The chromosome 16q-linked autosomal dominant cerebellar ataxia (16q-ADCA): A newly identified degenerative ataxia in Japan showing peculiar morphological changes of the Purkinje cell

The 50th Anniversary of Japanese Society of Neuropathology


Kinya Ishikawa, MD, PhD, Department of Neurology and Neurological Sciences, Graduate School, Tokyo Medical and Dental University, Yushima 1-5-45, Bunkyo-ku, Tokyo 113-8519, Japan. Email:


The chromosome 16q22.1-linked autosomal-dominant cerebellar ataxia (16q-ADCA) is a form of spinocerebellar ataxia (SCA) common in Japan. It is clinically characterized by late-onset purely cerebellar ataxia. The neuropathologic hallmark of 16q-ADCA is degeneration of Purkinje cells accompanied by an eosinophilic structure which we named “halo-like amorphous materials”. By immunohistochemistry and electron microscopy, the structure has been so far found to contain two components: the somatic sprouts from the Purkinje cells and presynaptic terminals of unknown origin. As far as we are aware, this peculiar morphological change of Purkinje cells has not been previously described. Further investigations may disclose unique pathological processes in SCA.


There is a considerable difference in frequencies of autosomal dominant cerebellar ataxias, also called spinocerebellar ataxia (SCA), in a small country such as Japan. However, overall, Machado-Joseph disease (MJD) and spinocerebellar ataxia type 6 (SCA6) are the two most prevalent SCAs in Japan. SCA1, SCA2 and dentatorubral-pallidoluysian atrophy (DRPLA), a form of SCA originally identified in Japan, are also present. These SCAs, caused by trinucleotide (CAG) repeat expansions, are diagnosed with relatively simple molecular genetic tests. While these SCAs with CAG repeat expansions are the major fraction of SCA, approximately 10–40% of all SCAs account for diseases for which mutations have not yet been identified.1

We have been pursuing a form of SCA in which any of the known CAG repeat expansions are excluded from its cause. We started investigation on six such families which showed slowly progressive, seemingly purely cerebellar, ataxia in every generation.2 We embarked on a genome-wide linkage analysis using approximately 300 microsatellite DNA markers to discover in which chromosome the mutation is located. After screening all autosomal chromosomes, we found a significant evidence of linkage to the long arm of chromosome 16 (16q22.1).2 Surprisingly, this locus had been already known for SCA4, a SCA with prominent sensory axonal neuropathy associated with pyramidal tract signs.3 While every SCA4 patient showed prominent sensory axonal neuropathy, none of our patients presented such a remarkable “extracerebellar” dysfunction. In addition, ages of onset were earlier in SCA4 than in our families.2,3 From these clinical distinctions, we coined a term, “chromosome 16q-linked autosomal-dominant cerebellar ataxia” (16q-ADCA) to keep it an open question whether this disease is allelic to SCA4 or a distinct disorder. To address this question, as well as to discover the nature of these two diseases, we needed to explore the mutation. Therefore, we started to collect families clinically similar to 16q-ADCA for further genetic investigation. This effort led us to encounter the first clinical and neuropathologic study of 16q-ADCA.


An index patient of this family was a 70-year-old male patient who had a 10-year history of slowly progressive ataxia.4 Clinical examination disclosed that he had no evidence of extracerebellar dysfunctions except for hearing impairment. Sensory functions were normal and peripheral nerve conduction study did not show abnormality, excluding clinical features of SCA4. MRI of the brain showed cerebellar cortical atrophy without obvious brainstem involvement.

Family history of this index patient revealed that more than 10 individuals in four successive generations had histories of unsteady or lurching gait and difficulty in articulation, both starting insidiously around the 5th and 6th decades of their lives and slowly progressing over 10 years. We were able to trace back to these patients, and found that they had somewhat uniform clinical features similar to the index patient. All the rest of the affected individuals had slowly progressive cerebellar ataxia without obvious extracerebellar features. We made a clinical diagnosis of this family as late-onset purely cerebellar ataxia compatible with 16q-ADCA.

During our study on this family, one patient who had slowly progressive cerebellar ataxia for 26 years died at the age of 96 from a natural cause. This patient also did not show any neurological abnormality, including her memory, except for cerebellar ataxia. We were able to examine this patient neuropathologically. Detailed description of this patient was described previously.4,5


The brain of this 96-year-old patient weighted 1200 g before fixation. On macroscopic examination after fixation, atrophy was noted only at the upper surface of the cerebellum. The cerebrum and the brainstem appeared fairly well preserved.

On histological examinations, the cerebellar cortex was noted as the region with obvious degeneration. The Purkinje cells had dropped out, whereas granule cells were still quite well preserved and the molecular layer also had its thickness preserved (Fig. 1).4 Not only had Purkinje cells significantly reduced in number, we also noticed that remaining Purkinje cells were often shrunken. Remarkably, a peculiar eosinophilic structure was found surrounding such shrunken Purkinje cells (Fig. 2a). This eosinophilic structure stained pale in both KB and modified Bielschowsky methods (Fig. 2b,c). While the combination of eosinophilic structure and shrunken Purkinje cell bodies was found elsewhere in the cerebellar cortex, we could not find this peculiar Purkinje cell change in other cerebellar diseases, suggesting that it may reflect a certain unidentified degeneration processes characteristic of 16q-ADCA. As the eosinophilic structure (appearing pale pink) surrounding condensed Purkinje cell bodies (appearing dark pink) was reminiscent of the halo in Lewy bodies, we named this peculiar change as, “halo-like amorphous materials”.

Figure 1.

A low magnification view of the cerebellar cortex of 16q-ADCA. While granule cell and molecular layers appear well preserved, a mild reduction of Purkinje cells is noted (HE stain).

Figure 2.

Conventional histological features of Purkinje cell change in 16q-ADCA. (a) A shrunken Purkinje cell body is surrounded by an eosinophilic structure which authors coined, “halo-like amorphous materials”. (b) The halo-like amorphous materials are not obviously stained with KB method. (c) The halo-like amorphous materials are also not stained with silver staining. Note that axons from basket cells attach to the outer margin. (a: HE stain; b: KB stain; c: modified Bielschowsky's stain. Bars = 50 µm).

Following our report of this peculiar Purkinje cell change, nearly 10 patients have been so far reported to show similar morphological changes in Purkinje cells.6 All the patients in who genetic tests for 16q-ADCA were performed harbored the same single-nucleotide C-to-T (−16 C > T) change in the puratrophin-1 gene specific to 16q-ADCA.7 Therefore, making the diagnosis of 16q-ADCA among numbers of cerebellar degenerations seemed to become feasible based on this neuropathologic hallmark, “halo-like amorphous materials”.


We next studied the halo-like amorphous materials immunohistologically to clarify what are the components of this peculiar change.4,5 First, we studied the cytosolic calcium binding protein calbindin D28k, which is expressed exclusively in Purkinje cells in the cerebellum. On immunohistochemistry for calbindin D28k, we observed various morphological changes of Purkinje cells. For example, numerous somatic sprouts stemming from a Purkinje cell body was occasionally seen (Fig. 3a). In such cases, a zone with calbindin D28k immunoreactivity appeared corresponding to the halo-like amorphous materials. On other occasions, calbindin D28k immunoreactive “granules” were found outside Purkinje cells (Fig. 3b,c). Sometimes, calbindin D28k immunoreactive puncta appeared to create a zone surrounding the Purkinje cell body, suggesting that remnants of somatic sprouts constitute at least a part of halo-like amorphous materials (Fig. 3b). Calbindin D28k-positive granules were also found distant from the Purkinje cells even though the halo-like amorphous materials themselves did not show obvious immunoreactivity against calbindin D28k (Fig. 3d). From these observations, we considered that the somatic sprouts from Purkinje cells are among the important constituents of the halo-like amorphous materials.

Figure 3.

Immunohistochemistry for calbindin D28k, a Purkinje cell-specific marker within the cerebellum, in 16q-ADCA. (a) A bizarre-shaped Purkinje cell shows numerous sprouts stemming from its cell body. Note a zone with calbindin D28k immunoreactivity is seen corresponding to the “halo-like amorphous materials”. (b) Numerous calbindin D28k immunoreactive puncta and granules are seen outside the Purkinje cell body, forming halo-like amorphous materials. (c) Some granules are seen outside a Purkinje cell. (d) Some calbindin D28k-positive granules are still seen even though the Purkinje cell body and the halo-like amorphous materials themselves did not show obvious immunoreactivity against calbindin D28k. (a–d: mouse monoclonal, anti-calbindin D28k antibody (Sigma, dilution 1:200), immunoperoxidase stain, Bars = 50 µm).

We next studied synaptic proteins since Purkinje cells are known to receive synaptic inputs from various types of neurons. For this purpose we studied synaptophysin, one of the pre-synaptic vesicle proteins. The numbers of synaptophysin-immunoreactive granules attaching to Purkinje cell bodies were not increased in SCA6 brains used as controls. On the other hand, such granules were remarkably increased in number in 16q-ADCA, creating a zone of synaptophysin-immunoractive structures surrounding Purkinje cell bodies (Fig. 4a). Such increased zones sometimes even extended up to the primary shaft of the Purkinje cell dendrites (Fig. 4b). This clearly added increased presynaptic terminals, conceivably originating from neurons other than Purkinje cells, as an important component of halo-like amorphous materials.

Figure 4.

Immunohistochemistry for synaptophysin (a,b), phospholyated neurofilament (c), and ubiquitin (d) in 16q-ADCA. (a) Synapthophysin, a 38 kDa presynaptic protein, is abundantly seen surrounding every Purkinje cell body. (b) On higher magnification, increased synaptophysin-immunoreactivity was sometimes even seen at the primary shaft of the Purkinje cell dendrites and axons. (c) Immunoreactivity for phosphorylated neurofilament was not obvious in the amorphous materials. (d) Ubiquitin-positive granules are seen outside the Purkinje cell body. (ML: the molecular layer, PL: the Purkinje cell layer; GL: the granule cell layer.). (a,b: rabbit polyclonal, anti-synaptophysin antibody [Sigma, dilution 1:200]; c: mouse monoclonal SMI31 [Sternberger, dilution 1:1000]; d: rabbit polyclonal anti-ubiquitin antibody [Dako, dilution 1:400]. All immunoperoxidase stains, bars = a: 25 µm; b–d: 50 µm).

Immunohistochemistry using SMI31, a marker for phosphorylated neurofilaments depicting axons, demonstrated that axon component is unclear in the halo-like amorphous materials (Fig. 4c), as indicated from the modified Bielschowsky's stain. Astrocytic processes, demonstrated by immunohistochemistry for glial fibril acidic protein (GFAP), were present only at the outside margin of the halo-like amorphous materials (figure not shown).

Finally, we examined 16q-ADCA by ubiquitin immunohistochemistry to examine the process of ubiquitin-related protein degradation system. We found several ubiquitin-positive granules within the halo-like amorphous materials (Fig. 4d). Because the structures and locations of ubiquitin-postive granules resembled those of calbindin D28k-positive granules (Fig. 3b–d), we speculate that some of the somatic sprouts stemmed from Purkinje cell bodies are labeled with ubiquitin, suggesting activation of such a protein degradation system in halo-like amorphous materials.

Through our present observations, we found that somatic sprouts of Purkinje cells and accumulation of synaptophysin-immunoreactive granules are two important features of halo-like amorphous materials. Somatic sprouts have been most often described in Menkes' disease8 but also in other conditions such as MELAS.9 However, the amorphous materials have not been described in any conditions other than 16q-ADCA.10 While an accumulation of synaptophysin-positive granules was seen in 16q-ADCA, synaptophysin immunoreactivity was found to be lost around the Purkinje cell soma in Menkes' disease (figure not shown). In accord with this contrast, loss of presynaptic terminals was seen under electron microscopy in Menkes' disease,11 whereas presynaptic structures were indeed seen surrounding the Purkinje cell soma in 16q-ADCA (Dr Mari Yoshida, Aichi Medical University, pers. obs.). Therefore, we consider that a certain mechanism that leads to the presynaptic terminal accumulation surrounding Purkinje cells is unique for 16q-ADCA.

However, we should note that an accumulation of synaptic proteins in the dentate nucleus is known as “the gurmose degeneration”,12,13 an eosinophilic amorphous structure surrounding the neurons of the cerebellar dentate nucleus, most commonly reported in progressive supranuclear palsy (PSP) and DRPLA. In these two conditions, the neurons of the dentate nucleus are degenerated, while synaptic terminals from Purkinje cells innervating to the dentate nucleus accumulate, forming grumose degeneration. Therefore, further investigations comparing grumose degeneration and halo-like amorphous materials may be needed to address similarities and differences in their pathological processes.


In summary, the 16q-ADCA seems to be a new SCA reported from Japan showing purely cerebellar ataxia and peculiar Purkinje cell degeneration. Recently, we identified a novel insertion repeat in an intron shared by two different genes, BEAN (brain expressed, associated with Nedd4) and TK2 (thymidine kinase 2).14 The length of this insertion inversely correlated with the age at onset in patients. Dissecting molecular mechanisms of 16q-ADCA, newly renamed as SCA31, would be an important theme to discover the pathologic basis of this peculiar morphological change.


We would like to thank Dr Taro Ishiguro (Tokyo Medical and Dental University) for assisting graphics in this article.

This paper is based on a long history of study discovering the clinical, genetic and neuropathological characteristics of 16q-ADCA, now renamed as SCA31. We would like to acknowledge all the people who participated in this study. Particularly, we are in debt to Dr Kiyoshi Owada (Tokyo Medical and Dental University), Drs Gen Ishida and Manabu Gomyoda (Matsue National Hospital), Drs Mari Yoshida and Yoshio Hashizume (Aichi Medical College), Dr Toshio Mizutani (Tokyo Metropolitan Neurological Hospital), Dr Kunihiro Yoshida (Shinshu University), and Drs Yuko Saito and Shigeo Murayama (Tokyo Metropolitan Geriatric Institute) for sharing their neuropathological samples. We also acknowledge Dr Asao Hirano (Montefiore Medical Center) for providing us specimens with Menkes' disease as a control.