A Genetic Defect Caused by a Triplet Repeat Expansion in Arabidopsis thaliana

See allHide authors and affiliations

Science  20 Feb 2009:
Vol. 323, Issue 5917, pp. 1060-1063
DOI: 10.1126/science.1164014


Variation in the length of simple DNA triplet repeats has been linked to phenotypic variability in microbes and to several human disorders. Population-level forces driving triplet repeat contraction and expansion in multicellular organisms are, however, not well understood. We have identified a triplet repeat–associated genetic defect in an Arabidopsis thaliana variety collected from the wild. The Bur-0 strain carries a dramatically expanded TTC/GAA repeat in the intron of the ISOPROPYL MALATE ISOMERASE LARGE SUB UNIT1 (IIL1; At4g13430) gene. The repeat expansion causes an environment-dependent reduction in IIL1 activity and severely impairs growth of this strain, whereas contraction of the expanded repeat can reverse the detrimental phenotype. The Bur-0 IIL1 defect thus presents a genetically tractable model for triplet repeat expansions and their variability in natural populations.

Copy number variation in tandemly repeated, short DNA sequences can underlie phenotypic variability in microbes (13), and several human disorders are caused by expansions of trinucleotide (or triplet) repeats (46). DNA polymerase slippage and unequal crossing over affect repeat numbers (7, 8), but how repeat variation arises and is maintained within a population is largely unknown.

To uncover cryptic genetic variation in Arabidopsis thaliana, we grew strains, collected from the wild, at 27°C and with short days—an environment not normally encountered by this species in its natural range (9). Most strains appeared normal (Fig. 1A), but the Bur-0 strain suffered from pleiotropic growth abnormalities and never progressed to the flowering stage. Because of the misshapen leaves (Fig. 1B), we refer to this phenotype as “irregularly impaired leaves” (iil). At 23°C, the iil defects disappeared (Fig. 1C), whereas the phenotype was even worse at 30°C (long days) (fig. S1) (10). Similarly, higher light intensity worsened the iil phenotype (Fig. 1, D and E). Leaf primordia arose normally (fig. S2, A and B), which suggested that the iil phenotype is due to abnormal leaf growth and expansion. The cellular architecture of the mesophyll was disrupted (Fig. 1F), whereas the overall organization of the vasculature was more normal (fig. S2, C to F).

Fig. 1.

The iil phenotype. (A) Four-week-old, normal-appearing Pf-0 plant grown at 27°C (short days). (B) Four-week-old Bur-0 plant at 27°C. Arrowheads indicate misshapen leaves, from about the seventh leaf onward. Inset shows progressive worsening of the phenotype after 10 weeks. (C) Ten-week-old, phenotypically normal Bur-0 plant at 23°C. (D) Six-week-old Bur-0 plant at 27°C, under photosynthetic photon flux density (PPFD) of 140 to 150 μmol m–2 s–1. (E) Bur-0 plant under PPFD of 200 to 210 μmol m–2 s–1. (F) Cross section of expanded leaves of Pf-0 (top) and Bur-0 at 27°C. Scale bars, 1.5 cm, except (F), 200 μm.

To determine the genetic basis of the iil defect, we crossed Bur-0 with another strain, Pf-0. F1 plants were normal, but the iil phenotype reappeared in about 25% of F2 plants. By linkage analysis in this population, we mapped the IIL locus to a 16.7-kilobase (kb) interval. In the sequenced strain Col-0, this region includes two genes fully, At4g13420 and At4g13430, and the first seven exons of At4g13410 (fig. S3). Because cDNA sequencing did not suggest causal polymorphisms in any of the three genes, we surveyed the mapping interval with the polymerase chain reaction (PCR). We discovered a dramatic expansion of a TTC/GAA triplet repeat from 23 copies in Col-0 and Pf-0 to more than 400 copies in the third intron of At4g13430 from Bur-0 (Fig. 2A). In humans, intronic TTC/GAA expansions reduce expression of the affected genes, with TTC repeats having weaker effects on transcript accumulation than GAA repeats (1114). Consistent with this, At4g13430 RNA levels were ∼50% lower in Bur-0 than in Pf-0 or Col-0 at 27°C (Fig. 2B). At 23°C, RNA levels did not differ between genotypes (Fig. 2B).

Fig. 2.

Identification of At4g13430 as IIL1. (A) Location of expanded TTC/GAA repeats in At4g13430. See supporting information for details and fig. S3 for final mapping interval. (B) Real-time reverse transcriptase polymerase chain reaction (RT-PCR) analysis of At4g13430 RNA in leaves of 4-week-old plants, with at least two technical and eight biological replicates. Means ± SEM. Expression levels of Pf-0 and Bur-0 plants at 27°C differed significantly (Student's t test, P < 0.0004). (C) Five-week-old Bur-0 plants at 27°C. (D) Bur-0 plants harboring 35S:IIL1 Bur-0 transgene. (E) Ten-week-old 35S:amiR-IIL1 Bur-0 plant at 23°C. (F) Five-week-old 35S:amiR-IIL1 Bur-0 plant at 30°C (long days). Scale bars, 1.5 cm.

To test whether reduced At4g13430 RNA levels were the cause for the iil defects, we overexpressed the mature Bur-0 transcript in Bur-0 (fig. S4A). These plants no longer had the iil phenotype, and their growth habit was similar to that of other strains (Fig. 2, C and D). On the basis of these results, we concluded that At4g13430 is IIL.

IIL encodes a chloroplast-localized protein (15). Many other genes encoding chloroplast- and endomembrane-localized proteins were down-regulated in Bur-0 plants grown at 27°C, and this group of genes was overrepresented among all down-regulated genes [chi-square test, P < 0.01 (table S1)]. However, transmission electron microscopy did not reveal any major abnormalities in the chloroplasts of iil plants (fig. S5). IIL is the A. thaliana protein that is most closely related to fungal isopropyl malate dehydratase, an enzyme involved in leucine biosynthesis, and is henceforth referred to as ISOPROPYLMALATE ISOMERASE LARGE SUBUNIT 1, IIL1 (16) (fig. S6). Although irrigation of plants with a leucine solution did not ameliorate the iil defect, analysis of the amino acid content of Col-0 plants with reduced IIL1 activity supports a similar role in A. thaliana (17).

Because a Col-0 line with a transferred DNA insertion at the 5′ end of IIL1 (10) did not have the iil phenotype, we reduced steady-state levels of IIL1 RNA by ∼50% (fig. S4B) using an artificial microRNA (amiRNA) (18). AmiR-IIL1 introduction caused only a mild phenotype in Bur-0 and Pf-0 at 23°C, with slower growth and slightly paler leaves (Fig. 2 and fig. S7A). At elevated temperatures, the phenotype of Bur-0 or Pf-0 amiR-IIL1 plants became more severe, including high mortality (Fig. 2F and fig. S7, B and C). These observations indicated that Bur-0 and Pf-0 require more IIL1 activity as temperatures rise.

Defects caused by amiR-IIL1 were milder in the Col-0 background at all temperatures (fig. S7, D and E). The weaker effects of amiR-IIL1 in Col-0, compared with Bur-0 or Pf-0, were consistent with the segregation of the iil phenotype in a Col-0 × Bur-0 F2 population. Only 23 out of 485 plants were affected, which suggested the presence of iil modifiers in the Col-0 genome. Further mapping in Bur-0 × Col-0 recombinant inbred lines (RILs) (19) revealed a region on chromosome 2 that may interact with the IIL1 interval to produce the iil phenotype (Fig. 3, A and B).

Fig. 3.

Mapping of an IIL1 modifier in Col-0. (A) Quantitative trait locus (QTL) analysis of iil phenotype in 164 Bur-0 × Col-0 RILs. Chromosome 2 has a LOD (logarithm of odds) peak (arrow) just below the significance threshold (red line). (B) Heat map for a two-dimensional genome scan with a two-QTL model. Upper triangle shows epistatic interactions (LOD scale on the left of the color bar), lower triangle additive interactions (scale on the right of the color bar). Although the LOD peak on chromosome 2 does not surpass the significance threshold on its own (A), this region appears to act additively with the IIL1 QTL on chromosome 4.

In mammals, expanded TTC/GAA repeats often display extensive somatic variability (2022). Likewise, we observed both between- and within-individual variability of repeat length in IIL1 (Fig. 4A). We exploited this variability to demonstrate that the triplet repeat expansion causes the iil phenotype in Bur-0. We found 17 phenotypic revertants among 30,000 Bur-0 plants; all 17 showed a reduction in repeat length (Fig. 4B). Because ethyl methane sulfonate (EMS) can promote the contraction of triplet repeats (23, 24), we also screened for suppression of the iil phenotype among 30,000 M2 progeny of selfed, EMS-treated Bur-0 plants. We isolated 156 phenotypic revertants, 27 of which had reduced triplet copy number (Fig. 4B). Although EMS in this system did not strongly promote repeat contraction, the EMS population contained many phenotypically normal lines that had retained the expanded triplet repeats.

Fig. 4.

Analysis of triplet repeat variation. (A) DNA blot analysis of IIL1 triplet repeat–containing region. Su13 and Su15 are two spontaneous revertants in the Bur-0 background. (B) PCR analysis of IIL1 repeat. Bur-0 [(B) ∼2.3-kb fragment with over 400 repeats] and Pf-0 [(Pf) ∼1.1 kb with 23 repeats] are always shown on the right. (Top) Sixteen spontaneous revertants are shown out of 17 spontaneous revertants in which phenotypic suppression was transmitted to the next generation. Asterisks indicate that ∼50% of progeny regained the iil phenotype. (Middle) M2 progeny of EMS-treated plants. All lacked the iil phenotype, but only a fraction had a shortened repeat. (Bottom) EMS M2 progeny retaining the iil phenotype. (C) IIL1 expression, measured by real-time RT-PCR. (D) RNA blots revealed a larger IIL1 transcript isoform in Bur-0 (arrows): (top) after 30 min', (middle) after 3.5 hours' exposure. (Bottom) Ethidium bromide staining of the RNA shown to indicate loading. (E) RT-PCR analysis (30 cycles) of cDNA from plants grown at 27°C, with genomic DNA (gDNA) as control. Amplified fragments are indicated on top. Intron retention is reduced in Su15 compared to Bur-0, and not detected in Pf-0. Sequencing of the PCR product confirmed that the adjacent introns were not retained. See table S2 for the sequences of primers used. (F) Distribution of IIL1 repeat copy number in 96 A. thaliana strains and in MN47 strain of A. lyrata (asterisk).

Stable phenotypic suppression was confirmed in the offspring of 13 spontaneous revertants. None had more than 120 triplet repeats left, which supported the hypothesis that the expanded triplet repeats caused the iil phenotype. In agreement, IIL1 expression was increased in the revertants compared with the Bur-0 parent (Fig. 4, C and D, and fig. S4C). Reducing IIL1 expression again in these lines with amiR-IIL1 resulted in a severe iil phenotype under higher temperatures (fig. S7F). We conclude that a decrease in IIL1 expression caused by the expanded triplets, coupled with higher requirements of IIL1 activity under elevated temperatures, is the basis of the iil phenotype in Bur-0 plants.

RNA blots revealed longer IIL1 transcript isoforms of minor abundance in Bur-0, independent of growth temperature (Fig. 4D). PCR analyses showed that this isoform retained only the intron containing the triplet expansion (Fig. 4E). The relatively weak effects are consistent with reports that GAA repeats are more likely than UUC repeats to affect splicing in mammalian cells (25). However, because normal splicing is not completely restored in a spontaneous revertant with an intermediate repeat length (Fig. 4E), the splicing defects alone cannot explain the observed differences in RNA expression levels. Nevertheless, inefficient splicing might contribute to reduced expression of mature IIL1 transcripts, along with transcriptional defects, epigenetic changes, and post-transcriptional silencing (2628).

If long repeats tended to be detrimental, as in the IIL1 case, one would expect that these are rare in the genome. Indeed, less than 1% of all triplet repeats in the reference A. thaliana genome have six or more copies (table S3), and there is no expressed gene with more than 41 copies (table S4). The Bur-0 allele of IIL1 itself seems to be rare, because we did not find it among 96 other A. thaliana strains nor in Arabidopsis lyrata (figs. S8 and S9). Several strains had either more (up to 36) or fewer than the 23 repeats in the Col-0 reference genome, and two strains had lost the triplets (Fig. 4F). These observations confirm the dynamic nature of the IIL1 triplet repeat. Copy number variability in normal individuals is common for triplet expansion disorders in humans and often underlies genetic anticipation (5, 6).

The A. thaliana Bur-0 allele of IIL1 presents a genetically tractable model for the study of triplet repeat expansion and contraction across multiple generations. The recovery of phenotypic revertants that had retained the expanded IIL1 repeat highlights the potential of the IIL1 triplet repeat for future studies. Some of the apparent second-site mutations might act downstream of IIL1, but others might ameliorate the effects of the triplet repeat expansion itself. In addition, our findings support the argument that simple sequence repeats could be associated with phenotypic variability of evolutionary significance (13).

Supporting Online Material

Materials and Methods

Figs. S1 to S9

Tables S1 to S4


References and Notes

View Abstract

Stay Connected to Science

Navigate This Article