TBK1 overexpression reduces neuronal loss in the P497S UBQLN2 mouse model of ALS/FTD

Abstract

Missense mutations in UBQLN2 are linked to dominant inheritance of amyotrophic lateral sclerosis (ALS) which is frequently accompanied by frontotemporal dementia (FTD). The encoded UBQLN2 protein functions to maintain proteostasis, collapse of which is increasingly being realized as the cause of many neurodegenerative diseases. During investigations of our UBQLN2 mouse models of ALS/FTD, we observed a significant decline in Tank-binding kinase 1 (TBK1) protein in end-stage mice. The decline could be significant, as haploinsufficiency of TBK1 expression is linked to ALS/FTD. To determine whether the reduction in TBK1 levels is responsible for driving pathogenesis, we crossed P497S UBQLN2 transgenic (Tg) mice with BacTBK1 Tg mice that overexpress TBK1 and analyzed the progeny for signs of pathology. The analysis revealed that double transgenic mice had a significant reduction in neurodegeneration in both the brain and spinal cord (SC) compared to P497S single-Tg mice. Double immunofluorescence staining of P497S mouse tissue revealed TBK1 colocalizes with UBQLN2 aggregates in spinal motor neurons. Biochemical extraction studies of mouse brain tissue revealed increased enrichment of TBK1 along with wild type and mutant UBQLN2 proteins in detergent-insoluble material, suggesting TBK1 gets sequestered by UBQLN2 aggregates. GST-pulldown assays revealed that UBQLN2 binds directly with TBK1, but paradoxically the P497S mutation was found to reduce TBK1 binding. Turnover studies indicated that loss of UBQLN2 destabilizes TBK1, providing another route for its reduction. These findings lead us to propose that efforts directed toward increasing TBK1 expression may provide a therapeutic approach for treating ALS/FTD caused by UBQLN2 mutations.

1. Introduction

Amyotrophic lateral sclerosis (ALS) is a rapidly progressive neurodegenerative disorder that causes loss of motor function. Approximately 50% of ALS patients also develop dementia. Disease progression is usually rapid, with approximately 70% of patients dying within 3 years of diagnosis, and the remainder having slower progression for an unknown reason (Levison et al., 2024). Pathologically, the disease is characterized by loss of both upper and lower motor neurons (Brown and Al-Chalabi, 2017; Masrori and Van Damme, 2020; van Es et al., 2017). Additionally, over 97% of ALS patients develop TDP-43 pathology, which is characterized by clearance of TDP-43 from its normal high nuclear concentration, often accompanied by displacement and aggregation in the cytoplasm (Kwong et al., 2007; Neumann et al., 2006).

Besides the pathologic clues, key insight into the etiology of ALS has come from the discovery that mutations in over 30 different genes are linked to hereditary ALS (Ghasemi and Brown Jr., 2018; Mead et al., 2023). Interestingly, many of these same mutations also cause frontotemporal dementia (FTD), suggesting ALS and FTD are part of a continuum in a spectrum of disorders. Interestingly, factors that regulate proteostasis comprise the largest category of genes in which mutations are found to cause ALS/FTD.

The studies we present here focus on UBQLN2 and TBK1, two genes in which mutations cause ALS and/or FTD, and play important roles in regulating proteostasis. Interestingly, UBQLN2 and TBK1 interact with one another, suggesting they may function together (Chen et al., 2020).

Mutations in UBQLN2 were first linked to ALS/FTD in 2011(Deng et al., 2011). Many additional mutations have since been discovered, most of which map in or near a proline-rich (PXX) repeat sequence that is only present in UBQLN2, and not in other UBQLN family members (Lin et al., 2022; Matthews and Whiteley, 2025). The UBQLN family of proteins are important regulators of proteostasis (Lin et al., 2022; Matthews and Whiteley, 2025). They principally function to shuttle ubiquitinated proteins to either the proteasome or autophagosome for degradation (Chang and Monteiro, 2015; Kleijnen et al., 2000; Ko et al., 2004; Lim et al., 2009; N’Diaye et al., 2009; Rothenberg et al., 2010). The proteins can also act as chaperones, by binding and shielding hydrophobic segments in polypeptides prior to protein assembly (Itakura et al., 2016; Lin et al., 2021; Onwunma et al., 2026; Whiteley et al., 2017; Wu et al., 2020). They also assist in protein refolding, an activity that has been linked to their ability to recruit heat shock and related proteins (Gavriilidis et al., 2018; Hjerpe et al., 2016; Kim et al., 2025; Phung et al., 2023; Teyssou et al., 2017; Zhang et al., 2021). Recent evidence suggests that they can also recruit ubiquitin ligases to aid in protein quality control (Chen et al., 2026; He et al., 2026; Liu et al., 2025; Scheutzow et al., 2025). Importantly, studies of ALS/FTD mutant UBQLN2 proteins in cells and animals have identified defects in most of the aforementioned functions, which are speculated to lead to collapse of the proteostasis network, and cause disease (reviewed by (Lin et al., 2022; Matthews and Whiteley, 2025; Renaud et al., 2019)). However, questions remain as to why the collapse causes ALS and FTD rather than another disease.

Mutations in TBK1 were first identified in 2015 as a cause of ALS/FTD (Cirulli et al., 2015; de Majo et al., 2018; Freischmidt et al., 2015). Several additional mutations have since been discovered. The majority of TBK1 mutations lead to loss of TBK1 protein expression, strongly implying haploinsufficiency of TBK1 expression drives ALS pathogenesis. TBK1 is a serine/threonine protein kinase that regulates interferon signaling and autophagy (Ahmad et al., 2016; Gurfinkel et al., 2022; Weidberg and Elazar, 2011). Complete loss of TBK1 is lethal (Bonnard et al., 2000), whereas conditional KO of the gene in the brain induces severe autophagy defects (Duan et al., 2019). TBK1 regulation of interferon signaling occurs mainly through phosphorylation of IRF3, leading to its dimerization and translocation into the nucleus, where it induces type I and III interferons. Likewise, TBK1 phosphorylates p62 and optineurin, two other ubiquitin-binding factors, promoting their interaction with the autophagic machinery and thereby stimulating autophagy (Ahmad et al., 2016). Interestingly, Chen et al. showed that TBK1 also binds UBQLN2 and this binding regulates IRF3 signaling (Chen et al., 2020). They further reported that TBK1 binds more strongly with ALS mutant UBQLN2 proteins carrying P497H and P506T mutations than with WT UBQLN2, and the stronger interaction somehow dampens IRF3 signaling.

We previously described the generation of two different UBQLN2 transgenic (Tg) mouse models of ALS/FTD, one carrying the P497S UBQLN2 mutation and the other carrying the P506T mutation (Le et al., 2016). We showed that both lines recapitulate key pathologic features found in human carriers of the UBQLN2 mutations, including motor neuron disease. During immunoblot characterization of the lines, we found that TBK1 levels were significantly reduced in SC lysates prepared from mice with end-stage disease, compared to age-matched Non-Tg controls (Fig. 1A–D). A similar decline in TBK1 was found in the SOD1^G93A mouse model of ALS (Duan et al., 2019). Interestingly, increased TBK1 through AAV9-Tbk1 overexpression did not delay onset of disease in the SOD1^G93A mice, but it did extend survival by 10 days (Duan et al., 2019). In this report, we describe experiments aimed at determining whether the reduction in TBK1 levels contributes to disease pathogenesis in the P497S UBQLN2 Tg line. Indeed, we show that restoration of TBK1 levels through TBK1 overexpression alleviates pathological signatures of disease in this model.

Fig. 1.

TBK1 and UBQLN2 levels in SC lysates from Tg P497S and P506T UBQLN2 end-stage mice. (A) SC lysates from different number of independent age-matched Non-Tg and P497S animals immunoblotted for the proteins shown. (B) Quantification of TBK1 levels in the blots shown in A. ***P < 0.001. (C) Similar to A but for age-matched Non-Tg and P506T mice. (D) Quantification of TBK1 levels in the blots shown in B. (E) Confocal microscopy showing colocalization of UBQLN2 and TBK1 proteins in motor neurons in the SC of P497S and P506T animals. Arrows indicate inclusions that were positive for both proteins. **P < 0.01.

2. Results

2.1. TBK1 colocalizes with UBQLN2 inclusions in spinal motor neurons (MN) of end-stage P497S and P506T UBQLN2 Tg mice

We conducted double immunofluorescence staining of SC sections from end-stage diseased P497S and P506T UBQLN2 Tg mice for TBK1 and UBQLN2 proteins. The staining revealed good colocalization of the TBK1 and UBQLN2 fluorescence signals, mainly in the inclusions that were found within MNs of both mutant lines, but less so in the inclusions that appeared to be located outside the cells (Fig. 1E). Line scan intensity measurements confirmed excellent co-alignment of the two fluorescence signals over the puncta in the MN (Fig. S1). This strong TBK1 staining in the SC, with the contradictory finding of reduced TBK1 by immunoblotting, raised the possibility that a fraction of TBK1 may redistribute to UBQLN2 inclusions in diseased spinal MN, potentially reducing the available soluble or functional TBK1 pool.

2.2. Manipulation of TBK1 levels in P497S UBQLN2 mice

We next designed a genetic strategy to test whether increasing TBK1 levels would attenuate disease in the P497S mutant line. Accordingly, we crossed P497S UBQLN2 Tg mice with Bac-TBK1 Tg mice that overexpress TBK1 (Fingert et al., 2017) and examined the double-Tg mice for evidence that disease had been attenuated (Fig. 2A). The Bac-TBK1 transgenic line carries a single extra copy of the human TBK1 gene under the control of its normal promoter (Fingert et al., 2017). Bac-TBK1 mice have normal lifespans and have no noticeable phenotype, apart from having a 13% reduction of retinal ganglion cells in the retina (Fingert et al., 2017). The success of the genetic strategy was evaluated using pathologic criteria, whose signature we found was linked to disease in the P497S line. Because the current P497S colony displayed an attenuated behavioral phenotype relative to the originally line (Le et al., 2016), we focused our examinations on alterations in neuropathological signatures of disease, which remained robustly detectable in 52-week-old animals of the current colony, including a reduction of neurons in the hippocampus and MNs in the SC, and evidence of TDP-43 pathology in spinal MNs.

Fig. 2.

Schematic and body weight analysis of the mouse groups used in the study. (A) Genetic strategy used to establish the four mouse genotypes used for the study; Non-Tg, TBK1, P497S, and TBK1/P497S. (B and C) Male (n = 6) (B) and female (n = 6) (C) body weights of the four genotypes over 52 weeks.

The transgenic cross between the P497S and TBK1 animals yielded progeny carrying all four permutations of the two parental transgenes with the expected Mendelian frequency of segregation for the four genotypes. Body weight measurements indicated that single-Tg TBK1 male and female mice showed a normal progressive increase in body weight with age that was not statistically different from the Non-Tg group (Fig. 2B and C). By contrast, the single-Tg P497S male and female mice had the lowest body weight of all the genotypes, with female P497S mice differing significantly from the Non-Tg group after 44 weeks of age (Fig. 2B and C). Interestingly, the double-Tg TBK1/P497S mice had slightly higher body weights than the single-Tg P497S mice, which was particularly noticeable for the female cohorts (Fig. 2B and C). All the mice were euthanized at 52 weeks of age after which their brains and SCs were collected for pathologic examination.

2.3. Restoration of TBK1 levels in double-Tg P497S/TBK1 mice

Immunoblots were performed on lysates prepared from the whole brain and SC of the 52-week-old animals to examine protein changes in the four genotypes. For this analysis, lysates from three independent animals for each of the four genotypes were run together on the same gel to directly compare protein levels across the genotypes. Male and female lysates were analyzed separately (Fig. 3 and Fig. S3). The UBQLN2 blots revealed the same pattern for both the brain and SC: —the presence of two prominent bands in all of the mice carrying the P497S UBQLN2 transgene, but only one, the larger-size band in those lacking the transgene (Fig. 3A, C and Fig. S3A, C). The smaller-size band is consistent with expression of the transgenic human UBQLN2 protein, whose length is smaller than the mouse UBQLN2 protein. Parallel blots of the same lysates revealed a variation between the brain and SC in TBK1 levels. In the brain, TBK1 levels were lower and found to be significantly reduced in the female single-Tg P497S UBQLN2 animals, when compared with the Non-Tg animals. By contrast, single-Tg TBK1 animals had elevated levels of TBK1 protein in the brain compared to the age-matched Non-Tg animals, consistent with overexpression driven by the additional TBK1 transgene. Importantly, TBK1 levels in both male and female P497S/TBK1 double Tg mice were restored to approximately the same level as the Non-Tg animals (Fig. 3 and S3). The blots for the SC lysates showed a similar pattern of TBK1 changes as found in the brain, except that TBK1 levels were not substantially reduced in the single-Tg P497S animals. It is possible that the lack of reduction of TBK1 levels in the SC compared to that seen in the line that had been backcrossed fewer times may be the reason for the weaker motor phenotype seen in the current line. However, similar to the findings in the brain, TBK1 levels in the SC were noticeably higher in the single-Tg TBK1 mice compared to the Non-Tg animals. Importantly, TBK1 levels were noticeably higher in the double-Tg TBK/P497S mice compared to single-Tg P497S animals, suggesting our genetic strategy to increase its expression in the mutant animals had worked in both the brain and SC. These results confirmed that the BacTBK1 transgene increased TBK1 levels in the P497S background, allowing us to determine whether its increase would attenuate neuropathological signatures of disease.

Fig. 3.

Analysis of protein expression in the brain and SC in male mice at 52 weeks of age. (A) Immunoblots of brain lysates from 3 independent mice for the four genotypes blotted for the proteins shown. (B) Quantification of protein levels shown in A. ns > 0.05, **P < 0.01, ****P < 0.0001. (C) Immunoblots of SC lysates for the four genotypes blotted for the proteins shown. (D) Quantification of protein levels shown in C. ns > 0.05, *P < 0.05, **P < 0.01, ***P < 0.001.

2.4. Transgenic overexpression of TBK1 alleviates motor neuron loss in P497S UBQLN2 mice

Our prior characterizations of the P497S line (Le et al., 2016; Wang et al., 2020) had indicated that at 52 weeks of age, P497S mice have significantly fewer MNs as well as prominent evidence of TDP-43 pathology in MNs of the SC. We therefore examined sections of the SC to determine if these pathologies still manifest in the current line, and if so, whether TBK1 overexpression had reduced these pathologies in the double-Tg animals. Accordingly, the number of MNs in the SC of 52-week-old animals was quantified using stereological principles for all four genotypes. The quantification revealed that both male and female single-P497S Tg mice used in the present study had significantly fewer 20–50% fewer MNs compared to age-matched Non-Tg mice (Fig. 4A–D). Single-Tg TBK1 mice had a similar number of MNs compared to the Non-Tg mice, showing that TBK1 overexpression does not cause MN demise. Intriguingly, the number of MNs in the double-Tg P497S/TBK1 mice was not significantly different to that of Non-Tg mice, strongly indicating increasing TBK1 levels prevents MN loss in animals expressing the P497S transgene (Fig. 4A–D).

Fig. 4.

Rescue of MN loss in the P497S Tg line by TBK1 overexpression. (A) SC sections stained with cresyl violet to identify MN for both male and female mice at 52 weeks of age for the four genotypes. (B) Immunofluorescent staining of the SC sections for both sexes for ChAT to confirm changes in MNs. (C and D) Quantification of MNs in male (C) and female (D) SC sections for the four genotypes. ns > 0.05, ****P < 0.0001.

The assessment of TDP-43 pathology in the mice, revealed single P497S Tg animals had significantly higher number of spinal MN that displayed robust TDP-43 pathology, evidenced by either nuclear clearance or displacement of the protein in the cytoplasm (Fig. 5A and B and S5A and B). By contrast, the double-Tg P497S/TBK1 mice had a dramatic reduction of such pathology, showing more uniform concentration of TDP-43 in the nucleus. Single-Tg TBK1 animals had a similar distribution and concentration of TDP-43 in the nucleus as Non-Tg animals, indicating that higher expression of TBK1 does not cause TDP-43 pathology (Fig. 5 and S5). Taken together, these findings strongly indicate that increasing TBK1 levels in the SC prevents MN demise and the development of TDP-43 pathology that is typically observed in mice expressing the mutant P497S UBQLN2 transgene.

Fig. 5.

Reduction in TDP-43 pathology and UBQLN2 inclusions in P497S mice overexpressing TBK1. (A) Immunofluorescent staining images of SC sections from male mice for the four genotypes for the indicated signals. Arrows signify MNs with clear evidence of TDP-43 pathology. (B) Quantification of MN in male SC sections with TDP-43 pathology. ns > 0.05, ****P < 0.0001. (C) Quantification of the number of UBQLN2 inclusions in a fixed area in male SC sections for the four genotypes. ns > 0.05, *P < 0.05, ***P < 0.001.

2.5. TBK1 overexpression reduces neuronal loss in P497S UBQLN2 Tg mice

To determine whether the neuroprotection conferred by TBK1 overexpression extended to the brain, we quantified the number of neurons in the CA1 and dentate gyrus regions of the hippocampus for all four genotypes (Fig. 6). We focused on these particular brain regions, as they were the ones where we found profound neuronal loss in previous studies of the P497S line (Le et al., 2016; Wang et al., 2020). The quantification, conducted by stereology counting of NeuN stained neurons, revealed significantly fewer neurons in both the CA1 and dentate gyrus regions of the hippocampus in single P497S Tg animals compared to Non-Tg animals, for both sexes (Fig. 6A–C), which is consistent with our previous findings. Interestingly, we found a slight, but non-significant reduction in the number of neurons in both the dentate gyrus and CA1 regions of the brain in the single-Tg TBK1 mice, suggesting that elevated TBK1 expression in the brain may adversely affect neuronal survival (Fig. 6C). Importantly, the neuronal counts for both the CA1 and dentate gyrus regions in the double-Tg P497S/TBK1 male and female mice were higher than those in the P497S animals, and statistical comparison showed they were not significantly different to the Non-Tg animals. These findings, when combined with those seen in the SC, support the idea that TBK1 overexpression protects neurons from demise caused by expression of the mutant P497S UBQLN2 protein across multiple regions of the CNS.

Fig. 6.

Rescue of neuronal loss in the hippocampus in P497S mice overexpressing TBK1. (A and B) NeuN staining of male (A) and female (B) dentate gyrus and CA1 regions of the hippocampus for the four genotypes. (C) Quantification of neurons in the dentate gyrus and CA1 regions for each genotype relative to Non-Tg mice. ns > 0.05, *P < 0.05, **P < 0.01.

2.6. Immunoblot analysis of neuropathology in double-Tg P497S/TBK1 mice

To obtain further evidence that disease was attenuated in the double-Tg P497S/TBK1 mice, we conducted immunoblots of the brain and SC lysates made from the 52-week-old animals to examine whether the double-Tg mice had a decline in particular markers of disease. Our primary focus was on proteins involved in autophagy because we found major disruption of this pathway, as witnessed by p62 accumulation, in the P497S mutant mice (Wu et al., 2020). Several immunoblots yielded weak signals and were therefore interpreted cautiously. Consequently, the conclusions presented here are based primarily on the most robust and reproducible findings.

Consistent with our prior findings that autophagy is disrupted in the P497S animals, p62 (SQSTM1) levels in the brain and SC were significantly increased in both male and female single-Tg P497S mice compared to Non-Tg animals (Fig. 3 and Fig. S3). However, the double-Tg P497S/TBK mice had only a minor reduction in p62 levels compared to the single P497S transgenic mice, which was non-significant, suggesting little rescue of the autophagy defect by TBK1 overexpression (Fig. 3 and Fig. S3). The levels of p62 in the TBK1 mice were not significantly different to those seen in Non-Tg mice. Examination of LC3 levels, another marker of autophagy, was uninformative as its levels were relatively unchanged across all four genotypes (Fig. 3 and Fig. S3). We also probed the lysates for ubiquitin reactivity because of the pronounced increase in ubiquitinated aggregates found in the P497S animals (Higgins et al., 2021; Le et al., 2016). As expected, single-Tg P497S male mice had a significant increase in anti-ubiquitin immunoreactivity in the brain, consistent with a buildup of misfolded proteins caused by mutant P497S UBQLN2 protein expression (Fig. 3A and B). This reactivity was significantly reduced in the double-Tg male group suggesting that TBK1 overexpression reduces accumulation of ubiquitinated proteins in P497S positive mice.

Previous studies had indicated that UBQLN2 regulates TBK1 levels through IRF3, but not IRF7 signaling (Chen et al., 2020). Accordingly we probed the spinal cord lysates with antibodies to detect IRF3 and IRF7 proteins. The blots revealed little difference in IRF3 levels in the SC in the four genotypes. IRF7 levels were variable across genotypes and sexes and did not show a consistent rescue pattern in TBK1/P497S mice. Specifically, in male spinal cord lysates, IRF7 appeared elevated in all three Tg groups compared with Non-Tg controls, whereas this pattern was not observed in females. The basis for this male-specific increase remains unclear.

2.7. Double-Tg P497S/TBK1 male, but not female mice, have fewer UBQLN2 aggregates than single-Tg P497S mice

A distinguishing pathologic feature found in the mutant P497S UBQLN2 Tg animals is deposition of UBQLN2 inclusions in a distinctive pattern in the brain, with notably high concentration around the dentate gyrus and CA1–3 regions in the hippocampus and throughout layers V to VI of the cortex (Higgins et al., 2021; Le et al., 2016; Wu et al., 2020). The majority of these UBQLN2 inclusions stained positive for p62, which is consistent with a defect in autophagy in the mice (Wu et al., 2020). Accordingly, we double-stained brain sections cut from all four genotypes for UBQLN2 and p62 to determine whether TBK1 overexpression had altered either the number of inclusions, or their colocalization. As expected, a major increase in UBQLN2 inclusions was found in the single-Tg P497S mice, in all three brain regions (DG, CA1, and Cortex) that were analyzed (Fig. 7 and Fig. S7). With rare exceptions, most of these inclusions were also positive for p62. Interestingly, the number of UBQLN2 inclusions in the double transgenic male mice were significantly reduced, but not eliminated, compared to single-Tg P497S animals, indicating TBK1 overexpression reduced the burden of UBQLN2 inclusions in the animals. By contrast, no reduction in aggregates was seen in the double-Tg female mice. Also, no obvious change in p62 colocalization was detected.

Fig. 7.

Reduction of UBQLN2 inclusions in the dentate gyrus, CA1, and cortex of P497S mice overexpressing TBK1. (A) Immunofluorescent staining of the dentate gyrus, CA1, and cortex regions in male mice for the different markers or proteins for the four genotypes. (B, C, and D) Quantifications of UBQLN2 inclusions in each region of the brain for the four genotypes (male). *P < 0.05, **P < 0.01, ***P < 0.001, ****P < 0.0001.

2.8. TBK1 protein is sequestered by UBQLN2 aggregates in P497S mice

We next investigated the potential reason for the decline in TBK1 levels in the P497S mutant mice. We considered two likely possibilities. One, UBQLN2 aggregates in the mutant lines may bind and sequester TBK1 thereby depleting its free pool from the cytosol. This idea would be compatible with the colocalization of the proteins that we observed in spinal MNs in the end-stage mutant mice (Fig. 1E). Two, UBQLN2 may function to stabilize TBK1, and the depletion of its free pool in the mutant mice may compromise this function. We entertained this possibility because we found evidence that the loss of UBQLN2 expression is a probable cause of the autophagy defect that manifests in the P497S mutant mice (Wu et al., 2020).

To examine the first possibility, we double stained cryosections cut from the brain and SC of 52-week-old P497S and control mice to see if TBK1 and UBQLN2 are colocalized in the mice used in the present study. Examination of the brain sections by confocal microscopy revealed weak signs of the colocalization of TBK1 with the UBQLN2 inclusions (Fig. 8A). The TBK1 staining in the tissue was diffuse, making it difficult to determine if TBK1 was localized in the UBQLN2 inclusions. Densitometry analysis of line scans drawn through the UBQLN2 puncta in the dentate gyrus was unconvincing (Fig. S8A). However, in the double transgenic TBK/P497S mice, a stronger indication that TBK1 had colocalized with the UBQLN2 aggregates was evident, both visually and by coalignment of their fluorescent images by line scan tracing (Fig. 8A and Fig. S8A). We speculate that the diffuse TBK1 signal found in the UBQLN2 inclusions in the single-Tg P497S mice is because of low TBK1 expression, which is distributed over the many UBQLN2 inclusions, and that their colocalization becomes more apparent with higher expression of TBK1 in the double-Tg animals.

Fig. 8.

UBQLN2 sequesters TBK1 in P497S mutant mice. (A, B) Immunofluorescent images of the hippocampus (A) and SC (B) for UBQLN2 and TBK1 for the four genotypes. (C) Immunoblots of pellets remaining from sequential extraction of cortical tissues from 32-week-old Non-Tg and P497S mice after re-solubilization in SDS sample buffer and analyzed for the proteins shown. (D) Quantification of the TBK1 signals in C showing the relative amount of TBK1 detected in the brain pellets. ns > 0.05.

More concrete evidence that TBK1 colocalizes with UBQLN2 inclusions was evident upon examination of the SC stained sections. Like in the brain, the staining of TBK1 and UBQLN2 in the SC of the Non Tg and TBK1 mice was weak and diffuse (Fig. 8B). Line scan examination for coalignment of their fluorescent signals suggested little overlap of their staining (Fig. S8A). By contrast, and similar to our previous findings, UBQLN2 inclusions in the single-Tg P497S mice were positive for TBK1, but only those that were present within MNs (Fig. 8B and Fig. S8A). No clear overlap in the staining of the two proteins was observed in MNs of the double transgenic TBK1/P497S mice, which like the single TBK1 mice had stronger UBQLN2 staining for reasons that are unclear (Fig. 8B and Fig. S8A). Taken together, the immunofluorescent staining, especially of the SC, suggested, but did not prove that TBK1 is sequestered by the UBQLN2 inclusions.

To further test whether TBK1 associated with UBQLN2 aggregates, we conducted immunoblots of the pellet fractions recovered, following progressive extraction of the same weight of brain tissue, isolated from the cortex of control and P497S mice using different detergents and 8 M Urea. Filter-trap analysis of these lysates revealed greater UBQLN2 reactivity in the P497S preparations (Higgins et al., 2021), consistent with the high content of UBQLN2 aggregates in the P497S animals. We further showed that the UBQLN2 aggregates in the P497S mice are highly resistant to detergent extraction by Triton X-100 and Sarkosyl, but can be partially solubilized with 8 M urea (Higgins et al., 2021). We took the pellet fractions that remained after urea extraction and analyzed them for TBK1 presence by immunoblotting. The pellet fractions recovered from three independent 32-week-old P497S mice all contained strong reaction with a UBQLN2 doublet whose sizes matched the endogenous mouse and transgenic human P497S UBQLN2 proteins, compared to their absence in the pellets made from the age-matched Non-Tg mice (Fig. 8C). Further blots showed that the P497S mouse preparations were highly enriched for p62 and ubiquitinated proteins (Fig. 8C). The strong detection of all of these proteins in the P497S preparations, and their virtual absence in preparations from the Non-Tg mice, together with entrapment of UBQLN2 reactivity in prior filter-trap assays, is strongly indicative of protein aggregates. Analysis of these same lysates for TBK1 presence revealed that TBK1 was present in both the control and P497S preparations, with higher amounts detected in the P497S animals (Fig. 8D). However, statistical analysis of the enrichment in the P497S mice showed that the increase was not significant (Fig. 8D). This is likely due to higher TBK1 reactivity seen in one of the Non-Tg controls, which also had higher ubiquitin reactivity. The presence of TBK1 in all the controls, suggests that the protein is highly prone to aggregate, independent of mutant UBQLN2 expression.

We repeated the biochemical fractionation studies to determine if insolubility of TBK1 and UBQLN2 proteins is also observed in brain tissue of 24-week-old mice. The blots for the partitioning of the two proteins in the different extraction buffers are shown in Fig. S8D. Like the samples from the 32-week-old animals, considerably higher amounts of UBQLN2 were detected in the pellet fractions for all four P497S samples, compared to negligible amounts in the four Non-Tg controls. Parallel blots of the same fractions for ubiquitin reactivity revealed high enrichment of ubiquitinated proteins in the P497S samples, consistent with aggregation of detergent-insoluble misfolded proteins in the brain tissue of the 24-week-old P497S animals. The TBK1 blots for the same pellet fractions showed higher amounts of TBK1 was present in the P497S samples compared to the Non-Tg samples. However, statistical analysis revealed the enrichment was not significant. Nevertheless, the results from these fractionation studies support the idea that TBK1 has increased insolubility and that it gets sequestered by UBQLN2 aggregates, which could result in its depletion in the P497S mice.

2.9. Analysis of TBK1 interaction with UBQLN2 by immunoprecipitation and pulldown assays

To further test the sequestration idea, we conducted immunoprecipitation and pulldown assays to determine whether TBK1 binds with UBQLN2 and whether ALS mutations in UBQLN2 alter this binding.

We first assessed if endogenous TBK1 and UBQLN2 proteins in HeLa cells coprecipitate with one another, following precipitation of each protein (Fig. S9 A). Immunoblots confirmed the successful immunoprecipitation of each protein (Fig. S9 A). However, no UBQLN2 was detected in the TBK1 precipitates. By contrast, a minor amount of TBK1 was detected in the UBQLN2 precipitates. However, the same amount was also detected in the Control IgG immunoprecipitates, leading to uncertainty regarding the coimmunoprecipitation of the two proteins. Repetition of the immunoprecipitation led to similar findings.

To further investigate their interaction, we conducted additional coimmunoprecipitation studies, this time analyzing UBQLN2 immunoprecipitates made from HeLa cells lacking UBQLN2 (the HeLa KO8 line in which CRISPR/Cas9 technology was used to specifically inactivate UBQLN2 expression (Wu et al., 2020)) that were either mock-transfected, or transfected with expression constructs encoding full-length untagged WT UBQLN2 protein, or the identical protein carrying one of 5 different ALS missense mutations, which we have routinely studied ((Chang and Monteiro, 2015; Deng et al., 2011; Gilpin et al., 2015; Lin et al., 2021; Phung et al., 2023; Wu et al., 2020)). A probe of the immunoprecipitates confirmed successful precipitation of UBQLN2 from all the cultures, except from the mock transfected cells, which was expected as they are devoid of UBQLN2 expression (Fig. S9B). Parallel blots of the immunoprecipitates revealed little, if any, TBK1 coimmunoprecipitation with WT UBQLN2. Similar weak TBK1 coimmunoprecipitation was seen for all of the 5 ALS mutants tested (Fig. S9B). Reexamination of the precipitates with different TBK1 antibodies yielded the same result, suggesting weak coimmunoprecipitation of TBK1 with UBQLN2, if any. The lack of strong evidence of the coimmunoprecipitation of the two proteins may be due to a lack of good antibodies for assessing the interaction.

We therefore conducted in-vitro GST-pulldown assays to assess whether TBK1 binds directly with UBQLN2 and, if so, whether the 5 different ALS mutations affect the binding. For these assays, purified GST alone or GST-fusion proteins, in which GST was fused with WT UBQLN2, or with UBQLN2 carrying each of the 5 different ALS mutations, were used as a bait and recombinant His-tagged FL TBK1 as a ligand (Fig. 9A). The pulldown results revealed GST-WT-UBQLN2-fusion protein bound 50-fold more TBK1-His than GST alone, demonstrating the specificity of the TBK1-UBQLN2 interaction (Fig. 9B, C). Interestingly, four of the five ALS mutant UBQLN2 fusion proteins bound less TBK1 than the WT-UBQLN2 bait. However, statistical analysis of the bindings indicated the reduction was only significant for the P497H and P497S mutants (Fig. 9C). It is important to note, that the reduction in binding by the latter mutant does not eliminate the possibility that the residual binding is still sufficient to facilitate its sequestration by the UBQLN2 aggregates. The pulldown data demonstrate that UBQLN2 can bind TBK1 directly, but they do not support a simple model in which the P497S mutation increases TBK1 binding with UBQLN2. Interestingly, the P509S mutant bound similar amounts of TBK1 as WT UBQLN2, suggesting that ALS-associated UBQLN2 mutations differ in the effects on TBK1 binding.

Fig. 9.

GST-pulldown assays demonstrating TBK1 binds directly to UBQLN2 and the changes in binding with ALS mutant UBQLN2 proteins. (A) Purified GST-tagged UBQLN2 mutant proteins, WT GST protein, and His-tagged TBK1 full length protein, stained with Coomassie blue that were used for the pulldown assays. (B) Immunoblots for the pulldown assay blotted for the proteins shown. (C) Quantification of TBK1 binding with UBQLN2 in the pulldown assays. ns > 0.05, *P < 0.05, **P < 0.01.

2.10. TBK1 turnover is regulated by UBQLN2

To examine the second possibility, that UBQLN2 may regulate TBK1 stability, we conducted cycloheximide-chase studies using WT HeLa and KO8 cells to determine if TBK1 protein stability is affected by the loss of UBQLN2 expression. Immunoblotting of the lysates confirmed UBQLN2 protein expression in the WT HeLa line and undetectable expression in the UBQLN2 KO8 line, as expected (Fig. 10A). Quantification of the TBK1 protein change in the two lines revealed a significant acceleration in TBK1 turnover in the KO8 cells, strongly suggesting that UBQLN2 stabilizes TBK1 (Fig. 10C).

Fig. 10.

Protein turnover of TBK1 analyzed by a cycloheximide (CHX) chase experiment. (A) Immunoblots of the turnover of TBK1 in WT HeLa and UBQLN2 KO8 HeLa cells. (B) Immunoblots of the turnover of TBK1 in HeLa KO8 cells transfected with expression constructs encoding either FL WT UBQLN2 or carrying the P497S ALS mutation. (C) Quantification of turnover of TBK1 in the WT or UBQLN2 KO HeLa cells without or upon transfection with expression constructs encoding either WT UBQLN2 or the P497S mutant. Turnover of TBK1 was significantly accelerated in the UBQLN2 KO8 cells (*P < 0.05, **P < 0.01) compared to normal HeLa cells (expressing endogenous UBQLN2), but not different upon transfection of the constructs encoding either WT or the P497S UBQLN2 mutant.

We next examined whether re-expression of WT UBQLN2 protein, or any of the 5 mutant UBQLN2 proteins, in the KO8 cells restores TBK1 stability. Indeed, re-expression of the WT UBQLN2 protein restored TBK1 stability, as its turnover in the transfected KO8 cells was similar to that of WT HeLa cells (Fig. 10B, C). Interestingly, all five ALS mutant UBQLN2 proteins stabilized TBK1 protein to a similar degree as WT UBQLN2 (Fig. S10A–E), suggesting that the ALS mutant proteins are not defective in this activity.

3. Discussion

Here we have shown that TBK1 protein levels are reduced in the brain in diseased P497S UBQLN2 Tg mice. By contrast, we found little alteration of its levels in the SC of the animals, although a significant decline was observed in an earlier stock of the line, as well as in a companion P506T UBQLN2 Tg line, both of which exhibited a more aggressive motor phenotype than the P497S line used in the present study. These observations lead us to speculate that there may be a correlation between the decline in TBK1 levels and pathogenesis, particularly motor dysfunction, induced by the UBQLN2 mutations. The foundation of this speculation comes from genetic findings that have linked TBK1 haplo-insufficiency to ALS/FTD pathogenesis. In accord with this idea, we found that TBK1 overexpression in the P497S line restores TBK1 levels in the brain and prevents neurodegeneration. Interestingly, the neuroprotective effects exerted by TBK1 overexpression extended to the SC in the P497S animals, evidenced by a reduction of MN loss and TDP-43 pathology. Taken together, our studies suggest that increasing TBK1 levels, or preventing its loss, may provide therapeutic benefit to delay or alleviate ALS/FTD pathogenesis caused by UBQLN2 mutations.

Insight into how UBQLN2 mutations trigger TBK1 loss would be important for devising strategies to prevent its loss for ALS/FTD therapy. Our studies identified two possible, non-mutually exclusive, mechanisms. The first, for which we found reasonably good evidence, is that the UBQLN2 aggregates that are deposited due to expression of the mutant P497S UBQLN2 protein, bind and sequester TBK1 protein, thereby depleting its free pool from the cytosol. The second mechanism, based on our findings that TBK1 turnover is accelerated in cells with inactivated UBQLN2 expression, is that TBK1 becomes destabilized in the mutant mice due to loss of the functional pool of UBQLN2. It is possible both mechanisms are involved.

Several lines of evidence suggest that UBQLN2 aggregates are responsible for binding and reducing TBK1 protein levels in the P497S mice. First, we found that the UBQLN2 inclusions, particularly those within MNs of the SC in both P497S and P506T UBQLN2 mice, are strongly decorated and exhibit high degree of colocalization with TBK1 staining, which is consistent with the sequestration of TBK1 by UBQLN2 aggregates. Puzzlingly, there was a lack of similar strong TBK1 colocalization with the UBQLN2 aggregates in the brain in the single-Tg P497S mice, despite having the greater reduction in TBK1 levels. However, we cannot rule out the possibility that TBK1 was present in the brain aggregates, but that its enrichment was simply not detected because its signal was diluted by the vast number of UBQLN2 aggregates present in the brain. Indeed, convincing evidence that TBK1 is present in the brain aggregates was seen in the double-Tg animals where many more UBQLN2 aggregates showed clear indication of TBK1 enrichment, presumably because of TBK1 overexpression. Second, and more compellingly, we found approximately 1.7-fold higher amounts of TBK1 protein co-sediments along with WT and mutant UBQLN2 proteins after detergent and urea extraction of cortical brain tissue obtained from 32-week-old P497S mice. Further analysis of the sediment revealed that it contains p62 protein and has high anti-ubiquitin content, which is consistent with properties expected of protein aggregates. Interestingly, lower but detectable amounts of TBK1 were also recovered in the sediment prepared from the Non-Tg animals, suggesting that TBK1 may be highly prone to aggregation. A similar trend of TBK1 insolubility was seen in 24-week-old mice suggesting a rise in its insolubility may begin at early age. Third, we found through GST-pulldown assays that TBK1 binds directly with WT UBQLN2. Interestingly, by these same assays we found that the P497S mutant UBQLN2 protein has 3-fold weaker binding with TBK1 than WT UBQLN2. Although the P497S mutant reduced TBK1 binding in vitro, this does not exclude the possibility that the P497S mutant is nevertheless responsible for recruiting TBK1 to UBQLN2 aggregates, as UBQLN2 inclusions contain both endogenous and transgenic UBQLN2 proteins (Fig. 8C), either of which could recruit TBK1 through multivalent or indirect interactions.

A second potential mechanism that could have triggered the decrease in TBK1 levels in the P497S mice is increased turnover of the protein due to loss of UBQLN2 function, based on the acceleration of its turnover that we observed in UBQLN2 KO cells. Such a mechanism may arise because the human P497S UBQLN2 mutant protein coaggregates with the endogenous mouse WT UBQLN2 protein, which, over time, could deplete all of the soluble WT UBQLN2 protein in the cell. We previously proposed that loss of UBQLN2 function by this mechanism is the likely cause of a defect in lysosomal acidification that we found in the P497S mutant mice (Wu et al., 2020). Importantly, we surmised that such depletion may not be sufficient to trigger ALS, because complete inactivation of UBQLN2 in rodents does not cause neuronal loss or motor neuron disease (Whiteley et al., 2021). Accordingly, we proposed that UBQLN2 mutations most likely cause ALS through both loss- and gain-of-function mechanisms (Wu et al., 2020).

If the two mechanisms we propose here were to operate simultaneously to reduce TBK1 levels, it would represent a pathogenic change that arises from both loss and gain of UBQLN2 function. Interestingly, in both of these scenarios, the UBQLN2 aggregates serve as the primary source for the loss or gain of UBQLN2 function, showing how they could serve a dual role in pathogenesis. Although both mechanisms are possible, we consider that the acceleration in TBK1 turnover is less likely to occur simply because UBQLN2 mutations are known to impede protein degradation (Chang and Monteiro, 2015; Deng et al., 2011; Liu et al., 2026). Another possibility is the decline stems from a reduction in TBK1 transcription and/or translation. Although we cannot rule out these latter possibilities, we consider them less likely because a similar decline was not seen in the BacTBK1/P497S double-Tg mice, which carries an extra copy of the entire TBK1 gene locus. The transgene should have been subjected to this same dysregulation.

Regardless of the exact mechanism by which TBK1 levels decline, our results provide compelling evidence that overexpression of TBK1 is sufficient to alleviate major signatures of disease in the P497S mouse model of ALS/FTD. It also suggests that strategies to increase TBK1 expression could provide a benefit in preventing or delaying disease caused by UBQLN2 mutations. However, we caution that any such attempt should be carefully controlled, because TBK1 duplication has been linked to Normal-Tension Glaucoma (Ritch et al., 2014). It might have been fortuitous that the BacTBK1 transgenic mouse used in our studies did not drive excessive TBK1 overexpression, otherwise it could have been detrimental. In our studies, the level of TBK1 overexpression appeared to be just sufficient to restore TBK1 levels in the brain to the same physiological level found in the Non-Tg animals. Importantly, this was sufficient to prevent neuronal loss through the study endpoint examined here.

We do not know the mechanism by which TBK1 overexpression prevents neuronal loss in the P497S animals. Because neuroprotection was observed despite incomplete and sex-dependent reduction of UBQLN2 inclusions, the protective effect of TBK1 is unlikely to be explained solely by reduced aggregate burden. Instead, we surmise it prevents neurodegeneration through TBK1’s normal function, such as its roles in regulation of autophagy and/or inflammation, both of which we found are highly disturbed in the P497S mice. Immunoblots were conducted to interrogate if the changes are mediated by IRF3 and IRF7 signaling, the two main downstream effectors that function to regulate interferon production and inflammation. These studies revealed little change in IRF3 in the four genotypes. Unfortunately, we did not have enough power to determine if the observed changes are significant. Some variation in IRF7 protein levels was observed across the four genotypes. However, the changes were inconsistent between sexes and did not demonstrate a clear rescue pattern in the TBK1/P497S mice. Consequently, the present data are insufficient to determine whether IRF7 signaling contributes to the protective effects associated with TBK1 overexpression. Quantification of p62 aggregates, which we used as a surrogate for measurement of impedance in autophagy, was little different in single and double P497S mice, suggesting TBK1 overexpression did not rescue the autophagy deficit. The available biochemical markers did not identify a single downstream pathway that accounts for TBK1-mediated neuroprotection. In particular, total p62 and LC3 levels did not indicate robust restoration of autophagy, and total IRF3/IRF7 levels are insufficient to infer pathway activity. Further studies should examine TBK1 kinase activity and phosphorylation of downstream substrates such as OPTN, p62, and IRF3.

Interestingly, our pathologic examinations of the P497S mouse SCs revealed extensive MN loss as well as visual signs of TDP-43 pathology (nuclear clearance and perinuclear aggregation of TDP-43) in the absence of a significant change in TBK1 levels. This observation should not be construed as lack of evidence for involvement of TBK1 in the pathologies. A strong indication that refutes this notion is that TBK1 overexpression by itself was sufficient to rescue the pathologies in the double-Tg mice. One reason for not detecting a change in TBK1 levels is that immunoblotting of whole SC lysates is not ideal for capturing changes that might occur in individual cells. Another possibility is that TBK1 activity, or its localization might have been altered, which will require more careful interrogation. Moreover, our findings that TBK1 levels decline significantly in the SC of P497S and P506T mice with more aggressive MN disease suggests that immunoblotting may only be suitable for detecting a change in its expression in animals with more advanced stages of disease.

The mechanism underlying the destabilization of TBK1 in cells lacking UBQLN2 expression is not known. Our most parsimonious explanation is that UBQLN2 binding with TBK1 is required to stabilize the protein. Indeed, through GST-pulldown assays we found WT UBQLN2 protein and all five missense UBQLN2 proteins analyzed bind directly with TBK1, although binding with the WT and the P509S mutant proteins were the only two proteins that exhibited significantly stronger binding compared with GST alone. The stabilization of TBK1 by UBQLN2 binding maybe related to UBQLN’s role as a chaperone (Lin et al., 2022). One place where this interaction may predominate is during synthesis, where UBQLN proteins are thought to function. Several roles have been assigned to this binding function:— (a) as a chaperone in facilitating protein translation, (b) in shielding hydrophobic domains before protein assembly, and (c) as a guardian in protein quality control. This could also explain why we observed only minor TBK1 coimmunoprecipitation with UBQLN2, as only a small fraction of TBK1 is synthesized at any given time. Alternatively, binding to the proteins may be triggered in response to some cellular signal. Future studies should focus on what role this interaction plays in cells.

It is important to note that our TBK1-UBQLN2 binding results differ substantially from a previous study which showed abundant coimmunoprecipitation of UBQLN2 with TBK1 (Chen et al., 2020). Those authors also reported increased binding of ALS mutant UBQLN2 proteins with TBK1, but statistical demonstration for the increase was not provided. We are unsure if the discrepancy in the strength of interaction observed between TBK1 with WT or ALS mutant UBQLN2 proteins in our two studies originates from differences in the protocol or cell lines used for the analysis. It is important to note that we analyzed the interaction between endogenous TBK1 and overexpressed UBQLN2 in UBQLN2 null cells, whereas they analyzed the interaction by overexpressing both proteins in cells that already expressed endogenous TBK1 and UBQLN2 proteins. If binding of UBQLN2 occurs during TBK1 synthesis, it would explain the discrepancy.

Several limitations should be considered. First, the current P497S cohort displayed an attenuated behavioral phenotype, preventing assessment of behavioral rescue. Second, total TBK1 levels in whole spinal cord lysates may not reflect cell-type-specific changes in TBK1 kinase levels or activity. Third, although TBK1 was detected in UBQLN2-positive inclusions and aggregate-enriched fractions, these data do not by themselves prove sequestration or loss of TBK1 function. Finally, the TBK1 turnover experiments were performed in HeLa cells and will require validation in more disease-relevant cell types.

In summary, we have shown here that TBK1 levels are reduced in vulnerable CNS tissues in UBQLN2 mouse models of ALS/FTD, and that genetic restoration of TBK1 levels attenuates neuronal loss and TDP-43 pathology in P497S UBQLN2 mice. Our studies support further investigation of TBK1 augmentation as a therapeutic strategy for preventing disease caused by UBQLN2 mutations.

4. Materials and methods

4.1. Animal research

All animal procedures were conducted according to the University of Maryland Baltimore Institutional Animal Care and Use Committee approved protocol, in full compliance of the recommendations described in the NIH Guide for the Care and Use of Laboratory Animals.

4.2. Mouse husbandry, breeding, and genotyping

The Tg P497S UBQLN2 mouse line used in this study was described previously (Le et al., 2016). The P497S mouse line was backcrossed with C57BL/6 J mice (The Jackson Laboratory, Strain 000664) for at least 10 generations prior to being used for the current experiments. The strong clasping phenotype that was evident in earlier generations of the line became less apparent, and for this reason no behavioral analysis was conducted. The construction and characterization of the Tg human Bac TBK1 mouse line was described previously (Fingert et al., 2017). P497S and TBK1 transgenic mice were crossed with one another and transmission of the Tg P497S and TBK1 human transgenes were identified by polymerase chain reaction analysis using mouse tail DNA using primer sets (5’ TCTGAGTGGCAAAGGACCTTA and CGACGCCGAGGTAGTGTTAGTTCCC for UBQLN2) and (5’ GGGGCACATTTCATTACCTC and AAACAGCATGAGGGTGAAGG for TBK1). All four genotypes (Non-Tg, single Tg TBK1, single-Tg P497S, and double-Tg TBK1::P497S) were obtained with the expected Mendelian frequency. The mice were segregated by sex, but not genotype, and housed at no more than 5 animals per cage. The mice were fed regular mouse chow and given clean water to drink ad lib. Animals that had difficulty in reaching the food were provided with Diet-Gel 76 A (Clear H₂O, Portland, ME) in the bedding.

All tests were conducted by investigators who were blinded to the animal genotypes. Body weight was measured bi-weekly from 6 to 52 weeks of age to the nearest 0.1 g. The mean body weight (SEM) for 6 animals for each sex and genotype is shown. The total number of animals used in the study was 75.

4.3. GST-pulldown assays

GST-tagged proteins (10 μg GST, GST-WT, or ALS mutant UBQLN2) were rotated with TBK1 protein (1 μg) in 800 μl 0.5% Triton X100 binding buffer (20 mM Tris pH 8.0, 150 mM NaCl, 10% glycerol, and 0.5% Triton X100) at 4 °C for 2 h. The samples were centrifuged at 8000 rpm for 5 min and the supernatants were added to 150 μl glutathione beads (Cytivia, Marlbrough, MA) that had been preblocked with 5% milk. The protein/bead mixtures were rotated at 4 °C for 2 h. The protein/bead mixtures were then washed two times with 0.5% Triton X100 binding buffer and once with binding buffer lacking the detergent. The beads were resuspended in gel loading buffer with SDS and heated at 95 °C for 10 min. All pulldown samples were resolved by SDS-polyacrylamide gel electrophoresis (PAGE) and analyzed by immunoblotting. The subsequent primary antibodies were used to analyze GST-pulldown assays by immunoblot analysis. The GST tag alone and GST-tagged proteins were detected with a mouse anti-GST monoclonal antibody (#MA4–004, Invitrogen). TBK1 protein was detected with a TBK1/NAK rabbit monoclonal antibody (#3504, Cell Signaling Technology).

4.4. Tissue isolation and cryosectioning

All the mice used for the study were euthanized at 52 weeks of age by overdose with isoflurane. Their brains and SCs were then dissected and immediately frozen on dry ice and then transferred to −80 °C for storage. For cryosectioning, the tissues were quickly thawed, and then quickly immersed in ice cold 4% paraformaldehyde (PFA) in 1 × PBS for at least 1 day and then transferred to a 25% sucrose solution in 1 × PBS. The tissues were then embedded in O.T.C. compound (Sakura Finetek USA, Inc., Torrance CA), which was then cut using a Leica CM 3050S cryostat (Leica Biosystems, Buffalo Grove, IL). Sagittal sections were cut for the brain tissue and transverse sections for the SC tissue, which were then transferred to slides and stored at −20 °C, prior to staining.

4.5. Staining of mouse neuronal tissue

MNs in the lumbar region of the spinal cord were quantified after staining of the appropriate sections with Cresyl violet using stereological principles, as described previously (Le et al., 2016). The number of NeuN-positive neurons in the dentate gyrus and CA1 regions of the hippocampus was by quantified using stereological principles, as described before (Le et al., 2016). At least 10 brain sections (12 μm thick) were counted.

The procedure for immunofluorescent staining of brain and SC sections was described previously (Le et al., 2016). The sections were stained with primary antibodies against the following proteins: mouse anti-UBQLN2 (#NBP2–25164, Novus Biologicals, Littleton, CO), rabbit anti-TDP-43 (UMY112, rabbit antibody against the C-terminal region of TDP-43 (NQAFGSGNNSYSGS), produced in house), goat anti-ChAT (#AB144P Millipore-Sigma, Burlington, MA), mouse (ab56416, Abcam, Eugene, OR) or rabbit anti-SQSTM1/p62 (CST-5411S, Cell Signaling Technology, Danvers, MA), rabbit anti-NeuN (ab177487, Abcam), mouse anti-ubiquitin (sc-8017, Santa Cruz Biotechnology, Santa Cruz, CA), mouse anti-IRF3 (#66670, ProteinTech), rabbit anti-IRF7 (#4920, Cell Signaling), rabbit anti-LC3A/B (4108 Cell Signaling Technology), mouse anti-GAPDH GTX627408 GeneTex, Irvine, CA), rabbit anti-p97/VCP (UMY475 home-made). Binding of primary antibodies was detected with the appropriate Alexa Fluor secondary antibodies (Invitrogen, Carlsbad, CA). Images were captured with a Hamamatsu digital C8484 camera attached to an inverted Zeiss fluorescent microscope with either 10×, 20×, or 40× dry objectives using iVision software (BioVision Technologies, Exton, PA), or using a Nikon Eclipse Ti A1R confocal microscopy with a 40× oil objective lens and 405, 488, 574 and 647 nm laser lines. Multi-wavelength fluorescent images of sections were merged using iVision software.

4.6. Quantification of UBQLN2 inclusions

Brain and SC sections were stained with primary mouse anti-UBQLN2 antibody (#NBP2–25164, Novus Biologicals) followed by goat anti-mouse Alexa Fluor 488 nm secondary antibody. Digital images of the stained sections were captured with a 40× objective using the inverted Zeiss microscope. The number of UBQLN2-positive inclusions in identical areas of the CA1 and dentate gyrus regions of the hippocampus and of the anterior horn in the SC in different animals were estimated using iVision software to identify, quantify, and segment the inclusions above a predefined size threshold. The data shown is the mean number of inclusions seen in three different animals for each genotype.

4.7. SDS-PAGE and immunoblotting

Protein lysates were made from the lumbar region of the SC or from one hemisphere of the brain by homogenizing the tissues in protein lysis buffer (50 mM Tris pH 6.8, 150 mM NaCl, 20 mM EDTA, 1 mM EGTA, 0.5% SDS, 0.5% NP40, 0.5% Sarkosyl, 10 mM sodium-orthovanodate, 2.5 mM sodium-fluoride and protease inhibitors). Protein concentrations of the lysates were measured using bicinchoninic acid assay (BCA) (ThermoFisher Scientific). Equal amounts of the proteins were mixed with SDS sample buffer and heated for 5 min at 95 °C prior to separation by SDS-PAGE. The isolated proteins were transferred onto 0.45 μm PVDF membranes (Immobilon-P, Millipore, Billeria, MA) for 24 min at 2.5 V or 13 min at 5 V using the Invitrogen Power Blotter XL (ThermoFisher Scientific). The membranes were probed for different proteins with appropriate antibodies using the chemiluminescence method (Fuller et al., 2024). The chemiluminescence signals from the blots were captured with a Fluoro-Chem M imager (Protein Simple, San Jose, CA) or Amersham Imager 600 RGB instrument (GE Health care, Chicago, IL) and the band intensities and the intensity of different bands was quantified using either Image J or Alpha View software (Protein Simple).

4.8. Statistical analysis

Student t-test and ANOVA tests were performed using MS Excel and GraphPad Prism 8.0 software (GraphPad Software Inc., San Diego, CA) for all statistical analyses. Error bars represent ± the standard error of the mean for body weights and standard deviation for all the other analyses. Significant differences are denoted by an * for (P ≤ 0.05), by ** for (P ≤ 0.01) and by *** for (P ≤ 0.005).

Appendix A. Supplementary data

Supplementary data to this article can be found online at https://doi.org/10.1016/j.nbd.2026.107611.

Data availability

All the data used for the study is contained in the article or files in the supplemental information. All other datasets used in the study are available from the corresponding author.