- © 2010 American Society of Plant Biologists
Abstract
DNA double-strand breaks are highly detrimental to all organisms and need to be quickly and accurately repaired. Although several proteins are known to maintain plastid and mitochondrial genome stability in plants, little is known about the mechanisms of DNA repair in these organelles and the roles of specific proteins. Here, using ciprofloxacin as a DNA damaging agent specific to the organelles, we show that plastids and mitochondria can repair DNA double-strand breaks through an error-prone pathway similar to the microhomology-mediated break-induced replication observed in humans, yeast, and bacteria. This pathway is negatively regulated by the single-stranded DNA (ssDNA) binding proteins from the Whirly family, thus indicating that these proteins could contribute to the accurate repair of plant organelle genomes. To understand the role of Whirly proteins in this process, we solved the crystal structures of several Whirly-DNA complexes. These reveal a nonsequence-specific ssDNA binding mechanism in which DNA is stabilized between domains of adjacent subunits and rendered unavailable for duplex formation and/or protein interactions. Our results suggest a model in which the binding of Whirly proteins to ssDNA would favor accurate repair of DNA double-strand breaks over an error-prone microhomology-mediated break-induced replication repair pathway.
INTRODUCTION
Plant cells must ensure the accurate replication and faithful transmission of three different genomes located in the nucleus, mitochondria, and plastids. Although the maintenance of nuclear genome stability in eukaryotes is the subject of an intense research effort, the systems that prevent DNA lesions and rearrangements in organelle genomes remain poorly understood. This is unfortunate as the small genomes of plant organelles encode a variety of proteins with critical functions in oxidative phosphorylation or photosynthesis (Knoop, 2004; Raubeson and Jansen, 2005) and any substantial alteration in their DNA could have devastating consequences for plant growth and development (Newton et al., 2004).
Despite the fact that plant organelle genomes are constantly exposed to reactive by-products of the electron transport machineries, they are surprisingly more stable than the nuclear genome at the nucleotide level (Wolfe et al., 1987; Drouin et al., 2008). Contrasting with their slow rate of nucleotide substitution, plant organelle genome organization is complex and highly dynamic. Indeed, DNA rearrangements resulting from spurious recombination between repeated sequences occur frequently in mitochondria and plastids (André et al., 1992; Maul et al., 2002). The prevention and repair of mutations as well as the regulation of recombination events require extensive genome maintenance systems. Plant organelle genomes encode very few, if any, of the proteins involved in these systems (Mackenzie, 2005; Day and Madesis, 2007). Their maintenance is thus mainly ensured by nuclear-encoded DNA replication, recombination, and repair proteins that are targeted to plastids and/or mitochondria. In recent years, some of these proteins have been identified, including the Arabidopsis thaliana and Physcomitrella patens RecA homologs (Shedge et al., 2007; Odahara et al., 2009), the Arabidopsis MutS-like MSH1 (Abdelnoor et al., 2003), the Arabidopsis organelle single-stranded DNA binding protein OSB1 (Zaegel et al., 2006), and the plastid-targeted Whirlies of Arabidopsis and maize (Zea mays) (Maréchal et al., 2009).
Whirlies form a small family of proteins present mainly in the plant kingdom (Desveaux et al., 2005). Different members of this family can localize to the nucleus, plastids, or mitochondria (Desveaux et al., 2005; Krause et al., 2005; Grabowski et al., 2008; Maréchal et al., 2008, 2009; Prikryl et al., 2008). Whirly proteins preferentially bind single-stranded DNA (ssDNA) (Desveaux et al., 2000, 2002, 2005) and perform numerous activities related to DNA metabolism, including the regulation of transcription (Desveaux et al., 2000, 2004; Xiong et al., 2009) and modulation of telomere length (Yoo et al., 2007). Interestingly, the maize Whirly protein WHY1 interacts with DNA from throughout the plastid genome but also binds to a subset of plastid RNAs and participates in chloroplast RNA metabolism (Prikryl et al., 2008). Resolution of the crystallographic structure of the Solanum tuberosum WHY1 protein revealed that Whirly proteins form tetramers resembling whirligigs. However, the precise mode of single-stranded nucleic acid binding, as well as the sequence specificity of Whirlies, has so far remained elusive (Desveaux et al., 2002).
The nuclear genome of Arabidopsis encodes three Whirly proteins: WHY1 and WHY3, which are targeted to plastids, and WHY2, which localizes to mitochondria (Krause et al., 2005; Maréchal et al., 2008, 2009). Recently, we demonstrated that in the absence of WHY1 and WHY3, the plastid genome of Arabidopsis becomes unstable and eventually accumulates rearranged DNA molecules causing chloroplast defects (Maréchal et al., 2009). This phenomenon is not unique to Arabidopsis, as rearranged DNA molecules also accumulate in the plastids of maize lines with reduced levels of Zm-WHY1 (Maréchal et al., 2009). The rearranged molecules correspond to deletion, duplication, and/or circularization events that are mediated by DNA microhomologies (Maréchal et al., 2009). Interestingly, microhomology-mediated DNA rearrangements (MHMRs) are also present in the nuclear genomes of humans, yeast, and bacteria (Hastings et al., 2009a). In humans, these MHMRs are associated with numerous copy number variants, as well as with several diseases (Hastings et al., 2009a, 2009b; Lupski, 2009; Conrad et al., 2010), and appear following DNA-related stresses, notably in response to double-strand breaks (DSBs) (Hastings et al., 2009a).
The similarity of the DNA rearrangements observed in plastids of Whirly mutants and in other species led us to suspect that the plastid genome rearrangements could be due to the error-prone repair of DNA following DNA damage and that the role of Whirly proteins in organelles could be to promote the accurate repair of DNA. We tested this hypothesis by treating wild-type and Whirly mutant plants with ciprofloxacin, a DSB-inducing genotoxic agent that targets the enzymes gyrases, which are localized in plant organelles (Wall et al., 2004). Our results indicate that DSBs lead to an accumulation of DNA rearrangements, including MHMRs, in both plastid and mitochondrial genomes. Importantly, plants lacking plastid or mitochondrial Whirlies accumulate higher levels of MHMRs than do the wild-type controls, suggesting that Whirlies might be components of the organelle repair machinery. To gain insights into the mechanisms by which Whirlies prevent DSB-induced rearrangements, we obtained several crystal structures of a mitochondrial Whirly protein in complex with ssDNA oligonucleotides. These structures reveal how Whirlies bind and protect the single-stranded form of DNA in a nonsequence-specific fashion. We propose a model in which Whirlies help maintain genome integrity under DNA stresses by binding to ssDNA regions of the genome and favoring conservative over error-prone repair pathways.
RESULTS
DSB Induces MHMR Accumulation in the Plastids of Arabidopsis
To test the hypothesis that Whirlies are important for the accurate repair of DNA lesions, Columbia-0 (Col-0; wild type) and a why1 why3 double knockout (formerly called KO1/3; Maréchal et al., 2009) line of Arabidopsis were treated with the quinolone ciprofloxacin. Quinolones are inhibitors of organelle-localized DNA gyrases (Wall et al., 2004) and induce DSBs in DNA (Gellert et al., 1977; Sugino et al., 1977; Snyder and Drlica, 1979; Chen et al., 1996; Rowan et al., 2010). The effect of ciprofloxacin on Arabidopsis wild-type plant survival was monitored and a LD50 (dose required to kill half the members of the tested population) of 0.5 μM ciprofloxacin was calculated (see Supplemental Figure 1 online). The error-prone repair of plastid DNA was assayed by monitoring the formation of DNA rearrangements using the PCR approach illustrated in Figure 1 and primer pairs scattered throughout the plastid genome (Maréchal et al., 2009).
PCR Strategy to Detect DNA Rearrangements.
Outward-facing PCR primers are used to monitor DNA duplication/circularization events, whereas inward-facing PCR primers detect deletions. In both cases, PCR amplification occurs only if a DNA rearrangement brings together the annealing sites of the primers. Gray areas represent repeated sequences.
As expected, treatment of Arabidopsis with ciprofloxacin induces leaf etiolation/variegation and inhibits seedling growth (Figure 2A) (Wall et al., 2004). At 0.25 μM ciprofloxacin, Arabidopsis wild-type plants displayed only slight growth delay and only 9% of the plants displayed etiolation/variegation (Figure 2B). At this same concentration, why1 why3 plants displayed greater sensitivity to ciprofloxacin, with 88% of why1 why3 plants displaying etiolation/variegation (Figure 2B). At 0.75 μM ciprofloxacin, both genotypes were seriously affected and true leaves were unable to develop. We then monitored the DNA rearrangements in the plastid genomes of these plants using the PCR approach described in Figure 1. Figure 2C shows that whereas DNA rearrangements were almost undetectable in untreated wild-type plants, they were abundant and diverse in untreated why1 why3 plants, in agreement with our previous findings (Maréchal et al., 2009). Treatment of plants with 0.25 μM ciprofloxacin led to a stronger increase in the number of DNA rearrangements in why1 why3 (13 additional PCR products) compared with wild-type plants (four additional PCR products) (Figure 2D). At 0.75 μM ciprofloxacin, both genotypes accumulated high levels of rearranged DNA molecules (Figure 2D).
DNA Rearrangements Accumulate in Plastids Following Treatment with the Gyrase Inhibitor Ciprofloxacin.
(A) and (E) Phenotypic effects of various concentrations of ciprofloxacin (A) or novobiocin (E) on wild-type (WT) and why1 why3 Arabidopsis plants. Plants were grown for 3 weeks on solid media containing the indicated concentrations of ciprofloxacin (CIP) or novobiocin (NOV).
(B) and (F) Histograms showing the average ± sd of plants with etiolated/variegated leaves at each ciprofloxacin (B) or novobiocin (F) concentration for wild-type and why1 why3 genotypes. At least three independent experiments were done. Plants with partially or fully white first true leaves were scored as etiolated/variegated. ND, etiolated/variegated leaves could not be counted at 0.75 μM ciprofloxacin and 100 μM novobiocin because they were very small or absent.
(C) and (G) Electrophoretic analysis of representative PCR performed with 13 outward- or inward-facing plastid genome-directed PCR primers on total leaf DNA of wild-type and why1 why3 plants treated with ciprofloxacin (C) or with 10 outward- or inward-facing plastid genome-directed PCR primers on total leaf DNA of wild-type and why1 why3 plants treated with novobiocin (G). Low cycle amplification of the YCF2 plastid gene was used as a loading control. The oligonucleotides used for each PCR are indicated. Individual bands were cut from the gel, cloned, and sequenced. DNA rearrangements are listed in Supplemental Data Set 1 online.
(D) and (H) Histograms showing the number of PCR products in wild-type and why1 why3 plants as a function of ciprofloxacin (D) or novobiocin (H) concentration.
To ensure that the PCR product accumulation was due to DSB induction rather than DNA gyrase inhibition, wild-type and why1 why3 plants were treated with novobiocin, a DNA gyrase inhibitor that also targets organelle-localized DNA gyrases (Wall et al., 2004) but does not produce DSBs (Gellert et al., 1976, 1977; Sugino et al., 1977). Figures 2E and 2F show that novobiocin inhibited seedling growth to the same extent in both wild-type and why1 why3 lines, with very little effect on leaf etiolation/variegation. However, no increase in the diversity or abundance of DNA rearranged molecules could be observed upon treatment with 100 μM novobiocin, the highest concentration tested, in either wild-type or why1 why3 plants (Figures 2G and 2H). This confirmed that DSB induction, rather than DNA gyrase inhibition, leads to DNA rearrangements in ciprofloxacin-treated plants.
DSBs Induce MHMR Accumulation in the Mitochondria of Arabidopsis
In contrast with Arabidopsis WHY1 and WHY3, no function in DNA metabolism has yet been ascribed to the mitochondria-localized protein WHY2. To verify if this protein has a role in maintaining mitochondrial genome stability under DNA stress conditions, we repeated the experiment with gyrase inhibitors on why2 knockout plants. No phenotypic differences could be observed between why2-1 (formerly called KO2; Maréchal et al., 2008) and wild-type plants in the presence or absence of ciprofloxacin (Figures 3A and 3B). We then monitored rearrangements in mitochondrial DNA following ciprofloxacin treatment using the PCR approach described above. Figures 3C and 3D show that rearranged DNA molecules are not abundant in wild-type or why2-1 plants at 0 μM ciprofloxacin concentration. However, treatment with 0.25 μM ciprofloxacin led to a large increase in rearranged products for why2-1 plants (18 additional PCR products) compared with wild-type plants (one additional PCR product) (Figure 2D). At 0.75 μM ciprofloxacin, rearranged DNA molecules accumulate to high levels in both wild-type and why2-1 plants.
DNA Rearrangements Accumulate in Mitochondria Following Treatment with the Gyrase Inhibitor Ciprofloxacin.
(A) and (E) Phenotypic effects of various concentrations of ciprofloxacin (A) or novobiocin (E) on wild-type (WT) and why2-1 Arabidopsis plants. Plants were grown for 3 weeks on solid media containing the indicated concentrations of ciprofloxacin (CIP) or novobiocin (NOV).
(B) and (F) Histograms showing the average ± sd of plants with etiolated/variegated leaves at each ciprofloxacin (B) or novobiocin (F) concentration for wild-type and why2-1 genotypes. At least three independent experiments were done. Plants with partially or fully white first true leaves were scored as etiolated/variegated. ND, etiolated/variegated leaves could not be counted at 0.75 μM ciprofloxacin and 100 μM novobiocin because they were very small or absent.
(C) and (G) Electrophoretic analysis of representative PCR performed with 20 outward- or inward-facing mitochondrial genome-directed PCR primers on total leaf DNA of wild-type and why2-1 plants treated with ciprofloxacin (C) or with 10 outward- or inward-facing mitochondrial genome-directed PCR primers on total leaf DNA of wild-type and why2-1 plants treated with novobiocin (G). Low cycle amplification of the COX1 mitochondrial gene was used as a loading control. The oligonucleotides used for each PCR are indicated. Individual bands were cut from the gel, cloned, and sequenced. DNA rearrangements are listed in Supplemental Data Set 1 online.
(D) and (H) Histograms showing the number of PCR products in wild-type and why2-1 plants as a function of ciprofloxacin (D) or novobiocin (H) concentration.
Figures 3E and 3F show that why2-1 and wild-type plants were similarly affected by novobiocin. No rearranged DNA molecules could be detected in plants treated with either 0 or 100 μM novobiocin (Figures 3G and 3H), thus confirming that mitochondrial DNA rearrangements in ciprofloxacin-treated plants depend on DSB induction. These results indicate that, upon induction of DSBs, why2-1 plants accumulate DNA rearrangements in the mitochondrial genome more readily than do wild-type plants.
DNA Rearrangements Are Mediated by Microhomologies
Cloning and sequencing of PCR products such as those shown in Figures 2C, 2G, 3C, and 3G led to the identification of 191 DNA rearrangement events (see Supplemental Data Set 1 online), some of which are presented in Table 1. The DNA rearrangements were classified in two groups: MHMR products that had at least 5 bp of microhomology at the junction and nonhomologous end joining (NHEJ)-like products that had <5 bp of microhomology at the junction (McVey and Lee, 2008). MHMR and NHEJ-like products constituted 83 and 17%, respectively, of all DNA rearrangements and were present in both deletion and duplication/circularization events (see Supplemental Data Set 1 online). Although MHMR junctions varied in sequence, they were frequently A/T rich and often contained homopolymeric tracts (Table 1). Some MHMR junctions appeared recurrently in different genotypes and/or in different plant treatments (see Supplemental Table 1 online) and were counted as independent events. This result suggests that DNA rearrangement hot spots exist. However, these may not all represent biologically relevant hot spots as DSBs could have a nonrandom distribution due to the binding of DNA gyrases at preferred locations (Fisher et al., 1981). The 49 mitochondrial MHMR junctions had a mean size of 12.6 ± 5.2 bp, whereas the 109 plastid MHMR junctions had a mean size of 14.5 ± 5.5 bp. Globally, the size of the MHMR junctions varied from 5 to 37 bp. No major difference was observed between the mean size of plastid junctions from ciprofloxacin-treated and untreated plants and that of mitochondrial junctions from ciprofloxacin-treated plants. This, plus the fact that MHMRs were more abundant in Whirly-deficient than in wild-type organelles (see Supplemental Table 2 online), indicates that a microhomology-dependent repair pathway is common to both organelles and is more active in Whirly-deficient plants. Importantly, rearrangements in the plastid genome of why2-1 mutant plants and in the mitochondria of why1 why3 mutant plants were at wild-type levels (see Supplemental Figure 2 online), thereby demonstrating the specificity of Whirly proteins for their respective organelle. Altogether, our data suggest that Whirlies could prevent break-induced DNA rearrangements of Arabidopsis organelle genomes.
Representative DNA Rearrangements Observed in Arabidopsis Organellar Genomes
DNA Rearrangements Mediated by Microhomologies Are Present as Substoichiometric Molecules in the Mitochondrial and Plastid Genomes
DNA gel blot analysis was performed on HindIII-digested DNA from wild-type and why1 why3 plants using isolated PCR products as probes (see Supplemental Figure 3 and Supplemental Table online). In untreated plants, only the bands corresponding to the wild-type genome were visible in wild-type or why1 why3 plants. Additional bands were detected in wild-type plants treated with 0.75 μM ciprofloxacin and in why1 why3 plants treated with 0.25 or 0.75 μM ciprofloxacin. The fact that the additional bands always have a lower intensity than the band corresponding to the main genome suggests that they are present in substoichiometric amounts compared with the main genome. The DNA gel blot analysis of undigested DNA (see Supplemental Figure 3 and Supplemental Table 3 online) revealed the existence of subgenomic-length molecules in wild-type and why1 why3 plants treated with 0.75 μM ciprofloxacin.
A similar analysis performed on why2-1 and wild-type plants (see Supplemental Figure 4 and Supplemental Table 3 online) indicated that the mitochondria of ciprofloxacin-treated plants also contain additional DNA molecules that are present at substoichiometric levels. Analysis of undigested DNA (see Supplemental Figure 3 and Supplemental Table 3 online) further revealed the existence of subgenomic-length molecules in mitochondria. Together with our previous observations, these results suggest that substoichiometric DNA rearrangements accumulate in wild-type, why2-1, and why1 why3 genotypes following ciprofloxacin treatment and that some of these rearrangements exist as subgenomic-length molecules.
Mitochondrial Recombination Mediated by Short Repeats (50 to 1000 Nucleotides) Is Not Affected in Whirly Mutants
We next investigated the effect of gyrase inhibitors and Whirly mutants on recombination mediated by short repeats (50 to 1000 nucleotides). This type of recombination is often detected in plant mitochondria (Mackenzie, 2005; Maréchal and Brisson, 2010; Woloszynska, 2010) and is modulated by a variety of proteins involved in maintaining mitochondrial genome stability, such as OSB1 (Zaegel et al., 2006), RecA3 (Shedge et al., 2007), RecA1 (Odahara et al., 2009), and Msh1 (Abdelnoor et al., 2003). DNA from wild-type, why2-1, and why1 why3 plants were analyzed by DNA gel blot and probed with mitochondrial repeats as described by Arrieta-Montiel et al. (2009). DNA gel blot analyses revealed that both ciprofloxacin (see Supplemental Figure 5 online) and novobiocin (see Supplemental Figure 6 online) trigger mitochondrial recombination mediated by short repeated sequences. However, different recombination patterns were obtained for ciprofloxacin- and novobiocin-treated plants. This was expected since inhibition of DNA gyrase by these compounds leads to different outcomes. Indeed, ciprofloxacin triggers recombination-dependent repair of DSB (López and Blázquez, 2009), whereas novobiocin triggers recombination-dependent restart of stalled replication forks (Woelfle et al., 1993). Interestingly, the recombination patterns were similar for all plant genotypes. This suggests that Whirlies do not modulate the mitochondrial recombination mediated by short repeated sequences (50 to 1000 nucleotides).
The DNA Rearrangements Are Induced Mainly by Stresses Targeting the Organelles
We next treated wild-type, why2-1, and why1 why3 plants with bleomycin and hydroxyurea. In Arabidopsis, bleomycin triggers a DSB stress response (Menke et al., 2001; Tamura et al., 2002; Chen et al., 2008), whereas hydroxyurea induces a replication stress response (Ferreira et al., 1994; Roa et al., 2009). Treating wild-type and Whirly mutant plants with bleomycin (see Supplemental Figure 7 online) or hydroxyurea (see Supplemental Figure 8 online) did not result in an increase in plastid or mitochondrial MHMR. Furthermore, no increase in mitochondrial recombination mediated by short repeated sequences was observed upon treatment of Arabidopsis plants with these compounds (see Supplemental Figure 9 online). These results suggest that bleomycin and hydroxyurea have a weaker effect than do ciprofloxacin and novobiocin on the formation of DNA rearrangements in Arabidopsis organelle genomes. As shown in other species (Baugnet-Mahieu et al., 1971; Shen et al., 1995; Morel et al., 2008), this could be due to the fact that bleomycin and hydroxyurea target both the nucleus and the organelles, whereas compounds such as ciprofloxacin and novobiocin target mainly the organelles (Ye and Sayre, 1990). However, we cannot exclude the possibility that DNA damage induced by bleomycin and hydroxyurea in the organelles is repaired by alternative mechanisms.
The Crystal Structure of St-Why2 Reveals How ssDNA Is Bound
To understand how Whirly proteins prevent MHMR accumulation in the organelle genomes of Arabidopsis, we sought to determine how these proteins bind ssDNA. We thus solved the crystal structure of St-WHY248-216, a close homolog of At-WHY2 that also localizes to mitochondria (Vermel et al., 2002), in the free form and bound to ssDNA (Cappadocia et al., 2008). The structures were obtained at 2.2- to 2.7-Å resolution (Table 2). The St-WHY248-216 construct, hereafter called WHY2, encompasses the entire ssDNA binding domain but lacks the mitochondria targeting sequence and the acidic/aromatic C terminus (Figure 4A). These regions, which are not required for high affinity binding to the ssDNA (Cappadocia et al., 2008), are predicted to be disordered in solution. For the ssDNA sequence, we chose ERE32, a 32-nucleotide ssDNA derived from the elicitor response element (ERE) because the interaction between this sequence and plant Whirlies is well documented (Desveaux et al., 2000, 2004). WHY2 and WHY2-ERE32 crystallized in different chemical environments, yet they both gave rise to isomorphous crystals of the F432 space group. In the two crystals, WHY2 tetramers are disposed around fourfold crystallographic axes (Figures 4B and 4C). In the WHY2-ERE32 structure, electron density corresponding to ssDNA was clearly visible (Figure 4C), enabling us to build a nine-nucleotide model. Although we could observe electron density for most nucleobases, we were unable to assign unambiguously the ERE32 sequence into the electron density. We thus modeled the thymine-rich ERE32 oligonucleotide as an all-thymine oligonucleotide.
Data Collection and Refinement Statistics
Crystal Structures of St-WHY2 in the Free Form and Bound to ERE32 at 2.2-Å Resolution.
(A) Schematic representation of St-WHY2. Filled boxes indicate the position of the mitochondria transit peptide (mTP), the Whirly domain, and the acidic/aromatic C-terminal tail (CT). A green dotted line represents the construct used for structure determination.
(B) Overall view of St-WHY2 in the free form in cartoon representation. The tetramer was generated by applying crystallographic fourfold symmetry. Protomers are colored in yellow, orange, pink, and green.
(C) Surface representation of the protein moiety in the St-WHY2-ERE32 complex. The tetramer was generated by applying the crystallographic symmetry along the fourfold axis. Difference electron density was calculated from a Fo-Fc simulated annealing omit map encompassing DNA, contoured at 3.0 σ, and colored in green. The density was carved at 8 Å around the protein model.
Shown in Figure 4C is the electron density corresponding to ssDNA that was observed primarily on the edges and in between the β-sheets of adjacent protomers. Such a positioning of the DNA was unexpected. Indeed, while most single-stranded nucleic acid proteins use the core of their β-sheets as a primary binding platform (Horvath, 2008), WHY2-ssDNA interaction relies mainly on the binding of the DNA between properly positioned domains. As a consequence, binding to ssDNA exploits, but also depends on, the fourfold symmetry of the Whirly protein. The root mean square deviation (RMSD) of merely 0.5 Å for all Cα atoms between the free and the DNA-bound forms of WHY2 indicates that the protein does not undergo major conformational changes upon DNA binding. This is consistent with a need for properly positioned ssDNA binding residues.
Supplemental Figure 10 online shows that whereas good electron density corresponding to ssDNA is observed near the edges of the β-sheets, this density fades out on the top of the β-sheets before reappearing near the other edge of the β-sheets. This suggests that the ssDNA is disordered and/or adopts multiple conformations on the top of the β-sheets. We propose that, in this region, long DNA molecules, such as the ERE32, either thread from one ssDNA binding site to the other or enter/leave the complex.
WHY2 Binds ssDNA through a Conserved Mechanism
Figure 5A indicates that ssDNA is maintained in an extended conformation in each ssDNA binding site. The single-strand preference of WHY2 arises from steric impediment for the binding of a second strand at several positions on the binding surface. The presence of abrupt twists in the DNA backbone, such as between nucleotides 2 and 3, also argues against binding of a double-stranded DNA (dsDNA) helix. The mode of ssDNA binding is dominated by stacking and hydrophobic interactions between adjacent nucleobases and between nucleobases and aromatic/hydrophobic protein residues, consistent with the salt-resistant binding of St-WHY2 to ssDNA (Vermel et al., 2002). The polar contribution to DNA binding involves seven hydrogen bonds plus one water-mediated interaction (Figure 5A). Few sequence-specific interactions between St-WHY2 and ERE32 were observed (see below). Indeed, most of the nucleobases have their edges, containing the sequence-specific binding moieties, exposed to the solvent, whereas the faces of the nucleobases make intimate contact with residues of the protein surface.
Mechanism of ssDNA Binding of Whirly Proteins.
(A) Protein–DNA interactions in the St-WHY2-ERE32 complex. The DNA and the DNA-interacting residues are in stick representation with carbon atoms colored in yellow and in gray, respectively. Nucleotides and protein residues are labeled. Asterisks, residues that contact DNA through their main-chain. +, residues that contact the DNA through both its main chain and side chain. Hydrogen bonds are represented as yellow dashed lines. A water molecule is represented as a red sphere.
(B) Sequence alignment of the Whirly domain of Whirly proteins from Arabidopsis (At-WHY1, At-WHY2, and At-WHY3) and S. tuberosum (St-WHY1 and St-WHY2) with schematic secondary structural elements from St-WHY2. Secondary structure conformations are denoted at the top of the sequence alignment. α, α-helix; β, β-strand; η, 310 helix. Similarity above 70% is depicted in yellow, whereas perfect conservation is depicted in red. Blue stars underneath the sequence alignment indicate residues of St-WHY2 that interact with ssDNA. The alignments were made using T-coffee (Notredame et al., 2000) and the figure prepared using ESPript (Gouet et al., 2003).
A structure-based sequence alignment reveals that most of the ssDNA binding residues are conserved among different plastid- and mitochondria-targeted Whirlies (Figure 5B). Structural alignment of St-WHY1 (PDB 1L3A) with St-WHY2 also indicates a strong structure conservation both at the monomeric level with an RMSD of 1.4 Å for 159 matched Cα atoms and at the tetrameric level with an RMSD of 1.9 Å for 642 matched Cα atoms. Furthermore, structural superposition of St-WHY1 with St-WHY2 reveals that the key ssDNA binding residues are properly positioned in St-WHY1 to contact ssDNA. This supports the hypothesis that plastid- and mitochondria-targeted Whirlies have closely related ssDNA binding interfaces and is consistent with a common role of Whirly proteins in both organelles. Importantly, ssDNA binding residues are also conserved in Arabidopsis proteins, suggesting that the mechanism of ssDNA binding of St-WHY2 is similar to the ones of Arabidopsis WHY1, WHY2, and WHY3 (Figure 5B).
WHY2 Binds Nucleobases That Differ in Size and in Functional Groups
The local disorder of the electron density around the nucleobases at their junction with the β-sheets, as well as the scarcity of sequence-specific contacts between St-WHY2 and ERE32, suggests that WHY2 could bind other ssDNA sequences with high affinity. We thus assayed the binding of St-WHY2 to different A/T-rich ssDNA, as plant organelle genomes are generally A/T rich (Fauron et al., 2004; Ravi et al., 2008). Quantitative electrophoretic mobility shift assays (EMSAs) revealed that, despite their different nucleotide sequences, the oligonucleotides ERE32, cERE32, rcERE32, and dT32 all interacted with WHY2 with similar nanomolar affinities (Figure 6A, Table 3). To gain structural insights on how WHY2 could accommodate these DNA sequences, we obtained the crystal structure of these oligonucleotides bound to St-WHY2 (Table 2) and compared them to that of WHY2-ERE32. Comparison of the unbiased difference electron densities of the four oligonucleotides reveals that they adopt similar conformations in the ssDNA binding site (Figure 6B), suggesting that they interact with WHY2 through a common mechanism. The structures also reveal how WHY2 can accommodate bases that differ in size and in functional groups. Indeed, Ser-62 and Asp-142, which make sequence-specific interactions with T7 and T3 in the WHY2-dT32 and WHY2-ERE32 structures, respectively, make equivalent interactions with A7 and A3 in the structures of WHY2-cERE32 and WHY2-rcERE32 (Figures 6C and 6D). These compensating interactions demonstrate the capacity of WHY2 to bind different DNA sequences with high affinity.
WHY2 Binds ssDNA with Limited Sequence Specificity.
(A) Representative EMSA results showing the binding of St-WHY2 to four different ssDNA sequences. Increasing amounts of WHY2 were incubated with target oligonucleotides ERE32, dT32, rcERE32, or cERE32 and the complexes resolved on a 10% (w/v) polyacrylamide gel. The sequences of these oligonucleotides can be found in Table 3.
(B) Crystal structures of four different ssDNA sequences bound by St-WHY2. DNA molecules are presented as stick models with carbon atoms colored in yellow. Fo-Fc simulated annealing omit maps encompassing the entire DNA are contoured at 2.5 σ (colored in gray) or at 5 σ (colored in green). DNA molecules are presented in the same order as in (A).
(C) and (D) Interactions between St-WHY2 and the edges of T3 (left panel) and A3 (right panel) (C) or the edges of T7 (left panel) and A7 (right panel) (D) in the WHY2-ERE32 and WHY2-rcERE32 structures, respectively. The representation and the orientation of the molecule are similar, thus revealing that compensating interactions enable WHY2 to bind DNA nucleobases that differ in size and in functional groups at these positions. A red sphere corresponds to a water molecule.
Analysis of the Binding of St-WHY2 to Four Different DNA Sequences by EMSA
WHY2 Binds to Single-Stranded Overhangs, Melts dsDNA, and Protects ssDNA
An important step in the repair of DSBs is the rapid resection of DNA by exonucleases to generate DNA with single-stranded overhangs (Persky and Lovett, 2008). Because the choice of a repair pathway depends on the identity of the repair proteins that bind resected DNA ends, we verified that Whirly proteins could bind ssDNA overhangs. Figure 7A indicates that St-WHY2 can bind 16, but not 8, nucleotide-long 5′ or 3′ overhangs. Intriguingly, ssDNA molecules of the same size as these overhangs are not bound with high affinity by Whirlies (Yoo et al., 2007). This suggests either that the dsDNA moiety contributes to the high-affinity binding or that high-affinity binding is achieved through partially melting of the dsDNA moiety. As Whirlies were shown to have much less affinity for dsDNA than ssDNA (Desveaux et al., 2000; Prikryl et al., 2008), and since our structural data suggest that dsDNA is a poor substrate for Whirlies, we tested whether Whirlies could melt dsDNA to bind additional nucleotides. We incubated a radiolabeled DNA duplex with WHY2 and monitored the WHY2-induced denaturation of dsDNA by electrophoresis. Figure 7B shows that more ssDNA is produced as the concentration of WHY2 is increased. This indicates that WHY2 can destabilize DNA duplexes, probably by binding with a higher affinity to the single-stranded form of DNA and driving the equilibrium toward the formation of ssDNA. This could explain the capacity of WHY2 to bind short ssDNA overhangs with high affinity.
WHY2 Binds Single-Stranded Overhangs, Destabilizes dsDNA, and Protects ssDNA against Nuclease Degradation.
(A) Representative EMSA results showing the binding of St-WHY2 to DNA duplexes with or without single-stranded dT8 or dT16 overhangs. Fifty nanomolar of WHY2 were incubated with target radiolabeled oligonucleotides and the complexes resolved on a 10% (w/v) polyacrylamide gel. Diagrams at the bottom of the gels illustrate the DNA used in the assay. An asterisk indicates the strand that is radiolabeled.
(B) WHY2 destabilizes a DNA duplex. A DNA duplex with one strand radiolabeled was incubated with increasing amounts of St-WHY2 or BSA. After protein denaturation, DNA was resolved on a 7.5% acrylamide gel.
(C) WHY2 protects DNA against mung bean nuclease degradation. Phage M13mp18 ssDNA either alone or prebound with St-WHY2 at a 1:10 protein/nucleotide ratio was incubated with mung bean nuclease for the indicated amount of time. After protein denaturation, DNA was resolved on an agarose gel. Black/white inverted images are shown. M represents the molecular weight markers.
(D) WHY2 protects DNA against the exonuclease activity of T4 DNA Polymerase. Radiolabeled ssDNA or dsDNA with a 16-nucleotide 3′-overhang was complexed with St-WHY2 and then incubated with T4 DNA polymerase for the indicated amount of time. After protein denaturation, DNA was resolved on a 10% acrylamide gel. ssDNA, YG16_3; 3′-dsDNA, YG16_3-YC duplex; dsDNA, YG-YC duplex.
In addition to its ssDNA binding capacity, WHY2 effectively protects DNA against nuclease degradation. As shown in Figure 7C, phage M13mp18 ssDNA preincubated with a saturating amount of WHY2 is protected against digestion by mung bean (Vigna radiata) nuclease, whereas free M13mp18 ssDNA is readily digested. Also, Figure 7D shows that WHY2 can protect the single-stranded overhang of a duplex against degradation by the exonuclease activity of T4 DNA polymerase. The fact that WHY2 protects ssDNA against both mung bean nuclease and T4 DNA polymerase likely suggests that WHY2 can prevent these nucleases from accessing the ssDNA.
DISCUSSION
Binding of Whirly Proteins to Nucleic Acids
There have been conflicting reports concerning the sequence specificity of Whirly proteins. Indeed, Whirlies were reported to bind three nuclear sequences that share little sequence similarity with each other: an ERE element (Desveaux et al., 2000, 2004), an Arabidopsis telomeric repeat (Yoo et al., 2007), and a distal element upstream of the Arabidopsis kinesin gene KP1 (Xiong et al., 2009). In plant organelles, Whirly proteins were shown to bind many regions of the plastid or the mitochondrial genome without obvious sequence consensus in vivo (Maréchal et al., 2008, 2009; Prikryl et al., 2008). This work demonstrates that Whirlies can bind ssDNA molecules that differ in nucleotide sequence using a binding mechanism involving mainly contacts with nonsequence-specific moieties of the ssDNA and allowing Whirlies to establish equivalent interactions with different nucleobases. Because Whirlies fused to green fluorescent protein accumulate abundantly in the organelles, whereas they are difficult to detect in the nucleus (Krause et al., 2005), we propose that, in the nucleus, the low concentration of Whirly proteins would enable them to bind only a limited number of sites that have a pronounced single-stranded character (e.g., A/T-rich sequences of transcribed regions or telomeric repeats). In the organelles, the abundance of Whirly proteins and their limited sequence specificity would enable them to bind all available ssDNA.
In addition to ssDNA, Whirlies were shown to bind RNA in maize (Prikryl et al., 2008). Furthermore, a close structural neighbor of plant Whirlies also binds RNA (Schumacher et al., 2006). Our crystallographic models are consistent with a role of Whirly proteins in RNA metabolism. Indeed, modeling of ssRNA in place of ssDNA leads only to small clashes, thus suggesting that RNA could be accommodated in an ssDNA-like conformation (see Supplemental Figure 11 online). However, RNA appears to be bound less strongly than DNA because Whirlies were found to be associated with DNA from throughout the genome but with only a subset of RNA molecules in vivo (Prikryl et al., 2008).
DSBs Can Be Repaired through a Microhomology-Mediated Break-Induced Replication Mechanism in Plant Organelles
Plants treated with ciprofloxacin accumulate rearranged DNA molecules in both plastids and mitochondria. Most of the DNA rearrangements contain microhomologies at their junction, suggesting that at least part of the ciprofloxacin-induced DSBs are repaired through an error-prone microhomology-dependent pathway. We previously proposed that a microhomology-mediated break-induced replication (MMBIR) pathway could explain the accumulation of MHMR in both untreated why1 why3 Arabidopsis and Why1 knockdown maize plants (Maréchal et al., 2009). The MMBIR pathway relies on the microhomology-dependent restart of DNA synthesis on a different template following collapse of the replication fork (Hastings et al., 2009a). Microhomologies are sufficient to initiate DNA synthesis as certain DNA polymerases can efficiently use mismatched primers or primers as short as 2 to 3 bp as substrates (Cannistraro and Taylor, 2007; Hastings et al., 2009a). Our present finding that MHMR events that appear in untreated plants are increased upon ciprofloxacin treatment suggests that a MMBIR pathway is involved in the error-prone repair of DSBs in organelles.
The use of a MMBIR pathway is consistent with our observation of both duplications/circularization and deletion events, whereas pathways such as microhomology-mediated end joining or single-strand annealing can account only for deletion events (McVey and Lee, 2008). A NHEJ repair pathway could be involved in the formation of NHEJ-like events. However, there is little support for the existence of such a repair pathway in plant organelles, particularly in plastids (Odom et al., 2008; Kohl and Bock, 2009). Furthermore, whereas MHMRs are detected in the absence or presence of ciprofloxacin, all but one NHEJ-like event are detected in ciprofloxacin-treated plants, thus raising the possibility that these events could be by-products of the ciprofloxacin-mediated DNA gyrase inhibition (Marvo et al., 1983). The fact that mismatches are detected in 91 out of 158 MHMR junctions relative to one or both parental strands (see Supplemental Data Set 1 online) is consistent with the low-specificity requirements of the MMBIR pairing process. Finally, MMBIR can account for complex cases of DNA rearrangements (Hastings et al., 2009a), including the formation of molecules containing multiple MHMRs, which have also been observed in our study (Table 1). Our results also suggest that MHMR products exist at substoichiometric levels. These DNA variants could accumulate to higher levels through a process known as substoichiometric shifting, which is well documented for mitochondrial DNA rearrangements (Small et al., 1987; Janska et al., 1998; Maréchal and Brisson, 2010; Woloszynska, 2010). The fact that, in why1 why3 plants, MHMR products present at low abundance in the chloroplasts of green sectors accumulate to high levels in the variegated sectors (Maréchal et al., 2009) suggests that this substoichiometric shifting could also occur in chloroplasts. A substoichiometric shift may also facilitate the transmission of rearranged molecules. Indeed, in the why1 why3 plants, variegation could readily be inherited provided that the female parent displayed a strong variegation phenotype (Maréchal et al., 2009). In this study, however, no increase in DNA rearrangements was observed in the progeny of plants treated with 0.25 μM ciprofloxacin. It is thus possible that DNA rearrangements, when present at substoichiometric levels, have poor transmission efficiency. Alternatively, germ cells may be less sensitive to ciprofloxacin treatment than are somatic cells. Globally, the characteristics of the MHMR events obtained in this study support the idea that MMBIR acts as a DNA repair pathway in plant organelles (Figure 8).
Model for the Repair of Organellar Double-Strand Breaks in the Absence or Presence of Whirly Proteins.
Upon formation of a DSB (1), the 5′ end of the broken DNA molecule is resected from the break, exposing a 3′ tail (2). At this step, the break can be repaired through homologous recombination in a Whirly-independent manner. Alternatively, if the Whirlies are absent or the repair machinery is overloaded due to numerous DSBs, the 3′ tail can anneal to any exposed ssDNA through microhomologies (3a). A D-loop forms and DNA polymerization proceeds from the microhomology junction (4a). A replication fork is established and lagging strand synthesis initiates while leading strand synthesis continues (5a). DNA synthesis continues until the end of the chromosome is reached (6a). Alternatively, if the Whirlies are present and the DSB level is low (3b), Whirlies could bind and protect ssDNA (either the 3′ tail and/or any exposed ssDNA), thereby promoting homologous recombination and accurate DNA repair. Arrowheads represent 3′ ends; a box symbolizes the microhomology between broken and unbroken DNA molecules; dashed arrows in tandem represent lagging strand synthesis.
[See online article for color version of this figure.]
Effects of DSBs and Whirly Mutants on Mitochondrial Recombination Mediated by Short Repeated Sequences (50 to 1000 Nucleotides)
Plants treated with ciprofloxacin or novobiocin displayed an increase in mitochondrial recombination mediated by short repeated sequences. As the plastid genome of Arabidopsis does not contain short repeated sequences capable of supporting aberrant homologous recombination (Maréchal and Brisson, 2010), we were unable to determine if ciprofloxacin or novobiocin affected homologous recombination in plastids. Whirlies do not appear to modulate mitochondrial DNA recombination mediated by short repeated sequences. This was expected for the plastid-localized proteins WHY1 and WHY3 and is consistent with our previous observations that DNA recombination products mediated by short repeats do not accumulate in untreated Arabidopsis plants lacking the mitochondria-localized WHY2 (Maréchal et al., 2008). As At-WHY2 appears to modulate only the DSB-induced mitochondrial MMBIR pathway, it is possible that these repair responses are mechanistically distinct.
However, we cannot rule out a role for WHY2 in homologous recombination because other proteins may function redundantly with the Whirlies in this process. Indeed, many different ssDNA binding proteins are located in mitochondria, including the OSB proteins (Zaegel et al., 2006) and the mitochondrial ssDNA binding proteins (Edmondson et al., 2005).
Whirly Proteins Share Common Characteristics with Eubacterial ssDNA Binding Proteins
The eubacterial ssDNA binding proteins SSBs are critical players during DNA replication, recombination, and repair processes. Often envisioned as simple ssDNA-coating proteins, they actually work as organizational scaffolds that recruit genome maintenance complexes when and where they are most required (Cox, 2007; Shereda et al., 2008). These proteins share a number of characteristics with the Whirly proteins even though they are not structurally related: (1) they are both tetrameric proteins that bind ssDNA with high affinity and modest sequence preference, (2) they use mainly hydrophobic/aromatic residues to contact ssDNA and, as a result, their binding to ssDNA is salt resistant, (3) they can destabilize dsDNA, (4) they can prevent degradation of ssDNA by nucleases, and (5) they both possess a conserved C-terminal acidic/aromatic tail. Usually tethered to the protein surface, the acidic/aromatic C terminus of the eubacterial ssDNA binding proteins SSBs is believed to be freed and rendered available for interacting with target proteins upon ssDNA binding (Lu and Keck, 2008; Shereda et al., 2008, 2009). Interestingly, structural comparison between St-WHY1 and St-WHY2 revealed that the C terminus in WHY1 and the ssDNA in WHY2 both interact with the same residues in the β-sheets. This demonstrates that the C terminus and the ssDNA can compete for the same binding surface on the Whirly proteins and that their interaction with the Whirly domain is mutually exclusive. The competition between ssDNA and an acidic/aromatic C-terminal tail is consistent with our previous results showing that a St-WHY1 protein mutated at its C terminus has an increased affinity for ssDNA (Desveaux et al., 2005). Our results are thus consistent with a model in which, upon binding of Whirly proteins to ssDNA, the acidic-aromatic C-terminal tail would become available for interacting with target proteins in an SSB-like mode.
Possible Roles for Whirly Proteins in the Repair of DSBs
Whirlies could maintain genome stability by favoring the repair of DSBs by error-free homologous recombination and/or by restricting the DSB repair by error-prone MMBIR. Based on our results, we propose that Whirlies could bind and protect resected DNA ends in a sequence-independent manner at break sites (Figure 8). Upon binding to resected DNA ends, Whirlies would prevent annealing of ssDNA overhangs to microhomologous sequences present in stretches of ssDNA elsewhere in the genome. Alternatively, or concomitantly, Whirlies could prevent MMBIR by binding to these ssDNA stretches, which may occur in many places in the genome, including in DNA secondary structures, in replication forks, in stalled transcription complexes, and in promoter regions (Hastings et al., 2009a). In addition, Whirlies might regulate the access of repair proteins to ssDNA by competing with them for binding to ssDNA, thereby influencing the choice of a repair pathway. Whirlies could even promote the accurate repair of DSBs by recruiting proteins involved in DNA repair through their acidic/aromatic C terminus in an SSB-like fashion.
No accumulation of rearranged DNA molecules could be detected in the mitochondria of why2-1 plants in absence of ciprofloxacin treatment. This could suggest that mitochondrial DNA is less subject to genotoxic stresses under normal growth conditions or that WHY2 fulfils a different function than do WHY1 and WHY3 in Arabidopsis. A partial functional redundancy by the OSBs and mitochondrial SSBs could also explain why, in the absence of DNA stresses, why2-1 plants do not show an increase in genome instability, contrary to what is observed in plastids of Arabidopsis and maize plants lacking plastid-localized Whirlies (Maréchal et al., 2009). This idea is consistent with the apparent lack of SSB homologs targeted to plastids. A functional redundancy could also explain our observation that, upon DSB induction, DNA rearrangements accumulate more readily in plastids of why1 why3 plants than in mitochondria of why2-1 plants. Further work is needed to determine if WHY2 works redundantly with other mitochondria-localized ssDNA binding proteins or if the protection against MHMR observed in why2-1 mutants could be due to a more general role of WHY2 in DNA metabolism.
Interestingly, TIF1, a distant homolog of the plant Whirlies (Saha et al., 2001) safeguards chromosomes from DNA damage in both nuclei of the ciliate Tetrahymena thermophyla. TIF1-deficient cells are hypersensitive to genotoxic stress and are defective in the activation of the intra-S phase checkpoint mediated by the sensor/transducer kinase ATR (Yakisich et al., 2006). These studies suggest that a role of Whirlies in the maintenance of genome stability may be conserved between plants and ciliates.
Our finding that MHMR products accumulate at high concentrations of ciprofloxacin even in the presence of Whirly proteins suggests that accurate repair of DNA may be bypassed if severe DNA damage occurs. Interestingly, in other organisms, MHMR was detected mainly when canonical repair pathways such as NHEJ or homologous recombination were overloaded or challenged, thus leading to the idea that microhomology-mediated DNA repair acts as a backup DNA repair pathway (Wang et al., 2003). In Arabidopsis organelles, multiple DSBs may cause depletion of repair proteins due to their recruitment to multiple damage sites, allowing the accumulation of MHMRs. Our observation that MHMR products accumulate to high levels at 0.75 μM ciprofloxacin both in the absence and the presence of Whirly proteins is consistent with this idea. However, the use of an error-prone repair pathway under DNA stress may be more than a backup pathway. Indeed, as MHMR enables DNA rearrangements including copy number variation (Hastings et al., 2009a), it promotes rapid changes in the genome. Some of these changes may confer selective advantages to plants under stress. A similar mechanism, termed stress-induced mutation, exists in bacteria and is also dependent on error-prone repair of DSBs (Ponder et al., 2005). The fact that certain plants maintain DNA rearrangements mediated by microhomologies in their organellar genome provides indirect support for their selective advantage (Ogihara et al., 1988; Kanno et al., 1993; Moeykens et al., 1995).
METHODS
Plant Material and Growth Conditions
Sterilized seeds of Arabidopsis thaliana ecotype Col (wild type), the why1 why3 double mutant, and the why2 single mutant were germinated on Murashige and Skoog basal media (Sigma-Aldrich) containing ciprofloxacin or novobiocin at the indicated concentrations and collected after 3 weeks of growth under long-day conditions (16 h light/8 h dark). Bleomycin and hydroxyurea treatments were performed as described (Chen et al., 2008; Roa et al., 2009). The why2-1 and why1 why3 mutant lines (formerly called KO2 and KO1/3, respectively) were described previously (Maréchal et al., 2008, 2009). An additional Arabidopsis line lacking WHY2 (why2-2; SALK_016156) was tested for its sensitivity to the DNA damaging agents and the generation of MHMR products during ciprofloxacin treatment (see Supplemental Figure 12C online) and similar results were obtained. A protein gel blot was performed using α-AtWhy2 polyclonal antibodies (Maréchal et al., 2008) to verify that both why2 lines are completely devoid of WHY2 protein (see Supplemental Figure 12 online).
Detection of DNA Rearrangements
DNA was isolated from plants using a cetyl trimethylammonium bromide DNA extraction protocol (Weigel and Glazebrook, 2002). PCR was conducted using the Taq polymerase (Genscript) according to the manufacturer's instructions. DNA rearrangement events were detected using both outward- and inward-facing PCR primers spaced by ∼5 to 30 kb. Thirteen and 10 PCR primer pairs scattered in the plastid genome were used for ciprofloxacin- and novobiocin-treated plants, respectively. Twenty and 10 PCR primer pairs scattered in the mitochondrial genome were used for ciprofloxacin- and novobiocin-treated plants, respectively. The sequences of the PCR primers are listed in Supplemental Table 4 online. PCR was performed on DNA samples from wild-type, why2, and why1 why3 plants and analyzed by gel electrophoresis. All visible DNA bands were isolated and cloned into the pDrive vector (Qiagen). The DH5α strain was used for transformation, and the plasmids from two randomly picked colonies were sequenced. When both sequencing products were identical, only one product was considered, whereas when sequencing products differed, both products were considered. Sequencing of visible bands confirmed that >96% of all cloned PCR products were due to DNA rearrangements, with the remaining events being caused by nonspecific annealing of the oligonucleotides at undesired locations. DNA rearrangements are listed in Supplemental Data Set 1 online.
Analysis of DNA Rearrangements by DNA Gel Blots
To evaluate the relative abundance of MHMRs compared with normal organelle genomes, rearranged DNA molecules were amplified by PCR using the oligonucleotides listed in Supplemental Table 4 online. Cloning and sequencing of PCR products confirmed the occurrence of DNA rearrangements (see Supplemental Table 3 online). PCR products were internally radiolabeled using [α-32P]dATP (6000 Ci/mmol; Perkin-Elmer Life Science) and Klenow polymerase (Fermentas) and were used as probes. DNA gel blots were performed as described previously (Maréchal et al., 2009). Detection of recombination mediated by short repeats (50 to 1000 nucleotides) was performed as described (Arrieta-Montiel et al., 2009). Briefly, PCR primers listed in Supplemental Table 4 online were used to amplify three mitochondrial repeats. The repeats were internally radiolabeled using [α-32P]dCTP (3000 Ci/mmol; Perkin-Elmer Life Science) and Klenow polymerase (Fermentas) and were used as probes.
St-WHY2 Cloning, Expression, and Purification
Details concerning cloning, expression, and purification of St-WHY2 are reported elsewhere (Cappadocia et al., 2008). Briefly, WHY248-216 was amplified from Solanum tuberosum total RNA and cloned into the plasmid vector pET-21a (Novagen), which encodes a C-terminal hexahistidine tag. St-WHY2 was expressed in Escherichia coli BL21 (DE3) strain with 1 mM isopropyl β-d-thiogalactopyranoside for 3 h at 37°C. The cells were lysed by alumina grinding and the lysate resuspended in 20 mM sodium phosphate buffer, pH 7.5, containing 500 mM NaCl and 25 mM imidazole. The recombinant protein was purified by applying the supernatant to a HiTrap Chelating column (GE Healthcare) followed by size exclusion chromatography on a Superdex 200 16/60 (GE Healthcare) in 20 mM Tris-HCl, pH 8.0, 100 mM NaCl.
Data Collection and Structure Determination
St-WHY2 in the free form and St-WHY2-ssDNA complexes were crystallized as previously described (Cappadocia et al., 2008). Briefly, protein-DNA complexes were crystallized by the hanging drop vapor diffusion method using a precipitant solution containing 100 mM Tris/HCl, pH 8.0, 13.5 to 25% (w/v) PEG6000, and 1 to 2 M LiCl. The free protein was crystallized by the hanging drop vapor diffusion method using a precipitant solution containing 100 mM MOPS, pH 7.0, 21 to 27% (w/v) PEG1000, and 100 to 400 mM NH4H2PO4. Diffraction data were collected using ADSC Quantum 315 CCD detectors at beamlines X25 and X29 at the National Synchrotron Light Source (NSLS) at the Brookhaven National Laboratory or using a MAR-225 CCD detector at beamline 22-ID at the Advanced Photon Source at the Argonne National Laboratory. Initial phases for the St-WHY2-ERE32 complex were obtained by molecular replacement using St-WHY1 (PDB 1L3A) as a search template. Phases were improved by iterative cycles of model building with Coot (Emsley and Cowtan, 2004) and refinement with CNS (Brünger et al., 1998) and Phenix (Adams et al., 2002). DNA nucleotides were modeled one nucleobase at a time from difference electron density maps. The refined structure of St-WHY2-ERE32 was used as a molecular replacement search template for structural determination of data sets collected from crystal of other St-WHY2-ssDNA complexes. Clear density was visible for protein residues 55 to 215 in St-WHY2. Since we could not attribute unambiguously the specific sequences of the ERE32, rcERE32, and cERE32 oligonucleotides to the electron density, we assigned these oligonucleotides as all-thymine, all-adenine, and all-adenine oligonucleotides, respectively. The DNA occupancy in each structure was calculated from the temperature factors and it ranges from 0.7 to 0.8. Test data sets were randomly selected from the observed reflections prior to refinement. The figures were prepared using PyMOL (DeLano, 2002).
EMSAs
The apparent dissociation constants for St-WHY2 bound to DNA were determined by EMSA. The ssDNA oligonucleotides were obtained from Integrated DNA Technologies and radiolabeled at the 5′ end using [γ-32P]ATP (6000 Ci/mmol; Perkin-Elmer Life Science) and T4 polynucleotide kinase (Fermentas) following the manufacturer's instructions. Binding reactions were done for 30 min at 20°C by incubating 15,000 cpm radiolabeled ssDNA (<100 pM) with the serially diluted St-WHY2 protein in a buffer containing 10 mM Tris-HCl, pH 8.0, 150 mM KCl, 0.5 mM EDTA, pH 8.0, and 15% (v/v) glycerol. Reactions were loaded on 10% (w/w) 29:1 acrylamide:bisacrylamide gels with 4.5 mM Tris, 4.5 mM boric acid, and 1 mM EDTA, pH 8.0, at 20°C under 150 V constant voltage for 20 min and then subjected to autoradiography. The autoradiograms were scanned and the intensity of the bands corresponding to bound and free radiolabeled ssDNA were quantified using ImageJ (NIH; http://rsb.info.nih.gov/ij/). The data was plotted and analyzed using Prism 5 Demo (GraphPad Software). The apparent dissociation constants were calculated by fitting the plot of the fraction of ssDNA bound versus protein concentration to the Hill equation.
The binding of St-WHY2 to ssDNA overhangs was also assessed by EMSA using DNA duplex formed by annealing two of the following oligonucleotides: YG, YC, YG8_3, YC8_5, YG16_3, and YC16_5. The sequences of these oligonucleotides are shown in Supplemental Table 4 online.
DNA Destabilization Assay
Oligonucleotide R2 was radiolabeled at the 5′ end using [γ-32P]ATP (6000 Ci/mmol; Perkin-Elmer Life Science) and T4 polynucleotide kinase (Fermentas) following the manufacturer's instructions and annealed with oligonucleotide R1. The annealed duplex was incubated with St-WHY2 or BSA for 30 min at 20°C. The reactions were stopped by the addition of a 40-fold excess of unlabeled R2 and SDS to a final concentration of 0.2%. Reactions were separated on 7.5% (w/w) 29:1 acrylamide:bisacrylamide gels with 4.5 mM Tris, 4.5 mM boric acid, and 1 mM EDTA, pH 8.0, at 20°C under 130 V constant voltage for 30 min. The sequences of oligonucleotides R1 and R2 are shown in Supplemental Table 4 online.
Mung Bean Nuclease Protection Assay
Two hundred nanograms of M13mp18 ssDNA (USB) either alone or prebound for 30 min at 20°C with St-WHY2 at a 1:10 protein/nucleotide ratio were incubated with 1U of mung bean (Vigna radiata) nuclease (New England Biolabs) for 0, 15, 30, or 60 min at 30°C. Reactions were stopped by incubation on ice and addition of SDS and EDTA to final concentrations of 0.1% (w/v) and 1.5 mM, respectively. Reactions were loaded on 0.7% (w/w) agarose gels and migrated under 100 V of constant voltage.
T4 DNA Polymerase Protection Assay
ssDNA or a dsDNA with a 3′ overhang of 16 nucleotides (3′-dsDNA) either alone or prebound for 30 min at 20°C with 1 μM St-WHY2 was incubated with 0.02 units of T4 DNA Polymerase (T4; Fermentas) for 0, 15, 30, or 60 min at 20°C. Reactions were stopped by incubation on ice and addition of SDS to a final concentration of 0.1% (w/v). Reactions were separated on 10% (w/w) 29:1 acrylamide:bisacrylamide gels with 4.5 mM Tris, 4.5 mM boric acid, and 1 mM EDTA, pH 8.0, at 20°C under 150 V constant voltage for 20 min and then subjected to autoradiography.
Accession Numbers
Sequence data from this article can be found in the Arabidopsis Genome Initiative, GenBank/EMBL, or The Institute for Genomic Research database under the following accession numbers: St-WHY2 (HM234504), At-WHY1 (AT1G14410), At-WHY2 (AT1G71260), and At-WHY3 (AT2G02740). The accession numbers for the mitochondrial and plastid genomes are NC_001284 (Unseld et al., 1997) and NC_000932 (Sato et al., 1999), respectively. The atomic coordinates of St-WHY2 in the free form, St-WHY2-ERE32, St-WHY2-dT32, St-WHY2-cERE32, and St-WHY2-rcERE32, have been deposited in the Protein Data Bank (www.rcsb.org) under ID codes 3N1H, 3N1I, 3N1J, 3N1K, and 3N1L, respectively. The ID code for the crystal structure of St-WHY1 in the free form is 1L3A.
Supplemental Data
The following materials are available in the online version of this article.
Supplemental Figure 1. Effect of Ciprofloxacin Treatment on Arabidopsis Plant Survival.
Supplemental Figure 2. DNA Rearrangements Accumulate Primarily in Plastids of Plants Lacking Plastid-Targeted Whirlies and in the Mitochondria of Plants Lacking Mitochondria-Targeted Whirlies.
Supplemental Figure 3. DNA Gel Blot Detection of DNA Rearrangements in Plastids Following Plant Treatment with Ciprofloxacin.
Supplemental Figure 4. DNA Gel Blot Detection of DNA Rearrangements in Mitochondria Following Plant Treatment with Ciprofloxacin.
Supplemental Figure 5. Analysis of Mitochondrial Recombination in Plants Treated with Ciprofloxacin.
Supplemental Figure 6. Analysis of Mitochondrial Recombination in Plants Treated with Novobiocin.
Supplemental Figure 7. Effects of Bleomycin on Plastid and Mitochondria DNA Rearrangements of Plants Lacking Plastid- or Mitochondria-Targeted Whirlies.
Supplemental Figure 8. Effects of Hydroxyurea on Plastid and Mitochondria DNA Rearrangements of Plants Lacking Plastid- or Mitochondria-Targeted Whirlies.
Supplemental Figure 9. Analysis of Mitochondrial Recombination in Wild-Type Plants Treated with Bleomycin or Hydroxyurea.
Supplemental Figure 10. Single-Stranded DNA Is Stabilized on the Edges of the β-Sheets of Whirly Proteins.
Supplemental Figure 11. Modeling of a St-WHY2-RNA Complex.
Supplemental Figure 12. Molecular Characterization of why2 Lines.
Supplemental Table 1. Related DNA Rearrangements Detected in the Mitochondrial and Plastid Genomes.
Supplemental Table 2. DNA Rearrangement Statistics.
Supplemental Table 3. DNA Rearrangement Products Detected in the Mitochondrial and Plastid Genomes and Used as Probes in DNA Gel Blots.
Supplemental Table 4. Oligonucleotides Used in This Study.
Supplemental Data Set 1. DNA Rearrangement Products Identified in the Mitochondrial and in the Plastid Genomes.
Acknowledgments
The assistance of Mathieu Coinçon, Sandra Grondin, and Bruno Piché is gratefully acknowledged. We thank James Omichinski for valuable discussions. We also thank the High Throughput Crystallization Service at the Hauptman-Woodward Institute, Buffalo, NY, for determining the initial crystallization conditions. Research carried out at the NSLS and at the APS was supported by the U.S. Department of Energy, Division of Materials Sciences and Division of Chemical Sciences, and by the U.S. Department of Energy, Office of Science, respectively. Assistance by X25, X29, and 22ID beamline personnel is gratefully appreciated. L.C. and A.M. were supported by fellowships from the Natural Sciences and Engineering Research Council of Canada (NSERC) and from the Fonds québécois de la recherche sur la nature et les technologies (FQRNT). This research was supported by grants from NSERC and FQRNT to both N.B. and J.S. and from the Canadian Institutes of Health Research to J.S.
Footnotes
The author responsible for distribution of materials integral to the findings presented in this article in accordance with the policy described in the Instructions for Authors (www.plantcell.org) is: Normand Brisson (normand.brisson{at}umontreal.ca).
↵1 Current address: Massachusetts General Hospital Cancer Center and Harvard Medical School, Charlestown, MA 02129.
↵2 These authors contributed equally to this work.
↵[C] Some figures in this article are displayed in color online but in black and white in the print edition.
↵[W] Online version contains Web-only data.
- Received September 15, 2009.
- Revised May 13, 2010.
- Accepted May 25, 2010.
- Published June 15, 2010.
References
