X-Linked Dystonia-Parkinsonism: Recent Advances

Abstract

Purpose of review

Our understanding of X-Linked Dystonia-Parkinsonism (XDP) has advanced considerably in recent years due to a wealth of new data describing its genetic basis, cellular phenotypes, neuroimaging features, and response to deep brain stimulation (DBS). This review provides a concise summary of these studies.

Recent findings

XDP is associated with a SINE-VNTR-Alu (SVA)-type retrotransposon insertion within the TAF1 gene. This element includes a hexameric DNA repeat expansion, (CCCTCT)_n, the length of which varies among patients and is inversely correlated to age of disease onset. In cell models, the SVA alters TAF1 splicing and reduces levels of full-length transcript. Neuroimaging data have confirmed previous neuropathology studies that XDP involves a progressive striatal atrophy, while further detecting functional alterations in additional brain regions. In patients exhibiting features of both dystonia and parkinsonism, pallidal DBS has resulted in rapid improvement of hyperkinetic movements, but effects on hypokinetic features have been inconsistent.

Summary

The discovery that XDP is linked to a polymorphic hexameric sequence suggests that it could share mechanisms with other DNA repeat disorders, while the transcriptional defect in cell models raises the possibility that strategies to correct TAF1 splicing could provide therapeutic benefit.

Introduction

X-Linked Dystonia-Parkinsonism (XDP) is a movement disorder which was first reported in the literature more than 40 years ago by Lee et al. [1], who described a dystonia-parkinsonism syndrome affecting males from the island of Panay, Philippines. In the many years since that first description, XDP has come to be recognized as a hereditary neurodegenerative disease thought to have arisen from a presumed founder mutation within the Panay population. During that time period, numerous studies have documented clinical observations in XDP patient cohorts, patterns of cell loss in limited numbers of XDP post-mortem brain samples, and the painstaking process of identifying the pathogenic gene variant. The culmination of those considerable efforts provided the foundation for much of what we currently know about XDP, but critical aspects of the disease have proven difficult to decipher.

The symptomatology of XDP that has been described in most published studies consisted of an unusual temporal evolution, involving hyperkinetic movements at early disease stages which shifted over time towards more hypokinetic features [2–4]. In these individuals, the first symptoms were typically a focal dystonia which generalized over the course of 5–10 years, as parkinsonian features emerged in parallel and appeared to predominate by 10 years on average, following the initial disease onset [3, 4]. During the dystonic phase there was often prominent involvement of the head and neck region, resulting in characteristic patterns of retrocollis, oromandibular dystonia, and tongue protrusion [4]. However, other studies reported XDP individuals exhibiting parkinsonism as the initial clinical manifestation [5], suggesting that the natural history of the disease may be heterogeneous and its full phenotypic spectrum has not yet been captured.

The neural substrates underlying these deficits have also not been completely defined. Neuropathological analyses of relatively few XDP post-mortem brain samples obtained from individuals who came to autopsy at different stages of disease detected an apparently progressive neuronal cell loss within the striatum [6–8], which appeared to reflect a selective dropout of striosomal medium spiny neurons (MSNs). To date there has not been any reported evidence of XDP-related degeneration of the substantia nigra, even in cases collected at advanced disease stages in which parkinsonian features predominated [7]. Moreover, despite the apparent similarity between the striatal lesion in XDP and the characteristic loss of MSNs which occurs in Huntington’s disease (HD), most clinical reports indicate that chorea is rarely observed in XDP individuals [4].

Yet until recently, the biggest obstacle in understanding the pathobiology of XDP has arguably been its genetics. Although the inheritance pattern suggested from the beginning that XDP was most likely an X-linked Mendelian disorder linked to a single gene variant within an isolate population, identifying that pathogenic sequence has been challenging. In 2003, Nolte et al. [9] published a landmark paper describing an XDP-specific haplotype, a discovery that resulted from intense efforts over many years to map the causal locus. That haplotype was subsequently confirmed and expanded by Makino et al. [10] in a separate XDP cohort. Together these two studies revealed that individuals with XDP share a set of seven variants: five single nucleotide substitutions, annotated as Disease-specific Sequence Change (DSC)-1,2,3,10,12; a 48-bp deletion; and a SINE-VNTR-Alu (SVA)-type retrotransposon insertion. The puzzling aspect of this haplotype was that it appeared to consistently remain intact during inheritance, i.e. there were no observed recombination events producing partial haplotypes in XDP individuals, nor were any of these variants ever detected in unaffected control subjects [9–11].

The seven sequence variants all cluster in and around the human TAF1 gene, which encodes TATA-binding protein (TBP)-Associated Factor-1 (TAF1), a core subunit of the TFIID complex which is part of the general transcriptional machinery [12, 13]. Three of the haplotype markers (DSC-10, 12, and the SVA) are localized within introns of TAF1, while the remaining four are present within an intergenic region 3’ to its terminal exon [9, 10]. This intergenic region has been designated as a Multiple Transcript System (MTS) based on the presence of multiple noncanonical exons which may be transcribed and spliced in various combinations, some of which may include transcripts derived from TAF1 exons [9, 10, 14]. Nevertheless, the combined presence of these variants in all reported XDP cases, as well as their positions in non-coding and/or unannotated gene regions, made it difficult to determine which, if any, may be pathogenic.

Using modern sequencing technologies to map the XDP causal locus

In the past five years, the pace of discovery in XDP has accelerated dramatically, fueled in large part by a massive coordinated effort from an international coalition of clinicians, investigators, and patient advocates. Publications resulting from this initiative have substantially advanced our understanding of XDP genetics. In one study, Aneichyk et al. [15] combined genomic analyses of a large XDP cohort with transcriptomic profiling of cell lines derived from a subset of these individuals. The DNA analyses incorporated multiple technologies to interrogate all classes of genomic sequence variation. The majority of probands shared an identical haplotype, presumed to be the founder, that spanned a broader segment of the X chromosome than previously reported and encompassed not only the seven known markers but 47 additional variants which segregated with disease in these individuals. There were also five recombination events detected in this cohort, creating at least seven derivative haplotypes which together narrowed the critical region shared by all probands to a segment of approximately 203.6 kb exclusive to TAF1. However, within this segment, all probands shared a total of 13 sequence variants, none of which fell within coding regions. Thus genomic analyses by themselves still did not pinpoint the causal variant.

The key to understanding the functional consequences of these variants was in the transcript structure of the shared region. TAF1 is a large gene, with at least 38 constitutive exons and numerous alternative exons, and previous attempts using PCR-based methods to map its many splice forms in various cell types produced conflicting results [10, 14]. Aneichyk et al. [15] assayed XDP and control fibroblasts, as well as neural cells differentiated from induced pluripotent stem cells (iPSCs) derived from these individuals. Using RNA sequencing to assemble the complete transcript structure of the region, they detected three XDP-related transcriptional defects, each of which involved a large TAF1 intron in which the SVA retrotransposon is inserted. In XDP neural progenitor cells, splicing at this intron was altered, resulting in aberrant truncated transcripts which terminated just proximal to the SVA insertion site. The initial segment of this intron was also aberrantly retained in mature mRNA, and transcription of downstream exons was reduced, thereby decreasing levels of the full-length transcript. Excision of the SVA by genome editing of XDP iPSC lines rescued these defects, restoring proper splicing and normalizing TAF1 transcript levels. A subsequent study of iPSCs derived from a different XDP cohort similarly reported that SVA ablation by genome editing increased TAF1 expression [16].

Another major advance was made possible by a detailed analysis of the SVA sequence itself. SVAs are one of many classes of retroelements within the human genome, and one of three (along with the Long Interspersed Nuclear Elements [LINEs] and Alu elements) still capable of mobilization from one genomic location to another [17, 18]. There are at least 3600 SVAs annotated within the human genome [19], yet only a small number appear linked to disease [20], and their highly homologous and repetitive sequences can complicate their characterization [21]. A recent study [22] obtained a complete sequence of the XDP-specific SVA from a single proband and compared that profile to the one reported in its original discovery [10]. The two were identical except for one domain: a hexameric sequence (CCCTCT)_n with a variable number of repeats. To assess the extent of this polymorphism, the hexamer was sized in a large number of probands, revealing that its length ranged from 35–52 repeats and showed a highly significant inverse correlation to the age of disease onset in these individuals. This observation represents the first direct link between sequence variation in XDP probands and disease manifestation, thereby establishing a causal role for the SVA in disease pathogenesis. Westenberger et al. [23] have since confirmed this correlation in an independent XDP cohort, while also observing in a subset of probands that hexamer length may modify the manifestation of dystonic vs. parkinsonian features.

These studies challenge previous conceptions of XDP in important ways. The genomic analyses performed up to this point had all indicated that the haplotype in XDP individuals is identical, but the variation detected within the SVA now reveals that not to be the case. Moreover, there has been a prevailing assumption that XDP is fully penetrant with an average age of disease onset typically within the fourth decade of life [3, 4], but studies of the hexamer suggest the picture to be more complicated [22, 23]. Based on these recent data, some individuals bearing the XDP-specific SVA with shorter repeat lengths may not develop clinically-relevant disease until later in life and/or follow different courses of progression. Defining the full contribution of SVA hexamer length to XDP pathogenesis will require a more detailed understanding of its dynamics during inheritance, its potential mosaicism among body tissues and particularly within the brain, and its possible influence on the natural history of disease. To the extent that other DNA repeat disorders have wrestled with similar questions, their experimental approaches may serve as valuable roadmaps for future XDP research.

The role of TAF1 in the central nervous system

The implication of current XDP studies is that disease pathogenesis may be due, at least in part, to a partial loss of TAF1 expression. But which TAF1 transcript(s) are responsible? The initial study by Makino et al. [10] identified a neuron-specific splice form of TAF1, designated nTAF1, which differs from the canonical transcript, cTAF1, by the addition of six nucleotides. Expression of nTAF1 in human post-mortem XDP striatal tissue, as measured by semi-quantitative reverse transcription-PCR (qRT-PCR), was decreased relative to levels in control tissue, though it should be noted that this finding was based on a single case. In iPSC-differentiated neural progenitors, nTAF1 expression was also decreased significantly in XDP vs. control cells when assayed by qRT-PCR [24], but transcriptome assembly in these same cell lines by RNA sequencing indicated that the levels of this transcript were extremely low and may be difficult to reliably quantify [15]. The latter study demonstrated instead that the primary consequence of the SVA-driven aberrant splicing was a significant decrease in the canonical transcript, cTAF1. Other analyses suggest that cTAF1 expression may also be moderately decreased in XDP fibroblasts and blood RNA compared to matched control samples [15, 23, 24, 25]. Distinguishing which TAF1 transcript(s) contribute to disease pathogenesis will ultimately require additional study of patient-derived biospecimens and, most importantly, post-mortem brain tissue. Yet these findings highlight how little is currently known about TAF1 protein isoforms, their differential distribution within the brain, their function(s) and targets in neural cells, and the physiological defects in neurons caused by their reduced expression.

To address these questions, it may be useful to consider in parallel the consequences of coding variation in TAF1, which a growing number of studies now suggest may be linked to intellectual disability [26–29]. Many of the probands bearing these variants have significant malformations at all levels of the neuraxis with severe clinical deficits present at birth. No study has yet performed any direct assessments to determine if these syndromes include movement deficits which overlap with the XDP phenotype. They may instead represent distinct clinical outcomes arising from different variants in the same gene, akin to how variation in ATP1A3 has been linked to rapid-onset dystonia-parkinsonism (RDP), alternating hemiplegia of childhood (AHC), and cerebellar ataxia, areflexia, pes cavus, optic atrophy, and sensorineural hearing loss (CAPOS) syndromes [30]. Nevertheless, these cases of TAF1 coding variants reinforce the notion that neural cells may be particularly vulnerable to perturbations in TAF1 protein, and identification of cellular pathways affected by these mutated forms may also provide clues as to sites of potential dysfunction in XDP.

Expanding the XDP phenotype using neuroimaging and clinical assessments

Just as recent molecular investigations have significantly advanced our understanding of XDP etiology, recent clinical studies are expanding our knowledge of its phenotypic manifestations. Perhaps most significant among these reports are neuroimaging studies indicating that the loss of striatal MSNs may not be the sole lesion in XDP [31–34]. Quantitative magnetic resonance imaging (MRI) of XDP individuals has detected not only striatal atrophy but also volume loss in the pallidum [31, 34], increased connectivity between the striatum and insular cortex [33], reduced cortical thickness in frontal and temporal cortices [34], grey matter pathology in cerebellum [34], as well as diffuse changes to white matter in multiple regions [31]. Although previous neuroanatomical studies did not detect a loss of nigrostriatal projections [7], there may be functional disturbances in this tract based on single-photon emission computed tomography (SPECT) imaging [32] and transcranial brain sonography [35]. Whether any of these imaging findings may serve as biomarkers remains to be determined and will require further investigation of larger cohorts coupled to detailed clinical assessments to stage disease progression.

In addition to neuroimaging, a series of neurophysiological studies have recently examined XDP-related signatures that may correlate with cognitive functions in these probands [36–38]. In contrast to motor symptoms, the nonmotor features of XDP are less well characterized, with only limited data available from surveys of relatively small cohorts and individual case studies [39–44]. A complete evaluation of nonmotor symptoms in XDP and their temporal progression will no doubt be a high priority for future analyses of the disease’s natural history and determination of prognostic biomarkers, and these objectives may be facilitated by the recent validation of a new XDP rating scale [45]. Such efforts to perform increasingly detailed clinical phenotyping of XDP patients are already underway, with some investigators reporting novel findings related to oculomotor abnormalities [46] and specific impairments in speech and swallow [47]. How these deficits track with disease progression is not yet known.

Looking ahead: the future of XDP treatment

In recent years there has been considerable progress in characterizing the efficacy of pallidal DBS in XDP. Following the initial clinical reports describing individual cases [48–52], prospective studies have summarized long-term outcomes related to neuromodulation in different XDP cohorts [53–55]. Although the patients in these groups were most likely at different disease stages at the time of intervention, the results demonstrated a consistent and relatively rapid improvement in dystonic symptoms as quantified by the Burke-Fahn-Marsden Dystonia Rating Scale (BFMDRS). However, pallidal DBS was less consistently effective at mitigating parkinsonian features based on scores from the Unified Parkinson’s Disease Rating Scale (UPDRS), with some patients exhibiting little to no improvement in these symptoms at the time of assessment. It should be recognized, however, that the application of DBS to XDP is still relatively recent and has involved only limited numbers of individuals. As more prospective studies continue to collect data from additional XDP patients undergoing DBS, its efficacy across the full spectrum of deficits may become more clear.

As with other forms of dystonia, the treatment options for XDP are generally focused on symptomatic management and currently consist of various combinations of neuromodulation, chemodenervation by botulinum toxin, and oral medications such as anticholinergic agents, benzodiazapines, baclofen, and levodopa [53–59]. Although this arsenal may currently seem small, the recent advances in molecular studies of XDP now offer hope that additional therapeutic strategies could someday be developed. Indeed, the discoveries of the SVA-mediated transcriptional defects [15] and the pathogenic repeat expansion [22] both suggest that XDP could ultimately benefit from approaches being developed to target diseases of RNA splicing [60, 61] and/or DNA repeat disorders [62, 63]. Bringing such possibilities to fruition will involve considerable work on all fronts, as we continue to learn more about the clinical manifestations of XDP, how they might be recapitulated in model systems, and how those models could be applied for drug discovery. Yet given the progress made in recent years, it is clear that this international coalition of clinicians, investigators, and patient advocates are more committed than ever to ensure that such translational research programs for XDP will soon become reality.

Conclusions

In recent years there have been important new studies which have revealed much about the molecular etiology and phenotypic expression of X-Linked Dystonia-Parkinsonism. These new data indicate that XDP is caused by a DNA repeat expansion within an intronic SVA-type retrotransposon insertion in TAF1. In patient cells, the SVA disrupts normal mRNA splicing to decrease levels of full-length TAF1 transcript. Within the brain, XDP is associated with progressive striatal atrophy but may also include functional disturbances in other regions as well. Treatment for XDP is currently limited to symptomatic management, but initial prospective studies have reported promising effects of bilateral deep brain stimulation in mitigating hyperkinetic features of the disease.

Key points.

• XDP is caused by a pathogenic DNA repeat expansion within a retrotransposon insertion in an intron of the human TAF1 gene

• The retrotransposon disrupts splicing of TAF1 mRNA and decreases levels of the full-length transcript, suggesting that the disease may be due, at least in part, to a partial loss of TAF1 function

• Although treatment options are currently limited, bilateral pallidal deep brain stimulation may be effective in controlling hyperkinetic symptoms in XDP