib-1 (S2L) mutation (Fig. 1D and E) indicating that decreased total DIB-1 protein does not drive tauopathy rescue. Rather the S2L mutation seems likely to be a change of protein function. To better understand the role of DIB-1 protein abundance in tau toxicity, we constructed several DIB-1 neuronal overexpression lines. One line had overexpression of wild-type DIB-1, while the other had overexpression of mutant DIB-1 protein with the single S2L point mutation. Overexpression of wild-type DIB-1 protein worsens behavioral defects in a low expressing tau C. elegans strain (Fig. 1F). Conversely, overexpression of S2L mutant DIB-1 protein significantly improves behavior in low expressing tau Tg animals (Fig. 1G). In order to further probe the role of DIB-1, we constructed a heterozygous DIB-1 loss of function strain, as homozygous loss of function is lethal [45], and observed behavioral rescue (Fig. 1H), suggesting that DIB-1 S2L is a dominant loss or reduction of function mutation. Behavioral rescue is slightly decreased compared to the dib-1 S2L mutant alone, suggesting that some normal dib-1 function is being preserved in the heterozygous null animals. Thus, these data support a model where tauopathy rescue occurs due to a loss or reduction of function in S2L mutant DIB-1, while preventing the embryonically lethal phenotype seen in the homozygous dib-1 null animals.

Mutations in dib-1 decrease total and phosphorylated tau, neurodegeneration, and lifespan defects in tau transgenic C. elegans

Transgenic tau C. elegans have significant total and phosphorylated tau burdens that lead to behavioral defects and neurodegeneration [7, 9, 46]. Our forward genetic screens may identify factors that alter tau levels or modulate toxic functions of tau without changing total tau [7]. To assess if behavioral improvement in dib-1 mutant strains was due to changes in tau, we used immunoblotting to quantify tau protein levels. Mutation of dib-1 decreases total tau protein, as well as tau phosphorylation at positions S202 and S396/404 (Fig. 2A–D), by over 50%. Overexpression of wild-type DIB-1 protein does not significantly increase tau burden (Supplementary Fig. S1B and C), indicating that DIB-1 OE may be driving increased tau toxicity independent of tau levels. To determine whether tau protein changes occurred due to suppression of tau transgene mRNA production, we quantified tau mRNA levels and found that they were unchanged or even slightly increased, excluding the possibility that reduced tau protein levels were due to a decrease in production or increase in turnover of mRNA (Supplementary Fig. S1D). Thus, decreased tau protein levels could be due to reduced tau transgene mRNA translation or stimulation of tau protein clearance. Translation as a whole appears unperturbed given the preservation of DIB-1 protein levels in the mutant (Fig. 1D and E), highlighting that DIB-1 may rescue tauopathy related behavioral defects by impacting tau proteostasis.

Figure 2.

Mutations in dib-1 decrease total and phosphorylated tau, neurodegeneration, and lifespan defects in tau transgenic C. elegans. (A) Representative immunoblots with tau (High) Tg C. elegans, as well as dib-1 S2L; tau (high) Tg C. elegans, for total tau, phosphorylated tau at pS396 and pS404, pS202, and tubulin (as a loading control). (B) Quantification of phosphorylated pS396/404 tau over total tau for tau (High) Tg C. elegans animals with and without dib-1 S2L mutation. (C) Quantification of phosphorylated pS202 tau over total tau for tau (High) Tg C. elegans animals with and without dib-1 S2L mutation. (D) Quantification of total tau over tubulin for tau (High) Tg C. elegans animals with and without dib-1 S2L mutation. Data are plotted as average ± SEM and analyzed using a one-way ANOVA followed by Tukey’s multiple comparisons test; **P< 0.01, ***P< 0.001, ****P< 0.0001. (E) Representative images of C. elegans with transgene juIs76 [unc25p::GFP] which drives the expression of GFP in D-type motor neurons in tau (high) Tg C. elegans with and without dib-1 S2L mutation. Red arrows are neurons, and yellow arrows represent absent neurons. (F) Quantification of total neurons lost in wild-type (N2), dib-1 S2L mutant, tau (high) Tg, and dib-1 S2L; tau (high) Tg C. elegans. Data are plotted as average ± SEM with individual data points and analyzed using a one-way ANOVA followed by Tukey’s multiple comparisons test; ns = not significant, ****P< 0.0001. (G) Mutation of dib-1 ameliorated lifespan deficits of tau Tg C. elegans; n = 105–125.

We have previously found that neurodegeneration can be quantitatively measured by scoring the 19 GABAergic inhibitory motor neurons in the ventral nerve cord using an unc-25::GFP reporter [7, 47]. Consistent with the observed behavioral improvements and decreased tau levels, the S2L mutation in dib-1 ameliorated the majority of neuronal loss in transgenic tau strains in day 1 adults, but did not completely prevent tau mediated neurodegeneration (Fig. 2E and F). Lifespan can also be used as an assessment of tau toxicity as tauopathy patients and tau transgenic animals demonstrate decreased lifespan due to neurodegenerative changes [7]. The S2L mutation in dib-1 increased lifespan of tau transgenic C. elegans but did not restore it to wild-type levels, matching findings in both behavioral phenotype and tau accumulation (Fig. 2G; Supplementary Fig. S2 and Supplementary Table S1). It is likely that decreasing dib-1 function via S2L mutation may have a deleterious effect on health separate from the tau-related behavioral rescue, which may be leading to the decreased lifespan observed in the dib-1 S2L animals. This apparent impact on the health of C. elegans makes its rescue even more interesting as it must also overcome any negative impacts of dib-1 loss of function on the animals.

dib-1 mutation drives broad changes in splicing

DIB-1/TXNL4A functions as a core component of the U4/U6.U5 tri-snRNP with a critical role in the gating of splicing kinetic decisions [13, 14, 18]. To further interrogate the nature of the dib-1 mutation, we performed bulk sequencing on polyA-enriched RNA from stage-matched day 1 adult C. elegans. The dib-1 mutants exhibit dramatic alterations to RNA splicing, measured as change of Percent Spliced In (ΔPSI), the change in the proportion of transcripts for a gene with a specific splicing event. The detected alterations (p adj < 0.01 and |ΔPSI| > 0.1) exhibit differential splicing of 413 events across 368 genes compared to wild-type animals (Fig. 3A and Supplementary Table S2A) [40]. dib-1 mutants also differentially express many genes (Supplementary Fig. S4A-B and Supplementary Table S2B). The vast majority of these splicing events, over 70%, are the increased inclusion of introns predicted to subject transcripts to degradation by NMD, consistent with a loss of conserved DIB-1/TXNL4A function in intron retention and gating of splicing kinetics (Fig. 3B) [13, 14, 18].

Figure 3.

dib-1 mutation drives broad changes in splicing. (A) A volcano plot of differential mRNA splicing in dib-1 mutants compared to wild-type controls (n = 8). Significant splicing events exhibited changes in the metric ΔPSI, the change in the proportion of transcripts for a gene with a specific splicing event, > 0.1 and p adj < 0.01 are color-coded by event type. (B) A pie chart of the number of total significant splicing events that increase intron retention and are predicted to subject transcripts to NMD relative to other significant intron retention and splicing events. (C) Heterozygous balanced prp-8 null mutation using hT2[bli-4(e937);let-?(q782,qIs48)] display improved behavior in Tau (High) Tg C. elegans; N = 3, n = 99–150; Student’s t-test; ****P< 0.0001.

To test whether generalized reduction in U4/U6.U5 tri-snRNP function might rescue tauopathy, we obtained a heterozygous loss of function mutant for PRP-8/PRPF8, another highly conserved protein localized to the U4/U6.U5 tri-snRNP [20–23]. PRP-8 plays a crucial role in splicing by coordinating rearrangements in the catalytic core of the spliceosome [48]. We examined the consequence of partial prp-8 loss of function and found that a heterozygous balanced prp-8 null strain demonstrates rescue of tauopathy related behavioral deficits (Fig. 3C and Supplementary Fig. S3A). Although tau transgenic animals in the heterozygous null prp-8 background have improved movement, we did not observe a decrease in tau protein levels (Supplementary Fig. S3B C), indicating that the PRP-8 driven alteration of tau toxicity appears independent of total tau protein. Thus, we observe that modifying distinct components of the tri-snRNP can rescue tau pathology.

Disruption of nonsense-mediated decay enhances dib-1 rescue

In an effort to identify the molecular pathways impacted by dib-1 S2L that may contribute to the rescue of tauopathy by U4/U6.U5 tri-snRNP disruption, we also performed bulk RNA sequencing on tau (high) Tg animals with and without the dib-1 S2L mutation (Supplementary Fig. S4A). Although the dib-1 S2L mutation drives significant gene expression outside of the context of tau (Supplementary Fig. S4B); consistent with our previous data, tau Tg animals with the dib-1 S2L mutation exhibit dramatically altered mRNA splicing compared to tau Tg animals (Fig. 4A and Supplementary Table S2C). In tau Tg animals, dib-1 S2L significantly (p adj < 0.01 & ΔPSI > 0.1) increases intron retention predicted to subject transcripts to NMD in 331 genes. These genes are significantly (P < 0.05) related to several critical biological pathways including regulation of transcription, adult lifespan, intracellular signaling, and DNA repair (Fig. 4B and Supplementary Table S2D). These biological pathways are consistent with known dib-1 phenotypes related to lifespan and offer compelling downstream molecular mechanisms connecting dib-1 function to tau toxicity. For example, previous research has shown the modification of DNA repair through the loss of two DNA glycosylases is protective in a C. elegans model of tauopathy [49]. The tau transgene itself is not a target for alternate mRNA processing in Tau Tg animals with mutant dib-1 (Supplementary Fig. S4C). Consistent with previous research, Tau Tg animals also exhibit disruptions to mRNA splicing significantly altering (p adj < 0.01 & ΔPSI > |0.1|) 20 events across 19 genes, including five events shared with dib-1 mutants (Supplementary Fig. S4D and Supplementary Table S2E). dib-1 S2L also drives alterations to gene expression in the context of tau Tg (Supplementary Fig. S4E and Supplementary Table S2F).

Figure 4.

Disruption of NMD enhances dib-1 rescue. (A) A volcano plot of differential mRNA splicing in tau (high) Tg animals with the dib-1 S2L mutation compared Tau (high) Tg animals (n = 8). Significant splicing events (p adj < 0.01 & |ΔPSI| > 0.1) are color-coded by event type. (B) Gene ontology analysis using the Database for Annotation, Visualization, and Integrated Discovery (DAVID) for biological processes over-represented by genes with significant (p adj < 0.01 and |ΔPSI| > 0.1) intron retention subjected to NMD in tau (high) Tg animals with the dib-1 S2L mutation compared Tau (high) Tg animal. (C) Thrashing assay of smg-2 null mutation in dib-1 S2L transgenic tau C. elegans; N = 3, n = 55–77. Ordinary one-way ANOVA with multiple comparisons; ns = not significant, **P< 0.01, ****P< 0.0001.

In continuing to probe a potential dib-1 mechanism related to U4/U6.U5 tri-snRNP activity, we turned to understanding the role of turnover of mis-spliced transcripts via NMD. NMD works both to prevent translation of incorrect mRNA transcripts through degradation and to regulate physiological gene expression through coupling with alternative splicing. NMD is hypothesized to be involved in a variety of diseases [50] and has previously been implicated in tauopathies. The highly conserved smg-2 gene functions as one of the canonical core NMD proteins that preferentially associates with mRNA transcripts with premature translation termination codons [51–53]. Knockdown of smg-2 has been observed to ablate activity of NMD due to its central role [54, 55]. We used a smg-2 knockout strain [54, 55] in the context of our dib-1 S2L mutation and observed that smg-2 knockout improves behavior in tau transgenic animals with dib-1 S2L mutations but not in tau Tg animals with wild-type DIB-1 (Fig. 4C). We observed a trend toward increased tau in the smg-2 null tau Tg strain (Supplementary Fig. S5A and B) but do not observe a further reduction in tau levels in the dib-1 S2L smg-2 null tau Tg strain (Supplementary Fig. S5C), suggesting that the further behavioral improvement may be due to modulation of tau toxicity rather than tau levels. Together, this suggests a target or set of targets of U4/U6.U5 tri-snRNP activity may connect dib-1 S2L and tau toxicity.

TXNL4A and PRP8 levels are decreased in human AD

To investigate the translational potential of our findings to human tauopathy disorders, we obtained human brain tissue samples from individuals with AD and nondemented age matched controls. TXNL4A, the human homologue of DIB-1, becomes depleted in the frontal cortex of many of the individuals with AD (Fig. 5A and B). PRP8 also becomes depleted in human AD cases, reinforcing the relevance of splicing-related proteins in tauopathies (Fig. 5C and D). To ensure that all proteins are not depleted in human AD, we stained for ALYREF, another protein identified in our forward genetic screens and observed no difference in the frontal cortex between individuals with AD and aged matched controls (Supplementary Fig. S6). To better understand the relationship between TXNL4A and AD, we stained for phosphorylated tau in the same samples using A180 antibody (Fig. 5E). We found that higher levels of TXNL4A were correlated with lower level of phosphorylated tau (P = 0.03). Additionally, we examined the correlation between TXNL4A levels and age of onset of AD symptoms (Fig. 5F) and observed that increased TXNL4A levels were correlated with a later age of onset (P = 0.03), indicating the overall potential translational relevance of TXNL4A and raising further questions about its role in AD resilience.

Figure 5.

TXNL4A and PRP8 levels are altered in human AD. (A) Representative images of TXNL4A IHC on frontal cortex for control and AD groups (upper scale bar = 500 microns, lower = 50 microns). (B) Quantification of TXNL4A staining. Data is plotted as mean ± SEM and analyzed using a t test. (C) Correlation between TXNL4A levels and phosphorylated tau levels as measured with the A180 antibody. Data analyzed using a Pearson correlation test. Plotted line is a simple linear regression, R²= 0.2238. (D). Correlation between TXNL4A levels and age of AD onset. Data analyzed using a Pearson correlation test. Plotted line is a simple linear correlation, R²= 0.2173. (E). Representative images of PRP8 IHC for control and AD groups; scale bar = 50 microns. (F) Quantification of PRP8 staining. Data are plotted as mean ± SEM and analyzed using a ttest.

Discussion

Using a classical genetics approach, we identified a single point mutation in the dib-1 gene that promotes clearance of pathological tau protein and tau-mediated neurodegenerative phenotypes. Subsequent interrogation of the U4/U6.U5 tri-snRNP complex uncovered prp-8 as another potent modulator of tau toxicity, highlighting that disruption of the tri-snRNP broadly leads to improvement in tau-driven neurodegeneration. S2L mutation of dib-1 ameliorates behavioral defects, decreases total and phosphorylated tau, prevents neuron loss, and rescues lifespan deficits in tau transgenic C. elegans. Overexpression of wild-type DIB-1 led to worsening behavioral deficits in tau transgenic C. elegans. Overexpression of mutant S2L DIB-1 in tau animals and heterozygous loss of wild-type DIB-1 in tau animals both displayed improvements in behavioral defects, indicating that the S2L mutation is loss or reduction of function in DIB-1, while preserving enough function to avoid the embryonic lethal phenotype in homozygous DIB-1 null animals. In accordance with DIB-1’s role as an essential tri-snRNP component, dib-1 S2L mutation profoundly alters splicing, specifically with respect to intron retention, with 292 intron retention events leading to NMD identified using RNA sequencing. The disruption of splicing suggests that the DIB-1 S2L mutation may modify tau toxicity through a specific set of splicing events or through a global RNA processing mechanism triggered by deficits in splicing.

In both animal models and authentic human disease, the data suggest that splicing becomes altered in AD. In the PS19 mouse model of tauopathy, mice showed dysregulation of RNA splicing, especially in synaptic transmission genes [56]. In humans, there have been a number of studies identifying alternative splicing events including increased intron retention and increased splicing deficiency in over 3000 genes [57–59]. Spliceosome components have been found in insoluble fractions of AD brains and increased intron retention can lead to decreased abundance of protein coding isoforms, highlighting splicing dysregulation as a potential factor in AD [59]. Alternative splicing, especially intron retention, may promote NMD. NMD works to prevent translation of incorrect mRNA transcripts through degradation and to regulate physiological gene expression through coupling with alternative splicing [50, 60, 61].

We demonstrate that the phenotype amelioration in tau Tg animals due to the dib-1 S2L mutation is increased when NMD is ablated. Enhanced rescue via NMD is especially interesting given that in a Drosophila melanogaster model of tauopathy, activity of NMD is reduced due to reduced RNA export and RNA accumulation within the nuclear envelope [62]. Activating NMD was sufficient to suppress neurodegeneration in tau transgenic animals [62]. NMD ablation alone does not improve tauopathy in C. elegans.

Individuals with AD exhibit disrupted splicing, specifically with respect to intron retention [59, 63], yet disrupting splicing ameliorates tau-driven behavioral deficits in C. elegans. This incongruence, coupled with evidence that NMD only enhances rescue in the context of dib-1 S2L mutation leads us to hypothesize that specific intron retention events in dib-1 mutants may be altering toxicity, and that these transcripts persist in the NMD ablated condition, further improving behavioral phenotypes. Specific intron retention events have been identified as drivers of other neurodegenerative diseases. TDP-43 is an aggregation-prone protein found in pathological inclusions of ALS and FTLD. Although TDP-43 drives broad RNA splicing changes [64], TDP-43 depletion induces cryptic exon inclusion in UNC13A, a critical gene for synapse function, and STMN-2, involved in microtubule dynamics [65, 66]. The identification of specific splicing changes in TDP-43 depletion highlights that a single, or few, splicing changes could be responsible for decreased tau toxicity, even in the face of broad splicing dysregulation. Alternately, the observed behavioral rescue could be due to a bulk effect from the production and subsequent retention of incorrectly spliced transcripts, especially those with intron retention events. Given the emerging role of tau as an RNA binding protein, it is possible that increased cellular intronic mRNA may be affecting tau toxicity.

To examine the translational relevance of these findings, we characterized the human homolog of DIB1, TXNL4A, in human postmortem brain tissue. In affected regions of the AD brain, TXNL4A becomes depleted from neurons. Levels of TXNL4A correspond with vulnerability to pathological tau accumulation as pTau burden negatively correlates with TXNL4A levels. Further, low TXNL4A levels correspond with an earlier age of AD onset.

Prior to this study, no investigation of DIB-1/TXNL4A in AD has been reported. Although it seems paradoxical that lower levels of TXNL4A would be correlated with earlier AD onset, lower age of onset, and higher tau burden, this pattern is seen with other nuclear speckle proteins related to tau toxicity including MSUT2, PABPN1, SON, and SRRM2 [9, 46, 67]. For sut-2/ MSUT2, the initial finding was also from a C. elegans screen and both MSUT2 and its binding partner PABPN were shown to be co-depleted in AD [9]. Loss of function in PABPN1 was identified as a cause of ocular pharyngeal muscular dystrophy, a secondary dementia with tau neuropathology [46]. Taken together these observations suggest a global disruption of nuclear speckles in more severe disease states.

In the case of TXNL4A in AD, the disease involvement seems clear and the C. elegans work strongly suggests the molecular mechanism must be through splicing changes, but we lack insight into the exact molecules mediating tauopathy clearance. The discrepant relationship between the abundance of pathological tau and the abundance of TXNL4A remains a mystery. We hypothesize that we may be observing a survivor bias, where the remaining neurons are those that were the healthiest, and those with the highest tau loads have degenerated. These findings, combined with its central role in splicing, suggest a more careful interrogation of the role of tau in aberrant splicing, potentially illuminating another pathway through which tau exerts its pathogenic effects. Future experiments should include attempts to identify these pathways affected by intron retention that could be modulating tau pathology, to devise more specific targets for eventual translation into humans.

Supplementary data

Supplementary data are available at NAR Molecular Medicine online.

Conflict of interest

None declared.

Funding

This work was supported by grants from the Department of Veterans Affairs [IK6BX006467 to BK and I01BX005762 to N.L.] and National Institutes of Health [T32AG052354 to K.L., F99AG088436 to R.E., R01AG066729 to N.L., R01AG084552 to R.K., R01NS064131 and R01AG084680 to B.K., and P30AG066509 and U19AG066567 to C.K. and C.L.], and the Nancy and Buster Alvord Endowment [to C.K.].

Data availability

Analyzed data are included within the manuscript and in Supplementary Information. RNA sequencing data are available on the GEO website under accession number GSE299236.