OUP user menu

CAG repeat expansion in the TATA box-binding protein gene causes autosomal dominant cerebellar ataxia

Hiroto Fujigasaki, Jean-Jacques Martin, Peter Paul De Deyn, Agnès Camuzat, Didier Deffond, Giovanni Stevanin, Bart Dermaut, Christine Van Broeckhoven, Alexandra Dürr, Alexis Brice
DOI: http://dx.doi.org/10.1093/brain/124.10.1939 1939-1947 First published online: 1 October 2001


At least 13 loci responsible for autosomal dominant cerebellar ataxia (ADCA) have been identified. Spinocerebellar ataxia 1, 2, 3, 6 and 7 are caused by translated CAG repeat expansions. However, in France, >30% of ADCAs are not explained by the known genes. Recently, analysis of the TATA box-binding protein (TBP) gene, one of the transcription factors known to contain a CAG/CAA repeat, in patients with progressive cerebellar ataxia revealed one sporadic case with 63 repeats. We examined this gene in 162 index cases with ADCA. An expanded repeat with 46 repeat units was detected in a single index case from Belgium. In this family, two affected members and six unaffected, but at-risk, individuals carried expanded alleles. Interestingly, the expanded repeat was stable during transmission. The main clinical features in six patients were cerebellar ataxia, dementia and behavioural disturbances with onset in their fourth to sixth decade. The main neuropathological finding was severe neuronal loss and gliosis in the Purkinje cell layer. Immunohistochemical analysis showed neuronal intranuclear inclusions containing expanded polyglutamine, indicating that this disease shares several features with other polyglutamine diseases. This study demonstrates that CAG/CAA repeat expansion in the TBP gene causes ADCA with dementia and/or psychiatric manifestations.

  • TATA box-binding protein
  • CAG/CAA repeat
  • autosomal dominant cerebellar ataxia
  • dementia
  • neuronal intranuclear inclusion
  • ADCA = autosomal dominant cerebellar ataxia
  • HDJ-2 = human DNAJ protein-2
  • NII = neuronal intranuclear inclusion
  • PCR = polymerase chain reaction
  • SCA = spinocerebellar ataxia
  • TBP = TATA box-binding protein


The autosomal dominant cerebellar ataxias (ADCAs) are a heterogeneous group of hereditary neurodegenerative diseases characterized by progressive cerebellar ataxia variably associated with other neurological signs (Harding, 1993). The genetic aetiologies for ADCA have been investigated intensively in recent years, and at least 13 loci responsible for ADCAs have been identified (Holmes et al., 1999; Hermann-Bert et al., 2000; Stevanin et al., 2000; Yamashita et al., 2000). It is known that several types of ADCA, such as spinocerebellar ataxia (SCA) 1, 2, 3, 6 and 7 (Orr et al., 1993; Kawaguchi et al., 1994; Imbert et al., 1996; Pulst et al., 1996; Sanpei et al., 1996; David et al., 1997; Zhuchenko et al., 1997), encoding polyglutamine stretches, as well as Huntington's disease (The Huntington's Disease Collaborative Research Group, 1993), spinal and bulbar muscular atrophy (La Spada et al., 1991) and dentatorubral–pallidoluysian atrophy (Nagafuchi et al., 1994) are caused by CAG repeat expansions. These disorders have several common features: (i) occurrence of the phenotype above a threshold which varies according to the gene; (ii) strong negative correlation between the age of onset and the size of the expansions; (iii) instability during transmission with a tendency toward expansion resulting in anticipation; and (iv) the presence of neuronal intranuclear inclusions (NIIs) containing the protein with expanded polyglutamine (Koyano et al., 1999; Zoghbi and Orr, 2000).

Recently, Koide and colleagues (Koide et al., 1999) examined the TATA box-binding protein (TBP) gene, one of the transcription factors known to contain a polymorphic CAG/CAA repeat ranging from 25 to 42 repeat units (Gostout et al., 1993; Imbert et al., 1994), for expansion in patients with ADCA and with sporadic progressive ataxia. Only one sporadic case with a CAG repeat expansion resulting from a de novo mutation was identified. The clinical phenotype was infantile-onset progressive cerebellar ataxia and mental deterioration (Koide et al., 1999). To determine whether the CAG repeat expansion in the TBP gene is also responsible for ADCA, we examined the lengths of the CAG/CAA repeat in 162 index cases with ADCA for which known genes had been excluded as causative. We found a short CAG repeat expansion in the TBP CAG/CAA repeat segregating with ADCA in a Belgian family, indicating that this expansion is responsible for the disease. We present the clinical phenotype and neuropathological findings in this family.



Index cases from 162 families with ADCA were selected after exclusion of CAG repeat expansions at the SCA 1, 2, 3, 6, 7 and 12 loci (Fujigasaki et al., 2001). The controls were 110 French subjects without neurological disorders. Blood samples were obtained with informed consent and genomic DNA was extracted from leucocytes or lymphoblastoid cells using standard methods.

Polymerase chain reaction (PCR) analysis of the CAG/CAA repeat lengths in the TBP gene

A portion of the TBP gene containing the CAG/CAA repeat was amplified by PCR in a 25 μl reaction mixture containing 200 ng of genomic DNA, 1 μM of each primer (5′-ATGCCTTATGGCACTGGACTG-3′ and 5′-CTGCTGGGACGTTGACTGCTG-3′), 300 μM of each deoxynucleotide triphosphate, 0.2 U of Taq DNA polymerase (Perkin-Elmer), and 10% dimethylsulphoxide in the buffer provided by the supplier. The cycling steps were 96°C for 3 min, 30 cycles of denaturation at 94°C for 45 s, annealing at 58°C for 30 s, extension at 72°C for 45 s, and final extension at 72°C for 7 min. PCR products were electrophoresed on 2.5% agarose gels and visualized by ethidium bromide. The number of CAG/CAA repeats was determined on an ABI 377 automated sequencer using GeneScan and Genotyper software (Perkin Elmer Applied Biosystems). Sequencing was performed using the BigDye Terminator Cycle Sequencing Ready Reaction Kit™ (Perkin Elmer Applied Biosystems) and analysis was done on the same sequencer with the Sequencing Analysis Software (Perkin Elmer Applied Biosystems).

Western blot analysis

Lymphoblasts from the proband and a normal control subject were solubilized in the lysis buffer as described previously (Koide et al., 1999), containing protease inhibitor cocktail complete™ (Boehringer Mannheim, Germany). Proteins were electrophoresed on a 15% SDS (sodium dodecylsulphate) polyacrylamide gel, and transferred onto a nitrocellulose membrane. Monoclonal antibody 1C2, known to recognize abnormally expanded polyglutamine and TBP (Trottier et al., 1995), was used as the primary antibody (dilution, 1 : 1000). Immunoreactive bands were visualized by chemiluminescence (Pierce, Rockford, Ill., USA).


The brain of patient III-2 was obtained at autopsy with consent from a relative after death. Myelin and fibrillary glia were stained on frozen coronal sections by the Spielmeyer's and Holzer's methods, respectively. Cresyl violet staining was used for cytology. Paraffin-embedded sections were obtained from the frontal cortex, temporal cortex, parietal cortex, occipital cortex, cingulate cortex, hippocampus, amygdala, cerebellar vermis and lateral lobe including the dentate nucleus, anterior and posterior basal ganglia, diencephalon, midbrain, pons, medulla oblongata and spinal cord at cervical, thoracic and lumbosacral levels. These sections were used for Nissl, heamatoxylin and eosin, and Bodian staining, and for immunohistochemistry.

For immunohistochemical analyses, anti-glial fibrillary acidic protein (Dako, Carpinteria, Calif., USA), anti-ubiquitin (Dako), anti-human DNAJ protein-2 (HDJ-2) (Lab Vision, Fremont, Calif., USA), anti-TBP (Santacruz, Calif., USA) and monoclonal antibody 1C2 (kindly provided by Dr Yvon Trottier, Université Louis Pasteur, France) were used. Immunostaining was performed under the conditions described previously (Takahashi et al., 2001).


Screening for CAG repeat expansions in the TBP gene in index cases

The geographical origins of the 162 index cases with previously excluded expansions at the SCA 1, 2, 3, 6, 7 and 12 loci were as follows: 155 families from Europe, including 123 of French origin, five from North Africa, one from Israel and one from the Western Indies. PCR followed by 2.5% agarose gel electrophoresis and GeneScan analysis revealed an expanded allele with 46 CAG/CAA repeats in a single index case with ADCA. In this family, six family members were affected in four successive generations (Fig. 1). The TBP CAG/CAA repeat lengths were examined in 15 available members of this family. The disease co-segregated with the expanded allele in the family. Eight members (two affected and six unaffected but at-risk individuals) carried 46 CAG/CAA repeats in the expanded alleles (Fig. 2). Direct sequence analysis in the index case confirmed the number of CAG/CAA repeats in the expanded allele. The structure of the repeat sequence is as follows: (CAG)3 (CAA)3 (CAG)9 CAA CAG CAA (CAG)26 CAA CAG. In other ADCA cases, the size of the CAG/CAA alleles ranged from 30 to 41 units, similar to the range observed in the French control population (Fig. 3).

Fig. 1

Simplified pedigree of a Belgian family with a CAG repeat expansion in the TBP CAG/CAA repeat sequence. Solid symbols, affected members; open symbols, unaffected members; hatched symbols, status unknown; squares, males; circles, females; arrow, the proband. Six individuals were affected in four successive generations. For reasons of confidentiality, parts of generation III and generation IV are not shown.

Fig. 2

PCR analysis in the family with the CAG repeat expansion in the TBP gene. PCR products were electrophoresed on 2.5% agarose gels and visualized with ethidium bromide. Two affected individuals (lanes 2 and 3) and six at-risk individuals (data not shown) carried expanded alleles. CAG/CAA repeat lengths are indicated at the bottom of the panel. Lane 1, 100 bp ladder; lane 2, normal control; lane 3, patient IV-2; lane 4, patient III-4.

Fig. 3

Distribution of the size of CAG/CAA repeats in the TBP gene in the French population. Two hundred and twenty control chromosomes were analysed. CAG/CAA repeat lengths in normal controls range from 29 to 40 units.

Western blot analysis

We performed Western blot analysis on extracts of lymphoblasts from the proband and a normal control subject carrying 39/46 and 39/40 CAG/CAA repeats, respectively. The monoclonal antibody 1C2 recognized an ~37 kDa immunoreactive protein corresponding to wild-type TBP in both samples. In the sample from the proband, this antibody also recognized a protein with an apparent molecular weight of 40 kDa in addition to the wild-type TBP. The signal intensity of the 40 kDa band which corresponds to the expanded protein was stronger than that of the 37 kDa band (Fig. 4).

Fig. 4

Western blot analysis of extracts from lymphoblasts. The monoclonal antibody 1C2 recognized an ~37 kDa immunoreactive protein corresponding to the wild-type TBP in both samples. In the sample from the proband (patient IV-2), this antibody also recognized an ~40 kDa protein in addition to the wild-type TBP. The signal intensity of the 40 kDa band which corresponds to the expanded protein was stronger than that of the 37 kDa band. Left, normal control; right, patient IV-2. The numbers of glutamines are indicated at the bottom of the panel.

Clinical features of the Belgian family

Patient IV-2

The proband, now 45 years old, developed minor difficulties in gait and speech at age 34 years. Her husband was aware of her personality change. She became gradually dependent for household activities. At age 35 years, neurological examination revealed dysarthria, dysmetria and gait disturbance. The Mini-Mental Scale Examination score was 24/30 and showed disorientation in time and space. In addition, she was euphoric. These symptoms progressed gradually. On examination at the age of 44 years, forced laughing and crying, and cogwheel rigidity on the left side were observed in addition to cerebellar ataxia. Brain MRI showed cerebellar and mild cerebral atrophy (Fig. 5A and B).

Fig. 5

Brain MRIs of patients IV-2 (proband) (A and B) and III-4 (C). Cerebellar atrophy and mild subcortical atrophy were observed in the brain MRI of the proband (A and B). Patient III-4 also showed marked cerebellar atrophy (C). (A and C) T1-weighted sagittal image; (B) T2-weighted axial image.

Patient III-3

The proband's father had had seizures during his childhood. At age 37 years, he had psychosis with paranoia, hypersexuality and violent behaviour. At age 39 years, he was admitted to a psychiatric hospital. Neurological examination revealed dysarthria with explosive speech, dysdiadochokinesis on both sides, dysmetria of both upper and lower limbs, gait disturbance with a wide base and choreoathetosis. Deep tendon reflexes were normal. At that time, he was diagnosed as having Huntington's disease. Psychosis regressed gradually under treatment, but cerebellar ataxia was progressive, and cognitive decline was observed. At age 50 years, he became bed-ridden because of a fracture of the femur. He died at age 52 years.

Patient III-4

A paternal aunt of the proband developed dysarthria and gait disturbance at the age of 55 years. At age 59 years, neurological examination revealed dysarthria and gait disturbance with a broad base. She was also disoriented in time and space, and did not recognize her entourage. At age 61 years, bilateral hyper-reflexia and an extensor plantar reflex on the left side was observed. Her score on the Mini-Mental Scale Examination was 16/30 with disturbance of long-term memory, disorientation in time and space, graphic disturbances and dyscalculation. Brain MRI showed marked cerebellar atrophy (Fig. 5C). She had become wheelchair-bound at that time. On neurological examination at the age of 63 years, she presented nystagmus on lateral gaze and mild cogwheel rigidity. At age 65 years, she developed left hemiplegia due to middle cerebral artery infarction. She died of pneumonia in the same year.

Detailed clinical histories of other family members were not available. Patient I-1 was admitted to a psychiatric hospital at age 48 years. Individual II-1 who was an obligate carrier died before the manifestation of neurological symptoms. The clinical features of this family are summarized in Table 1.

View this table:
Table 1

Summary of the clinical features in the Belgian family with CAG repeat expansion in the TBP gene

Patient (years)Age at onset (years)Age at deathCerebellar ataxiaDementiaPsychosisRigidityChoreoathetosisSeizure
ND = not determined. *Patient in a psychiatric institution; †no detailed information available.
III-1ND55ProgressiveUnknownParanoia, hypersexualityNDND
III-33752ProgressivePresent†Paranoia, hypersexuality Violent behaviour++
Deficit in long-term memory


The brain of patient III-4 weighed 1012 g on autopsy. There was marked atrophy in the cerebellum. Bilateral cerebral cortical and mild subcortical atrophy were also observed. There was no atrophy in the pons. Small infarctions were observed in the right frontal lobe and in the basal ganglia on the right side. On frozen coronal sections passing through the frontotemporal lobe and the rostral thalamus, mild cortical atrophy and large amounts of subpial corpora amylacea were detected. There was mild cerebral cortical atrophy with diffuse microspongiosis in the second cortical layer and in the entorhinal cortex. There was focal gliosis in the white matter and in the cortex. A lacuna in the basal ganglia was also found. The cerebellum was severely atrophied, with gliosis in the molecular layer and severe loss of Purkinje cells with Bergmann's gliosis. The inner granular layer was rarefaied. There was fibrillary gliosis in the dentate nucleus. The loss of Purkinje cells was confirmed, and silver impregnation showed the presence of empty baskets around the Purkinje cells that had disappeared. In contrast, neuropathological change of the brainstem was mild. In the midbrain, many Marinesco bodies were seen in pigmented neurones in the substantia nigra. Neither neuronal loss nor gliosis was observed in the pontine nuclei. The nuclei of the reticular formation were normal and there was mild neuronal loss in the locus coeruleus. In the medulla oblongata, fibrillary gliosis without neuronal loss was seen in the inferior olivary nucleus. Spheroids were present in the gracilis nucleus. The neuropathological findings are summarized in Table 2.

View this table:
Table 2

Summary of neuropathological findings in patient III-4

Brain regionNeuronal lossGliosisSpongiosisNII
– = unaffected, + = mild, ++ = moderate, +++ = severe.
Frontal cortex+++
Temporal cortex+++
Parietal cortex++
Occipital cortex+++
Cingulate gyrus++
Enthorhinal cortex+++
Basal ganglia
Caudate nucleus
Globus pallidus
Red nucleus
Substantia nigra
Oculomotor nucleus
Pontine nuclei+
Locus coeruleus+
Reticular formation
Inferior olivary nucleus+
Molecular layer++
Purkinje cell layer++++++
Granule cell layer+
Dentate nucleus+++
Spinal cord
Anterior horn+


Immunohistochemical analyses were performed to determine whether there were NIIs as in other polyglutamine diseases. Monoclonal antibody 1C2 recognized NIIs in the pyramidal cells of CA1 (Sommer's sector) in the hippocampus and the subiculum, the supraoptic nucleus in the hypothalamus, the occipital cortex, the superior temporal gyrus, the superior frontal gyrus, the putamen, the cerebellar dentate nucleus, the griseum pontis, the nucleus ambiguus in the medulla oblongata and the anterior horn of the spinal cord. In the dentate nucleus, 1C2 also recognized perinuclear aggregates. In addition, several 1C2-positive nuclei were seen in the granule cells of the cerebellar granular layer. NIIs were also recognized by anti-TBP, anti-ubiquitin and anti-HDJ-2 antibodies. TBP-positive nuclei were observed in neurones in the striatum (Fig. 6).

Fig. 6

Neuropathology and immunohistochemistry of autopsied patient (III-4). (A) Cerebellum: loss of Purkinje cells and gliosis of Bergmann glia (arrowheads) and of the molecular layer (asterisks), cresyl violet, scale = 100 μm. (B) Dentate nucleus of the cerebellum: NII recognized by anti-TBP antibody, scale = 10 μm. (C) Striatum at low magnification: numerous nuclei are positively stained by anti-TBP antibody, scale = 10 μm. (D) Striatum: NII recognized by anti-ubiquitin antibody, scale = 10 μm. (E) Striatum: NII recognized by monoclonal antibody 1C2, scale = 10 μm. (F) Striatum: NII recognized by anti-HDJ-2 antibody, an arrowhead indicates the invagination of the nuclear membrane, scale = 10 μm. Arrows indicate NIIs.


We have illustrated the clinical phenotype and neuropathological findings of an ADCA family associated with a CAG repeat expansion in the TBP gene. These results confirm that CAG repeat expansions in the TBP gene cause ADCA, and demonstrate that TBP CAG expansions are not only associated with de novo mutations resulting from large CAG expansion as previously described (Koide et al., 1999). However, the frequency of this locus is very low in the population tested (1/162). There were no French families with this mutation. Compared with the French families (n = 256) accounted for by known loci (n = 133, unpublished data) and those excluded in this study (n = 123), this locus must be very rare in the French population (95% confidence interval: 0–1.4%).

It is known that CAG/CAA repeat length in the TBP gene is highly variable. The number of glutamines encoded by CAG/CAA repeats ranges from 25 to 42 in the American population (Gostout et al., 1993). The basic structure of this repeat is also known: (CAG)3 (CAA)3 (CAG)n CAA CAG CAA (CAG)n CAA CAG (Gostout et al., 1993). In this family, it was surprising that the CAG/CAA repeat number in the expanded allele was identical in all individuals carrying the mutation, indicating that it was stably transmitted through 13 meioses (seven paternal and six maternal). Direct sequencing revealed that the basic structure was conserved even in an expanded allele, and that the expansion occurred in the second polymorphic CAG repeat. This result suggests that CAA interruptions contribute to the stability of CAG/CAA repeats as seen in SCA1 and SCA2 genes (Zoghbi and Orr, 2000) and confirms that the mechanism of this expansion differs from the partial duplication observed in the case described previously (Koide et al., 1999).

The clinical features of this family were variable. Age at onset ranged from 34 to 55 years, in contrast to the infantile onset in the sporadic case with a large CAG repeat expansion described previously (Koide et al., 1999), suggesting a negative correlation between age at onset and the CAG/CAA repeat length. However, as all patients carried the same expansion, age at onset is not only determined by the size of the CAG/CAA repeat. All affected members presented progressive cerebellar ataxia and dementia or psychosis, but other clinical signs were variably associated. Choreoathetosis was seen in one affected member. In summary, the main clinical features of this family were progressive cerebellar ataxia, dementia and psychosis. The phenotype is, therefore, closer to DRPLA (dentatorubral–pallidoluysian atrophy) than to other ADCAs in which dementia is less frequent and often mild (Iwabuchi et al.,1999).

As in other polyglutamine diseases, lesions were selective despite the fact that TBP plays an essential role in all eukaryotic cells (Lee and Young, 1998). This suggests that the susceptibility of neurones to polyglutamine expansions in TBP may vary according to brain region. The main lesion in the brain was located in the cerebellum, especially in Purkinje cells, and to a lesser extent in the cerebral cortex. In contrast, neurodegeneration in the brainstem was very mild compared with some other ADCAs. Although the clinical features are reminiscent of DRPLA, the neuropathological findings were very different (Iwabuchi et al., 1999).

We found NIIs that contained expanded polyglutamine, heat shock protein and ubiquitin as in other polyglutamine diseases. NIIs were not restricted to affected regions but were also seen in unaffected parts of the brain. In addition, perinuclear inclusions were found in neurones of the dentate nucleus as seen in Purkinje cells of SCA2 and SCA6 brains (Ishikawa et al., 1999; Huynh et al., 2000; Zoghbi and Orr, 2000). Western blot analysis confirmed that the mutant TBP was expressed, and was even more strongly recognized by the monoclonal antibody 1C2 than wild-type TBP, suggesting a conformational change in the pathological protein, as was postulated for other polyglutamine diseases (Zoghbi and Orr, 2000).

This study demonstrates that CAG repeat expansions in the TBP gene cause ADCA with dementia and/or psychiatric manifestations, but is a rare cause of dominant ataxia in France. The main neuropathological findings are severe neuronal loss and gliosis in the Purkinje cell layer. Interestingly, phenotypically the disease shares many features of other diseases with polyglutamine expansions: (i) manifestation above a threshold number of repeats; (ii) negative correlation between the number of CAG/CAA repeats in the expansion and age at onset; and (iii) the presence of ubiquitinated NIIs containing the pathological protein. However, the expansion, as in SCA6 (Zoghbi and Orr, 2000), is stable during transmission.


We wish to thank Drs Patrice Verpillat for statistical analysis, Jan Dumon for providing genetic information on the family, Yvon Trottier for the gift of the 1C2 antibody, Olaf Riess and Ann Löfgren for providing DNA samples, and Merle Ruberg for critical reading of the manuscript. This work was supported by the VERUM Foundation and l'Association Francıaise contre les Myopathies, France (A.B.), and the Fund for Scientific Research Flanders, Belgium (J.J.M. and C.V.B). J.J.M., P.P.D.D. and C.V.B. are supported by the Born-Bunge Foundation. H.F. is supported by a fellowship from EGIDE, France, and B.D. by a Ph.D. fellowship of the FWO, Belgium.


View Abstract