Dystonia is a neurological disorder characterized by sustained or intermittent muscle contractions that result in abnormal, often repetitive movements, postures, or both [1]. DYT1 or DYT-TOR1A dystonia is the most common early-onset generalized dystonia [2]. Most patients carry a heterozygous three-nucleotide (ΔGAG) deletion in exon 5 of the DYT1/TOR1A gene, leading to the loss of a glutamate residue in the torsinA protein [3]. Although numerous studies suggest that this mutation results in a loss-of-function of torsinA, a toxic-gain-of-function effect cannot be ruled out [4, 5]. Nevertheless, the precise mechanism by which the mutation leads to dystonia remains largely unknown. Dyt1 knockin mice harboring the corresponding in-frame ΔGAG deletion in the endogenous Dyt1/Tor1a gene have been developed to mimic the human condition [6, 7]. Although these mice exhibit several dystonia-like phenotypes, such as motor and sensory deficits, abnormal gait, and hind limb muscle co-contraction [7–10], the lack of overt dystonia phenotypes in most genetically modified models has hindered progress in understanding the disease [11–17].

The cerebellum—and specifically its Purkinje cells—plays a key role in controlling movement and posture, with multiple animal studies implicating these cells in dystonia pathogenesis [15, 18–29]. Dyt1 knockin mice show altered Purkinje cell morphology [30, 31] and abnormal firing patterns, with increased large-conductance calcium-activated potassium (BK) current and elevated BK channel protein levels [32]. Most importantly, acute knockdown of torsinA in the adult cerebellum using shRNAs induces overt dystonia in adult mice—an effect not observed when knockdown is performed during development or directed to the striatum [33], suggesting the likely involvement of cerebellar dysfunction in the pathogenesis of DYT1 and other dystonias.

The present study aims to validate cerebellar acute torsinA knockdown models of DYT1 dystonia. Small hairpin RNAs (shRNAs) or small interfering RNAs (siRNAs) are known to cause significant off-target effects that may alter the expression of unintended genes [34]. Here, we developed an alternate acute torsinA knockdown (AAV-cre KD) mouse model by bilateral stereotaxic injection of AAV5-CMV-cre-GFP into the cerebellum of Dyt1 ^loxP/loxP mice, with wild-type (WT) mice receiving AAV5-CMV-GFP as controls. Cre-mediated recombination subsequently reduced torsinA expression in the cerebellum. The motor behavior of these mice and the electrophysiological properties of the cerebellar Purkinje cells were characterized.

The strategies used to knock down torsinA in the cerebellum are shown in Figure 1A. Instead of shRNA targeting torsinA mRNA, we took advantage of the cre-loxP system and the floxed Dyt1 locus we developed earlier [35]. We used an AAV virus expressing Cre and GFP under the control of a general promoter, CMV. GFP was included to assess the efficiency of AAV transduction. Under UV light, strong fluorescent signals were observed in the cerebellum of AAV-injected mice (Figure 1B), confirming widespread AAV transduction. The effect of Cre-mediated recombination was assessed indirectly through a Western blot analysis of torsinA protein levels. Western blot showed a significant decrease in torsinA expression level in the cerebellar lysates of AAV-transduced animals compared to control mice (Figure 1C). In contrast, the levels in the cerebral cortex, brainstem, and striatum were unaffected (Supplementary Figure S1), suggesting successful inactivation of the endogenous Dyt1 gene specifically in the cerebellum.

We observed the home cage behavior to determine whether the knockdown of torsinA resulted in overt dystonia (Supplementary Video S1). The presence of dystonia and its severity were assessed using a previously published Dystonia scale [33, 39]. Briefly, 0 = normal behavior; 1 = abnormal motor behavior, no dystonic postures; 2 = mild motor impairment, dystonic-like postures when disturbed; 3 = moderate impairment, frequent spontaneous dystonic postures; 4 = severe impairment, sustained dystonic postures. The KD mice showed overt dystonia in five out of seven mice within 9 weeks, as evidenced by abnormal hindlimb postures (Figure 2A) and increased overall dystonia scores in the KD mice (χ² = 7.55, p = 0.006, Figure 2B). In addition to these general dystonic postures, a distinct and dramatic phenotype emerged in three of the five dystonic mice. This behavior was most clearly observed and captured on video during the later stages of the experiment, as the animals’ general dystonia scores reached their peak severity. These mice exhibited recurrent episodes of vigorous, bidirectional spinning, particularly upon tail suspension (Supplementary Video S2). To provide a rigorous analysis, we quantified this behavior from the video recordings. The onset of spinning was remarkably rapid, with a latency of less than one second. The behavior was forceful and sustained; in the two most severely affected animals, the rotational speed reached up to 5.4 rotations per second (RPS), and the spinning bouts consistently lasted for more than 10 s. The third affected mouse displayed a similar but milder spinning phenotype. None of the WT control mice injected with an AAV5-CMV-GFP virus exhibited these phenotypes (Supplementary Videos S3, S4).

To contextualize these findings, we statistically compared the penetrance of overt dystonia in our AAV-cre KD model (71.4%, 5/7 mice) with established models. This penetrance rate is significantly higher than that reported in constitutive Dyt1 ΔGAG knock-in mice [6, 7], which typically exhibit no overt dystonia (0%; p < 0.01, Fisher’s exact test). Furthermore, the high penetrance in our model is comparable to that observed in acute cerebellar shRNA knockdown models (p > 0.05) [33], supporting the validity of acute cerebellar suppression of torsinA in recapitulating motor symptoms.

To determine the effect of torsinA KD on Purkinje cell firing, we measured spontaneous firing and intrinsic excitability of the Purkinje cell in acute brain slices. Cell-attached recording of Purkinje cells revealed a 27% decrease in spontaneous firing frequency (control: 19 cells/3 mice; KD: 29 cells/5 mice; p = 0.04; statistical power = 1.0; Figures 3A,B) while coefficient of variation (CV) was not significantly altered (p = 0.55; Figure 3C). Decreased spontaneous firing in KD mice is consistent with our earlier finding in Dyt1 knockin mice [32].

Depolarizing current steps were injected into the Purkinje cell of KD and control mice to assess intrinsic excitability. We injected a brief, positive current into the cell, which causes the membrane potential to become more positive, or depolarized. The action potentials produced are indicative of intrinsic excitability. KD mice exhibited 27% fewer action potentials than controls (p = 0.0004, Figures 3D–G). Decreased intrinsic excitability indicated that the KD Purkinje cells would be less likely to fire action potentials, which could explain the decreased spontaneous firing frequency.

We developed an alternate torsinA KD model to validate whether cerebellar torsinA KD can be a robust dystonia phenotypic model. Using the cre-loxP system-based approach, we observed overt dystonia in 5 out of 7 mice, similar to the shRNA KD model [33]. None of the 8 WT mice injected with AAV5-CMV-GFP showed overt dystonia, supporting that these symptoms are specific to torsinA knockdown rather than artifacts of the viral vector or surgery. Dyt1 knockin mice show altered Purkinje cell morphology [30, 31] and abnormal spontaneous firing, with increased BK current and elevated BK channel protein levels [32]. BK channels have been implicated in dystonia [46–49]. Together, these results support a key role of the cerebellum in DYT1 dystonia. However, 3 of the 5 dystonic AAV-cre KD mice showed circling or spinning phenotype, which is not reported in the shRNA KD mice [33] and question the validity of cerebellar torsinA KD mice to model DYT1 dystonia.

Our KD mice and the previous shRNA KD mice show overt dystonia, supporting a key role of the cerebellum in DYT1 pathogenesis. In contrast, models with conditional torsinA knockout (pKO) or mutant knockin (Pcp2-KI) in Purkinje cells show better motor performance [50, 51]. The discrepancy may result from developmental compensation, as shRNA-mediated torsinA KD in young mice does not lead to overt dystonia [33]—or from the involvement of granule cells and deep cerebellum nuclei, which are affected in the KD models but spared in pKO and Pcp2-KI mice. Notably, during development, conditional knockin of ΔGAG mutation in the mouse cerebellum with engrailed1-Cre (En1 Cre) does not produce overt dystonia [4]. Future studies should use En1-creERT1 [52] or Pcp2-creERT2 [53] lines to inducibly knock out the Dyt1 gene in adult mice to resolve these discrepancies.

Although both AAV-cre KD mice and shRNA KD mice support an essential role for the cerebellum in DYT1 pathogenesis, it remains unclear whether dystonia originates in the cerebellum. For example, Pcp2-KI mice showed normal Purkinje cell firing [41] in contrast to the global Dyt1 knockin mice [32], suggesting that the cerebellum may act as a downstream node rather than the origin of dystonia. The Purkinje cell abnormality in Dyt1 KI mice [32] likely originated from the striatum [41], and the overt dystonia in acute KD mice supports the idea that the cerebellum functions as a node downstream of the striatum via a disynaptic pathway linking these two structures [54]. Remarkably, conditional knockout of torsinA in the striatum mimics beam walking deficits in global Dyt1 knockin mice [7, 36, 38], whereas acute shRNA KD in the adult striatum and globus pallidus does not produce overt dystonia [33]. The exact role of the cerebellum and basal ganglia in DYT1 dystonia remains to be fully delineated.

We found significantly reduced spontaneous firing in vitro in the AAV-cre KD mice, which is consistent with the in vivo finding from shRNA KD mice [33]. We examined further and found significantly decreased intrinsic excitability of AAV-cre KD Purkinje cells. Reduced intrinsic excitability would likely decrease the spontaneous firing of Purkinje cells both in vivo and in vitro. The molecular basis of reduced intrinsic excitability remains unknown, but we reported previously that BK channel activity was significantly increased [32], which could contribute to the decreased intrinsic excitability. Increased BK channel activity allows more potassium ions to move out of the cell, making the Purkinje cells less likely to fire. The AAV-cre KD mice are expected to reduce torsinA protein levels in other cerebellar neurons, including deep cerebellar nuclei, granule cells, and other interneurons. The effect of torsinA KD in these neurons remains to be investigated. Interestingly, aberrant outputs of glutamatergic neurons in deep cerebellar nuclei appear to mediate dystonic movement in the Prrt2 mouse model [55], suggesting the importance of deep cerebellar nuclear neurons and their modulation by Purkinje cells.

We detected an unexpected bidirectional circling and spinning behavior during tail suspension in severely affected mice, resembling spinning behavior observed in a shRNA KD model targeting Sgce mRNA in the cerebellum, where motor deficits, spinning, and myoclonic-like jerky movements were alleviated by alcohol consumption [56]. The mutation of the SGCE gene causes myoclonus-dystonia or DYT11 dystonia [57]. The spinning behavior is unique to Sgce shRNA KD mice compared to other models of dystonia, including torsinA shRNA KD mice [56]. Although there is no spinning symptom in DYT11 dystonia patients, the alcohol rescue in the Sgce KD mice suggests that the spinning during tail suspension may be related to myoclonus-dystonia manifestation in mice, since DYT11 patients can get temporary relief by alcohol ingestion [58–60].

It is interesting that both Sgce shRNA KD and AAV-cre KD mice showed spinning behavior. ε-sarcoglycan, the product of the Sgce gene, is known to interact with torsinA. TorsinA binds to mutant ε-sarcoglycans and promotes their degradation [61]. Mutant mice with double mutations in the torsinA gene and Sgce show an earlier onset of motor deficits [62]. The common pathway affected in Sgce shRNA KD and AAV-cre KD mice remains to be determined.

On the other hand, vertigo, dizziness, and imbalance are associated with lesions of the vestibulo-cerebellar, vestibulo-spinal, or cerebellar ocular motor systems [63]. Acute knockdown of torsinA and Sgce might have damaged some of these systems, leading to the spinning behaviors in these mice. In the torsinA and Sgce KD mice without spinning behavior, it is plausible that vestibulo-cerebellar and cerebellar ocular motor system impairments exist, which may contribute to imbalance or even overt dystonia in torsinA and Sgce KD mice. Future research should explore localized or refined cerebellar AAV manipulations to avoid perturbations to the vestibulo-cerebellar and cerebellar ocular motor systems. It will also be interesting to study how vestibulo-cerebellar and cerebellar ocular motor system dysfunction contributes to dystonia pathogenesis and treatment. Vestibular dysfunction has been linked to focal or cervical dystonia [64–68]. A patient develops cervical dystonia soon after ear surgery that causes vestibular hypofunction [69]. Another report detailed an idiopathic cervical dystonia patient with benign paroxysmal positional vertigo [70]. To our knowledge, there are no reports of vestibular dysfunction in DYT1 and DYT11 dystonias, therefore, the acute torsinA and Sgce KD models may not accurately represent DYT1 and DYT11 dystonias. This is the major weakness of the current study and the use of acute gene KD models in dystonia research in general. Additionally, while our analysis of the spinning phenotype provided valuable data on its severity, a prospective study designed to systematically track the emergence and progression of this specific behavior from earlier time points would be necessary to understand its relationship with the overall dystonia pathology fully. Additional limitations include the fact that acute KD of torsinA in adults does not occur in DYT1 patients, and recent studies have pointed to a developmental origin of DYT1 dystonia [71–73]. This underscores the need for transgenic models with developmental genetic manipulations to replicate the human condition more accurately.