Introduction

Chorea-acanthocytosis (ChAc, OMIM 200150) is an uncommon neurodegenerative disorder. First described more than 30 years ago by Levine1 and Critchley,2 the disease is mainly characterised by gradual onset of hyperkinetic movements and aberrant erythrocyte morphology.3 Diseases that combine neurological features with acanthocytosis have been termed ‘neuroacanthocytoses’. Other disorders in this group include abetalipoproteinemia (OMIM 200100), hypobetalipoproteinemia (OMIM 107730) and McLeod syndrome (OMIM 314850). ChAc is an autosomal recessive condition, although autosomal dominant transmission has been reported.1, 4, 5 The onset of neurological symptoms is usually delayed until 25–45 years of age and the disease follows a progressive course.3 There is as yet no effective long-term treatment for this disorder.

As well as chorea, which closely resembles that observed in Huntington's disease, patients can suffer from motor and vocal tics, dystonia, parkinsonism and ocular motor disorders.3,6,7 Orofacial dyskinesia often leads to tongue and lip biting and dysphagia. Epileptic seizures and cognitive as well as psychopathological abnormalities are also seen.3,8 Post-mortem studies reveal degeneration of the basal ganglia, which results in atrophy of the putamen and caudate nucleus. In contrast to Huntington's disease, however, the cerebral cortex is usually spared.3

The term acanthocytosis is used to describe the unusual spiky appearance of a proportion of red blood cells in the patients’ blood. Reported percentages of acanthocytes in affected individuals’ blood vary widely, but are usually in the 5–50% range.3 As there is no standard clinical method for performing an acanthocyte count, it is not clear how much variability can be accounted for by differing laboratory procedures.9 The broader significance and molecular basis of acanthocytosis is unclear.

We reported linkage of chorea-acanthocytosis in 11 families to a 6 centimorgan interval flanked by the markers GATA89a11 and D9S1843 on chromosome 9q21 in 1997.10 We subsequently identified a novel gene in the ChAc critical region and found 16 different mutations in these same 11 families.11 Another group who independently identified the gene reported one other mutation in three Japanese families.12 The CHAC gene is organised in 73 exons spanning about 250 kb of the genome. We identified two main transcripts: transcript A comprises exons 1–68 and 70–73, encoding the 3174 amino acid protein chorein, whereas transcript B contains exons 1–69 only.

Chorein seems to belong to no known human gene family. Computer searches of its sequence do not identify any known structural motifs or domains. Even the number of putative transmembrane domains in chorein diverges widely between different prediction programs. Other than the assumption that chorein plays a similar role to its structural homologue in Saccharomyces cerevisiae,13 little can be said of its potential function or how its absence leads to the chorea-acanthocytosis phenotype. As the first step towards addressing these issues and to obtain an idea of the mutational spectrum of this disorder, we conducted a large-scale screen of 43 unrelated probands with ChAc. We report 54 novel mutations in CHAC and discuss their composition and distribution within the gene.

Materials and methods

Patients

Forty-three probands from 15 countries were included in this study. Thirty-three probands were the only known affected member of their family; the other 10 came from families with multiple affected members. No families that have been reported to display autosomal dominant inheritance were included in this study. The clinical diagnosis of ChAc was based on the presence of acanthocytes in patients with a movement disorder in whom McLeod syndrome had been excluded. The clinically very similar family of Johnson in whom acanthocytes had not been found14 was also included. Peripheral blood was collected for DNA isolation after patients had given informed consent. Where possible, families were haplotyped and only those with results consistent to linkage to the CHAC locus were included in this study.

Mutation analysis

For 14 probands, genomic DNA was extracted from peripheral blood in Oxford using the Nucleon kit (Scotlab, Lanarckshire, UK). The remainder were extracted using a variety of procedures. All 73 translated CHAC exons plus flanking splice sites were amplified from the proband DNA. Primers used were mainly as published previously,11 but some were redesigned for improved mutation analysis (primer sequences and PCR conditions listed in Supplementary Information). PCR amplification was carried out with standard reaction mixes containing a 9 : 1 ratio of AmpliTaq GoldTM: Pfu Turbo DNA polymerases (PE-Applied Biosystems, Cheshire, UK; and Stratagene, Amsterdam, Holland, respectively). Thermocycling was carried out as follows: 95°C for 15 min; a touchdown of 14 cycles (95°C for 30 s; annealing temperature +7.5°C with −0.5°C per cycle for 30 s; 72°C for 30 s); 25 cycles of 95°C for 30 s, the annealing temperature for 30 s, 72°C for 30 s; 72°C for 7 min. Amplification products from probands were combined in a 3 : 1 ratio with the appropriate wild-type homozygous amplicon. This mixture was then annealed and analysed by denaturing high-performance liquid chromatography (DHPLC), as described previously.15 PCR products showing a variant DHPLC pattern were purified using a QIAquick PCR purification kit (QIAGEN, Surrey, UK). These were then sequenced in both directions by means of the BigDye Terminator Cycle Sequencing Ready Reaction kit (PE-Applied Biosystems) and were run on an ABI PRISM 377 DNA Sequencer (PE-Applied Biosystems). For 11 exons (exons 2, 16, 29, 33, 35, 48, 49, 50, 66, 67 and 72), analysis by DHPLC was not appropriate, either because the amount of product was not sufficient or because the amplicon contained a high frequency polymorphism. These were therefore sequenced directly as above. The presence of each mutation was verified in the proband and any available family members by restriction digestion or sequencing (Table 1).

Table 1 CHAC disease mutations
Full size table

Restriction enzyme analysis of controls

The DNA of 192 control individuals (European Collection of Cell Cultures, Wiltshire, UK) was analysed to check for the presence of any missense or exonic splice-site mutations by restriction enzyme analysis (Table 1). Where the mutations did not provide a restriction fragment polymorphism, a diagnostic restriction site was introduced by PCR mutagenesis (primer sequences listed in Supplementary Information).

Haplotype analysis

Polymorphic markers GATA89A11, D9S1674, GATA89C08, GGAT13B07, D9S153, AFMA101XD1, D9S1867 and D9S1843 flanking the CHAC gene were used in order to define haplotypes and establish linkage to the CHAC locus. Amplification conditions were as described previously.10

Southern blot analysis

Where it was repeatedly not possible to amplify a certain exon in a proband, Southern blot analysis was performed on BamHI, HindIII and EcoRI digests of the patient's genomic DNA according to standard procedures.16

Results

The entire coding and flanking intronic sequence in CHAC was screened for mutations by DHPLC in 43 unrelated patients with chorea-acanthocytosis. This revealed 57 different mutations likely to cause disease in 39 probands (Table 1). Only three of these mutations have been previously reported, in other families.11 All available family members were screened for the relevant mutation(s) using the verification method outlined in Table 1; in each case, the mutation(s) co-segregated consistently with the affection status. In seven patients only one heterozygous mutation was found; in four patients with typical symptoms of ChAc, no disease mutations were found.

Nonsense mutations

Seventeen different base substitutions that result in premature termination codons (PTCs) were identified (Table 1). Three of these mutations (1549G>T, 3109A>T, 5920G>T) result in substitution to a TAA stop codon; three other mutations (3157C>T, 6419C>G, 9219C>G) result in substitution to a TAG stop codon. The remaining 11 mutations in this group result in substitution to a TGA stop codon: the majority of these (7/11, 64%) arise in CGA arginine codons.

Splice-site mutations

Fifteen different mutations are predicted to result in altered CHAC mRNA splicing (Table 1). Twelve of these are located in introns: they directly alter the highly conserved AG or GT dinucleotides of the intronic splice-acceptor or splice-donor sites, respectively. Mutation 883-1_892del spans intron 11 and exon 12. Although it deletes 10 nucleotides of exon 12 and could therefore shift the reading frame of the CHAC mRNA, we consider its primary effect to be on splicing, since the new intron/exon junction formed (attctttagTATTTCAGTATTAT → attctttaTTAT) is widely divergent to the consensus splice-acceptor sequence. The remaining two splice-site mutations (7806G>A & 9474G>A) alter the 3′ nucleotide in exons 55 and 72, respectively. The first substitution occurs within the sequence CCGgtaata → CCAgtaata: although both trinucleotides CCG and CCA code for proline, this exon/intron junction no longer conforms to the splice-donor consensus. The same is true of mutation 9474G>A in proband 8: the substitution AGGgtaaat → AGAgtaaat does not affect the coding for arginine 3158, but does abolish the splice-donor site. Both exonic splice-site mutations were found to be absent in 192 control individuals using the verification methods indicated in Table 1. Unfortunately, as RNA was not available from these families, it was not possible to determine the precise effect of these mutations on splicing. However, input of all these base changes into the SpliceView program (http://l25.itba. mi.cnr.it/∼webgene/www.spliceview.html) revealed that, in each case, the new junction sequence was too divergent from the consensus to score as a splice-site (Table 1).

Missense mutations

Only two different base changes resulting in a potentially pathogenic, non-conservative amino-acid substitution were identified (3283G>C & 7378T>C, Table 1). Neither of these missense mutations was found when screening 192 normal individuals using the verification methods outlined in Table 1. Mutation 3283C>G results in the substitution of alanine to proline at position 1095: this position is occupied by an aliphatic amino acid in putative chorein homologues T08G11.1, CG2093, VPS13 and KIAA1421 of Caenorhabditis elegans, Drosophila melanogaster, Saccharomyces cerevisiae and Homo sapiens, respectively. Mutation 7378T>C results in the substitution of tryptophan to arginine at position 2460, which is occupied by a large aliphatic or aromatic amino acid in the aforementioned homologues.

Insertion/deletion mutations

Twenty different insertions or deletions leading to a shift in the reading frame, introducing a PTC, were identified (Table 1). These comprise 11 1-bp deletions, two 1-bp insertions, two 2-bp deletions, three 4-bp deletions and one deletion each of five and 14 base pairs. Mutation 2029_2031delins27 involves the deletion of the trinucleotide CAT and the insertion of the 27-nucleotide sequence ATATACTAATATCTGCTTCTTTTGGAC. This mutation results in a net increase of 24 nucleotides, so does not shift the reading frame: however, the third trinucleotide that is introduced is a PTC. In probands 6 and 42, it was not possible to amplify exon 23 and exons 70–73, respectively. Southern analysis of the appropriate regions revealed that proband 6 was homozygous for an approximately 7 kb deletion covering exon 23, and that proband 42 was homozygous for a deletion of at least 13 kb removing the three terminal exons of CHAC (Figure 1).

Figure 1
figure 1

Southern blot analysis of gross deletions in CHAC. (A) analysis of proband 6 EX23del mutation. A 563-bp cDNA probe, containing exons 22–26, was used. A scaled restriction map of the normal locus is shown on the left; hybridisation of the probe to an EcoRI digest of the proband family is shown on the right. The abnormal restriction fragment is arrowed; restriction fragments that are absent in the proband are labelled a and b. (B) analysis of proband 42 EX70_73del mutation. A 710-bp cDNA probe, containing exons 65–68 and 70–73, was used. Scaled restriction maps of the normal locus are shown on the left; hybridisation of the probe to HindIII and EcoRI digests of a control (+) and proband 42 is shown on the right. Restriction fragments that are absent in the proband are labelled c–g.

Full size image

Exonic polymorphisms

In addition to the mutations listed above, 16 exonic variants were identified (Table 2). Nine of these represent silent mutations, in that the nucleotide substitutions do not alter the amino-acid coding. The remaining seven changes were concluded to be neutral variants, despite leading to amino-acid substitutions. This was because they were either present only in unaffected relatives, or on the same chromosome as disease-causing mutations. Interestingly, one of the so-called ‘silent’ mutations (1020A>G) in exon 13 occurs in the nucleotide sequence GAAGTAAAT → GAG GTAAAT. The new sequence scores 84 on the SpliceView program as a new splice-donor site (the splice-donor site in intron 13 that is usually used scores 79). Proband 18 is heterozygous for this change and an S539X mutation in exon 18 (Table 1). The exon 13 change was also found on 1 of 384 control chromosomes, so it is doubtful that this represents a pathogenic variant. However, we cannot exclude the possibility that it could lead to the ChAc phenotype in combination with a heterozygous nonsense mutation.

Table 2 Exonic polymorphisms identified in the CHAC gene
Full size table

Discussion

Mutation analysis of the CHAC gene in a cohort of 43 unrelated probands revealed 57 different mutations likely to cause chorea-acanthocytosis in 39 pedigrees. Fifty-four of these disease mutations have not previously been reported, indicating a strong allelic heterogeneity with no single mutation causing the majority of ChAc cases in the population. For four probands in this study with typical symptoms, no disease mutations in CHAC were detected. As there is little evidence for locus heterogeneity, it could be that the mutations are located within the 5′ or 3′ untranslated regions, or within intronic sequences not screened in this study; small deletions and rearrangements preventing amplification of one allele will also not be detected with this methodology. For the seven probands with only one heterozygous mutation detected, this is almost certain to be the case. Since three gross deletions have already been identified (12 and present study) and the CHAC gene is spread over such a large genomic region (about 250 kb), it is conceivable that many ChAc patients could harbour heterozygous deletions. Such deletions will only be detected by comprehensive Southern blot analysis of the entire gene or quantitative PCR.

CHAC mutation distribution

Figure 2 shows that there does not seem to be any ‘clustering’ of mutations identified in CHAC so far. Nine mutations in this study (Table 1) and two from our previous screening11 are predicted to affect only transcript A, which contains exons 70–73. Proband 42 is in fact homozygous for a transcript A-specific mutation, namely the deletion of exons 70–73 (Figure 1). Given that this patient displays symptoms typical for ChAc, we can deduce that these exons are essential for some functions of chorein and that transcript B, which lacks these exons, cannot compensate for it.

Figure 2
figure 2

Spectrum of CHAC mutations identified. The coding sequence of CHAC transcript A is illustrated; exons are denoted by grey and black bars. Mutations identified in previous studies are referenced. The number of probands in which identical mutations were identified is shown in parentheses.

Full size image

Recurrent CHAC mutations

Three mutations reported here were found in multiple probands within this study, or had been identified in another family in a previous study in this laboratory. Probands 12 and 39 are heterozygous and proband 27 is homozygous for the mutation R208X, which was also found in family CHAC3 reported earlier.11 Probands 1 and 34 have mutations (R3037X and 9429_9432del, respectively) that had initially been identified in families CHAC7 and CHAC11.11 As we could not perform haplotype analysis on certain probands in this study, it is not possible to determine whether they are related. Family CHAC3 and proband 12 both originate from Italy and so it is conceivable that their mutation was inherited from a common ancestor. However, proband 27 originates from Japan and proband 39 has inherited the mutation from a Danish mother. As it is unlikely that these individuals are related to the other pedigrees, it seems logical to conclude that the R208X (622C>T) mutation has arisen at least three times independently. Supporting this hypothesis is the fact that the cytosine involved is part of a CG dinucleotide and as such is especially vulnerable to methylation and spontaneous deamination to thymine. The same explanation is appropriate for the probable recurrence of the R3037X (9109C>T) mutation in the Mexican family CHAC7 and the German proband 1; and may also explain the preponderance of R>X substitutions in this cohort. Proband 34 and CHAC11 both originate from the UK and therefore inheritance of the 9429_9432del allele from a common ancestor cannot be ruled out. The deletion is of the tetranucleotide AGAG within the tandem repeat AGAGAG, however, and so it is plausible that this mutation has arisen twice through replication slippage. The existence of a smaller deletion within the same region in proband 29 (9431_9432del, deletion of the dinucleotide AG) supports the latter hypothesis.

Predicted effects of CHAC mutations on function

Forty of the 57 CHAC mutations identified in this study (70%) are nonsense or insertion/deletion mutations that lead to PTCs and therefore predict absence or marked reduction of mutated CHAC transcript via nonsense-mediated decay. This is in accordance with the recessive inheritance of chorea-acanthocytosis, where lack of chorein is the primary cause of the disease.

Fifteen out of the 57 CHAC disease mutations described here (26%) are changes affecting splicing. A likely explanation for this relatively high percentage of splice-site mutations is the large number of splice sites contained within this gene: the 143 splice sites of CHAC presumably present quite a large target for mutation. In the absence of RNA from affected patients, it is difficult to predict exactly what effect these mutations have at the transcript level. However, as every splice-site mutation described here alters the respective exon/intron junction such that it does not score on the SpliceView program, it is likely that normal splicing will be markedly reduced in each case.

Only two of the CHAC mutations that are predicted to cause disease in this study (2/57, 4%) are missense mutations. They are non-conservative substitutions and were not found in 384 control chromosomes, which suggests that they do not represent benign polymorphisms. The fact that just two out of nine amino-acid variants reported here (Tables 1 & 2) are believed to be pathogenic is perhaps an indication that much of chorein is tolerant to substitution.

Although it has been established that mutations in CHAC are the primary cause of chorea-acanthocytosis, it is not yet known how chorein functions or how its abolition brings about the neurological and erythrocyte changes that make up the ChAc phenotype. There seems to be no significant genotype-phenotype correlation and so it is not possible to identify what clinical consequences are seen when different parts of the protein are ablated. As there are also no identifiable domains to give clues to chorein's normal function(s), information has to be derived from studies on its homologues. Only two known proteins show significant homology to chorein. TipC, which is believed to play a role in Dictyostelium's morphogenetic signalling pathway,17 has homology mainly to chorein's N- and C- termini. Vps13p in Saccharomyces cerevisiae is the closer homologue: it shows 41% similarity with chorein, rising to 54–58% in the N- and C-terminal regions. Yeast with mutations in vps13 (soi1) show mislocalisation of some trans-Golgi network (TGN) and vacuolar proteins.13 Further studies suggested that Vps13p functions at two steps in the cycling of Kex2p and other proteins between the TGN and the prevacuolar compartment.18

If we extrapolate from these findings we can speculate that chorein may play a similar role in protein trafficking. Proteins which could be sorted by chorein include furin, the mammalian homologue of Kex2p,19 and cytoskeletal components. It is known that mice lacking the erythrocyte membrane protein 4.1 exhibit neurobehavioural deficits and abnormal erythrocyte morphology.20 Perhaps mislocalisation of a similar protein or proteins in the absence of functional chorein could cause the haematological and neurological abnormalities seen in ChAc.

With only data from CHAC homologues to draw on, it is clear that determination of chorein function is in its infancy. In this study we identified 54 novel mutations in the CHAC gene. Functional analysis of these mutants is now needed in order to elucidate the exact role that chorein plays, both in the pathogenesis of ChAc and within the context of normal brain processes.

Appendix 1

Appendix 1 CHAC primersa
Full size table