Sensory-motor circuit is a therapeutic target for dystonia musculorum mice, a model of hereditary sensory and autonomic neuropathy 6

Abstract

Mutations in Dystonin (DST), which encodes cytoskeletal linker proteins, cause hereditary sensory and autonomic neuropathy 6 (HSAN-VI) in humans and the dystonia musculorum (dt) phenotype in mice; however, the neuronal circuit underlying the HSAN-VI and dt phenotype is unresolved. dt mice exhibit dystonic movements accompanied by the simultaneous contraction of agonist and antagonist muscles and postnatal lethality. Here, we identified the sensory-motor circuit as a major causative neural circuit using a gene trap system that enables neural circuit-selective inactivation and restoration of Dst by Cre-mediated recombination. Sensory neuron–selective Dst deletion led to motor impairment, degeneration of proprioceptive sensory neurons, and disruption of the sensory-motor circuit. Restoration of Dst expression in sensory neurons using Cre driver mice or a single postnatal injection of Cre-expressing adeno-associated virus ameliorated sensory degeneration and improved abnormal movements. These findings demonstrate that the sensory-motor circuit is involved in the movement disorders in dt mice and that the sensory circuit is a therapeutic target for HSAN-VI.

Multipurpose gene trap system allowed neural circuit-selective gene switch and identified a therapeutic target for HSAN-VI.

INTRODUCTION

Hereditary neurological diseases usually involve complex abnormalities in neuronal networks, and it is challenging to determine the neuronal circuits that are affected by disease pathogenesis and that have potential as therapeutic targets. Hereditary sensory and autonomic neuropathies (HSANs) are clinically and genetically heterogeneous disorders of the peripheral nervous system (PNS) that are characterized by progressive degeneration of sensory and autonomic neurons (1). Mutations in the Dystonin (DST) gene, which is also known as bullous pemphigoid antigen 1 (BPAG1), cause hereditary sensory and autonomic neuropathy type VI (HSAN-VI). DST generates tissue-selective isoforms, DST-a, DST-b, and DST-e (2, 3) that are expressed in neural, muscular, and cutaneous tissues, respectively. DST-a and DST-b are cytoskeletal linker proteins that belong to the plakin family (4, 5). Distinct mutations in the DST gene result in a spectrum of symptoms with variations in severity, characteristics, and age of onset, possibly resulting from distinct patterns of DST isoform deficiency (6). Although the disease phenotype of HSAN-VI is heterogeneous, sensory and autonomic abnormalities have been commonly described, indicating impairment of the PNS to be crucial for the manifestation of HSAN-VI (7–12).

Dystonia musculorum (dt) mice arose spontaneously and exhibit sensory neuron degeneration in dorsal root ganglia (DRG) at an early postnatal stage and progressive movement disorders, such as ataxia and dystonic movement (13). dt mice have a loss-of-function mutation in the Dst gene (14, 15); therefore, they have been used to investigate the pathogenic mechanisms of HSAN-VI and to develop approaches to treat the disease (16, 17). In addition to neurodegeneration in the PNS, dt mice also exhibit abnormalities in the central nervous system (CNS): neurofilament accumulation and neurodegeneration in the spinal cord and brainstem (17–19) and reduced proliferation of oligodendrocyte progenitor cells (20). Therefore, it is not clear to what extent neurodegeneration in the PNS contributes to aspects of the dt phenotype, such as motor deficits and postnatal lethality.

Neuronal circuit-selective gene inactivation and restoration of a single allele are promising approaches to determine the neuronal circuit(s) or cell type(s) involved in the pathogenesis of hereditary neurological diseases. However, only a limited number of such studies have been conducted (21). Here, we applied a multipurpose Dst gene trap allele that enables cell type–selective inactivation and restoration of a target gene through inverting a gene trap cassette by Flip-excision (FLEX) technology (22, 23). Using sensory neuron–selective restoration and inactivation of the Dst gene, we demonstrated that impairment of the sensory-motor circuit is responsible for the movement disorder in dt mice and that it is a potential target for treating HSAN-VI. The FLEX-mediated multipurpose gene trap system was effective for revealing the pathogenic mechanism of this genetic disease and is a valuable tool for determining causative neuronal circuits in various inherited neurological diseases.

RESULTS

Sensory neuron–selective inactivation or restoration of Dst gene

We generated Dst^Gt mice harboring the FLEX-mediated multipurpose allele whose gene trap cassette is in the Dst locus (Fig. 1A) (23). Gene trap cassette disrupts endogenous expression of Dst gene and generates a Dst-βgeo fusion protein, which can be detected by X-galactosidase (X-gal) staining. The gene trap cassette is flanked by inversely oriented heterotypic target sites for flippase (FLP) and Cre recombinases. Cre recombination leads to the irreversible switch from the functional Dst^Gt-inv allele to the Dst^Gt-DO mutant allele or from the Dst^Gt mutant allele to the functional Dst^Gt-inv allele (Fig. 1A). Dst^Gt or Dst^Gt-DO homozygotes and compound heterozygotes of Dst^Gt and Dst^Gt-DO exhibit the dt phenotype, and we refer these mice as Dst GT mice.

Fig. 1. Wnt1-lineage selective deletion and restoration of Dst expression.

(A) The multipurpose gene trap cassette locates in the intron of the Dst locus. The gene trap cassette contains splice acceptor (SA) sequence, the reporter gene βgeo, and poly-A (pA) termination signal. In this FLEX system, the gene trap cassette is flanked by pairs of inversely oriented target sites of FLP recombinase (Frt and F3: half circles) and Cre recombinase (loxP and lox5171: triangles). FLP- or Cre-mediated recombination irreversibly switches mutant Dst^Gt allele to functional Dst^Gt-inv allele and functional Dst^Gt-inv allele to mutant Dst^Gt-DO allele. (B) Mating scheme to generate Wnt1-Cre;Dst cGT and Wnt1-Cre;Dst cRescue mice. (C) X-gal staining visualized gene trap cassette inversions in 1-month-old Wnt1-Cre;Dst^Gt-inv/wt mice. βGeo (blue) was expressed in dorsal root ganglion (DRG) and sympathetic ganglion (SG), while the spinal cord (SC) lacked βgeo expression. In the brain, βgeo was expressed in the cerebellum (CB), pontine nucleus (Pn), and midbrain, while scarcely detected in the cerebral cortex (Ctx), diencephalon (Die), and medulla oblongata (MO). The dotted line indicates the edge of the SC. Boxed area is enlarged to show the DRG. DCN, deep cerebellar nucleus; SN, substantia nigra. (D) In situ hybridization in each group at 3 to 4 weeks of age. Dst mRNA was lost in DRG neurons of Wnt1-Cre;Dst cGT mice but restored in DRG neurons of Wnt1-Cre;Dst cRescue mice. (E) The number of Dst-positive neurons in the DRG [Ctrl (n = 8 mice); Wnt1-Cre;Dst cGT (n = 3); Wnt1-Cre;Dst cRescue (n = 6)]. (F) Western blotting for Dst and β-actin (Actb) in the DRG and SC of 2-week-old mice. (G) Quantification of Dst levels normalized to Actb (n = 3 mice). Scale bars, 200 μm [(C) and (D)]. Data are presented as mean ± SE. P > 0.05 [not significant (ns)] and ***P < 0.005, using analysis of variance (ANOVA) with Tukey’s test in (E) and (G). WT, wild type.

To analyze Dst functions in the PNS, brainstem, and cerebellum, we used Wnt1-Cre transgenic mice (24), in which Cre-mediated recombination occurs in neural crest derivatives, including sensory neurons, sympathetic ganglionic neurons, and Schwann cells, and in the developing midbrain-hindbrain (Table 1) (25–27). For conditional gene trap (cGT) experiments, male Wnt1-Cre;Dst^Gt-DO/wt mice were crossed with female Dst^{Gt-inv/Gt-inv} mice (Fig. 1B). A quarter of their offspring were Wnt1-Cre;Dst cGT mice (Wnt1-Cre;Dst^Gt-DO/Gt-inv), and we used the mice of the other three genotypes as littermate controls (Ctrl). For conditional rescue (cRescue) experiments, male Wnt1-Cre;Dst^Gt-DO/wt mice were crossed with female Dst^Gt/wt mice to generate Wnt1-Cre;Dst cRescue mice (Wnt1-Cre;Dst^Gt-DO/Gt; Fig. 1B). One-eighth of their offspring were Wnt1-Cre;Dst cRescue mice, one-eighth were Dst GT mice (Dst^Gt-DO/Gt, positive control with dt phenotype), and the other six genotypes had a wild-type phenotype and were used as Ctrl.

Table 1. List of Cre driver mice.

Mouse strain	Targets	MGI	Reference
Wnt1-Cre	PNS	MGI:2386570	(24)
	Sensory neurons
	Sympathetic ganglionic neurons
	Schwann cells
	CNS
	Cerebellum and midbrain
Avil-Cre	PNS	MGI:4459942	(34)
	Sensory neurons
	Sympathetic ganglionic neurons (after postnatal stage)
	CNS
	No target
En1-Cre	PNS	MGI:2446434	(38)
	No target
	CNS
	Cerebellum and midbrain
	Spinal cord interneurons

To validate sensory neuron–selective inversions of the gene trap cassette, we performed X-gal staining on sections of Wnt1-Cre;Dst^Gt-inv/wt mice because the βgeo reporter is expressed from the recombined Dst^Gt-DO allele in Wnt1-lineage cells (Fig. 1C). Most sensory neurons in the DRG and sympathetic ganglionic neurons were labeled by X-gal staining. However, positive signals were very scarce in the spinal cord. The efficiency of Cre-mediated gene trap inversions in DRG neurons was similar at the cervical, thoracic, and lumbar levels (fig. S1A). In cerebellar circuits, positive signals were detected in Purkinje cells, the deep cerebellar nucleus, and the pontine nucleus. In the midbrain, X-gal–labeled cells were also broadly distributed. In contrast to the midbrain-hindbrain, X-gal staining in the forebrain was below detectable levels, including in the cerebral cortex, striatum, and diencephalon. Dst is widely expressed in the PNS and CNS (23); therefore, these data indicated that Cre-mediated gene trap cassette inversions occurred in Wnt1-lineage cells in the PNS and midbrain-hindbrain of Wnt1-Cre;Dst^Gt-inv/wt mice.

In situ hybridization confirmed sensory neuron–selective inactivation and restoration of Dst expression in Wnt1-Cre;Dst cGT mice and Wnt1-Cre;Dst cRescue mice, respectively (Fig. 1, D and E). In Ctrl mice, Dst mRNA was detected in DRG sensory neurons and the spinal cord. Dst mRNA was selectively deleted in the DRG of Wnt1-Cre;Dst cGT mice. In contrast, Dst mRNA was restored in DRG of Wnt1-Cre;Dst cRescue mice, while the spinal cord of these mice still lacked Dst mRNA. Inactivation and restoration of Dst expression exhibited similar trends at the cervical, thoracic, and lumbar levels (fig. S1B). In the brain, Dst expression was inactivated in the deep cerebellar nucleus and midbrain of Wnt1-Cre;Dst cGT mice and restored in the same regions of Wnt1-Cre;Dst cRescue mice (fig. S2). Dst protein was also examined by Western blotting (Fig. 1, F and G). In Wnt1-Cre;Dst cGT mice, Dst protein bands were significantly diminished in the DRG but were detected in the spinal cord to the same extent as in Ctrl mice. Conversely, Dst protein was significantly restored in the DRG of Wnt1-Cre;Dst cRescue mice compared with Dst GT mice but remained diminished in the spinal cord of these mice. Therefore, Dst mRNA and Dst protein in sensory neurons were deleted and restored in Wnt1-Cre;Dst cGT mice and Wnt1-Cre;Dst cRescue mice, respectively.

Phenotypic characterization of Wnt1-Cre;Dst cGT mice and Wnt1-Cre;Dst cRescue mice

Next, we characterized phenotypes of Wnt1-Cre;Dst cGT mice and Wnt1-Cre;Dst cRescue mice. Almost all Dst GT mice die until approximately 1 month of age. The life span of Wnt1-Cre;Dst cRescue mice significantly increased compared with that of Dst GT mice. A total of 68% of Wnt1-Cre;Dst cRescue mice survived to more than 100 days, 32% survived to 200 days, and the longest survived to 18 months. Wnt1-Cre;Dst cGT mice had a shorter life span than Ctrl mice, and 79% of them died by 200 days of age (Fig. 2A). At 3 weeks of age, body weight gain stopped in Dst GT mice. Wnt1-Cre;Dst cGT mice also had significantly smaller body sizes than Ctrl mice (Fig. 2B). Body weight gain was also slower in Wnt1-Cre;Dst cRescue mice compared with Ctrl mice, and their body size was indistinguishable from that of Dst GT mice at 3 weeks of age. This trend of low weight continued with age. These data indicate that loss of Dst functions in Wnt1-lineage cells is responsible for postnatal lethality and growth arrest in dt mice. Differences in the phenotypes of individual cGT or cRescue mice may result from variation in the efficiency and timing of Cre recombination during development. Impaired motor performance and dystonic postures are characteristic phenotypes of dt mice (17); therefore, we assessed motor performance using the rotarod test (Fig. 2C). Wnt1-Cre;Dst cGT mice showed significantly impaired motor performance compared with Ctrl mice as assessed by their ability to stay on an accelerating rotarod. Conversely, Wnt1-Cre;Dst cRescue mice showed significantly superior motor performance compared with Dst GT mice but did not reach the level of Ctrl mice. The motor impairment in Wnt1-Cre;Dst cGT mice and phenotypic rescue in Wnt1-Cre;Dst cRescue mice continued in mice over 3 months of age (fig. S3A). Next, we assessed body balance using a balance beam test. Normal Ctrl mice walked smoothly along the beam; however, Wnt1-Cre;Dst cGT mice could not kept their balance and immediately fell (movie S1). In sharp contrast, Wnt1-Cre;Dst cRescue mice walked smoothly along the beam and reached the goal (movie S1). We also assessed uncoordinated limb movements and dystonic postures in the tail suspension test. During tail suspension, Wnt1-Cre;Dst cGT mice showed hindlimb clasping, abnormal extension of both forelimbs and hindlimbs, and sometimes twisting of the trunk, which appeared approximately 10 days after birth (Fig. 2D). In contrast, Wnt1-Cre;Dst cRescue mice showed normal posture during tail suspension (Fig. 2D) and rescued from dystonic movement and walked in the cage normally, compared with Dst GT mice (movie S2). As a sensory test, the plantar test was performed to evaluate the thermal sensitivity of mice (fig. S3B). Thermal nociceptive function was impaired in Wnt1-Cre;Dst cGT mice, and it was rescued in Wnt1-Cre;Dst cRescue mice.

Fig. 2. Phenotypic characterization of Wnt1-Cre;Dst cGT mice and Wnt1-Cre;Dst cRescue mice.

(A) Survival curve indicates a shorter life span of Wnt1-Cre;Dst cGT mice (yellow dashed line, n = 18 mice) than Ctrl mice (black solid line, n = 36). Wnt1-Cre;Dst cRescue mice (blue solid line, n = 19) showed a longer life span than Dst GT mice (red dotted line, n = 16). (B) Body weight (grams) of male mice at 3 weeks of age [Ctrl (n = 23 mice); Wnt1-Cre;Dst cGT (n = 7); Dst GT (n = 16); Wnt1-Cre;Dst cRescue (n = 9)]. Wnt1-Cre;Dst cGT mice showed significant decrease in body weight compared with Ctrl mice. Body weight of Wnt1-Cre;Dst cRescue mice was also light compared with Ctrl mice and not significantly different from Dst GT mice. (C) In the rotarod test, latency to fall (seconds) was measured to assess motor performance in each group at 3 to 4 weeks of age [Ctrl (n = 28 mice); Wnt1-Cre;Dst cGT (n = 8); Dst GT (n = 4); Wnt1-Cre;Dst cRescue (n = 16)]. The motor performance of Wnt1-Cre;Dst cGT mice was more impaired than Ctrl mice. The motor performance of Wnt1-Cre;Dst cRescue mice was improved than Dst GT mice. (D) Tail suspension test in each mouse. Ctrl mice maintained a normal posture during tail suspension. Wnt1-Cre;Dst cGT mice displayed abnormal postures such as hyperextended and clasped hindlimbs and truncal twists similar to Dst GT mice. On the other hand, Wnt1-Cre;Dst cRescue mice maintained a normal posture similarly to Ctrl mice. Data are presented as mean ± SE. P > 0.05 (ns), *P < 0.05, and ***P < 0.005, using ANOVA with Tukey’s test in (B) and (C).

Electromyography (EMG) analysis has shown that Dst mutant mice exhibit frequent co-contractions of agonist and antagonist muscles (17, 23). EMG recordings were performed from the forelimb triceps and biceps brachii muscles of Ctrl, Dst GT, Wnt1-Cre;Dst cGT, and Wnt1-Cre;Dst cRescue mice in the awake state (Fig. 3A and fig. S4, A and B). Co-contractions between the triceps and biceps brachii muscles were more frequently observed in Dst GT mice, compared with Ctrl mice and Wnt1-Cre;Dst cGT mice. In contrast, co-contractions were less frequently observed in Wnt1-Cre;Dst cRescue mice than Dst GT mice (Fig. 3A). We also confirmed these observations by cross-correlograms between the triceps and biceps brachii muscle activities (Fig. 3B). There were no peaks in Ctrl and Wnt1-Cre;Dst cRescue mice and a high peak at around 0 ms in Dst GT mice. In Wnt1-Cre;Dst cGT mice, there was a low peak at around 0 ms. Therefore, the deletion of Dst in Wnt1-lineage cells was insufficient to fully reproduce co-contractions observed in Dst GT mice. On the other hand, restoring Dst expression in Wnt1-lineage cells rescued co-contraction between the agonist and antagonist muscles, a hallmark of dystonic movements and postures.

Fig. 3. EMG analysis of simultaneous contractions in Wnt1-Cre;Dst cGT mice and Wnt1-Cre;Dst cRescue mice.

(A) Rectified EMG of the triceps (blue) and biceps (magenta) brachii muscles in Ctrl, Wnt1-Cre;Dst cGT, Dst GT, and Wnt1-Cre;Dst cRescue mice at 6 to 12 weeks of age. Co-contraction between the triceps and biceps brachii muscles (represented by black vertical lines based on Co_Tri) was frequently observed in Dst GT mice (frequency of co-contraction = 61.0%), less frequently in Wnt1-Cre;Dst cGT (39.0%) and Wnt1-Cre;Dst cRescue mice (33.5%), and least frequently in Ctrl mice (26.5%). (B) Cross-correlograms between triceps and biceps brachii muscle activity were calculated from rectified EMG of Ctrl, Wnt1-Cre;Dst cGT, Dst GT, and Wnt1-Cre;Dst cRescue mice. Synchronization between triceps and biceps brachii muscle activity was observed as a hump at around 0 ms in Dst GT mice but not in other mice.

Sensory-motor circuit in Wnt1-Cre;Dst cGT and Wnt1-Cre;Dst cRescue mice

Sensory neurodegeneration was investigated in Wnt1-Cre;Dst cGT mice and Wnt1-Cre;Dst cRescue mice. Neurofilament (NF) accumulation in the CNS and PNS is a pathological hallmark of dt mice (17–19, 28). Abnormal NF accumulation and induction of activating transcription factor-3 (ATF3), a marker of neural injury, are observed in DRG sensory neurons of Dst GT mice throughout the cervical, thoracic, and lumbar levels (Fig. 4, A and C) (29). Such pathological alterations also occurred in DRG sensory neurons of Wnt1-Cre;Dst cGT mice. Meanwhile, NF accumulation and ATF3 expression were scarcely observed in these neurons of Wnt1-Cre;Dst cRescue mice. In the brain of Wnt1-Cre;Dst cGT mice, we observed NF accumulations in the cerebellum and midbrain while scarcely observed in their medulla oblongata and cerebral cortex (fig. S5, A and B). In contrast, NF accumulation decreased in the cerebellum and midbrain of Wnt1-Cre;Dst cRescue mice, compared with Dst GT mice (fig. S5, A and B). In sympathetic ganglion neurons of Wnt1-Cre;Dst cGT mice, ATF3 was up-regulated, whereas ATF3 was not expressed in those of Wnt1-Cre;Dst cRescue mice (fig. S6, A and B). These data showed that Dst loss of function leads to neurodegeneration via cell-autonomous mechanisms. Large-size DRG neurons are more vulnerable than small-size DRG neurons in dt mice (30). We observed decreased numbers of proprioceptive and mechanosensory neurons in the DRG of Dst GT mice, while thermal nociceptive neurons were less affected (fig. S7, A to F). We found that in Dst GT mice and Wnt1-Cre;Dst cGT mice, ATF3 was more frequently expressed in proprioceptive neurons than in thermal nociceptive neurons (fig. S8, A and B). Proprioceptive neurons in the DRG directly form synapses with spinal motor neurons to mediate monosynaptic reflexes. Parvalbumin (PV) and vesicular glutamate transporter 1 (VGluT1) have been used as markers for proprioceptive DRG neurons and proprioceptive synapses on spinal motor neurons, respectively (31, 32). VGluT1 labeling of proprioceptive axon terminals in the anterior horn was significantly decreased in Dst GT mice and Wnt1-Cre;Dst cGT mice compared with that in Ctrl mice (Fig. 4, B and C). Meanwhile, proprioceptive axon terminals were more abundant in Wnt1-Cre;Dst cRescue mice than Dst GT mice (Fig. 4, B and C). In the spinal cord of Ctrl mice and Wnt1-Cre;Dst cRescue mice, many VGluT1-positive axon terminals contacted with choline acetyltransferase (ChAT)–positive motor neurons, whereas fewer terminals contacted in Dst GT mice and Wnt1-Cre;Dst cGT mice (fig. S9A). Wnt1-Cre;Dst cGT mice had fewer PV-positive proprioceptive neurons and Trkb-positive mechanosensory neurons compared with Ctrl mice, while Wnt1-Cre;Dst cRescue and Ctrl mice had comparable numbers (fig. S6, C to F). No significant differences in the number of ChAT-positive motor neurons were observed between Ctrl, Wnt1-Cre;Dst cGT, Dst GT, and Wnt1-Cre;Dst cRescue mice (Fig. 4, B and C).

Fig. 4. Histological analysis of sensory-motor circuit in Wnt1-Cre;Dst cGT and Wnt1-Cre;Dst cRescue mice.

(A) In Dst GT and Wnt1-Cre;Dst cGT mice, ATF3 was expressed in some DRG neurons together with neurofilament (NF) accumulation (arrowheads) at 3 to 4 weeks of age. Increases in ATF3 and NF were rarely observed in Wnt1-Cre;Dst cRescue mice. The dotted line indicates the edge of the DRG. (B) VGluT1-positive axon terminals were observed in the lumbar SC. In the anterior horn (AH), ChAT-positive motor neurons were surrounded by VGluT1-positive axon terminals. There were fewer VGluT1-positive axon terminals in the AH of Dst GT and Wnt1-Cre;Dst cGT mice than in Ctrl mice, but they were normally distributed in Wnt1-Cre;Dst cRescue mice. The dotted line indicates the edge of the SC. (C) Quantitative data showed a significant increase in numbers of ATF3-positive cells and NF-accumulating cells in Dst GT and Wnt1-Cre;Dst cGT mice [ATF3, Ctrl (n = 5 mice); Wnt1-Cre;Dst cGT (n = 3); Dst GT (n = 5); Wnt1-Cre;Dst cRescue (n = 6); NF, Ctrl (n = 6); Wnt1-Cre;Dst cGT (n = 3); Dst GT (n = 4); Wnt1-Cre;Dst cRescue (n = 3)]. Quantitative data showed a significant decrease in number of VGluT1-positive axon terminals in the AH of Wnt1-Cre;Dst cGT mice. VGluT1-positive axon terminals were significantly increased in Wnt1-Cre;Dst cRescue mice than Dst GT mice [Ctrl (n = 12); Wnt1-Cre;Dst cGT (n = 3); Dst GT (n = 6); Wnt1-Cre;Dst cRescue (n = 5)]. The number of ChAT-positive cells was statistically same between groups [Ctrl (n = 8); Wnt1-Cre;Dst cGT (n = 3); Dst GT (n = 5); Wnt1-Cre;Dst cRescue (n = 5)]. Scale bars, 50 μm (A) and 200 μm (B). Data are presented as mean ± SE. P > 0.05 (ns), *P < 0.05, **P < 0.01, and ***P < 0.005, using ANOVA with Tukey’s test in (C).

Next, we performed electrophysiological analyses to assess functional changes in the sensory-motor circuit. We stimulated the tibial nerve and recorded EMG in the plantar muscle. An M response, a direct response, is produced by motor neuron activation. A Hoffmann reflex (H-reflex) is produced by a spinal cord pathway consisting of group Ia afferents from the muscle spindles and the motor neurons (Fig. 5A). In Ctrl mice, the M response and H-reflex sequentially occurred following stimulation. In Wnt1-Cre;Dst cGT mice, attenuated H-reflex was recorded with delay (Fig. 5B and fig. S10). The H-reflex was rescued in Wnt1-Cre;Dst cRescue mice than in Dst GT mice (Fig. 5B and fig. S10). The frequency of successful H-reflexes was lower in Dst GT and Wnt1-Cre;Dst cGT mice than in Ctrl mice. The H/M amplitude ratio (H response maximum amplitude/M response maximum amplitude) in Dst GT and Wnt1-Cre;Dst cGT mice was significantly reduced compared with that in Ctrl mice (Fig. 5C). The latencies of both the M response and H-reflex in Dst GT mice and Wnt1-Cre;Dst cGT mice were significantly longer than in Ctrl mice, while Wnt1-Cre;Dst cRescue mice showed normal M response and H-reflex latency (Fig. 5C). These data demonstrated reduced amplitude and delayed response of the H-reflex in Wnt1-Cre;Dst cGT mice. In contrast, the H-reflex in Wnt1-Cre;Dst cRescue mice showed normal amplitude and latency. These data indicate that the proprioceptive feedback circuit is morphologically and functionally disrupted or rescued following inactivation or restoration of Dst expression in sensory neurons, respectively. We therefore consider that loss of Dst function in the sensory circuit leads to disruption of proprioceptive sensory feedback and movement disorders.

Fig. 5. Functional analysis of the sensory-motor circuit in Wnt1-Cre;Dst cGT and Wnt1-Cre;Dst cRescue mice.

(A) A diagram of the H-reflex mediated by proprioceptive sensory neurons and motor neurons. The M response was evoked by electronic stimulation of motor fibers in the tibial nerve of mice. The H-reflex is transmitted via Ia afferent fibers derived from proprioceptive neurons. (B) Representative EMG images of M response and H-reflex. In Ctrl mice, electrical stimulation (Sti) evoked M response and H-reflex sequentially (black solid line). In Wnt1-Cre;Dst cGT mice, H-reflex was attenuated and delayed (yellow dashed line). In Wnt1-Cre;Dst cRescue mice, stimulation normally induced H-reflex (blue solid line), whereas it was attenuated and delayed in Dst GT mice (red dashed line). (C) Quantitative data of H/M amplitude ratio, H-reflex latency (milliseconds), and M response latency (milliseconds) in each mouse group at 3 to 4 weeks of age. H/M amplitude ratio was significantly decreased in Dst GT mice and Wnt1-Cre;Dst cGT mice compared to Ctrl mice, whereas Wnt1-Cre;Dst cRescue mice had a normal level of H/M amplitude ratio as in Ctrl mice [H/M amplitude ratio and H-reflex latency, Ctrl (n = 29); Wnt1-Cre;Dst cGT (n = 6); Dst GT (n = 7); Wnt1-Cre;Dst cRescue (n = 6); sample sizes are numbers of successful evocation of H-reflex]. The latencies of H-reflex and M response in Dst Gt mice and Wnt1-Cre;Dst cGT mice were significantly longer than in Ctrl mice, while latencies of both responses in Wnt1-Cre;Dst cRescue were statistically same as in Ctrl mice [M response latency, Ctrl (n = 32); Wnt1-Cre;Dst cGT (n = 11); Dst GT (n = 12); Wnt1-Cre;Dst cRescue (n = 6); sample sizes are numbers of measurement of M response]. Data are presented as mean ± SE. P > 0.05 (ns) and ***P < 0.005, using ANOVA with Tukey’s test in (C).

Manipulation of Dst separately in the sensory-motor circuit and cerebellar circuit

Our data suggested that disruption to the sensory-motor circuit underlies the dystonic movements in dt mice. However, Cre-mediated recombination also occurred in the cerebellar circuits of Wnt1-Cre;Dst^Gt-inv/wt mice (Fig. 1C). Therefore, it is still possible that inactivation of Dst in cerebellar circuits also contributes to abnormal movements in dt mice. Dst expression was separately manipulated in PNS neurons and cerebellar circuits using Advillin (Avil)–Cre mice and En1-Cre mice to distinguish involvement of the PNS and cerebellum (Table 1). Avil-Cre mice produce Cre-mediated recombination in PNS neurons, including sensory neurons of DRG and trigeminal ganglia after the late developmental stage but rarely in the brain and spinal cord (33). We therefore used Avil-Cre knock-in mice, which have been used for PNS neuron–specific gene manipulations (33–37). In Avil-Cre;Dst^Gt-inv/wt mice, Cre-mediated gene trap switching was restricted to DRG neurons and was not observed in cerebellar circuits (fig. S11A). Dst mRNA was significantly decreased in DRG neurons of Avil-Cre;Dst^Gt-DO/Gt-inv (Avil-Cre;Dst cGT) mice, whereas Dst expression was restored in DRG neurons of Avil-Cre;Dst^Gt-DO/Gt (Avil-Cre;Dst cRescue) mice (fig. S11, B and E). En1-Cre knock-in mice have been used to manipulate Dst expression in the cerebellum and spinal interneurons (38, 39). We confirmed inversion of the gene trap cassette in cerebellar circuits, including in Purkinje cells, deep cerebellar nucleus neurons, and the pontine nucleus in En1-Cre;Dst^Gt-inv/wt mice, while gene trap inversion did not occur in DRG neurons (fig. S11A). Although En1-Cre mice label large population of spinal interneurons (39), we did not observe X-gal–positive cells in the En1-Cre;Dst^Gt-inv/wt spinal cord. Therefore, it seems that En1-lineage spinal interneurons do not express Dst. Moreover, Dst expression in DRG neurons was not inactivated in En1-Cre;Dst cGT mice and not restored in En1-Cre;Dst cRescue mice (fig. S11, B and E).

Avil-Cre;Dst cGT mice exhibited sensory neurodegeneration accompanied by NF accumulation and ATF3 induction, whereas Avil-Cre;Dst cRescue mice exhibited an ameliorated phenotype (fig. S11, C to E). Avil-Cre;Dst cGT mice did not display decrease of Dst mRNA and NF accumulation in the cerebellum/midbrain regions, unlike Wnt1-Cre;Dst cGT mice (fig. S12, A and B). En1-Cre;Dst cGT mice did not exhibit NF accumulation or ATF3 induction in DRG sensory neurons (fig. S11, C to E). However, decrease of Dst mRNA and NF accumulation were observed in the cerebellum/midbrain regions of En1-Cre;Dst cGT mice, as in Wnt1-Cre;Dst cGT mice (fig. S12, A and B). Avil-Cre;Dst cGT mice exhibited hindlimb clasping in the tail suspension test, whereas Avil-Cre;Dst cRescue mice could maintain normal posture (Fig. 6A). Freely moving Avil-Cre;Dst cRescue mice did not display dystonic movements (movie S3). Avil-Cre;Dst cGT mice and Avil-Cre;Dst cRescue mice showed impaired and rescued motor performance, respectively, as assessed by the rotarod test (Fig. 6B). On the other hand, postures and motor performance were not impaired in En1-Cre;Dst cGT mice and not rescued in En1-Cre;Dst cRescue mice (Fig. 6, A and B, and movie S4). The number of VGluT1-labeled proprioceptive synapses around motor neurons was decreased in Avil-Cre;Dst cGT mice, whereas Avil-Cre;Dst cRescue mice had abundant proprioceptive synapses, similar to Ctrl mice (Fig. 6, C and D, and fig. S9B). The number of PV-positive proprioceptive neurons also decreased in Avil-Cre;Dst cGT mice than Ctrl mice, whereas Avil-Cre;Dst cRescue mice had similar number of these neurons as Ctrl mice (fig. S13, A and B). In Avil-Cre;Dst cGT mice, some proprioceptive neurons expressed ATF3 (fig. S8, A and B). Electrophysiological analysis also demonstrated that the H/M amplitude ratio was small in Avil-Cre;Dst cGT mice and that latency of the H-reflex in Avil-Cre;Dst cGT mice was significantly longer than that in Ctrl mice (Fig. 6E). Unlike Wnt1-Cre;Dst cGT mice, Avil-Cre;Dst cGT mice did not exhibit prolongation of the M response latency (Fig. 6E). Conversely, the H/M amplitude ratio of Avil-Cre;Dst cRescue mice was normal, while the latencies of the M response and H-reflex did not rescue compared with Ctrl mice (Fig. 6E). These data indicate that disruption of the sensory-motor circuit and not the cerebellar circuit is responsible for dystonic movements in dt mice.

Fig. 6. Manipulation of Dst expression separately in sensory neurons and cerebellar circuits.

(A) During tail suspension, Avil-Cre;Dst cGT mice showed hindlimb clasping at 3 to 4 weeks of age, while Avil-Cre;Dst cRescue mice maintained normal postures. En1-Cre;Dst cGT and En1-Cre;Dst cRescue mice neither developed nor rescued abnormal postures, respectively. (B) In a rotarod test, Avil-Cre;Dst cGT mice displayed shorter latency to fall (seconds) than Ctrl. Avil-Cre;Dst cRescue mice performed better than Dst GT mice [Ctrl (n = 14 mice); Avil-Cre;Dst cGT (n = 4); Dst GT (n = 4); Avil-Cre;Dst cRescue (n = 8)]. En1-Cre;Dst cGT and En1-Cre;Dst cRescue mice showed no differences compared to Ctrl and Dst GT mice, respectively [Ctrl (n = 6); En1-Cre;Dst cGT (n = 7); Dst GT (n = 4); En1-Cre;Dst cRescue (n = 3)]. (C) In the lumbar SC of each group at 3 months old, VGluT1-labeled terminals were decreased in Avil-Cre;Dst cGT mice than Ctrl but normal in Avil-Cre;Dst cRescue mice. (D) Quantitative data on numbers of VGluT1-positive terminals [Ctrl (n = 6 mice); Avil-Cre;Dst cGT (n = 3); Avil-Cre;Dst cRescue (n = 3)]. (E) Quantification of electrophysiological data in mice at 3 months of age. H/M amplitude ratio was reduced in Avil-Cre;Dst cGT mice than Ctrl but normal in Avil-Cre;Dst cRescue mice [H/M amplitude ratio and H-reflex latency, Ctrl (n = 12); Avil-Cre;Dst cGT (n = 5); Avil-Cre;Dst cRescue (n = 8); sample sizes are numbers of successful evocation of H-reflex]. Avil-Cre;Dst cRescue mice show delayed M response and H-reflex [M response latency, Ctrl (n = 13); Avil-Cre;Dst cGT (n = 12); Avil-Cre;Dst cRescue (n = 8); sample sizes are numbers of measurement of M response]. Scale bars, 100 μm (C). Data are presented as mean ± SE. P > 0.05 (ns), *P < 0.05, **P < 0.01, and ***P < 0.005, using ANOVA with Tukey’s test in (B), (D), and (E).

Treatment of Dst^Gt mice with a postnatal AAV injection

Intraperitoneal administration of AAV9 vectors allows efficient gene transfer to DRG neurons in neonatal mice (40). Therefore, we attempted to ameliorate the movement disorder in Dst^Gt homozygous mice through systemic delivery of an adeno-associated virus (AAV). AAV9–green fluorescent protein (GFP) or AAV9-Cre was intraperitoneally administered to Dst^Gt homozygous mice with an ICR background at an early postnatal stage (Fig. 7A). After tissue clearing using clear unobstructed brain/body imaging cocktails and computational analysis (CUBIC), we observed the projecting axons of GFP-labeled DRG neurons in the dorsal funiculus (Fig. 7B). In transverse sections of the spinal cord, GFP-labeled axons were predominantly observed in the dorsal horn, with some located in the anterior horn (Fig. 7C). We also observed GFP-labeled axons in the medullar oblongata where dorsal funiculus and trigeminal nerve fibers project and sparse GFP-labeled cells in other brain regions (movie S5). The life span of the Dst^Gt homozygous mice was approximately 3 to 4 weeks, and a single injection of AAV9-Cre extended their life span: Approximately half of these mice survived for 4 months, and some (~10%) survived for more than 10 months (Fig. 7D). Dystonic movements and motor performance were improved in the AAV9-Cre–treated Dst^Gt homozygous mice (Fig. 7E and movie S6), and footfalls were reduced on the horizontal ladder floor (Fig. 7F). It should be noted that motor performance of AAV9-Cre–treated mice gradually worsened at 2 months after treatment, while their motor symptom still seemed to be milder than nontreated Dst^Gt mice (movie S7). Dst expression was restored in the DRG neurons of the AAV9-Cre–treated Dst^Gt mice, while Dst mRNA was not detected in AAV9-GFP–treated Dst^Gt mice (Fig. 7, G and H). Treatment with AAV9-Cre ameliorated sensory neurodegeneration and increased numbers of proprioceptive neurons and proprioceptive synapses around motor neurons (Fig. 8, A and B, and fig. S9C). The H/M amplitude ratio and the H-reflex latency were rescued in Dst^Gt homozygous mice treated with AAV9-Cre compared with nontreated Dst^Gt homozygous mice (Fig. 8C). Two months after AAV9-Cre injection, ATF3 was slightly increased in the DRG of Dst^Gt homozygous mice (fig. S14). However, PV-positive proprioceptive neurons were observed as normal, and some PV-positive neurons expressed ATF3 after 2 months of AAV treatment (fig. S14). Therefore, we considered that the decline in motor performance after 2 months of AAV9-Cre treatment is related to dysfunction of proprioceptive neurons rather than neuronal loss.

Fig. 7. Presymptomatic restoration of Dst expression using viral vectors rescued disease phenotypes.

(A) Intraperitoneal (ip) injection of AAV9-Cre or AAV9-GFP into neonatal Dst^Gt/Gt mice. (B) A reconstructed transparent SC image from AAV9-GFP–injected wild-type mice at 2 months of age. GFP labeled the DRG and sensory tract in the dorsal funiculus (DF, arrow). The dotted line indicates the edge of DRG. (C) GFP-labeled axons are distributed in the dorsal horn (DH). In the AH, axons of proprioceptive neurons were also labeled with GFP. (D) The survival curve indicates a life extension of AAV9-Cre–treated Dst^Gt/Gt mice (blue solid line, n = 8 mice) compared with Dst^Gt/Gt mice (red dashed line, n = 12). (E) AAV9-GFP–treated Dst^Gt/Gt mice exhibited twist movements with hyperextension of forelimbs and hindlimbs (arrowheads). AAV9-Cre–treated Dst^Gt/Gt mice showed normal posture 3 weeks after injection. (F) Motor performance was assessed by counting number of footfalls on the horizontal ladder floor. Impaired motor performance of Dst^Gt/Gt mice was represented with many footfalls compared with Ctrl mice. AAV9-Cre treatment significantly improved motor performance of Dst^Gt/Gt mice at 3 weeks of age while not reaching the Ctrl level [Ctrl (n = 3 mice); Dst^Gt/Gt (n = 3); Dst^Gt/Gt + AAV9-Cre (n = 4)]. (G) In situ hybridization image in the DRG from each group at 3 weeks of age. Dst mRNA was detected in DRG neurons of Dst^Gt/Gt mice injected with AAV9-Cre, but Dst mRNA was diminished in DRG neurons of Dst^Gt/Gt mice injected with AAV9-GFP. (H) Quantitative data on numbers of Dst-positive neurons in the DRG [Ctrl (n = 3 mice); Dst^Gt/Gt (n = 4); Dst^Gt/Gt + AAV9-Cre (n = 5)]. Scale bars, 500 μm (B), 50 μm (C), and 100 μm (G). Data are presented as mean ± SE. P > 0.05 (ns), *P < 0.05, and ***P < 0.005, using ANOVA with Tukey’s test in (F) and (H).

Fig. 8. Presymptomatic treatment with AAV rescued the sensory-motor circuit of Dst^Gt mice.

(A) Histological analysis of neurodegeneration in each group at 3 weeks of age. In DRG of Dst^Gt/Gt mice injected with AAV9-GFP, ATF3 and NF accumulation was abundantly observed, and there were few VGluT1-positive axon terminals around motor neurons in the AH of lumbar SC. In DRG of Dst^Gt/Gt mice injected with AAV9-Cre, ATF3 and NF accumulation was scarcely observed, and VGluT1-positive axon terminals were abundant. PV-positive proprioceptive neurons were few in the DRG of Dst^Gt/Gt mice injected with AAV9-GFP while normally observed in Dst^Gt/Gt mice injected with AAV9-Cre. (B) Quantification of numbers of ATF3-positive cells [Ctrl (n = 4 mice); Dst^Gt/Gt (n = 4); Dst^Gt/Gt + AAV9-Cre (n = 5)], NF-accumulating cells [Ctrl (n = 3); Dst^Gt/Gt (n = 3); Dst^Gt/Gt + AAV9-Cre (n = 5)], PV-positive cells [Ctrl (n = 4); Dst^Gt/Gt (n = 5); Dst^Gt/Gt + AAV9-Cre (n = 5)], and VGluT1-positive axon terminals [Ctrl (n = 3); Dst^Gt/Gt (n = 3); Dst^Gt/Gt + AAV9-Cre (n = 5)]. (C) Quantification of electrophysiological data in mice at 2 to 3 weeks of age. H/M amplitude ratio and H-reflex latency in Dst^Gt/Gt mice were rescued by injection with AAV9-Cre. [H/M amplitude ratio and H-reflex latency, Ctrl (n = 8); Dst^Gt/Gt (n = 9); Dst^Gt/Gt + AAV9-Cre (n = 4); sample sizes are numbers of successful evocation of H-reflex; M response latency, n = 10 Ctrl; n = 16 Dst^Gt/Gt; n = 6 Dst^Gt/Gt + AAV9-Cre; sample sizes are numbers of measurement of M response]. Scale bars, 50 μm (A). Data are presented as mean ± SE. P > 0.05 (ns), *P < 0.05, and ***P < 0.005, using ANOVA with Tukey’s test in (B) and (C).

DISCUSSION

In this study, we used a FLEX-mediated multipurpose gene trap system that enables tissue-selective inactivation and restoration of Dst expression to demonstrate that disruption of the sensory-motor circuit is involved in the movement disorder of dt mice. Dt mice exhibit dystonic movements and postures accompanied by synchronized muscle activities between agonist and antagonist muscles. Inactivation of Dst in PNS neurons leads to disruption of sensory-motor circuits and impaired motor performance. Conversely, restoring Dst in PNS neurons ameliorated sensory neurodegeneration, suppressed synchronized muscle activities, rescued dystonic movements, and extended life span. The therapeutic effects of postnatal Dst restoration by a single injection of AAV-Cre advocate the efficacy of gene therapy for HSAN-VI. Overall, we have identified the sensory circuit as a therapeutic target of HSAN-VI after birth.

The FLEX-mediated multipurpose gene trap system useful for elucidating pathogenesis

The FLEX-mediated multipurpose gene trap system in a single allele is a useful experimental system to investigate gene function (22, 41). Here, we applied this technology to elucidate pathogenic mechanisms and therapeutic targets of the inherited neurological disease HSAN-VI. The short life span of dt mice makes it challenging to analyze the mechanisms that underlie the dt phenotype. However, sensory neuron–selective inactivation of Dst reproduced the disease phenotypes of dt mice and circumvented postnatal lethality. Therefore, cGT is helpful for determining neural circuits and cell types involved in disease pathogenesis and for developing animal models that can avoid lethality when loss-of-function mutations in a gene directly cause lethality. However, the specificity and efficiency of Cre recombination need to be considered when interpreting experimental results. For example, Wnt1-Cre;Dst cGT mice exhibited more severe symptoms than Avil-Cre;Dst cGT mice. Such phenotypic differences in cGT mice might be caused by spatiotemporal differences in Cre recombination between Wnt1-Cre and Avil-Cre mice (42). Wnt1-Cre–mediated recombination occurs in a wide range of neural crest derivatives, including sensory neurons, sympathetic ganglionic neurons, and Schwann cells (27), whereas Avil-Cre–mediated recombination occurs in sensory neurons and postganglionic neurons of the autonomic nervous system (34, 35). In the present results, the M response latency was prolonged in Wnt1-Cre;Dst cGT mice, while it was unchanged in Avil-Cre;Dst cGT mice. This difference is probably due to differential recombination in Schwann cells. We have previously demonstrated the significance of Dst in Schwann cells for maintaining myelin sheaths in the PNS (43). Conditional Dst deletion in Schwann cells using a P0-Cre transgene resulted in late-onset neuropathy with prolongation of M response latency. This suggests that lack of Dst in sensory neurons, not in Schwann cells, is the primary cause of early postnatal neurodegeneration and abnormal movements in dt mice. On the other hand, Dst in Schwann cells is essential for the maintenance of myelin sheaths and normal nerve conduction in the PNS at the adult stage. Patients with HSAN-VI suffer from autonomic dysfunction and sensory neuropathy (7–11). Neuronal degeneration in the autonomic nervous system has been reported in dt mice (44, 45). We are now investigating the significance of autonomic dysfunctions in dt mice. Abnormalities of multiple PNS cell types may be associated with the pathogenesis of HSAN-VI.

The PNS as a therapeutic target for dt mice and HSAN-VI

Here, we investigated the target to ameliorate the dt phenotype through conditional restoration of Dst in several tissues. Dst cRescue with Wnt1-Cre and Avil-Cre but not En1-Cre improved motor performance and prolonged life span, indicating that the Wnt1 and Avil lineages of PNS neurons rather than the cerebellum/midbrain are therapeutic targets of dt mice. Therefore, PNS neurons are a potential therapeutic target for dt mice, a mouse model of HSAN-VI. Our results are also supported by the previous study showing that transgenic expression of Dst under a neuron-specific promoter partially rescues the dt phenotype (16). Notably, a single administration of AAV-Cre at the early postnatal stage improved the survival and functional rescue of the sensory circuit, leading to improved motor symptoms and prolonged life span. These results indicate that the sensory circuit of dt mice is a target for gene therapy after birth. Previous studies have proposed gene therapy using viral vectors as a therapeutic strategy for sensory neuropathy (46, 47). A single dose of AAV encoding the survival motor neuron protein has been successfully used to treat patients with spinal muscular atrophy, an inherited motor neuron disease (48). The success of postnatal AAV therapy in dt mice shows that the same strategy has promise for treating HSAN-VI and other hereditary neuropathies. However, the size limit of an exogenous gene (5 kb) in AAV makes it difficult to insert the large DST-a gene (18 kb) into the AAV genome for gene delivery. There is, however, the possibility of applying genome editing methods to enable treatment using AAV injection (49, 50). HSAN-VI presents a spectrum of symptoms due to the complexity of the DST gene, which generates tissue-selective isoforms, which need to be considered in therapy development (3, 6). In addition to the neural DST-a isoform associated with HSAN-VI, there is an epidermal DST-e isoform involved in epidermolysis bullosa simplex (51, 52). Furthermore, we also found late-onset protein aggregate myopathy and cardiomyopathy in Dst-b isoform–specific mutant mice (29). Life-span recovery by Dst cRescue in the PNS with Wnt1-Cre and AAV9-Cre was incomplete. Thus, defects in Dst isoforms other than PNS may also be associated with the disease phenotype of dt mice. When considering the treatment of these DST disorders, the location of each DST mutation and the target tissue expressing the mutated DST isoform must be considered.

Causative neuronal circuits in dystonic movement disorders

As the name suggests, dt mice exhibit dystonic movements with co-contractions between agonist and antagonist muscles. The neural circuit mechanisms of dystonia remain unclear. Previous studies of mouse genetic models of DYT1-TorsinA (TOR1A), a causative gene for hereditary dystonia, and manipulation of cerebellar circuits have shown that neuronal circuits in the cerebellum and striatum are involved in dystonia (53–59). The dystonia-causal Tor1a mutation Dyt1ΔE impairs the formation of inhibitory synapses against proprioceptive afferent synapses on motor neurons (60), but their involvement in motor deficits remains to be elucidated. In some mouse models, disruption of proprioception causes ataxia and abnormal posture (61–64). The recent report of conditional deletion of Tor1a in the spinal cord and DRG describes impaired monosynaptic reflexes in the spinal cord and recapitulates early-onset generalized torsional dystonia (65). The lack of Tor1a in the spinal cord was also considered to contribute to dystonia because DRG-specific Tor1a deletion alone does not reproduce dystonia. Together, the neural circuits responsible for dystonia are probably diverse and unclear.

Traditionally, dystonia has been considered a disorder of the basal ganglia, but previous studies suggest that dystonia is a network disorder involving a wide range of sensory-motor regions. In the present study, we demonstrated that normalization of the sensory-motor circuit suppressed co-contraction between agonist and antagonist muscles and rescued dystonic movements and postures in dt mice. Furthermore, we found that Dst deletion in sensory neurons caused disruption of the sensory-motor circuit, ataxia, and twisting movements during tail suspension but did not fully reproduce co-contractions. These results indicate that disruption of the proprioceptive circuit contributes to dystonic movement in dt mice; however, disruption of other circuits is also necessary to fully reproduce dystonia with co-contraction. Dystonic movements caused by sensory loss resulting from spinal cord or PNS damage are called pseudo-dystonia to distinguish them from dystonia associated with basal ganglia abnormalities (66). Although the contributions of sensory circuit abnormalities to dystonia are still to be elucidated, our study adds to the current understanding of the neural circuit mechanisms of dystonia.

MATERIALS AND METHODS

Animals

We used Dst^Gt(E182H05) mice (Dst^Gt; mouse genome informatics number, MGI:3917429) (23) derived from the embryonic stem (ES) clone (ID: E182H05) obtained from German Genetrap Consortium (22). Dst^Gt mice were crossed with several Cre-driver mice (Table 1) including neural crest and midbrain/cerebellum-selective Wnt1-Cre transgenic mice [H2az2^{Tg(Wnt1-cre)11Rth}; MGI:2386570] (24), sensory neuron–selective Avil-Cre mice [Avil^tm2(cre)Fawa; MGI:4459942] (34), and midbrain/cerebellum-selective En1-Cre mice [En1^tm2(cre)Wrst; MGI:2446434] (38). The Dst^Gt line was backcrossed to C57BL/6NCrj (Charles River Japan Inc.) or ICR (SLC Japan) for at least 10 generations. The inbred C57BL/6NCrj strains were used for mouse genetics. Because ICR mice have larger body size and produce more offspring than C57BL/6NCrj mice, ICR strains were used for infections with viral vectors. The functional Dst^Gt-inv allele (MGI:7423689) was inverted from Dst^Gt allele by FLP recombinase, and mutant Dst^Gt-DO allele (MGI:7423690) was inverted from Dst^Gt-inv allele by Cre recombinase (23). Mutant mice of Dst^Gt, Dst^Gt-inv, and Dst^Gt-DO were maintained in each line. Since heterozygotes of Dst^Gt or Dst^Gt-DO are fertile, homozygotes of Dst^Gt or Dst^Gt-DO were obtained by the heterozygous mating. To perform Cre-mediated conditional inactivation or restoration of the Dst expression, gene trap mice harboring multipurpose Dst alleles were crossed with each Cre driver mouse. Male and female mice were analyzed in this study, but only male mice were used for body weight comparisons. The animal experiments were approved by the Internal Review Board of Niigata University (permit numbers: SA00521 and SA00621) and the Institutional Animal Care and Use Committee of the National Institutes of Natural Sciences (21A014), and the guidelines of Niigata University Animal Care and Use Committee were followed. Mice were maintained at 23° ± 3°C, 50 ± 10% humidity, and 12-hour light/dark cycle with food and water available ad libitum. Humane endpoints were applied as per veterinary recommendation, and the criteria used to determine when mice should be euthanized included severe dehydration, immobility, and desertion by feeding mother.

X-gal staining

The X-gal staining was performed as described previously (23). Tissues were fixed with a mixture of 2% paraformaldehyde (PFA) and 0.2% glutaraldehyde, immersed in the fixative overnight, and transferred to 20% sucrose in 20 mM phosphate-buffered saline (PBS; pH 7.4) until they sank. The tissues were frozen in liquid nitrogen and embedded in Tissue-Tek OCT compound (Sakura Finetek Japan, Tokyo, Japan), and 16-μm-thick sections were cut on a cryostat (Leica CM1850 UV; Leica, Wetzlar, Germany; HM525 NX; Thermo Fisher Scientific). Sections were washed for 10 min in PBS and incubated with X-gal staining solution {5 mM K₄[Fe(CN)₆], 5 mM K₃[Fe(CN)₆], 20 mM Tris-HCl (pH 7.3), 2 mM MgCl₂, 0.2% NP-40, 0.1% sodium deoxycholate, and 0.1% X-gal in PBS} at 37°C for 3 days. We used tissues from littermate Dst^Gt-inv/+ mice as negative controls.

Tissue preparations and in situ hybridization

For tissue preparation, mice were euthanized with an intraperitoneal injection of pentobarbital sodium (100 mg/kg body weight) and then perfused with 4% PFA in 0.1 M phosphate-buffered (PB) solution (pH 7.4) by cardiac perfusion with 0.01 M PBS followed by ice-cold 4% PFA in 0.1 M PB (pH 7.4). Dissected tissues were immersed in the same fixative overnight. For cutting the spinal cord and DRG sections, the specimens were rinsed with water for 10 min and decalcified in Morse solution (135-17071; Wako, Osaka, Japan) overnight. Tissues were dehydrated through an ascending series of ethanol and xylene and then embedded in paraffin (P3683; Paraplast Plus; Sigma-Aldrich, St. Louis, USA). Consecutive 10-μm-thick paraffin sections were cut on a rotary microtome (HM325; Thermo Fisher Scientific), mounted on Matsunami Adhesive Silane (MAS)-coated glass slides (Matsunami Glass, Osaka, Japan), and air-dried on a hot plate overnight at 37°C.

In situ hybridization was performed on paraffin sections described in the previous study (67). The following probes were used: mouse Dst-plakin (GenBank accession number: NM_001276764, nucleotides 2185 to 3396), mouse PV (also known as Pvalb, NM_013645, nucleotides 92 to 885), rat TrkA (a gift from K. Abe, accession number M85214, nucleotides 80 to 427), mouse Trkb (a gift from S. Nakamura) (68), rat TrkC (a gift from K. Abe, accession number L14445, nucleotides 1 to 311), mouse Vglut1 (also known as Slc17a7, NM_182993, nucleotides 618 to 1266) (69), mouse CGRP (also known as Calca, GenBank accession number, NM_001289444, nucleotides 567 to 914), mouse Galectin-1 (also known as Lgals1, GenBank accession number, NM_008495, nucleotides 12 to 769), and mouse Trpm8 (GenBank accession number, NM_134252, nucleotides 2151 to 2968).

Immunohistochemistry

Analyses of immunohistochemistry (IHC) were performed on paraffin sections. For IHC, deparaffinized sections were treated with microwave irradiation in 10 mM citric acid buffer (pH 6.0) for 5 min and incubated overnight at 4°C with the following primary antibodies: rabbit polyclonal anti-ATF3 antibody (1:2000; Santa Cruz Biotechnology; sc-188), rabbit polyclonal anti-VGluT1 antibody (1:1000; Frontier Institute, Hokkaido, Japan; Rb-Af500), goat polyclonal anti-VGluT1 antibody (1:1000; Frontier Institute; Go-Af310), goat polyclonal anti-ChAT antibody (1:500; Millipore; AB144P), guinea pig polyclonal anti-CGRP antibody (1:300; Frontier Institute; GP-Af280), mouse monoclonal anti-neurofilament-M antibody (1:500; 1C8) (70), and mouse monoclonal anti-PV antibody (1:1000; Sigma-Aldrich; PARV-19, P3088), diluted in 0.1 M PBS with 0.01% Triton X-100 (PBST) containing 0.5% skim milk. Sections were then incubated in horseradish peroxidase–conjugated secondary antibody (1:200; Medical & Biological Laboratories (MBL), Nagoya, Japan) diluted in PBST containing 0.5% skim milk for 60 min at 37°C. Between each step, sections were rinsed in PBST for 15 min. After rinsing sections in distilled water, immunoreaction was visualized in 50 mM Tris buffer (pH 7.4) containing 0.01% diaminobenzidine tetrahydrochloride and 0.01% hydrogen peroxide at 37°C for 5 min. Sections were dehydrated through ethanol-xylene and coverslipped with Bioleit (23-1002; Okenshoji, Tokyo, Japan). For immunofluorescent staining, sections were incubated in mixtures of Alexa Fluor 488– or Alexa Fluor 594–conjugated secondary antibodies (1:200; Invitrogen, CA) for 60 min at 37°C. Mounted sections were air-dried and coverslipped. Digital images were taken with a microscope (BX53; Olympus, Tokyo, Japan) equipped with a digital camera (DP74, Olympus) and a confocal laser scanning microscopy (FV-1200, Olympus). Tag image file format (TIFF) files were processed with Photoshop software (Adobe, San Jose, USA).

Western blotting

Western blotting was performed as previously described (52). Frozen spinal cord and DRG were homogenized in ice-cold homogenization buffer [0.32 M sucrose, 5 mM EDTA, 10 mM Tris-HCl (pH 7.4), and phosphatase inhibitor cocktail tablet; Roche] and centrifuged at 4500 rpm for 10 min at 4°C, and the supernatants were collected. The protein concentration was determined using the bicinchoninic acid Protein Assay Reagent (Thermo Fisher Scientific). Lysates were mixed with an equal volume of 2× SDS sample buffer [125 mM Tris-HCl (pH 6.8), 4% SDS, 20% glycerol, and 0.002% bromophenol blue] for a final protein concentration of 0.5 to 1 μg/μl and denatured in the presence of 100 mM dithiothreitol at 100°C for 5 min. SDS–polyacrylamide gel electrophoresis was performed with 5 to 10 μg per lane on 5 to 20% gradient gels (197-15011; SuperSep Ace; FUJIFILM Wako Pure Chemical Corporation, Osaka, Japan) running at 10 to 20 mA for 150 min. The gels were blotted onto an Immobilon-P transfer membrane (Millipore, Billerica, MA, USA). After blocking with 10% skim milk for 3 hours, blotted membranes were incubated with the following primary antibodies: rabbit polyclonal anti-Dst antibody (gifted from R. K. Liem) (71) which recognizes the plakin domain of Dst and mouse monoclonal anti–β-actin antibody (1:2000; AB_2223041; clone C4, Merck Millipore). Each of the first antibodies was incubated overnight at 4°C. Then, membranes were incubated with peroxidase-conjugated secondary antibodies for 1 hour at room temperature: anti-rabbit immunoglobulin G (IgG) (1:2000; AB_2099233; catalog no. 7074, Cell Signaling Technology, Beverly, MA, USA) or anti-mouse IgG (1:2000; AB_330924; catalog no. 7076, Cell Signaling Technology). Tris-buffered saline [10 mM Tris-HCl (pH 7.5) and 150 mM NaCl] containing 0.1% Tween 20 and 10% skim milk was used to dilute primary and secondary antibodies, and Tris-buffered saline containing 0.1% Tween 20 was used as the washing buffer. Immunoreactions were visualized by enhanced chemiluminescence (GE Healthcare, Piscataway Township, NJ, USA). Images were acquired, and signal intensities from bands were determined using a luminescence image analyzer (C-Digit, LI-COR, Lincoln, NE, USA).

Behavioral tests

We performed behavioral tests to evaluate motor performance. In the rotarod test (O’Hara & Co.) (29), we measured the latency to fall from the rotating rod (30-mm diameter with an acceleration from 10 to 150 rpm). Each trial was conducted for 3 min. In each mouse, two trials were conducted in a day. In the balance beam test, we evaluate the ability of balance and motor performance (O’Hara & Co.) (72). Mice were placed on the end of the 1-m beam (10 mm in diameter) resting 50 cm above the table on two poles and containing a block box placed at the other end of the beam as the finish point. In the horizontal ladder test, mice were placed on the metal rungs (3 mm in diameter with 9-mm distance), and then footfalls of the hindlimbs from the ladder floor were counted from the side view.

The severity of dystonia was assessed in a home cage using locomotor disability scores, as previously described (16). The scores were graded as follows: D0 (no motor abnormalities), D1 (slightly slowed or abnormal motor behavior, no dystonia), D2 (mild impairment, sometimes limited ambulation, dystonic postures when disturbed), D3 (moderate impairment, frequent spontaneous dystonic postures), D4 (severe impairment, sustained dystonic postures, and limited ambulation), and D5 (prolonged immobility in dystonic postures).

To evaluate thermal sensitivity, we performed a plantar test (O’Hara & Co.). Each mouse was placed on a glass floor and surrounded with a clear plastic cage. Radiant thermal stimuli were applied to a hind paw from under a glass floor, and the latency of paw withdrawal was measured. The level of heat stimulation was kept constant throughout the test among the animals.

Measurement of co-contraction by EMG

EMG recordings from the triceps and biceps brachii muscles of the forelimb were performed in the awake state as previously described (17, 23). For EMG recordings from Dst GT mice, these mice were prolonged by intraperitoneal injection with 5% glucose in saline and rearing with MeiBalance mini in agarose (Meiji, Tokyo, Japan). Briefly, mice were anesthetized with isoflurane (1.0 to 1.5%) and fixed in a stereotaxic apparatus. A small U-frame head holder was mounted on the mouse’s head as described previously (23, 73, 74). Then, the EMG recording electrodes (bipolar wire electrodes; tip distance, 1 to 2 mm) made of 140-mm-diameter Teflon-coated seven-stranded stainless steel wire (A-M Systems) were implanted in the bellies of the triceps and biceps brachii muscles of the forelimb. The wires were passed subcutaneously and soldered to connectors attached to the U-frame. After recovery from the surgery, the awake mouse was positioned painlessly in the stereotaxic apparatus using the U-frame head holder. The EMG signals from the triceps and biceps brachii muscles were amplified (×1000), filtered (0.25 to 1.5 kHz), and archived to a computer at a sampling rate of 10 kHz. The EMG was recorded for 20 s when mice moved their limbs intermittently. The EMG was rectified, and co-contraction of the triceps and biceps brachii muscles was analyzed by the following two methods using Igor software (WaveMetrics). (i) EMG was resampled at 1 kHz (1-ms bin, decimation by averaging). Local maximal points (peaks) of rectified EMG were marked during significant changes (>mean + 1.65 SD calculated from 20-s recording) in the triceps and biceps brachii muscles and counted (Tri and Bi, respectively). If a local maximal point of triceps brachii EMG coincided with 3 bins (3 ms) centered at a local maximal point of biceps brachii EMG, then it was considered as co-contraction defined from the triceps brachii muscle (Co_Tri). The similar calculation was performed starting from the biceps brachii muscle (Co_Bi). Percentage of co-contraction was calculated as (Co_Tri/Tri + Co_Bi/Bi)/2 × 100. (ii) Cross-correlograms of rectified EMG were constructed between the triceps and biceps brachii muscles.

Measurement of H-reflex

H-reflex was measured as previously described (43). Mice at 2 weeks to 3 months of age were anesthetized by an intraperitoneal application of ketamine (60 mg/kg) and xylazine (10 mg/kg) during all electrophysiological experiments. Depth of anesthesia was checked repeatedly throughout the experiment by pinching the paws. Ketamine/xylazine was additionally administrated when a withdrawal reflex was elicited. For H-reflex recordings, the tibial nerve was exposed and kept moist in a mineral oil pool made with skin flaps as described previously (75, 76). The nerve was hooked to a custom-made bipolar stimulating electrode (Unique Medical, Tokyo, Japan). A pair of needle electrodes were inserted in the interosseous muscles of the hindpaw for EMG recordings. The signals were fed into a computer equipped with a Cambridge Electronic Design (CED) Power 1401 board and analysis software (Spike 2 ver 6.17; Cambridge Electronic Design Ltd., Cambridge, UK). Repetitive electrical pulses (0.5-ms duration at 0.2 Hz) were applied to the tibial or sciatic nerves by an electric stimulator (SEN-3301, Nihon Kohden, Tokyo, Japan) through a stimulus isolator (SS-203J, Nihon Kohden, Tokyo, Japan). For H-reflex recordings, electrical stimuli of graded intensity were applied to nerves for determining the intensity that evokes the maximal amplitude of M response and H-reflex. A total of 25 repetitive stimuli at an intensity that evokes the maximal amplitude were delivered to evoke constant M response and H-reflex responses. The first five responses were discarded to allow reflex stabilization. Latencies of each response and conduction velocity (meters per second) were measured. The maximum amplitude of H-reflex (H_max) and M responses (M_max) was used to calculate the H_max/M_max ratio. H-reflex and M response were measured from both hindlimbs. Data of H_max/M_max ratio and H-reflex latency were excluded when the H-reflex was not evoked.

AAV vector preparation and injection

AAV vectors were prepared by using the AAV Helper Free Expression System (Cell Biolabs Inc., San Diego, CA) as previously described (77). We chose AAV9 serotype according to the previous report to deliver exogenous gene to DRG neurons (40). The packaging plasmids (pAAV-9 and pHelper) and transfer plasmids (pAAV-CAG-EGFP and pAAV-CAG-Cre) were transfected into human embryonic kidney 293T cells using the calcium phosphate method. A crude cell extract containing AAV vector particles was obtained from transfected cells, and then AAV vector particles were purified by serial ultracentrifugation with cesium chloride. The purified particles were dialyzed with PBS containing 0.001% Pluronic F-68 (Sigma-Aldrich, St. Louis, MO), followed by concentration with an Amicon 10K molecular weight cut-off (MWCO) filter (Merck Millipore, Darmstadt, Germany). The copy number of the vector genomes (vg) was determined by real-time quantitative polymerase chain reaction. AAV9 vectors in PBS (40 μl, 1.0 to 1.4 × 10¹² vg/ml) were intraperitoneally administrated into ICR Dst^Gt/Gt mice at 1 to 3 days of age using a 31-G thin-walled needle.

Tissue clearing protocol and imaging

The brain and spinal cord were made transparent for imaging according to the improved CUBIC protocol (78). The tissues were dissected from mice following perfusion with PBS, pH 7.4, and 4% PFA in 0.1 M PB. The tissues were postfixed overnight in 4% PFA in 0.1 M PB at 4°C and washed with PBS. For 5 days, the samples were immersed in CUBIC-L [10% polyethylene glycol mono-p-isooctyphenyl ether (12969-25, Nacalai Tesque, Kyoto) and 10% N-butyldiethanolamine (B0725, Tokyo Chemical Industry, Tokyo) in water] with shake at 37°C. CUBIC-L was exchanged at 2 days after the immersion. The samples were then washed three times with PBS at room temperature for 2 hours and then stained for 5 days at room temperature with rat monoclonal anti-GFP antibody labeled with Alexa Fluor 647 (1:100; D153-A64, MBL) in buffer (0.5% Triton X-100, 0.25% Casein, and 0.01% sodium azide). After staining, the samples were washed three times with PBS at room temperature for 2 hours and postfixed with 1% PFA in 0.1 M PB at room temperature for 5 hours. Samples were immersed in 50% CUBIC-R [45% 2,3-dimethyl-1-phenyl-5-pyrazolone (D1876, Tokyo Chemical Industry) and 30% nicotinamide (N0078, Tokyo Chemical Industry), pH adjusted to approximately 8 to 9 with N-butyldiethanolamine (B0725, Tokyo Chemical Industry, Tokyo) in water] diluted in water at room temperature for 5 hours and then gently shaken in CUBIC-R at room temperature overnight. The samples were immersed in new CUBIC-R and kept until the microscopic observation. Fluorescent images were acquired with Light Sheet Fluorescence microscopes (MVX10-LS, Olympus).

Quantification and statistical analysis

VGluT1-positive varicosities were counted in the field of the anterior horn of spinal cord sections with MetaMorph software version 7.10.2. (Molecular Devices, CA, USA). Morphometric analysis was performed on at least three sections per mouse with no blinding. As a statistical analysis, one-way analysis of variance (ANOVA) was performed to compare each group. Statistical analysis was performed using the Easy R (EZR, Saitama Medical Center, Jichi Medical University, Japan) (79). Unless otherwise noted, sample size is the number of animals.

Supplementary Materials

The PDF file includes:

Figs. S1 to S14

Legends for movies S1 to S7

Other Supplementary Material for this manuscript includes the following:

Movies S1 to S7