Diverse dystonin gene mutations cause distinct patterns of Dst isoform deficiency and phenotypic heterogeneity in Dystonia musculorum mice

Dystonia musculorum (dt) mice were originally characterized as spontaneously occurring mutants that display severe degeneration of sensory neurons in the dorsal root ganglion (DRG) and progressive motor symptoms such as dystonia-like movements and ataxia (Duchen et al., 1964). The causative gene for dt mice is dystonin (Dst), also known as bullous pemphigoid antigen1 (Bpag1) (Brown et al., 1995; Guo et al., 1995). Because of the large size of the Dst locus (encoding over 100 exons) and obvious motor phenotypes, many untargeted Dst mutant mouse strains have been established, including many spontaneously occurring mutants [Dst^dt (Duchen et al., 1964); Dst^dt-alb (Messer and Strominger, 1980); dt^Frk (Pool et al., 2005); Dst^dt-23Rbrc (Horie et al., 2016); dt-MP (Seehusen et al., 2016)], a transgene insertion-induced mutant [Dst^Tg4 (Kothary et al., 1988)] and chemically induced mutants [Dst^dt-33J (MGI: 1889074); Dst^dt-36J (MGI: 2681971)]. Several mutants were also intentionally generated, including a gene-targeting knockout [Dst^tm1Efu (Guo et al., 1995)], a gene-trap mutant [Dst^Gt(E182H05) (Horie et al., 2014)] and nuclease-mediated mutants [Dst^em13Dcr (MGI: 6258920); Dst^em14Dcr (MGI: 6258923)]. However, a limited number of these genomic DNA mutations have been identified precisely, including Dst^Tg4, Dst^dt-alb, Dst^dt-23Rbrc, dt-MP, Dst^tm1Efu and Dst^Gt(E182H05). Different transcription start sites and alternative splicing produce several Dst isoforms in a tissue-selective manner. The three major Dst isoforms that are predominantly expressed in neural, muscular and epidermal tissues are Dst-a/Bpag1a, Dst-b/Bpag1b and Dst-e/Bpag1e, respectively, each of which has its own distinct cytoskeleton-binding domains (Leung et al., 2001; Horie et al., 2017). All known Dst isoforms include a plakin domain and belong to the plakin family of proteins. The Dst-e isoform, also known as BPAG1 or BP230, localizes to the inner plaque of hemidesmosomes (HDs), adhesion complexes between basal epithelial cells in the epidermis and basement membranes (Nievers et al., 1999; Walko et al., 2015). Since Dst knockout (Dst^tm1Efu) mice display structural abnormalities of HDs, subepithelial bulla and retardation of wound healing, Dst-e protein is also considered to maintain epidermal integrity (Guo et al., 1995). Further support for this idea is that the Dst-e isoform is well known to be a self-antigen in bullous pemphigoid (Künzli et al., 2016), a distinctly different disease that results in symptoms strikingly similar to those from epidermolysis bullosa simplex (EBS).

Loss-of-function mutations in human DST are related to hereditary sensory and autonomic neuropathy type VI (HSAN-VI) and/or the skin blistering disease, EBS. Patients with HSAN-VI harbor mutations in DST-a and suffer from reduced sensation to pain, touch, vibration, and autonomic disturbances such as reduced sweating, absent pupillary light reflexes, cardiovascular dysregulation and gastrointestinal dysmotility (Edvardson et al., 2012; Manganelli et al., 2017; Fortugno et al., 2019). Patients with EBS carrying loss-of-function mutations in DST-e display skin blisters (Groves et al., 2010; Takeichi et al., 2015; He et al., 2017; Turcan et al., 2017). Because the DST locus generates several DST isoforms, it is assumed that the heterogeneity of DST-related diseases may be due to mutations that affect a single transcript or multiple transcripts among DST isoforms. Although EBS and HSAN-VI have been described as distinct pathologies in most cases, a recent report describes a mixed case: a patient harboring the compound heterozygous mutation within the specific exon in DST-a and an exon common to both DST-a and DST-e, suffering from both neurological and skin manifestations (Cappuccio et al., 2017). It has also been reported that patients with DST-a2-specific mutations show late-onset HSAN-VI (Manganelli et al., 2017). Therefore, to understand the diverse spectrum of DST-related diseases, it is important to investigate phenotypic heterogeneity in dt strains.

In this study, we have characterized the deficiencies for Dst isoforms and phenotypic heterogeneity in two Dst mutants: Dst^dt-23Rbrc and Dst^Gt mice. In the Dst^dt-23Rbrc allele, a nonsense mutation was identified in the plakin domain, which is a common domain in all Dst isoforms (Horie et al., 2016). In the Dst^Gt allele, the gene-trap construct was inserted within the coding region of the actin-binding domain (ABD) at the N-terminus shared by Dst-a and Dst-b (Horie et al., 2014), but not contained in Dst-e. Our results indicate that diverse mutations within the Dst locus cause distinct loss-of-function patterns among Dst isoforms and distinct pathological outcomes in neural and cutaneous tissues. These results demonstrate phenotypic heterogeneity in dt mice and DST-related diseases and substantiate the necessity for additional studies of Dst mutations.

The three major Dst isoforms are predominantly expressed in neural, muscular and epidermal tissues (Dst-a, Dst-b and Dst-e), all of which have distinct cytoskeleton-binding domains (Fig. 1A) (Leung et al., 2001; Horie et al., 2017). Each Dst isoform contains various domains, including cytoskeleton-binding domains: plakin repeat domains (PRDs), ABDs composed of calponin homology (CH) domains, coiled-coil rods (CC-ROD), Gly-Ser-Arg (GSR) repeats, EF-hand (EFh) domains, GAS2-related (GAR) domains, and end-binding protein 1 and 3 (EB1/3) domains (Künzli et al., 2016). Dst-a and Dst-e isoforms are mainly expressed in neural and cutaneous tissues, respectively. Quantitative polymerase chain reaction (qPCR) analysis was performed to quantify the expression levels of Dst transcripts in neural and cutaneous tissues from Dst^dt-23Rbrc and Dst^Gt mice. Dst-a mRNA from brain extracts were significantly reduced in both Dst^dt-23Rbrc and Dst^Gt homozygous mice relative to wild-type (WT) mice (Fig. 1B,C). Conversely, a significant reduction in Dst-e mRNA in skin extracts was observed in Dst^dt-23Rbrc homozygous mice (Fig. 1D) but not in Dst^Gt homozygous mice (Fig. 1E).

In the N-terminus of Dst-a and Dst-b, structural diversity is generated by three different promoters, resulting in six different isoforms, termed Dst-a1, -a2 and -a3 and Dst-b1, -b2 and -b3 (Fig. 1A) (Jefferson et al., 2006). Dst-a1, -a2 and -a3 are suggested to display distinct subcellular localization and functions (Jefferson et al., 2006). In the brain of Dst^Gt mice, there was a remarkable decrease in the expression levels of Dst isoforms 1 and 2 compared with those in WT mice, whereas there was no significant change in the expression of Dst isoform 3 (Fig. 1F-H). The selective deficiency of Dst-a1 and Dst-a2 isoforms in the brains of Dst^Gt homozygous mice was expected because the gene-trap cassette is located downstream of Dst-a1 and Dst-a2 transcription initiation sites and upstream of the Dst-a3 promoter. In the brains of Dst^dt-23Rbrc homozygous mice, there was a decrease in the expression levels of all Dst isoforms compared with those in WT (Fig. S1).

In situ hybridization was performed to examine the distribution of Dst transcripts in neural and skin tissues. We used a Dst-plakin probe, which detects all Dst isoforms, for accurately comparing Dst expression levels for all isoforms across tissues. In the nervous system of control mice, Dst mRNA is strongly expressed in neurons in the spinal cord and sensory neurons in the DRG (Fig. 2A,D). In both Dst^dt-23Rbrc homozygous mice (Fig. 2B,E) and Dst^Gt homozygous mice (Fig. 2C,F), Dst expression in these neurons was markedly reduced compared with that in control mice. In the skin of control mice, Dst mRNA was detected on epidermal cells (Fig. 2G). Dst mRNA was diminished in the epidermis of Dst^dt-23Rbrc homozygous mice (Fig. 2H), but was unaltered in Dst^Gt homozygous mice (Fig. 2I). Data from our in situ hybridization analyses are consistent with our qPCR data: Dst^dt-23Rbrc homozygotes display reduced expression in both neural tissue and skin, whereas Dst^Gt homozygotes have reduced expression only in neural tissue, not in the skin. Next, we used a Dst-SR probe, which hybridizes with the spectrin repeats of Dst-a and Dst-b isoforms, but not with those of the Dst-e isoform. A similar distribution pattern of Dst mRNA was observed in the spinal cord and DRG (Fig. S2A-F). Dst mRNA in the epidermis was below detectable levels using the Dst-SR probe for all genotypes (Fig. S2G-I).

To detect Dst protein expression in the brain and skin tissue, we performed western blot analysis using a rabbit polyclonal anti-Dst antibody, which recognizes the shared plakin domain. In WT brain extracts, several bands were detected. These bands were not detected in brain extracts of Dst^dt-23Rbrc homozygous mice and only faintly detected in those of Dst^Gt homozygous mice (Fig. 3A). In the skin of WT and Dst^Gt homozygotes, a single Dst-e band was detected. The band was undetectable in the skin of Dst^dt-23Rbrc homozygotes (Fig. 3B).

A lack of sensory and autonomic nerve fibers has been observed in the skin of patients with HSAN-VI that harbor mutations involving DST-a but not DST-e (Manganelli et al., 2017). We previously reported that DRG sensory neurons of Dst^dt-23Rbrc and Dst^Gt homozygous mice undergo neurodegeneration (Horie et al., 2014, 2016). To investigate dt skin pathology, cutaneous innervation in postnatal dt mice was first assessed by immunohistochemistry (IHC) using anti-class III beta-tubulin (TUBB3) antibody (TuJ1) and anti-PGP9.5 antibody, which stain small nerve fibers in the skin (Lauria et al., 2004). In the control mice, TuJ1- and PGP9.5-immunoreactive sensory nerve fibers were detected in the epidermis (Fig. 4A,D,G; Fig. S3A). In both Dst^dt-23Rbrc and Dst^Gt homozygous mice, TuJ1- and PGP9.5-immunoreactive fibers were almost absent at week 3 (Fig. 4H,I; Fig. S3B,C). TuJ1-immunoreactive nerve fibers were observed at 1 week (Fig. 4B,C); however, there were fewer nerve fibers in Dst^dt-23Rbrc homozygous epidermis (Fig. 4B) than in Dst^Gt homozygous epidermis. Over time, TuJ1-immunoreactive fibers gradually decreased in week 2 in both Dst^dt-23Rbrc and Dst^Gt homozygotes (Fig. 4E,F). Loss of TuJ1-immunoreactive fibers in the epidermis was statistically analyzed by two-way analysis of variance (ANOVA) (genotypes and postnatal stages) and one-way ANOVA (genotypes) in each postnatal stage (Fig. 4J). Two-way ANOVA revealed statistically significant differences between WT and both dt mutants (control versus Dst^dt-23Rbrc homo, P<0.01; control versus Dst^Gt homo, P<0.01), but differences between Dst^dt-23Rbrc homozygotes and Dst^Gt homozygotes were not significant (P=0.09). Furthermore, the decrease in nerve fibers was significant between weeks 1 and 2 (P<0.01) and weeks 2 and 3 (P<0.05). One-way ANOVA at 1 week showed significantly fewer nerve fibers in Dst^dt-23Rbrc homozygotes than in the control (P<0.05), whereas no significant difference was observed between the control and Dst^Gt homozygotes (P=0.87). Statistically significant differences between the control and both dt mutants were observed at weeks 2 and 3 (control versus Dst^dt-23Rbrc homo, P<0.01; control versus Dst^Gt homo, P<0.01; one-way ANOVA). The autonomic nerve fibers around the sweat glands in the footpad were remarkably decreased in both Dst^dt-23Rbrc and Dst^Gt homozygous mice (Fig. S4B,C) compared with control mice (Fig. S4A) at 3 weeks of age. These results indicate that both sensory and autonomic nerves undergo degeneration leading to skin denervation in Dst^dt-23Rbrc and Dst^Gt homozygous mice.

Because mutations in DST-e result in skin fragility and blistering in patients with EBS (Groves et al., 2010), we investigated skin blisters by histological analyses. Hematoxylin and Eosin (HE) staining demonstrated that the morphological arrangement of the back skin was largely intact in Dst^Gt homozygotes and most Dst^dt-23Rbrc homozygotes (Fig. 5A-C). Skin integrity was also investigated by keratin immunostaining (Fig. 5D-F). The frequency of blistering in the skin of Dst^dt-23Rbrc homozygotes was as low as that in the control and not significantly different (control mice, 0.49±0.60%, n=3; Dst^dt-23Rbrc homozygotes, 0.82±1.54%, n=4; P=0.36, Student's t-test). These data are consistent with a previous study, which reported that Dst knockout mice have less severe blistering in haired skin than in tail skin (Guo et al., 1995). The frequency of blistering in the Dst^Gt homozygous skin was also similar to that in the control (control mice, 0.29±0.50%, n=3; Dst^Gt homozygotes, 0.17±0.15%, n=3; P=0.37, Student's t-test).

Dst-e encodes one of the structural components of HDs, which are adhesion complexes between basal keratinocytes and basement membranes. In HDs, Dst-e localizes to the inner plaque together with plectin (Fig. 6A) (McMillan et al., 2003; Walko et al., 2015). Dst-e and plectin interact with keratin 5/keratin 14 intermediate filaments (IFs), tethering them to the inner plaque of HDs. Dst-e and plectin also form links with transmembrane proteins α6β4 integrin and collagen XVII (Bpag2/BP180; also known as Col17a1) in HDs, with the latter two molecules largely visualized as the outer plaque. In the basement membrane, α6β4 integrin and collagen XVII interact with laminin 332, which extends as thin fibers across the electron-lucent lamina lucida (LL) and finally binds to collagen VII anchoring fibrils (AFs) in the lamina densa (LD). Collagen VII AFs then extend into the dermis (Fig. 6A). Ultrastructure of HDs was visualized using transmission electron microscopy (TEM). Electron-dense cytoplasmic outer and inner plaques were observed in the HDs of control mice (Fig. 6B,D). On the cytoplasmic side of the plasma membrane in the basal keratinocytes, keratin IFs visibly extended to HD inner plaques. In the skin of Dst^dt-23Rbrc homozygous mice, structural abnormalities in HDs were observed (Fig. 6C): the inner plaques of HDs were almost completely absent and the keratin IFs hardly invaded into the HDs, suggesting that the keratin IFs were not anchored underneath the HDs. However, the outer plaque of the HDs and connections across the basement membrane displayed no apparent abnormalities. These abnormalities of HDs in Dst^dt-23Rbrc homozygotes were observed throughout the epidermis. In contrast, Dst^Gt homozygotes displayed no apparent HD abnormalities (Fig. 6E).

In this study, we demonstrated that the expression of Dst isoforms is differentially disrupted in Dst^dt-23Rbrc and Dst^Gt mice. In Dst^dt-23Rbrc homozygotes, both Dst-a and Dst-e transcripts were reduced; probably because of nonsense-mediated mRNA decay induced by a newly introduced stop codon (Figs 1 and 2; Figs S1 and S2). Nonsense mutation in the Dst^dt-23Rbrc allele results in a drastic reduction in Dst protein to below detectable levels in both neural and cutaneous tissues (Fig. 3). In Dst^Gt homozygotes, Dst-a1 and Dst-a2 transcripts are reduced, whereas Dst-a3 in the nervous system and Dst-e in the skin are unaffected (Figs 1 and 2; Fig. S1). The expression patterns of Dst isoforms account for the residual Dst protein expression patterns in Dst mutants (Fig. 3). TEM analyses of the skin revealed that only Dst^dt-23Rbrc homozygotes show a loss of keratin filament contacts underneath the HDs, and that there is no such abnormality in the skin of Dst^Gt homozygotes (Fig. 6). Our data suggest that Dst-e plays a critical role in HD integrity and that Dst^dt-23Rbrc mice are an excellent model for investigating disease mechanisms, not only for HSAN-VI but also for EBS. Our data describe important aspects of phenotypic heterogeneity in DST-related diseases.

The structures of Dst-a and Dst-b are quite similar, and are considered to function as cytoskeletal regulators (Dalpé et al., 1998; Ryan et al., 2012). Indeed, neurons derived from HSAN-VI patient induced pluripotent stem cells have abnormal morphologies, such as short neurites (Manganelli et al., 2017). In addition to the regulation of cell shape, intracellular functions of Dst have also been described; Dst knockdown with siRNA impairs vesicular transport and maintenance of the Golgi apparatus (Ryan et al., 2012; Poliakova et al., 2014). Such traffic functions are essential for neurons, including neurite outgrowth, axonal transport and synaptic transmission. Among the domains in Dst-a and Dst-b, a tubulin-binding domain in the C-terminus is critical because the C-terminal truncated mutations occur in patients with HSAN-VI (Edvardson et al., 2012). The ABD at the N-terminus of Dst-a and Dst-b is also critical because a point mutation allele has been reported in patients with HSAN-VI (Fortugno et al., 2019). Although the role of Dst-a in neuronal function has been well investigated, it remains an open question as to how much the Dst-b isoform in skeletal and heart muscles contributes to dt phenotypes of mutant mice (Horie et al., 2016). Therefore, we have generated and are analyzing Dst-b-specific mutant mice.

Disturbance of sensory and autonomic nervous systems has been commonly reported in patients with HSAN-VI (Edvardson et al., 2012; Manganelli et al., 2017; Fortugno et al., 2019). In both Dst^dt-23Rbrc and Dst^Gt homozygotes, neurodegeneration of sensory and autonomic nervous systems is observed (Fig. 4; Fig. S4). Degeneration of sensory neurons in the DRG has been most frequently described for dt mice (Duchen et al., 1964; Horie et al., 2014), suggesting that sensory functions are severely disrupted in dt mice. Disturbance of the autonomic nervous system is also a possible factor that alters the life span and phenotypic severity (Tseng et al., 2011). In addition, we assume that abnormal mastication accompanied by the degeneration of motor neurons in the trigeminal motor nucleus is another factor altering systemic conditions in dt mice (Hossain et al., 2018,b).

There is variation in the severity and onset of HSAN-VI manifestations: late-onset patients with HSAN-VI carrying mutations that selectively disrupt DST-a2 have been reported (Manganelli et al., 2017), whereas the first-described HSAN-VI family harboring frameshift mutations affecting all DST-a isoforms displayed more severe phenotypes, including infant death (Edvardson et al., 2012). In Dst^Gt homozygotes, Dst isoforms 1 and 2, but not isoform 3, are disrupted by gene-trap insertion in the genomic region encoding the ABD, whereas all Dst isoforms have a nonsense mutation in Dst^dt-23Rbrc homozygotes (Fig. 1; Fig. S1). In our observations, sensory denervation in the skin of Dst^dt-23Rbrc homozygotes appeared to occur earlier than that in the skin of Dst^Gt homozygotes (Fig. 4). Although functional differences between all three N-terminal Dst isoforms are not fully understood, our histological observations are consistent with a recent study suggesting that Dst-a3 plays a compensatory function in the neural phenotypes of dt^Tg4 homozygous mice, which have defects in Dst-a1 and Dst-a2, but still express Dst-a3 (Lynch-Godrei et al., 2018). Because transgenic restoration of Dst-a2 in dt mice ameliorates the degeneration of sensory neurons in the DRG and partially extends the life span (Ferrier et al., 2014), DST-a2 seems to be the most crucial isoform for neurological symptoms in HSAN-VI, whereas the other DST-a isoforms may have redundant roles. Although Dst^Gt homozygotes and Dst^dt-23Rbrc homozygotes show almost the same motor abnormality, we previously reported that the dystonic score of Dst^dt-23Rbrc homozygotes is slightly lower than that of Dst^Gt homozygotes (Horie et al., 2016). This observation suggests that factor(s) other than Dst-a3 isoform expression also affect the dt phenotype. One possible factor is genetic background. We used mutant lines with different mouse backgrounds: Dst^dt-23Rbrc homozygotes were on a mixed background of C57BL/6 and C3H/HeN, and Dst^Gt homozygotes were on a C57BL6 background. It is known that mouse background can affect the mutant phenotypes (Olds-Clarke and McCabe, 1982; Lenk and Meisler, 2014; Terumitsu-Tsujita et al., 2019).

Mutations in DST-e result in EBS with skin fragility and blistering (Groves et al., 2010; McGrath, 2015). In the first reported cases of EBS linked to DST mutations, the defects were limited to nonsense mutations in the CC-ROD, which is a DST-e-specific domain (Groves et al., 2010; Liu et al., 2012). These patients experienced generalized trauma-induced spontaneous skin blisters. Electron microscopy analyses of both mutations identified a lack of HD inner plaques. Keratin filaments extend to where the inner plaque should be, as though connected, but the presence of connection points is not apparent. Keratinocytes cultured from patients with EBS have adhesion and migration defects (Michael et al., 2014). Patients with EBS have fragile skin due to DST-e mutations; mild physical trauma, such as rubbing or scratching, and defects in wound healing processes cause skin blisters. In our analyses, Dst-e protein levels were undetectable in the skin of Dst^dt-23Rbrc mice (Fig. 3); blisters were rarely observed (Fig. 5). One possible reason for the lack of skin blistering in mutant mice is that the short lifespan of dt mice hampers long-term analysis. Dst-e-specific mutant mice should make it possible to analyze the mechanism of skin blistering in these mice. Because Dst knockout mice show a delay in cutaneous wound healing (Guo et al., 1995), it would be interesting to investigate the delay in wound healing in Dst^dt-23Rbrc mice in future studies. We also expect that dt-MP (deletion of exon 39 to intron 61; Seehusen et al., 2016) will have a skin phenotype (ultrastructural abnormality in HDs), as well as a neural phenotype, because they disrupt all Dst-a, Dst-b and Dst-e isoforms. The genes other than Dst-e that cause EBS are keratin 14 (Bonifas et al., 1991; Coulombe et al., 1991), keratin 5 (Lane et al., 1992) and plectin (Chavanas et al., 1996; Gache et al., 1996; McLean et al., 1996). Plectin (PLEC) encodes huge intermediate filament-binding protein and its mutation leads to EBS with muscular dystrophy (Chavanas et al., 1996; Gache et al., 1996; McLean et al., 1996). Conditional deletion of plectin in epidermal cells causes skin blistering in mice (Ackerl et al., 2007). By comparing the severity of EBS manifestation with their causative mutations, one should be able to identify genes and proteins that contribute to maintaining skin strength and integrity.

To understand the pathogenesis of HSAN-VI and EBS, it is important to identify gene mutations in patients with HSAN-VI, patients with EBS and various dt mutant models, and to determine their precise phenotypes. It may be useful to check for the existence of HD abnormalities in patients with HSAN-VI and dt mutants by TEM analyses to diagnose cutaneous manifestations, and to restrict the genomic region needed for accurate mutation analyses. In addition, we believe that conditional knockout and rescue experiments using the multifunctional Dst^Gt allele will be useful for identifying the neural circuits responsible for movement disorders, and organs or cell types involved in the systemic manifestations in dt mice, paving the way for a viable treatment of HSAN-VI.

Animal care and experimental protocols were approved by the Animal Experiment Committee of Niigata University and were carried out in accordance with the Guidelines for the Care and Use of Laboratory Animals of Niigata University (approved number SA00521). Mice were maintained at 23±3°C, 50±10% humidity, 12 h light/dark cycles and food/water availability ad libitum. Dst^dt-23Rbrc mice [RBRC01615, Mouse Genome Informatics (MGI) number: 6119708] (Horie et al., 2016) and Dst^Gt(E182H05) mice (MGI: 3917429) (Horie et al., 2014) were used in this study. The Dst^Gt(E182H05) allele was abbreviated as Dst^Gt. Homozygous Dst^dt-23Rbrc and Dst^Gt mice were obtained by heterozygous mating. The day of birth was recorded as postnatal day 0 (P0). Genotyping PCR for the Dst^Gt allele was performed as previously described (Horie et al., 2014). For genotyping PCR of the Dst^dt-23Rbrc allele, we used an improved PCR primer set, wherein the MnlI site within the primer sequence was disrupted as follows: dt23F3 primer, 5′-CCTGGCTATGCTCCAGGAAAT-3′ and dt23R3 primer, 5′-GCCACGCCATTAATCCAAGG-3′. PCR conditions and restriction fragment length polymorphism protocols were as previously described (Horie et al., 2016).

qPCR was performed as previously described (Hayakawa-Yano et al., 2017). Total RNA was extracted from the brain and back skin using an RNeasy Mini Kit (Qiagen, Hilden, Germany). One microgram of RNA template was used for cDNA synthesis with oligo (dT) primers. Real-time PCR was performed using a StepOnePlus Real-Time PCR system (Thermo Fisher Scientific, Waltham, MA, USA) and the following cycling conditions: 95°C for 2 min, followed by 40 cycles of 95°C for 15 s, 60°C for 40 s and 95°C for 15 s. Expression levels of Dst-a, Dst-e and Dst isoforms 1, 2, and 3 were analyzed using the ΔΔCT method. β-Actin (Actb) and glyceraldehyde-3-phosphate dehydrogenase (Gapdh) were used as internal controls to normalize the variability of expression levels for Dst-a and Dst-e. For semi-quantitative reverse-transcription PCR, the PCR reaction was performed using the same cDNA samples with the following cycling conditions: 94°C for 2 min, followed by 27-30 cycles of 94°C for 30 s, 58°C for 30 s and 65°C for 30 s. The following primers were used: Dst-a forward, 5′-AACCCTCAGGAGAGTCGAAGGT-3′ and Dst-a reverse, 5′-TGCCGTCTCCAATCACAAAG-3′; Dst-e forward, 5′-TGAGAATAGCAAACTTAGCGGGA-3′ and Dst-e reverse, 5′-CGGCCTCCTTAACTTTCGG-3′; Dst isoform 1 forward, 5′-TCCAGGCCTATGAGGATGTC-3′ and Dst isoform 1 reverse, 5′-GGAGGGAGATCAAATTGTGC-3′; Dst isoform 2 forward, 5′-AATTTGCCCAAGCATGAGAG-3′ and Dst isoform 2 reverse, 5′-CGTCCCTCAGATCCTCGTAG-3′; Dst isoform 3 forward, 5′-CACCGTCTTCAGCTCACAAA-3′ and Dst isoform 3 reverse, 5′-AGTTTCCCATCTCTCCAGCA-3′; Actb forward, 5′-GGCTGTATTCCCCTCCATCG-3′ and Actb reverse, 5′-CCAGTTGGTAACAATGCCATGT-3′; Gapdh forward, 5′-AGGTCGGTGTGAACGGATTTG-3′ and Gapdh reverse, 5′-TGTAGACCATGTAGTTGAGGTCA-3′.

For tissue preparations, mice were euthanized via intraperitoneal injection with pentobarbital sodium (100 mg/kg body weight), and then perfused with 4% paraformaldehyde (PFA) in 0.1 M phosphate-buffered solution (PB) (pH 7.4). The tissues were fixed by cardiac perfusion with 0.01 M phosphate-buffered saline (PBS) followed by ice-cold 4% PFA in 0.1 M PB (pH 7.4). Dissected tissues were immersed in the same fixative overnight. To cut 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 then dehydrated using an ascending series of ethanol and xylene washes, and embedded in paraffin (P3683; Paraplast Plus; Sigma-Aldrich, St Louis, MO, USA). Consecutive 10-μm-thick paraffin sections were cut on a rotary microtome (HM325; Thermo Fisher Scientific), mounted on MAS-coated glass slides (Matsunami Glass, Osaka, Japan) and air-dried on a hot plate overnight at 37°C. Paraffin sections were deparaffinized in xylene, rehydrated using a descending series of ethanol washes, and then rinsed in distilled water. For HE staining, sections were stained with Mayer's Hematoxylin (131-09665; Wako) for 10 min and with 0.2% Eosin Y in ethanol for 5 min.

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-PGP9.5/ubiquitin carboxy-terminal hydrolase L1 (UCHL1) antibody (1:1000; UltraClone, Isle of Wight, UK, purchased from Cosmo Bio Inc., Tokyo, Japan), mouse monoclonal anti-tubulin β3 (TUBB3) antibody (1:2000; RRID:AB_10063408, clone TuJ1; BioLegend, San Diego, CA, USA), mouse anti-keratin antibody (1:100; KL1; Immunotech, Marseille, France) and mouse monoclonal anti-neurofilament-M (NF-M) antibody (1:500; 1C8; Watanabe et al., 2006), all 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; 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, immunoreactivity was visualized in 50 mM Tris-HCl buffer (pH 7.4) containing 0.01% diaminobenzidine tetrahydrochloride and 0.01% hydrogen peroxide at 37°C for 5 min. Sections were then dehydrated through graded ethanols and xylene solutions and placed on coverslips with Bioleit (23-1002; Okenshoji, Tokyo, Japan). Digital images were taken with a microscope (BX53; Olympus, Tokyo, Japan) equipped with a digital camera (DP74, Olympus), and the TIF files were processed with Photoshop software (Adobe, San Jose, CA, USA).

In situ hybridization was performed on paraffin sections as described in previous studies (Takebayashi et al., 2000; Hossain et al., 2018a,b). Dst-plakin probe [GenBank accession number NM_001276764, nucleotides (nt) 2185-3396] and Dst-SR probe (GenBank accession number NM_001276764, nt 15994-17059; Horie et al., 2014) were used. The Dst-plakin probe recognizes Dst-a, Dst-b and Dst-e isoforms and the Dst-SR probe recognizes only Dst-a and Dst-b isoforms. After staining with Dst-plakin probe, spinal cord sections were subsequently immunostained with anti-NF-M antibody and other sections (DRG and back skin) were counterstained by Nuclear Fast Red.

Western blotting was performed as previously described (Zhou et al., 2018). Frozen brain and skin tissues were homogenized using a Teflon-glass homogenizer in ice-cold homogenization buffer (0.32 M sucrose, 5 mM EDTA, 10 mM Tris-HCl, pH 7.4, phosphatase inhibitor cocktail tablet; Roche, Mannheim, Germany), centrifuged at 1900 g for 10 min at 4°C and the supernatants were collected. The protein concentration was determined using bicinchoninic acid protein assay reagent (Thermo Fisher Scientific). Lysates were mixed with an equal volume of 2× sodium dodecyl sulfate (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 1.5-2 μg/μl and denatured in the presence of 100 mM dithiothreitol at 100°C for 5 min. SDS-polyacrylamide gel electrophoresis (PAGE) was performed with 20 μg per lane for brain samples or 30 μg per lane for skin samples on 5-20% gradient gels (197-15011, SuperSep™ Ace; FUJIFILM Wako Pure Chemical Corporation, Osaka, Japan) running at 10-20 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 h, blotted membranes were incubated with the following primary antibodies: rabbit polyclonal anti-Dst antibody (gifted from Dr Ronald K Leim; Goryunov et al., 2007), which recognizes the plakin domain of Dst, and mouse monoclonal anti-Actb antibody (1:2000; AB_2223041, clone C4; Merck Millipore). Each primary antibody was incubated overnight at 4°C. Then membranes were incubated with peroxidase-conjugated secondary antibodies for 1 h at room temperature: anti-rabbit immunoglobulin G (IgG) (1:2000; AB_2099233, Cat. #7074; Cell Signaling Technology, Beverly, MA, USA), anti-mouse IgG (1:2000; AB_330924, Cat. #7076; Cell Signaling Technology). Tris-buffered saline (10 mM Tris-HCl, pH 7.5, 150 mM NaCl) containing 0.1% Tween-20 and 10% skim milk was used for the dilution of primary and secondary antibodies, and Tris-buffered saline containing 0.1% Tween-20 was used as the washing buffer. Immunoreactions were visualized by ECL (GE Healthcare, Piscataway Township, NJ, USA) and a luminescence image analyzer (C-Digit; LI-COR, Lincoln, NE, USA).

For TEM analysis, tissue preparation was performed as previously described (Eady, 1985). Tissue blocks (back skin and foot sole skin) were fixed using 2.5% glutaraldehyde in 0.1 M PB (pH 7.4), cut into small pieces (∼3×3 mm), and further fixed in the same fixative over 1 day at 4°C. Next, samples were fixed with 1% osmium tetroxide in 0.1 M PB, dehydrated using a graded series of ethanol washes, transferred into propylene oxide and embedded in epoxy resin (Epon 812; TAAB, Berkshire, UK) for polymerization over 24 h at 60°C. Ultrathin sections with a thickness of 70 nm were prepared using an ultramicrotome (Ultracut N; Reichert-Nissei, Tokyo, Japan), stained with 1% uranyl acetate for 10 min, followed by 1% lead citrate for 5 min (Richardson et al., 1960), and observed under a TEM (H-7650; Hitachi, Tokyo, Japan) at an accelerating voltage of 80 kV.

To quantify qPCR data, Student's t-test was performed. Quantification of TuJ1-immunoreactive fibers was performed with MetaMorph software (Meta Series Software ver. 7.10.2; Molecular Devices, San Jose, CA, USA). The area of TuJ1-immunoreactive fibers within the epidermis was normalized by the surface length of the skin sections. Morphometric analysis was performed on three sections per mouse, with three or more mice per group. For statistical analysis, one-way and two-way ANOVA were performed to compare controls, Dst^dt-23Rbrc homozygotes and Dst^Gt homozygotes across postnatal 1, 2 and 3 weeks. Statistical analysis was performed using ANOVA4 on the Web (https://www.hju.ac.jp/~kiriki/anova4/).

We thank Dr Ronald K. H. Liem (Columbia University) for the anti-Dst antibody; Dr Hayato Ohsima and Dr Kotaro Saito (Niigata University) for the anti-PGP9.5 antibody; Dr Ryota Hayashi for discussions; and Dr Yukiko Mori-Ochiai, Mr Yuya Imada, Ms Satoko Yamagiwa and Mr Seiji Takahashi for technical assistance.