A Longitudinal Study of Sex Differences in a TDP‐43 Mouse Model Reveals STI1 Regulation of TDP‐43 Proteinopathy and Motor Deficits

ABSTRACT

Amyotrophic lateral sclerosis (ALS) is a disease influenced by a complex interplay of age, genetics, and sex. Most ALS cases are sporadic, and individuals with this disease show elevated levels of TDP‐43 in their central nervous system and aggregated cytoplasmic inclusions containing TDP‐43 in neurons. Misfolded and aggregated proteins like TDP‐43 can be refolded or marked for degradation by molecular chaperones and their co‐chaperone partners. In this study, we use a mouse model of ALS that mildly overexpresses human wild‐type TDP‐43 in neurons to explore how aging affects the onset of motor abnormalities and proteinopathy in male and female mice. We found that the age‐dependent onset of motor symptoms is more pronounced in male mice, despite both sexes sharing similar TDP‐43 pathology. Further, we found that reducing the activity of STI1, an Hsp90 co‐chaperone, was associated with reduced mislocalized TDP‐43 in the brain and spinal cord and partially rescued some motor deficits. By contrast, overexpressing STI1 seemed to be deleterious, exacerbating the levels of C‐terminal TDP‐43 fragments in the cytoplasm, worsening motor abnormalities and reducing lifespan. Our findings reveal that sex is a key biological factor in an ALS mouse model of TDP‐43 overexpression and provide novel insights on the role of STI1 and proteostasis in mediating TDP‐43 pathology.

Amyotrophic lateral sclerosis (ALS) is a fatal neurodegenerative disease characterized by abnormal accumulation of the RNA‐binding protein TDP‐43. Using a mouse model that mildly overexpresses human wild‐type TDP‐43 in neurons, we show that male mice exhibit earlier onset of motor symptoms despite comparable TDP‐43 pathology between sexes. Reducing the levels of STI1, a stress‐inducible co‐chaperone of Hsp90, decreased mislocalized TDP‐43 and improved motor outcomes, while STI1 overexpression worsened pathology and reduced survival. These findings reveal sex‐specific disease dynamics and suggest that modulating the proteostasis network may offer new therapeutic avenues in TDP‐43‐related neurodegeneration.

Abbreviations

AD

alzheimer's disease

ALS

amyotrophic lateral sclerosis

a‐Syn

alpha synuclein protein

Aβ

amyloid beta protein

BOS

base of support

DnaJ

DNA J protein

FTD

Frontotemporal dementia

GAPDH

glyceraldehyde 3‐phosphate dehydrogenase

HET

heterozygous

Hsp40

heat shock protein 40

Hsp70

heat shock protein 70

Hsp90

heat shock protein 90

Hsp90β

heat shock protein 90 β isoform

MW

molecular weight

NeuN

neuronal nuclei

PBS

phosphate‐buffered saline

pTDP‐43

phosphorylated TAR DNA‐binding protein 43

RRID

Research Resource Identifier

siRNA

small interfering RNA

STI1/STIP1/HOP

stress inducible phosphoprotein 1; also known as Hsp70‐Hsp90 organizing protein

TDP‐43

TAR DNA‐binding protein 43

TGA

transgenic

TPR

tetratricopeptide repeat

WT

wild‐type

1. Introduction

Amyotrophic lateral sclerosis (ALS) is a devastating neurodegenerative disease with an average life expectancy of 2–5 years after diagnosis (Masrori and Van Damme ²⁰²⁰). Given that the highest incidence of ALS occurs between ages 60–79, age is a major risk factor (Feldman et al. 2022). Up to 97% of ALS patients present with TAR‐DNA binding protein (TDP‐43) proteinopathy in neurons of the brain and spinal cord (Brown and Al‐Chalabi ²⁰¹⁷). ALS‐associated TDP‐43 proteinopathy is characterized by mislocalization of the protein from the nucleus to the cytoplasm and post‐translational modifications such as phosphorylation, ubiquitination, and truncation of the prion‐like C‐terminal domain, forming 35 and 25 kDa toxic fragments (Prasad et al. 2019). Mislocalized TDP‐43 often aggregates into insoluble inclusions with prion‐like properties (Nonaka et al. 2013). In physiological conditions, TDP‐43 regulates mRNA processing for thousands of genes by controlling splicing, stabilization, and transport of mRNA (Ratti and Buratti 2016; Mackenzie et al. ²⁰⁰⁷).

ALS can present as either familial (inherited) or sporadic (non‐inherited) forms. Approximately 10% of ALS cases are familial, caused by mutations in genes such as SOD1, FUS, C9orf72, TARDBP (which encodes TDP‐43), and other known ALS genes (Mackenzie et al. 2007, 2010). Pathogenic TARDBP missense mutations such as p.M337V, p.N345K, and p.I383V, located within the prion‐like C‐terminal domain, enhance TDP‐43 mislocalization, aggregation, and functional deficits, contributing to motor neuron toxicity in familial ALS (Rutherford et al. 2008). The remaining ~90% of cases are sporadic, with no clear family history and largely unknown etiology. In sporadic ALS, spinal cord tissues of patients exhibit elevated protein and mRNA levels of wild‐type TDP‐43 (Swarup et al. 2011), suggesting a possible gain of function mechanism, though loss of TDP‐43 function is another possible driver of pathogenesis (Vanden Broeck et al. 2014). While familial ALS cases with TARDBP mutations have clearer mechanistic links to TDP‐43 dysfunction, the upstream factors driving TDP‐43 pathogenesis in sporadic ALS remain elusive. Aberrant cellular stress, impaired proteostasis, and altered chaperone activity are emerging as potential contributors to TDP‐43 pathology in these cases (Ruegsegger and Saxena 2016; Medinas et al. ²⁰¹⁷). Despite this distinction, TDP‐43 pathology is a hallmark in the vast majority of ALS cases, regardless of familial or sporadic classification (Mackenzie et al. 2010; Swarup et al. ²⁰¹¹).

A significant biological factor in ALS is sex, with males having a higher prevalence and incidence than females. Globally, ALS incidence is 1.91 per 100,000 person‐years in males compared to 1.36 in females (a~40% higher rate), and prevalence is 5.96 versus 3.90 per 100,000 (a~53% higher rate) (Xu et al. 2020). Some models of ALS recapitulate these differences (Torres et al. 2020; Günther et al. ²⁰¹⁴; Cacabelos et al. ²⁰¹⁶); however, most ALS mouse model research has focused on SOD1‐related and familial ALS, which are rare ALS variants in human populations (Todd and Petrucelli 2022). Multiple mouse models of wild‐type TDP‐43 overexpression have been developed in recent years that demonstrate ALS‐like phenotypes in a dose‐dependent manner. Yet, how biological sex contributes to the onset of symptoms in these models has not been systematically studied (Xu et al. 2010; Ebstein et al. ²⁰¹⁹; Cannon et al. ²⁰¹²; Wils et al. ²⁰¹⁰; Shan et al. ²⁰¹⁰).

TDP‐43 accumulation, localization, and toxicity are regulated by a special class of molecules called molecular chaperones (Webster et al. 2017), which are crucial for protein synthesis, maintenance, and degradation (Hartl et al. 2011). Heat‐shock protein 90 (Hsp90) and 70 (Hsp70) are two central molecular chaperone systems that interact with and regulate TDP‐43 (Zhang et al. 2010). Molecular chaperones, including Hsp70 and Hsp90, typically function together with co‐chaperones, which regulate their recognition and recruitment of client proteins and ATPase activity. Stress‐inducible phosphoprotein 1 (STI1/STIP1/HOP), a co‐chaperone for both Hsp70 and Hsp90, interacts with TDP‐43 and can regulate its toxicity in cellular models (Lin et al. 2021). However, how this chaperone system modulates TDP‐43 toxicity in vivo remains to be established.

In the present study, we used a mouse model of ALS that mildly overexpresses human wild‐type TDP‐43 in neurons and found remarkable sex differences related to motor abnormalities, despite similar mislocalization of TDP‐43 for both sexes. Moreover, STI1 appears to bidirectionally regulate TDP‐43 toxicity, with lower activity levels being associated with partial improvements in motor function, and higher levels correlating with exacerbated ALS‐like phenotypes. Our results shed light on sex differences related to increased accumulation of TDP‐43 and suggest a role for STI1 and the associated molecular chaperone system in regulating TDP‐43 toxicity.

2. Methods

2.1. Ethics Statement

Mouse breeding and housing was done at the University of Western Ontario animal facility in accordance with the appropriate Canadian Council of Animal Care (CCAC) guidelines and regulations. All procedures were approved under Animal Use Protocols (2020‐162, 2020‐163).

2.2. Animals

TAR4 (hemizygous) mice overexpressing human wild‐type TDP‐43 under the Thy1 neuronal promoter (Wils et al. 2010) were obtained from Jackson Laboratory (B6; SJL‐Tg((Thy1‐TARDBP)4Singh/J); RRID: IMSR_JAX: 012836; JAX stock #012836) on a C57BL/SJL background. STI1 knockout mice (STI1^−/+), STI1 mice expressing a hypomorphic allele (STI1‐ΔTPR1; ΔHET), and transgenic mice overexpressing STI1 (STI1^TGA) were on a C57BL/6j background and were generated as previously described (Beraldo et al. 2013; Lackie, Razzaq, et al. ²⁰²⁰). In brief, STI1^−/+ have 50% less STI1; STI1‐ΔTPR1 (ΔHET) mice are missing the TPR1 domain for one allele, and total STI1 levels are reduced by about 40%; and STI1^TGA mice have about 3 folds more of STI1. TAR4 mice were bred with STI1^−/+, STI1‐ΔTPR1 (ΔHET), or STI1^TGA mice. From each breeding, we analyzed littermate TAR4 mice with the following genotypes: TAR4 × STI1^−/+ breeding (TAR4‐STI1^−/+ and TAR4); TAR4 x STI1‐ΔTPR1 breeding (TAR4‐ΔHET and TAR4); and TAR4 x STI1^TGA breeding (TAR4‐STI1^TGA and TAR4). To generate homozygous TAR4 mice (TAR4/4) with various STI1 expression levels, TAR4 mice were bred with TAR4‐STI1^−/+ and TAR4‐STI1^TGA mice. All control mice used were littermates. Given that ALS has a higher incidence and prevalence in human males (Xu et al. 2020), both male and female mice were used for experiments, and sex was analyzed as a biological variable.

Mice, aged approximately 21 days, were kept in standard plexiglass cages with unrestricted access to food (Harlan) and water. Housing conditions included controlled temperature (22–25 C) and humidity (40%–60%), with a 12‐h light/dark cycle running from 7:00 a.m. to 7:00 p.m. Animals were grouped in sets of 2–4 per cage.

Sample sizes were determined a priori based on standards in the field (Pillai‐Kastoori et al. ²⁰²⁰) and prior published studies (Janickova et al. 2017; Tullo et al. ²⁰²⁴). A total of 274 mice were used in this study. At the start of the experiment, the following numbers of animals were assigned to each genotype group: 38 WT, 12 WT‐STI1^TGA, 85 TAR4, 35 TAR4‐STI1^−/+, 36 TAR4‐STI1^TGA, 12 TAR4‐ΔHET, 27 TAR4/4, 18 TAR4/4‐STI1^TGA, and 11 TAR4/4‐STI1^−/+.

2.3. Blinding Procedures

Experimenters were blinded to genotype during motor behavior testing and during scoring of hind limb clasping videos. Other aspects of motor behavior analysis were performed unblinded. Western blot experiments and biochemical analyses were conducted without blinding.

2.4. Subcellular Fractionation and Western Blotting

Mice were anesthetized using ketamine (100 mg/kg) and xylazine (25 mg/kg) in sterile saline (0.9% sodium chloride) and perfused using ice‐cold PBS. One hemisphere was dissected into cortical, striatal, hippocampal, and brainstem regions and flash‐frozen using dry ice. Similarly, a laminectomy was performed to isolate the spinal cord, which was dissected into cervical, thoracic, and lumbar parts before being frozen. To isolate the cytoplasmic and nuclear fractions, NE‐PER Nuclear and Cytoplasmic Extraction Reagents (Cat#78833, ThermoFisher Scientific) were used following the manufacturer's instructions. The remaining NE‐PER‐insoluble pellet was resuspended in 4 M urea, vortexed, sonicated, and rocked for 60 min in a 4°C cold room before being centrifuged at 16,000×g for 10 min, which constituted the insoluble fraction. Phosphatase inhibitors (1 mM NaF and 0.1 mM Na3VO4) and protease inhibitor cocktail (1:100, Cat#539134‐1SET, Calbiochem) were used in all buffers to preserve protein and phosphorylation status.

The Pierce BCA Protein Assay Kit (ThermoFisher Scientific, Cat#23227) was used to quantify protein concentration. 10–12 μg of protein was resolved in 4%–12% Bis‐Tris Protein Gels (Invitrogen, Cat#NW04125BOX) or homemade 10% Bis‐Tris gels, and later transferred using the BioRad Semi‐Dry Transfer System onto PVDF membranes. Membranes were covered with 0.1% Amido Black, rinsed with deionized water, and exposed prior to blocking with 5% BSA/1× TBS‐T to serve as a protein loading control for nuclear and urea fractions. Following blocking, membranes were left in primary antibodies overnight in a 4°C room using the following dilutions: anti‐TDP‐43 C‐terminal (1:1000, Cat#12892‐1‐AP, RRID: AB_2200505, Proteintech), anti‐Phospho TDP‐43 (Ser409/410, 1:1000, Cat#22309‐1‐AP, RRID: AB_11182943, Proteintech), anti‐STI1 (1:4000, in‐house antibody generated by Bethyl Laboratories; Beraldo et al. ²⁰¹³), anti‐Hsp90 (1:1000, Cat#4874, RRID: AB_2121214, Cell Signaling Technology), anti‐Hsp70 (1:1000, Cat#ab2787, RRID: AB_303300, Abcam), anti‐GAPDH (1:1000, Cat#ab9483, RRID: AB_307273, Abcam), and anti‐Hsp90β (1:1000, Cat#5087, RRID: AB_10548761, Cell Signaling Technology). Membranes were later washed and incubated for 1 h using the following secondary antibodies: anti‐rabbit HRP (1:10000, Cat#170‐6515, RRID: AB_11125142, Bio‐Rad) and anti‐mouse HRP (1:5000, Cat#SAB3701095, RRID: N/A, Sigma‐Aldrich).

3. Motor Assessment

3.1. Inclusion and Exclusion Criteria

All mice were included unless they exhibited signs of illness (e.g., ulcerative dermatitis) that could interfere with behavioral testing. Affected mice (∼8/274) were excluded during treatment, in accordance with veterinary recommendations. Severely ill mice were euthanized (CO₂ inhalation followed by cervical dislocation) in accordance with institutional animal care guidelines and the approved animal use protocols (2020‐162, 2020‐163). No animals were replaced following exclusion.

3.2. Wire Hang

Wire hang was performed using a metal grid wire that was placed 45 cm above a large tub filled with soft bedding. Mice were placed on the wire and gently rocked to ensure a firm grip before being flipped and suspended for a maximum of 60 s. Five trials were conducted with at least 15 min of rest time in between; the average time to fall was used for data analysis (Janickova et al. 2017).

3.3. Rotarod

Rotarod was performed using AccuRotor Four Animal Rotarod (Omnitech Electronics Inc). Mice were placed on the rod which gradually increased in rotation speed from 5 to 35 rpm over 5 min as described before (Janickova et al. 2017). A 2‐day protocol was used, with the first day consisting of 10 trials and the second day of 4 trials. There was at least a 30‐min inter‐trial interval between each trial. The average latency to fall time (s) was used from the second day for analysis.

3.4. Open Field

Open field test was conducted using automated locomotor boxes (Omnitech Electronics Inc., Columbus, USA). Mice were first habituated to the testing room for at least 30 min prior to the test. After habituation, the mice were placed in the center of the open field box and left for 30 or 60 min. Sum distance traveled (cm) was automatically collected and split into 5‐min intervals as previously described (Martyn et al. 2012).

3.5. Righting Reflex

Righting reflex was only assessed in TAR4/4 (homozygous) mice, for hemizygous mutants did not develop a severe enough phenotype for this deficit. In brief, mice were flipped to their side and timed until they could right themselves onto all four limbs. Mice were given a maximum of 10 s to fully right themselves.

3.6. Gait Analysis

Gait abnormalities were measured using the Noldus Catwalk 7.1 and Noldus Catwalk XT (version 10.6) systems (Noldus Information Technology B.V., Wageningen, Netherlands). Data acquisition and analysis was done as described before (Martins‐Silva et al. ²⁰¹¹). In brief, mice were acclimated to the dark room for 30 min before starting the experiment. A minimum of 3 runs were recorded for each mouse and later processed using the Noldus software.

3.7. Clasping Reflex

Clasping reflex was measured using an established protocol (Zhu et al. 2016). In brief, clasping reflex was measured by suspending mice on a metal rod for 30 s and scoring their clasping reflex from 0 to 3. A score of 0 meant that the hindlimbs and toes were fully splayed; 1 if one of the limbs was partially retracted toward the body without contacting it; 2 if both hindlimbs were retracted toward the body without contacting it; and 3 if both hindlimbs were fully retracted toward the body.

3.8. Longitudinal Study Design

To examine the natural progression of ALS‐like features in the TAR4 model, we implemented a longitudinal study design spanning four age‐defined time points—4–6 months, 7–9 months, 12–14 months, and 15–17 months of age—that correspond to human adulthood, middle age, and advanced age. These brackets were chosen based on established mouse–human age equivalencies (Fox et al. 2007), in which 3–6 months of age in mice approximates human ages of ~20–30 years and reflects mature adulthood. This stage offers a stable neuromuscular baseline, minimizing confounds from developmental variability seen in juvenile (< 3 months) mice (Fox et al. 2007). The next timepoint of 7–9 months (~33–37 human years) represents the transition into midlife, followed by 12–14 months (~38–47 years) and 15–17 months (~50–55 years), which reflect middle age and early senescence. This design allows for detection of early disease onset and tracking of ALS‐like pathology across aging stages that closely mirror the natural history of ALS in humans (Atsuta et al. 2009).

3.9. Statistical Analyses

All data were statistically analyzed using GraphPad Prism 10 software. Normality was assessed using the Shapiro–Wilk test. Parametric tests were used unless otherwise stated. For comparisons between two groups, an unpaired two‐tailed Student's t‐test was performed; if normality was not met, a Mann–Whitney U test was used instead. One‐way ANOVA was used for three or more groups, followed by Tukey's post hoc test where appropriate. When comparing two or three groups over time, two‐way ANOVA was used if there were no missing values, followed by Šidák‐corrected multiple comparisons for two‐group analyses and Tukey's post hoc test for three‐group analyses. If missing values were present, linear mixed‐effects models were used instead. Outliers were identified using the ROUT test with a false discovery rate of 5%. Four mice (TAR4 n = 2; TAR4‐STI1^TGA n = 1; TAR4‐STI1^−/+ n = 1) were excluded from the open field analysis due to being identified as outliers based on this criterion. All data are represented as mean ± SEM. In cases where data variability is minimal, error bars may not be visible on the graphs.

The study was limited by the COVID‐19 pandemic disrupting behavioral testing and thus preventing some experiments from being conducted for certain cohorts of mice at specific time points, accounting for differences in the number of mice at each time point. This is the reason why some data is missing for hind limb clasping and gait analysis experiments. Given the resulting variation in the number of mice across time points, additional analyses were conducted to confirm that cohort variability did not skew our results. Mice that completed testing at every time point were analyzed using repeated‐measures to confirm that same group differences are observed as in the mixed‐effects model. Despite the smaller sample size for the repeated measures model, statistical power remained sufficient to detect the same differences, mitigating concerns about skewed results due to cohort variability.

4. Results

4.1. TAR4 Mice Present With Sex‐Specific Motor Deficits

Given that age and sex are risk factors for ALS, we first investigated sex‐specific motor phenotypes in hemizygous TAR4 mice, which can reportedly live up to 26 months of age (Janssens et al. 2013). Homozygous TAR4/4 mice do not survive beyond 4 weeks after birth; thus, motor function was not as extensively assessed in this model.

Compared to control wild‐type littermate mice, motor deficits related to motor coordination and endurance were observed in TAR4 males at 4–6 months of age (p = 0.013), as measured by their ability to run on an accelerating rod without falling, and this deficit persisted to 7–9 months (p = 0.009) and 15–17 (p = 0.005) months of age (Figure 1c). Additionally, TAR4 rotarod performance progressively declined between 4–6 and 12–14 (p = 0.002) and 7–9 to 15–17 (p = 0.027) time points when examining within‐group differences (Figure 1c). Although female TAR4 mice did not differ from wild‐type controls at any individual time point on the rotarod task, within‐group comparisons revealed a significant decline in TAR4 female performance from 4 to 7 to 12–14 (p = 0.007) and 7–9 to 15–17 (p = 0.004) months of age, which was not observed in wild‐type females (Figure 1d). To assess changes in strength, the wire hang task was used by putting mice on a wire grid and timing their latency to fall when positioned upside down. Male TAR4 mice showed an age‐dependent decline in wire hang performance when compared to wild‐type controls, with differences appearing at 12–14 (p = 0.003) months that continued to decline to 15–17 (p = 0.034) months of age (Figure 1e). Within‐group analysis of male TAR4 mice showed a decrease in wire hang performance between 4–6 and 12–14 (p = 0.004) and 7–9 to 15–17 (p = 0.009) time points (Figure 1e). Wire hang performance was not affected in female TAR4 mice when compared to wild‐type mice (Figure 1f). Importantly, there were no significant weight differences between male and female TAR4 mice when compared to wild‐type controls, suggesting that the observed motor deficits were not due to weight (Figure S1a,b). Together, these data indicate that male TAR4 mice have deficits related to motor coordination, balance, and strength that progressively worsen over time—deficits that mirror those commonly observed in patients with ALS (Schell et al. 2019)—whereas female TAR4 mice present less or even no deficits.

FIGURE 1.

TAR4 mice show sex‐specific motor deficits. (a) Longitudinal timeline for hemizygous TAR4 mice. (b) At each timepoint mice were analyzed on rotarod, wire hang, catwalk gait analysis, hindlimb clasping, open field, and weight tracking. (c, d) Rotarod test of TAR4 and wild‐type male and female mice. The average amount of time spent on the rod is shown across four time points. (e, f) Wire hang task of male and female TAR4 and wild‐type mice. (g, h) Stride length in millimeters measured by catwalk gait analysis test for male and female TAR4 and wild‐type mice. (i, j) Swing time measured in seconds (s) by catwalk gait analysis test of male and female TAR4 and wild‐type mice. (k, l) BOS measurement in centimeters (cm) is shown for male TAR4 and wild‐type mice of front and hindlimbs. (m, n) Female TAR4 and wild‐type BOS. Mixed‐effects model with multiple comparisons was used to analyze differences within groups (black lines) and between groups (asterisks above time points). All data are represented as mean ± SEM. *p < 0.05; **p < 0.01; ***p < 0.001; ****p < 0.0001. Numbers represent mice tested in each timepoint.

Because gross motor function tests revealed a sex difference in the TAR4 model, we investigated whether there were also differences in more subtle motor functions—such as gait, locomotive activity, and clasping reflexes—which may manifest as early deficits in ALS models (Pitzer et al. 2021; Mancuso et al. ²⁰¹¹) and are relevant to ALS given their association with motor impairments seen in humans (Schell et al. 2019). Stride length, a gait parameter of how far a mouse moves its paw with each step, was found to be decreased (4–6: p = 0.038; 7–9: p = 0.048; 12–14: p = 0.0009; 15–17: p = 0.007) in male TAR4 mice at every time point measured when compared to wild‐type males (Figure 1g). There was also a significant decline (p = 0.002) in their overall stride length from 7–9 to 15–17 months of age that was not seen in wild‐type controls (Figure 1g). This decrease in stride length suggests that TAR4 males are displaying neuromuscular dysfunction and weakness. Interestingly, female TAR4 mice also showed changes related to stride length at 7–9 months of age (p = 0.029) and later at 15–17 months (p = 0.0004), but not at 12–14 months (when male TAR4 mice showed the greatest difference: ≈20 mm shorter than wild‐types; Figure 1h). Swing time (s), the time it takes for the mouse to swing a given paw, was also assessed. Male TAR4 mice showed decreased swing times at all time points compared to control mice (4–6: p = 0.0001; 7–9: p = 0.0002; 12–14: p = 0.003; 15–17: p = 0.007; Figure 1i). This decrease suggests that TAR4 mice are compensating for balance issues by adopting a faster stride. By contrast, female TAR4 mice only showed significantly lower swing times at the advanced age of 15–17 months (p = 0.014; Figure 1j). Stability and walking pattern can be assessed by measuring the base of support (BOS) of front and hind limbs, which is represented by the width across two paws during a stance. Male TAR4 mice had an increased BOS of the hind limbs at 4–6 months (p = 0.033) and front limbs at 7–9 months of age (p = 0.011; Figure 1k,l), whereas BOS was not affected in female TAR4 mice (Figure 1m,n). This increase in hind limb BOS for male mice suggests that they adopt a wider stance to improve balance and stability when walking. Clasping reflex, the ability of mice to splay their hindlimbs outwards when held by the tail, was significantly impaired in male TAR4 mice across all time points (7–9: p = 0.002; 12–14: p = 0.091; 15–17: p = 0.019; Figure S1c). This decrease in clasping reflex suggests a decline in reflexive motor control. Female TAR4 mice also revealed compromised clasping reflex at 7–9 (p = 0.0001) and 12–14 (p = 0.0002) months of age; however, this deficit was not seen at the time point of 15–17 months (Figure S1d).

Hyperactivity has been noted in ALS/Frontotemporal dementia (FTD) mouse models, which likely represent clinical features of FTD (Scekic‐Zahirovic et al. ²⁰²¹; Alfieri et al. ²⁰¹⁶; Ke et al. ²⁰¹⁵). Our open field testing revealed that male TAR4 mice traveled significantly longer distances over 30 min than wild‐type mice at 7–9 (p = 0.027) and 15–17 (p = 0.003) months of age, whereas TAR4 females did not significantly differ compared to wild‐types (Figure S1e,f). This increase in distance traveled for male TAR4 mice suggests that they are hyperactive. There was also a significant decline in distance traveled between 4–6 and 12–14 months for all groups, regardless of sex and genotype. Notably, wild‐type mice showed non‐significant (p > 0.05) trends for worsened motor performance with aging, especially on the rotarod, wire hang, gait analysis, and open field tasks, which have been reported by previous researchers (Fahlström et al. 2012; Shoji et al. ²⁰¹⁶; Yanai and Endo ²⁰²¹). Taken together, our data revealed sex differences in the TAR4 model related to gait, clasping reflexes, and activity, with males showing earlier and more robust changes than females across all time points.

4.2. Male and Female TAR4 Mice Share Similar TDP‐43 Histopathology

Previous research characterizing TDP‐43^wt mouse lines primarily focused on TAR4/4 and TAR6/6 homozygous mice (Wils et al. 2010; Janssens et al. ²⁰¹³; Scherz et al. ²⁰¹⁸), whereas information on hemizygous TAR4 mice with subtle TDP‐43 overexpression (0.5–1 fold) was limited. To characterize the TAR4 hemizygous line biochemically, we fractionated cortical and cervical spinal cord tissues into cytoplasmic, nuclear, and urea‐soluble proteins and analyzed them separately by sex (Figure 2a). We focused our experiments on 15–17 month‐old mice, when motor deficits are most pronounced and peak pathology is expected. In male TAR4 mice, full‐length (p = 0.0002), 35 kDa (p = 0.0001), and 25 kDa (p = 0.0357) truncated cytoplasmic species of TDP‐43 were significantly elevated in the cortex compared to wild‐type controls (Figure 2b,c). Similarly, full‐length TDP‐43 (p = 0.035) and 35 kDa C‐terminal fragments (p = 0.003) were significantly elevated in the spinal cords of male TAR4 mice (Figure 2d,e). Conversely, nuclear TDP‐43 levels in the cortex were similar for mutants and controls (Figure 2f,g). Furthermore, a significant increase in aggregated TDP‐43 levels (p = 0.009) was observed in the cortical urea‐soluble protein fraction (insoluble proteins) for TAR4 males (Figure 2h,i).

FIGURE 2.

Male and female TAR4 mice present with elevated levels of cytoplasmic and insoluble TDP‐43 species. (a) Mouse cortex and spinal cord tissues were fractionated into cytoplasmic, nuclear, and insoluble proteins. Western blotting analysis of cytoplasmic, nuclear, and RIPA (whole lysate) proteins confirmed effective fractionation using GAPDH and NeuN antibodies. (b, c) Western blot of cytoplasmic TDP‐43 proteins from male cortices with quantifications relative to GAPDH signal. (d, e) Western blot of cytoplasmic TDP‐43 proteins from male spinal cords with quantifications relative to GAPDH signal. (f, g) Western blot of nuclear TDP‐43 protein from male cortices with quantification relative to amido black signal. (h, i) Western blot of urea‐soluble TDP‐43 protein from male cortices with quantification relative to amido black signal. (j, k) Western blot of cytoplasmic TDP‐43 proteins from female cortices with quantifications relative to GAPDH signal. (l, m) Western blot of cytoplasmic TDP‐43 proteins from female spinal cords with quantifications relative to GAPDH signal. (n, o) Western blot of nuclear TDP‐43 protein from female cortices with quantifications relative to amido black signal. (p, q) Western blot of Urea‐soluble TDP‐43 protein from female cortices with quantification relative to amido black signal. Data are presented as mean ± SEM and were analyzed using unpaired t‐tests or Matt‐Whitney U test (n = 3–5 mice/group). *p < 0.05; **p < 0.01; ***p < 0.001; ****p < 0.0001.

Similar results were observed in female TAR4 mice: cytoplasmic full‐length TDP‐43 (p = 0.035), 35 kDa (p = 0.006), and 25 kDa (p = 0.013) C‐terminal fragments were increased in the cortex (Figure 2j,k), as well as in the spinal cord (full‐length TDP‐43 p = 0.0005; 35 kDa p = 0.0007; Figure 2l,m). Nuclear TDP‐43 levels were similar in the cortex of female mutants and controls (Figure 2n,o), and aggregated TDP‐43 (urea‐soluble) was increased (p = 0.047) in TAR4 females compared to controls (Figure 2p,q).

To determine whether baseline levels of TDP‐43 were different in male and female mice, we compared samples from both sexes on the same blots. We found no significant differences in total TDP‐43 steady‐state levels between the sexes (Figure S2). These results indicate that TDP‐43 localization, fragmentation, and solubility changes in male and female TAR4 hemizygous mice are similar at 15–17 months of age.

4.3. Homozygous Mutants (TAR4/4) Present With TDP‐43 Pathology, Reduced Body Weight, and Impaired Righting Reflex

We also analyzed righting reflex and TDP‐43 pathology in homozygous TAR4/4 mice. Given that male and female TAR4/4 mice exhibited similar onset of symptoms, sex was not considered as a variable; male and female mice were mixed for behavioral analyses, whereas Western blot analyses were conducted in male mice only to maintain consistency across biochemical assays. At 18–20 days old, the TAR4/4 mice weighed significantly less (p = 0.0003) than their wild‐type littermates and presented with righting reflex deficits (Figure S3b,c; p = 0.0014), confirming that increased levels of TDP‐43 significantly accelerate the motor phenotypes. Biochemical analysis revealed that, even at an early age, cytoplasmic levels of TDP‐43 (3–4‐fold increase; p = 0.002), 35 kDa (p = 0.002), and 25 kDa (p = 0.0001) are significantly elevated in the TAR4/4 mice compared to controls (Figure S2d,e). Insoluble TDP‐43 levels were similarly increased (Figure S2h,i; p = 0.008). Overall, these results suggest that after a certain threshold of increased TDP‐43 expression, mislocalization, and aggregation, toxicity accelerates drastically, leading to early death.

4.4. Reducing STI1 in TAR4 Mice Can Correct Cytoplasmic TDP‐43 Levels in Both Sexes

Toxicity and aggregation of TDP‐43 are regulated by STI1 in vitro (Lin et al. 2021). Specifically, deleting STI1 in yeast, SN56 cells, and mouse embryos can decrease endogenous TDP‐43 levels. However, whether changing STI1 levels in vivo can mitigate TDP‐43 proteinopathy is unknown. To investigate this, TAR4 mice were crossed with heterozygous STI1 knockout mice to generate double mutants (TAR4‐STI1^−/+). In male TAR4‐STI1^−/+ mice, there was a significant reduction of cytoplasmic full‐length TDP‐43 (p = 0.027), 35 kDa (p = 0.011), and 25 kDa (p = 0.029) fragments in the cortex, with no changes to phosphorylated TDP‐43 (pTDP‐43; Figure 3a–c). Female TAR4‐STI1^−/+ mice only showed a reduction in 25 kDa C‐terminal fragments (p = 0.018; Figure 3d–f). Cytoplasmic fractions obtained from the cervical spinal cord showed a reduction in full‐length TDP‐43 for both male (p = 0.006) and female (p = 0.011) TAR4‐STI1^−/+ mice (Figure 3g,h). Urea‐soluble fractions obtained from the cortices showed no changes in male TAR4‐STI1^−/+ mice, but females had a reduction in full‐length TDP‐43 (p = 0.049) and an increase in pTDP‐43 (p = 0.041; Figure 3k–p). Nuclear fractions obtained from mice with reduced STI1 levels did not show any TDP‐43 expression changes (data not shown). Together, these results suggest that reducing STI1 levels can regulate TDP‐43 proteinopathy with sex‐specific outcomes: males showed broad reductions of cytoplasmic TDP‐43 species in the cortex, whereas females showed only decreased cytoplasmic 25 kDa in the cortex, but also decreased insoluble TDP‐43 and increased insoluble pTDP‐43. These results indicate the potential sex‐specific regulatory effects of STI1 reduction on TDP‐43 trafficking, aggregation, and post‐translational modifications.

FIGURE 3.

Reducing STI1 levels in TAR4 mice regulates cytoplasmic TDP‐43 levels in males and females at 15–17 months of age. (a–c) Western blot of cytoplasmic TDP‐43 proteins from the cortices of male TAR4 and TAR4^−/+ mice quantified relative to GAPDH and full‐length TDP‐43. (d‐f) Western blot of cytoplasmic TDP‐43 proteins from the cortices of female TAR4 and TAR4‐STI1^−/+ mice quantified relative to GAPDH and full‐length TDP‐43. (g, h) Western blot of cytoplasmic TDP‐43 proteins from the spinal cords of male TAR4 and TAR4^−/+ mice quantified relative to GAPDH. (i, j) Western blot of cytoplasmic TDP‐43 proteins from the spinal cords of female TAR4 and TAR4^−/+ mice quantified relative to GAPDH. (k–m) Western blot of urea‐soluble TDP‐43 and pTDP‐43 from the cortices of male TAR4 and TAR4^−/+ mice quantified relative to amido black and full‐length TDP‐43. (n–p) Urea‐soluble TDP‐43 and pTDP‐43 from the cortices of female TAR4 and TAR4^−/+ mice quantified relative to amido black and full‐length TDP‐43. Data are presented as mean ± SEM and were analyzed using unpaired t‐tests or Matt‐Whitney U test (n = 4–5 mice/group). *p < 0.05; **p < 0.01.

4.5. Reducing STI1 in TAR4/4 Mice Only Partially Reduces Mislocalized TDP‐43

Homozygous transgenic mice with reduced levels of STI1 (TAR4/4‐STI1^−/+) showed no changes in their weights and righting reflex scores compared to TAR4/4 mice (Figure S4a,b). However, there was a significant reduction in full‐length cytoplasmic TDP‐43 (p = 0.007). No changes were observed for the urea‐soluble fraction (Figure S4e,f). This suggests that partially reducing STI1 can influence cytoplasmic TDP‐43 levels in mice with both high and moderate TDP‐43 expression levels, but does not consistently reduce insoluble TDP‐43.

4.6. STI1 Overexpression in TAR4 Mice Increases TDP‐43 C‐Terminal Fragments in Both Sexes

Overexpression of STI1 in vitro can have varying dose‐dependent effects on TDP‐43 in cellular models: moderate overexpression reduces TDP‐43 toxicity, and high overexpression exacerbates it (Lin et al. 2021). To test how higher STI1 expression levels affect TDP‐43 in vivo, we increased STI1 levels by 3‐fold in TAR4 mice using STI1 overexpressing mice. Overexpressing STI1 did not affect cytoplasmic full‐length and phosphorylated TDP‐43 levels in the cortices of male and female TAR4‐STI1^TGA mice; though it led to a significant increase in 35 kDa fragments (p = 0.0001) in males and an increase in both 35 kDa (p = 0.01) and 25 kDa (p = 0.017) for females when compared to sex‐matched TAR4 mice (Figure 4a–f). Interestingly, full‐length TDP‐43 levels were significantly reduced in male TAR4‐STI1^TGA but not female spinal cord (p = 0.001; Figure 4g–j). Insoluble levels of TDP‐43 were unaltered for both male and female TAR4‐STI1^TGA mice; however, there was a significant increase in pTDP‐43 in females (p = 0.007; Figure 4k–p). Nuclear fractions obtained from STI1 overexpressing mice did not show any TDP‐43 expression changes (data not shown). These findings suggest that moderate STI1 overexpression in TAR4 mice results in a complex modulation of TDP‐43 processing during aging, with distinct sex‐dependent outcomes. Moreover, cytoplasmic TDP‐43 fragments were increased for both sexes in the cortex, but there was a reduction in spinal cord TDP‐43 exclusively in males. Insoluble levels of TDP‐43 were unaltered, but its phosphorylation was increased in females, indicating that the effects of STI1 overexpression on TDP‐43 are multifaceted and influenced by sex.

FIGURE 4.

Overexpressing STI1 leads to increased mislocalization of TDP‐43 C‐terminal fragments in male and female cortices at 15–17 months of age. (a) Western blot of cytoplasmic TDP‐43 proteins from the cortices of male TAR4 and TAR4^TGA mice. (b, c) Cytoplasmic TDP‐43 species quantified relative to GAPDH and full‐length TDP‐43. (d–f) Western blot of cytoplasmic TDP‐43 proteins from the cortices of female TAR4 and TAR4‐STI1^TGA mice quantified relative to GAPDH and full‐length TDP‐43. (g, h) Western blot of cytoplasmic TDP‐43 proteins from the spinal cords of male TAR4 and TAR4^TGA mice quantified relative to GAPDH. (i, j) Western blot of cytoplasmic TDP‐43 proteins from the spinal cords of female TAR4 and TAR4^−/+ mice quantified relative to GAPDH. (k–m) Western blot of urea‐soluble TDP‐43 and pTDP‐43 from the cortices of male TAR4 and TAR4^TGA mice quantified relative to amido black and full‐length TDP‐43. (n–p) Urea‐soluble TDP‐43 and pTDP‐43 from the cortices of female TAR4 and TAR4^TGA mice quantified relative to amido black and full‐length TDP‐43. Data are presented as mean ± SEM and were analyzed using unpaired t‐tests or Matt‐Whitney U test (n = 4–5 mice/group). *p < 0.05; **p < 0.01; ****p < 0.0001.

4.7. STI1 Overexpression in TAR4/4 Mice Increases Insoluble TDP‐43 and Cytoplasmic C‐Terminal Fragments

Homozygous transgenic mice (TAR4/4‐STI1^TGA) showed similar results to the hemizygous line in their cytoplasmic cortical fractions, with elevated STI1 levels leading to a significant increase in 35 kDa (p = 0.0002) and 25 kDa (p = 0.0023) fragments when compared to TAR4/4 mice (Figure S5b,c). Insoluble TDP‐43 levels were also significantly elevated (p = 0.001), as well as insoluble 25 kDa (p = 0.004) fragments in mice with more STI1 (Figure S5d,e). These data further demonstrate that STI1 overexpression seems to worsen TDP‐43 proteinopathy, similar to previous results with amyloid‐β (Lackie, Marques‐Lopes, et al. ²⁰²⁰) and α‐synuclein (Lackie et al. 2022).

4.8. Higher STI1 Levels Worsened Male Motor Performance and Increased Morbidity in Females

To understand if STI1‐dependent changes in TDP‐43 influence motor function, we studied the motor performance of male and female TAR4 mice with less or more STI1 across multiple time points. Despite the observable biochemical changes in TAR4‐STI1^−/+ male at 15–17 months, their performance on the rotarod task was unaltered compared to TAR4 mice with normal levels of STI1 (p = 0.8255), and there was a slight decline in overall performance from 7–9 to 15–17 months of age (p = 0.04; Figure 5a). Interestingly, TAR4‐STI1^TGA male mice performed significantly worse than TAR4 controls on the rotarod task at 7–9 months (p = 0.0085) and their performance significantly dropped from 4–6 to 12–14 months (p = 0.001; Figure 5a). Female mice with reduced STI1 performed significantly better on the rotarod task than TAR4 controls at 12–14 months of age (p = 0.002), though their performance still declined from 7–9 to 15–17 months (p = 0.004; Figure 5b). Females overexpressing STI1 had no significant changes in rotarod performance. The wire hang experiment revealed no changes in performance for TAR4‐STI1^−/+ males, but TAR4‐STI1^TGA males presented a significant deficit at every time point (4–6: p = 0.0001; 7–9: p = < 0.0001; 12–14: p = 0.0002; 15–16: p = 0.0001; Figure 5c). Additionally, TAR4‐STI1^TGA performance significantly decreased with increasing age, as identified by within‐group comparisons for 4–6 and 12–14 (p = 0.009) months of age, as well as 7–9 and 15–17 (p = 0.007) months of age. Female mice with altered STI1 levels had no changes in wire hang performance (Figure 5d). The survival of male mutants was unaltered by either STI1 condition, but TAR4 females overexpressing STI1 did present with a significantly lower probability of survival compared to TAR4 controls (p = 0.0157; Figure 5g,h). Overall, decreasing mislocalization of TDP‐43 in males by moderately reducing STI1 did not improve motor performance for rotarod, wire hang, and hindlimb clasping tasks. However, overexpressing STI1 in males quickened and worsened the onset of motor deficits for two tasks, suggesting that increasing STI1 levels is deleterious in the context of TDP‐43 proteinopathy. Importantly, to confirm that STI1 overexpression alone does not affect wire hang and rotarod performance in mice, we tested STI1 overexpressing wild‐type mice (WT‐STI1^TGA) on these tasks and observed no changes (Figure S7a,b). Interestingly, overexpressing STI1 in females did not worsen their motor performance but did reduce survival, and reducing STI1 led to minor improvements in rotarod and clasping‐related tasks.

FIGURE 5.

Reducing cytoplasmic TDP‐43 levels in TAR4 mice does not rescue motor deficits in males. (a, b) Rotarod performance as measured by latency to fall in seconds (s) is shown for TAR4, TAR4‐STI1^−/+, and TAR4‐STI1^TGA males and females over time, respectively. (c, d) Wire hang performance as measured by latency to fall in seconds (s) is shown for TAR4, TAR4‐STI1^−/+, and TAR4‐STI1^TGA males and females over time, respectively. (e, f) Hindlimb clasping scores are shown at final time point of 15–17 months of age for males and females, respectively. (g, h) Probability of survival is shown for TAR4, TAR4‐STI1^−/+, and TAR4‐STI1^TGA males and females, respectively. Mixed‐effects model with multiple comparisons was used to analyze datasets (a–d); differences within groups are represented as black lines and between groups as asterisks above time points. Ordinary one‐way ANOVA with multiple comparisons was used for analysis of datasets (e, f). Simple survival analysis (Kaplan–Meier) with multiple comparisons was used for analysis of (g, h) analysis. All data are represented as mean ± SEM. *p < 0.05; **p < 0.01; ***p < 0.001; ****p < 0.0001. Numbers represent number of mice at each timepoint.

4.9. Reduction of STI1 Levels and Activity Can Partially Correct Motor Deficits in Males

To further evaluate whether TDP‐43 burden can be improved by lowering STI1 levels, we used a hypomorphic STI1 allele (STI1 ΔTPR1), which has been shown to dramatically reduce the activity and function of STI1 (Lackie, Razzaq, et al. ²⁰²⁰). We have previously shown that one STI1 ΔTPR1 allele (termed ΔHET) can rescue behavioral deficits in synucleinopathy mouse models by reducing the levels of toxic alpha‐synuclein (Lackie et al. 2022). To test if this allele can further reduce TDP‐43 pathology and related motor deficits, we generated male TAR4 mice with one ΔTPR1 allele and found that it also corrects mislocalized full‐length TDP‐43 (p = 0.016), but not insoluble TDP‐43 in the cortex (Figure 6a–d). Genetic backgrounds are well‐known to greatly affect the symptom severity in mouse models of ALS (Lutz 2018). In our study, we observed a difference in motor performance for TAR4 mice on a C57BL/6j background (Figure 6) compared to the mixed C57BL/SJL background (Figures 1, 2, 3, 4, 5). Specifically, on the C57BL/6j background, TAR4 mice do not exhibit an age‐dependent decline in rotarod performance, but do present with lower baseline (4–6 months) performance on wire hang (Figure 6e,f). Despite this discrepancy, motor performance on the rotarod did not differ between TAR4 and TAR4‐ΔHET mice (Figure 6e). However, there was improvement in TAR4‐ΔHET wire hang performance that reached significance at 7–9 (p = 0.044) and 12–14 months (p = 0.023) when compared to plain TAR4 males (Figure 6f). However, both TAR4 and TAR4‐ΔHET performance significantly decreased from 4–6 to 12–14 months, suggesting an age effect.

FIGURE 6.

Hypomorphic STI1 allele is sufficient to correct some motor deficits in TAR4 males. (a, b) Western blot of cytoplasmic TDP‐43 proteins from TAR4 and TAR4‐ΔHET male cortices quantified relative to amido black. (c, d) Western blot of urea‐soluble TDP‐43 levels from TAR4 and TAR4‐ΔHET male cortices quantified relative to amido black. (e) Rotarod performance as measured by latency to fall is shown for TAR4 and TAR4‐ΔHET males. (f) Wire hang performance as measured by latency to fall is shown for TAR4 and TAR4‐ΔHET males. (g–j) Catwalk gait analysis performance for TAR4 and TAR4‐ΔHET mice showing swing time (s), stride length, front BOS, and hind BOS. (k) Stride length for each paw of TAR4 and TAR4‐ΔHET mice at 15–17 months of age. (l) Hindlimb clasping reflex scores for TAR4 and TAR4‐ΔHET mice. Unpaired t‐tests were used for (a–d) analysis (n = 5 mice/group). Mixed‐effects model with multiple comparisons was used to analyze datasets (e–j); differences within groups are represented as black lines and between groups as asterisks above time points. Two‐way ANOVA with multiple comparisons was used for (k) analysis. Unpaired t‐test was used for analysis of dataset (l). All data are represented as mean ± SEM. *p < 0.05; **p < 0.01; ***p < 0.001; ****p < 0.0001. Numbers represent number of mice at each timepoint.

Gait analysis performance showed no differences for swing time and stride length for both groups (Figure 6g,h). BOS of forelimbs was also unaltered; however, TAR4‐ΔHET males' hindlimb BOS was significantly different from TAR4 males at 12–14 months (p = 0.017; Figure 6i,j), suggesting that TAR4‐ΔHET mice have better hindlimb support. Thus, to further investigate this, we performed a cross‐sectional analysis of separated limbs' stride length at 12–14 months and saw significant improvements in their strides (RF: p = 0.016; LF: p = 0.045; RH: p = 0.006; Figure 6k). However, hindlimb clasping reflex showed no improvements for TAR4‐ΔHET mice (Figure 6l). Finally, hyperactivity seen in male TAR4 mice did not change in TAR4‐ΔHET mice (Figure S8a). Overall, these results indicate that the ΔHET allele reduces TDP‐43 mislocalization in the brain and appears to modestly improve performance in the wire hang and catwalk gait analysis tasks.

5. Discussion

Our study reveals that the onset and progression of gross and fine motor dysfunction is more severe in TAR4 male than female mice and can also be modulated by STI1, providing new insights on sex regulation of TDP‐43‐related phenotypes and proteostasis control. We demonstrate that TDP‐43 mislocalization and aggregation is similar in mild TDP‐43 overexpressing TAR4 males and females; however, male mice present with more pronounced and earlier motor phenotypes compared to female mice. We also show a complex relationship between STI1 levels and TDP‐43 in both sexes. We found that increased STI1 in general worsens motor performance in male mice and reduces female lifespan. Conversely, reduced levels of STI1 mitigate TDP‐43 mislocalization in both sexes, and dysfunctional STI1 partially restores motor deficits in males. Similar biochemical changes were found in mice with two copies of the transgene (TAR4/4), manipulations of STI1 did not alter phenotypes in this more aggressive model of TDP‐43 toxicity.

Previous work reported rotarod deficits in TAR4 mice only at 14–16 months of age; however, in these publications, the sex of the animals was not reported (Wils et al. 2010; Janssens et al. ²⁰¹³). In our analysis, we clearly show that motor deficits manifest as early as 4–6 months in this model, with sex differences appearing on multiple motor tasks. It is possible that motor phenotypes emerge earlier than 4 months in these mutants; however, we did not investigate this possibility, as behavioral testing in younger mice can be unreliable due to ongoing neuromuscular development (Fox et al. 2007). Female sex is known to delay the onset of ALS in humans and in mouse models with mutated TDP‐43 (Blasco et al. 2012; Bargsted et al. ²⁰¹⁷; Watkins et al. ²⁰²⁰). Our data support these findings in a mild wild‐type TDP‐43 overexpression model, which may more accurately recapitulate ALS phenotypes. Indeed, TAR4 females only exhibited a partial and delayed decline in motor deficits, underscoring the importance of sex‐specific analysis in ALS studies.

Previous studies have shown TDP‐43 mislocalization, microgliosis, astrogliosis, and spastic paralysis in TDP‐43^WT mouse lines (Wils et al. 2010; Scherz et al. ²⁰¹⁸), but TDP‐43 protein levels in TAR4 mice were not thoroughly examined. By using subcellular fractionation and direct comparisons between sex‐matched TAR4 and wild‐type mice, we quantified TDP‐43 cellular distribution in this model and found a 50%–100% increase in full‐length TDP‐43 in cytoplasmic and insoluble protein fractions for both sexes. The Thy‐1 promoter used to drive TDP‐43 expression in TAR4 mice has an estrogen response element, with previous studies attributing some sex‐specific outcomes in transgenic mice to this element (Forner et al. 2021; Sadleir et al. ²⁰¹⁵). Another ALS model overexpressing human wild‐type TDP‐43 with a Thy1.2 promoter reported TDP‐43 expression differences between male and female mutants in RIPA fractions, with females reportedly expressing less TDP‐43 (Shan et al. 2010). However, we did not observe these differences in TDP‐43 levels across cytoplasmic, nuclear, and insoluble protein fractions in our study. To account for possible confounding effects of different baseline TDP‐43 expression levels in either sex, we directly compared both sexes on the same blots, but still did not observe any differences in TDP‐43 levels for all fractions. Our results thus demonstrate that while both sexes share similar TDP‐43 pathology, female motor functions remain largely unencumbered. Whether this resilience is due to sex hormones or other factors remains to be defined by future studies.

Protein quality control components are known to be sequestered to inclusions in protein‐misfolding diseases, including sporadic ALS (Watanabe et al. 2001). The STI1 co‐chaperone for Hsp90 and Hsp70 interacts with TDP‐43 and can regulate its toxicity (Lin et al. 2021). Although STI1 levels generally decline with age in mammals (Lackie, Razzaq, et al. ²⁰²⁰), they are found to be increased in humans with Alzheimer's and Parkinson's disease (Lackie et al. 2022; Ostapchenko et al. ²⁰¹³) and, to some extent, in TDP‐43 proteinopathies such as FTD and ALS (Umoh et al. 2018; San Gil et al. ²⁰²⁴). Overexpression of STI1 in both the 5xFAD Alzheimer's mouse model and M83 synucleinopathy mouse models escalated protein aggregation and worsened behavioral phenotypes (Lackie, Marques‐Lopes, et al. ²⁰²⁰; Lackie et al. ²⁰²²). These findings support the notion that imbalances in particular chaperone machinery components may detrimentally affect protein aggregates by stabilizing and facilitating the conversion of misfolded proteins (Tittelmeier et al. 2020). Our observation that STI1 overexpression in the TAR4 model is associated with ~30% increased mortality in females, worsened motor performance in males, and an increased trend for aberrant TDP‐43 post‐translational modifications further supports this notion. Notably, female TAR4 mice overexpressing STI1 exhibited elevated levels of pathological 35 kDa (~30% increase) and 25 kDa (~100% increase) TDP‐43 fragments in the cortex, along with increased pTDP‐43 (~15% increase), which may mechanistically underlie their heightened mortality. Given that STI1 facilitates client transfer between Hsp70 and Hsp90, increased STI1 levels may alter the efficiency of TDP‐43 processing by these chaperones, potentially exacerbating the aggregation of misfolded TDP‐43 and enhancing its toxicity. Alternatively, STI1 may act as a holdase (Rutledge et al. 2024), which could facilitate the accumulation of TDP‐43 independent of chaperones.

The in vivo overexpression of STI1 in our study did not recapitulate the protective effects observed in vitro (Lin et al. 2021). This discrepancy mirrors earlier findings in Alzheimer's disease models, where STI1 and Hsp90 overexpression reduced Aβ toxicity in worms, but proved detrimental in mammalian systems (Lackie, Marques‐Lopes, et al. ²⁰²⁰). Additionally, another mouse study found that overexpressing Hsp90 co‐chaperones in wild‐type mice alone can create a pro‐aggregation environment (Criado‐Marrero et al. ²⁰²¹). Collectively, these studies highlight how the complex biology of mammalian systems can produce outcomes that diverge from those observed in simpler models and in cell culture, especially in response to chaperone imbalances.

A recent longitudinal proteomic study analyzing cortical tissues from the rNLS8 mouse model of ALS—a model that presents with human wild‐type TDP‐43 mislocalization and aggregation—found a significant increase in molecular chaperones at early stages of disease (San Gil et al. 2024). The short‐lived nature of this proteostatic increase may reflect early cellular protective effects that cannot be maintained in later stages of disease. This notion was confirmed by modulating the levels of DNAJB5, a Hsp70 co‐chaperone, and observing detrimental motor and pathological effects from its reduction in mice, and protective effects from its overexpression in cells (San Gil et al. 2024). These findings are supported by other studies demonstrating anti‐aggregation properties of DnaJ (Hsp40) family co‐chaperones in ALS models, possibly due to the Hsp40‐Hsp70 complex being a crucial component for misfolded protein clearance (Chen et al. 2016; Takeuchi et al. ²⁰⁰²; Novoselov et al. ²⁰¹³). Nonetheless, our findings reinforce the idea that fine‐tuning proteostasis components may be much more complex than increasing the levels of one specific co‐chaperone. Rather, it seems that molecular chaperones and their partners have varying specificities for misfolded proteins and can differentially affect their fate (Tittelmeier et al. 2020).

The concept of the epichaperome, a maladaptive network of chaperones and co‐chaperones, has garnered attention in neurodegenerative disease and cancer research, as it appears capable of stabilizing pathological protein aggregates (Rodina et al. 2016; Inda et al. ²⁰²⁰). Epichaperome assembly seems to be context‐dependent, with stress‐inducing cellular conditions possibly reshaping Hsp90's interactions and functions that may lead to the formation of this maladaptive complex (Chiosis et al. 2023). Specifically, post‐translational modifications such as phosphorylation of Hsp90 in diseased states seem to enhance its binding with co‐chaperones, including STI1, leading to new gain‐of‐function mechanisms that may contribute to aberrant protein‐to‐protein interactions and dynamics that favor epichaperome assembly (Roychowdhury et al. 2024). Interestingly, selectively targeting either STI1 or Hsp90 can effectively collapse the epichaperome (Rodina et al. 2016), suggesting that these molecules act as hubs or nodes for epichaperome formation. A brain permeable pharmacological inhibitor of the epichaperome, PU‐AD (icapamespib), is safe to use in humans (Silverman et al. 2022) and has been effective in correcting protein misfolding in mouse models of Alzheimer's disease (Inda et al. 2020). Mechanistically, the inhibitor has a high affinity for the pathologically altered Hsp90 ATP binding site, forsaking normal Hsp90 proteins (Silverman et al. 2022). As such, a better understanding of how particular chaperones and co‐chaperone dynamics alter in disease can help uncover novel avenues of precision therapeutics, where specific conformations of pathogenic proteins can be targeted.

Lin et al. (2021) found that decreasing STI1 in vitro (SN56 neuronal culture and yeast cells) and mouse embryos can reduce endogenous TDP‐43 levels. This STI1 reduction was also toxic for yeast cells with TDP‐43 proteinopathy, enhancing insolubility of TDP‐43 and cellular death. While we did not observe an increase in TDP‐43 toxicity for mutants with reduced STI1 levels, our results further confirmed the conserved role of STI1 in modulating TDP‐43 levels in mammals, as a reduction in STI1 led to decreased cytoplasmic TDP‐43 levels by about 50%. Moreover, this general reduction of cytoplasmic TDP‐43 levels in TAR4‐STI1^−/+ mice suggests that TDP‐43 may be a client protein for STI1/Hsp90; decreasing STI1 levels is known to reduce client proteins of Hsp70 and Hsp90 in vivo (Lackie, Razzaq, et al. ²⁰²⁰). Alternatively, it is possible that STI1 creates a favorable condition for maintaining TDP‐43 mislocalization in vivo, thus reducing STI1 levels can facilitate TDP‐43 clearance. Importantly, alterations in STI1 levels were accompanied by corresponding changes in Hsp90β expression, likely reflecting their functional interdependence within the chaperone network (Figure S9). Given STI1's role as a co‐chaperone of Hsp90, these changes may result from compensatory proteostasis mechanisms or transcriptional co‐regulation. This makes it challenging to discern whether the effects observed in this study are due to STI1, Hsp90β, or the combined effects of reductions in both proteins.

Using the STI1‐ΔTPR1 model, we have previously shown a rescue of cognitive and pathological deficits in two models of a‐Syn toxicity (Lackie et al. 2022). Unlike the heterozygous STI1 knockout, the ΔTPR1 allele leads to both a reduction in STI1 levels and activity. The addition of this allele in TAR4 males was associated with a partial reduction in the penetrance of motor symptoms, suggesting that motor performance in the TAR4 model can be potentially corrected with additional interference to the STI1 gene. While we see STI1 gene dose‐dependent effects in this study, the exact mechanism behind these effects is not well understood. In SN56 cells, Hsp70, Hsp90, and STI1 co‐immunoprecipitate in a complex with TDP‐43 (Lin et al. 2021), but the exact order by which this interaction occurs and with which TDP‐43 domain(s) remains to be elucidated. C‐terminal TDP‐43 fragments were also found to co‐immunoprecipitate with Hsp70 and Hsp90 in human neuroblastoma cells, and reducing either chaperone with siRNA exacerbated C‐terminal levels, suggesting that both Hsps partake in the degradation of these fragments (Zhang et al. 2010). Given that we observed TDP‐43 C‐terminal fragment modulation in response to altered STI1 levels, it is possible that STI1 and its partners affect cleavage and stability of TDP‐43 in vivo, though in an inverse relationship to that found in cellular models. Given that STI1 facilitates client transfer between Hsp70 and Hsp90, changes in STI1 levels may alter the efficiency of TDP‐43 processing by these chaperones in vivo.

In summary, we found a partial relationship between STI1 levels and TDP‐43 toxic effects and aggregation in mice. We also found strong sex‐related phenotypes that need to be better understood, suggesting the female mice are more resilient to increased TDP‐43 levels. Future experiments targeting the imbalance of proteostasis, for example, with inhibitors of the epichaperome, may be warranted to further determine potential treatments for TDP‐43 proteinopathies.

Author Contributions

Vladislav Novikov: investigation, writing – original draft, writing – review and editing, visualization, methodology, validation, formal analysis, data curation, conceptualization, funding acquisition. Latiyah T. C. Timothy: investigation. Jue Fan: investigation, conceptualization. Kareem Sadek: investigation. Matthew F. Cowan: investigation. Kate M. Onuska: data curation, formal analysis, software, writing – review and editing. Martin Duennwald: conceptualization, methodology, supervision, funding acquisition, investigation, writing – review and editing. Vania F. Prado: conceptualization, methodology, funding acquisition, investigation, project administration, supervision, resources. Marco A. M. Prado: conceptualization, investigation, funding acquisition, writing – review and editing, methodology, project administration, supervision, resources.

Peer Review

The peer review history for this article is available at https://www.webofscience.com/api/gateway/wos/peer‐review/10.1111/jnc.70204.

Supporting information

Figure S1: jnc70204‐sup‐0001‐FigureS1.pdf.

Data S2: jnc70204‐sup‐0002‐DataS1.xlsx.

Novikov, V. , Timothy L. T. C., Fan J., et al. 2025. “A Longitudinal Study of Sex Differences in a TDP‐43 Mouse Model Reveals STI1 Regulation of TDP‐43 Proteinopathy and Motor Deficits.” Journal of Neurochemistry 169, no. 8: e70204. 10.1111/jnc.70204.

Data Availability Statement

The data that support the findings of this study are openly available in figshare at https://figshare.com/s/93be6b11a56b8d16b054, https://figshare.com/s/3804a7a8f2fa9e011a40.