Brain Advance Access originally published online on August 25, 2004
Brain 2004 127(11):2385-2405; doi:10.1093/brain/awh278
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
Brain Vol. 127 No. 11 © Guarantors of Brain 2004; all rights reserved
Invited Review |
Trinucleotide repeats and neurodegenerative disease
Institute of Neurology, National Hospital for Neurology and Neurosurgery, London, UK
Correspondence to: Professor N. W. Wood, Institute of Neurology, National Hospital for Neurology and Neurosurgery, Queen Square, London WC1N 3BG, UK. E-mail: n.wood{at}ion.ucl.ac.uk.
Received March 14, 2004. Revised May 11, 2004. Accepted June 10, 2004.
Summary
Major insights have been attained into the molecular pathology of the trinucleotide repeat neurodegenerative diseases over the past decade. Genetic definition has allowed subclassification into translated polyglutamine diseases, which are due to CAG repeat expansions, and a more heterogeneous group in which the trinucleotide repeat remains untranslated. The polyglutamine disorders are due to a toxic gain of function of mutant expanded proteins. Neuronal intranuclear inclusions (NIIs) characteristically occur. Protein misfolding, interference with DNA transcription and RNA processing, activation of apoptosis and dysfunction of cytoplasmic elements have all been invoked in the toxic process. The end result is apoptotic cell death with many aspects of neuronal function being perturbed. Promising progress has been made into arresting and reversing neurodegeneration in both cellular and animal models. The molecular mechanisms underlying the untranslated group remain less clear. Impedance of gene transcription secondary to abnormal DNA structures formed by repeats, modification of chromatin gene packaging and dysfunction at the RNA level have all been suggested as possible pathological mechanisms. These diseases remain irreversible. It is hoped that clarification of the molecular pathogenic mechanisms will provide the tools for future therapeutic intervention.
Key Words: trinucleotide repeat; polyglutamine expansion; neuronal intranuclear inclusion; transcription factors; chromatin
Abbreviations:
AR = androgen receptor; BDNF = brain-derived neurotrophic factor; CACNAIA = 
1A subunit of a voltage-gated calcium channel; cAMP = cyclic AMP; CBP = CREB binding protein; CRE = cAMP response element; CREB = CRE binding protein; DRPLA = dentatorubralpallidoluysian atrophy; FRDA = Friedreich's ataxia; hCD2 = human CD2 reporter gene; HD = Huntington's disease; HDAC = histone deacetylase; HDAC = histone deacetylase; hsp = heat shock protein; Htt = human huntingtin; htt = murine huntingtin; ISC = iron–sulphur cluster; NII = neuronal intranuclear inclusion; polyQ = expanded polyglutamine repeat; SCA = spinocerebellar ataxia; TBP = TATA binding protein; yfhlp = yeast frataxin homologue 1 protein
Introduction
The number of neurodegenerative trinucleotide repeat diseases is increasing, as is the complexity of their molecular pathology. Broadly, they fall into two distinct classes. Firstly there are the polyglutamine repeat disorders, in which the expanded repeat is translated into an expanded polyglutamine (polyQ) tract, which results in formation of protein aggregations within the cell. The presence of the polyQ tract is thought to result in a toxic gain of function of the mutant protein. The second class, in which the trinucleotide repeat is present in an untranslated region of a gene, has more varied molecular mechanisms, including gene repression. Table 1 summarizes the two classes. This review discusses the molecular mechanisms of neuronal death in these neurodegenerative diseases rather than providing a detailed description of clinical features.
|
The polyglutamine repeat diseases
The archetypal polyQ repeat disease is Huntington's disease, which will be emphasized as it illustrates many of the pathogenic principles of the group and is the most common. Others include the spinocerebellar ataxias (SCAs) 1, 2, 3, 6, 7 and 17, dentatorubral pallidoluysian atrophy (DRPLA) and Kennedy's disease.
George Huntington described Huntington's disease (HD) in 1872. The HD locus was localized to chromosome 4p16 in 1983 (Gusella et al., 1983
). The genetic defect of an unstable expanded CAG repeat in the first exon of the huntingtin (Htt) gene was defined in 1993 (Huntington's Disease Collaborative Research Group, 1993). Interestingly, a CAG repeat expansion within an alternatively spliced exon of the junctophilin 3 gene has been found to be responsible for an HD-like phenotype in African–Americans (Holmes et al., 2001).
Genetic features
All of the polyQ repeat diseases are inherited in an autosomal dominant fashion except for Kennedy's disease, which is X-linked recessive. An expanded CAG repeat within the disease genes encodes a polyQ repeat in the mutant proteins. The phenomenon of anticipation, in which subsequent generations have earlier onset and more severe disease, occurs in the polyQ diseases and can be explained by enlargement of the expansion over generations and in this group particularly occurs with paternal transmission.
The unstable CAG repeat in the Htt gene is translated into a polyQ stretch near the N-terminus of the protein resulting in mutant Htt. The length of the repeat dictates the age of onset and severity of disease features (Andrew et al., 1993). In normal individuals the gene contains eight to 39 repeats (Huntington's Disease Collaborative Research Group, 1993). However, the presence of 37 or more is found in patients, suggesting that the disease is not fully penetrant in the overlap range (Rubinsztein et al., 1996). Age of onset is inversely correlated with length of the trinucleotide repeat (Trottier et al., 1994; Brandt et al., 1996; McNeil et al., 1997). Repeats of 40–50 are found in cases with onset in middle age (90–95% of cases), while over 60 repeats usually results in juvenile HD. However, CAG repeat length does not fully explain the differences in age of onset. Other genes may play a role (Kehoe et al., 1999) and polymorphisms surrounding the CAG repeat could be important (Vuillaume et al., 1998).
Dentatorubralpallidoluysian atrophy (DRPLA) is caused by an expanded CAG repeat in the 5' coding region of the atrophin-1 gene, located on the short arm of chromosome 12 (Koide et al., 1994; Nagafuchi et al., 1994). The disease is most common in Japan. Age of onset is tightly correlated to repeat length, with different clinical syndromes manifest at differing ages of onset. There is a degree of overlap in the manifestations of DRPLA and HD. The length of the repeat on the normal allele is 7–34 and on the expanded 58–88 (Ikeuchi et al., 1995; Komure et al., 1995).
There are at least 25 different SCAs. The causal mutation in eight is expansion of a trinucleotide repeat. SCA 1, 2, 3, 6, 7 and 17 are due to a translated CAG expansion whereas SCA8 and 12 are suggested to be due to an untranslated CTG and CAG expansion respectively (Table 2). Prior to the genetic definition of these disorders, they were classified as autosomal dominant cerebellar ataxias according to clinical criteria (Harding, 1982) (Table 3). The SCA genes encode a group of proteins known as the ataxins. Ataxins 1, 2, 3 and 7 are expressed ubiquitously in nervous and other tissues. The normal cellular function of the ataxins is unknown. TATA binding protein (TBP) is a ubiquitous transcription factor and harbours the polyQ expansion in SCA 17. The polyQ tracts are near the N-terminus in TBP and ataxins 1, 2, 7 and near the C-terminus in ataxin-3. In the presence of an expanded polyQ tract the ataxins and TBP form ubiquitinated NIIs. SCA 6 is less closely related to the other polyQ SCAs. The expanded CAG repeat occurs at the 3' end of the CACNA1A gene encoding the 1A subunit of the P/Q-type voltage gated calcium channel (Zhuchenko et al., 1997). SCA 6 has the shortest CAG repeat responsible for disease and is much more stable both mitotically and meiotically than the CAG repeats found in other SCAs.
|
|
Also known as X-linked recessive spinal and bulbar muscular atrophy, Kennedy's disease is due to the pathological expansion of a CAG repeat in exon 1 of the androgen receptor (AR) gene located on the long arm of the X chromosome. It was the first trinucleotide repeat disease described (La Spada et al., 1991
). Kennedy's disease is primarily a disorder of motor neurons with degeneration in the spinal cord and brainstem (Sobue et al., 1989). Onset is usually from the second to fourth decade with slow progression and long survival. The normal range of the CAG repeat is 11–33, patients having between 38 and 62 (Fischbeck et al., 1999). The length of CAG repeats correlates well with the age of onset of disease (Igarashi et al., 1992). Anticipation occurs albeit less commonly than in other polyQ diseases, reflecting the greater stability of the CAG repeat (Watanabe et al., 1996). Molecular mechanisms
The molecular processes involved in polyQ toxicity are broadly summarized in Fig. 1.
|
The proteins
The molecular mechanisms involved in polyQ neurodegeneration are complex. The cellular roles of the wild-type and mutant proteins are becoming clearer, although they remain far from fully defined.
The Htt gene has 67 exons, with the CAG repeat in the first. Htt is a protein of 3140 amino acids with a molecular mass of 349 kDa (Huntington's Disease Collaborative Research Group, 1993). It has no major homology to any other known protein, although it contains a motif involved in cytoplasmic transport (Andrade and Bork, 1995). Htt is ubiquitously expressed (Strong et al., 1993; Trottier et al., 1995). In the brain it is found predominantly in neurons with little expression in glial cells (Li et al., 1995; Sharp et al., 1995). During development it appears in the mammalian brain by gestation day 15 (Bhide et al., 1996). The precise localization of wild-type Htt within cells is still not certain. Most studies suggest that it is predominantly located in the cytoplasm (Aronin et al., 1995; DiFiglia et al., 1995; Sharp et al., 1995; Ferrante et al., 1997; Sapp et al., 1997; Jones, 1999) but there is evidence of localization to the nucleus (De Rooij et al., 1996). Htt may shuttle between the nuclear and cytoplasmic compartments (Ross et al., 1999).
Htt is highly conserved through evolution, suggesting an important role (Sharp and Ross, 1996). In knockout mice in which the homologous murine (Hdh) gene is inactivated, nullizygous embryos die at 8–10 days of gestation, indicating the essential role for Htt in development (Duyao et al., 1995; Nasir et al., 1995; Zeitlin et al., 1995).
Htt associates with clathrin-coated vesicles, microtubules and cytoplasmic granules and may have a role in intracellular transport (DiFiglia et al., 1995; Gutekunst et al., 1995; Sapp et al., 1997). In studies of axonal transport, Htt was antero- and retrogradely transported (Block-Galarza et al., 1997), giving further credence to a role in vesicle function. Indeed, mutant Htt containing an expanded polyglutamine tract inhibits fast axonal transport (Szebenyi et al., 2003).
It is thought that mutant Htt harbouring the expanded polyQ expansion has a toxic gain of function. mRNA from both the normal and expanded allele are found in HD lymphoblasts and brain (Ambrose et al., 1994) and mutant Htt mRNA is present in patients homozygous for the CAG expansion (Huntington's Disease Collaborative Research Group, 1993), ruling out a loss of gene transcription as the cause of the disease. The occurrence of homozygous HD patients suggests that mutant Htt maintains its normal function in addition to its toxic effects. Homozygous knockin mice with 50 CAG repeats in the homologous murine Hdh gene develop normally, implying that mutant htt retains the developmental function of wild-type htt (White et al., 1997). A complicating factor has emerged in that the postembryonic inactivation of the murine Hdh gene causes a progressive neurological phenotype with accompanying neurodegeneration (Dragatsis et al., 2000). Therefore, a subtle loss of function may act concomitantly with the toxic gain of function of mutant Htt in mice. It is possible that this proposed loss of function and the more clear toxic gain of function may have adverse effects at different molecular levels. Indeed, a similar mechanism may occur in Kennedy's disease as expression of the mutant AR gene is reduced in transient transfection assays in a CAG length-dependent fashion (Choong et al., 1996). This may account for the lower level of target gene activation that occurs in the presence of the mutant expanded AR (Mhatre et al., 1993). The implication is that a loss of function of the AR is also relevant to pathogenesis. Interestingly, there are examples of homozygous cases of HD, SCA 3 and SCA 17 (Lang et al., 1994; Squitieri et al., 2003; Toyoshima et al., 2004), all of which have a more severe phenotype than found in the usual heterozygotes. The molecular correlate may be that the double dose of mutant polyQ protein is more toxic to the cell, in keeping with the toxic gain of function hypothesis. Also possible, however, is that the low-level loss of function of the mutant protein, which may be relatively insignificant in the face of polyQ toxicity, becomes more significant when all of the protein is mutant. It is also possible that the polyQ load within cells is crucial. In a Hdh knockin mouse, a CAG expansion of approximately 72 repeats expanded preferentially in striatal cells with age. Therefore, the increased levels of polyQ in these cells may account for the selective degeneration of striatal neurons with time (Kennedy and Shelbourne, 2000). Dramatic increases in CAG repeat length in the striatum may occur early on in the disease course (Kennedy et al., 2003).
The difficulties in elucidating the function of the ataxins begin with defining their precise cellular localization. Nuclear and cytoplasmic locales have been suggested, with the possibility of shuttling between compartments. Ataxin-1 is localized predominantly in the nucleus in neuronal tissue and in the cytoplasm in the periphery. In neuronal tissue it is expressed at high levels in regions that are affected in SCA 1 (cerebellar Purkinje cells) and in areas that are not (hippocampus) (Servadio et al., 1995). The nuclear presence of mutant ataxin-1 is essential for disease, as transgenic mice do not develop it if the nuclear localization sequence of ataxin-1 is disabled (Klement et al., 1998). Ataxin-1 is not dispensable as knockout mice have learning defects and deficits in hippocampal function (Matilla et al., 1998). Homozygous polyQ mutant ataxin-1 knockin mice had a phenotype of impaired memory with no pathological changes (Lorenzetti et al., 2000). Therefore, prolonged exposure to mutant ataxin-1 at endogenous levels is probably needed to produce a neurological phenotype. Pathogenesis is likely to be a function of polyQ length, expressed protein levels and duration of neuronal exposure to the mutant protein.
Wild-type ataxin-2 is located in the cytoplasm and within the brain is expressed in cerebellar Purkinje cells and other brainstem nuclei (Huynh et al., 1999). Ataxin-3 is ubiquitously expressed and is present in the cytoplasm in normal and diseased brain and in transfected cells (Nishiyama et al., 1996; Paulson et al., 1997a; Schmidt et al., 1998). In contrast, in neuroblastoma cell lines wild-type ataxin-3 was found predominantly in the nucleus, associating with the inner nuclear matrix. A putative nuclear localization sequence was proposed and transport into the nucleus occurred independently of an expanded polyglutamine tract (Tait et al., 1998). Wild-type ataxin-3 may therefore shuttle between the nucleus and cytoplasm.
Ataxin-7 distributes to the nucleus and associates with the nuclear matrix and nucleolus in transfected cells. A nuclear localization sequence has been identified (Kaytor et al., 1999). Wild-type ataxin-7 colocalizes with an ATPase component of the 26S proteosome within discrete nuclear foci and high levels of ataxin-7 are found in cerebellar Purkinje cell nuclei (Matilla et al., 2001). In transgenic mice mutant ataxin-7 N-terminal fragments translocate from the cytoplasm and accumulate in the nucleus in an age-dependent manner (Yvert et al., 2001).
Atrophin-1 is ubiquitously expressed and predominantly cytoplasmic. There is predominantly neuronal expression in the brain, which is highest in the cerebellum, hippocampus, substantia nigra and pontine nuclei (Yazawa et al., 1995). Atrophin-1 mRNA levels are similar in normal and DRPLA brain (Nishiyama et al., 1997), consistent with the gain-of-function hypothesis. Atrophin also acts as a transcriptional corepressor in Drosophila (Zhang et al., 2002) and may therefore be involved in regulating gene expression.
NIIs
Neuronal intranuclear inclusions (NIIs) characteristically occur in polyQ repeat disease. In vitro, N-terminal mutant Htt forms aggregates in a ß-pleated sheet conformation similar to amyloid, with a fibrillar structure on electron microscopy, and no such aggregation occurs when a normal polyQ stretch is present (Scherzinger et al., 1997). PolyQ tracts above a certain length threshold are thought to form a ‘polar zipper’ and to associate via formation of hydrogen bonds (Perutz et al., 1994). Htt is a substrate for transglutaminase in vitro with the rate of reaction and polymer formation increasing with polyQ number (Kahlem et al., 1998). It has been possible to prevent mutant Htt aggregation in vitro using specific monoclonal antibodies to the polyQ repeat and a variety of chemical compounds (Heiser et al., 2000). Additionally, a transglutaminase inhibitor used in mutant N-terminal Htt transgenic mice, extended survival and reduced the appearance of some movement disorders (Karpuj et al., 2002). Conversely, NII levels were increased in HD transgenic mice with transglutaminase 2 (the major brain transglutaminase) deleted, although both generalized and brain weight loss were reduced with prolonged survival of the mice (Mastroberardino et al., 2002). Therefore, the role of transglutaminase in aggregate formation remains unclear, although it may play a role in regulating cell death in HD.
In a neuronal cell culture model, transfection of expanded mutant cDNA Htt resulted in the formation of NIIs in a polyQ length- and time-dependent manner. The longer the repeat the greater the number of NIIs (Kim et al., 1999). Other workers have found this phenomenon in various cellular models (Cooper et al., 1998; Li and Li, 1998; Lunkes and Mandel, 1998; Lunkes et al., 1999). The formation of NIIs and toxicity in rat striatal neurons when an expanded polyQ tract was linked to an otherwise inert protein demonstrated that the polyQ stretch was the toxic moiety (Moulder et al., 1999).
Cellular transfection studies have repeatedly shown that truncated portions of mutant Htt translocate to the nucleus, are more likely to aggregate and are more toxic than the full-length protein (Cooper et al., 1998; Hackam et al., 1998; Lunkes and Mandel, 1998; Martindale et al., 1998; Li et al., 2000). Indeed, introducing aggregated polyQ peptides directly into mammalinn nuclei is toxic (Yang et al., 2002). In some studies, blocking nuclear accumulation of mutant Htt resulted in less toxicity (Saudou et al., 1998). Addition of a nuclear export signal to mutant Htt reduced toxicity while increasing nuclear localization resulted in increased toxicity (Peters et al., 1999). Others have shown that both nucleus and cytoplasm are sites for toxic effects (Hackam et al., 1999). In the Hdh knock-in mouse mutant Htt is translocated into the nucleus with subsequent formation of N-terminal aggregations, (Wheeler et al., 2000). The mechanism for translocation of mutant human Htt to the nucleus is uncertain although there may be a nuclear localization sequence at the N-terminus (Hackam et al., 1998).
In transgenic mice, truncated N-terminal fragments rather than full-length mutant Htt may be more toxic but direct comparison of different transgenic models is difficult, due to varying construction of transgenes, mouse strains and expression levels. Generally, however, transgenic mice with shorter N-terminal constructs have resulted in more marked phenotypes, with neuronal aggregations forming and cellular toxicity occurring earlier than in those with full-length Htt. There is more widespread formation of aggregations in N-terminal mutant Htt transgenic mice, whereas in a Hdh knockin mouse changes occurred more specifically in the striatum. It may be that the provision of ready-processed N-terminal fragments aids toxicity. Full-length mutant Htt, whether it is derived from a transgene or the endogenous murine Hdh gene, seems to require long repeats to result in toxicity. Importantly, however, whatever genetic manipulation is used to generate a murine HD model, N-terminal fragments of mutant Htt are always found within the aggregations (Mangiarini et al., 1996; Davies et al., 1997; White et al., 1997; Reddy et al., 1998, 1999; Hodgson et al., 1999; Schilling et al., 1999a; Shelbourne et al., 1999; Li et al., 2000; Wheeler et al., 2000; Yamamoto et al., 2000; Lin et al., 2001). Further, co-expression of mutant N-terminal Htt accelerates the disease process initiated by full-length mutant Htt (Wheeler et al., 2002).
Similarly, ubiquitinated NIIs occur in DRPLA (Table 4), Kennedy's disease (Table 5) and the polyQ SCAs. Ataxin-1 aggregates are found in regions of SCA 1 brain that specifically degenerate and are found in cell lines overexpressing mutant ataxin-1 (Skinner et al., 1997). In transgenic mice overexpressing a mutant SCA 1 allele with expression targeted to cerebellar Purkinje cells, a phenotype of progressive ataxia developed with formation of NIIs (Clark and Orr, 2000). Similar transgenic mice, without the self-association domain found in ataxin-1, developed a similar phenotype and Purkinje cell pathology but without NIIs (Klement et al., 1998), suggesting that NIIs are not the primary pathology.
|
|
NIIs of ubiquitinated mutant ataxin-3 aggregations occur in SCA 3 neurons (Paulson et al., 1997a
, b). Mutant ataxin-3 binds to the nuclear matrix, resulting in a conformational change and exposure of the polyQ tract (Perez et al., 1999). Exposure of an expanded polyQ may therefore result in more marked aggregation and binding of other polyQ-containing nuclear proteins. Overexpression of both full-length and truncated mutant ataxin-3 fragments in cell lines, transgenic mice and Drosophila resulted in cellular toxicity, apoptosis and death with the associated formation of NIIs (Ikeda et al., 1996; Paulson et al., 1997b; Warrick et al., 1998; Evert et al., 1999). The correlation between NII formation and cell death was poor, suggesting that NIIs are not the primary toxic insult. Indeed, in SCA 7 knockin mice neuronal dysfunction occurred several weeks before NII formation (Yoo et al., 2003).
Ubiquitinated NIIs have been reported in SCA 2 and SCA 7, brains although the exact distribution and relationship to disease remains unclear. SCA 7 NIIs are ubiquitinated in areas such as the inferior olive, which is severely degenerated in this disease, but also in the cerebral cortex, which is unaffected (Holmberg et al., 1998; Yvert et al., 2001), an observation that is borne out in transgenic mice (Yvert et al., 2001). In SCA 17, ubiquitin- and TBP-positive NIIs were found in the putamen and frontal cortex (Nakamura et al., 2001).
The ubiquitination of NIIs suggests that the ubiquitin-proteosome proteolytic pathway of the cell attempts to clear aggregates. Consistent with this, the 20S proteosome relocates to aggregates in SCA 1 (Cummings et al., 1998) and SCA 3 (Chai et al., 1999b). Transgenic mice that express mutant ataxin-1 but not a ubiquitin-protein ligase have significantly fewer NIIs but the Purkinje cell pathology is markedly worse. Thus, preventing ubiquitination of mutant ataxin-1 results in increased toxicity to the cell, and ubiquitination of mutant ataxin may play an important detoxifying role (Cummings et al., 1999). The ubiquitin–proteosome pathway is also important in the metabolism of ataxin-3 aggregations and inhibition of the proteosome complex results in accumulation of aggregates in a repeat-dependent manner (Chai et al., 1999b). Proteosome subunits also colocalize with mutant ataxin-7 NIIs in SCA 7 transgenic mice (Yvert et al., 2001), transfected cells and brain (Zander et al., 2001). A specific proteosome subunit is depleted in regions of the brain affected by neurodegeneration in SCA 7 (Matilla et al., 2001) and may account for impaired proteolysis of NIIs. Interestingly, downregulation of mutant ataxin-7 expression after early development in transgenic mice did not prevent progression of retinopathy, suggesting that the toxic effect may be irreversible (Helmlinger et al., 2004).
NIIs form in SCA 6 Purkinje cell cytoplasm and are not ubiquitinated. There is, however, marked degeneration of cerebellar Purkinje cells in SCA 6 and transfection of cells with mutant CACNA1A resulted in the formation of perinuclear aggregates and apoptotic cell death (Ishikawa et al., 1999). The presence of an expanded polyQ tract results in altered function of the calcium channel, which may be responsible for neurodegeneration (Toru et al., 2000). Two other disorders are caused by mutations in the CACNA1A gene (Ophoff et al., 1996). Familial hemiplegic migraine is generally caused by missense mutations and episodic ataxia type 2 by nonsense mutations. There is a definite clinical overlap between these disorders (Denier et al., 1999; Jen et al., 1998; Yue et al., 1997).
Neuronal aggregates and their role in neurodegeneration
Much has been made of the formation of neuronal aggregates in polyQ neurodegeneration. It is not certain that the aggregates are of primary pathogenic importance in the polyQ repeat diseases, as suggested above in relation to the SCAs. They may merely represent end products of an upstream toxic event. Some have suggested that inclusions serve a protective role (Saudou et al., 1998).
Aggregate formation does not necessarily result in cell death. For example, they have been seen in the dentate nucleus of HD cerebellum, a region of the brain unaffected by neurodegeneration in HD (Becher et al., 1998). Cellular models show the discrepancy between NII formation and cell death. In rat primary striatal neurons, NII formation is not a prerequisite for cell death but mutant Htt must be present in the nucleus to cause apoptosis (Saudou et al., 1998). Apoptosis in neuroblastoma cell lines was increased in the presence of mutant Htt but the correlation with NII formation was not tight (Lunkes and Mandel, 1998; Lunkes et al., 1999).
There is also evidence in transgenic mice for NII formation without cell death and vice versa. The first transgenic mouse modelling HD had a phenotype reminiscent of HD, extensive striatal NIIs but minimal cell death (Mangiarini et al., 1996). Other mice have striatal loss with minimal or absent NIIs (Reddy et al., 1998; Aronin et al., 1999; Hodgson et al., 1999). Interestingly, when a 146 CAG repeat was introduced into a hypoxanthine phosphoribosyltransferase transgene (Ordway et al., 1997), abundant NIIs developed throughout the brain and the animals developed a late onset neurological phenotype with multiple seizures and early death. These mice demonstrate that polyQ incorporation into a cytoplasmic protein (hypoxanthine phosphoribosyltransferase) renders it toxic, causing translocation into the nucleus with subsequent formation of NIIs and neuronal dysfunction.
A conditional model of HD has been developed based on mice expressing an N-terminal mutant Htt fragment with a CAG expansion of 94 repeats. Reducing expression of the mutant Htt in symptomatic mice led to a disappearance of NIIs and an amelioration of the behavioural phenotype. This does not clarify whether toxicity is caused by NIIs but, importantly, shows that reduction of mutant Htt expression reverses the pathological changes and phenotype (Yamamoto et al., 2000). Recently, interference with polyQ aggregation in HD transgenic mice with disaccharide molecules ameliorated the HD phenotype (Tanaka et al., 2004). Polypeptides have been identified that bind to mutant Drosophila huntingtin and interfere with aggregation in a Drosophila model for HD, with a reduction in lethality (Apostol et al., 2003). Thus, interfering with mutant htt aggregate formation may have therapeutic potential (Bates, 2003). Interestingly, there have been suggestions that eliminating mutant Htt in mice may result in sensitization to the remaining mutant htt with hastening of the disease (Auerbach et al., 2001), which may be relevant to strategies aimed at reducing mutant Htt in humans.
Molecular chaperones and NII
Abnormal folding of polyQ-containing proteins with impaired protein processing has been invoked as a mechanism of aggregation and inclusion formation. A class of proteins known as molecular chaperones is involved in this process. Examples are heat shock protein (Hsp) 70 and Hsc70. A co-chaperone class known as Hsp40, with members human DNAJ1 (Hdj1) and Hdj2, facilitates Hsp70 action. Hsp40 and Hsp70 are both involved in the clearance of misfolded proteins via the ubiquitin-proteosome pathway (Bercovich et al., 1997). Hdj2 and Hsc70 were associated with nuclear aggregates in HD cell culture and transgenic mice models and overexpression of Hdj1 and Hsc70 reduced aggregate formation. Hsp70 levels were upregulated in both the cellular model and in transgenic mice (Jana et al., 2000). Therefore, mutant Htt may elicit a stress response causing molecular chaperone expression, perhaps in an attempt to clear aggregates. On the other hand, sequestration of protein chaperones into aggregates may result in misfolding and reduced clearance of other crucial cellular proteins. Whether chaperones are directly involved in aggregation itself is not clear. Notably, a recent study in a Drosophila model of HD found the presence of suppressor genes capable of modifying the phenotype with one of the predicted proteins being Drosophila Hdj-1 (Kazemi-Esfarjani and Benzer, 2000).
Overexpression of protein chaperones in mutant ataxin-1-transfected HeLA cells reduced aggregate formation (Cummings et al., 1998), consistent with the hypothesis of protein misfolding leading to aggregate formation in polyQ diseases. Hdj-2 is associated with NIIs in SCA 1 (Cummings et al., 1998), SCA 3 (Chai et al., 1999a, b) and in SCA7 transgenic mice (Yvert et al., 2001), transfected cells and SCA 7 brain (Zander et al., 2001). Overexpression of Hsp70 in SCA 1 transgenic mice also suppressed neurodegeneration and improved motor function (Cummings et al., 2001). Additionally, in a SCA3 Drosophila model the molecular chaperone Hsp70 was colocalized with NIIs and its overexpression ameliorated neurodegeneration (Warrick et al., 1999). Hsp40 also ameliorated toxicity in a SCA 3 Drosophila model (Chan et al., 2000). Further, these proteins are also contained within aggregates in Kennedy's disease (Table 5). This provides further evidence for a protein-misfolding model for polyQ toxicity.
Apoptosis in polyglutamine diseases
It is becoming increasingly apparent that apoptosis is crucial to neurodegeneration in polyQ repeat diseases. Cellular models of HD have shown that toxicity is mediated by increased sensitivity to apoptotic stress (Saudou et al., 1998; Kim et al., 1999). Both wild-type and mutant Htt are cleaved at sites within the N-terminal by the pro-apoptotic caspases (Goldberg et al., 1996; Wellington et al., 1998). Therefore, the apoptotic pathway may be self-perpetuating, with the cleavage of mutant huntingtin activating apoptosis (and caspases) and hence generating more toxic N-terminal fragments. Wild-type Htt itself has been proposed to protect cells from apoptotic signals. (Rigamonti et al., 2000). Preventing cleavage of mutant Htt resulted in reduced aggregate formation and reduced toxicity in apoptotically stressed cells (Wellington et al., 2000). Inhibiting caspases can abrogate the formation of NII and prolong the survival of cells (Kim et al., 1999; Moulder et al., 1999; Wang et al., 1999; Wellington et al., 2000), although different inhibitors have different effects (Kim et al., 1999). Interestingly, a caspase-1-deficient mouse used in a neurotoxic model of HD protected against neurotoxicity, implicating apoptosis in neurodegeneration (Andreassen et al., 2000). It has been demonstrated that caspase activation occurs in HD brain and lymphoblasts (Sanchez et al., 1999; Sawa et al., 1999). Lately, activated dsRNA-dependent protein kinase has been found at high levels in areas of the brain specifically affected in HD (Peel et al., 2001). dsRNA-dependent protein kinase has been implicated in the stress response, triggering apoptosis. Atrophin-1 is also cleaved by caspase, resulting in nuclear translocation and toxicity (Nucifora et al., 2003).
Promising inroads have been made into slowing HD disease progression by interfering with apoptotic processes. Inhibiting caspase-1 in transgenic mice expressing mutant Htt has delayed the appearance of NIIs, neurotransmitter alterations and the onset of symptoms (Ona et al., 1999). HD-transgenic mice treated with minocycline, which is known to downregulate caspase expression in other disease models, delays the onset of symptoms and prevents the upregulation of caspase-1 and -3 expression (Chen et al., 2000).
Interactions of polyglutamine containing proteins with other cellular proteins
The polyQ proteins interact with other cellular proteins and mutant Htt, atrophin-1, ataxins and AR have several interactions in common (Tables 4
–7). Wild-type Htt is known to associate with several predominantly cytoplasmic proteins (Table 6), giving clues to its usual cellular functions. These interactions may all be perturbed in HD. Indeed, Htt-associated protein-1, glyceraldehyde-3-phosphate dehydrogenase and calmodulin show enhanced binding to mutant Htt, which may result in toxicity. A Hdh expanded PolyQ knockin mouse showed progressive formation of nuclear aggregates in the striatum and formation of presynaptic inclusions with synaptic vesicles. Given that N-terminal mutant Htt bound to synaptic vesicles in a CAG length-dependent manner and impaired glutamate uptake in vitro, binding of mutant Htt to vesicles in vivo may play a role in toxicity (Li et al., 2000).
|
|
Aggregation of mutant polyQ proteins results in the recruitment of other polyQ-containing proteins. Therefore, sequestration of endogenous proteins containing polyQ tracts may occur (Preisinger et al., 1999
). Several transcription factors contain short polyQ repeats (see below). Altering the balance of transcription factors within the nucleus may have profound effects on gene expression with potentially toxic effects. A variety of transcription factors have been found to colocalize with NIIs and several interact directly with the mutant proteins (Tables 4–7 and below).
Mutant polyglutamine proteins and gene expression
A common transcriptional activator cyclic AMP (cAMP) response element (CRE) binding protein (CREB), along with its coactivator CREB binding protein (CBP), has been strongly implicated in mutant Htt-induced gene repression (Fig. 2). Mutant Htt causes repression of CBP-regulated gene expression, which can be overcome in cell culture by overexpressing CBP (Nucifora et al., 2001). Mutant Htt may also promote CBP degradation (Jiang et al., 2003). TAFII130 is a coactivator for CREB-regulated transcriptional activation. TAFII130 binds directly to polyQ stretches with the consequence of impaired CREB-dependent transcriptional activation, which again can be rescued by overexpression of TAFII130 (Shimohata et al., 2000). Mice with the Creb gene disrupted postnatally developed neurodegeneration in the hippocampus and striatum (Mantamadiotis et al., 2002). Suggestions have been made that transcription from genes regulated by the cAMP response element may be impaired in HD due to reduced cAMP levels secondary to impaired energy metabolism in the cell (Gines et al., 2003). CREB/CBP recruitment to NIIs or the interaction with mutant polyQ proteins also occurs in DRPLA, SCA 3, SCA 7 and Kennedy's disease (Tables 4–7). Thus, disruption of CREB/CBP-mediated gene expression may be a common mechanism of neurodegeneration in polyQ repeat diseases.
|
CREB-binding protein and p300/CBP-associated factor have intrinsic histone acetyl transferase activity. Acetylation of DNA histones is generally associated with transcriptional activation. PolyQ Htt binds to CBP and p300/CBP-associated factor histone acetyl transferase domains, inhibiting the enzymatic activity, which may impair gene expression. Notably, in an in vivo Drosophila neurodegeneration model, inhibition of histone deacetylases (HDACs) arrests the degenerative process (Steffan et al., 2001
). Recent work in HD transgenic mice has confirmed that inhibition of HDACs, hence increasing the amount of acetylated histones, improves motor impairment and increases brain histone acetylation (Ferrante et al., 2003; Hockly et al., 2003). Thus, HDAC inhibitors may be beneficial in human disease.
Htt also interacts with the ubiquitous transcription factor Sp1, and coexpression of Sp1 and TAFII130 in cultured striatal cells overcame the downregulation of dopamine D2 receptor gene transcription caused by mutant Htt (Dunah et al., 2002). Interestingly, this study also showed that soluble mutant Htt prevented Sp1 binding to DNA from both presymptomatic and HD post-mortem brain. Thus, trapping of transcription factors within NIIs may not entirely account for pathogenic effects of mutant Htt. Rather, gene expression may be affected by soluble mutant Htt–transcription factor interactions. For example, TBP, CBP and SP1 remained diffusely localized within the nucleus in HD mouse models despite the presence of NIIs, with no difference in staining patterns between wild-type and HD mouse brains (Yu et al., 2002).
To investigate the functional effect of transcription factor interactions, oligonucleotide microarrays have been employed to screen striatal mRNA in transgenic mice modelling HD. There were diminished levels of less than 2% of the mRNAs tested. These did, however, encode components of neurotransmitter, calcium, and retinoid signalling pathways involved in striatal neuron function. Similar changes in gene expression were also seen in another HD mouse model (Luthi-Carter et al., 2000). Further, gene expression changes in mutant atrophin-1 transgenic mice overlap with those found in HD transgenics (Luthi-Carter et al., 2002), suggesting common mechanisms in the neurodegenerative process. It is becoming increasingly apparent that both wild-type and mutant Htt have important roles in gene transcription. Consistent with this is the regulation of BDNF, a neurotrophic factor necessary for striatal neuron survival, at the transcriptional level by Htt. BDNF expression is downregulated in the presence of mutant Htt (Zuccato et al., 2001). Other genes are upregulated in the presence of mutant Htt [e.g. a ribosomal protein Rrs1 (Fossale et al., 2002)].
The mutant ataxins also have effects on gene expression. In SCA 1 transgenic mice, six neuronal genes involved in signal transduction and calcium homeostasis and abundantly expressed in Purkinje cells were downregulated. This downregulation occurred before detectable pathology. Such interactions with genes important in cellular homeostasis are likely to have toxic effects. Similar effects were also evident in SCA 1 human tissues (Lin et al., 2000). Altered RNA processing may also play a role in the pathogenesis of SCA 1 as in vitro ataxin-1 was found to bind to RNA in a manner inversely related to polyQ length (Yue et al., 2001).
A Drosophila model of SCA 1 has been screened for genes that modify neurodegeneration (Fernandez-Funez et al., 2000). Modifiers included molecular chaperones and genes involved in transcription and RNA processing. In a SCA 3 cellular model both upregulation and downregulation of genes occur suggesting that loss of normal ataxin-3 as well as the gain of function of mutant ataxin-3 may be important in pathogenesis (Evert et al., 2003).
There have been more suggestions recently that expanded polyQ tracts may act independently of protein context. This has been alluded to above in relation to the toxic gain of function of hypoxanthine phosphoribosyltransferase when an expanded polyQ tract was introduced into this protein. For example, polyQ expansions present in atrophin-1 and Htt transgenic mouse constructs resulted in similar changes in gene expression, as determined by microarray analysis. Interestingly, several of the genes affected in these DRPLA and HD transgenic mice were also affected in SCA 7 and Kennedy's disease mouse models (Luthi-Carter et al., 2002). Further, CAG repeat expansions found in the junctophilin-3 and TBP genes are associated with an HD-like phenotype (Stevanin et al., 2003). These findings also suggest a commonality of neurodegenerative mechanisms in these diseases.
As in HD, the interaction of the polyQ repeat containing atrophin-1, AR and ataxins with a variety of nuclear proteins may have important pathological consequences (Tables 4, 5 and 7).
Discussion
The pathogenesis of the polyQ repeat diseases is complex. Certainties are the genetic defects and the association of NIIs with disease. The aggregations may or may not be primarily responsible for neurodegeneration. The multiple interactions of wild-type and mutant polyQ proteins with a variety of proteins involved in diverse cellular functions could result in cellular dysfunction, apoptosis, neurodegeneration and disease. It remains unclear which of the multiple molecular changes that occur in the polyQ diseases are of primary pathogenic importance and which are involved in continuation of the pathological process. Recent work has suggested dysfunction of CREB/CBP-mediated gene expression may be a common mechanism involved in polyQ-mediated neurodegeneration. Whether this represents a common pathway initiating the degenerative process or occurs as a result of toxicity ‘upstream’ remains to be clarified. A major challenge is to try to collate evidence from in vitro and in vivo models and indeed patients to further define pathogenic mechanisms and to develop therapies. Indeed, therapies are already being explored in HD based on evidence derived from experimental models (Beal and Ferrante, 2004).
Neurodegenerative trinucleotide repeat diseases with untranslated repeat expansions
This class is separate to the polyQ repeat diseases and includes SCA 8, SCA 12 and Friedreich's ataxia (FRDA). In these, the expanded repeat is not translated into expansion within a mutant protein. Certainly in FRDA there is downregulation of the frataxin gene in association with a GAA repeat expansion. There is some evidence for gene downregulation in SCA 8. The mechanism responsible for perturbed gene expression in these diseases remains unclear.
SCAs with untranslated trinucleotide repeats (SCA 8 and SCA 12)
There is an expanded CTG repeat in the 3' untranslated region of the SCA 8 gene (Koob et al., 1999). There remains debate as to whether the expanded CTG repeat found in association with the SCA 8 locus represents a rare associated polymorphism (Stevanin et al., 2000; Worth et al., 2000; Schols et al., 2003) or is causative for the disease (Day et al., 2000; Cellini et al., 2001). The CUG containing SCA 8 transcript may represent an antisense transcript to a gene coding for an actin organizing protein named kelch-like protein 1 (Nemes et al., 2000). The suggested SCA 8 transcript overlaps the transcription and translation start sites of KLH1 and it is hypothesized that the SCA 8 antisense transcript regulates expression of KLH1, with the presence of an expanded CUG repeat perturbing such regulation. An expanded CAG repeat has been linked to the 5' untranslated region of the protein phosphatase 2 regulatory subunit B gene in SCA 12 (Holmes et al., 1999).
FRDA
Nicholaus Friedreich described the disease in 1863. It is the most common of the hereditary ataxias with a prevalence of 1 in 50 000 and is another of the untranslated trinucleotide repeat diseases.
Genetic features
FRDA is an autosomal recessive disease usually with onset before the age of 25 (Harding, 1981). In 1996 the most common genetic abnormality was defined as a homozygous expanded GAA trinucleotide repeat in the first intron of the frataxin gene located on the long arm of chromosome 9 (Campuzano et al., 1996). Frataxin itself is a mitochondrial protein thought to be involved in iron metabolism (see below). Expanded GAA repeats inhibit frataxin expression. Frataxin transcription and expression are inversely proportional to the length of the expanded GAA repeat, this being particularly true for smaller expanded repeats (Campuzano et al., 1997). This may explain the correlation of the severity of some clinical features and age of onset with the shorter of the two expanded repeats (Durr et al., 1996; Filla et al., 1996; Lamont et al., 1997; Monros et al., 1997; Montermini et al., 1997c). The residual expression from the allele with the shorter expansion may be important in modulating disease severity.
The normal repeat length is less than 39 and patients generally have repeats ranging from 100 to 1700 (Cossee et al., 1997; Montermini et al., 1997a, b, c). However, atypical patients often have GAA repeats of similar length to patients with more classical features (Gellera et al., 1997; Geschwind et al., 1997; Montermini et al., 1997c). Other factors, such as environment, modifier genes and somatic mosaicism, may all play a role in such variation. Expanded GAA repeats are unstable both meiotically and mitotically. The repeat tends to contract on paternal transmission, shorter repeat arrays being found in sperm than in leucocytes (Pianese et al., 1997). Expansion or contraction can result following maternal transmission (Monros et al., 1997; Pianese et al., 1997). Approximately 2% of patients are compound heterozygotes, with the GAA repeat expansion on one allele and a point mutation on the other (Campuzano et al., 1996; Cossee et al., 1999). Generally N-terminal mutations result in milder phenotype than that seen with mutations at the C-terminal of the protein.
The frataxin gene and GAA repeats
Frataxin mRNA is found most abundantly in heart, liver, skeletal muscle and pancreas. Expression is tissue-pecific and mirrors the disease distribution (Campuzano et al., 1996; Jiralerspong et al., 1997; Koutnikova et al., 1997). In the CNS, expression is highest in the spinal cord,with lower levels in the cerebellum and very little expression in the cerebral cortex. Frataxin is developmentally expressed in mice (Jiralerspong et al., 1997; Koutnikova et al., 1997) and the frataxin knockout mouse dies in utero (Cossee et al., 2000). Embryonic lethality of frataxin nullizygous mice can be rescued with expression of human frataxin (Pook et al., 2001).
A viable frataxin knockout mouse has been made using conditional gene-targeting techniques (Puccio et al., 2001). In these mice the frataxin gene has been disabled in muscle and neural tissue, with the development of aspects of FRDA related to the frataxin deficient tissues. In the muscle knockouts there was a hypertrophic cardiomyopathy. In the neural knockouts progressive ataxia with neurodegeneration in the cerebellum and dentate nucleus occurred. There was also a sensory proprioceptive neuropathy. There was impaired activity of the respiratory chain complex and subsequent iron accumulation within mitochondria. These animals provide a model to further investigate the sequelae of frataxin deficiency and in which to test possible therapies.
Mechanisms for reduced frataxin expression in FRDA
Some inroads have been made into mechanisms underlying GAA repeat-induced gene repression. Transient transfection experiments, using a two-exon construct derived from the HIV gp120 gene with 9–270 GAA repeats present in the only intron, showed that both transcription and replication were impaired in a manner dependent on GAA repeat length and orientation (Ohshima et al., 1998). Pronounced stalling at the replication fork in yeast occurs when the expanded GAA strand serves as the lagging strand template (Krasilnikova and Mirkin, 2004). The propensity of GAA repeats to form triplexes and the inhibitory effects on transcription of such structures have been postulated as a possible mechanism for the reduction of frataxin expression (Wells et al., 1988; Ohshima et al., 1996; Mariappan et al., 1999). DNA triplexes formed by GAA repeats have been shown to be ‘sticky’, due to the formation of a single-stranded region within the triplex that has the propensity to self-associate to other such strands in other GAA triplexes. Sticky DNA may impair transcription (Sakamoto et al., 1999). GAA repeats have been shown to cause pausing of RNA polymerase in vitro in a length-dependent manner, due to the transient formation of an intramolecular DNA triplex initiated by RNA polymerase itself (Grabczyk and Usdin, 2000). More recent in vitro experiments have shown that sticky DNA is capable of impairing transcription and of sequestering RNA polymerase to the sticky repeat (Sakamoto et al., 2001b). Therefore, a model of GAA repeat-induced transcriptional repression has been suggested, in which sticky DNA sequesters transcription factors. Such an effect has not been demonstrated in vivo.
A 65 GAAGGA hexanucleotide repeat, found in several related individuals not suffering from FRDA, does not affect transcription and expression in the gp120 reporter gene assay. Neither triplexes nor formation of sticky DNA occurred with this hexanucleotide repeat (Ohshima et al., 1999; Sakamoto et al., 1999) and, as expected, transcription was not impaired (Sakamoto et al., 2001b). Therefore, a pure GAA repeat seems necessary to cause disease as imperfections may interfere with the pathological properties. Indeed, inclusion of GGA imperfections in long GAA repeats reduced sticky DNA formation dependent on the proportion of GGA imperfections in the array, relieved in vitro transcriptional repression and reduced repeat instability (Sakamoto et al., 2001a).
A GAA repeat frataxin knockin mouse has been made, with 230 GAA repeats placed in the first intron of the frataxin gene. There was a small reduction in frataxin expression in homozygous mice, leaving a residual expression level of approximately 75% in the tissues examined. There was no associated phenotype or accumulation of tissue iron. Compound heterozygous mice carrying both a GAA repeat expansion and a non-functional frataxin allele were similarly unaffected, despite frataxin expression of 25–30% of the wild-type levels (Miranda et al., 2002). Therefore, frataxin levels of less than 25% are probably required for a phenotype to develop in mice and a repeat expansion of more than 230 may be necessary to repress frataxin expression to this level. Indeed, in human FRDA lymphoblastoid cell lines, frataxin expression determined by immunoblotting ranged from 4 to 29% compared with normal (Campuzano et al., 1997).
Recent work has shown that GAA repeats may induce gene silencing in association with changes in the chromatin packaging of DNA. A (GAA)200 repeat expansion linked to a human CD2 (hCD2) reporter gene in transgenic mice was found to heritably silence expression of this gene. Since the GAA repeat was linked to an untranscribed region of the hCD2 transgene, the inhibitory transcriptional effects of GAA repeats were excluded. The GAA repeat induced repression of hCD2 manifested as the phenomenon of ‘position effect variegation’ of gene expression, in which, rather than hCD2 expression being silenced in all cells, a proportion of cells were silenced with the remaining cells actively expressing hCD2. This form of gene silencing is characteristic of that found when genes are placed near to densely packaged chromatin (known as heterochromatin). In cells in which hCD2 was silenced, there was condensation of chromatin packaging at the promoter of the gene. Further, this silencing was modified by heterochromatin protein 1ß, a protein known to be a major component of heterochromatin (Saveliev et al., 2003). It is possible that the phenotypic variation seen in FRDA may be due to variegation of frataxin gene expression (Fig. 3), the degree of silencing and therefore variegation perhaps influenced by the length of the repeat and/or by other modifiers, for example heterochromatin protein 1ß. This work raises the possibility of therapies aimed at modifying chromatin, which may be effective in the treatment of FRDA. For example, condensation of chromatin packaging of DNA is associated with histone deacetylation and, as discussed above, agents are available that inhibit HDACs; these may overcome frataxin repression in FRDA. Interestingly, in cell lines containing constructs of the entire human FRDA functional sequence, upregulation of frataxin expression occurred with butyric acid treatment (Sarsero et al., 2003), an agent known to be an HDAC inhibitor.
|
A CTG expansion behaved similarly in the above transgenic mouse model, causing gene silencing and chromatin condensation (Saveliev et al., 2003
). Other trinucleotide repeat diseases, such as SCA 8, SCA 12 and myotonic dystrophy 1, and perhaps also extending to diseases in which other repeats are found in untranslated regions [e.g. SCA 10 (Matsuura et al., 2000) and myotonic dystrophy 2 (Liquori et al., 2001)], may share a similar molecular pathogenic mechanism. Indeed, it is known that the CTG expansion in myotonic dystrophy 1 alters the chromatin structure of an enhancer present in the 3' region of the myotonic dystrophy protein kinase gene, rendering it inaccessible to transcription factors (Otten and Tapscott, 1995; Klesert et al., 1997), and that CTG repeats are capable of efficiently recruiting nucleosomes, the basic structural element of chromatin (Wang et al., 1994; Wang and Griffith, 1995). Interestingly, the genetic defect in facioscapulohumeral muscular dystrophy (FSHD) is the deletion of subtelomeric repeats on chromosome 4. It has been suggested that these repeats normally serve to repress nearby genes, perhaps through an effect on chromatin packaging. With deletion of these repetitive regions in facioscapulohumeral muscular dystrophy, the nearby genes are suggested to become dysregulated (Gabellini et al., 2002).
The frataxin protein
Frataxin is a mitochondrial protein of 210 amino acids and is localized to the inner mitochondrial membrane (Campuzano et al., 1997; Koutnikova et al., 1997). The N-terminal mitochondrial targeting sequence is proteolytically cleaved within the mitochondrion (Koutnikova et al., 1998). The precise role of frataxin within the mitochondrion remains unclear and several functions have been suggested, which may not be mutually exclusive. Studies with yeast frataxin homologue 1 protein (Yfh1p) provided the first evidence that frataxin is involved in mitochondrial iron balance. Disrupting Yfh1p resulted in a large mitochondrial iron excess, impairment of oxidative phosphorylation, increased sensitivity to oxidant stress and damage to mitochondrial DNA (Babcock et al., 1997; Wilson and Roof, 1997). Reactive oxygen species formed within the iron overloaded Yfh1p-deficient mitochondrion are thought to be toxic to mitochondrial DNA and proteins and also to nuclear DNA (Karthikeyan et al., 2003b).
Lately, further characterization of Yfh1p has clarified its role in iron–sulphur cluster (ISC) synthesis. ISCs are cofactors for proteins involved in metabolic processes involving electron transfer and the mitochondrion is essential to ISC synthesis. Frataxin is a component of the ISC synthetic machinery acting early in the process, perhaps loading the system with iron, in concert with other proteins of the ISC assembly mechanism e.g. Isu1, Isu2 and Nfs1 (Muhlenhoff et al., 2002, 2003; Gerber et al., 2003). Export of iron from the yeast mitochondrion is also dependent on ISC synthesis (Chen et al., 2002b). Microarray analysis of gene expression in human cells has confirmed that several genes in the ISC biosynthetic pathway, in addition to several in the sulphur amino acid pathway, are frataxin-dependent and in frataxin-deficient cells there is downregulation of genes involved in ISC metabolism (Tan et al., 2003).
Frataxin may be involved in the response of the cell to oxidative stress. FRDA fibroblasts are hypersensitive to oxidant stress and inhibitors of apoptosis and iron chelators prevented death in these cells, suggesting that oxidative stress induces apoptosis and hence neurodegeneration (Wong et al., 1999). In a ce