Friedreich's ataxia (FRDA) is characterized by progressive sensory ataxia with onset before 25 years of age, areflexia, loss of position and vibration senses, dysarthria, and extensor plantar responses.1 Patients have primary degeneration of dorsal root ganglia (DRGs), associated with axonal degeneration of the posterior columns, spinocerebellar tracts, and corticospinal tracts, and large myelinated fibers in peripheral nerves.2 In later stages, the cerebellum may be affected; however, other regions of the nervous system remain unaffected. Two-thirds of patients have cardiomyopathy. Although the rate of progression is variable, the mean age of loss of ambulation is 25 years, and patients often die prematurely.1
FRDA is an autosomal recessive disease, and most patients are homozygous for expanded GAA triplet-repeat sequences (E alleles) in intron 1 of the FXN gene.3 E alleles interfere with transcription in a length-dependent manner,4 resulting in deficiency of the mitochondrial protein frataxin.5 E alleles contain 66 to 1,700 triplets, and the severity of disease and rate of progression correlate with repeat length.6–8
DRG neurons are hypersensitive to frataxin deficiency as seen in neuronal-specific, conditional, frataxin knock-out mice.9 However, this mouse model did not accurately mimic the situation in FRDA, because the mice had normal levels through 1 month of life followed by a complete lack of frataxin, as opposed to FRDA patients who have a relative frataxin deficiency, estimated to be 5 to 29% of normal.5 The cause of the progressive neurodegeneration in FRDA remains unexplained.
E alleles are commonly detected by conventional polymerase chain reaction (PCR) or Southern blot analysis; however, these assays estimate only the repeat length of the “constitutional or most common” allele (E allele). Using small-pool PCR (SP-PCR), a sensitive assay that utilizes low levels of genomic DNA (typically 6–600pg) to detect GAA triplet repeats derived from individual FXN molecules (genes), we have previously shown that E alleles are highly unstable in somatic cells in vivo. Long E alleles (>500 triplets) showed a strong contraction bias in peripheral leukocytes from patients,10 whereas short E alleles (<250 triplets) had a tendency to expand.10, 11 In carriers of borderline alleles (44–66 triplets), somatic instability was required for the development of the FRDA phenotype.11
We hypothesized that progressive degeneration of DRGs may stem from somatic instability, resulting in a progressive accumulation of further large expansions in an age-dependent manner. Here, we demonstrate that the GAA triplet-repeat sequence in FRDA displays age-dependent, progressive accumulation of large expansions in DRGs. Our data support the paradigm developed to explain the progression of Huntington's disease and myotonic dystrophy, and indicate that this may be a common theme in the pathogenesis of disparate triplet-repeat diseases.
- Top of page
- Materials and Methods
To detect tissue-specific patterns of somatic instability, we performed SP-PCR analysis10, 11 of multiple tissues derived from autopsies of six FRDA patients. Onset of disease occurred between 5 and 11 years of age (8 ± 2.2), and disease duration ranged from 12 to 38 years (26 ± 11.5). All patients died of sequelae of cardiomyopathy at 17 to 47 years of age (34 ± 13.3), and the median postmortem time was 10 hours. Initial analysis using conventional PCR demonstrated that one of the six patients (Patient A-1) was homozygous for E alleles of the same size (950/950 triplets in blood). Because this would allow accurate determination of the magnitude of allelic size variation, we performed a comprehensive study of all available tissues from Patient A-1 (Fig 1; Table), and subsequently performed analysis of select tissues from the other five autopsies.
Figure 1. Small-pool polymerase chain reaction (SP-PCR) showing greater prevalence of large expansions in dorsal root ganglia (DRGs). (A) Representative Southern blots showing distribution of somatic variability in multiple tissues of Patient A-1 (each lane contains “small pools” of 12––20 individual FXN molecules derived from the indicated somatic tissue). Arrowheads indicate the position of the constitutional allele determined by conventional PCR. Braces to the left of the gels indicate large expansions in DRG and cerebellum, and the bracket to the left of the gel indicates the location of large contractions in blood. DNA size markers are shown as dashes along the right margin, which indicate the relative positions of 1, 2, 3, and 4kb from the bottom of the gel. The lanes marked B indicate “blank reactions” performed with no added DNA template. (B) Frequency distribution (plotted on the y-axis) of expansions (magnitude plotted on the x-axis as increase in size [%] over the constitutional allele) seen in various tissues derived from Patient A-1. This graph represents the cumulative analysis from multiple gels as depicted in (A) containing reactions with 11 to 24 individual molecules per lane. All data points to the right of the bold line, plotted at 20%, represent large expansions (>20% of the constitutional allele size).
Download figure to PowerPoint
Table . Somatic Instability of the Expanded GAA Triplet Repeat in Multiple Tissues
|Tissues||Molecules Analyzeda||ML||Large Contractionb||Large Expansionb|
|Blood||1347||567 (42.1%)||206 (15.3%)||3 (0.2%)|
|Cerebrum||1249||274 (21.9%)||17 (1.4%)||19 (1.5%)|
|Cerebellum||1008||273 (27.1%)||25 (2.5%)||12 (1.2%)|
|DRG||1014||362 (35.7%)||31 (3.1%)||65 (6.4%)|
|Spinal cord||269||99 (36.8%)||22 (8.2%)||1 (0.4%)|
|Brainstem||853||290 (34%)||45 (5.3%)||2 (0.2%)|
|Heart||711||193 (27.1%)||31 (4.4%)||11 (1.5%)|
|Total||6451||2058 (31.9%)||377 (5.8%)||113 (1.8%)|
SP-PCR analysis of 6,451 individual FXN molecules derived from 7 tissues of Patient A-1 showed 2,058 molecules (31.9%) that differed significantly in length (>5%) from the constitutional E allele (see the Table). Of these, 377 (5.8%) represented large contractions (≤50% of the constitutional allele length), and 113 (1.8%) were large expansions (≥20% of the constitutional allele length). There was a greater frequency of large expansions in DRGs (p < 0.001; see the Table). More than half (57%) of all 113 instances of large expansions occurred in DRGs, and DRGs were the only tissues with more large expansions (6.5%) than large contractions (3.1%). Most large expansions represented gains of 20 to 40%, although occasional increases above 70% were also seen (see Fig 1). In contrast, all other tissues showed a much lower frequency (≤1.5%) of large expansions. The frequency of large contractions was also not uniform across tissues; it was low in DRGs, cerebellum, cerebrum, and heart; intermediate in brainstem and spinal cord; and greatest in blood (see the Table).
The observed mutational spectrum was classified into six different sized bins: small (5–20%), medium (20–50%), and large (≥50%) contractions; small (5–20%) and large (≥20%) expansions; and unchanged, if variation was less than 5%. Homogeneity tests (α = 0.01) were performed using likelihood ratio χ2 decomposition and Kolmogorov–Smirnov two-sample test. DRGs were unique for the presence of large expansions and blood for large contractions. Two additional tissue subgroups were identified based on the inability to reject homogeneity in all six bins: brainstem and spinal cord, and cerebrum and cerebellum. This indicates that tissue-specific factors determine the type of somatic instability in different tissues. DRGs were notable for the low prevalence of large contractions and high prevalence of large expansions.
To confirm the observation of DRG-specific expansions, we analyzed somatic instability from five additional autopsies using the following tissue samples: five DRG, three cerebellar, three spinal cord, and two blood sample. These tissues were derived from patients who had constitutional E alleles ranging in size from 350 to 1,030 triplets. Large expansions were conservatively estimated as those that were ≥20% of the larger of two constitutional alleles. Again, the cumulative frequency of large expansions was significantly greater in DRGs compared with all other tissues (6.9 vs 1.5%; p < 0.0001).
The frequency of large expansions in DRGs, which ranged from 0.5% at 17 years to 13.9% at 47 years, showed a significant correlation with patient age (r = 0.78; p = 0.028; Figs 2A, B) and tended to correlate with disease duration (r = 0.74; p = 0.056). No correlation was found with the length of the constitutional E allele (p = 0.15 and p = 0.16 for the longer and shorter E alleles, respectively). No such tendency for age-dependent, large expansions was detected in cerebellum or spinal cord, which were available from four of these individuals (p = 0.32 and p = 0.1, respectively; see Fig 2C). These data indicate that somatic instability in FRDA results in the progressive accumulation of large expansions in DRGs.
Figure 2. Age-dependent increase of large expansions in dorsal root ganglia (DRGs) from Friedreich's ataxia (FRDA) patients. (A) Representative small-pool polymerase chain reaction (SP-PCR) results showing increasing levels of large expansions (>20% of the constitutional allele size) in DRG samples derived from patients aged 17, 24, and 47 years (indicated by braces). Each lane contains “small pools” of 10 to 15 individual FXN molecules. Arrowheads indicate the location of the respective constitutional alleles in the three patients. Note that Patient A-1 (24 years) is homozygous for the same sized E allele as determined by conventional PCR. DNA size markers are shown as dashes along the right margin, which indicate the relative positions of 1, 2, 3, 4, and 5kb from the bottom of the gel. The lanes marked B indicate “blank reactions” performed with no added DNA template. (B) Large expansions (%; y-axis) are shown plotted against age (in years; x-axis) for DRG samples (solid circles) derived from six FRDA patients. Their ages are 17, 24, 26, 36, 46, and 47 years. This graph represents the cumulative analysis from multiple gels as depicted in (A) containing reactions with 10 to 15 individual molecules per lane. (C) Large expansions (%; y-axis) are shown plotted against age (in years; x-axis) for spinal cord (triangles) and cerebellar (squares) samples derived from four FRDA patients. They are four of the six patients shown in (B), and their ages are 17, 24, 36, and 47 years. This graph represents the cumulative analysis from multiple gels containing reactions with 12 to 20 individual molecules per lane. As opposed to the DRG (shown in B), there was no age-dependent increase in large expansions in cerebellum and spinal cord (p = 0.32 and p = 0.1, respectively).
Download figure to PowerPoint
- Top of page
- Materials and Methods
The age-dependent accumulation of large expansions in DRGs may be, at least in part, the reason for their progressive degeneration in FRDA patients. The observation of tissue-specific, age-dependent expansion of triplet repeats at the primary site of pathology was also noted in patients with Huntington's disease12–14 and myotonic dystrophy,15–17 wherein patients have larger repeats in their striatum and muscle, respectively. Notwithstanding the absence of such a phenomenon in SCA1 patients,18 our present observations in FRDA lend further support to this paradigm for disease pathogenesis. Although DRG are inherently hypersensitive to frataxin deficiency, the progressive accumulation of large expansions may serve as a modulator of pathogenesis, thereby causing their early and progressive degeneration.
Whereas the expanded (CAG · CTG)n triplet repeats in Huntington's disease and myotonic dystrophy have a tendency for further expansion,12–17 E alleles in FRDA have a tendency to contract in somatic cells.10 Therefore, the large GAA triplet-repeat expansions in DRGs are even more significant and suggest the existence of a unique mechanism that permits expansions in certain somatic cells. It is also notable that three of the four tissues with the lowest frequency of large contractions were DRGs, cerebellum, and heart; that is, tissues that are primarily affected in FRDA. Therefore, the combination of a high frequency of large expansions and paucity of large contractions in DRG may help modulate their involvement in FRDA.
Although our data do not show that the age-dependent expansions are occurring specifically in neurons, it is striking that on analysis of several central nervous system subregions we detected large expansions only in the primary site of pathology in FRDA. There is no detailed analysis of age-dependent pathology in DRGs of FRDA patients. Analysis of the two transgenic mouse models of FRDA9, 19 demonstrated age-dependent increase in numbers of DRG neuronal bodies with signs of pathological involvement (accumulation of one or more vacuoles), which coincided with the development of gait abnormalities, but these changes clearly preceded the development of overt neuronal degeneration. The age-dependent accumulation of large expansions in the DRGs is compatible with two nonmutually exclusive models of pathogenesis. It is possible that the large expansions are within neurons, and thereby directly determine cellular susceptibility. Long after the development of physical signs, the FRDA mouse models showed 1 to 7 vacuolated neurons in 50 to 60% of DRG sections, which is estimated to represent 10 to 30% of the neurons. Because our patients span a large age range (17–47 years) it is conceivable that they would have a growing proportion of “sick” DRG neurons, and the progressive accumulation of large expansions from 0.5% at 17 years to 13.9% at 47 years supports this idea, given that our PCR assay only (mostly) identifies living cells. Indeed, it is plausible that we may have underestimated the prevalence of large expansions because they could be underrepresented in older individuals. However, a second model of disease pathogenesis is also possible, wherein large expansions target glial cells, thus indirectly resulting in progressive neuronal pathology, similar to what was recently reported for SCA7.20
A potential caveat is that somatic instability may be due to the progressive gliosis that occurs in DRG of FRDA patients. However, we recently found a similar frequency (7.2%) and magnitude (20–40% gain) of expansion occurring in an age-dependent fashion in DRGs of mice that were transgenic for a randomly inserted YAC containing the entire human FXN gene with 190 GAA repeats.21 Because instability occurred in the absence of any pathology, neither reactive gliosis nor oxidative damage associated with FRDA pathology are required for the age-dependent expansion of GAA triplet repeats. Therefore, the large expansions are unlikely to be a consequence of the progressive pathology in DRG, and our data are compatible with the possibility that they may precede the development of tissue pathology.
The incremental somatic instability in postmitotic tissues indicates that erroneous DNA replication alone is unlikely to be the mechanism. It remains to be seen whether mismatch repair proteins, critical for the development of somatic instability of (CAG · CTG)n repeats,22 also mediate somatic instability of the GAA triplet-repeat sequence. However, neither greater levels of FXN expression nor oxidative DNA damage are likely to be the cause of the expansions seen in DRG, because the heart, which also shows high frataxin expression3, 5 and oxidative damage,23 did not show similar expansions. This may also explain the invariably later development of cardiomyopathy compared with DRG pathology.
In summary, the expanded GAA triplet-repeat sequence in FRDA displays age-dependent and tissue-specific instability in vivo. The progressive accumulation of large expansions specifically in DRG, coupled with a lower frequency of large contractions, suggests that the progressive involvement of DRG in FRDA may be, at least in part, due to somatic instability.
- Top of page
- Materials and Methods
This research was supported by the NIH (National Institute of Neurological Disorders and Stroke, NS047596, S.I.B.), Muscular Dystrophy Association (MDA 3973, S.I.B.), Oklahoma Center for the Advancement of Science and Technology (OCAST; HR05-009, S.I.B.); Friedreich Ataxia Research Alliance (FARA, S.I.B.); Idea Network of Biomedical Research Excellence (INBRE, D.E.); Consejo Nacional de Ciencia y Technologia (CONACYT, A.R.), Fondo per gli Investimenti della Ricerca di Base (FIRB; S.C.), The Wellcome Trust (M.P.), and a postdoctoral fellowship from the National Ataxia Foundation (I.D.).
We thank the patients and their families for participation in our research program.