Replication-mediated instability of the GAA triplet repeat mutation in Friedreich ataxia

Abstract

Friedreich ataxia is caused by the expansion of a polymorphic and unstable GAA triplet repeat in the FRDA gene, but the mechanisms for its instability are poorly understood. Replication of (GAA*TTC)n sequences (9-105 triplets) in plasmids propagated in Escherichia coli displayed length- and orientation-dependent instability. There were small length variations upon replication in both orientations, but large contractions were frequently observed when GAA was the lagging strand template. DNA replication was also significantly slower in this orientation. To evaluate the physiological relevance of our findings, we analyzed peripheral leukocytes from human subjects carrying repeats of similar length (8-107 triplets). Analysis of 9400 somatic FRDA molecules using small-pool PCR revealed a similar mutational spectrum, including large contractions. The threshold length for the initiation of somatic instability in vivo was between 40 and 44 triplets, corresponding to the length of a eukaryotic Okazaki fragment. Consistent with the stabilization of premutation alleles during germline transmission, we also found that instability of somatic cells in vivo and repeats propagated in E.coli were abrogated by (GAGGAA)n hexanucleotide interruptions. Our data demonstrate that the GAA triplet repeat mutation in Friedreich ataxia is destabilized, frequently undergoing large contractions, during DNA replication.

Figure 1

(GAA•TTC)_n constructs used to analyze the effect of differential orientation of replication on repeat instability. Several lengths of the (GAA•TTC)_n sequence were cloned into the PstI/XbaI sites of pUC19 in both orientations with respect to the unidirectional ColE1 origin of replication (arrow). Repeat-containing constructs are depicted in the ‘GAA’ or ‘TTC’ orientations, with repeat lengths of n = 9, 21, 48, 82, 105 and 111 in the ‘GAA’ orientation, and n = 8, 21, 48, 79, 105 and 108 in the ‘TTC’ orientation. Additionally, two interrupted tracts of 101^* and 95^* repeats were cloned in the ‘GAA’ and ‘TTC’ orientation, respectively. The black boxes flanking the repeat represent minimal flanking sequence from intron 1 of the FRDA gene.

Figure 2

Slowed replication and large contractions of the (GAA•TTC)_n repeat when GAA is the template for lagging strand synthesis. Closed diamond = GAA-82, closed square = TTC-79, closed triangle = random sequence control and all error bars reflect ± SD (A) Representative PCR products generated from colonies obtained by plating the glycerol stock of a single culture (16 h) are shown. The repeat tract was significantly more unstable and prone to contractions when GAA was the template for lagging strand synthesis. Arrows indicating the position of GAA-40 and TTC-40 represent the cutoff (∼50%) used for defining small versus large contractions. (B) Slower growth of E.coli and significantly blunted log phase when GAA was the template for lagging strand synthesis compared with TTC or random sequence control. (C) Slower plasmid DNA replication when GAA was the template for lagging strand synthesis compared to TTC or random sequence control, as determined by dot blot analysis. RDU, relative densitometric units. (D) Percentage of colonies containing (GAA•TTC)_n repeats of altered length (% instability) after 12, 16, 20 and 24 h of culture indicates that there was a significant increase in instability over time for GAA-82 versus TTC-79 (P = 0.01 at 20 and 24 h). (E) Large contractions (>50% loss of initial repeat length) were significantly more frequent with GAA-82 versus TTC-79 (P < 0.001). The median length of contraction products (indicated by horizontal lines) was significantly shorter for GAA-82 (25.5 repeats) than for TTC-79 (49.5 repeats) (P < 0.001). Only contraction events are shown in the graph, with the magnitude of change (in triplets) plotted on the y-axis. (F) Large contractions accumulated throughout the log phase when GAA was the template for lagging strand synthesis.

Figure 3

SP-PCR analysis of (GAA•TTC)_n alleles with 78–105 uninterrupted repeats shows instability and a contraction bias in somatic cells in vivo. (A) A representative Southern blot showing multiple ‘small pools’ of FRDA molecules from human genomic DNA containing 9 and 78 GAA triplet repeats. Somatic instability in vivo comprised frequent small contractions/expansions and some large contractions into the normal, non-disease size range. (B) Summary of mutations detected following SP-PCR analysis of 2520 individual FRDA molecules containing (GAA•TTC)_n with n = 78, 81, 91, or 105 uninterrupted repeats. The x-axis represents the magnitude of change (%) from the constitutional (most common) GAA triplet repeat length, determined by sequencing, with negative and positive readings indicating contractions and expansions, respectively. Non-mutant bands are not plotted.

Figure 4

The threshold length for the initiation of somatic instability of the GAA triplet repeat at the FRDA locus is between 40 and 44 uninterrupted triplets. SP-PCR was performed on 5593 individual FRDA molecules, with allele sizes ranging from 8 to 66 uninterrupted GAA triplet repeats, i.e. spanning the normal and PM allele range. Representative Southern blots of SP-PCR amplifications are shown. (A and B) SP-PCR analysis of (GAA)₃₀ and (GAA)₃₉ alleles showed complete stability in vivo. (C and D) SP-PCR analysis of (GAA)₄₄ and (GAA)₆₆ showed somatic instability in vivo.

Figure 5

(GAGGAA)_n hexanucleotide interruption stabilizes the GAA triplet repeat at the FRDA locus in somatic cells in vivo. (A) Representative SP-PCR amplifications of a (GAA)₁₀₇ allele containing uninterrupted GAA triplet repeats at the FRDA locus showed somatic instability. (B) Representative SP-PCR amplifications of a (GAA)_114^* allele interrupted by a (GAGGAA)₅ hexanucleotide sequence showed no appreciable somatic instability.

Figure 6

(GAGGAA)_n hexanucleotide interruptions stabilize the (GAA•TTC)_n triplet repeat, despite slower bacterial growth, when propagated in plasmids in E.coli. All error bars reflect ± SD. (A) Growth curves generated from OD₆₀₀ measurements at each time point from 9 to 24 h of growth, in triplicate, are shown for E.coli containing pure GAA-111 (filled diamond) and pure TTC-108 (filled square) plasmids. (B) Growth curves generated from OD₆₀₀ measurements at each time point from 10–24 h of growth, in triplicate, are shown for E.coli transformed with interrupted GAA-101^* (filled diamond) and pure GAA-111 (filled square) plasmids. (C) Repeat instability was determined for pure and interrupted (asterisk) (GAA•TTC)_n repeat sequences in both orientations after 22 h of culture. The bar graph shows the percentage of colonies containing repeat tracts of altered length (% instability). GAA-111 and TTC-108 were significantly more unstable than GAA-101^* and TTC-95^*, respectively. The stabilizing effect of the hexanucleotide interruption was more pronounced in the GAA orientation.

Figure 7

Model depicting the genesis of large contractions during lagging strand synthesis of GAA triplet repeat sequences (in eukaryotic replication). ‘H’, helicase; ‘PCNA’, proliferating cell nuclear antigen; RPA, replication protein A. Polδ (large rectangle) replicates the leading strand at the advancing fork. Polα/primase (gray filled circle) initiates Okazaki fragment synthesis, and polδ elongates the nascent Okazaki fragment during lagging strand synthesis. The single-stranded lagging strand template is bound by RPA). Non-repeat sequence and the complementary TTC repeat sequence are shown as solid black lines, and the GAA triplet repeat is shown as a dotted line. When the GAA triplet repeat sequence expands beyond the length of an Okazaki fragment (the threshold length of 40–44 triplets at the FRDA locus), individual fragments would have to be initiated, elongated and ligated within the length of the repeat tract. We propose that the 50-fold reduced affinity of RPA for purine-rich sequences [compared with pyrimidine-rich sequences (34)] would allow the single-stranded GAA strand to adopt stable/metastable secondary structures (‘?’), which would result in bypassing of a variable number of GAA repeats in the nascent Okazaki fragment, thus resulting in the experimentally observed contractions. Large contractions (and the under-representation of intermediate sized contractions) in vivo involving alleles with 78–105 repeats could stem from the minimum length required for the secondary structure(s) to be stable.