Genetic Background Modulates the Phenotype of a Mouse Model of DYT1 Dystonia

DYT1 dystonia is a debilitating neurological disease characterized by involuntary twisting movements. The disease is caused by an in-frame deletion (GAG, “ΔE”) mutation in the TOR1A gene that encodes the torsinA protein. Intriguingly, only 30% of mutation carriers exhibit motor symptoms despite the fact that functional brain imaging studies show abnormal brain metabolism in all carriers. Because genetic modifiers may be a determinant of this reduced penetrance, we examined the genetic contribution of three different inbred strains of mice on the DYT1 mutation in animals that are homozygous (Tor1aΔE/ΔE) or heterozygous (Tor1aΔE/+; disease state) for the disease-causing ΔE mutation. We find that the DBA/2J, C57BL/6J, and CD1-ICR contribution of genes significantly alter lifespan in Tor1aΔE/ΔE mice, which die during the first few days of life on the 129S6/SvEvTac (129) background. The C57BL/6J (B6) strain significantly decreases life expectancy of Tor1aΔE/ΔE animals but, like 129S6/SvEvTac Tor1aΔE/+ mice, congenic C57BL/6J Tor1aΔE/+ mice do not exhibit any motor abnormalities. In contrast, the DBA/2J (D2) strain significantly increases life expectancy. This effect was not present in congenic DBA/2J Tor1aΔE/ΔE mice, indicating that the extended lifespan of F2 129/D2 mice was due to a combination of homozygous and heterozygous allelic effects. Our observations suggest that genetic modifiers may alter the penetrance of the ΔE mutation, and that mapping these modifiers may provide fresh insight into the torsinA molecular pathway.


Introduction
Dystonia is defined as abnormal involuntary movements that are prolonged, twisting in nature and frequently stereotypic and repetitive. Dystonia occurs as an isolated symptom without evidence of brain injury (''primary'' dystonia) or as a consequence of pathologic insults to the basal ganglia or related structures (''secondary'' dystonia). Primary and secondary dystonia may be treated with similar medications (e.g. anticholinergics) and both respond to deep brain stimulation therapy [1]. These facts, and the characteristic dystonic postures that result from diverse etiologies, suggest that primary and secondary forms of dystonia may share a common downstream abnormality, perhaps a stereotyped disruption of basal ganglia output (from the internal segment of the globus pallidus/substantia nigra pars reticulata).
The most common genetic form of primary dystonia, DYT1 dystonia, is a neurodevelopmental disorder caused by an in-frame deletion (GAG, ''DE'') in the TOR1A gene that results in the loss of a glutamic acid in the C-terminus of torsinA [2,3]. DYT1 dystonia is dominantly inherited but abnormal movements affect only 30% of mutation carriers. Despite this incomplete penetrance, 2-deoxyglucose studies show that all carriers exhibit abnormal brain metabolism, with increased metabolic activity in the cerebellum, putamen/globus pallidus, and supplementary motor cortex [4,5]. Similarly, magnetic resonance diffusion tensor imaging (DTI) shows white matter abnormalities associated with reduced integrity of the cerebellothalamocortical motor pathway in all DYT1 mutation carriers [6,7]. These clinical studies highlight that the apparent penetrance of a mutation depends greatly on the phenotype being assessed, and demonstrate that all DE mutation carriers have abnormally functioning brains.
The factors that determine conversion from sub-clinical ''endophenotype'' to overt disease remain unknown. Similarly, nearly all animals harboring monogenic mutations show significant phenotypic variability, likely due to multiple intermingling factors such as environment, allelic heterogeneity and stochastic effects, as well as the presence of modifier genes [8]. Indeed, a focus on the effect of this ''genetic background noise'' [8] is emerging in an effort to understand what makes some individuals more susceptible than others to certain disease-causing mutations. The features of DYT1 dystonia (monogenic mutation, incomplete penetrance) suggest that this disease may be an excellent model system in which to examine these issues. Possible genetic modifiers of the torsinA pathway include torsinB, which has redundant functions [9], and other torsinA-interacting proteins, including LAP1, LULL1 [10] and printor [11]. Importantly, identifying factors that modulate DE-torsinA phenotypes has the potential not only to provide insight into disease mechanism, but also may suggest alternative strategies for disease treatment and prevention.
Given the many factors that can modulate disease phenotypes, it can be exceedingly difficult to model diseases with limited penetrance, such as DYT1 dystonia. To date, etiologic mouse models of DYT1 dystonia do not have any obvious dystonic features or evidence of pathology such as neuronal loss, including transgenic mice expressing human mutant torsinA (hMT) [12,13,14], and heterozygous knock-in mice in which the DGAG mutation has been introduced in the endogenous mouse Tor1a gene [15,16]. Furthermore, homozygous mutant torsinA expression results in perinatal lethality [17] preventing behavioral analysis of these mice. Therefore, mouse models of DYT1 dystonia suffer from an ''all or none'' effect of mutated torsinA in mice. We set out to explore ways to: 1. Amplify any behavioral abnormalities in the disease state mouse (heterozygous) or 2. Temper the effects of homozygous Tor1a DE/DE mouse (increase lifespan to observe effects).
The lack of a consistent or clearly apparent phenotype may be due in part to the variability in mouse backgrounds used in these studies. Modifier genes present in certain strains may act to suppress or exacerbate the effects of the DE mutation. Numerous studies demonstrate that genetic background alters both baseline and pharmacological responses in mice [18]. Modifier loci have already been mapped for several neurological diseases in both human and mouse. These diseases include tremor, dystonia, epilepsy and Huntington's disease. For example, in the kinesiogenic mouse model of dystonia, Scn8a medj mice exhibit striking phenotypes on the C57BL/6J background with paralysis and lethality by one month of age, while the C3H inbred strain background causes a progressive dystonia and ataxia, but a normal lifespan [19]. To begin to identify genetic modifiers of the torsinA pathway, we utilized mice homozygous for the DE mutation (Tor1a DE/DE ). These mice, which die on the day of birth [16], have histologically normal-appearing brains, but electron microscopic (EM) analysis shows a selective disruption of the neuronal nuclear envelope (NE; referred to as ''blebs'') [16]. We used these phenotypes (animal death, NE blebs) as an in vivo read-out of torsinA function, and explored whether they were modified when placed on distinct genetic backgrounds, a strategy similar to how the rough eye phenotype is utilized in drosophila genetic studies. We pursued an F1 intercross screening strategy to identify background strains that amplify or suppress death or NE blebs. Subsequently, we generated congenic mice on these different genetic backgrounds to test for an effect of background on the behavioral phenotype of Tor1a DE/+ mice. We find that despite the effect of C57BL/6J alleles on the survival of Tor1a DE/DE mice, these alleles did not cause a behavioral phenotype in the congenic Tor1a DE/+ mice (disease genotype).

Materials and Methods
Male mice were housed in groups of 5 and maintained on a 12hour light/dark schedule (lights on at 7:00 pm). Food and water were provided ad libitum. Behavioral testing occurred during the dark phase between 7:00 am and 5:00 pm. Animal testing was conducted in accord with the National Institutes of Health laboratory animal care guidelines and with the University Committee on Use and Care of Animals at the University of Michigan approval. The University of Michigan's Institutional Unit for Laboratory Animal Medicine (ULAM) provides veterinary care to all animals used on campus. We ensured that all animals used in this study were healthy and experienced minimal discomfort. All protocols were approved prior to experimentation. Specifically, experiments were described in protocol 10292.

Generation of Tor1a DE/+ congenic mouse strains and intercross breeding strategy
The Tor1a DE/+ mouse, with a targeted deletion of glutamic acid (DE) in the encoded protein torsinA, was generated as previously described [16]

Genotyping
Tail samples from mice were excised and boiled in 300 ml 50 mM NaOH at 95uC for 50 minutes. Denatured tails were vortexed and mixed with 30 ml 1 M Tris pH 8.0 buffer to neutralize and centrifuged for 10 minutes at maximum speed. Premixed PCR beads (PuReTaq Ready-To-Go PCR Beads, GE healthcare) were resuspended in 24 ml of primer mix at final concentration of 3.0 mg/ml and 1.0 ml of tail lysate supernatant. Primer sequences and PCR parameters for genotyping are listed in Table 1.

Sequencing genomic DNA
For sequencing of the Tor1a mouse gene, DNA was extracted and purified from mouse tail samples using Qiagen DNeasy Blood and Tissue kit per manufacturer's instructions. Sequencing was done using an ABI Model 3730 sequencer with the following primers: 59-AAC AGA GCC AAC ACT CTG G-39 (forward) and 59-TGC TGT ACA AGA TCC TCC-39.

Behavioral Studies
All mice were kept on a reverse light dark cycle. Behavioral tests were performed during the dark period when animals were most alert. Independent cohorts were used for the baseline open-field and drug challenge open-field studies. Only male mice were used for behavioral studies.
Open-field test. The open-field test was used to assess animals' locomotor activity. Animals (n = 9-15 for each genotype) were placed in one of the five square open field boxes 43643 cm 2 with two sets of 16 pulse-modulated infrared photobeams (MED Associates) that records the animal's location and path (horizontal activity), as well as the number of rears (vertical activity) located inside sound-attenuating cabinets with fans. Illumination of the test room was the same as the mouse colony room. Mice were examined at 6 months, 9 months and 12 months of age. For baseline experiments, mice were placed in the open-field chamber for 60 minutes. Data were analyzed as distance traveled (cm) and rearing in 5-minute bins over time.
Rotarod. Rotarod (Ugo Basile, model 47600) was used to assess the mouse's ability to maintain balance and coordination (n = 10-15 for each genotype). The apparatus consists of five 3 cm diameter drums with six flanges dividing the drum, accommodating up to five mice. Mice were placed in one of the five allocated slots on the rotarod and latency to fall was measured. There were two components to this test -training and challenge. Rotarod training occurred over 3 consecutive days. For training, animals were placed on the rotarod as it accelerated from 4 rpm to 40 rpm over 6 minutes. The trial ended when the mice either fell off the rod or 400 seconds elapsed. Four trials were performed on each of the three days. On the testing day, mice were placed on the rotarod at 3 different fixed speeds, 4 trials per speed, for a total of 12 trials. On this day, the trial ended when the mouse fell off the rod or 300 seconds elapsed.
Balance Beam. Mice were trained to cross a square 80 cm long sanded plastic beam 5 mm wide, which was elevated 50 cm above base level (n = 9-14 for each genotype). At the start of each trial, mice were placed on clear open platform. A dark box at the opposite end of the beam provided motivation for the mouse to cross the beam. Traversal time and number of foot slips were measured as mice traversed the beam. Mice were tested daily with two trials on four consecutive days. The protocol was adapted from Shokkattai and colleagues [20].

Drug Treatment
Three drugs dissolved in saline were administered by i.p injection -scopolamine, GBR1290 and quinpirole. All drugs were purchased from Sigma. Animals' locomotor and rearing activities were monitored following a 30-minute habituation period and immediately after drug injection for the locomotor activating drugs: scopolamine (1.0 mg/kg, i.p.; n = 10-14 for each genotype receiving drug) and GBR12909 (5.0 mg/kg, i.p.; n = 5-8 for each genotype receiving drug and 4-7 for each genotype receiving vehicle). For the locomotor-depressing quinpirole (0.1 mg/kg, i.p.; n = 10-11 for each genotype receiving drug and 6-7 for each genotype receiving vehicle), there was no habituation period prior to injection and activity was monitored immediately after the challenge.

Data Analysis
The effects of strain on lifespan of Tor1a DE/DE mice were analyzed with survival curves using the Kaplan and Meier method created with GraphPad Prism 4.0. Two or more survival curves were compared using the logrank test, and Chi-square test was used to generate p values.
Behavioral data were subject to tests for homogeneity (Leveine test) and normal distribution (Kolmogorov-Smirnov test). All data were analyzed with the Student t-test, repeated measure ANOVA, or two-way ANOVA. Nested repeated measure ANOVA was used when data were collected in multiple trials in more than one session. For repeated measure ANOVA, all data were also subject to Mauchly's sphericity test, and F-ratios adjusted if violation occurred. All statistical analyses were performed using SPSS 19.0. A critical value for significance of p,0.05 was used throughout the study. Data are plotted 6 SEM.

Results
Genetic background modifies the phenotype of mutant torsinA (Tor1a DE/DE ) mice To test for the presence of genetic modifiers of the DE-torsinA phenotype, we pursued a two-step strategy. We first intercrossed the 129-Tor1a DE/+ mice with B6, D2 or CD1 wildtype mice. These strains were chosen because they are genetically dissimilar from each other and from 129 [21]. We then intercrossed the F1 129/ ''X'' heterozygous offspring and assessed the phenotype of the homozygous F2 Tor1a DE/DE offspring. On average, these offspring will be 50% 129 and 50% novel strain. All strains yielded F2 progeny in the expected Mendelian ratios, indicating that D2 or CD1-ICR genes do not alter the ability of Tor1a DE/DE animals to survive through gestation and birth. The resulting litters were observed twice daily during the first two postnatal (P) days and once daily thereafter to determine the duration of survival. All pups were genotyped at death or between P2 and P3 and any remaining Tor1a DE/DE animals were monitored. Tor1a DE/DE mice on 129 (n = 21), 129/B6 (n = 26), 129/D2 (n = 53) and 129/CD1 (n = 20) backgrounds were generated and closely observed.
We find that each of the three background strains has a distinct effect on the DE-torsinA phenotype. 129/D2 Tor1a DE/DE mice live significantly longer than 129-Tor1a DE/DE mice. The D2 background significantly increases median survival to 3.5 days, compared to 1.5 days on 129 background (x 2 (1) = 14.60; p,0.0001 (129 vs 129/D2); Figure 1A). The longest-surviving 129/D2 Tor1a DE/DE animal lived for 21 days, and 13.2% live longer than 7 days. Mice that live through the end of the first postnatal week develop abnormal motor behavior, exhibiting tremor and prolonged twisting movements during gait, particularly of the hindlimbs ( Figure 1C; Video S1). To test whether we could further enhance the lifespan of 129/D2 Tor1a DE/DE mice, we continued to backcross the DE mutation for 10 generations onto the D2 background. However, we find that F10 D2?Tor1a DE/DE mice do not live significantly longer than mice on the 129 background (x 2 (1) = 0.62; p = 0.43). In contrast to the D2 background, the B6 background significantly decreases median lifespan to 0.5 days (x 2 (1) = 16.31; p,0.0001 (129 vs. 129/B6)). The CD1 background also significantly alters lifespan, increasing the median survival from 1.5 days (pure 129) to 2.5 days (x 2 (1) = 5.207; p,0.05 (129 vs CD1/ 129). Unlike D2 background effect, however, all Tor1a DE/DE animals die by P3.5. These data are consistent with existence of genetic factors that modify the DE-torsinA related phenotypes ( Figure 1A).

Gene sequencing for polymorphism (Aspartic acid/ Histidine 217)
The only genetic factor linked to the penetrance of the DE-torsinA phenotype in humans is a coding polymorphism of torsinA itself. Penetrance for DE-torsinA gene carriers whose wild type torsinA contains a histidine (H) at position 216 is 3% compared to 35% for the more common aspartic acid (D) at the same position [22,23]. However, this polymorphism cannot account for the effects of the D2 or B6 background, since in congenic Tor1a DE/DE mice, both Tor1a-DE alleles derive from the original 129 ES cell line used for gene targeting. It could however, account for any behavioral differences that may be observed in the heterozygous disease state Tor1a DE/+ mice. Therefore, we sequenced the Tor1a allele from 129, B6 and D2 mice to determine whether there were differences at this position (217 in the mouse protein) that would inform our choice of background strain for modeling the disease in heterozygous mice. We find that all three strains carry an aspartic acid (D), preventing us from exploiting this polymorphism in this context ( Figure 1B).

B6?Tor1a DE/+ (DYT1) mice do not have baseline motor abnormalities
The contribution of genes from the B6 strain of mice significantly shortened lifespan in the resulting F2 129/B6 Tor1a DE/DE mice (compared to the pure 129 background). Therefore, we did all behavioral analysis on congenic B6?Tor1a DE/DE mice with the rationale that this more susceptible strain might reveal motor abnormalities we did not observe previously in heterozygous Tor1a DE/+ mice on the 129 background (there are no significant differences in open field or rotarod between 129-Tor1a DE/+ and 129-Tor1a +/+ mice at 6, 9, and 12 months of age -n = 16 WT and 14 mutants; data not shown). While we realize that the decreased lifespan may be the result of an interaction between 129 and B6 genes, we chose to backcross the 129 Tor1a DE/+ heterozygous mice to the B6 background to reduce any background noise that may occur on a mixed background. Furthermore, for all behavioral testing we only used adult mice greater than 3 months of age to ensure that any developmentally dependent phenotype would be fully manifest.
Open field. To test motor activity in B6?Tor1a DE/+ mice, we placed naïve male mice in the open field and measured horizontal and rearing locomotor activity over 60 minutes. This assessment was performed at 6, 9 and 12 months of age (n = 10 Tor1a +/+ and 9 Tor1a DE/+ at 6 months; n = 15 Tor1a +/+ and 10 Tor1a DE/+ mice at 9 and 12 months of age). No significant differences were observed between B6?Tor1a DE/+ and B6?Tor1a +/+ mice at any of the ages tested. All animals habituated to the open field at the same rate and performed similarly as assessed by total distance traveled and total rearing (Figure 2A-I).
Rotarod. To assess balance and coordination we performed rotarod testing at 12 months of age (n = 15 Tor1a +/+ and 10 Tor1a DE/+ mice). The ability to perform this task is measured by assessing the latency to fall from the rotarod (with longer latency representing improved performance). During the three-day training component of the task (2 trials per day), both groups of mice exhibited significant improvement on the accelerating rotarod, but no difference was found between the two groups, indicating that both genotypes learned the task at the same rate. Similarly, there was no significant difference in performance on the ''testing day'' when the animals are tested at the fixed speeds of 40 rpm, 35 rpm and 30 rpm (4 trials per speed). These data demonstrate that in the rotarod task, the DE mutation does not significantly impair motor learning or gross motor skills such as balance and coordination ( Figure 2J-K).
Beam-walking test. To further assess fine motor behavior and balance in the B6?Tor1a DE/+ mice, we used the beam-walking paradigm. A separate cohort of seven-month old B6?Tor1a naïve mice were trained to traverse a 5 mm square plexiglass beam on two trials for three consecutive days, and latency to cross was measured (with longer latency indicating impaired performance). Latency to cross was measured for the training and testing days, and on the fourth ''testing day'' we also quantified the number of foot-slips. We noted a significant improvement in the time it took for mice to traverse the beam (one-way rm-ANOVA, main effect of day: F[2.83, 63] = 10.12, p = 0.00) however, no significant difference was observed for latency between wild type (n = 14) and mutant mice (n = 9) during training or on the testing day. Similar to the findings for latency, no significant difference was observed between wild type and mutant mice for foot-slips ( Figure 2L-M).

B6?Tor1a DE/+ (DYT1) mice do not display altered responses to drug challenges
Anti-muscarinic drugs can be effective in treating the symptoms of DYT1 dystonia and DYT1-torsinA transgenic mice are reported to show an abnormal interplay between the dopaminergic and cholinergic systems in electrophysiological studies [24,25,26]. To probe these neurochemical systems, we challenged Tor1a +/+ and Tor1a DE/+ mice with the muscarinic receptor antagonist, scopolamine, and the dopamine transporter reuptake inhibitor, GBR12909, measuring the behavioral response to these drugs in the open field.

Discussion
Our study is the first to investigate the effects of genetic background on the phenotype of torsinA mutant mice, a feature of particular interest because of the reduced penetrance of the disease, some of which may relate to genetic modifiers in the human population. We identified background strains that are able to suppress (D2 and CD1) or enhance (B6) the lethality of 129-Tor1a DE/DE mice. These genetic backgrounds produced survival times ranging from less than 12 hours to up to 3 weeks, depending on the parental strains used for the F2 intercross ( Figure 1). Future mapping of the genes responsible for these effects may provide insight into the torsinA pathway, which remains poorly understood. Alternatively, it is possible that these variants alter lifespan independently of the torsinA pathway, for example by making the pups more (or less) able to withstand the effects of torsinA dysfunction.
A barrier to progress in dystonia research is the lack of an animal model with overt abnormal movements, and a key aim of our experiments was to identify background strains that might enable us to develop such a model. Strikingly, the long-lived D2?Tor1a DE/DE pups exhibited a noticeable tremor, abnormal limb placement and limb weakness, and a delayed righting reflex (Video S1). While this abnormal motor behavior likely results from torsinA-related neural dysfunction, these pups do not feed well and appear generally ill, a confounding factor that complicates the interpretation of this phenotype. This finding was nevertheless encouraging, and we tried to build on it to create healthy mice that display abnormal motor function (like the disease). Since an increase from one to two mutated torsinA alleles causes early lethality in the majority of mice, we tried to temper this phenotype by further backcrossing to the apparently more permissive D2 background and analyzing D2?Tor1a DE/DE mice. The variability and enhanced lifespan diminished when Tor1a DE/+ animals were backcrossed more than 10 generations to a congenic D2 background. In fact, the lifespan of Tor1a DE/DE on the D2 background was indistinguishable from those on the original 129 background. These findings indicate that the extended lifespan of F2 129/D2 mice was due to a combination of homozygous and heterozygous allelic effects. The debilitated nature of these mice precluded further study of these animals.
We find that 129-Tor1a DE/+ mice have no apparent behavioral abnormalities. Because the B6 background significantly decreased the survival of torsinA mutant mice, we explored whether this background would enable us to detect phenotypic effects of the DE mutation in the heterozygous animals. We generated congenic B6?Tor1a DE/+ mice and tested them in several behavior paradigms: open field (with and without pharmacological challenge), rotarod, and balance beam. Similar to the 129S6/SvEvTac background, however, there were no significant differences between Tor1a DE/+ and Tor1a +/+ on the B6 background.
A previous study of Tor1a DE/+ mice on a mixed 129/B6 background reported hyperactivity (significantly increased distance traveled) in a 10-minute open field test and normal rearing activity [15]. Interestingly, we find a trend in the opposite direction, with B6? Tor1a   behavior on the rotarod and in the beam walking test. They report that wild type and mutant mice perform equally well on the rotarod, however mutant mice show significantly more footslips than controls on the balance beam test. There was no difference in latency to cross the beam, however. There were some differences in the execution of experiments between our studies presented here and those of Dang et al. First, we examined spontaneous locomotor activity in the open field in 5-minute bins for 60 minutes, while Dang et al. examined only the first 10 minutes. Second, Dang et al. examined balance and coordination with several different sized beams on the beam walking test (both square and round beams ranging in size from 17 mm diameter/ width to 7 mm width, while we trained and tested mice on a more difficult 5 mm width square beam. These data, and our finding of normal behavior in the open field and beam walking suggest that DE-torsinA may cause subtle behavioral abnormalities. However, it is also possible that the confounding effects of mixed background are responsible for the subtle abnormalities identified by Dang et al. There are limitations of genetic backcrossing that may be relevant to studies of torsinA. Although nearly all loci become homogenous by approximately the tenth generation of backcrossing, the mutated allele and closely linked flanking sequence from the original background persist. With each successive backcross the flanking sequence surrounding the gene of interest shortens, but frequently several MB of the original background remain. TorsinB (a close homolog of torsinA) is located adjacent to torsinA, so all of the backgrounds tested almost certainly carry the original 129-torsinB allele. A previous study demonstrated that torsinA and torsinB share redundant functions in multiple cell types, raising the possibility that torsinB influences disease penetrance [9]. This may explain, at least in part, the absence of a behavioral phenotype in Tor1a DE/+ mice.
Similar considerations pertain to the one reported genetic modifier of disease penetrance in DYT1 dystonia. A nonsynonymous SNP in the coding sequence for residue 216 encodes aspartic acid (D) in 88% and histidine (H) in 12% of control population alleles, and the D216H allele is reported to significantly reduce disease penetrance when present in the normal allele in trans to the mutated allele [23,27]. We found the ''D'' allele at the analogous murine residue (217) in all of the strains used in this study, further demonstrating the highly conserved features of the Tor1a gene and indicating that this SNP does not affect penetrance in Tor1a DE/+ mice. Several other SNPs are present in the 59 and 39 UTR regions of TOR1A, as well as a single-base-pair deletion in the 39 UTR (G-del), but these features are not known to be associated with the penetrance of DYT1 dystonia. On the other hand, several SNPs in the 39 UTR may influence the onset and propensity to spread in adult-onset primary dystonia [28,29].
Etiological animal models (i.e., those based on known causes of human disease) may offer insight into disease pathogenesis even if they do not replicate the outward symptoms of the disease. In fact, diffusion tensor imaging (DTI) of B6?Tor1a DE/+ mice demonstrates microstructural abnormalities in the cerebellothalamocortical and thalamocortical tracts [22] similar to those observed in nonpenetrant DYT1 carriers [6,30,31] indicating that these mice model non-manifesting carriers. Thus, while genetic modifiers seem likely to account for at least part of variable penetrance and expressivity of the DE-Tor1a allele, additional factors may be required to convert Tor1a DE/+ mice (or patients) from nonmanifesting to overt disease. For example, stress exposure or excessive motor activity (e.g., prolonged wheel running) may be required ''second hits''. Future studies aimed at identifying such factors will therefore be required to generate torsinA mutant mice that exhibit abnormal movements, a critically needed reagent if we are to use animal models to dissect the neurobiological substrates of dystonic movements.