Inherited Isolated Dystonia: Clinical Genetics and Gene Function

Abstract

Isolated inherited dystonia—formerly referred to as primary dystonia—is characterized by abnormal motor functioning of a grossly normal appearing brain. The disease manifests as abnormal involuntary twisting movements. The absence of overt neuropathological lesions, while intriguing, has made it particularly difficult to unravel the pathogenesis of isolated inherited dystonia. The explosion of genetic techology enabling the identification of the causative gene mutations is transforming our understanding of dystonia pathogenesis, as the molecular, cellular and circuit level consequences of these mutations are identified in experimental systems. Here, I review the clinical genetics and cell biology of three forms of inherited dystonia for which the causative mutation is known: DYT1 (TOR1A), DYT6 (THAP1), DYT25 (GNAL).

Electronic supplementary material

The online version of this article (doi:10.1007/s13311-014-0297-7) contains supplementary material, which is available to authorized users.

Introduction

Dystonic movements were defined in 1984 as a syndrome consisting “of sustained muscle contractions, frequently causing twisting and repetitive movements, or abnormal postures [1].” While this definition proved extremely useful, the explosion of interest in, and study of, movement disorders since that time has highlighted a variety of ways that dystonia can manifest clinically that are not captured by this definition. For example, dystonic movements are not always prolonged enough to cause twisting and turning, and may be intermittent, leading to tremulous-type movements. The original definition also failed to capture some characteristic features of dystonic movements, including their striking directionality, repetitive stereotyped nature, and their relation to action. These limitations led to the creation of an international panel of investigators that proposed an updated definition of dystonic movements to incorporate this new understanding: “Dystonia is a movement disorder characterized by sustained or intermittent muscle contractions causing abnormal, often repetitive, movements, postures, or both. Dystonic movements are typically patterned, twisting, and may be tremulous. Dystonia is often initiated or worsened by voluntary action and associated with overflow muscle activation [1].”

Among the terms used to describe different types of abnormal movements, “dystonia” is perhaps the most confusing, since it refers to both a type of abnormal movement (described above) and a disease entity [2]. As a type of abnormal movement, dystonia can occur in a wide range of neurological disorders that damage the motor system, ranging from Parkinson disease in the aged to metabolic diseases that manifest in the first years of life. When it occurs in the context of underlying disease, dystonia is typically accompanied by signs and symptoms characteristic of the associated illness. In contrast, when used to describe a disease entity, “dystonia” refers to an illness in which dystonia is the only neurological sign, with the exception of tremor, and is not accompanied by obvious central nervous system (CNS) injury. These two clinical contexts in which dystonia appears have been termed “primary” (isolated dystonia, no overt CNS lesion) and “secondary” (dystonia in the context of other CNS-damaging insult). However, these terms lack clarity and precision because they incorporate both clinical-phenomenological and etiologic features into single terms (e.g., primary dystonia denotes both isolated dystonia and the absence of any CNS lesion or exogenous cause). To overcome this limitation, a new nomenclature has been proposed with separate axes to describe the clinical (“Axis I”) and etiological (“Axis II”) features. In this new nomenclature, the Axis I clinical terms are “isolated” and “combined,” and imply nothing about etiology. The Axis II etiological categories include inherited, acquired (e.g., brain injury) and idiopathic. This review focuses exclusively on the genetics and cellular mechanisms of the three forms of inherited isolated dystonia for which the causative gene has been identified and validated extensively: DYT1 (caused by mutations in TOR1A encoding the AAA + protein torsinA), DYT6 (caused by mutations in THAP1 encoding transcription factor THAP1) and DYT25 (caused by mutations in GNAL encoding the G-protein Gα_olf). Variations in CIZ1 [3] and ANO3 [4] have been reported more recently in subjects with isolated dystonia. However, the evidence in favor of a pathogenic role for these genes is not yet sufficiently conclusive, so they will not be considered further in this review (see [5] for further discussion of these genes).

Clinical Genetics

The clinical characteristics of isolated dystonia of genetic or idiopathic etiology were most comprehensively described by Bressman and colleagues in Ashkenazi Jewish subjects [6, 7], but their findings are broadly applicable. These studies demonstrated that age-at-onset of dystonia is a critical factor predicting likelihood of genetic etiology and degree of spread from the initially affected body region. In general, this work demonstrates that compared to adult-onset disease (typically in 40s–50s), younger age-at-onset (particularly during childhood) is more likely to be of genetic etiology and spread to involve a greater number of body parts.

DYT1

DYT1 dystonia is a dominantly inherited disease that typically begins in school-aged children, with a mean age at onset of ~12 years. Characteristically, it initially affects an arm or leg, and subsequently spreads to involve additional limbs and/or the trunk. Up to 50% of affected subjects may ultimately develop generalized involvement. The tendency to generalize distinguishes DYT1 dystonia (and other childhood-onset genetic etiologies) from adult onset idiopathic isolated dystonia, which typically remains focal or segmental. DYT1 dystonia typically spares facial and laryngeal structures (present in ~11–14% of cases), and involves the neck in a minority of cases (present in ~25% of cases) [8–10]. This clinical picture of DYT1 dystonia is highly stereotyped, but exceptional cases and families have been reported, including with much earlier or later onset, onset in the larynx, or a phenotype of isolated writer’s cramp [11, 12].

In 1997, an in-frame 3 base-pair deletion (∆gag; ∆E) in the TOR1A gene encoding the AAA + protein torsinA was discovered as the cause of DYT1 dystonia [13]. While best described in Ashkenazi Jews, the DYT1 mutation has arisen de novo several times in diverse popoulations [14–16]. Clinical studies subsequent to mutation identification confirmed earlier findings based on haplotype analysis, showing that the mutation is roughly 30% penetrant, and exhibits considerable variation in symptom expressivity. Mutation-carrying subjects exhibit abnormalities of brain metabolism as assessed by fluorodeoxyglucose positron emission tomography analysis, regardless of clinical status [17]. Mutation carriers who do not develop dystonia by their early 20s almost always remain symptom-free for life, suggesting the presence of a developmental window of susceptibility during which torsinA function is critical for brain function. Based on these findings, diagnostic testing for the DYT1 mutation is recommended for subjects developing dystonia before age 26 years, unless other factors (e.g., a relative with early-onset dystonia) are present [10].

Several investigators have sought genetic modifiers of disease penetrance and expressivity, and examined the potential role of torsinA in other forms of dystonia. A torsinA coding polymorphism acting in trans reduces the risk of disease penetrance [18], and may alter the risk of non-DYT1 adult-onset idiopathic dystonia [19]. These effects may relate to the propensity of this polymorphism to cause torsinA to misfold [20]. Genetic background appears to confer only a modest influence on the clinical phenotype of the DYT1 mutation [21]. No studies have identified genetic or environmental factors that contribute to the striking variability in disease severity among manifesting subjects. Studies examining a possible association of torsinA sequence variation with adult-onset idiopathic isolated dystonia are mixed [22–26]. No evidence has been found linking TOR1A to susceptibility to causes of dystonia such as neuroleptic drugs or perinatal asphyxia [27]. Similarly, studies exploring a role for the DYT1 mutation in musician’s dystonia, one form of use-associated “occupational” dystonia, found either no association [28] or identified the mutation in a single individual [29]. Definitive identification of environmental factors that impact mutation penetrance or expressivity would require a prospective study of asymptomatic mutation carriers, but such a study has not yet been reported. In contrast to other inherited CNS diseases (e.g., spinocerebellar ataxias), the variable expressivity in DYT1 dystonia relates almost exclusively to the distribution and severity of dystonia; additional neurological findings, with the possible exception of tremor, do not complicate the clinical phenotype.

DYT6

DYT6 dystonia is also inherited dominantly. This form of dystonia was characterized and mapped initially in the Amish-Mennonite population [30, 31]. Similar to DYT1 dystonia, the disease in this population typically begins in childhood, with a mean age at onset of ~16 years. In contrast to DYT1 dystonia, however, roughly 50% of DYT6 subjects initially develop dystonia of cranial muscles (larynx, tongue, face) or the neck, with only ~5% initially manifesting in the leg (a common site of onset in DYT1 dystonia) [31]. While symptoms commonly spread and ultimately involve the leg(s) in many patients, leg involvement is typically mild, and disability in DYT6 dytstonia arises most commonly from craniocervical involvement. Penetrance in the Amish-Mennonite population is ~60%, and women appeared more frequently affected.

In 2008, two mutations in the THAP1 gene encoding the transcription factor thanatos-associated protein 1 (THAP1) were identified as the cause of DYT6 dystonia [32]. In this report, the Amish-Mennonite population were found to have an insertion-deletion (“indel”) mutation and a German family was found to have a missense mutation (F81L) that appeared to impair DNA binding. Subsequently, a large number of different mutations in THAP1 have been identified in a wide range of ethnicities [33–37], linking this gene to a far larger number of inherited cases of dystonia than had been suspected previously. The number of reported mutations led to the creation of a locus-specific database (http://www.umd.be/THAP1/{Blanchard), which by 2011 already included 56 different families, most of which have unique mutations. THAP1 mutations span the gamut from those that lead to very early truncations (e.g., at amino acid 3) to missense, but no genotype-phenotype correlation is apparent. In addition to the typical dominantly inherited mutations, two families with apparently recessively inherited mutations have been described [38, 39], furthering the evidence for a loss-of-function mechanism in DYT6 dystonia. The clinical characteristics of the many non-Amish-Mennonite subjects described subsequent to the initial discovery of THAP1 conforms to the initially described clinical phenotype, with either initial involvement in or frequent spread to craniocervical regions, though it has become apparent that the age at onset may extend into the late 40s [35]. Cases of focal dystonia (torticollis and blepharospasm) have been found to harbor mutations in THAP1 [36, 40], but this an uncommon manifestation of DYT6 dystonia [41], and genotype-phenotype correlations demonstrates that the THAP1 variations in these cases are predicted to be benign [42]. It is also possible that these patients will ultimately develop segmental disease. Despite the differences in pattern of involvement between DYT1 and DYT6 dystonia, intraoperative microelectrode recordings exhibit a common electrophysiological signature [43], indicating shared mechanism at the level of neural circuits.

DYT25

The identification of TOR1A and THAP1 mutations were major advances that spawned a range of exciting research effort focused on identifying molecular and cellular underpinnings of dystonia. Yet these genes cause isolated dystonia that is primarily of childhood-onset; isolated dystonia much more commonly begins in adulthood. Against this backdrop, the identification of GNAL mutations (DYT25) is particularly exciting [44], being the first gene linked to adult-onset focal/segmental dystonia—a condition commonly seen by movement disorders neurologists. Indeed, the discovery of this gene will allow studies exploring potential links between childhood- and adult-onset dystonia—an important question in dystonia research.

Just since 2012, 5 independent groups have identified GNAL mutations in patients with adult-onset dystonia, typically beginning in or involving the neck. Fuchs and colleagues initially reported GNAL mutations in 8 families, each of which had a unique variant. These mutations included nonsense, frameshift, missense and deletion mutations [44]. The average age at onset was ~31 years in the 28 patients, most of whom (82%) had neck onset or eventual neck involvement (93%). Dystonia remained focal in about half of these patients, with the rest showing spread to contiguous regions (e.g., developing segmental dystonia). Subsequent reports [45–49] demonstrated yet additional GNAL mutations in patients with a similar clinical phenotype (i.e., adult-onset, cervical predominant). While the age at onset is typically in the fifth to sixth decade, GNAL mutations lead to a rather broad range of age at onset, spanning from age 7 to 54 years. These reports broadened the ethnic diversity in which mutations are found, which now includes Caucasian, Asian and African American subjects. Interestingly, mutations in TOR1A, THAP1 and GNAL can all be found within one ethnic group—as demonstrated in Amish-Mennonites—emphasizing the genetic heterogenetity of dystonia even in narrowly defined populations, and the importance of not assuming genetic etiology based upon ethnicity [48].

Gene Function

The brains of subjects with isolated dystonia do not exhibit overt histopathological change, and subtle abnormalities (e.g., those recently reported in cerebellum [50], or others recently reviewed [51]) are of uncertain significance or yet to be independently validated. This situation creates conceptual challenges distinct from those encountered in neurodegenerative disease, where the cellular effect to be studied—cell death—is defined clearly. Consequently, much of the research on the genes linked to isolated dystonia has focused on defining the cellular pathways in which they participate, and the consequences for these pathways provoked by pathogenic mutations. There has also been considerable effort at manipulating isolated dystonia genes in vivo to create a mouse model of the disease. Here, I focus almost exclusively on the cell biological aspects of isolated dystonia genes, referring to results from animal models only where they bear directly on the cell biological aspects of gene function, including whether pathogenic mutations act via a gain- or loss-of-function. Comprehensive reviews of human pathophysiological studies of primary dystonia [2] or the behavioral and electrophysiological features of mouse models of isolated dystonia, including synaptic alterations in torsinA mutant mice [52], are available elsewhere. The first discovered cause of isolated inherited dystonia, mutations in TOR1A encoding torsinA (DYT1), were identified more than a decade before the next identified gene, THAP1 (DYT6). Consequently, most work has been done on torsinA, the review of which will comprise the majority of the following discussion.

TorsinA (DYT1)

TorsinA Function

TorsinA is expressed widely in neuronal and non-neuronal tissues and is a member of the AAA+ (ATPase associated with a variety of cellular activities) protein family. AAA+ proteins typically function as oligomers and use the energy of ATP hydrolysis to disassemble protein complexes [53]. Within this family it is related closely to the ClpB/Hsp104 proteins, which protect cells from denaturant stress by unfolding damaged proteins [54], and can regulate prion protein conformation [55]. AAA+ proteins contain a conserved ATPase domain spanning ~250 residues, which allows them to derive energy from ATP hydrolysis. These proteins typically assemble into ring-shaped oligomers (often hexamers) and function via a cycle of substrate binding ➙ non-covalent substrate modification ➙ substrate release, thus acting as a type of chaperone. This cycle is typically linked to ATP binding and hydrolysis: the binding of ATP leads to a high affinity substrate interaction, and modified substrate is released following ATP hydrolysis. In fact, experimental mutations in the AAA+ domain that prevent ATP hydrolysis often cause AAA+ proteins to lock onto substrates (so-called “substrate trap” E/Q mutations [53]). AAA+ proteins modify substrates in a wide variety of cellular contexts. For example, the AAA protein NSF (N-ethylmaleimide sensitive factor) dissociates the SNARE complex after the completion of membrane fusion [56]. AAA+ proteins utilize this chaperone-like activity in a diverse range of biological functions, including but not limited to membrane trafficking, powering cellular motors, and protein chaperoning [53]. Thus, the inclusion of torsinA in the AAA+ protein family indicates how torsinA likely acts on substrates, but provides little insight into the cellular role of torsinA.

Clues to the role of torsinA come from its localization in the endoplasmic reticular/nuclear envelope (ER/NE) endomembrane space. TorsinA resides in the lumen of the ER/NE endomembrane system as a peripheral membrane protein glycoprotein [57, 58]. The ER is an intracellular membrane compartment through which all membrane localized or secreted proteins must traffic. The ER has several well defined subdomains, which include the smooth and rough ER, and the NE, with which it is continuous. The subcellular localization of torsinA and identification of its interacting proteins within the ER/NE space indicates that it functions in both major subcompartments. Exogenously expressed wild type torsinA resides predominantly in the ER but concentrates in the NE to a somewhat greater extent than typical ER proteins [59]. Multiple laboratories have observed that artificially engineering a “substrate trap” mutant of torsinA that prevents it from dissociating from protein partners causes it to accumulate abnormally in the NE [59–61]. Moreover, torsinA interacts with inner nuclear membrane-localized protein lamina-associated polyprotein 1 (LAP1, also known as torsin-1A-interacting protein 1) and the main ER-localized protein lumenal domain like LAP1 (LULL1, also known as torsin-1A-interacting protein 2) [62]. Both LAP1 and LULL1 have been shown to function with torsinA. Loss of torsinA function causes characteristic abnormalities of nuclear membrane structure (reviewed below) and similar abnormalities occur in LAP1 knockout mice [63, 64]. Moreover, binding of either LAP1 or LULL1 greatly stimulate the ATPase activity of torsinA [65]. These data indicate a novel situation in which torsinA function can be independently regulated in different domains of the ER/NE space by LAP1 (at the NE) and LULL1 (in the main ER).

The precise biological role that torsinA plays within the ER/NE space is not well understood, but several themes have emerged (Fig. 1). Within the ER, several reports have pointed to a role for torsinA in protein trafficking or protein quality control. TorsinA is reported to participate in ER-associated protein degradation (“ERAD”)—a process whereby misfolded ER proteins are exported from the ER space and degraded [66]. Studies in Caenorhabditis elegans [67] also point to a role in protein quality control pathways. Other studies indicate that torsinA regulates the trafficking of polytopic membrane proteins, including the dopamine transporter [68], or ER-based mechanisms controlling protein secrection [69, 70]. These pathways are tightly interrelated, so these different findings may well reflect a core function of torsinA that manifests in a variety of ways depending upon the experimental approach. A recent study exploring torsinA function in mammalian neurons in vivo using a novel set of torsinA mouse mutants is consistent with these findings in in vitro systems and lower organisms. This work demonstrates that loss of torsinA function causes activation of ER stress pathways and neurodegeneration in a set of discrete set of sensorimotor brain regions linked previously to the pathophysiology of DYT1 dystonia [71].

Fig. 1.

Summary of TorsinA Function and Effects of Loss-of-Function. TorsinA is a AAA + ATPase that likely exists as a hexamer. The DYT1 mutation impairs torsinA ATPase activity, and may also exert a dominant negative effect on wild type torsinA. TorsinA resides in the endoplasmic reticulum/nuclear envelope (ER/NE) luminal space. The major known functions, interacting proteins, and consequences of loss-of-function in the ER (blue box) and NE (yellow box) are listed

A growing body of work also highlights potential roles for torsinA within the NE. The potential importance of NE-localized torsinA function to DYT1 dystonia is underscored by the finding that ΔE-torsinA abnormally concentrates at the nuclear membrane [59–61]. While the specific function of NE-localized torsinA is unclear, one possibility is regulation of the connections between proteins that tether the nucleus to the cytoskeletal network [72, 73]. Nucleo-cytoskeletal (N-C) connections are critical for the control of nuclear movements, and such movements are essential for several CNS developmental events, including neurogenesis, neural tube closure, and neural migration [74]. Neurons are highly polarized cells, so nuclear positioning is likely to be a critical, if poorly understood, aspect of normal function. N-C connections depend upon nesprin and SUN proteins. Nesprins are outer nuclear membrane proteins that participate in the N-C link by binding to the cytosolic cytoskeletal network. Nesprins are anchored to NE through binding to SUN protein family members, which themselves are localized to the inner nuclear membrane. TorsinA has been reported to interact with SUN1 [75] and Nesprin-3, which itself is mislocalized in murine torsinA null fibroblasts and DYT1 patient fibroblasts. Overexpression of torsinA in a neuronal cell line inhibits neurite outgrowth [76], an effect that may relate to its interactions with nesprin-3 [77].

TorsinA has also been linked to striking changes in nuclear membrane morphology, which may relate to the N-C coupling machinery. Loss of torsinA function in vivo causes the formation of membranous “blebs” that emerge from the inner nuclear membrane, and similar structures are seen following overexpression of a “substrate-trap” E/Q version of torsinA (see above) [61], in vivo deletion of LAP1 [64], and disruption of N-C connections [78]. These membranous blebs appear similar to those formed during the egress of herpesvirus— a process in which torsinA has been implicated [79]. Viruses typically act by usurping normal cellular processes, suggesting a function for trans-membrane nuclear transport independent of the normal nuclear pore-based route. In drosophila, torsinA has been linked to the formation of such structures implicated in the transport of ribonucleoprotein granules [80, 81], whereas in mouse these torsinA-related structures have been linked to the transport of nuclear ubiquitinated nuclear proteins [71].

The ER and NE effects of torsinA are not mutually exclusive, and multiple abnormalities of the ER/NE system likely contribute to the cellular dysfunction underlying DYT1 dystonia, especially considering the presence of functionally validated torsinA-activating proteins in both compartments.

Effect of the DYT1 Mutation

A seminal question of disease pathogenesis is determining whether the DYT1 mutation acts via a gain- or loss-of-function mechanism. Early studies showed that the ΔE-mutation does not appear to grossly alter torsinA protein stability or solubility [57]. Further, wild type and ΔE-torsinA show no difference in distribution on a sucrose gradient; both forms are found in a range of fractions, suggesting that the ΔE mutation does not grossly disturb the ability of torsinA to multimerize [57]. Potentially consistent with a gain-of-function mechanism, multiple investigators [20, 57, 60, 82, 83] find that overexpressed DYT1-torsinA or another potentially disease-related variant [84] concentrate in ER-associated clumps that are largely or completely segregated from ER markers. These immunoreactive clumps of mutant torsinA immunostaining are not insoluble aggregates [57] and they appear to include whorled double-membrane structures [60, 82]. It is important to note that these structures only form at high levels of overexpression [20, 59,85] and have not been observed in patient or mouse tissues [71, 86, 87]. These points, and the fact that similar membranous structures form from the overexpression of ER membrane proteins [88], suggest that the formation of these structures may not reflect properties of torsinA in vivo at endogenous levels.

In contrast, accumulating evidence—including mouse genetic, biochemical and cell biological studies—indicates that the DYT1 mutation impairs torsinA function. TorsinA knockout and homozygous DYT1 knock in mice phenocopy each other [63], both displaying perinatal lethality and characteristic deformities of the neuronal nuclear membrane; these data genetically define ΔE as a loss-of-function mutation. Biochemical studies show that the DYT1 mutation impairs the ATPase activity of torsinA [65]. This effect appears related to an intrinsic effect on torsinA protein, as well as by impairing its interaction with LAP1 and LULL1 [65, 89]—proteins required for its enzymatic activity [65]. The DYT1 mutation also decreases the levels of torsinA protein—an effect observed in skin fibroblasts from DYT1 subjects and tissue from DYT1 knock-in mice [63, 90]. AAA+ proteins function as multimers, and additional evidence suggests that disease mutant torsinA may also act via a dominant-negative mechanism [59, 61, 68]. There are also numerous examples of cell biological effects (including several of those described above) observed when overexpressing wild type torsinA that are reduced or abolished by the DYT1 mutation e.g., [67–69, 91]. The recent creation of mouse mutants recapitulating the overt twisting movements that characterize dystonia provide further evidence for a loss-of-function mechanism [71]. This study demonstrated that conditional CNS deletion of torsinA or isolated CNS expression of DYT1 mutant torsinA (without wild type protein, e.g., Tor1a^ΔE/– in the CNS) both cause such overt movement abnormalities and a characteristic histopathology involving regions implicated in human DYT1 subjects. These findings show that the DYT1 mutation impairs torsinA function in an in vivo context, and link this hypofunction to dystonic-like twisting movements.

THAP1 (DYT6)

THAP1 is an atypical zinc finger protein with a highly conserved N-terminal domain, termed THAP, which defines a family of over 100 proteins in diverse species, including 12 proteins in humans. This metal-coordinating domain encodes a DNA-binding domain that recognizes an 11-nucleotide sequence within the promoters of target genes [92, 93]. Crystallization of a drosophila THAP protein indicates that the protein binds DNA through a bipartite interaction, implying that THAP1 dimerizes, as is typical of zinc finger transcription factors. Indeed, THAP1 has been demonstrated to dimerize, an interaction that requires a 13-amino acid region of its coiled-coil domain [94]. As noted above, there are a large number of mutations reported to occur across the whole protein [95], including some that are recessively inherited. These mutations include early truncations (e.g., at amino acid 3) and deletions that disrupt dimerization [94]. The consequence of many of the missense mutations is less clear, with some investigators finding that they disrupt DNA binding [32] whereas others do not observe this effect [96]. Some data suggest that missense mutations may disrupt the stability of the DNA binding domain, or potentially change the specificity of DNA binding [96]. Considered together, it seems overwhelmingly likely that DYT6 mutations act by impairing THAP1 function.

THAP1 is expressed widely throughout the brain. Similar to torsinA, THAP1 mRNA levels show marked changes during early postnatal development, attaining stable expression in mouse brain at about 2 weeks of age [97]. Very little is known about THAP1 target genes. Only 1 report [98], published prior to the discovery of the connection between THAP1 and DYT6, has explored this question, implicating cell cycle genes of the pRB/E2F pathway. In that report, either overexpression or knockdown of THAP1 inhibited the proliferation of endothelial cells. Other work suggested an interaction with Par-4 protein in nuclear PML bodies and found THAP1 overexpression to promote apoptosis in HeLa cells [99]. There are no reports of THAP1 function (or loss-of-function) in neurons, however, which will be essential to determine the relevance of these studies the pathophysiology of DYT6 dystonia.

Several reports have explored a connection between THAP1 and other dystonia proteins. Two reports demonstrate that THAP1 can bind the TOR1A promoter [100, 101]; both find that DYT6 mutations impair this association. One of these studies [100] demonstrated that THAP1 could enhance the expression of a luciferase reporter linked to the TOR1A promoter, but no differences in torsinA mRNA or protein were observed in fibroblasts derived from DYT6 dystonia subjects. Given its nuclear localization, it is plausible that THAP1 could interact with torsinA-related pathways, including via pRB or LAP1, which localize to the inner nuclear membrane by binding A-type lamins. DYT3 is a form of dystonia known as X-linked dystonia Parkinsonism, or “Lubag,” and is predominantly seen in Filipino men. THAP1 is also reported to bind O-GlcNAc transferase, a gene present in the DYT3 locus, but whether this is indeed the DYT3 gene remains to be determined [102].

DYT25 (GNAL)

There are several links between the dysregulation of dopamine signaling and dystonia, including dopa-responsive dystonia (DYT5) caused by mutations in the genes encoding GTP cyclohydrolase or tyrosine hydroxylase, tardive dystonia caused by drugs that block dopamine receptors, and the frequent occurrence of dystonic movements that complicate levodopa therapy in Parkinson disease. The relationship of these observations (and dopaminergic dysfunction) to primary dystonia has been uncertain, as subjects with primary dystonia do not typically respond to medications that modulate dopaminergic signaling. The discovery of GNAL mutations in DYT25 are the most direct link between dopamine signaling and primary dystonia (a concept reviewed in detail in Goodchild et al. [103]). GNAL encodes Gα_olf—a protein that plays a critical role in striatal dopamine function, particularly in transducing signals downstream of D1 dopamine receptors in the “direct” striatonigral projecting GABAergic

The dopamine receptors (and other G protein-coupled receptors) utilize heterotrimeric G protein complexes to transduce signals to downstream effector molecules subsequent to dopamine binding. Stimulation of D1-type receptors (D1 and D5) activates a stimulatory G protein complex, leading to upregulation of adenylate cyclase and cAMP generation, whereas stimulation of D2-type receptors (D2, D3, D4) activates an inhibitory G protein complex with opposite effects. GNAL encodes Gα_olf (so named because it was identified in olfactory neurons), a G protein that is ~80% identical in amino acid sequence with GαS—the predominant CNS stimulatory G protein [104]. While GαS mediates stimulatory responses throughout the brain, this protein is essentially replaced in the striatum by the homologous Gα_olf, which is expressed in both direct (dopamine receptor D1 expressing) and indirect (dopamine receptor D2 expressing) neurons, as well as in cholinergic interneurons [105, 106].

Most work has focused on the role of Gα_olf in D1 signaling (in “direct” pathway striatonigral neurons), but it also transduces signals downstream of adenosine 2A receptors (A2A) in “indirect” pathway striatopallidal neurons [105]; its function in cholinergic neurons is poorly defined, but potentially critical considering role of these cells in primary dystonia [107]. Characterization of Gα_olf null mice has established key roles for this protein in D1-mediated molecular and behavioral responses. Gα_olf null mice are hyperactive at baseline, implicating loss of Gα_olf with a hyperkinetic state [108]. These mice show marked deficiencies in locomotor and molecular responses linked to D1 activation, including a blunted increase in movement to D1 agonists or acute cocaine, and reduced activation of cAMP, PKA, ERK or c-fos [109–111] following D1 treatment. This link to D1 signaling is particularly intriguing considering that D1 hypersensitivity is strongly implicated in l-dopa induced dyskinesias in Parkinson disease in connection with abnormal plasticity [112]—a pathogenic mechanism thought to be important in the pathogenesis of primary dystonia [113–114]. The connection to D1 signaling is complex, however, as comparisons between heterozygous Gα_olf (representing the likely molecular situation in DYT25) and D1 dopamine receptor mice show several interesting differences at the molecular and behavioral level. For example, Gα_olf heterozygous mice show a decreased behavioral response and cAMP increase to cocaine or d-amphetamine, whereas similar behavioral changes are not observed in D1 receptor heterozygous mice [106, 111]. These and related differences (reviewed in detail in Herve [104]), and the role of Gα_olf in A2A signaling and in cholinergic neurons, highlight the need for further studies identifying which of these roles (or combination of effects) is key for dystonia pathogenesis.

Conclusion

There are several reasons for optimism that the next several years will witness accelerated progress in our understanding of the pathophysiology of inherited isolated dystonia, including the identification of cellular mechanisms or specific molecules that are rational therapeutic targets. The identification of GNAL provides a specific target of investigation for the large group of investigators actively studying striatal neurotransmission. The identification of THAP1 is likely to yield clear molecular pathways involved in DYT6, as straightforward and practical strategies exist for the identification of transcription factor target genes, and such studies are benefiting tremendously from rapid advances in genetic technologies (e.g., Chip-Seq). Moreover, the development of the first mouse model based on one of these genes (torsinA) with overt abnormal movements will enable a range of studies not previously possible. As these lines of work gain momentum, it will be particularly important and interesting to explore whether these genes function in a common molecular pathway. Presently, there is no convincing data to suggest this is the case, though future studies may reveal that distinct molecular effects may nevertheless yield a common cellular or circuit defect around which investigators could focus therapeutic efforts.