Neuronal TDP-43 pathology marks most cases of amyotrophic lateral sclerosis (ALS) and frontotemporal dementia (FTD), but how does a single protein fit with such a wide range of symptoms? Researchers led by Jemeen Sreedharan at King's College London addressed this question by using CRISPR gene editing to place a single pathogenic mutation in mouse TDP-43, creating animals that express a human disease mutation within its normal mouse genomic context. As reported in the March 19 Nature Neuroscience, the mice struggled to pay attention and barely learned, not unlike people with FTD. Their motor functions seemed normal. The mutant TDP-43 accumulated in the nuclei of cortical neurons, where it skewed the splicing of many transcripts, including its own and those of tau. Notably, neuroprotective gene expression climbed in the cortical neurons of mice whose behavior problems were mild, while neurotoxic gene expression climbed in the cortical neurons of animals whose behavior problems were severe. Such expression differences among almost genetically identical mice suggest that epigenetic mechanisms are at play, and offer up a fresh batch of therapeutic targets to explore.

  • TDP-43-Q331K knock-in mice have no motor problems but are inattentive and don’t learn.  
  • The severity of the behavioral phenotype varies, and correlates with gene expression changes.  
  • They have extensive splicing changes, including longer forms of tau.

“There have been many TDP-43 models reported, but TDP-43-induced toxicity can be severe with a narrow gene dose-response,” wrote Ronald Klein of Louisiana State University in Shreveport. “This new knock-in mouse with endogenous expression levels should be more relevant than overexpression models, and the data support that TDP-43 can be a driver of disease.”

While the ALS/FTD spectrum encompasses wide variation in the nature and severity of symptoms, they all emerge from a shared pathology of TDP-43 cytoplasmic inclusions. This pathology occurs in sporadic forms of ALS/FTD and in rare cases is caused by pathogenic mutations. Some transgenic mouse models created to understand the role of TDP-43 overexpress wild-type or mutant forms, others force it to accumulate in the cytoplasm, and their pathological and behavioral phenotypes vary substantially (see Alzforum TARDP models).

First author Matthew White and colleagues set out to generate a more physiologically relevant model. Specifically, they wanted to mutate endogenous mouse TDP-43 without disrupting the context of the gene. They used CRISPR to place a point mutation in the mouse TDP-43 gene, which switched a lysine for a glutamine at a position equivalent to amino acid 133 of the human gene. The human ALS-causing Q331K mutation resides in a highly conserved region in the protein’s C-terminus that is reportedly involved in protein-protein interactions. Notably, an existing model slightly overexpressing this same mutant under control of the mouse prion promoter develops ALS-like motor neuron loss (Arnold et al., 2013).

The new CRISPR Q331K-TDP-43 mice had a subtler phenotype. Their most common symptom—less time per day spent walking—cropped up around four months of age, mostly in males. Further breeding generated fewer males in each litter, suggesting the mutation was deleterious for male embryos. ALS and FTD are known to disproportionately affect men, and the researchers chose to limit their analyses to male animals for the rest of the study.

Not only were the male mutant mice reticent walkers, they were also uncoordinated, falling off a spinning rod more easily than control mice. However, the researchers discovered that neither of these deficits were caused by inherent motor problems. Rather, the mutant mice ate more than control mice, and the extra ounces they put on appeared to make them sluggish and unbalanced. The researchers drew this conclusion because mutant mice put on a calorically restricted diet performed as well as control mice. A look at the motor neurons of five-month-old mice supported this idea: The cells appeared normal in quality and number, had no obvious signs of TDP-43 pathology, and neuromuscular junctions appeared fully intact.

Still, the researchers wondered if gene expression changed in the motor neurons. They isolated motor neurons from five-month-old mice by laser capture and analyzed their transcriptomes. The 31 gene expression changes they found in Q331K-TDP-43 knock-in neurons versus controls included upregulation of agrin, a protein involved in clustering acetylcholine receptors at neuromuscular junctions. The most dialed-up gene was for aldehyde oxidase 1, an enzyme that converts retinaldehyde to retinoic acid to protect neurons following axonal injury. The researchers hypothesized that this might explain the resilience of Q331K-TDP-43 motor neurons.

White next put the mice through behavioral tests to tease out FTD-like symptoms. To assess learning and attention, he used the five-choice serial reaction time task. In this test, one of five spots on a touchscreen is briefly illuminated, and if a mouse taps it with its nose, it is rewarded with a mouse-size milkshake. The mutant mice took longer to learn this task than controls. Once they did, they tapped as accurately as control mice, however, they often just did not choose, suggesting an attention deficit. Other tests ruled out a deficit in motivation. Mutant mice also performed poorly on the novel-object-recognition test of memory. Though the behavior deficits were statistically significant on a group level, the researchers noticed a wide range of performance among the mice.

Losing Their Marbles?

Some mutant TDP-43 knock-in mice buried marbles (left, MB+), while others left them in the open (right, MB–). [Courtesy of White et al., 2018.]

To dig deeper into these behavior findings, the researchers examined neurons in the frontal cortices of the mice. Those from the knock-ins contained 45 percent more TDP-43 protein within the nucleus than did control neurons, but no cytoplasmic inclusions. In transcriptomic analyses, 171 genes were up- and 233 genes downregulated in frontal cortices of knock-ins compared to controls. Interestingly, TDP-43 itself was upregulated by 14 percent. The TDP-43 protein is known to manage its own expression by binding to its transcript, suggesting that this autoregulatory mechanism was defective in the knock-in mice.

Another example of downregulation was the gene encoding parvalbumin, a calcium-buffering protein of GABAergic inhibitory interneurons whose expression is down in ALS patients (Nihei et al., 1993). When the scientists immunostained the frontal cortices of their knock-in mice, they found 25 percent fewer parvalbumin-positive neurons. While loss of these interneurons could possibly explain the mice’s attention deficit, it is unclear if these neurons are gone or simply not making parvalbumin.

Because TDP-43 plays a role in splicing, the researchers also searched the gene expression data set for splicing differences and found 138 changes in 106 genes in the knock-in mice. Once again, the TDP-43 transcript itself was affected; mutant knock-ins retained 80 percent more of the mRNA-stabilizing intron 7. For its part, mutant tau mRNA ended up with more splice variants including exons 2 and 3. They encode tau’s N-terminus and correlate with tau localization to the somatodendritic compartment. However, immunostaining the mice’s frontal cortices showed no difference in tau localization or aggregation. The researchers did identify a TDP-43 binding site located just upstream of exon 2 within an intron of the tau gene, suggesting tau is a bona fide TDP-43 splicing target. Other known TDP-43 splicing targets were altered as well, including Sort1. Together with the increase in TDP-43 expression, the findings point to a gain of function for the Q331K-TDP-43 mutation.

By 20 months of age, gene expression was changed much more drastically in the mutant mice. More than 1,000 genes were expressed differently between mutant and wild-type mice. Downregulation of genes involved in parvalbumin transcription and GABAergic function suggested deepening problems with inhibitory interneurons with age. Multiple ALS/FTD-associated genes were downregulated, including Tbk1, Chmp2b, and Erbb4. Expression of genes involved in neurodegeneration and neuroinflammation rose with age.

Changes with Age. Differences in gene expression (top) and splicing (bottom) between wild-type and mutant mice grew with age. [Courtesy of White et al., Nature Neuroscience, 2018.]

Could differential gene expression explain why some mice outperformed others on the behavior tests? To address this, the researchers took advantage of the variation in responses when they placed marbles in the animals’ cages. Control mice instinctively buried marbles, but some of the knock-ins had no interest in doing so. Comparing RNA sequencing data between mutant "buriers" and "non-buriers," White identified 410 gene expression and 61 splicing differences. These changes were distinct from those that distinguished controls from the knock-ins as a whole. Strikingly, neither expression levels nor splicing of TDP-43 distinguished the groups, suggesting other genes modified the behavior.

One standout was ataxin-2. Compared with normal controls, this gene was downregulated in marble buriers and upregulated in non-buriers. Aaron Gitler’s group at Stanford University previously reported that ataxin-2 exacerbated TDP-43 toxicity in flies and yeast, and expanded CAG repeats in the gene elevated ALS risk in people (Aug 2010 news and Apr 2017 news on Becker et al., 2017). Another interesting hit was the chromatin modeling gene Arid4a; it was down in buriers and up in non-buriers. Loss-of-function mutations in this gene suppress TDP-43 toxicity in flies (Sreedharan et al., 2015). The buriers also upregulated genes involved in translation and axonal repair, including the gene encoding myelin basic protein.

Marbles Stratify. Gene expression profiles differed between knock-in mice depending on whether they did (MB+) or didn’t (MB-) bury marbles. [Courtesy of White et al., Nature Neuroscience, 2018.]

Sreedharan proposed that when mutant TDP-43 somehow derails autoregulation of its own transcript, this elevates TDP-43 to evoke a gain of toxic function. This stands in contrast to the idea of loss of function caused by TDP-43 mislocalization to the cytoplasm. For example, a new study led by Christine Vande Velde at the University of Montreal reported that TDP-43’s sequestration in cytoplasmic aggregates in human cell culture prevented it from splicing HNRNPA1, which led to retention of an exon that promoted toxic fibrillization of the HNRNPA1 protein (Deshaies et al., 2018). However, Sreedharan’s mice never developed cytoplasmic inclusions of TDP-43, a pathology that he suggested might only occur in the end stages of disease. Instead, he proposed that epigenetic differences between the animals could promote more neuroprotective gene expression profiles in some mice, thus explaining their varied behavioral phenotypes.

“The variability in some of the behavioral deficits in the authors’ mutant mice could be a key advantage and offer an opportunity to harness that variation to discover disease modifiers,” commented Gitler and Lindsay Becker at Stanford. They were particularly interested in the role of ataxin-2. “Now the challenge (and opportunity) will be to chase down the mechanisms regulating ataxin-2 expression levels, be it at the transcriptional, post-transcriptional, translational, or post-translational level. If we can identify the upstream regulators of ataxin-2 levels, then targeting those regulators could provide a therapeutic opportunity.”

Sandrine Da Cruz of the University of California, San Diego, said this new study is an important contribution to the field of ALS/FTD research, as it represents the first knock-in mutant TDP-43 mouse model. She was impressed by the behavioral stratification experiments, but cautioned that the meaning of the marble-burying assay is unclear. Some researchers interpret it as a test of anxiety, she said, adding that it would be exciting to see if similar gene expression differences dictate variability on other behavioral tasks.

Da Cruz was surprised at the absence of a motor phenotype, as other animal models of TDP-43, including one that Da Cruz helped generate, have pronounced motor dysfunction. Sreedharan told Alzforum that with age, the animals do seem to start to develop motor problems as neuroprotective defenses wear down and inflammatory responses mount. The endogenous expression levels of the mutant gene could explain why it takes longer for motor neurons to succumb to the toxicity in this model than in overexpression models, he added.

One aspect that needs further investigation is how the Q331K mutation affects autoregulation of the TDP-43 transcript. Da Cruz commented that studies from multiple labs implicate the 3' UTR, which differs in sequence between human and mouse. Therefore, whether the mutation inflicts similar autoregulatory changes in the human and mouse versions of the gene is unclear, she said.

Expressing the mutant TDP-43 gene within its normal genetic environs was a smart move, Gitler and Becker added. “This new mouse model will be useful in investigating the TDP-43 disease cascade in a more physiological system,” they wrote. “For example, the selective vulnerability of parvalbumin interneurons would not be nearly as convincing in a transgenic mouse line carrying an exogenous, strong promoter with its own pattern of differential expression.”—Jessica Shugart


  1. This is a smart approach because there will be no overexpression artifacts.

  2. This is an important new paper. The authors created a TDP-43 knock-in mouse with the ALS-associated mutation Q331K. Other labs have used BAC (bacterial artificial chromosome) systems or large regions of the human TARDBP genomic region to introduce mutant TDP-43 into mice at near-normal levels (Swarup et al., 2011; Mutihac et al., 2015). Another group knocked in human TDP-43 cDNA with an ALS-causing mutation into the mouse Tardbp locus (encoding TDP-43) (Stribl et al., 2014). There are also knock-in TDP-43 rat models and many other transgenic TDP-43 mouse models where TDP-43 is expressed at much higher levels. The advantage of this model is that it is a single base-pair knock-in into the endogenous mouse Tardbp locus.

    Similar to other TDP-43 mouse models without high levels of TDP-43 overexpression, the authors demonstrate mild motor and cognitive impairment phenotypes as well as gene expression changes. The behavioral impairments are convincingly demonstrated by an array of behavioral tests.

    The TDP-43 mouse models with strong behavioral and pathological phenotypes also have very high levels of TDP-43 overexpression. This new mouse model will be useful in investigating the TDP-43 disease cascade in a more physiological system. For example, the result showing selective vulnerability of parvalbumin interneurons is very interesting. This kind of cell-type-specific result would not be nearly as convincing in a transgenic mouse line using an exogenous, strong promoter with its own pattern of differential expression. 

    The authors performed RNA analysis by laser capture of motor neurons in the spinal cord and by using whole-tissue analysis in the frontal cortex. They found differential gene expression and splicing changes in both contexts, but much more so for the whole-tissue analysis in the frontal cortex. The whole-tissue analysis has the added confound of possible differences in the cellular makeup of the tissue (such as selective death of parvalbumin neurons). However, this technique is very often used and the data are still interesting and informative.

    The variability in some of the behavioral deficits in the authors’ mutant mice could be a key advantage and offer an opportunity to harness that variation to discover disease modifiers. For example, the results comparing mutant mice that are deficient in marble burying to those who are not is particularly intriguing. One particular gene caught our attention. The authors found that ataxin-2 expression is reduced in the knock-in mice that are able to bury marbles and elevated in those that are defective in marble burying behavior. Ataxin-2 has previously been connected to TDP-43 and human disease. Lowering levels of ataxin-2 is sufficient to suppress TDP-43 toxicity in yeast, flies, and mice (Elden et al., 2010; Becker et al., 2017). In humans, intermediate-length polyQ expansions in ataxin-2 increase risk for ALS (Elden et al., 2010). The intermediate-length polyQ expansions increase the stability of ataxin-2 (Hart and Gitler, 2012), which is expected to increase levels of ataxin-2.

    The authors’ results showing that lower ataxin-2 levels correlate with less severe TDP-43 phenotypes and higher ataxin-2 levels correlate with more severe TDP-43 phenotypes are interesting. Now the challenge (and opportunity) will be to chase down the mechanisms regulating ataxin-2 expression levels, be it at the transcriptional, post-transcriptional, translational, or post-translational level. If we can identify the upstream regulators of ataxin-2 levels, then targeting those regulators could provide a therapeutic opportunity.


    . Therapeutic reduction of ataxin-2 extends lifespan and reduces pathology in TDP-43 mice. Nature. 2017 Apr 20;544(7650):367-371. Epub 2017 Apr 12 PubMed.

    . Ataxin-2 intermediate-length polyglutamine expansions are associated with increased risk for ALS. Nature. 2010 Aug 26;466(7310):1069-75. PubMed.

    . ALS-associated ataxin 2 polyQ expansions enhance stress-induced caspase 3 activation and increase TDP-43 pathological modifications. J Neurosci. 2012 Jul 4;32(27):9133-42. PubMed.

    . TARDBP pathogenic mutations increase cytoplasmic translocation of TDP-43 and cause reduction of endoplasmic reticulum Ca²⁺ signaling in motor neurons. Neurobiol Dis. 2015 Mar;75:64-77. Epub 2014 Dec 17 PubMed.

    . Mitochondrial dysfunction and decrease in body weight of a transgenic knock-in mouse model for TDP-43. J Biol Chem. 2014 Apr 11;289(15):10769-84. Epub 2014 Feb 10 PubMed.

    . Pathological hallmarks of amyotrophic lateral sclerosis/frontotemporal lobar degeneration in transgenic mice produced with TDP-43 genomic fragments. Brain. 2011 Sep;134(Pt 9):2610-26. Epub 2011 Jul 13 PubMed.

Make a Comment

To make a comment you must login or register.


Research Models Citations

  1. TDP-43 (Q331K)
  2. TDP-43 (Q331K) Knock-In (Line 52)

News Citations

  1. ALS—A Polyglutamine Disease? Mid-length Repeats Boost Risk
  2. Two For One? ASOs for Ataxin Allay ALS and SCA2 in Mice

Paper Citations

  1. . ALS-linked TDP-43 mutations produce aberrant RNA splicing and adult-onset motor neuron disease without aggregation or loss of nuclear TDP-43. Proc Natl Acad Sci U S A. 2013 Feb 19;110(8):E736-45. PubMed.
  2. . Patterns of neuronal degeneration in the motor cortex of amyotrophic lateral sclerosis patients. Acta Neuropathol. 1993;86(1):55-64. PubMed.
  3. . Therapeutic reduction of ataxin-2 extends lifespan and reduces pathology in TDP-43 mice. Nature. 2017 Apr 20;544(7650):367-371. Epub 2017 Apr 12 PubMed.
  4. . Age-Dependent TDP-43-Mediated Motor Neuron Degeneration Requires GSK3, hat-trick, and xmas-2. Curr Biol. 2015 Aug 17;25(16):2130-6. Epub 2015 Jul 30 PubMed.
  5. . TDP-43 regulates the alternative splicing of hnRNP A1 to yield an aggregation-prone variant in amyotrophic lateral sclerosis. Brain. 2018 May 1;141(5):1320-1333. PubMed.

Other Citations

  1. Alzforum TARDP models

Further Reading

Primary Papers

  1. . TDP-43 gains function due to perturbed autoregulation in a Tardbp knock-in mouse model of ALS-FTD. Nat Neurosci. 2018 Apr;21(4):552-563. Epub 2018 Mar 19 PubMed.